Packet- Speed- More Speed

191

description

Packet- Speed- More Speed and applications

Transcript of Packet- Speed- More Speed

Page 1: Packet- Speed- More Speed
Page 2: Packet- Speed- More Speed

Packet: Speed, More Speedand Applications

A Collection of Advanced Packet Methodsand Activities from ARRL Publications

and Other Sources

Compiled by Rich Roznoy, K10F

Production Staff:Paul LappenJoe Shea

Cover:Design: Sue Fagan

Published by:The American Radio Relay LeagueNewington, CT 06111 USA

Page 3: Packet- Speed- More Speed

Copyright © 1997 byThe Ame rican Radio Relay League , Inc.

Copyright secured under the Pan-AmericanConvention

International Copyr ight secured

This work is Publication No. 193 of the RadioAmateur's Library, published by the Leag ue.All rights reserved. No part of this work maybe reproduced in any form except by writtenpermiss ion of the publisher. All rights oftranslation are reserved .

Printed in the USA

Quedan reservados todos los derechos

Second EditionFirst print ing

ISBN: 0-87259-605-2

Page 4: Packet- Speed- More Speed

ForewordThe 2nd edition of Packet : Speed. More Speed is indeed "more ." Picki ng up

where the Ist edit ion left off, it provides, all in one conve nient volume , an impres­sive compi lation of articles and other informa tio n on advance d packet applications .The best news is that there's 110 carryover fro m the I st edition: If yo u have a copyof the 1st edition, there 's no dupli cati on in this book.

Resourceful amateurs have used commercial techn ology developed du ring thepas t 30 years to make packet radio one of our hobby ' s most exc iting modes. In theea rly days, receiving error free data was the principal goal. Today that is accom­plished eas ily . No w the challe nge is to ha ve thi s error free data represent so methingmore than plain text messages. With the new ap plica tions now ava ilable, data mayrepresent maps, geographic locat ions, weather, DX stations and much more. Aseries of imp ress ive new uses is possible due to the sk ill and generosity of BobBruninga, WB4APR, who deve loped the APRS (A utomatic Packet ReportingSystem) . He coupled a packet modem and a hand-held transcei ver with a GPS(Global Positio ning System), allowing us to track rea l-time eve nts such as the pathof a balloon as it dr ifts across the countryside . Mark and Keith Sproul, KB2 ICI andWU2Z respectively, are developing add-on applica tio ns for APRS . Skywarnwea ther reporting, DX Clus ter mon itoring and high resolutio n map dis plays are ju sta few of these inn ovat ions.

As 1200 baud has become ro utine, innovators like Da le Heatherington, WA4D SYand TAPR (Tucson Amateur Packe t Radio) are drawin g on commercial technolo­gies to give us a 50-fo ld increase in speed . If you, like many of us, don 't have ahigh speed modem, you sti ll ca n enjoy an exci ting new use for your l 200-baudstation. Fuji -OSCAR 29 offers us the firs t rea lly user-friend ly sate llite BBS. Youdo n't need any special equipment; a 1200-b aud TNC used for yo ur local BBS wi llwork just fine .

As yo u can see, there are a lot of exci ting new thin gs to do on packet. Thi s bookprovides you with the most co mplete single so urce of advanced packet infor matio navailable. Let us know what yo u are doin g and what mat erial you would like to seein futu re ed itio ns. A handy feedback form is at the bac k for your con venience.

Da vid Sum ner, KIZZExec utive Vice President

Newi ngton, Connec ticutMarch 1997

iii

Page 5: Packet- Speed- More Speed

Contents

Chapter 1: Operating1-1 "Amateur Radio Meets the Internet" by Steve Ford, WB81MY

from January 1995 QST

1-4 "Mess ages via Packe t/Internet Gateways" by Stan Horzepa, WA1LOUfrom June 1995 QST

1-5 "KISSe s, POPs and Pings" by Steve Ford, WB81MYfrom June 1995 QST

1-8 "Exploring the 9600 -Baud PACSATs" by Andre w Cornwall, VE1CORfrom June 1995 QST

1-11 "DX PacketClusters" by Steve Ford, WB81MYfrom March 1996 QST

1-15 "Chasing DX with APRS" by Stan Horzepa, WA1 LOUfrom October 1996 QST

1-16 "Pactor-II, Impressions and Update Information one year after" by Tom Rink and Dipl.-Ing. Martin Clas,DL1ZAM

from May 1996 Dig ital Journal

1-19 "The HAL P38 DSP, and other thoughts" by Hal Blegen, WA7EGAfrom August 1995 Digital Journal

1-21 "Fuji-OSCAR-29"from the AMSAT Web site

1-22 "WinAPRS: Windows Automatic Posit ion Reporting System" by Mark Sproul , KB21CI and Keith Sprou l,WU2Z

from the 15th ARRL and TAPR Digital Communications Conference

1-28 "Automatic Radio Direction Finding Using MacAPRS & WinAPRS" by Keith Sproul, WU2Zfrom the 15th ARRL and TAPR Digita l Communications Conference

1-33 "Packet and Internet" by James Wagner, KA7EHKfrom the 15th ARRL and TAPR Digital Communications Conference

1-36 "SUNSAT: A Mic ro Satell ite Unde r Construction in South Afr ica"from the Universitat Kaiserslauten Web Page

1-38 "ACARS: Packet for Ai rplanes" by Donald Cox, AA3EKfrom November 1996 QST

1-41 "SCS PTC-II Mult imode Controller with Pactor-II" by Steve Ford, WB81MYfrom January 1997 QST

1-43 "Geminids Packet Meteor-Scatter Test Results" by Stan Horze pa, WA1LOUfrom April 1996 QST

1-44 "On-A ir Measurements of HF Data Throughput , Results and Reflecti ons," by Ken Wickwi re, KB1JYfrom March 1996 Digital Journal

1-55 "Portable Packets" and "Software For the BP-2 and BP-2M" by Phillip Nicho ls, KC8DQFfrom March 1997 QST and from Tigertronics Web page

'''9600-Ready' Radios: Ready or Not?" by Jon Bloom, KE3Zfrom May 1995 QST

"A Comparison of HF Digital Protocols" by Tim Riley; Dennis Bodson, W4PWF; Stephen Rieman; andTeresa G. Sparkman

from July 1996 QST

"Measuring 9600-Baud Radio BER Performance" by Jon Bloom, KE3Zfrom March 1995 QEX

2-7

2-12

Chapter 2: Theory/Design2-1

iv

Page 6: Packet- Speed- More Speed

2-20 "The WA4DSY 56 Kilobaud RF Modem" by Dale A. Heatherington , WA4DSYfrom the 14th ARRL and TAPR Digital Communications Conference

2-32 "Data Radio Standard Test Methods, by Burton Lang , VE2BMQ, and Donald Rotolo , N21RZfrom the 14th ARRL and TAPR Digital Communications Conference

2-37 "javAPRS: Implementation of the APRS Protocols in Java" by Steve Dimse, K04HDfrom the 15th ARRL and TAPR Digital Communications Conference

2-44 "On-Air Measurements of MIL-STD-188-141A ALE Data Text Message Throughput Over Short Links"by Ken Wickwire, KB1JY

from the 15th ARRL and TAPR Digital Communications Conference

2-48 "CLOVER- The Technology Grows and Matures" by Bill Henry, K9GWTfrom the 15th ARRL and TAPR Digital Communications Conference

" RMNC/FlexNet: The Network of Choice in Western Europe" by Eric Bertrem, F5PJEfrom Apr il 1995 QST

"Availability of Seventy 9600 Baud Packet Channels on Two Meters" by Bob Bruninga, WB4APRfrom the 14th ARRL and TAPR Digita l Communications Conference

"Building a Packet Network" by Karl Medcalf, WK5Mfrom the 14th ARRL and TAPR Digital Communications Conference

"The Puget Sound Amateur Radio TCP/IP Network" by Steve Stroh, N8GNJfrom the 14th ARRL and TAPR Digital Communications Conference

3-13

3-2

3-5

Chapter 3: Networking3-1

"An Easy Path to Packet: the IMP" by Tony Marchese, N2YMWfrom December 1995 QST

"Radio-TNC Wiring Diagrams" from What is your TNC Doing? by Gloria E. Medcalf, KA5ZTX

"Introducing the Ottawa PI2 (Packet Interface 2) Card" by the Ottawa Amateur Radio Club , Inc.from the Ottawa ARC Web Site

"Modifications for the Alinco DR-1200" by David Stinson , AB5S and Bill Leahy, K0ZLfrom the Internet

"The AX384 & AX576 GMSK Radio Modems" by Matthew Phillips, G6WPJ and John Ferguson , G8STWfrom the GMSK Web Site

"W3IWI/TAPR TAC-2 (Totally Accurate Clock) Project" by Tom Clark, W31WIfrom the TAPR Web Site

"Emulating the 'Totally Accurate Clock" by Tom Clark, W31WIfrom the Goddard Space Flight Center FTP Site

4-31

4-23

4-27

4-3

4-18

4-20

Chapter 4: Construction4-1

AcknowledgmentsThis book is the product of many individuals. Our appreciation goes out to all who made contributions by supplying

information and granting permission to use their material. A special thank you goes to Gloria Medcalf, KA5ZTX, authorof What is your TNC Doing?, for sharing her "Radio-TNC Wiring Diagrams" with us.

v

Page 7: Packet- Speed- More Speed

About the

American Radio Relay League

The seed for Amateur Radio was plant ed in the 1890s, whenGuglielmo Marconi began his experiment s in wireless telegra­phy . Soon he was joined by dozen s, then hundreds , of otherswho were enthusiastic about sending and receiving messagesthrough the air-some with a commercial interest, but otherssolely out of a love for this new communications medium. TheUnited States government began licensing Amateur Radio op­erators in 1912 .

By 1914, there were thousand s of Amateur Radio opera­tor s-hams-in the United State s. Hiram Percy Maxim, a lead­ing Hartford, Connecti cut , invento r and indu striali st saw theneed for an organization to band togeth er this fledgling gro upof radio experimenters. In Ma y 1914 he founded the Ameri canRadi o Relay League (ARRL) to meet that need .

Toda y ARRL, with more than 170,000 mem ber s, is thelar gest organiz ation of radio amat eurs in the United States. Th eLeague is a not-for-profit organi zation that :

• prom ote s interest in Am ateur Radio communicati onsand experimentation

• represents US radio am ateurs in legislati ve matters, and• maintains fraternalism and a high standard of conduct

among Amateur Radio operators.At League headquarters in the Hartford suburb of

Newington, the staff helps serve the need s of members. ARRLis also International Secretariat for the International AmateurRadio Union , which is made up of similar societies in more than100 countries around the world.

ARRL publishes the monthly journal QST, as well as news­letters and many publications covering all aspects of AmateurRadio . Its headquarters station, W IAW, transmits bulletins ofinterest to radio am ateurs and Morse code pract ice sess ions .The League also coordinates an ex tensive field org aniza tion,whi ch includes voluntee rs who provide technical inform ationfor radi o am ateurs and publ ic-service activi ties . ARRL also

vi

represents US am ateurs with the Federal Co mmunica tio ns Co m­mission and other go vernment age nc ies in the US and abroad .

Me mbership in ARR L means much more than receiving QSTeach month . In addition to the services already described, ARR Loffer s membership servic es on a person al level, such as theARRL Volunteer Exa miner Coo rdinator Program and a QSLbur eau.

Full ARRL membership (available only to licensed radioam ateurs) gives you a voice in how the affairs of the organizationare governe d. League poli cy is set by a Boa rd of Direct ors (onefro m eac h of 15 Divisions) . Eac h year, half of the ARRL Boardof Direc tors stands for elec tio n by the full memb ers they repre­sent. Th e day-to-d ay operation of AR RL HQ is managed by anExecutive Vice Pre sident and a Ch ief Financial Officer.

No matter what aspect of Amateur Radi o attracts yo u, ARRLmemb ership is relevant and important. The re wo uld be no Ama­teur Radi o as we know it today were it not for the ARRL. Wewo uld be happ y to welcome you as a member ! (An Amateu rRadi o license is not required for Associate Me mbership .) Formor e inform ation about ARRL and answers to any questions yo umay have about Amateur Radi o, write or call :

ARRL Educati onal Ac tivities De pt225 Main StreetNewington CT 06111-1494(860) 594-0200

Prospe ct ive new ama teurs ca ll:800-32-NEW HAM (800 -326-3942)

E-mai l: [email protected]: http ://www.arrl.org/

Page 8: Packet- Speed- More Speed
Page 9: Packet- Speed- More Speed

Chapter 1Operating

Page 10: Packet- Speed- More Speed

By Steve Ford, WB81MYAssistant Managing EditorInternet: sford @arrl.org

the Internet

11msitting at the key board of afr iend's packet sta tio n. He'sa re lat ively new packetee r,and I' ve been tryin g to co n­

vert hi m to the wonders of TCP/IP. Tha t'sthe Am ateur Radio version of the Int ern et ,the glob al information super highway.

It' s been tou gh sledding, thou gh . Firstwe had to set up the softw are in his co m­put er. The n I had to acquaint him with anentire ly new way of navigatin g his localnet wor k.

"TCP/IP is all well and good, but wha tca n I do wit h it?" he asked . "What happensafter yo u go hom e to Connect icut and I' mleft with this?"

"Well, we ca n cha t and sen d electron icmail ," I repli ed.

"Yeah, right. We tried that with the regu­lar packet BBSs, re me mbe r? It took a weekfor my messages to reach you-whe n theyreached yo u at all. Beside s, yo u sa id myTCP/IP net work was loc al. How co uld Ipossibl y connect to you?"

Thi s is exac tly the moment I was wai tingfor. It was time to unveil my sec ret weapon.I tapped furious ly on the keys and soo nestabli shed a telnet sess ion to a stat io n onhis network.

"What are you-""Sit tight. You 'll see. "He peered closely at the scree n as I

entered :

tel net 44 .88.4 .35

I just in stru cted the loc al sta tion to find

a ro ute to 44 .88.4.35-my home TC P/ IPaddress .

My friend shook his head. "You' ve askedthat stat ion to do the impossible. Our net­wor ks don 't connect. "

"A re you sure ?" I said wit h a smile."What if thi s station has a way to pass dat ato and from the Int ern et ? What wo uld hap ­pen then? "

His eyes wide ned. "A ga teway!"" Hrnmrn . . . co uld be !"A thousand miles away, my packet sta­

tion is faithfull y monitoring 145.53 MHz.Another station belonging to Bill Lyman,NI NW P, is on the same freq uency . N INWPis also linked directly (by wire) to theIntern et.

Sure eno ug h, the lin k request fina llyarr ives at N INWP via the Int ern et. Hi sco mputer " looks" at my IP add ress andknows exactly what to do. A co nnec t re­quest blasts out on 145.53 MHz. The virt ualcircuit is esta bli shed and, hal fway ac rossthe co ntinent, I'm rewarded with the mes­sage show n in Figure I

" Incredible!" my friend gas ps. "So thi sis ho w the fab led gateways wo rk. I thoughtthey were just ru mors. Thi s is unbeliev­able!"

The Elusive Goa l

As amateur pack et rad io has spreadthroughout the world, one goa l has rem ainedelu si ve: find ing a fas t, re liable way to tran s­fer information between di stant points. Ato ne time we thou ght that we co uld build

high -speed RF packet net works encirclingthe Earth, but thi s is not likel y to happ enanyt ime soo n. Th e cost is out rageou s andthe required maintenance would be too muchfor mos t indi vidu als or groups to bear.

We 've been us ing HF packet to helpbridge the gaps, but its performance lea vesmuch to be desired . Other HF di gital modesha ve improved the situat ion, but the y're allat the mercy of the changeable nature of HFprop agat ion . None of them offer the spee dthat even a 1200-bit/s VHF/UHF packetlin k ca n ach ieve.

Digita l Amate ur Rad io sate llites havebeen pressed into service to transfe r packetmail. T he sate llites are reasonably depend ­able, but they are not geo sta tionary (s ta­tion ed at fixe d point s in the sky fro m ourper specti ve on the gro und). Instead , theyorbi t the Eart h at rel atively low altitudes.Thi s means that packet stat ions have on ly afew oppor tunities each day to get dat a toand from the sa tellites .

It wasn' t long before man y ham s beganto con sid er nonamateur means to reach thesame end. Th at ' s whe n the y began taking aser ious loo k at the Int ernet.

Opening the Gateways

Internet conn ections are as clo se as thenea rest te lephone line s. With the prop erhardware and so ftware, it isn't difficult tointerface amateur pac ket radio stations tothe Internet. Many of the se become the so ­ca lled "gateways."

So me gateway stat ions are set up at

Figure 1-By using an Internet gateway on my friend 's TCP/IP network, I've connected to my home packet station­f rom more than 1,000 miles away!

Figure 2- This is a brief glimpse of the international asoconference bridge. It's especially hot and heavy during theevenings and on weekends.

Operating 1-1

Page 11: Packet- Speed- More Speed

Finding Your Nearest GatewayWhen this article went to press, there were more than 30 Amateur Radio/ Internet

gateway systems in the United States. The trick is finding them. The informationdoes exist , although most amateurs are unaware of it. That's about to change.

The list of gateways-including the services they offer, packet frequencies, callsigns and so on-can be found in a text file called resource. If you have full Internetaccess , you can transfer (ftp) the file into your computer and read it. Connect to:

ftp.radnet.com

Look in the gatewa ys directory .We've also placed this file on the ARRL BBS (203-666-0578) . It's called

RESOURCE.TXT. You can obtain the file from the ARRL InfoServer, too. You cantap the InfoServer by using e-mail from your Internet account, or from any of the on­line services such as America On-Line, CompuServe, and so on. See "Exploring theInternet-Part 2" in the October 199 4 OSTlor instructions on using the ARRLInfoServer.

Once you have the file, read through it and find the gateway closest to you. Ifyou're lucky, you'll find a gateway in the city where you live. If not, you'll need tolocate a gateway that you can reach via your local or regional packet network (eitheron TCP/IP or standard packet). Be careful about trying to reach distant gatewaysover packet networks. Even on TCP/IP, the throughput gets worse the farther youget from home. You might be able to establish a link , but it will be too slow to beusable.-WBBIMY

colleges, uni versities or businesses . Be ­cause the Int ernet connections already ex­ist, the peopl e in charge allow the gatewayopera tors to use th em . Oth er gateway sta­tions are operated by ham s who have Internetaccess at their homes for person al or busi­ness use . (T his is the case wi th the N INWPgateway .)

Once the se gateway stations tran sferpacket radio da ta to the Inte rnet, the infor­mati on moves at high speeds to almostanyw he re in the world . Data also po urs outof the Int ernet throu gh the ga te ways andends up on am ateur packet networks .

Although it isn't a common-carrier ser­vic e like the teleph on e compan ies, the reli­ability and speed of the Int ernet are impres­sive. You ca n think of the Int ern et as amultilane superhighwa y. The amateur packetnetwork is liken ed to the slower surfacestree ts . And what abo ut the gateway sta­tions? Th ey're the entra nce and ex it ram ps.

Internet ga teway stations ha ve been pop­pin g up all over the place in recent years .Hams are usin g them to act as lin ks betweenpacket net works. For example , a pac ketbull etin board sys te m (P BBS) in New En ­gland may pass messages to the West Coastvia an Internet gateway, rather th an re ly ingon the HF mod es or sate llites . Thi s is oftenknown as a wormhole link. Messages thattook days to reach th ei r fin al destinationsarrive within hours or eve n minutes. Otherhams tap th e gate ways and ex plo re on theirow n, usin g th e Int ern et as a packet versionof the proverbia l magic carpet.

TCP/IP

If yo u 've been arou nd packet radio lon geno ugh, there 's a fa ir cha nce you've heardabout TCPlIP. TC P/IP is an Amate ur Radi oadapta tion of the protocol s used on theInt ernet. Amateur TC P/IP networks (re ­fe rred to co llectively as the AMPRNet ) havebeen growing throug ho ut the nat ion . Whe n

you consider the similarity bet ween ama ­te ur TCP/IP and the Internet, it's easy tounderstand why most gateway sta tions areon TCP/I P rather th an "regular" (AX .25)packet.

Yo u' ll find TCP/ IP ac tivity pri ma ri ly on2 meters and 70 ce ntimeters-usua lly onthe sa me freq uenc ies where yo u ' d ex pec t tohear regul ar packet. If you try to co py aTCP/ IP sig nal wi th AX.25 protocol, how­ever, you 're likel y to see gibberish on yo urmon itor.

If yo u' re already involve d in packet ,co nvert ing to TCP/ IP is straightforward .All you need is the NOS software (in one ofits man y versions). Your te rmi na l nod e co n­tro ller (TNC) mus t be ca pable of op eratingin the KISS mode, but all TNCs manufac­tured in the past eight years or so incl udeth is feature.

Setti ng up NO S is difficult fo r some.Yo u have to learn Unix-type jargon, forexample , and change a number of fi les toadapt the software to the net wor k as itex is ts in your area. To make matters worse ,most ve rs io ns of NOS are not especial lyfr iendly .

Yo u ca n pu rchase books that wi ll helpyo u throug h the pr ocess. I re commendNOS Intro, by Ian Wade, G3NRW. It ' s avail­able fro m the ARR L (see the ARR L Publi­cat ions Cata log in this iss ue).

You can also ge t special NOS so ftwarepackages that are des igned to "c on figure"themselves automatically . T he progra msask you to enter several items of informa­tion, an d th en proceed to set them sel ves upin your compu ter. One such pa ck age,MFNOS wi th "Autogen " , is availab le fro mthe Connect icu t Digi ta l Rad io Associatio n.The software is written for IBM PC s andco mpa tibles.

If you have fu ll Internet access , yo u ca nftp (transfer) MFNOS with "Autogen" toyo ur computer b y co n nec t i ng to

ftp .radnet.com. Log in as "anonymous" andenter your Inte rnet address whe n you'reasked for a password .

You' ll find the followi ng files in themfnos directory :

mfnos!ct.zip (for Connecticut stations only)mfnos!44.zip (for all other stations)!readme.1st (insta llation direct ions)

If yo u have qu esti ons about MFNOS,se nd e-ma il to: n l nwp@ a3bee2.radne t.com.

If you don't have Internet access, yo uca n se nd a blank hig h-de nsi ty di skette a longwith a se lf-addressed, stamped di sk enve ­lop e to:

Bill Ly ma n Jr , N I NWP219 S Orchard StWallingford, CT 06492

I've a ls o p laced mf nos!4 4. zfp andmfnoslct.zip on the ARRL BBS . If you havea modem, call 203-666-0578 and down loa dit.

Once yo u have the softwa re running inyo ur computer, all you need is yo ur own IPaddress. These are ass igned by IP addresscoordinators th rou ghout the country. Yo u' llfi nd a list of coordina tors in ARRL bookssuch as the Ope rat ing Manual and YourPac ke t Companion. The list can also befo und on ha m-radio-oriented computer bul ­letin boards (incl uding the ARRL BB S) andseveral on-line services.

If you don 't want to jump into TCP/IPjust ye t, take heart. So me gateways offeraccess via AX .25 packet. You can learnabout the services eac h ga teway offers byreadi ng thro ugh the gate wa y "reso urce " file.See the sideba r, " Finding Your NearestGateway."

Using the Gateways

The ty pes of ac tiv ity you can enjoythro ugh a ga tew ay dep end on the ga tewayoperator. Some ga teways allow access onlyby speci f ic ama te ur s ta t ions (suc h asPBBSs ). Others limit indi viduals to send­ing and receivi ng mail , or joining rea l-ti me"conferences." Here ' s a qui ck rundown ofwha t yo u might en count er:

o Electronic Mail: Some ga tew ays al­low yo u to pass packet mail usin g theInternet as a high- speed re lay . You ca nsend TCP/IP e-mail (k now n as SM TP) fromone network to ano ther. Yo u may even beable to se nd AX .25 pack et mail through thegateways . Ask the operato r to tell yo u howto create addresses for e-ma il trav eling viahi s or her gateway. Th e procedures maydiffer dep ending on whether you're us ingTCP/ IP or AX .25 pac ke t.

1-2 Chapter 1

Page 12: Packet- Speed- More Speed

Gateway operators sometimes shy awa yfrom offering e-mail capability because ofpos sible legal complications (violations ofthird-party traffic rules and so on) . Otherssolve this problem by holding and review­ing all e-mail messages that arrive from theInternet before passing them onto the packetnet work.

o FTP: Thi s function is available toTCP/IP packete er s, but few gateways offerit. FTP stands forfil e tran sfer protocol. It' sa means to move files efficiently from sta­tion to station. TCP/IP packeteers use it tosend computer programs. images . you nameit!

One problem with FTP ac tivity via thegate ways is the slow data rates on the radiosides of the link s. Even at 9600 bit s/s, tran s­ferring a large fi le could take a lon g timeand seriously bog down the gateways.

o Telnet: This is another TCP/IP-onlyfunction . By using telnet you can access anamateur TCP/IP station rem otely throughan Internet gat eway . Th at' s what I wasdoing in the example show n at the begin­ning of thi s artic le.

If you're skeptical, try to reach my homestation . Establ ish a te lnetlink to your near-

est gateway station (if it offe rs tel net capa­bility) . then ask it to telnet to 44 .88 .4 .35 . A" host unknown" message means that thesystem was unable to find a route to me . Ifthis happens. teln etto N1NWP-1.AMPR.ORG,then telnet to WB8IMY. "Failure with .. ."usually means that my station was not onthe air. Try again later.

o QSO Bridge :Thi s is a live , keyboard­to-keyboard roundtable that's on 24 hours aday. If your gateway of fer s access to thebridge. you can enjoy conversations withhams throughout the world. Most of themare sitting at the keyb oard s of packet sta­tion s just like yours . Thanks to the Internetand the gateway stations, ever yone can gettogether and enjoy th e conversati onalequi valent of a food fight! Figure 2 showswhat the QSO bridge look s like as seenthrough the N INWP gateway.

The Future of GatewaysYou can expect to see more gateways

appearing on the air in the near future . Thecost of priv ate Internet access is falling, sogateway stations will bec ome mor e afford­able . Gateways are especially attract ive to

Technicians like N INWP who have an in­terest in computers and want to expandtheir communication hori zon s.

As the gateways proliferate. more net­works will link in thi s fashion. Th e benefitto hams will be much faster movement oftraffic bet ween regi onal networks-andeve n within networks where Internet worm­hole s can fill troubl esome gap s.

Some hams may grouse at the idea ofusing a nonamateur network as part of anAmateur Radio communication system.Until a radio-based alternative is found ,however , the Internet stands as a valuableresource for expanding Amateur Radiopack et activity. Rem ote packet stations atdisaster sites, for example, could link to anInternet gateway and pass traffic to distan tpoints with extraordinary efficiency.

If you can reach an Internet gateway.explore it and see what it has to offer. Don'tforget to send a "thank you" note to thegateway operator when you get a chance.His or her effort-and often money-makesyour enjoyment possible!

Many thank s to Bill Lyman , N INWP,jorhis contribution to thi s article.

Operating 1-3

Page 13: Packet- Speed- More Speed

Packet PerspectiveConducted By Stan Horzepa , WA1LOU

One Glen Ave , Wolcott , CT 06716·1442Internet: ho rzepa @gdc.com CompuServe: 70645 ,247

Messages via PacketlInternet GatewaysThere are misconce ptions and misi nfor ­

mation floating aro und packet radio landabo ut how to use the ga teways that connec tamateur packet radi o and the Intern et.Because no two gateways are exactly alike,it' s not possible to prov ide spec ific instru c­tion s that explain how to use them all. On theother hand , instructions that are too ge nera lwi ll be so lack ing in information that theread er will get litt le or nothing out of them .

The third alte rnative (the one I' ve cho­sen to use here) is to provide speci fic in­structions on how to use a part icular pac ket­to-I nternet gateway. After reading them,you sho uld understand how to use one gat e­wa y and be able to appl y that knowled ge toother gateways you may want to use . Al­though other ga tew ays you encounter maynot work exactly the same as the one de­scribed here , they'll be similar eno ug h sothat yo u wo n ' t have to be completely re­educated whe n you use a different gateway.

Jim Durham , W2XO, Gi bson ia, Penn­sy lvania, runs a popular packet-to-Intern etgateway and has ge nerous ly agree d to al­low me to use his gateway as the examplefor properl y using pac ket -to-Intern et gate ­ways. Don't wo rry; Jim wrote these inst ruc ­tions, so yo u can be assured that they wo rk(and wer en 't man gled by me). Her e goes :

Using the W2XO Gateway

The gateway is configured to make mail ­ing from the ham packet radio net work toand from the Intern et as straightfo rward aspossible and in accordance with standa rdmai l gateway practices. You use the subjectline s and message text as usual (you don 'tneed any special lines in the text or subjectof the message) . However, Internet usershave to regi ster with me so that I can assignan ali as, as ex plained below.

Mailing from theInternet to Packet Radio

On the Int ernet, the co nve ntio na lmeth od of mailing to so meo ne on anothe rnet work is to repl ace the "at" sym bo l ( @)in the no n-Inte rnet address with a pe rcent­age sig n (%) and fo llow the mod ified no n­Int ern et address with an @ and the ga tewayhost ' s ad dress. Th e non-Intern et address

1-4 Chapter 1

johndoe@wha tnet wo uld be entered onan Internet mailer as johndoe%whatn et @hostname.domain whe re hostname.domainis the gatew ay host' s address.

To se nd a message to packet radio fro mthe Interne t, all yo u have to do is repl acethe ham packet radi o address' @ with apercentage sign, foll ow that with an @ andthen the gateway' s Intern et address (in thi sca se, w2 xo.pgh .pa .us). To mail a messagefro m the Internet to the pack et radioaddress of WAIYUA @WI EDH. CT.USA.NOA M , you 'd rep lac e its @ with %, fol­low it with @, and then the gatewa y 'sInternet add res s. WA I YUA % W IEDH.CT.USA.N OAM@ w2xo.pg h.pa .lIs is theres ult.

(My packet BBS is W2XO.#S WPA .PA .USA. NO AM and my Internet host nam e isw2xo.pgh.pa.us . These are , as the mansays, "similar but differ en t." Don 't co nf usethem ! One is a packet BBS hierarchica laddress ; the other is an Int ern et address .Because I' m part of the " us" do ma in onInternet and my co mpute r' s host name isw2 xo , they look simi lar.)

That' s the Internet to packet side ; nowfor the other side.

Mailing from Packet Radio to theInternet

The amateur pack et radio message fo r­mat allows only six cha rac te rs in therecipient ' s address. I use an "alias" to getaround thi s limitation th at prevent s youfrom using an Internet addre ss likeLwi ll ia [email protected] dir ectly. If theInternet recipient is a ham , I use therecipient 's call sign. If the rec ipient isn 't aham , I create a six-c ha rac te r alias in theform 3PTYXX ("third-p art y" ).

To send a message to an Intern et rec ipi­ent fro m the amate ur packet radi o net work .yo u' d address the message to call sig nor alias @W2XO.#SWPA .PA.USA .NOAMand the message will auto ma tica lly be fo r­wa rded to the Internet ad dress co rres po nd­ing to that call sig n or alias .

For examp le, suppose yo u want to se nda message fro m amate ur packet radio tojoe@somecompu ter. comon the Intern et. IfJoe is a ham whose ca ll s ign is WA IYUA.yo u' d sen d yo ur message to Joe using the

address WA IYUA@W2XO .#SWPA .PA .USA .NOAM and the ga teway will look upWA IYUA in its files and mail the messageto the Intern et address joe @som ecomputer.com . If Joe isn ' t a ham , I ass ign him an aliaslike 3PT YXX . and yo u send your messa geto Joe usin g the address 3PTYXX@ W2XO.#SWPA. PA .USA.NOAM, and the ga tew aywill look up 3PTYXX in its fil es and mailthe message to the Internet addressjoe @som ecomputer .com .Many amateurs be lieve that a messagethat or iginates on packet rad io (or tra vel san y part of the way o n packet) in the US ,and is then re layed overseas via theInte rne t, fal ls into the category of interna­tional third-party traffic, mea ning it canon ly be handled with cou ntr ies where third­party agreements are in effect-but this isnot true . Accord ing to the Regu latory in­format ion Branch at ARRL HQ, as long asthe ra dio traffic is internal to the US orexchanged with a co untry with which wehave a third-pa rty agreement, it is com­pletely legal. That it came from or is des­tined for a country with wh ich we do nothave an agreement is imma te ria l, as longas it travels to or from that country viameans other than Amateur Radio (eg , theinternet, postal mail , telephone , etc).

Third-party traffic is def ined by the FCCas communication between the control op ­erator of one station and the control op ofanother station on beha lfof a th ird party , sothe rules deal with the locations of the con ­trol operators involved, not of the th ird party .Regard ing inte rnationa l rad io commun ica ­tion, the reason for the rule is to preventlos s of reven ue to countries where te lecom­mu nicat ions is a state monopoly. So if themessage travels via the Intern et in thosecountries , their te lecom munication systemshave not been bypassed .-WAtLOU

Seeking Super Packet Tricks

Last month ' s co lumn wa s devoted tosupe r pack et tric ks. There wa s a lot of in­te rest in the trick s I di scu ssed and I'd liketo bri ng yo u mo re in the future . so se nd methe supe r pack et tri cks yo u use to get mor eou t of packet radi o (o r j ust to ge t so me thi ngo ut of packet rad io) . I'll pre sent them he reand you ' ll beco me ric h and fa mo us. We ll.maybe not ric h. but how abo ut becomingfamou s by enric hing others '?

Page 14: Packet- Speed- More Speed

KISSes. POPs and PingsIf you're looking for a new challenge, try unraveling the mysteries of TCP/IP.By Steve Ford, WB81MY

As sistant Managing EditorInternet: sford@arr l.orgAMPRNET: 44.88.4.35

Figure 1-A typical ftp (file transfer protocol)session. The characters that actually appear on my

computer mo nitor are shown in bold type.

SYN sentEstablished220 wb8isz .ampr.org FTP version 890421.1e ready atSat Aug 1318:22:2619

3 files 13,617,152 bytes free. Disk size 33,400,832bytesGet comp lete, 200 bytes received

It looks complicated at first glance, but all it'stelling you is that WB8ISZ has three files available fortransfer. The name of each file is shown along with itssize and the date it was placed on his disk. If you havesome experience with computers, this may lookfamiliar. (You've just issued the "directory" commandthat's common to most machines.)

I happen to know that "space.exe" is a game thatWB81SZ has written for my particular computer.Regardless of whether it' s an ASCII or binary file, I cantransfer a copy by simply using the get command.

get space.exe

Now I'm ready to log in. At the command prompt Isend: user anonymous. My terminal displaysWB8ISZ's response.

The Attracti on of TCP/IPWhen a gro up of pac ket stations

decide to adopt TCP/IP as a way ofcommunica ting among themsel ves,the resulting network looks a lot likethe Internet. We ca ll it theAMPRNET-Amateur Packet RadioNet work. It operates in much the samefashion as the Internet and , as I've al­rea dy mentioned, it uses the sam e pro­tocols. The primary difference is thatamateur networks are usuall y muchslower and more lim ited in sco pe.

Some TCP/IP networks are verysmall- j ust a loose gro up of dedicateden thusiasts who enjoy swapping infor­mation . At the oppos ite end of thescale, there are vast net wor ks that of­fer coverage to packetee rs in seve ralstates . They achieve th is cov eragethro ugh the use of special TCP/ IPnodes (of ten called switches) that actas relays .

Several of these large networks in­cl ude gateways to transfer informationto and from the Internet. The samegateways can also funct ion as worm­holes- Internet pipelines that link di s­tant sections of large networks.

But what makes TCP/IP so special?Here are just a few of the high points . ..

Mail- TCP/IP mail is sent from sta­tion to station through the network.There are no BBSs invo lved. You needonly prepare a message and leave it inyour ow n TCP/I? "mail box ." Withinseconds your computer will attempt tomake a connection to the target stationand deliver the mes sage. If the targetstation isn 't on the air, your com puterholds the message and keeps trying.When you check your computer and seethat the message is no longer in themailbox, you ca n be sure it has arrive dsafe ly at the other station.

NOS. That's why you 'll find so many avail­able. The re ' s GRINOS, JNOS, TNOS,MFNOS and so on. All of these NOS vers ionsuse the same TCP/IP protocols . Eve n theoriginal Unix j argon remains.

But why go to so much trouble? After youget the software running and the hardwareperking along , what do you have?

Simply this: an Amateur Radio vers ionof the Internet.

19:5702/1919502:3004/0119522:25 04/16195

sys tem he cre ated wa s or igi nally called NETIn time he rewrote the software and it de­buted to the ham pop ulace as NOSNET, orj ust NOS for short (Net Operating System) .Not long thereafter, NOS was adapted to runon Macintoshes and other computers.

Of course, hams can 't resist the urge totinker. Ta lented programmers followedPhil ' s lead and created their own versions of

When the link is established, my terminal displays:

No problem. Most systems allow you to use yourcall sign as the password.

pass wb8imy

When I see 230 Logged in, it's time to check hiscomputer to see what he has to offer. AlII have to do issend dir and my screen displays:

200 Port command oka y 150 Opening data connec­tion for LlSTIpublic

switch.map 1,500tcp/ip.doc 10,000space.exe 20,000

331 Enter PASS command

ftp wb8isz

The Internet and AmateurT CP/IP

If yo u've been fort unate enough tosurf the Internet , you 've alreadydab bled in TCP/IP . (Although ifyou're using a she ll pro gr am tha tmakes the Internet more user friendly,many of the nut s and bolt s of TCP/IPare transpare nt to yo u.) As compli­cated as it may sound, TCP/IP is sim­ply a group of packet protoco ls thatmake it possible fo r the Internet toshuffl e informat ion throughout theworld .

But what' s a protocol? A protocolis a standardized way of doing some­thin g tha t ma ny par ties -people orcomputers-agree to recognize . Forexample , yo ur ham club may useRob ert 's Rules of Order to ma intainsanity duri ng meetings. (Maki ng "mo­tion s" to vote and so on .) We ll, tho serule s are a type of protocol. When youhear Internet users say they ' re goingtoftp a fi le, you ' re hearing a TCP/IPterm. FTP stands for File Tran sferProtocol. Yo u' ll also hear referencesto SMTP (Si mp le Ma il Transfer Pro ­tocol ) Telnet and others. All these pro ­toco ls reside under the grand umbrellawe call TCP/IP.

TCP/IP was pretty much confinedto the Internet and Uni x-type com ­puter sys tems until Phil Karn, KA9Q,dec ided to adapt it for use on ham net­works and IBM PCs. The operating

11re yo u getting weary of the same

~ old packet? Connect to the localbu lletin board (BBS), grab yourwait ing mail , read a few bulletins,

and log off. Now tha t's hamming (yawn)!Hey , I have nothing aga inst this type of

packeteering. I check my local BBS almos tevery day and read the bulletins, too . But af­ter a while (about a decade, in my case) youget a thirst to do ' something more .

If you want to expand your packethorizons, try Transm ission ControlProtocol/ Internet Protocol-othe r­wise known as TCP/IP . Assuming thatyou already own a packet station (FMtransceiver, computer and terminalnode controller [TNCj), the additionalcost to run TCP/IP is .. .nothing. Nada .Zip . Nil.

Ope rat ing 1-5

Page 15: Packet- Speed- More Speed

File Transfers-r-You can use the file­tra nsfer protocol (FTP) to easily transfer fi les(ga mes , images, or whatever) to any stationon the network (see Fig ure I) .

Data Handling-c-Resue: than spewingdata at random intervals like sta nda rd pac ket,TCP/IP stations take the smart approach andautomatically adapt to net work delays. As thenetwork slows down, TCP/IP sta tions sensethe change and len gthen their tran smissiondelays accord ingly. (They do n' t transmit asoften.) As the network speeds up, the TCP/IPstations sho rten their delays to match the pace(they transmit more freq uently) . This kind ofinte lligent network sharing virtually guaran­tees that all packets wi ll reac h their des tina­tio ns with the greatest efficie ncy the networkcan provide.

Di re ct Addressing-i-Evetv TCP/IP sta­tion in your netw ork has an address. These IPaddresses are ass igned by volunteer coordi­nators throu ghout the country (see thesidebar, "Y our Own IP Address").

My address , for exa mple, is 44. 88.4 .35.Rea ding from left to right , 44 tell s you thatthis is the add ress of an Amateur Radio TCP/IP station (ra ther than a non ham Internet site) .

The 88 designates my part of the NewEngland reg iona l network (Connecticut). The4 get s even more spec ific , pointin g to my littl ecorner of the state (known as a subn et) ,Finally, the 35 is my unique address in thesubnet.

But you don't need to know the addresses ofyourfe llow TCPIIP users! All of thi s info rma­tion is contained within your NOS software ina fi le called DOMAIN. TXT. When I want tosen d a message to, say , WS 10 , I si mply ad­dre ss the message to WS I O. When it 's timeto send the message, NOS wi ll search thro ughDOMAIN.TXT and locate WS 10 . When itdoes, it'll use the IP add ress it finds the re.

By analyzing the IP address , the switchesand other stations probe throu gh the networkto crea te a link to WS I O. Once the "ci rcuit"is established, the da ta flows and the messageis de livere d. All of thi s takes place wit houtyour lifting a finger!

Man y ver sion s of NOS cont ainDOMA IN.TXT. It may be a sma ll fi le thatincludes onl y the use rs in a particul ar area .On the other hand, it co uld be the gra nddaddyof all DOMAIN.TXTs tha t lists eve ry ama­teur TCP/IP user in the world! If your NOS

does n ' t have a DOMAIN file, yo u' ll need tofind a copy . See the sidebar, "Gett ing Startedin Four Steps ."

Multitask ing-s-With TCP/IP you can doseveral things sim ultaneously . For example,you can se nd a message , recei ve a messageand tran sfer a file-all at the same time!

Flexibility-NOS provides an excelle ntplatform for develop ing new pro tocols tha twill dramatically expand the capabilities ofthe AMPRN ET . Some versions already sup­port Go pher and an amate ur vers ion of theWo rld Wide Web is on the way .

Ping! Is Anyone Home?

Is yo ur buddy on the air this af ternoon?Should you bother try ing to get a message

Your Own IPAddress

Mark Bitterlich , WA3JPYCharles Layno, WB4WORSteven Elwood, N7GXPMike Nickolaus, NF0NGary Grebus, K8LTDave Trulli, NN2ZBob Applegate, WA2ZZXJ. Gary Bender, WS5NBob Bellini , N21GUDave Brown, N2RJTBob Foxworth , K2EUHEarl Petersen, KF7T1Bill Healy, N8KHNJohn Ackerman, AG9VJoe Buswell, K5JBRon Henderson, WA7TASTom Kloos, WS7SDoug Crompton, WA3DSPBob Hoffman, N3CVLKarl Wagner, KP4QGCharles Greene, W1CGMike Abbott, N4QXVSteven Elwood, N7GXPJeff Austin, K9JAJack Snodgrass , KF5MGKurt Freiberger , WB5BBWRod Huckabay, KA5EJXMatt Simmons, KG7MHJim DeArras, WA40 NGJon Gefaell, KD4CQYBernie McDonne ll, NP2WRalph Stetson, KD1RSteven King, KD7ROBob Donnell , KD7NMThomas Landmann, N9UDLRich Clemens, KB8AOBReid Fletcher, WB7CJO

NC (eastern)NC (western)NDNENHNJ (northern)NJ (southern)NMNY (eastern)NY (western)New York City and Long IslandNV (southern)NV (northern)OHOKOROR (northwest and Vancouver, WA)PA (eastern)PA (western)PRRISCSDTNTX (northern)TX (southern)TX (western)UTVAVA (Charlott esville)VIVTWA (eastern)WA (western)WIWVWY

Terry Neal, AA6TNBob Meyer, K6RTVDon Jacob, WB5EKUGeoffrey Joy, KE6QHBrian Kantor, WB6CYTDouglas Thom, N60YUFred Schneider, K0YUMBdale Garbee, N3EUABob Ludtke, K9MWMBill Lyman, N1NWPRichard Cramer, N4YDPButch Rollins, NF3FBrian Lantz, K0 4KSDoug Reed, N3AIAJohn Shalamskas , KJ9USteven King, KD7ROChuck Henderson , WB9UUSKen Stritzel, WA9AEKDale Puckett, K0HYDAllan Dayton , N0KFOJames Dugal, N5KNXGo~on LaP~~, N 1 MGO

Bob Wilson, KA1XNHoward Leadmon, WB3FFVCarl Ingerson, N1DXMThomas Landmann, N9UDLJeff King, WB8WKAAndy Warner, N0REN

Stan Wilson, AK0BJohn Martin , KB5GGOSteven Elwood, N7GXP

Before you can get on the air with TCP/IP, you need your own AMPRNET IP address. These addresses are assigned by thevolunteer coordinators listed below. This list is subject to change without notice. If you have Internet ftp capability , you 'll findupdated AMPRNETcoordinator lists at ftp.ucsd.edu.

AK John Stannard, KL7JLAL Richard Elling, KB4HBAR Richard Duncan, WD5BAZ David Dodell, WB7TPYCA, Antelope Valley/Kern County Dana Myers, KK6JQCA, Los Angeles- Jeff Angus, WA6FWI

San Fernando ValleyCA, Orange CountyCA, SacramentoCA, Santa BarbaralVenturaCA, San Bernardino and RiversideCA, San DiegoCA, Silicon Valley-San FranciscoCO (north)CO (south)CO (west)CTDCDEFLGAHI and Pacific islandsID, eastern WAIL (central and southern)IL (northern)KSKYLAMA (center and eastern)MA (western)MDMEMI (upper peninsula)MI (lower peninsula)MNMOMSMT

1-6 Chapter 1

Page 16: Packet- Speed- More Speed

Getting Started in Four Steps

1. If you don't own a packet station now, you'll need to build one. All thaI's requiredis a comput er, a 2-meter FM transceiver and a terminal node controller, or TNC. TheTNC must be capabl e of operating in the KISS mode (Keep It Simple, Stupid-nokidding!) . All TNCs made within the last eight years or more include KISS. Beware thatsome TNC emulation systems such as DIGICOM, Poor Man 's Packet and BA YCOMmay not be compatib le with TCP/IP software.

2. Get a version of NOS. If you can determine the version thaI's most popular onyour network, it's a good idea to use the same. II's much easier for the locals to helpyou if you're using similar software (see Step 4).

You' ll find NOS software on many on-line services such as CompuServe (in theHamNETforum) and on a number of ham-oriented computer bulletin boards. (Yes, theARRL BBS has several NOS versions availabl e. Call 203-666-0578 and download oneof them!) If you have Internet capability, you can ftp many vers ions of NOS fromoak. oak land.edu or ft p.ucsd.edu. NOS is also offered for a nominal price by TAPR(Tucson Amateur Packet Radio). For a complete list, send a self-add ressed stampedenvelope to: TAPR, 8987-309 Tanque Verde Rd, No. 337, Tucson , AZ 85749-9399.

3. Get an IP address. Contact one of the coordinato rs shown. Some coordinatorsmay prefer to issue a temporary "test" address at first. If you find that you like TCP/IP,they may ask that you send a message to them via the network and request a perma­nent address. They're not making you jump through hoops for their enjoyment. II' s justthat they don' t want to assign permanent addresses to hams who are not going to beactive on the network for the long term .

4. Install your NOS and get on the air. I strongly recommend that you get outsidehelp from an experienced TCP/IP operator for this step. You must configure yourAUTOEXEC.NOS file with your IP address , routing parameters (who will relay yourpackets?) and severa l other bits of information. This can be a frustrat ing, time­consuming exercise if you don't know what you're doing!

Local TCP/IP operators are your best resource , by far. They'll be able to help youget your stat ion on the air in the shortest time. They 'll also have the most up-to-dateversio n of DOMAIN.TXT for your particular network. TCP/IP packeteers are almostevangelical in their quest for new conve rts, so you'll likely find several willing to assist.If you're already active on standard packet , drop a message on your local bullet in boardand ask for assistance . TCP/IPers usually check the local BBSs regularly .

If you must install NOS by yourse lf, pick up a copy of NOSintro by Ian Wade,G3NRW. This book will supply much of the informat ion you need. See your favoritedealer or the ARRL Publicat ions Catalog in this issue.

to him, or tr an sfer a file to his co mputer?TCP/IP gives you an easy way to find out.Using WS 10 as our example, I can determineif his station is active by issuing a pin g.

ping ws to

My station sends a response req uest toWS 10 . It weaves throu gh the network untilit arrives at his stat ion. Hi s computer re­spon ds and I receive a rep ly, alon g with thetim e (in milliseconds) that it took for my pin gto get there and back.

44.88.4.23 : ech o reply 7000 ms

All thi s jargon basicall y tell s me thatWS lOis on the air (his address is shown) andthat my ping required 7000 milli secon ds(7 seconds) to ma ke the trip.

Give Him a FingerNo , not that finger! Let ' s say that you want

to find out more about WB81MY witho utboth erin g to se nd a message and wa it for myrep ly . The quick way to do thi s is through thefinger co mm and . It ' s a terrifi c fea ture fornosy packeteers!

f inger wb8imy@wb8imy

Assumi ng th at you have access to myTCP/IP network, yo ur fin ger requ es t will

travel to my sta tion. Lik e many TC P/IP op ­era tors, I have a spec ial text file that providesa brief rundown on my sta tion . When mycomputer recei ves a finger requ est, this file isse nt aut om ati call y.

Connect with Telnet

The telnet func tion allows yo u to do key­board-to-keyboard work. You can co nnect toanother station on the network and acc ess hisor her mailbox, ge t a list of stat ions heard andso on. If th e othe r op erator is present atthe keyboard, yo u ca n even enjoy a "li ve"co nve rsa tion.

Many TCP/I P Int ern et gateways allowyo u to use teln et to co nnec t to an Intern etRelay Ch at (IRe ). This is the equiva lent ofa conversa tional foo d fight in cy berspace!(Everyone is talking to each oth er at once.) Ifyo u don 't mind the confusion , yo u' ll enjoytapping into an IRe.

TCP/IP Q&A

Q: Can I use any TNC with TCP/IP?

A: As long as it incl udes the KISS mode,yes. (Most TNCs do .) To pu t yo ur TN C inKISS, you may have to enter the co mmandKISS ON followed by RESET. To get out of

KISS , so me TNCs requi re you to hold the ALTkey and type 192. Hold ALT again and type255 . Release the ALT key and then pre ss itone more tim e while typin g 192. These pro ­cedures vary , so co nsult you r TNC manu al.

Q: Does it have to be a 9600-bitls TNC?

A: No. You' ll find plenty of 1200-bit/sTCP/IP activity, primari ly on 2 meters. If youcan upgrade to a 9600-bi t/s station, so muchthe better. TCP/I P rea lly shines at high dat ara tes ! Mos t 9600-bit/s TCP/IP acti vity see msto take pla ce on the 440-M Hz band.

Q: What kind of computer do I need?

A: I recommend an AT -c lass IBM PC or corn­patible such as a 286 , 386 or 48 6. So me ver­sio ns of NOS wi ll run on old XTs , but otherswo n' t. Some TCP/IP operators also enjoyusin g Apple Macintosh machines.

Q: Do I have to leave my computer on24 hours a day to receive mail?

A: If yo u ca n leave your TCP/IP statio n run ­ning co ntinuo us ly, yo u' ll be able to se nd andrecei ve mail at any ti me. Thi s is convenientfor everyone concerne d . So me packeteer sbuy used PCs (such as relatively chea p 286mac hines) and dedi cate them to their TCP/IPstations. They ju st pa rk them in a co rne r withthe other equ ipm ent and lea ve them on da yand night.

Th is approach isn' t practical for everyone,thou gh. That's where POP comes into play .POP stands for Post Office Protocol . If an­other station on your net work is active 24hou rs a day, he or she may be will ing to act asa depository for yo ur incoming mail-other­wise known as a POP se rve r. By activatingthe POP fun ct ion in yo ur NOS software, yo urstation will automatica lly contac t yo ur POPse rver and grab any waiting mail. Th is usu­ally takes place imm edi ate ly after yo ur sta­tion co mes on the air.

Q: Can I stilI connect to my friends onstandard (AX. 25) packet if I'm runningNOS?

A: Absolutely! You ca n still connect to yourAX.25 BBS and yo ur friends can connect toyou. For exa mp le, the vers ion of NOS I useallows me to make an AX.25 co nnec t to mylocal BBS by typing the fo llow ing co mma nd:

c2mw1nrg

The lett er "c " sta nds for "co nnec t" while"2m" is the designator my NOS uses to accessmy TNe. Finally, "w Inrg" is the call sign ofmy local packet BBS. When I enter this com­mand, NOS will use my TNC to make a stan­dar d packet connect ion to WI NRG .

Some AX .25 pac ket BBSs eve n have"ports" to the TCP/IP network . This mean stha t yo u ca n use yo ur TC P/IP telnet co mmandto access the BBS wit hou t resorting to thecommand line shown abov e, and wi tho utleaving your TC P/IP network frequency . It ' sas though the BBS is a split person alit y-it 'sa TCP/IP sys te m and a standard pac ket BBS!

Spec ia l than ks to John Ackerma n, AG9V,fo r his ass is tance in the pr epa ration of thisarticle .

Ope rating 1-7

Page 17: Packet- Speed- More Speed

Exploring the 9600-BaudPACSATsLook to the heavens for a new packet radio challenge.

By Andrew Cornwall, VE1COR5 Belmore OrWellington, NS B2T 1J4Canada

[tJ ur Amateur Radio packet satellite"fleet" is impressive . There arepresently eight PACSATs in orbitserving thousands of hams world­

wide. Five of these satellites operate at datarates of 1200 bit/so(Most of the 1200-bits/ssatellites require special PSK TNCs inground sta tions , not ordinary packet TNCs.)The remaining three-OSCARs 22, 23 and25-use 9600 bit/s .

The 9600-bit/s PACSATs are the mostpopular. Why? All of the PACSATs travelin relative ly low orbits. This means thatyou can use them for only about 15 or 20minutes before they disappear below yourhorizon. If you onl y have a short amountof time on the satellite, you want to com­municate at the highest data rate you canmanage . You can acc omplish eight timesas much in 15 minutes at 9600 bitls as youcan at 1200 bit/s!

somewhat less than 800 km for UO-22 andKO-25 , and about 1,300 km for KO-23. Asthey pass overhead, ground stations withintheir 5,000-km-wide!ootprillts hav e an op­portunity to uplo ad and download packetdata files con sis ting of personal mes sages,bulletins, computer programs and graphics.The files range in size from a few hundredbyte s to over 300 kbytes. Even at 9600 bitlsyou may need a few pas ses to downl oad alarge file.

" Broadcast" P rotocol

Despite the sign ific ant amount of datathat can be captured during a pas s, there iscon siderable competition among groundstations about ex actly whi ch data the satel­lite should receive or send! There are typi­call y two or th ree do zen stations within asatell ite ' s moving footprint , all makingtheir requ est s.

The PACSATs sort out the pileups bycreating two queues (waiting lines)-onefor uploading and another for downloadi ng.The upload queue can accommodate twostations and the download queue can take asmany as 20 . Once the satellite adds a groundstation to the queue for down load ing , thegro und station moves for ward in the lineuntil it reaches the front, whereupon thesate llite services the req uest for several sec­ond s.

Unless the file you want is small, youwon 't get it all in one shot. After the satel­lite completes its transmission to your sta­tion , you're bounced back to the end of thequeue once again. In some cases yo u won' tautomaticall y be put in queue again , but willju st fall out of line entirely . When this hap­pen s, you r station must signal the sate llite ,sometimes repeatedly, to get back to the endof the queue. For uploading, the satellite

Profile of the Big Three

UoSAT-OSCAR 22 was designed andbuilt by the Uni versity of Surrey (England)in 1991. KITSAT-OSCAR 23 and KITSAT­OSCAR 25 were designed by the KoreanAd vanced Institute of Science and Technol­ogy (KAIST) with assistance from the Uni­versity of Surrey. Th ey were launched in1992 and 1993 , respectively .

All three satellites have store-and-for­ward messaging as their primary function.That is, they accept mes sages and hold themuntil they are do wnloaded by the recipients.They al so tak e still pictures of the Earth fordownloading as computer graphics, and ru ntelemetry for scientif ic experiments.UO-22 , KO-23 , and KO-25 use thePACS AT protocol and operate in Mode J,which means they rec ei ve on 2 meter s andtran smit on 70 cm. FM is used fo r both theuplinks and downlinks.

One of the fascinating aspects of thesesatellites is their small size . Th ey pack allof the equ ipment for radios, po wer, com­puter, image capturing, thermal contro l, andphysical stabili zation in a 9-in ch, solar-cell­clad cube . Two-meter and 70 -cm ant ennasare attached at the top and bottom of thecube . The UHF transmitter output is about5W.

Each sat ellite orbits the Earth in a paththat crosse s both poles at an altitude of

1-8 Chapter 1

MSPE [KO-23) aD.Eile ~etup Q,irectory Fill Satellite Send Msg tlelp

12c'lll . If!> w e 8 File 12ECB downloaded1Z11E~ Z~ 8 8 File 12E~F heard

I 12EBII 18 37 85408 Saving file 12E87r--In28 1 B B llato: Request fi ll of file 12E28

NO - 1 UEiCORI 12CE0 2 9 9 Br-oadc as t; q u e u e full.I 12E8A 9 0 0 File 12CE0 hear-dI 12E9F 9 o • 0 File 12C7A hea r-d

12D03 16 0 8 Auto: Reques t fill of file 12E28File 12CE0 hear-d

:H~~~4 :. 28 1l!lb433 File 12EBA heard2D 3 44 8 8 Auto: Request fill of file 12E28

Sat Apr 01 22: 42:23 1995 Up: 54/21 : 7 EllAC= 717 F:78960 L:77648 d:0 s:0 b:207205Sat Apr 01 Z2:42:38 1995 Up: 54/21:8 EDAC= 717 F:78192 L:77648 d:0 s:0 b : 207206Sat Apr 01 Z2:42:53 1995 Up: 54/21:8 EDAC= 717 F:78192 L:77648 d:0 s:0 b:207206Sat Apr 01 22 :43 :08. 1995 Up: 54/21:8 EDAC= 717 F:78192 L:77648 d:0 s:0 b :207207Sat Ap,' 01 22:43:23 1995 Up: 54/21 :8 EllAC= 717 F:78192 L:77648 d:0 s:0 b:207208Sa t Apr 01 22:43:38 1995 Up: 54/21 : 9 EDAC= 717 F:77488 L:77488 d:0 s:0 b:207209Sa t Apr 01 22: 43:53 1995 Up: 54/21:9 EDAC= 717 F:77488 L:77488 d:0 s:0 b:207210Sat ApI' 01 22: 44 : 08 1995 Up : 54/21: 9 EDAC= 717 F:77488 L:77488 d:0 s:0 b:207212Sat ApI' 01 22 :44 : 23 1'195 Up: 54/21:9 EDAC= 717 F:77488 L:77488 d:0 s:0 b:207213Sat Apr 01 22: 44:38 1995 Up: 54/21:10 EDAC= 717 F:77488 L:77488 d:0 s:0 b :20721Sat Al' r 01 22 :44 : 53 1995 Up: 54/ 21 :10 EDAC= 717 F:77488 L:77488 d:0 s:0 b:20721Sat Apr 01 22 : 45: 08 1995 Up: 54 /21:18 EDAC= 717 F:78248 L:77488 d :8 s:8 b:28721Sa t Ap,' 81 22:45:23 1995 Up: 54 /21:18 EDAC= 717 F:78240 L:77488 d:0 s:8 b:28721Sat Apr 81 22: 45:38 1995 Up: 54/21:11 EDAC= 717 F:77488 L:77488 d :8 s:8 b:20721Sat Apr 81 22: 45:53 1995 Up: 54/21 :11 EDAC= 717 F:77488 L:77488 d:8 s:8 b:20721Sat ApI' 01 22:46:23 1995 Up: 54/21:11 EDAC= 717 F:77488 L:77488 d:8 s:8 b :20722Sat Apr 81 22: 46:38 1995 Up: 54/21 :12 EDAC= 717 F:77488 L:77488 d:8 s:0 b :28722Sat ApI' 81 22:46:53 1995 Up: 54/21:12 EDAC= 717 F:78248 L:77488 d:0 s:0 b :28722e-:]!J1IfBU UEJ BDR OH6ICII'tl )(8TL Deli'ii'inMUPL OH4l<UI~ IIB2YLJr Tpe n 2a : GU6EFB

BED" IIB2RE1t HZ3~ DFSDP C8SUL F1TIU G3CD)(~ Lll1BB UEJEGO. ZU"'D UEJ PRHI DIR 44 holDS I AUTO 12E28 183:-. I 1:548631 I D:48934 I F:3?5225 IFigure 1-A lot is going on in th is sample of a WISP screen during an actualKO-23 pass, The list near t he bottom shows the ground stations waiting fordownloads of directories (\0), message s, or fi les , GU6EFB, bottom right, isuploading a file,

Page 18: Packet- Speed- More Speed

Figure 3-My unusual PACSAT station setup uses two computers. With second­hand PCs, it's also cost-effective. The alternative is to use one properly equipped386 or 486 PC running WISP in Microsoft Windows to control the equipment andcapture the data. Most hams prefer the single-computer approach. I just happenedto have two computers available!

computer equipment. There is not ye t a" sta ndard" satell ite ground station, andnearly every rig has some aspect of homebrew con struction.

There are a number of affordable 9600­bit/s packet TN Cs on the market. The ch al­len ge, howe ver , is to get your radio to trans­mit and recei ve the high-speed dat a sig nalwitho ut di stortion . Unlike l 200-bit/spacket, you ca n' t simply run the tran smitaudio into the microphone jack. Instead , thedat a signal must be injected at the modula­tor. Th e receive signa l must be tapped at theFM discrim inator. For most kinds of rad iosthe mod ificat ion is fairly simple . Neverthe­less, many ground station operators haveexperienc ed the trepidation of expo sing theguts of ex pens ive rad ios, wi th so lderingiron in hand .

So me radios claim to be "9600 ready"without modi ficat ion, but some work sig­nificantl y bett er than others . (See ''' 9600­Read y Radi os ' : Read y or No t?", by JonBloo m, KE3Z, in the May 1995 QST.) Andeve n the radios that are truly 9600 capablemay not be suita ble for PACSAT opera tion.So, many satell ite packeteer s still rely onrig s they modify them selves, such as theYaesu FT-736, Ken wood TS -790 or theICOM IC-820H (o nly a slight mod is neces­sary fo r the IC- 820H ).

Th e good news is that you don 't needelabo rate ant enn as or high output power.My station is curr entl y set up to work the9600-baud PACSAT satellites exclusivel y.I run only 25 W for the 2-meter uplink. Myantenna sys tem cons ists of a single logperiodic dipole arra y with a duplexer tosepa rate the 2-meter and 70 -cm signa ls .Because the elevation beamwidth ofthe LPDA is quite wid e, I've been succes s-

Operating 1-9

CAT~Interfa ce

ControlComputer

VHF/UHFTransce iver

Most -MountedPreamp

~st> 9600- bit/s TNC~ TNC Int er face

Dat aComputer

AC Power toAll Devices

AC Power

~D Event

~nmer

PACS AT station. The gro und sta tion soft­ware communic ates with the satell ite, re­ceiv ing the broadcast messages, fill ingholes in file s and so on. For years the mos tpopular ground station software was actu­all y two pro gram s: PB for do wn loading andPC for uploading. Within the last year orso , howe ver , a new program kno wn as WISPhas taken the PACS AT world by storm.Written for Mi cro soft Windo ws, WISP of­fer s a much "friendlier" en vironment thanPB or PC (see Figure I). It will also total lyautomate your station, if you choose to doso. Both PB/PC and WISP are av ai lablefrom AMSAT, PO Box 27, Was hington , DC20044. AMSAT also sells a valu able guidefor anyone who wan ts to try the PACS ATs .It ' s the Digital Sa tell ite Guide and it in­cludes the PB/P C softwa re. Send a self-ad­dressed, stam ped en velop e to AM SAT formore in form at ion.

Through so ftwa re such as WISP, and theaddition of a special tracking card or exter­nal controll er, yo ur computer can auto mateall vit al functi on s. It can run the TNC, tunethe radio (and con tinu ally retun e it to trackthe satellite sign al, compensating for Dop­pler shift), and aim the antenna to track thesa te ll ite as it trav e ls acros s the sky, fromhorizon to hori zon . With a full y auto matedgro und station, ther e is litt le for an ope ra torto do. Ind eed, many stations run unattendedduring a pass. In advance of a pass, the hu­man operator of a ground sta tio n wo uldschedule the co mputer to upload and down­load particular messages and ot he r fil es.Item s rec eived are stored on the gro undstation's hard di sk to be read or otherwiseused later.

Gro und sta tion s working the 9600-b it/sPACSATs use var iou s kind s of radio and

Figure 2- Tilting t he log periodicantenna to take advantage of its wideverti cal beamwidth avo ids the need foran expensive azimuth/elevation rotor.Despite the odd angle, the antenna alsoprovides good terrestria l VHF and UHFperformance.

atte mpts to receive fro m the two gro undstations in the up loa d qu eu e un til their vari­ous transacti on s are co mp leted .

If the sate llite di sappears be fore yo u re ­ceive the complete file, the re 's no need towo rry. Your PACS AT ground station soft­ware "reme mbers" which part s of the fi leyou still need from the bird. When it ap­pears again , your software can request thatthese "holes" be filled .

And while all of thi s going on, yo u 'rereceiving data that othe r sta tio ns have re ­qu ested! No t only do you ge t the file yo uwa nted, yo u also rec ei ve a large porti on ofthe data that other ham s have requ ested. Infact , yo u may rece ive a num ber of messagesand files with out tran sm itt ing a sing le wa ttof RF . All yo u have to do is listen. Th at' swhy they call it "broadc ast" protocol.

Station Equipment and Software

The computer is at the heart of an y

Page 19: Packet- Speed- More Speed

The VE ICOR Ground Station .

The block diagram (Figure 3) shows the current VE1COR ground station forworking the 9600-bit/s PACSATs. Although my setup is a bit unusual (note the twocomputers), it's adequate to communicate reliably with KO-23, KO-25 and UO-22.

My first ground station consisted simply of a dual-band omnidirectional VHF/UHF vertical antenna, VHF/UHF duplexer, mast-mo unted UHF preampl ifier, VHF/UHF transceiver, 9600-bitls FSK modem and TNC, and an IBM-compatible PCrunning PB and PG software. The download capability of the this setup was around100 kbytes per pass. As noted below, additions to the initial ground station haveincreased operating convenience and efficiency. My present download performancewith KO-23 is 300 to 600 kbytes , sometimes more than 700 kbytes .

Event TimerAn event timer automates the operation of the station. It's programmed to turn

on ac power to the station for the period between five minutes before and eightminutes after each pass. When powered on, the computers boot and run therequired software . A Radio Shack event timer is programmed for seven on/off cyclesper day, allowing contact with KO-23 on all of the stronger passes (those where thesatellite is at least 15° above my local horizon). In order to reduce current load, thetimer does not supply power to the computer monitors, which can be left off .

TNC InterfaceThe TNC interface is a home brew exper iment consisting of severa l op-arnp

stages to buffer and match the impedances of the modem and the transceive r. Theinterface provides a small treble boost to the audio from the received signal. There isalso a manual switch that decouples the modem from the transceiver, so thetransceiver can be used for terrestrial communications . Most stations would not needa TNC Interface. A single PC is dedicated to talking to the TNC and running thePACSAT communication software.

Computer-Assisted Transceiver (CA T) ControlA commercial CAT interface adjusts for signal voltage differences between the

microcompu ter's serial port and the transceiver's CAT port . Many functions of thetransceiver can be controlled by a microcomputer. In this instance, the computeradjusts the transceiver's receive frequency to offset Doppler shift. The secondstat ion computer does this by monito ring the transceiver's deviation meter andchanging the receive frequency to keep the meter at center scale. Tracking thereceived frequency throughout a pass significantly increases download performan ce.

Log-Periodic AntennaA log-periodic dipole array offers mult iband capab ility, gain of about 12 dBi,

relatively wide beamwidth, and compact size. The added gain improves communi­cation throughput. The antenna is directional, however, and must be aimed at thesatellite as it crosses the sky.

Antenna Rotor ControlAn inexpensive, light-duty azimuth rotator turns the log-periodic antenna in the

horizontal plane. Elevation tracking is not necessary because of the wide vert icalbeamwidth (60°) of the verticall y polar ized antenna . Pointing the antenna 22° upwardworks quite well. During some passes, when the satellite goes direct ly overhead, thesignal drops off for a couple of minutes. With the majority of passes, however, signalinterruption is not signifi cant.

The same IBM-compatible computer that tunes the transceiver for Doppler shiftalso controls the rotator. A home brew rotato r controller replaces the factory -suppliedmanual cont rol box. The microcomputer turns the rotor to predete rmined bearings atspecific times during a pass. A batch of bearings are computed beforehand by asatellit e tracking program and stored in computer files.

ful by tilt ing the antenna upward at a fixedelevation angle and usi ng only a single ro­to r for az imuth tracking (see Fig ure 2).Dur ing my earl y expe rime nts I was abl e towork UO-22 and KO-23 with ju st a simp le2-m ete r/70-c m, omnidirectio nal verticalantenna ! (See the sidebar, "T he VE1 CORGround Stati on ." )

1-10 Chapter 1

A Challe nging Hobby!

A common bond among amateurs onsate lli tes ge nerally is the realizatio n tha tso meday the ir favo rite bird s will ceas e tofunct ion. Th is is inevitab le, and do n' t counton an yon e to take a ride on the shutt le to fixa broken ham satelli te . For example, when

KO-23 was launched, its life ex pectancywas on ly three to five years bec ause of theexpected degradation caused by its orb itbeing in the Van Allen radiation belt. Coun­tering thi s pessimisti c outlook, however, isthe strong intere st to launch new satellites .In late March, a Russian rocket fa iled in anattempt to launch two new PACSATs, onebui lt by the Uni ver sity of Me xico and theother by the Technion-Israel Institute ofTechno logy (see the "A mateur Sate lliteCom munication" co lumn in this issue fordetail s). On the other hand, the JapanAma teur Sa te lli te Organ iza tion plans tolaunch a 9600-bi t/s PACSAT ca lled JAS- 2in 1996 . At about the same time, AMSATplans to launch a multipurpose super-satel­lite known as Pha se 3D. Other organiza­tions are also wor king on Amateur Radiosate llite project s. (Did you read the ar ticle"Phase 3D : The Ulti mate Ea sySa t" in theMay 1995 New Ham Co mpanion? If not, goback and read it-especially the last fewparagraphs !)

Within the inevitable budget constraints,volunteer satellite designers and assemblersattempt to use compo nent pa rts that arehard ened to withstand the mass ive g-forcesand vibration of the launc h, the rigors ofcontinual cos mic radiation, and the cycl icextremes of temperature . Each success fulpass of a sate llite testifies to the abi lity ofthe crew that bro ught it into existe nce andwho continue to main tain system operation .

Data satellites occasionally ex per iencetemporary difficult ies . KO-23 and KO-25both "crashed" (in the computer sense ) inDecember 1994. KAIST was successful inre loading the sate llites ' software andrebooting. KO-23 crashed again after onl ynine hours, and KO-25 stayed operatio nalfor about a week before crashing again.Both satellites were functioning whe n thisarticle went to press.

Be forewarned tha t pursu ing the am a­teur sat ellite hobby , whether voice, CW , ordigital, can be inconvenient and frus trat ing .Communicati on is possib le at only ce rtain ,oft en rel ati vel y brie f, periods of the day.Working in the dig ital mode present s theadditional chall enge of sett ing up and oper­ating extra gear consi sting of microcomput­ers, software , and the TN C. The comp lexityof digital ground station operation is espe­ciall y ev ident if a problem occur s. For ex­ample, when confronted with the problemof poor per form anc e. is it due to the an­tenn a, rotor, antenna preamp, tra nsmiss ionlines, rad io, TNC or what? Or is the prob ­lem wit h the sate llite itsel f?

Despite the se trib ulations, use rs ofUO-22, KO-23 . and KO-25 frequent ly postmessages reflec ting their enj oyment. Topara phrase one PACSAT de votee in York­shire, Engl and: Thi s is no hobb y. It is anobsession!

Adaptedfrom an article originally writtenfo rthe Halifa x, Nova Scotia. Amateur Radio ClubNewslette r.

Page 20: Packet- Speed- More Speed

By Steve Ford, WB81MY

OX Pack,tClu,t,p,The time-saving solution for busy DXers and contesters!

lIDlove to spe nd hours tro lli ng the HF bands. I rea llywould . Wit h a hot cup of coffee at my side, I could

• spend these chilly winter evenings searching for DXon my favorite ba nds . And the n there are the co n­

tests. Yo u can burn through several hours hu nting for that las tstate, section, co unt ry or whatever.

In tr uth, however, I'm typical of most middle-aged andyounger hams-the majori ty of the ham popu lation today. Oureven ings and weekends are fi lled with school or fami ly activities.We have to squeeze Amateur Rad io into those few precious mo­ments when we can park ourselves in fro nt of our radios .

So how do you reconci le the desperate need to pursue DX orcontests, and sti ll fin d time to meet all your other co mmitments?Wo uldn't it be nice if someone co uld at least do the hun ting foryou? That's the most time-cons uming part. With the rep ort s oftheir "reconnaissance" in hand, you cou ld choose which "prey"you wanted to sta lk .

But who would be crazy enough to vo lunteer for such a job?As it turns out, there are thousands of such "volunteers" through­out the nation .. .and thanks to some nifty software developed byDick Newe ll, AK I A, they 're co nnec ted to DX PacketClusters!

As Close as your TNCIf you have a packet radio setup (a 2- meter FM tra nsce ive r,

terminal node contro ller [TNC] and a co mputer or da ta term inal ),you're ready to enjoy the advantages of PacketClus ters. APacketCluster is a network of relayi ng node s devoted to contest­ing and DX hunting . Some PacketCluster networks are sma ll,single-node systems wit h a dozen users or so (see Figure I ). Othernet works are vast, wit h ma ny nodes interco nnected using VHF /UHF or Internet links (see Figure 2) .

Regard less of its siz e, a Packe tCluste r exists for one funda ­mental purpose: sharing inf ormation. When a ham sen ds a pieceof information to a PacketCluster node (the appearance of a rare

Figure 1-Asm a llPacke tClustermay becomprised of asin gle node.Everyone whocon nects to thenode shares thesame OX orcontestinformation .

DX stat ion, for example), the node relays it to everyo ne con­nec ted to the network. If you're on the network, you share the"profits ."

Busy hams like me can connec t to their nearest PacketClusternodes and, withi n seconds, receive lists of all the most recent DXsig htings . In the bl ink of an eye I can see which bands are hot atthe moment, and which DX stat ions are on the air (and at whatfrequ enci es). And if I'm the lucky guy who stumbles acros s a DXsignal firs t, I can post the info rmation for eve ryo ne else to see.(Share and share alike!) If I' m dabbling in a co ntes t, I can connectto the PacketCluster and dete rmine where to find the sta tions Ineed to boost my sco re.

The BasicsYo u connect to a PacketCluster in the same way you'd con­

nec t to any other pac ket station. However, the information you' llrece ive will be very different !

Th e sec re t of using a PacketCluste r is knowin g the variouscommands . The most common ones are show n in Table I .

My routine is to connec t to my nearest node and imm ediatelyask for a list of the latest DX sightings using the SHOW/OX com­mand (Figure 3). Then, I take a peek at solar condi tion s by send­ing SHOWIWWV. By sending these two commands, I receive acapsule summary of band conditions and DX activity.

If I' m tuning through the bands I might stumble on a DX sta­tion worthy of a spo t on the clu ster . Let ' s say I hear SV3AQR on

Figure 2- 0ther PacketCluster networks are huge, coveringentire stat es or reg ions. The individual nodes often use UHFradio links or Internet connections to exchange informati on.

Operating 1-11

Page 21: Packet- Speed- More Speed

A/F x

Al x Y

Al x y

DIR/OW N

prefi x(s) is z, eg, SE/NEED/RTTY/BAND=(10) ZS9 .

Indicate that your computer/terminal is notANSI -compatible .

Indicate that you are not in your shack .Set your QTH as location x.Display requested information.Display names of files in archive fi le area.Display names of files in bulletin file area .Display phys ical configuration of clust er.Display station conn ected to node whosecall sig n is x.

Display names of nodes in clusters , numberof local users, number of total users andhighest number of connected stations.

Displ ay available show commands.Display the last five OX announcements.Display the last five OX announcements forfrequ ency band x.

Disp lay the last n OX anno uncements.Display the last n OX announcements forfrequency band x.

Display names of files in general files area .Display mail-forwarding database.Display head ing and distance to countrywhose prefi x is x.

Disp lay status of inact ivity function andinactivity timer value .

Displ ay your stati on's long itude and latitude.Display the longitude and latitude of stationwhose call sign is x.

Display last five entries in cluster's log.Display last n entr ies in cluster's log.Display MUF for count ry whose prefi x is x.Display needed countries for station whosecall sign is x.

Display stat ions needing country whosepref ix is x.

Display needed countr ies for mode x wherex equals CW , SSB , or RTTY.

Display system not ice .Display pref ix(s ) starting with letter(s) x.Display QSL information fo r station whos ecall sign is x.

Display sunrise and sunse t times for countrywhose pref ix is x.

Display call signs of stations connected tothe cluster .

Display version of the cluster software.Display last five WWV propagationannouncements .

Tal k to anoth er stati on .Tal k to station whos e call sign is x. Send<CTRL-Z> to terminate tal k mode .

Send one-line mess age y to station whosecall sign is x.

Displ ay a file .Display file named y stored in fil e areanamed x.

Displ ay n lines of file named y stored in filearea named x.

Update a custom database .Update the database named x.Add text to your entry in the databasenamed x. .

Upload a file .Upload a file named x.Upload a bul let in named x.Upload a file named x.Announce and log WWV propagationinformati on.

Announ ce and log WWV prop agationinformation where xxx is the solar flux,yy is the A-index, zz is the K-index anda is the forecast.

W SF= xxx, A=yy ,K=z z,a

SH/I

SH/LOCSH/LOC x

UPDATEUPDATE/xUPDATE/xlAPPEND

UPLOADUP xUP/B xUP/F xWWV

SHINE x

SH /S x

T xy

SH /NE /x

SH/LOGSH/LOG nSHIM xSHINE x

SH/NOSHIP xSH/QSL x

TYPETY l x y

TY l xln y

SH/U

SHIVSH /W

TALKT x

SH/FISH/FOSHIH x

SH /DXlnSH /DXln x

SH/COMSH/DXSH /D X x

SE/NOA

SE/NOHSE/Q xSHOWSH/ASH/BSH/CSH /C x

SH/CL

DXla x y z

SE/NEEDl x Y

OXOX xyz

Table 1Common PacketCluster CommandsANNOUNCE Make an announcement.A x Send message x to all stations connected to

the local node .Send message x to all stations connected tothe cluster.

Send mess age y to stations connected tonode x.

Send mess age y to stations on distributionlist x.

Disconnect from cluster.Disconnect from cluster.Enter the conference mode on the local node.Enter the conference mode on the local node.Send <CTRL-Z> or IEXIT to termi nateconference mode .

Enter the conf erence mode on the cluster.Send <CTRL-Z> or IEXIT to terminateconference mode.

Delete a message.Delete last message you read .Delete message numbered n.List activ e messages on local node.List all active messages on local node.List act ive messages addr essed to "all."List the n most recent active messages.List active messages added since you lastinvoked the DIR command.

List active messages addressed from orto you.

Announce OX station .Announce OX station whos e call sign is x onfrequency y followed by comment z, eg,OX SP1N 14.205 up 2.

Announce OX stat ion whose call sign is x onfrequency y followed by comment z withcredit given to station whose call sign is a,eg, DX/K 1CC SP1N 14.205 up 2.

Find file .Ask the node to find file named x.Display a summary of all commands.Display help for command x.Read message.Read oldest message not read by you .Read message numbe red n.Read file named y stored in file areanamed x.

Reply to the last message read by you .Reply to the last message read by you .Reply to and delete the last message read by

youSend a mess age.Send a private messag e.Send a publ ic message.Set user-specific para meters .Indicate that your compu ter/te rminal isANSI-compatible.

Indicate that your computer/terminal isreverse video ANSI -compatibl e.

SE/H Indicate that you are in your radio shac k.SElL ab c d e f Set you r station's latitude as: a degree b

minutes c north or south and longitude ddegrees e minutes f east or west.

Set your name as x.Store in databas e that you need country(s)whose prefi x(s) is x on CW and SSB. eg,SE/NEED XX9.

SE/NEED/BAND Store in database that on frequency band (s)=(x)y x, you need country(s) whose prefix(s ) is y,

eg , SE/NEED/BAND=(10)YA.Store in data base that in mode x (where xequa ls CW, SSB or RTTY ), you needcountry(s) whose prefi x(s) is y, eg,SENEED/RTT Y VA.

SE/NEEDl x/BA ND Store in database that in mode x (where= (y)z x equals CW, SSB or RTTY) on frequency

band(s) y, you need country(s) whose

FINDFILEFI xHELP or?HELP xREADRRnR/x y

REPL YREPREP/D

DELETEDEDE nDIRECTORYDIR/ALLDIR/BULLETINDIR/nDIR/NEW

CONFERI F

SE/A/ALT

BYEBCONFERENCECONFER

SE/N xSE/NEED x

SENDSIPS/NOPSETSE/A

1-12 Chapter 1

Page 22: Packet- Speed- More Speed

SHOW/IRC

SHOW/PUB

SHOW/FLUX

SHOW/ZONE

SHOW/QSLRECSHOW/RULESSHOW/IOTASHOWITODAYSHOW/COUNTYSHOW/EXTRASHOW/INFOSHOW/MICSHOW/LADDERSHOW/SWL

SHOW/NCDXF

SHOW/DXCCSHOW/FCC

SHOW/CONTESTSHOW/COORDSHOW/DEALER

SHOW/OXNODESSHOW/BUREAU

Table 2Additional SHOW CommandsNote: Some of these commands may not work on your localPacketCluster network.Command Definition

SHOW/BUCKMASTER Buckmaster US call-sign listSHOW/RUMORS Anything and everything!SHOW/OBLAST Russian oblast informationSHOW/PREFIX Listing of countries and zones by prefixSHOW/ALLOC ITU allocation tableSHOW/STARS Planetarium dataSHOW/BAND Frequencies available for each class of

licenseListing of countries and zones by prefix orzone

Database of known OX PacketCluster nodes.World and Russian Oblast QSL bureauaddresses

Calendar and info on OX-related contestsLongitude and latitude of US/OX locationsAmateur Radio dealer telephone/FAXdirectory

ARRL DXCC Country ListFCC Field Office Directory and FCC InfoGuide

Historical sunspot/solar flux data andglossary

Required IRCs for QSL returns andIRC/postal info

Information on the Northern California OXFoundation

Directory of Amateur Radio publications fromvarious countries

Information on QSL cards received by usersFCC Rules and Regulations Part 97Islands on the Air (Award)Did you know? Events of yesteryear!Directory of US countiesElement 4B Extra Class Q&AInternational Q Signals, ARL MessagesMIC wiring data prepared by KC4LWIPacketCluster DXCC "Ladder of Success"Shortwave Listeners Guide

Above and BeyondPacket Clusters have other useful features. Yo u can use the OIR

command to see a list of bulletins just as you would on a packetBBS. U sing the R (R EAD) command wi ll allow you to read anybulletin you w ish. Unlike packet bulletin boards, however,PackerClu sters can rel ay bulletins and messages onl y w i thi n thenetwor ks they serve .

A s I 'm watching the Pa ck etCluster, I see a spot for aDXpedition on Sable Island. Hmmm... I could use that one! Iwonder what band s would provide the best propa gation from mylocation to Sable I sland ? W hy not ask the PacketClus te r'! A ll Ihave to do i s send: SHOW/M CYO.

" M " stands for maximum usable f requ ency (or M UF) and CYOi s the call sign pre fi x fo r Sabl e Island . Here ' s how thePacket Clu ster responds :

Sable-Is Propagation: Flux: 137 Sunspots: 90 Rad Angle: 29

2 1.250 MHz. I can add this i nformation to the netwo rk by send­in g: SV3AQR OX 21.250.

W ithin seconds the spot will appear on the screensof everyoneelse who is connected to the network.

On e questi on that comes up often is, " How do I know if a D Xstation is wo rth y of an announcement on the Cluste r?" After all ,what's choice D X to one ham is garden variety to another. Myadvice is to observe the acti vi ty on your network before yo u startpostin g spots of yo ur own . On some networ ks, any thing goes.Other networks, however , concentrate on rarer contacts. Youmight receive a sarcastic respon se if you post a spot for, say, aFrench statio n on 20 meters from the east coast. (A contact withFrance is not rare D X to mo st hams.)

In ei ther case, make your spots accurate. Be sure you've en­tered the call sign and frequency correc tly. If the D X stat io n isworking split, say so. Just a bri ef comment such as "LISTENINGUP 30" is suffic ient.

If there is a hot contest in progress, you' ll know right awaywhen you connect to the PacketCluster (see Figure 4). Just thenature of the spot l i sti ngs will tip you off. In addi tion, the nodemay send a sig n-on message telling you that the network is in the"contest mode." This means that certain PacketClu ster f unctionsare di sabled during the contest. Generally speak ing, yo u shouldonly post in forma ti on about contest contacts when the network isin this mode.

23572

2352Z23 54Z23 5922354Z2355Z

. :fil e f dlt S.ettlngs Blone I fanster. He lp

•nd : c kca pe** COHNECTED to KC8 PEIi s t eve t Weltone to YCCC Pac ke t c Iu s t er- nod e - Cheshi r e C1l uster: 30 nod es. 5 local / 2 21 tota l us e rs 11iUC use r s 758 up t Iee 6 05 :22~8IHY de KC8PE 24-S.p-1995 153 8 Z Type It or ? f or help >

how/d x15 40Z

1 8134.1 9U/ F5FHI 2 Jt- Se p-1995 1 5 4 02 <WA1YTW- 6>1 41 8 8. 2 X5BYZ 2Ja-Se p-1995 15 38 2 <K2SX>1 41 8 8. 2 X50A 2.1t-Sep-1995 1 5382 YU7KttD clai ns nee co untry <K2SX>211180 . 0 A71CW 2 .1t-Se p-1995 1 5322 RTTY <WA4QOH>1!l216. 5 OH1X 2"'- Se p-1995 1529 2 <HH3Q>~81HY d. KC8PE 24 - Sep 1539 2 >

ho..,/ uuuDat e Hour SF I A K Forec a s t

4 -Sep -1995 15 75 8 1 ULOW/QUI ET-A CTI U[ ;ULOW-lO U/QUI ET- UHSE ( H4 VI C>Jt-S e p - 1995 12 75 8 3 UlOW/QUI ET- ACTI U[ ;ULOW-LO U/QUI ET- UHSE ( H"'VI C>It-S e p- 1995 O. 75 8 3 Ul OW/ QUI ET- ACTI UE; UlO W-lOWlQU I ET- UHSE <H4YI C>Ji- Se p- 1995 06 75 8 4 UlO WlQUI ET- ACTI UE; UlO W- lOWlQU I ET- UHSE <H4YI C>4-Sep - 1995 03 75 8 1 Ul OW/QUIET- ACTI UE; UlO W- lO WlQUI ET- UHSE <H4YI C>

LIll81HY d. KC8PE 2!1-Sep 1 5392 ) .. .~C8PE Uol l'o/lNRGJ READMAIL Ud(J:C8PE1 I ~

_1IoI'1NRG IIY£ ICISS0N

", .

Figure 3-When I connect to my local PacketCluster node(KC8PE), I usually ask for a list of the most recent OX spots.I follow up with a request for WWV solar activity reports.

file fdlt llenl ngs fIlane I ra nslers Helpc KC8PE

• • COHHECTE D t o KC8PEIi St eu e t lIe l c oAe to yecc Pac ke tcluster nod e - Cheshire Cllus te r : 49 ncu es , 3 local I 422 t ota l us er s Hax us e r s 158 Upt i ne 7 21 : ..7~8 I HY de KC8PE 19 -Hou-1 995 23522 Typ e H or ? fo r help }

X de KY1H: 3808 . 7 UE2CUA pq f or anyo ne t hat still n eed s 235UZX de U3HUQ: 38 18 . 0 KE3ttX epa 23552o ALL de UB2EAR <2354Z> : HEED PACI FIe - PSEo Al l de KD1GU <2354Z> : HEED HS AHD KYX de H1CC: 14 26 6 . 9 KH6RS PAC ( Ny Sweep !)X de tF2l: 350 1 . 9 7Q7A qs x upX de KB3BlL : 38 24 .0 AA.ltRX IC YX de KY1H: 38.1t6.7 1JA1S nhX de KY1H: 38 67 . 7 K1RH cto nLL de KB2FGO <2359Z> : any one hear HO or ttn R??X de KY111 : 388 1 .2 K2YY Fr esh nea t ! sny

Figure 4-Here's a snapshot of my local PacketClusternetwork during the 1995 ARRL November PhoneSweepstakes. Notice that several spots have been posted for"rare" sections.

Operating 1-13

Page 23: Packet- Speed- More Speed

Dist: 995 km Hops: 1

MUF (90%): 9.0 (50%): 10.7 (10%): 13.1

Th e MUF ca lculations te ll me that prop agati on to the island ismost re lia ble at 9 MHz (90%), but drops rapidly as the frequencyincreases. The 30- me ter band (10 MHz) wo uld be the best bet.

DX Pac ketClusters also allow yo u to chat with ot her stationsin the network. You ca n send a simple greeting by using the TALKco mma nd:

TALK KX4V Hello, Rick . Nice job working 5U7Ml

IfRick wa nts to talk to me at len gth, he ca n use the same TALKco mma nd to es ta blish a lin k between o ur sta tions . ThePacketCluster will continue to show us new DX sig htings as theyappear, but everythi ng we type wi ll be se nt to eac h other.

If yo u wa nt to see how the Cluster ne twork is configured andwho's co nnected to it, j ust send the SHOW/C (show configuration)co mmand. Th e PacketCluster will respond with a complete list ofevery sta tion co nnected to the net wor k grouped by the node they'reusin g. Here ' s a typical examp le:

PacketCluster Configuration:

You can send TALK messages, or ent er into a conver sation,with any ham on the list. The only exceptions are call signs inparenthesis . These hams are con nected to the Cluster, but they'reaway from their keyboards temporarily . How does the systemknow this? Actually , it doesn't-unless you tell it. (Yes, there isa command for this, too!) And like packet bulletin boards, youdon 't need to send the full command ever y time. Instead, you canuse the abbreviated form such as SH/OX rather than SHOW/OX.

There are many other function s available, depending on thesophistication of your local network . See Table 2 for some of themore versatile and interesting SHOW commands.

Where's My PacketCluster?There are PacketCluster networks in most urban areas, and

even in some lightly pop ulated regions ofour country. Ron Rueter,NV6Z, ma intains a list of active Cluster networks, but it' s toolarge to publish here . If you have Internet access, you can obtainthe list from the ARRL InfoServer. Simply address an e-mailmessage to: [email protected]. Leave the subject line blank. In thebody of your message enter the following on separate lines :

SE ND CLUSTER .TXT

KC8PE N1API KS1L K1WJL KC1SJ WB81MY

K2TR KA2EXB (K2QE) N2JJ NJ1 F K2VV (KB2HUN)K20NP KQ2K WK2H KA2HTU WS2U N2EKU

W1RM (WB1AIU) NT0Z KA1BSA AB1U KB1LE N1JBHN1GLA KG1D-1 (NJ2L) (NA11) KB1HY K1FRDNX1L KB1BE W1CKA KB1CQ (W1GG) K1ZJ HWB1GUY K1KI

Node Connected stations QUIT

The server will send you the complete list right away . You 'llalso find CLUSTER.TXT on the ARRL Hiram BBS at 860-594­0306. List updates can be found at the following Internet ftp site :pinsight.c om in the directory /pub/K6PBT.

A nd what if there are no Packet Clusters in your area? Con­sider starting yo ur own ! You can obtain more information aboutPacketCluster software from the manufacturer: Pavilion Soft­ware, 8 Mount Royal Ave, Marlborough, MA 0 1752 . Enclose aself-addressed, stamped envelope . Ig!n--I

1-14 Chapter 1

Page 24: Packet- Speed- More Speed

------- - - -------- ------ --- - - - ----$ tan Horzepa, WA tLOU*

Wireles s WWW Page of the MonthWhile I am on the subject of VHF/UHF

DXing, it is appropria te that the Wire lessWorld Wide Web page for this month isthe North East Weak Sig nal VHF Gro up homepage at http ://uhavax.hartford.edu/newsv hf.This page has a fine selection of files and soft­ware related to VHF/UHF DXi ng and contest­ing. It also has a long list of links to otherVHF/UHF-re lated sites throughout the world.

len t for the Contro ls-Filte rs-Direct series ofcommands in Ma cAPRS or WinAPRS, so dif­fere ntia ting between the direct and digi­peated/ gatewayed statio ns on a Maci ntosh orWindo ws platform req uir es more work .

After yo u have used MacA PRS orWinAPRS for a while, you get a feel for whichsta tions are with in your normal limits, as wellas the HF stations that are be ing ga tewayed toyour LAN . Whe n an unfamiliar stat ion ap­pears on your map , you can check how you arereceiving it by using you r mou se to double­click its icon to dis play its path . For exam ple,today the path of K I HJC in Ca ndia , NewHampshire , appears as K IHJC>APRS,NR IN -2 ,WIDE.

Thi s indi cates that I am receiving K IHJCdirect ly . An asterisk in the path wou ld ind i­cate otherwise . For example, if an asteri skfollowed NR 1N-2 or WIDE, it would indica tethat I was receivi ng K I HIC via digi pea terNR IN-2 or via a digipeater with an alias ofWIDE. Similarly, if GATE* appeared in thepath, it would indicate tha t the station wasbeing gatewayed to me.

That exp lains how you can use APRS as aVHF /UHF propagation tool. Now let me goand look at my A PRS map . Maybe I can worksta te number 27!

NORTH EASTWE/Jd( SIGNALVER.MOM. CT

North E ast Weak Signal VHF Gro up Home Page

N.E .W.S .

GROUP

It.VHF-UHF-SH F /

R.EGIOMAl 'i..CLUB ;,,, - .

The North East Weak Signal VHF Group home page.

Wheat vs ChaffTo use thi s tool correct ly, you must inter­

pret what your APRS map is disp laying. On atyp ica l day, an APRS map displays a lot ofstations within a ISO-mile radius of your sta ­tion as well as pockets of activ ity sca tteredacross the continent. Such a di spl ay does notnecessarily mean that the band is open .

Most of the statio ns within a ISO-mile ra­dius (give or take mile age depending on thelay of your APRS LAN) are displayed on yourmap because APRS digipeaters are relayingtheir posit ions to your stat ion. Similarly, thestations beyond the ISO-mile radius are likelyto be HF A PRS statio ns whos e positions arebein g rela yed to 2 meters by an A PRS HFgateway statio n on your A PRS LAN. In anycase , you are not receiving most of the se sta­tions directly; repeaters and gateways aredo ing the work .

Onl y when you rece ive a distant stationdirectly is it an indica tion of unus ual propaga­tion conditions . So, how do you differenti atethe d ire ctly recei ved statio ns from thedigipea ted and gate wayed stations?

If you use the DOS version of APRS, youcan filter all but the direct stations by invok­ing the Con trol s-Filters-Di rect series of com ­mands. Once invoked, APRS only displays thestations you hear directly. There is no eq uiva -

are WA2JNF in Brooklyn, New York, 85 milesto my southwest and W ITDG in Hinsda le, NewHampshire, 90 miles to my northeast. How­ever , this morning, my comp uter moni tor dis­played stations such as WA IYKN on CapeCod , 135 miles to the east , NRlN in Warner,New Hamp shire , 133 miles to my northeast ,and N2MSM on the South Jer sey shore, 174miles to my southeast, indicat ing that there wastropo inversion prop agation afoot.

' One Glen AveWolcott, CT 067 16-1442e-mail [email protected]

Propagation Tool TimeWhen I started playing wi th WB4APR' s

Automatic Packet Reporting System (AP RS),I thought that it migh t be a suitable tool forchasing 2-meter DX . APRS is software tha tgraphically represents the position of mobil eand stationary packet- radio station s and otherobjects on a map disp layed on your computermoni tor. Most A PRS activity occ urs on thesame channel (for example, 145.79 MHz on2 meters) and, as a result, if yo u con tinuouslymonitor that channel, you can tell that the bandis open when stations begin app ear ing on yourmap that don 't normally appear there .

I was sold on using APRS as a 2-meterpropagatio n alerting mechanism when oneafternoon last July, stations in Georgi a starte dpopping up on my APRS map . I gues sed thatsporadic-E pro pagation was in the works, so Iswitched to SSB and worked a number of sta­tion s in Florida for my 26th state on 2 meters.

The oute r limits of my APRS coverage areaduring normal 2-meter propagation cond itions

Chasing DX with APRSTwo meters is open this morning . There is ~-----------------------------------,

a troposph eric (or "trope ") inversion along theEas t Coast that is enhancing propagation outto a rang e of I00 to 200 miles. Yo u wouldnever know this by mon itoring the 2-meterSSB ca lli ng frequency (144.200 MHz) asthere is nary a sig nal to be heard there . So ,how do I know that 2 meters is open? APRS isshowing the way .

Long before the marriage of home com­puter s and Amate ur Radio, my main interestwas in the VHF/U HF world. One of my firstradios was a Heat hkit 2-meter " Benton Har -bor Lun chbox," also known as a "Twoer." Itwas a one -channe l AM transcei ver wit h asuperregenerative receiver. "Super" was amis nomer. It cer tainly was not ofClark Ken t' slineage as it left a lot to be desire d in the se­lect ivity department. It was so nonselectivetha t one strong sig nal seemed to fill the whole2-meter band !

But, becau se of my limi ted college incom e(or shou ld I say "superincome"), I used to haulmy Twoer aro und in my car looking for thehigh spots in New Haven County to work 2­meter DX. It was tough goi ng, but I managedto work a handful of states because the re­ceiver was sens itive, albeit non selective.

Then came FM , but I won 't get into that (ifyou 're interested, see "FM /RPT" in QST be­tween June 1979 and Marc h 1990). Anyway,I was bit by the VHF /UHF bug a long time agoand I still chase 2-meter DX whenever it isavai lable.

Operating 1-15

Page 25: Packet- Speed- More Speed

Pactor-IIImpressions and Update Information one year afterBy Dr. Tom Rink, OL2FAK & Oipt.-tng. Martin etas, OL1ZAMRoentgenstrasse 36, 0-63454 Hanau, Germany

I. IntroductionPACTOR-II was introduced together with the new multimodeDSP controller PTC-II about one year ago. It includes severaladvantages that cannot be found in any other digital mode,such as a powerful convolutional code with a real Viterbidecoder to increase the robustness with weak signals, or anewly developed on-line data compression system (Pseudo­Markow Coding), which, along with the run-length encoding,roughly doubles the effective throughput. Therefore, the inter­est in the PTC-II units has always been tremendous, whichcaused the manufacturer to run out of stock several times. Upto now, more than 1000 modems are sold in all continents,about half of them in the commerc ial market, and the demandis still growing fast. Many tests have been performed by com­mercial customers, like relief organizations and even the mili­tary, using highly sophisticated equipment such as ionosphericsimulators, to compare all available digital modes concerningbandwidth, speed and robustness. They, as well as the radioamateurs who tested it with real band conditions, clearlyproved PACTOR-II to be the fastest and also the most robustdigital narrow band mode for data transmission on short waveavailable at the moment. Information can still be transferredwhen a signal is 18 dB below the noise level. This, for exam­ple, allowed a QSO between Germany and a mobile Australianstation with absolutely inaudible signals and only 16 mW of HFpower. In good conditions, PACTOR-II easily exceeds an effec­tive throughput of 1000 bits per second. The required signalbandwidth is just 450 Hz (at minus 50 dB), regardless of theactual modulation form and the transferred speed. Unlike otherDPSK modes, PACTOR-II tolerates a high frequency offsetwhen connecting and also a high frequency drift in an estab­lished link, as newly developed frequency and phase trackingalgorithms are used, which still work in the above describedborderline conditions. Initial frequency offsets of up to +/-80 Hzare automatically compensated by the PTC-II without the needof any user access. This process can be observed on the multi­color tuning display. For operating PACTOR-II, high tuningaccuracy or frequency stability of the transceivers are hencenot required. Detailed information on the technical basics, thePACTOR-II protocol and the PTC-II hardware can be found ina four part's series on PACTOR-II, which was published in theJanuary to April 1995 issues of the Digital Journal.

The PTC-II is much more than just a multimode modem fordata communication on short waves. It features the powerfulprocessor MC68360 which also includes 4 SCC's implementedas a RISC system. Four different communication channels arethus simultaneously supported by the unit. The 60 MHz versionof the special DSP 56156 is used for the HF port and also formany other tasks, such as audio processing, etc. The firmwareis always being expanded in order to improve the system andto add new modes and functions. All new firmware releases areavailable for free, and can be downloaded from the SCS mail­box, which is available 24 hours a day at (+49)-6184-900427.Since the PTC-II came into the market about one year ago,many new features have been implemented , like a new and forDSP operation optimized CW-decoder, the above mentionedaudio DSP filter, an automat ic output power control in order tomatch the needs of the current link, and a complete remotecontrol capability for Icom, Kenwood, SGC and Yaesu radiosusing the special transceiver control port. This last mentioned

1-16 Chapter 1

feature, for example, allows one to scan pre-defined frequen­cies without the need of an external computer. All transceivercontrol commands as well as the required handshake proce­dures are automatically generated by the PTC-II. As the volt­age levels are compatible in most cases, an additional inter­face between the PTC-II and the transceiver is usually notrequired.

A lot of new features will still be added this year, for examplePacket Radio, a complete host mode and the support of addi­tional modes, like FAX and SSTV. Explaining the details of allpossible applications would surely exceed the range of this arti­cle, therefore we have to restrict to three of the novelties, whichwere already added last year: The audio processing function,the transceiver control option and the enhanced CW operation.

II. DSP Audio Filter OperationIndependent of its function as a DSP multimode controller, thePTC-II can be used for special processing and filtering of anyaudio signal. It is thus also a comfortable stand-alone DSP fil­ter, suitable for SSB operation, CW listening, and many othertasks. The high processing power of the PTC-II has proven tobe very advantageous for this audio filter function. In com­parison with the usual simpler and cheaper DSP audio denois­er units, much more computing effort can be used to obtain anoptimum result. The commands controlling the DSP audio filterfunction can be found in a special sub-menu called 'Audio'.

The AF signal is supplied to the PTC as usual via the HF radioconnector, and thus no changes of wiring compared to the 'nor­mal' RTTY/AMTOR/PACTOR operation is required. Theprocessed and filtered AF is available at another pin of thatconnector and additionally at the Mini-DIN-connector (8-pinTRX-Remote-Control). All functions of the Audio sub-menu thatevaluate the AF input signal, use a 4-stage signal level match­ing (22 dB control range) for the 16-Bit AID converter, in orderto keep the quantisation effect as low as possible and to allowa large effective dynamic range. The PTC-II therefore adjustsitself automatically to the average signal level delivered by thetransceiver. The maximum level of the AF output signal is +/­500 mV. For the first test, a 600 Ohm earphone can be con­nected directly to either of the two AF outputs. Nevertheless, asmall AF amplifier with volume control is recommended to beused, which enables comfortable speaker operation as well asthe connection of any kind of earphones. Such a unit, whichadditionally provides remote controllable, bi-directional digitaland analog user ports, will be available as an option later thisyear (see below). As the unregulated supply power of the PTC­II is also available at the Mini-DIN-connector, the unit candirectly be connected there, and does not need any additionalwiring.

Here is a short explanation of the most important commands ofthe Audio sub-menu. 'Notch' activates the automatic N-timesnotch filter. All systematic signals are heavily attenuated. Theused algorithm leads to considerably less signal distortion ofspeech signals compared to simpler DSP notch filters. 'Peak'activates the automatic N-times peak filter. This may be con­sidered as a phase-linear auto-correlation filter, of a very highorder and large dynamic range. This filter enables slow CW­signals within the SSB bandwidth to be found that are BELOW

Page 26: Packet- Speed- More Speed

the level discernible by the human ear. The filter algorithm putsa very narrow band filter on all systematic signal components.Uncorrelated noise is heavily attenuated . The active mode isalways displayed on the dot-matrix display of the unit as well ason the connected terminal. The command 'CWfilter', activatesthe CW filter, using a center frequency and bandwidth that canbe freely defined. The filter is designed as FIR with a linearphase-change, so that even with a bandwidth of 30 Hz it doesnot ring. The transfer function is not designed for maximumslope steepness, but a shape leading to a signal easily read­able for the human ear, and the best obtainable signal to noiseratio.

III. Transceiver Remote ControlThe commands required for the transceiver control functioncan be found in the sub-menu 'TRX', which is entered using the'TRX' command (without argument). This command may alsocontain an argument, where all applicable commands from theTRX-menu are allowed . In this case, the PTC will carry out thecommand without switching into the TRX sub-menu . The con­trol commands can thus considered to be 'fed through' .

Example: 'TRX Frequency 14079.0 <Enter>'

changes the frequency of a connected transceiver directly to14079.0 kHz - without having to enter the TRX-menu. In the fol­lowing, we introduce the most important commands of theTRX-menu:

The 'Channel' command allows up to 16 channels to bedefined . Every channel consists of a channel number, its fre­quency (in kHz), the scan status and an optional short com­ment. Information on the current entries can be obtained usingthe 'List' command. Such a frequency list could look like the fol­lowing example:

CHANNEL-LIST: Ch Frequency (kHz) Scan Comment========

1: 14079.000 YES DL2FAK Main QRG on 20 m2: 14077.000 NO Test QRG3: 3584 .000 YES DL2FAK on 80 m

The 'Channel' command (without argument) behaves similar tothe List command. All user defined channels are listed. If theChannel command is followed by ONE argument , consisting ofa number between 1 and 16, the PTC switches the connectedtransce iver to the frequency of the given channel. If, forinstance, the command 'CH 3 <Enter> ' is given, then (con­sidering the above example) the transceiver would be switchedto 3584.000 kHz. The definition of a channel is carried out byputting two or three arguments after the 'Channel ' command .

Example: 'c 10 14076.5 EA5FIN STBY FREQUENCY <Enter>'

defines the frequency 14076.5 kHz as channel number 10, withthe comment 'EA5FIN STBY FREQUENCY'. The commentdoes not contain essential information and may be omitted .The frequency input is always in kHz. The decimal point afterthe megahertz position is optionally allowed (e.g. '14.076.50').The last decimal point is processed as a kilohertz decimalpoint. There are up to three positions allowed after this decimalpoint. A frequency accuracy of 1 Hz is thus anticipated, which ,however, is not supported by some transceivers. The frequen­cy input 0 kHz erases the channel from the frequency list.

The 'TYpe' command is used to set the transceiver type for theconfiguration of the PTC-II interface . There are up to threearguments allowed. The first one defines the transceiver (cur­rently Icom, Kenwood, SGC and Yaesu, but this list will beextended). The second value is the baud rate. When usingIcom equipment, the PTC requires a number as the third argu­ment (maximum 2 figures) that represents the transceiver

address number. With Kenwood and Yaesu equipment, thethird argument is the VFO number (A or B) that shou ld beaddressed by the PTC. The 'Scan' command has two differe ntfunctions : An argument 1 or 0 switches the scanner on and off,respectively. It is thus the 'main switch' of the scanner. If, as argu­ment, the word 'Channel' (minimum abbreviation: 'C') follows, achannel number of the frequency list can be entered to toggle thescan status between 'YES' and 'NO' for the respective channel.This feature hence allows a channel to be scanned or skipped.The 'DWell' command sets the dwell time of the PTC scanner oneach channel in 100-ms steps. A dwell time of 30, for example,means that the scanner will pause on each channel for exactly 3seconds. The 'Wait' command defines the time (in seconds) thatthe scanner waits at the respective channel after a disconnectoccurred before the scanner starts again. The 'Offset' commandapplies a frequency offset to EVERY channel of the channel listbefore it is sent to the transceiver. The valid range is -5.000 to+5.000 kHz. This allows the PACTOR mark frequency to beentered, even whilst in SSB mode. If. for instance, low-tones areused (1200/1400 Hz) and USB position, the transceiver is set on14077.60 kHz in order to transmit the mark frequency of14079.00 kHz. As the transceiver displays the frequency of the(imaginary) carrier, the frequency of the audio-mark-tone (1400Hz) must be added to the carrier frequency in USB, to get theactual mark frequency.'If, on the other hand, any mark frequencyis taken from a BBS list, the mark tone frequency must be sub­tracted in order to find the correct frequency to tune the SSBtransceiver to. If the offset value is defined as -1.4 kHz, the PTC­II carries out the required frequency correction for the mark fre­quency automatically. It is thus only necessary to enter thedesired mark frequency, and the correct offset is automaticallyapplied. With regard to the first above mentioned example, onecan give the command 'F 14079.0 <Enter>'. The PTC-II thenswitches the transceiver to 14077.6 kHz, which automaticallyleads to the required transmit and receive mark frequency of14079.0 kHz.

The 'Down' and 'Up' commands allow the microphone down or upkey to be activated (simulated by an FET switch), which is con­nected to the corresponding pin of the HF transceiver socket. Thisway it is possible to adjust the frequency without accessing theserial interiace. As argument following the command, a numberbetween 1 and 60000 is entered. This represents the number ofkey pulses initiated by the PTC-II. With no argument, a singlepulse is given. 'Ptime' sets the time (in milliseconds) for these upand down keying pulses. For example, a Ptime-value of 50means that the respective switch in the PTC-II is closed andopened for 50 ms per pulse.The commands 'Dump' and 'Transfer' are used to send an ASCIIstring or HEX dump to the transceiver. This way, any commandor remote control string supported by the transceiver can bepassed through the PTC-II, for example from an external controlsoftware.

A description of the full TRX command set as well as more detailson the above mentioned commands can be found in the updateinformation file, which is available at the SCS phone mailbox or inthe Packet Radio and PACTOR network.

IV. Improved CW OperationThe CW routine is implemented using a highly sophisticated DSPtechnique. The demodulator utilizes the much talked about tech­nique of the auto-correlation filter in the Audio menu (see above).This enables even weak signals to be reliably detected withoutany tuning problems. (An ideal filter for a CW signal with a speedof 60 cpm exhibits a % bandwidth of only 20 Hz that - with con­ventional methods - requires extremely exact and stable tuningfor good results.) The auto-correlation method is also the basis ofthe AGC used in the CW demodulator, with a dynamic range ofapproximately 40 dB. The AGC allows a constantly good receiveperformance, independent of the audio input signal level.

Operating 1-17

Page 27: Packet- Speed- More Speed

For easy and fast accessing operation, the CW terminal offersthe use of a number of so-ca lled hotkeys: Pressing the Break­In key twice switches between direct transmission of the key­board input (immediate transmission mode), and a delayedswitch over (delayed transmission mode) . This delayed switch­ing allows text to be 'typed ahead ' whilst reading the other QSOpartner's transmission. The text in the buffer can then be trans­mitted by pressing the Break-In-Character once. The transmis­sion is then only blocked again when no characters aretransmitted for 6 seconds. This way, the user may continue towrite after the buffered text is sent , without having to press anyothe r key. The renewed blocking of the transmitted text (it isthen being redirected into the buffer instead of being immed i­ately transm itted) is shown by the PTC with the message ">>>"written into the Echo-Window. After switching to the CW ter­minal, one is always in the direc t transm ission mode, and theautomatic speed adjustment mode is act ivated . '<CTRL-F>'switches between automatic and manual RX speed adjustment(fixed speed mode) . On switc hing betwee n these two possibili­ties, the present spee d is take n without change . Thus the auto­matic adjus tment can be left on for a while to detect the correctspeed, then '<CTRL-F>' can be pressed to keep this setting.'<CTRL -U>' ('Up') increases and '<CTR L-D>' ('Down') decreas­es the deco der speed by 1/16 of the actual value. This is im­po rtant when the automatic RX speed adjustme nt is turned off.The operation with a fixed speed has definite advantages inweak signals and in signa ls with heavy fad ing. The decode rwithstands speed errors of 40 percent without any problem.Hence even with a fixed speed sett ing, virtua lly no readingerrors are found. The CW speed, regard less of whether it wasautomatically detected or manually set, is shown at the dot­matr ix-display.

V. A Look to the FuturePacket Radio and the host mode are certainly the next featuresto be added to the PTC-II. For the Ham Radio exhibition inFriedrichshafen/Ge rmany, which takes place in June this year,the 1200/2400 Baud as well as the 9600 Baud plug-in modemsare planned to be avai lable, together with the corres pondingfirmware update. At the same time, the above mentioned Re­mote-Control-Ampl ifier-Unit (RCU) for the PTC-II will be avail­able. This sma ll box is con nected to the PTC- II using the 8-pinMini-DIN con necto r. It has two major features: At first, it pro­vides an 8 W audio-amplifier with volume control, speaker andheadphone connector to suppo rt and simplify the audio denois­er operation with the PTC- II. As a second feature, it provides amixed-mode user-port interface to the PTC-II, accessible fromthe local terminal as well as from each com munication channel(HF or Packet). It consists of eight digital inputs, eight digitaloutputs, eight analog inputs and eig ht analog outputs.Connected to any imaginable application , remote-control orremote data acqu isit ion is possible using a subse t of user­friend ly commands. A detailed descr iption of the RCU will bepublished separately.

(Note : please see the editorial comment regarding Pactor-II inthe Last Word column-de N2HOS)

1-18 Chapter 1

Page 28: Packet- Speed- More Speed

I'm a'self-made advertising pariah. I have not been approached todo a product review since I submitted scientific evidence, a fewyears back, proving that on RTTY, a PK232 was only slightly moreefficient than my dog Brownie who learned baudot by sleeping inthe furnace room next to a Teletype model 15.

I don't like to work on computers. It has been months since I lastpulled the cover on my 486. The anxiety I associate with trying toget WINDOWS to talk with all the ports and interrupts so that amouse, modem, sound card, FAX, cd rom, printer, joystick and apair PK232s will allwork has me stressed to the point where themere sight of a phillips screwdriver gives me diarrhea. It's no won­der that the UPS delivery guy thought I was acting a little strange.All that was visible to him was a mailing label indicating that thepackage came from HAL COMMUNICATIONS but to my, moreexperienced eye, what it said, DANGER, THIS PACKAGE CON­TAINS A POISONOUS REPTILE!" My P38 HF RADIO DSPMODEM had arrived.

Installation instructions: Turn off the PC and stuff the board into anyempty slot. Naturally, the pint-sized fan that my good friend JeffFlashnerd mounted on my processor chip to solve the meltdown­lockup problem stuck up into the only open slot and the P38, notone of your whimpy little serial port boards, needed the whole slot.The first sunshine in my otherwise rainy life was, after shuffling thedeck to get an open slot, the computer still worked! Unless yourrig cannot use open collector switching for FSK keying, no switch­es or jumpers have to be configured to get the board to work.

While this is not about software but trying to talk about a softwaredriven device without mentioning software is like describing a fishwithout mentioning water. One unique features of the P38 is thelack of E-PROMs. The basic run-time software for the micro­processor is downloaded each time you initialize the control p.ro­gram - it's just a file. This should significantly reduce the handlingand distribution overhead of software upgrades, allowing the

The HAL P:;8and other thoughtsby Hal Blegen , WA 7EGA

2021 Smythe Road· Spangle, WA 99031

The HAL P38: Digital Modes on a card

DSPnewest version can be downloaded from aBBS or an FTP site on the Intemet.

HAL's software runs under 'plain-vanilla,DOS but for mouse-bigots there are severalauthors offering WINDOWS compatible,P38 software. The only compatibility prob­lem I found was that different authors usedunique names for the run-time files andsome renaming was required (RAGCHEW

worked fine until I installed EXPRESS at which point I had to rein­stall RAGCHEW to get it to work again).

It was refreshing to run the install program and not have the screenall cluttered up with dire warnings about insufficient free RAM andEMM386 incompatibility. The programs don't require much diskspace and, best news yet, THE P38 DOESN'T NEED EITHER ANINTERRUPT OR A COMM PORT! The hex memory address for1/0 is user selectable so you can avoid conflict with common cardslike Sound Blaster or the ever-popular Dental-Surgery adapter. Iinstalled it and it worked, the first time!

The interface to the radio is dead simple. In recognition of the factthat most HF operators aren't smart enough to run AFSK withouttransmitting 5 khz worth of birdies, the P38 comes with a built-inFSK keying line. On most radios, unless you want to runCLOVER, all you need is FSK, Audio in, and PD. I'll probably godown in flames for saying so, but the average human whoragchews and works a little DX needs CLOVER like a guy in rub­ber life raft needs a chainsaw. CLOVER is for adults.

Noticeably absent was any sort of scope driver output. The P38operating screen displays a set of bar graphs for real-time tuningthat takes about 10 seconds to confirm what a scope shows at aglance. Solid state tuning indicators annoy me so much that I haveresurrected an old Flesher TU-170 just to drive a tuning scope.

The P38 does CLOVER, AMTOR, PACTOR, ASCII and BAUDOT.It doesn't do the G thing (You know, the one that was named by Dr.Ruth?). At Dayton somebody asked Bill Henry why the board did­n't do PACKET. He just laughed and pointed out that, "The P38was designed to work on HF."

PACTOR on the P38 was another happy experience. In fact, ifthey leave things along and don't invent 10 more modes to confusethe issue, FEC PACTOR may well replace BAUDOT. It's just amatter of getting a few of us old die-hardsto buy some new equip­ment. With 5 watts out, I linked with a ZL (no timing problems onmy elderly ICOM 751 using the default parameters) and once Ichanged from 250 to 400 hz bandpass, the P38 switched to 200baud which boosted the throughput to about 10 pages of copy perminute and made an instant convert out of me. When I went toSSB filters at 2.8 khz, the DSP filters completely ignored anotherstation that was in the passband even though the interfering signalwas pumping my AGC to levels above the linked signal level!

A rumor about the P38 that peaked my interest enough to buy (Ipaid -no cozy deals), had to do with some supposed compar­isons to HAL's, top-of-the-line ST-8000. The P38 was supposedto be "only 1 db down" from the 8000. I didn't know what thatmeant but that I thought I would find out.

Error correcting modes give perfect copy under varying conditions.The down side, since only one device can control a link, is that get­ting any useful data in real-time comparisons between boxesrequires more than just a couple of radios. Rather than spend a lotof time with multimode boxes in ARQ LISTEN modes I figured thatoverall demod effectiveness was unchanged on all modes exceptCLOVER so I went back to the old standard, RTTY-BAUDOT to domy comparisons.

Operating 1-19

Page 29: Packet- Speed- More Speed

On VOLTA contest weekend, I connected a HAL ST8000, an AEAPK232, a DOVE-TRON MPC1000, and the P38 to a single,buffered audio source and, armed with VCR to record audio, wentin search of the crummiest signals I could find. No surprise. TheST8000 still wins. The P38 was an improvement over my early­serial-number PK232's and held it's own against a 15-year-oldDovetron. On over-the-pole, warbly dx signals (20 meters) therewasn't enough difference between the 8000 and the P38 to costme a single contest OSO. At the point where both units startedtaking hits, the ST8000 usually recovered several characterssooner than the P38 but I was impressed.

Next, I dropped down to 80 meters. Yeah, I know, NOBODY runsRTTY on 80 meters. On the low bands, propagation does somefunny things . On an otherwise loud signal , the mark and spacecan fade separately. There is also a lot of pulse stretching thatcan be a challenge . I remembered that Mike, N7RY maintains arig on autostart at 3612.5 and I started playing his message-of­the-day file. The MPC-1000 (an AM unit) and the ST-8000 (run­ning FM) never missed a beat. The PK232 fumbled now andthen but the P38 might as well have curled up next to the furnaceand gone to sleep. Not good.

I would have expected HAL COMMUNICATIONS to know allabout selective fading and multipath so Monday morning I was onthe phone -amazing, straight through, first time, to Bill Henry.Since I've heard a lot of variations on the l'll-look-into-lt routine. Ididn't expect very much. When I saw the E-mail saying the engi­neering department had more-or-less had confirmed my findingsand were working on the problem, I was sorta impressed . Whenabout a month later a new version of the code showed up on theinternet, I was really impressed .

There is a point to be made here that does­n't fit the format of a product review. As faras the ability to put copy on your screen, Idon't see a lot of practical differencesbetween multi-mode boxes, what you'rebuying is just different bells and whistles .You might as well just look at the price.What I do think is important , is the generalattitude that I found when I called the man­ufacturer. "Yep," they said, "We agree withyou and we can fix it." What a concept!

Nope, the P38 still cannot out perform theST8000 , especially on 80 meters, but then,what kind of dummy would actually expecta $400 board that comes with free softwareto beat a top-end, mil-spec demod that sellsfor $4000. Right now, the P38 DSP is anevolving product. The current version per­forms very well in its market niche and, as itlooks to me, the hardware I am now usingwill be able to utilize future evolutions with­out having to disman tle my computer everytime I upgrade. I like that.

See ya on the air, Hal WA7EGA

1-20 Chapter 1

Page 30: Packet- Speed- More Speed

Fuji-OSCAR-29

• Satellite Summary

Name: Fuji-OSCAR-29 aka Fuji-2 and JAS-2Callsign: 8J1JCSNASA Catalog Number: 24278Launched: August 17, 1996Launch vehicle: Japanese H-II NO.4Launch location: Tanegashima Space Center of NASDA, Tanegashima Island, JapanWeight: 50 kgOrbit: Polar LEO (Low Earth Orbit)Inclination:Size: 44 cm wide x 47 cm highPeriod:

Features:

,:: BBS Message System (digital store-and-forward)::..:: Analog Communications Transponder,~ Att itude Control~ Digi-Talker:__Testing of newly developed solar cells in space

Beacon (100 milliWatt) Telemetry Format

435.795 MHz - CW435.910 MHz - PSK digital - Digi-Talker

Digital Transponder - Mode JD (1 Watt)

_. Uplinks: AFSK (FM) 1200 bps, AX.25, Manchester EncodedG 145.850 MHz=:J 145.870 MHz (the only 9600 bps uplink frequency)~ 145.890 MHz~ 145.910 MHz

Downlink: BPSK 1200 bps or FSK 9600 bps- 435 .910 MHz (also Digi-Talker frequency)

Analog Transponder - Mode JA (1 Watt)

Uplink: 145.900 - 146.000 MHzDownlink: 435.800 - 435 .900 MHz

• References

-= Steve Ford, WB8IMY, "JAS-2 In Orbit!," QST, Oct. 1996, p. 94..=Fujio Yamashita, JS1UKR and Hideo Ono, JA1BU, "JAS-2 Comes to Life as FO-29 ," The

AMSAT Journal, Vol. 19, No.5, Sep/Aug 1996, p. 1. Operating 1-21

Page 31: Packet- Speed- More Speed

~WinAPRSTM

Windows Automatic Position Reporting SystemA Windows™ Version of APRSTM

Mark Sproul, [email protected]

Abstract

Keith Sproul, WU2Zksprou [email protected]

WinAPRS is a Windows version of the popular APRS, Automatic Position ReportingSystem. WinAPRS is fully compat ible with APRS TM, The DOS vers ion, andMacAPRSTM, the Macintosh version . Due to the larger amounts of memory available inthe Windows operating system, WinAPRS, just like MacAPRS has many additionalfeatures not available in the DOS version.

WinAPRSWinAPRS is growing rapidly. Just like APRS and MacAPRS, the users are findingmore and more things to do with this technology. We (Bob Bruninga and the SproulBrothers)are committed to keeping the on-air protocols the same and are working withmany different groups to expand and add many different capabilities to the APRSgroup of programs. One of the recent developments along these lines is a largeinterest from several National Weather Service groups across the country.

WinAPRS uses the exact same map files as MacAPRS, and will also use the map filesfrom DOS APRS. Most of the source code of WinAPRS is the exact same code asMacAPRS, so it has been around for a few years, and has been thoroughly tested.See the discussion below about the development system used forMacAPRSlWinAPRS.

WinAPRS is a full Windows-95 32-bit app lication that follows the Windows UserInterface Guidelines. It runs under Windows-95 and Windows-NT, and will run underWirldows 3.1 and 3.1 .1 if you have the Win32 DLLs installed that allow Win95applications to run under the older versions of Windows.

1-22 Chapter 1

Page 32: Packet- Speed- More Speed

History of APRS

1992APRSTM was first introduced by Bob Bruninga, WB4APR, in the fall of 1992 at the ARRLComputer Networking Conference in Teaneck, New Jersey. [1]. We, (Mark and Keith)were at this conference and saw Bob's program. Keith commented that he wanted todo some of this, but when we asked how much a GPS (Global Positioning Unit) cost,we got an answer of $3000!. We decided to wait.

1993APRS started gaining popularity. There were several articles in different magazinesand many new uses for this growing technology . The article that caught a lot ofattention was about using APRS to track the football from the Naval Academy to theArmy-Navy game in Philadelphia. [2]

1994In the fall of 1993, just about a year later, Keith started working on MacAPRS. [3] Hecontacted Bob Bruninga in February of 1994 and went to see him, with a workingversion of APRS that ran on a Macintosh . This version had many enhancements overthe basic APRS features, including Call sign look-up from CD-ROM, and multiple mapsopen at the same time.

When Bob introduced APRS, all of his maps were made by hand! Keith, having hadexperience in college doing Cartography programming, refused to do maps by handand did all of the maps for MacAPRS using USGS (US Geological Survey) map data,available on CD-ROM. Soon after Keith's visit to Bob, he started using the USGS CDstoo. This improved the map quality greatly.

1995By this time, APRS, and MacAPRS were becoming very popular and the uses of thistechnology had expanded much beyond the original concepts. The APRS programshave been used for Fox Hunting, Balloon Tracking, Weather Networks, DX Clustermonitoring, and many other applications. [4] [5]

At the Dayton Hamvention in April of 1995 Mark and Keith presented more and moreof the fancy capabilities of MacAPRS. During 1995, we were invited to give talks atother hamfests and clubs in the New York/New Jersey/Connecticut area. During thistime, one of the more common questions was "... do you have a WINDOWS version? ..."

One of the more 'popular' features was the fact that MacAPRS did not really have anylimitation as to the number of points that could be in a map. The DOS version, whichwhen it first came out, was limited to 1,500 points had been upgraded so that it couldhandle 3,000 points , But the typical MacAPRS maps STARTED at 10,000 points, withsome maps as large as 300,000 points. Other features that people were interested inthat were not in the DOS version were the interface to the many different types of call­sign databases on CD-ROM.

Operating 1-23

Page 33: Packet- Speed- More Speed

At the Dayton Hamfest, we started getting more and more pressure from the HamRadio community to do a Windows version. This PRESSURE got really severe at theARRL DCC in Arlington, Texas.

When Keith got back from the ARRL DCC in Texas, we had long talks about doing aWindows version. Mark made the comment:

"I have never had so much peer pressure in all of my life... "

At this time, several critical items came together. CodeWarrior, the developmentsystem that the Sproul brothers used for MacAPRS came out with support fordeveloping Windows programs on the Macintosh. Mark Sproul, who is portingMacAPRS to Windows finally succumbed to the pressure from APRS users. Whenthese things happened, we determined that it was realistic to port the alreadydeveloped Macintosh code to Windows and decided to do a Windows version ofAPRS. On September 15th, Keith went to down to see Bob Bruninga to discuss doinga Windows version. On September 16th, the following announcement was put up onthe Internet:

MacAPRSTM for Windows(WinAPRSTM)

Automatic Position Reporting System for Windows

September 16, 1995NORTH BRUNSWICK, NJ: Mark Sproul (KB2ICI) and Keith Sproul (WU2Z) authors ofMacAPRSTM, the Macintosh version of Bob Bruninga's (WB4APR) popular packet radiomapping system announced today that they will be porting their Macintosh version toWindows. This will be the official version and has the backing of Mr. Bruninga. Thecurrent plans are for beta release by Christmas 1995 and for the final release to be atthe Dayton Hamvention in May of 1996.

APRS is a multi-faceted system used primarily within Amateur Radio for tracking manydifferent types of things. APRS is used for tracking Weather, for tracking moving cars,boats, weather balloons, and many other things. It can also be used as GraphicsInformation System for many different aspects of Amateur Radio.

The original version of APRS was developed by Bob Bruninga, WB4APR, to run underDOS and was introduced at the 1992 ARRL Computer Networking Conferences.MacAPRS was released at the Dayton Hamvention in 1994.

The Macintosh version is written entirely in C and will port easily to Windows. Keithand Bob have worked hard at keeping the two versions compatible and by using all ofthe C code already developed for the Macintosh version, it will ensure completecompatibility on the Windows version. In addition, the two versions will use the exactsame map file format so all of the wonderful maps that the Mac users have will beimmediately usable by the Windows version.

When asked about future plans, Mark said, "When we finish with the Windows version,we are planning on doing an X-Windows version as well."

1-24 Chapter 1

Page 34: Packet- Speed- More Speed

October 14, 1995One day less than one month after deciding to do WinAPRS, we had the mapsdrawing on a Windows computer and put screen-dumps of these maps up on the Webfor all to see.

- file Edit jiettings M8P Qispl8y Window

. ... ..y

December 22, 1995As promised in the original announcement, we released WinAPRS before Christmas.This release was to about 20 people.

January 28, 1996We released a public beta version to the ham radio community. We showed WinAPRSpublicly for the first time at the Wharton Hamfest near Chicago, Illinois.

May 1996Again, as promised in the original announcement, we released WinAPRS version1.0.0 at Dayton Hamfest 1996! This release had more features in it than we originallyexpected to have done at this time.

Operating 1-25

Page 35: Packet- Speed- More Speed

Development System of WinAPRS and MacAPRS

METROWORKS CODE WARRIORThe Development system that we have been using for MacAPRS is Code Warrior byMetroworks. This development environment is a full C/C++ development system forthe Macintosh. MacAPRS was written entirely in straight 'C', with no C++ at all.In September 1995, Metroworks added the capability to compile code and createexecutable files for the Intel processors. You still have to write the code for theoperating system that you want, i.e. it will NOT take the Macintosh program and simplyre-compi le it for Windows. You MUST write Windows code for the Windowsapplications and Macintosh code for the Macintosh applications. However, theroutines that are not machine dependent end up being exactly the same.

What we have for done for the MacAPRSlWinAPRS system is to create two differentapplications that use most of the same code. For example, doing the math for drawingmaps from a map file is the same no matter what platform it is on. Similarly, decodingdata from a TNC is the same, etc. The source code that is different mostly involves theuser interface.

All of the source code is written on the Macintosh. It is then compiled on the Mac. Thenthe executable file is transferred via TCP/IP-EtherNet to the Windows computer. TheCode runs on the Intel processor, but the source-level debugging is done on theMacintosh via the network.

The source code for the entire MacAPRSlWinAPRS project is written with what iscalled CONDITIONAL COMPILE flags . This means that a specific section of sourcecode mayor may-not get compiled, depending on what flags are set. We haveMacintosh Flags, Windows Flags, and several other internal flags. The objective of thesystem is to have as much of the code to be common, i.e. compiled in ALL cases, andas little as possible to be specialized code, i.e. compiled ONLY for Mac, or ONLY forWindows. By doing this, we have a much easier system to maintain, and a much morecompatible system across different platforms

X-APRS, APRS for X-Windows (UNIX)At the Dayton Hamfest in May, we had a SUN workstation running a very preliminaryversion of X-APRS (X-Windows is the Graphical User Interface for UNIX computers).This too is being done with the conditional compiles described above. Doing thedevelopment this way allows us to use code that has been around a long time that hasbeen fully tested, thus speeding up development time. We hope to have X-APRS outsometime next year. (1997)

FUTUREAPRS, MacAPRS, WinAPRS and X-APRS are continuing to evolve. These programshave proven themselves to be useful in many more applications than originallyimagined. This type of system is a system that takes full advantage of the technologyavailable only in portable radio communications and cannot be replaced with theIntemet.

1-26 Chapter 1

Page 36: Packet- Speed- More Speed

References

[1] Automatic AX.25 Position and Status ReportingBob Bruninga, WB4APRAmerican Radio Relay League, 11th Computer Networking ConferenceTeaneck, New Jersey, November 7, 1992

[2] Up Front In QST, December 1994, p 14

[3] MacAPRS, Automatic Position Reporting System, A Macintoshversion of APRS,Keith Sproul, WU2Z and Mark Sproul, KB21CIAmerican Radio Relay League, 13th Digital Communicat ions Conference,Bloomington, Minnesota, August 19-21, 1994. pp 133-145

[4] Advances in APRS TechnologyKeith Sproul, WU2Z and Mark Sproul, KB21CIProceedings of the 1995 TAPR Annual MeetingSt. Louis, MO, March 1995, pp 55-59.

[5] Graphical Information Systems and Ham Radio(The Future of A.P.R.S. Technologies)Keith Sproul, WU2Z and Mark Sproul, KB21CIAmerican Radio Relay League, 14th Digital Communications Conference,Arlington, Texas, September 8-10,1995. pp 108-117

Internet ResourcesWeb sites with APRS Information

http://aprs.rutgers.edu/APRS/http://www.tapr.org/tapr/html/sigs.html

Operating 1-27

Page 37: Packet- Speed- More Speed

s:!Automatic Radio Direction FindingUsing MacAPRSTM & WinAPRSTM

Automatic Position Reporting System

Keith Sproul, [email protected]

http://aprs.rutgers.edu/APRS/

Abstract

Radio Direction Finding has been around for almost as long as radio itself. Doppler­based RDF systems have been around for quite awhile too . In the recent past, peoplehave developed computer interfaces to Doppler-based RDF systems. APRS has theability to display the RDF information on maps, giving the user a graphical way to viewthe RDF patterns.

Over the last few years, the call sign databases available on CD-ROM from severalcompanies have become more and more sophisticated. There are also databases ofcommercial frequencies and locations ava ilable.

Most of us involved in Amateur Radio have experienced situations where we need totrack down the cause of an unwanted radio signal, i.e. stuck microphone, improperlytuned equipment, or even a jammer.

With all of the available technology, we should be able to develop a system that zerosin on a location and automatically shows us the possible transmitters in the area.

Computerized Radio Direction Finding

Doppler RDF units have been around for many years. Several years ago, peoplestarted trying to get the output of these RDF units to feed directly into a computer. Oneof the early versions of this was simply a method for reading the status of the LEDs onthe RDF unit via a computer interface. Later on, these interfaces became moresophisticated. The current RDF units have serial ports that report not only the direction,but also signal strength indicators. The direction vectors are also reported in muchhigher accuracy resolution.

This year at the Dayton Hamfest, Agrelo Engineering introduced the DFjr. This unit is acomplete computerized RDF unit. During the development of this unit, Agrelo workedwith the developers of APRS to ensure smooth operation of their unit and the APRSsoftware.

The 'normal' mode of operation of the DFjr is to have it in a car for doing RDF work.However, this unit also can be configured to be hooked up to a TNC so that each time

1-28 Chapter 1

Page 38: Packet- Speed- More Speed

it hears a signal on the frequency it is monitoring, it will transmit the RDF informationover Packet, using APRS protocols.

,.:Agrelo DFjr, Computerized Doppler RDF Unit

Computerized RDF and APRSAPRS will take the output of the RDF units and display the information on any of theAPRS maps. This gives you a geographical representation of the RDF data. If youhave more than one RDF/APRS station participating, then you can- get real-timeintercept vectors. The first picture below shows WinAPRS and the vectors from a DFjr.The second picture below shows MacAPRS and two stations reporting RDF vectors.

i/i

!• N2AIG-7

,_ 0

.. .._.-

...)

; ,. 0-

r

../i

iJ .,

)

~.- .__. ; ~....... ...

/,<

....""-... ../,1"'<'<'-.0

.•..-p --------..........__•

....T!'"• • • - ..... .~.....::.. ,

-;."

r r

",

.- -, '~ ""' - " '':'''-- . - ._ '- -._'-.- .... _:. .; ' -

Operating 1-29

Page 39: Packet- Speed- More Speed

Washington st"

.I"t

'---=-- '

-- ----

'. ....

.. r'.'.

'\,---:----+--------.....-~~RDFl

i..r'__.~....t- .

-- ----r ' .- -I "--'

..'--

----1-.

--- ---' r : .

1\",

I I I I

Combining RDF and Call-sign Databases

Once the RDF information lets you know the area of interest, you can find all of thestat ions in the area with the help of the call-sign databases on CD-ROM, MacAPRSand WinAPRS can search through the database and show you all of the stationslocated in that general area. This is done via a database containing the latitude /longitude of all of the post offices in the US. Some of the CD-ROMs are starting to addthe Zip+4 latll on to their databases. The Buckmaster CD was the first to do this. (This,alone, makes their CD one of the best available for this type of use).

The user can then search for all of the call signs reported to be in this area. The usercan select how big of an area to search. The initial search is done on the latllon of thezipcode. This is done for speed. Then, once this group of data has been selected, it isfurther enhanced using the Zip+4 data, if available . The chart below shows theinformation obtained from the Buckmaster Hamcall CD.

The table below shows one page of approximately 110 people found within a 1 mileradius of the intersection point shown above. Realize that this is the FIRST pass basedon the 5-digit zipcode. The table shows the actual distance from the intersection of theRDF vectors to each station based on the its zip+4 lat/lon. If the CD-ROM databaseyou are using has the Zip+4 location data, you can double click on each one of the

1-30 Chapter 1

Page 40: Packet- Speed- More Speed

stations in the list and it will show you exactly where that person lives on the map .(Within the accuracy of the Zip+4 system which is generally about 1/2 block).

.DO~:o;: · : ' ·:: .:C:·.: <.:.: .,. =::=..;:::.,··oc ..:'...:.:.::... :=~: :. ~:;::·· --,·r" . :.": " Selected Call Silln list .. ....... --"_..- _.- ._. . . .- _.. ·91~

!&J.l LC F i r s t Na-e Las t Harae Stree t Cl t y S t Z i p OOB L ie Issue lie Exp l refr~tVOlh'"'Oi s t

KB70AT N Jef fr ey L Pu l l en 4002 S 170 th Seattl e ~A 98 188-32:54 19:54 1017 199 10820 200 10820 200 Ki ng I 0 .2 ~K87CH. P Eric M E.... y 550 5 S 152nd Apt 3 4Tukwl 10 ~A 981 88 -78 15 19090708 19930928 20030028 206 Ki n9 I 1.3

K870TF P James L Qu inton 3705 S 172nd Sea Tac ~A 98188- 3028 19490721 199 20002 20020002 206 K i ng I 0 .4

K87RJA T Steve P Olson 3749 S 194th Sea tac ~A 98188-:5300 197407 1:5 1992 1229 2002 1229 200 K i ng I 1.:5

K87AYJ . P Howard T ~ew ..II"" 15325 Sunwood 8~ "BJOli!i I a ~A 98 188-5720 19410929 199407 16 20030 119 206 K i n9 I 0 .3 \~K87SAQ T Ther esa A Kenn edy 37 14 S 152nd 27 'Iukwr l o ~A 98188 195 8070 6 19930 119 20030 119 200 K ing I 0 .1 gKB7TTn T Uaughan F Phi I pot 40 11 S 152 St Tuk un l o ~A 98 188-2231 19240913 19930 41 3 200304 13 200 K in9 i 1.1 '"~KB7U'n T Zi to Joan Hallstr om 17047 35 t h Ave S Sea t ac ~A 98188-3608 19330820 19930504 20030504 206 King I 0 .4 :>

KB7VTO N Gr eQOry 5 Berglund 37:54 S 172nd Seatac ~A 981 88- 3027 1904 0:50 I 19930921 2003092 1 200 K i n9 I 0 .3.:~:;~KC7AHT G Noncy B Sc h I rnme I tfton 04 5 S Center 260 Seat t j e ~A 981 88 195202 15 1111140 104 20040 104 206 K i ng I 0 .1 ,"

KC7AVZ T Jos on E Parvu 3738 S 164th Sea Tee ~A 981 88-3040 1117802 12 11194 02 15 200402 15 200 Ki ng I 0 .5 .1',':

KC7AZX T Thomasson 10432 32nd Ave S Seatac ~A 98188- 302 1 19570805 111940222 2~""

Mar gare t K 200 Ki n9 I 0 .7 :;;rKC'7AZY T Norma H Thonloss on 10432 32nd St Seatac ~A 98188 193 10214 191140222 20040= 206 Ki ng I 0 .1 .:;;

KC7BNS T Don i e'! n Fisher 17343 Mil i tory Ad ssec tcc '·JA 981 88- 3651 19590727 19940329 20040329 200 K i ng I 0 .3 <,v,: ,;"CUS T Sidney u Anderson 3408 S 175th Seatt le ~A 981 88 -3662 192 8 1125 19940531 2004053 1 206 K in9 I 0 .5 ·:;i:KC70BN P Dougl as ~ Hans 15037 45 Th Ave S Fukwr l c ~A 98188 19470825 199407 05 20040705 206 King I 0 .1

:'j

':;,KC7FBP T James E Mi tche l l 1723 0 n il i tory Ad $eat tI e ~A 98188-3048 19710318 . 199:50:50:5 200408 17 206 K ing I 0 . 2

);~KC7HPt1 T Todd J Rogers 3054 S 150th Sea t tJ e ~A 98 188-2 107 1118305011 191141221 2004 122 1 206 King I 1.4 )-KC7HVZ P Lloyd L Cr ab t r ee 1862 5 39th Ave S Sea tac ~A 98188-5007 1935 1112 19950900 2004 1229 206 Kin9 I 1.1 ':.;KC7 1GO T Diosdado A Alejo 35 11 S 160th St B1 Seat tl e ~A 98 188- 2634 197401 27 19950 117 200:50117 206 King I 0 .7

KC71GA P T i na M Pa t ton 173 4 1 32 Ave SA l02ieo to e ~A 9818 8-4430 1958 1029 19115020 1 200 50 117 206 K i n9 I 0 .6 .:;;KC7 1UC T Mic hae l S Ward 3200 S 170 t h St 40ESea tt I e ~A 98188-4072 19070319 199:502 09 200:502011 206 Ki ng I 0 .7

".:,:,

KC7KLlJ T Binyamin Y Lev ine 1080 1 33rd Ave Sea Tac ~A 118 188-3132 19461105 111950425 200:5042:5 206 King I 0 .5 ~ ;;j ,

KC7LON T Diane L De nee-r ee- 4024 F S 158th Se c t t l e ~A 991 88 19480202 19950522 20050522 206 K i ng I 0 .1 :'~<::KC7rFC T Quent in I.J Aapp 38 00 S 179th St Sea t tl e ~A 98188-4107 19301015 199507 14 200:507 14 206 KinQ I 0 . 7

~KC7tUJ T Jana E Ward 3200 S 176 th St 40ffi ea t tI e ~A 9818 8-4072 1970 1208 191150819 200:508 19 206 K i~ I 0.7

\!l

Conclusion

This kind of Geographical Information System has many potential uses within the ham­radio community. This type of search is not limited to ham-radio databases only. Thereare databases available that contain similar information about commercia ltransmitters. These databases not only include latitude and longitude, but also actua lfrequencies etc. Over a year ago, when I started doing demonstrations of this type ofcapabil ity, many people wanted to have it immediately. However, at that time, thecomputerized RDF units where either done as build-it-yourself kits, or for the most part,were just not available. Now, with the DFjr from Agrelo Engineering, this type ofautomatic RDF Unit is easily available and affordable. This type of technology willallow us to do semi-automatic Radio Direction Finding for such things as trackingdown interference problems etc.

Operating 1-31

Page 41: Packet- Speed- More Speed

References

[1] MacAPRS, Automatic Position Reporting System, A Macintoshversion of APRS,Keith Sproul, WU2Z and Mark Sproul. KB21CIAmerican Radio Relay League. Digital Communications Conference.Bloomington, Minnesota, August 19-21,1994. pp 133-145

[2] Graphical Information Systems and Ham Radio(The Future of A.P.R.S. Technologies)Keith Sproul. WU2Z and Mark Sproul, KB21CIAmerican Radio Relay League, 14th Digital Communications Conference,Arlington, Texas, September 8-10, 1995. pp 108-117

Internet ResourcesWeb sites with APRS Information

http://aprs .rutgers.edu/APRS/http://www.tapr.org/tapr/html/sigs.html

Agrelo DFjr

http://home.navisoft.com/agrelo/ae.htm

1-32 Chapter 1

Page 42: Packet- Speed- More Speed

PACKET AND INTERNET

James Wagner, PhD - KA7EHK31677 N. Lake Creek Drive Tangent, OR 97389

(541) 928-7869 [email protected] http://www .proaxis.com/-wagnerj

ABSTRACT

Debate is one of the interesting aspects of the packet bbs system. One of the recent debate issues is reallyquite important to all of us. It concerns the question of bbs mail forwarding by methods other than theham RF network. Whichever side proves to be "right", (and it is possible that both may be right) , theanswers to this debate will have an impact on all packet users . KEY WORDS: BBS, NETWORK,FORWARDING, INTERNET

INTRODUCTION

A major debate has raged in the packet bulletin board system over the last year or so. This debateconcerns the use of alternate forwarding methods for moving packet messages. In particular, the use of(commercial) internet has been a major issue.

The issues in this debate warrant presenting here because they do represent ideas which are having , andwill continue to have, major impact on the future of packet radio .

THE ISSUES, SIMPLIFIED

One viewpoint in this debate argues that the use of alternate forwarding methods (telephone, internet,etc) will result in a deteriorating RF network. The logic is that when alternative methods are used, thereis no longer pressure to upgrade existing networks, fix broken ones, or maintain the ones we have . Theargument continues with the idea that a network which is allowed to deteriorate will not be there whenemergencies arise . And, when emergencies arise, it is also likely that portions of the internetinfrastructure will fail. The result will be failure of the ham packet system to perform in emergencies.

The other viewpoint argues that the ultimate responsibility of bulletin board operators is to move "mail".If the ham RF infrastructure is not capable of moving that mail, this argument continues, then they havethe responsibility to find some method which will allow the mail to be moved . If that method happensto involve the telephone system or internet (ie, "wire line"), then so be it.

Network Quality

There are a number of factors which combine to represent the quality of a packet network. Of course, notall users have the same idea of quality . None the less, there are a number of general things:

Operating 1-33

Page 43: Packet- Speed- More Speed

a. links need to be reliableb. unexpected disconnects should not occurc. reasonable throughput must be available most of the timed.hardware is physically reliable

When any of these factors gets worse, usually we perceive the quality of the network to be "worse". Anetwork can remain physically the same, but be perceived as deteriorating because it is not able tohandle an increasing message load, for example.

Thus; for a network to remain as a high quality one, just keeping the hardware working is not enough . Ifthe network cannot support increased demands placed on it, then it is not doing its job. Unfortunately,many users consider bbs forwarding to be the culprit , rather than the "miner's canary" which warns us ofimpending difficulty. To such users, its probably just fine that their local bbs forwards over internetbecause it makes things seem to work better.

It is also an unfortunate human attribute that often, the quiet wheel does not get improved . If bbs sysopsuse other methods for forwarding messages, then a visible pressure for network improvement goes away.

Mail Movement

For many bbs sysops , mail movement is their entire reason for participating in ham radio. And, forsome, at least, ham radio is just another access method for their bbs. When you have bbs sysops of thissort, what technical methods are they going to choose for linking? Certainly, the most familiar ones . Ifthey are more comfortable with wire-line , that is what they will choose.

While "Clover", for example, may do an excellent job of handling messages via HF, it is probable thatmore bbs sysops are familiar with wire-line modem technology than they are with Clover. So, is it anywonder that non-ham message passing technology is frequently the method of choice?

It is also likely that arguments that "it is not ham radio" will be quite ineffective. It is probable that manyof the bbs sysops in this category do not have a big stake in ham radio and that this argument results in abig "so what?"

On June 30, 1996, WORLI wrote the author: "Yes, there is still a small amount of traffic handled viasatellite, HF digital modes, and long haul vhf/uhf links. In fact, all PRESENTED traffic is easilyhandled. However, very little traffic is presented to the radio network; it is instead moved viacommercial networks. When I first started speaking out on this issue, 18 months ago, about 50% of thelong haul traffic was still being carried by radio. That percentage has now reached 0."

The practice of wire-line forwarding is actually having a far bigger impact than it might seem. Whenstrategically located bbs, in widely separated locations, forward almost instantaneously to each other,bulletins arrive in the other area more rapidly and get distributed to those stations which do use RFforwarding. Since the messages are already there when attempted by traditional means, they arerejected. This "capture" phenomenon results in an artifically forced reduction ofRF forwarding .

1-34 Chapter 1

Page 44: Packet- Speed- More Speed

CONCLUSIONS

It seems probable that both sides are "correct". The sad part is that the cases of network deteriorationseem to be growing more numerous with the use of forwarding over wire-line . It is also likely, however ,that the cause-and-effect is not so clear. Bbs forwarding is moving off the RF network because it isdeteriorating in many places and the deterioration is accelerating because there is less reason to keep itup. In other words, it is likely that these two effects go hand-in-hand and neither is the cause of theother.

What does seem fairly clear is that bbs sysops who move their forwarding off the RF network are notdoing hams much of a favor. This is, in fact, one area where users can apply pressure, encouragement,and support to sysops. Hams with solid HF experience can help a sysop to set up a reliable forwardingsystem using Clover, Pactor, Amtor , or packet. Hams with good VHFIUHF network experience can helpto make the bbs VHFIUHF packet equipment as good as it can be.

Likewise, bbs sysops AND users can apply pressure and offer assistance to their local ham clubs andpacket organizations, and to node operators . Make sure that network capability improvement is planned ,that groups involved in packet networking get together and figure out what is needed on a regional basis,and make sure that there is a solid commitment to carrying through on those plans.

Failing all else, bbs sysops in some areas of the country are rerouting messages to avoid forwardingthem to bulletin boards which use wire-line forwarding . This is certainly a drastic measure but it is oneof the few ways available to avoid the capture effect previously described. As unpleasant as this measuremay be, the health of our network(s) may depend on it!

BIOGRAPHY

James Wagner is an electrical engineer & programmer employed by Kalatel Engineering in Corvallis,OR. His work is in the area of embedded controllers and design of system components for the closed­circuit video security industry . He has a BA in Physics from Oregon State Univ, an MS in ElectricalEngineering, also from O.S.U, and a PhD in Electrical Engineering from Colorado State University. Hehas been employed by Tektronix and by the College of Oceanography of Oregon State University. Hisinterest in ham radio began in the 1950's but did not actually get a license until 1979. He has been theadvisor to the Oregon State University ARC and the node-op of their packet node. He is the author of"The Amateur Packet Radio Handbook".

Operating 1-35

Page 45: Packet- Speed- More Speed

From the UniversitatKaiserslauten Web Page

,

SUNSAT: AMicro SatelliteUnderConstruction InSouth Africa

SUNSAT is a 60 kg, 45 by 45 by 62cm micro-satellite being designed, built andtested by 22 M.Eng. students at the ElectronicSystems Laboratory in the Department ofElectrica l and Elec tro nic Engineering atStellenbosc h Universi ty.

Deta ile d de sign sta rte d in January1992, led by Computer and Control Systemlecturers. SUNSAT was originally designedfor a sun-synchronous-type orbit on the Ariane4 Helios mission, which is ideal for the mainimaging payload. However, when launch costsbecame prohibit ive, alternatives were sought.

NASA scientists have learned a greatdeal about the earth by detailed studies of themagnetic field and the gravitational field, andhav e arr anged for the Dani sh Oers te dmicrosatellite to be launched as a secondarypayload on a USAF Delta II from VandenbergAir Force Base on the P91-1 Argos missionin Janu ary 1996. NASA and Ste llenboschhave now agreed to carry SUNSAT into thesame orbit instead of a cou nterweight forOersted.

In exchange for the launch, Sunsat willcarry a precis ion GPS receiver and a set ofLaser retro-reflectors. These wi ll enableNASA to study fine orbital perturbations forgravity field recovery, and for cross verifica­tion of GPS and NASA's laser trackingnetwork.The orbit will be the same as Oer­sted, namely polar, 400 by 840 km. The equa­torial crossing will initially be approximately15:00 UTC, and drift an hour earlier everyseventy days.

SUNSAT is a comp lex microsatellite.Its developers expec t they will not have timeand manpower to utilize all its possibilit ies,and hope that other amateurs and universitieswill become interested in using it once it is

1-36 Chapter 1

fully commissioned. Sincethis is their first satellite,they recognize this may takemany months to get right.

The Amateur Radiocommunications payloadcomprises a packet radio ser­vice, a 2 m band "parrot"speech transponder, and aMode-S transponder.

The use of an imag­ing system necessitates atti­tude stabilization. Coarse at­titude stabilization will be bya gravity grad ient boom andby mag netorqueing and isimproved by small reactio nwhee ls du rin g imaging.Continuous spacecraft atti­tude sensing is provide d bymagnetometers, sun sensors,visible wave length horizonsensors, and a star sensorprovide I miliradian accuracy when imagingfrom the sun-synchronous orbit. The averagepower of 25 W enables images of South Af­rica and elsew here to be taken on a daily ba­sis for real time downlinking.

Availability of excellent linear SiliconCCD sensors able to operate in the visual andnear-IR band led to a 3-color sensor systemwith bands similar to SPOT 4 and LANDSAT6. These permit "biomass" production moni­toring, which is of continuing interest in a"water-short" country like South Africa, forexample. A linear CCD sensor with 3456 pix­els of 10.7 microns spacing was also chosen.The optical assembly is mounted in a tubewhich can be rotated forward or rearward forstereo images.

The communications payload providesduplicated synthesized transm itters and re­ceivers for the 2-m and 70-cm Amateur Ra­dio bands and nearby frequencies.

A 23-cm receiver will operate as a fastuplink, or be coup led to the S-band downlinktransmitter to provide a straight-t hrough tran­sponder.

The high resolutio n data will be trans­mitted in real time via the S-Band downlinkto reception stations at Stellenbosch andJohannesburg . Small-area images stored in theRAM disk can be down-linked at much lowerrates. For example, a 40 kbyte image cover­ing a 4 km x 4 km area can be downloaded at9600 baud in about 100seconds. The SUNSATteam plans to be able to supply such images

Page 46: Packet- Speed- More Speed

on request to amateurs once the satellite is fullyoperational.

At 800 km altitude, a 5 deg elevationfootprint has a diameter of 5,080 km and whichspans 45 deg in longitude. Radio range variesfrom 800 to 2,800 km compare d to the geosta­tionary range of 36,000 km. Data communi­cation with 10 watts or lower powered trans­mitters and a dipole antennas is practical, per­mitting data interchange with low cost terres­trial transceivers . Since large quantities of datacan be stored in the satellite, global data trans­fer is possible. AX25 data protocols will beused to ensure error-free operat ion.

The 5 watt EIRP S-ba nd downlink willproduce a 14.4 dB SIN ratio in a 40 MHzbandwidth at 2000 km slant range for a 4.5 mdiameter parabolic dish that has a 100 degreeKelvin receiv ing station which is current whatis being planned for Stellenbosch. By addingan L-band receiver and appropriate switch-

ing, a transponder capable of I MByte/s with2 m diameter grou nd stations can be imple­mented . Application of the syste m for Ama ­teur Radio gateway service is possible.

The Amateur Radio payload definitionwas approved at the SA-AM SAT Spacecon 91Conference. Store and forward digital packetradio will be provided , including 1200 baudAFSK for compatibility with terrestrial equip­ment common in South Africa.To provide suf­ficient uplink chan nels, one of the 2 m bandreceivers has four IF sections displaced in 25kHz steps, and connected to 1200 baud mo­dems. Three 9600 baud modems compatiblewith the G3RUH standard are carried, and canbe switched to various receivers and transmit­ters. Both the 2 m up/down and 2 m up/70 emdown options will be included, together withfull bulletin board facili ties. The AMSATPacsat Standard Protocols can be supported .

The 2-m and 70-cm downlinks can beswitc hed to 10 watts output, producing a 0.5~V sig na l (50 ohm) at 43 5 MHz and1.5 ~V signal at 145 MHz with 0 dBi receiveantenna at full range. This power level willbe used over critical areas to provide signalto noise ratios approac hing 15 dB for easyrecep tion. At other times the power will bereduced.

A 2 m "parrot" mode repea ter is in­tended especially for Novice category users(under the age of 16). Up-linked speech willbe digitally stored and re-transmitted on thesame frequency. Novice school users will thushear the re-transmission and know that theyare getting through. The need to learn andapply opera ting protocols will definitely beexper ienced!

Stay tuned to the AMSAT News Ser­vice (ANS) bulletins for more informationabout SUNSAT.

Operating 1-37

Page 47: Packet- Speed- More Speed

By Donald Cox, AA3EK

ACARS: Packet lor AirplanesA new digital communication mode for commercial airliners and business jets

Dyou remember the flight attendant on your last airli ne

I • fli ght reading off the li st of connecting gate numbersas you prepared to l and? That information wasmost likely passedto the flight crew using a new digi­

tal communication link called ACARS-Aircraft Communicationsand Reporting System. As comple x as it may sound, it 's simi lar inmany respects to amateur packet radio.

Until a decade ago, alm ost all radio links between the groundand commercial aircraft used voice communication to re layweather, position, aircraft performance and departure/arrival re­port s. The airlines supported the creation of an organization,Aeronautical Radio Inc (A RINC), to run a network of ground sta­tions wi th which to communicate wit h airl iners anywhere in thewor ld. VHF, HF and satelli te links were used to stay in touch wi thaircraft.

The expansion in air traff ic, reduction in the size of fI ight crew s,and the automation of aircraft cockpit and contro l systems gener­ated the need fo r a faster and more effic ient system for handlingcommunicat ion with aircraft aloft. The resul t was the creation ofACARS , adigital data link system designed to use existing ground

stati on and aircraf t radio equipment, and to enhance air-gro und­air comm unication.

The Heart of ACARSOn board the aircraft , the heart of the ACARS system is a com­

puter that receives data fr om the cockpit or other aircraft systemsand passes it to the VHF radio for transmi ssion to the ground. I n thecockpit, the pi lots have their own terminal and keyboard to disp layincoming messages and to compose outgoing requests and reports.A printer provides the crew wit h a hard copy of the messages.

The ACA RS computer can also be linked to other on-boardavionics systems. When tied to these systems, the ACARS systemcan automatically pass aircraft and engine performance data to theground. Many airl inesand aerospacefirmsrouti nely coll ect perfor­mance data that ispassed to their engineering staffs for analy sis andearly identification of problems or perform ance improvements.

ACARS messages currentl y carry weather report s, arrival anddeparture time report s and aircraf t system data dumps. Plans callfor ACARS-deri ved systems to be used eventually to handle mostof the air traffic control messages. Ins tead of a contro ller on the

ACARS Hardware, Software and ResourcesDecoders and Demodulators

A number of commercial demodulators are available to re­ceive VHF-band ACARS messages , including severa l under$99. All use the audio output of any VHF receiver/sca nner ca­pable of covering the 129 to 132-MHz AM aircraft bands.

AEA ACARS, produced by Advanced Electronics Applica­tions, is an IBM PC-compatible serial-port decoder and DOS­based software system . A 386 or higher CPU is recommended.A software-only versio n of the system is also available to usewith AEA PK-900 or DSP-232 multi-mode control lers, or AEAFAX decoders . The software includes options to suppress thedisplay of messages received with parity errors , write messagesto disk, print them, change screen colors , and review and man­age log fi les. A 132-page manual documents ACARS messageformats and the extensive reference information necessary forunderstanding the messages .

The AEA ACARS serial decoder and software package isavailable from many AEA product distributors, or from AEA, POBox C2160, 2006 196th St SW, Lynnwood, WA 98036; tel 206­774-5554. AEA ACARS software for PK-900, DSP-232 or AEAFAX product owners is also avai lable separate ly.

Lowe Airmaster 2.0 is distributed by Lowe Electronics anduses the same basic DOS software as the AEA system, but adifferent seria l-port decoder. Like the AEA software , messagefields are broken apart and displayed separately and there aremessage printing and logging options. It comes with a 24-pagemanual.

Universal Radio's M-400 is a stand-alone hardware readerthat decodes ACARS messages and displays them in the raw,transmitted format. Their M-1200 PC card and M-8000 hard­ware systems also receive ACARS and display it in the samemanner. Universa l's product line also includes the ACT-1 PCseria l port decoder and DOS-based software system . It comes

1-38 Chapter 1

with a 33-page manual and a copy of the book UnderstandingACARS . The ACT-1 software includes several message dis­plays and filtering options and displays message fields individu­ally. Universal Radio M-400, the M-1200 PC card, ACT-1 , andthe M-8000 are available from Universal Radio, 6830 Ameri­cana Parkway, Reynoldsburg , OH; tel 800-431-3939.

Books

Understanding ACARS by Ed Flynn, third edition, publishedin 1995 by Universal Radio. 92 pages.

Complete technical details on the ACARS message formatand system architecture is documented in a series of ARINCtechnical publications , available for $65 to $80 from ARINCDocument Section M/S 5-123, 2551 Riva Rd, Annapolis , MD21401-7465 ; tel 410-266-4117; fax 410-266-2047 .

ARINC Characteristic 724B-2 : Aircra ft CommunicationsAddressi ng and Reporting System (ACARS), November 1993.A follow-up to ARINC Characteristic 597-5 cover ing second­generation ACARS units.

ARINC Characteristic 620-2: Data Link Ground SystemStandard and Interface Specification, December 1994. Com­plete technical reference on ACARS message formats .

ARINC Characteristic 635: HF Data Link Protocols, August1990. Technica l reference on HF ACARS system design, mes­sage and data link formats.

Cyberspa ce

Web sites with ACARS info rmation include http ://barbie.epsilon.n l/-bart, http://web.inter.nl .netlhcc/Hans.Wildschut, http://www.u-net.com/-morfis/acars.htm andhttp://www.arinc.com.

Page 48: Packet- Speed- More Speed

3C01 POSWX 1722/18

.,5D5P5R5U5Y5Z7A7B10-4951545780-89C1F3H1QOQ1Q2Q3Q4Q5Q6Q7QAQBQCQDQEQFQGQHQKQLQMQNRARB

Table 2ACARS Message ExamplesMessage label: RA Translation: United flight 837 fromGATE ASSIGN Newark to San Francisco will arriveUA837 EWRSFO at gate 82 at 1844 GMT. The aircraftGATE 82 FREQ 129.5 is to turn off its auxiliary powerEON 1844 APU OFF unit upon arriva l and contact United

operations on129 .5 MHz.

Message label: 80 Translation: Delta flight 1722 fromBlock ID: 7 Msg. no: 1553 Dallas to NY is over Nottingham, MDFlight ID: DL1722 (On) at 2214 GMT, flying at 37,000Message content: feet , with 13,200 Ib of fuel on board, airKDFW/KLGA .N922D temp of -61 deg, wind from 246 deg at/POS OTT /OVR 2214/ALT 25 kt, speed Mach 0.746 with a smooth370/FOB 0132/SAT 61 ride and clear sky./WND 246025/MCH746/TRB SMOOTH /SKY CLEAR

Translation: Departure message fromContinental flight 1854, which departedDetroit at 1646 GMT for Newark.

Translation: Message from groundwith list of gates for connecting flights atChicago . The flight will arrive at gateM20.

Translation: weather advisory messagefrom ground to American Airlinesaircraft N41063 near Washington, DC,advising of thunderstorms near LongIsland.

Message label: QFFlight ID: C01854Message content:­DTW1646EWR

Message label: 40Flight ID: 0023Message content:­ARR ORDARR M20 BAG lABDEST GATEDALLAS/FT W H6DENVER H15KANSAS CITY K15LOS ANGELES H1

Message label: C1Flight ID: DDAA 2Message content: -10012 FROM " D39 13 13AN N41063/GL IAD-WX RADAR SHOWSTSTMS DVLPG ALNGTHE SOUTH SHORE OFL.I . ALSO ONE LG CELLTO FL600 OVE HUOMOVG SE-20KTS .

Table 1ACARS Message TypesLabel Message Type

Data transcei ver autotuneATIS requestTemporary suspension of ACARSAircrew initiated position reportWeath er requestAircrew revision of previous ETA or dive rsion reportAir line designated downlin kAircrew initiated engine data/takeoff thrust reportAirc rew ente red miscel laneo us messageUser-defined functionsGround GMT request/GMT updateAircrew initiated vo ice contact request/voi ce go-aheadAlternate aircrew init iated position reportAircrew addressed downlinksPrinter messageDedicated transceiver advisoryOptional aux iliary ter minal messageLink testDeparture/arrival reportETA reportClock update advisoryVoice circuit busyUnable to deliver uplinked messageVoice to ACAR S channel changeo verDelay messageOut/fuel reportOff reportOn reportIn/fuel/destinat ion reportOut/fu el/destination reportOff/ destination reportOut/return in reportOut reportLanding reportArr iva l reportArrival information reportDive rsion reportCommand/response uplinkCommand/response downlink

Message label: 40 Translation: aircraft advising groundFlight ID: 0023 B maintenance that the tray tab le in seatMessage content: 7B is broken .HX TO ORD 28JUU1745ZPOSN-07B KWRD-SEATARM-6847 007H TRAYTABLE WILL NOT RETRACT/

Depending on where you li ve, a simple omnidirectional antennasuch as a ground plane or J-pole is perfectl y adequate. An outdoorantenna is best, but even an indoor antenna can bring you pl enty ofaction. Somehavealso reported successusing only telescopin g whipsand rubber -duck antennas, but they're the opti ons of last resort.

Yo u can expect to recei ve ACARS tran smi ssion s from high-

Operating 1-39

gro und talking to the flight crew via voice radio, the controllerwill tr ansmit a digital message to the flight crew to tell them toclimb or turn . M ost new commerc ial aircraft are equipped w i thA CARS systems, as well as a sig nifica nt number of pri vate busi­ness j ets. A i r l i nes are alrea dy experi ment ing wi th the use of in ­flight satel lite communication links and HF li nks to carry AC A RSmessages. Thi s would all ow gl obal coverage and would reduce theneed fo r an extensive network of gro und statio ns.

You Can Receive ACARSMost commerci al recei vers and scanners capable of receiving

the VHF aircraft-band A M tran smi ssions can recei ve the 129 to132-MHz ACARS frequencies. A growi ng number of modernA mateur Radio 2-me ter tr anscei vers can also cove r th i s band.A CARS f requenci es in the US in clude 131.550 , 130.025 , and129.125 M Hz; in Europe, 131.725 and 131.525 ; in As ia, 13 1.450;and A i r Canada has used 131.475 MHz. The 13 1.550- M H z fre­quency i s considered the primary channel in the US. ACARS-l iketran smi ssions on HF have been observe d on 6.646, 10.027 and13.339 M Hz on fl ig hts trave ling the North A tlant ic ro utes.

VHF ACARS recepti on i s easiest for most hams. A ll you needi san A CARS demodulator conn ected to the speaker output of yourscanner or transcei ver and your computer. Th e demodulator tr ans­lates the ACARS audio tones in to data your computer can under­stand. Speciali zed softwa re runn in g on yo ur PC displays theA CA RS information on your monitor. See the sidebar, "ACARSHard ware, Softwa re and Resource s."

Message label : 5ZBlock ID: 4 Msg. no: 1846Flight ID: US0065Message content:­/ENG/0938/350/248/740/432/M19/M44/199/199/860/850/454/433/855/880/2750/2700

Translation : a data dump from USAirflight 65 showing it at 93,800 Ib of fuel,35,000 feet altitude , 248 kt speed,Mach 0.74, true airspeed 432 kt,indicated & true outside temp of -19 and- 44 degrees C. Pairs of data for theplane's two engines follow: EPR of1.99/1.99, N1 =86.0/85 .0%, EGT =454/433 deg C, N2 =85.5/88.0%,fuel flow = 2750/2700 Ib/hr.

Page 49: Packet- Speed- More Speed

AEA ACARS software decodes received data and displays theresu lts on yo ur mon itor.

altitude aircraft ISOto 200 mile s away. If you' re lucky enough tolive near an airport, you might be able to receive ACARS ground­to-air uplink messages .

Understanding ACARS MessagesThe VHF ACARS frequencies carry a large volume of traffic

between aircraft and gro und stations. The types of information inthe tran smission s varies widely . It can range from simple arrivalor departure reports , to lengthy downlinks of navigation, engineand performance dat a. Messages might include weather observa­tions and forecasts; departure clearances and checklists ; flightplans; navigation posit ions; aircraft and engine perfo rmance data ;arri val , departure and delay reports; equipment malfunction re­port s; crew reports; and connecting gate lists.

Like amateur packet, ACARS transmissions are very short,burst-type signals that last only a fractio n of a second. Each mes­sage is broken up into a number of subfields . The seven-characteraddress field contains the regi strati on number of the aircraft. Theregi stra tion number is the official number assi gned to the aircraftby the government. International agreements dictate a letter prefixindicating the country of origin. For US aircraft, all registrationnumbers beg in with "N." Shortwave radio listeners and ham radiooperators will notice the similarity between these letters and theradi o call sign allocations for the same countries . The followinggive some examples of regi stration numbers for airliners fromseveral countries: NI4245, USA ; C-FDSN, Canada; G-BNLR,Britain; D-AIBE, Germany; JAS097, Japan; HB-IGC, Switzer­land; and PH-BFH, The Netherlands.

A number of references and comput er databases have compila­tions of registration number s for aircra ft arou nd the world . These

1-40 Chapter 1

1~ II' .511 Z •• 1mI1II • I ~ .11JII77 1.- 11,17" Z .~ s:z 7 1S1II1 "II .1 •I~ 11'17 .11 I "71111 • 1., "li7iio'"I~ 11,17 .11 Z •__ • I .u: ,-.. NIl~ 11,17 •• I lU _A 7 *1> I .. ... I .,

1 l~ 11 ,17 . 2 • • •ISII C _ .7111 IP( Pll TU - 5 H!to zCl~ 11.11.. I IIISI-> 5 11 .1 I I' - l c_UII~ 11.11 .11 I """- .: K IITOIo~l~ 18.11 .11 r 1I-C11U ole 0 ZrY> I.PlIlIIl IlIl'" >.... ZIl17KZlI1'.US - ' OT

> .x_) '_'.TC Q)Wx - •• -Tl~ 18.11 .11 2 .. . IS U _" • •4.> l .lIDOIl1.,..195 18 .11 .26 2 .11653 l1li HI 2 JlIlI5 _6 ••BnII••1IA .122C246 5311946KIA

'_11 8952318CL 122 lATA HOT _UAIL! 38 .9113 - 17 .463223 9 1458 8 .2286 22 39 .1115 - 17 . 5851 23 12 6482 -

1.,...-115 18 .18 .32 2 .1l719lIU 52 I 1983 US11M1 "II 11111~ 18 .18.33 2 .1II553lM HI 3 __.....12 .829& 3D 39. 1182 - 11 .

1666 231 6 11511-19 .8296 3111~195 18.18.35 2 .. .1I31U _" 3 - L- 8_ 12h

The ACT-1 ACARS soft ware f rom Universal Radio displaysthe data in a column fo rmat.

can be used to quickl y ident ify the type of aircraft sending an ACARSmessage , the aircraft operator and other related information.

A series of two-charac ter message labe ls have been defined todes ignate the type of message being sent. These includ e a numberof fixed-format message s with key information like arrival/depar­ture times (so- called on/off time s) and fuel loaded, as well as anumber of labels with variab le formats tha t can be defined by theaircraft operators. These labe ls are cruc ial to understanding thetype of information in the message ; they are frequently the ent ireextent of the message themselves. For example, if the messagelabel is "51," the aircraft is requesting the ground to update its on­board clock. A short list of these label s is shown in Table I.

Several examples of ACARS com munications between theground and aircraft are shown in Ta ble 2. ACARS messages usea shorthand sty le of abbrevia tions and a good reference is neces­sary to fully understand the message contents .

Give It a Try!ACARS monitoring is a fun and relati vely inexpensive addi­

tion to your ham activities . By following the message traffic , you' llgai n an educational insight into airline operations, how flig hts areplanned and flown, the type s of dai ly challenges they experie nce ,and much more . Best of all, you' ll quickly appreciate the com­plexity and sophistication of toda y' s airlines and air traffic contro lsys tems.

PO Box 11130Wash ington. DC 20008-0330e-mai I74507.3446@compuserve .com

Page 50: Packet- Speed- More Speed

SCS PTC-II Multimode Controller with PACTOR-IIBy Steve Ford, WB8/MYMa nag ing Edito r

It' s astonishing to think about how farwe 've come in HF digital communicati on sin such a shor t time. Prior to 1982, Baud otRTTY was the norm. It was a gre at modefor ca sual con ver sation , but it lacked themeans to det ect errors . If the RTTY signalyou were cop ying fell victim to noise, fad­ing or int erference, the result was gibberishon your screen. (Before the PC revolution,the "screen" was oft en telet ype paper! )

AMTOR burst (pun intended) onto thescene in' 82. It gave us text that was almosterror-free, and filled the band s with thoseodd chi rp- chirp sounds . A few years laterHF pac ket came into its own . Although farfrom ide al in terms of dealing with the va­garies of HF, packet at least gave us theability to exch ang e the enti re ASCII char­acter se t, as well as binary files.

Special Communications Systems (SCS)created PACTOR, starting with PACTOR-I,which hit the airwaves in 1991 with sophis-

PACTOR-III once heard a ham describe PACTOR­

II as being PACTOR-I on stero ids. Theremay be some truth to thi s ana logy!PACTOR-II re tains man y of the featuresthat made PACTOR-I so popular: Supportof the compete ASCII character set;memory ARQ to reduce the number ofretransmission s; on-line data compressionto spee d transfe rs; automatic data-r ate se­lection; and independence from sidebandselection (it doesn 't matter if you operatewith your rig set to LSB or USB) .

What PACTOR-II brings to the part y isconsiderable. Using differential phase-shi ftkey ing (OPSK) and innova tivecon volutional coding (with a true Viterbidecoder), PACTOR-II boasts data tran sferrate s at least three tim es fas ter thanPACTOR-I. Under optimum band condi­tions, the rate could soar to six times fas ter.In term s of effective da ta rate s, that tran s­lates to about 800 bits /so PACTOR-II hasthe opti on of usin g Huffman coding for datacompressi on (the same as PACTOR-I ), orPseudo-M arkov coding (PMC) for evengreater performance and efficiency.

One of the most astonishing aspec ts ofPACTOR-II is that it achieves all thi s won­drou s performanc e while using only 500 Hzof spectrum (at-50 dB). You need fast hard ­ware and digital signa l pro cessin g (OSP)tec hniques to make all of thi s practical.That ' s why you won 't find PACTOR-II in­cluded in the current cro p of multi modecontro lle rs. In fac t, the only pl ace you' llfindPACTOR-II at the moment is in thePTC-II.

ticated data compression and the remark­able ability to "reco nstruct" cor rupte d data.Reaction from the ham community wasswift and positive . It wasn 't long beforePACT OR-I became a staple in virtually allmulti mode controllers .

BOTTOM LINEA top -of-the-line digita l box at a top­

of-th e-line price . This one's for seriousdigital enthusiasts.

The PTe-IIWhen I first eyeballed the front panel of

the PTC-II, one image came to mind: AChri stm as tree . There are no fewer than30 LEOs populat ing the panel , along with adot-matrix LED readout. They're all quiteinform ative, tellin g you whether the data isflowing without errors, which compressionscheme is in use, if you have messages wait­ing in the mailbox and so on. Inpractic e I found that I didn 't have time to fo­cus on anyone of them. Only the dot-m atrixMODE display and the tuning indicator sawmuch use. The MODE display is handy whenyou're tuning around in the listen mode. If Icame upon a PACTOR signal, the displaywould immediately indicate whether it was aPACTOR-I or PACTOR-II conversation .

The tuning indicator is a real gem! It issimply a row of dual-color LEOs with aTUNE j...EO in the center. As you tune acrossa PACTOR-II signal, for example, the LEOson the right- and left-hand ends flash red, thengreen, as you "arrive" on frequency. At thatpoint you make slight adjustments of yourVFO until the TUNE LED is centered. Witha little practice I found that I could tune in aPACTOR-II signal in a matter of a few sec­onds.

On the back panel are two packet radioports. (Yes, the PTC-II supports simul -taneousHFNHF communication-up to three chan­nels at once!) You can' t operate packet withthe standard PTC-II , though. If you' wantpacket capability, you must purchase addi­tional modules from SCS. A 300 baud modemis not yet available, but SCS offers an AFSK

Just when it see med tha t the HF digitalwaters had calmed a bit, CLOVER andG-TOR appe ared . Both modes offered su­perior performance, but ham s seemed con­ten t with PACTOR-I and were reluctant tomake a change . Even so , CLOVER andG-TOR foun d ded icated, if somewhat lim­ited, audie nces .

As we prog ress toward the end of the20th century, we have no fewer than six digi­ta l modes compet ing for the hearts andminds of HF-acti ve amateurs. If you listento the HF digital subbands for any length oftime , you' ll discover that while the vener­able RTTY is still quite popular for digitalcontest ing and OXing, PACTOR-I reign s asthe undi sput ed king of the "burst modes."

In the current "To wer of Babel" digitalclimate, will hams acce pt a seventh mode? Ifthey' re offered an HF digital mode that prom­ises the ultimate in performance, will theyflock to it in numbers sufficient to dethronePACTOR-I? That's what the folks at SCS­the creato rs of PACTOR- II- are betting on.

modem for 1200 and 2400 baud, plus an FSKmodem for 4800,9600, 19,200 and higher baudrates.The labeling of the HF ports can be a littleconfusing. There is a port labeled AUDIO andanother labeled CONTROL. The AUDIO portprovides the audio inputs and outputs , as wellas the push-to-talk lines. CONTROL refers tothe interfac e between the PTC-II and yourtransceiver's computer-control port, if ithas one. For basic operating you need only con­struct a cable for the AUDIO port, althoughyou can do some very clever things with theCONROL port, as we' ll discuss later.

Communicating with your computer is aseasy as running an RS-232 cable between thePTC-II and your computer's serial port. ThePTC-II package includes a freeware programcalled Plus'Termon a 3.5-inch diskette. I triedto load Plus'Term using its installation batchfile and had a devil of a time. I eventually gaveup and worked around the problem.

To my dismay I discovered that Plus'Termis strictly a DOS program. Asa guy who's beenspoiled by Windows, I found it painful to use.The good news is that the terminal programalready available in Windows works very wellwith the PTC-II. The even better news is thatWilfried Max, OLi XAM,hasdeveloped a full­featured Windows programspecifica lly for thePTC-II. (You can get a demo version by send­ing a formatted 3.5-inch diskette with $10 cash(no checks) to Wilfried Max, OLi XAM,Lisbeth Bruhn Str 18, 0 -2 1035 Hamburg,Germany. For more information see the SCSsite on the Web at http://www.scs-ptc.com!soft ware.htm!.)

Operating 1-41

Page 51: Packet- Speed- More Speed

The only serious weak spot in the PTC-I1package is the manual. The slender volumedoes an excellent job of explaining PACTOR­II, but gives poor coverage to other aspects ofthe PTC-I1.For example, I didn 't know that thePTC-I1could send and receive CW until I dis­covered this feature by accident while readingthrou gh the list of commands. Some of theunit' s most interesting features are describedvaguel y, leaving users very much on their own.

On the AirAfter hearing and reading so much hype

about PA CTOR-I1, I was dying to give ita try. My first problem was finding aPACTOR-I1 station. With the PTC-I1 in thelisten mode , I scanned through the 20-meterdigital subband, but all of the "burst" linkswere using PACTOR-I. I wasn 't entirely sur­pri sed . There are only a handful of PACTOR­II stat ions in the US and a limited number inEurope.

After about 30 minutes of hunting Ifinall y copi ed a link with the ON40B mail­box. As I tuned in the signal, the MODE indi­cator on the PTC- I1 flashed PT-2 . Bingo!When the station signed off, I sent a connectreque st to the mailbo x. Two seconds later wewere connected and the data stream was flow­ing nicely. The band was in poor shape, andcond ition s were worsening by the minute . Ireque sted a list of his station equipment just ashis signal took a deep fade . To my astonish­ment , the list began march ing across my moni­tor despite the fact that I could no longer hearhis signal!

Later the same afternoon I witnessed adramatic demonstrat ion of PAC TOR -I1'sability to withstand inter ference. I had con ­nected to FG5FU and was downloading in­formation when a RTT Y station fired up righton top of us. The link seemed totall y unaf­fected by the RTT Y interference . In fact , thetext continued to zip by faster than I couldread!

Of cou rse, the PTC-I1 feat ures its ownmailbo x, with a 512- kB message capacity(expandable to 2 MB with the optional staticRAM extension). It offers some versatile re­mote-control functions as well. The mailboxaccept s connection s in PACT OR-I or II, orAMTO R.

It is import ant to point out thatPACTOR-I1, as implemented in the PTC-I1, isentirely "backward comp atibl e" withPACTOR-I. That is, you can work PACTOR ­I stations and they can work you. When youconnect to a PACTOR- I1station , the exchangebegins in PACTOR-I and then automatically"upshifts" to PACTOR-I1 .

AMTOR conversations were even harderto find than PACTOR-II. I spent the better partof an afternoon searching for stations runnin gAMTOR and came up empty. That even ing Ifinall y encountered a few on 80 meters. Onceagain the PTC-I1 did an outsta ndi ng job ofmaintaining links in the face of considerablenoi se and inte rfer ence. After using the PTC ­II on both PACTOR mod es, howev er ,AMTOR see med almost primiti ve.

The PTC-I1 had a surpri se in store on

1-42 Chapter 1

RTT Y. The DSP fil te r performa nce thatproved itself so we ll on the othe r modesdid a supe rb job with RTTY. I encountereda nu mber of sta tions using RTTY on20 meters and enjoyed sev eral live con ­ve rsations (all of my PACTOR-I1 con­nect ions were with mailbox sys tems ). ThePTC -I1provided wo nderful copy even whensigna ls sank into the noi se. This unit mightbe a gre at contest box!

CW performance was abo ut what yo uwould expect. If the code was sent per­fectly , the PT C-II copi ed pe rfec tly . Ifthe fist was poor, the co py wa s poor. ThePTC -I1's tuning indicator made CW rece p­tion quick and easy. Yo u simp ly tu ne yourrig unti l the LED s begin "bouncing" fromleft to right in sy nc with the sign al.

Fascinating ExtrasThe PTC-I1includes a numbe rofbell s and

whistles that you won' t find on other control­lers.

I had to chuckle when I read that thePTC-I1can function as a DSP audio fi lter. Nokidding! The unit provides an audio output line(on the CONTROL port) that you feed to asmall amplifier and speaker. (The audio fromyour rig is already feeding into the HF trans­ceiver port.) At your keyboard you enter theAUDIO menu and select either an adjustable­bandwidth CW filter, a "denoiser" or a notchfilter. I noticed dramatic performance with theCW filter; the skirts seemed incredibly sharp.The denoi ser worked best on CW, while givinga somewhat "hollow" sound to SSB signals.The automatic seek-and-destroy notcher did afine job of eliminating "tuner-uppers."

And what about that CONTROL port? Ifyour transce iver prov ides compute r contro l,you' re in luck. The PTC -I1can use it to do oneor both of the follow ing:

• Contro l the frequency of your radio fromyour computer keyboard, or remotely fromanother station.

• Set up an automatic scan on freque nciesyou select in advance . You can enter up to 16frequency "channels" and command the PTC­II to scan through all of them. As it does so, itcomma nds your radio to jump from one fre­quency to another. For example, you can con­figure the PTC- II to scan on 16 different fre­quencies from 80 meters through 10 meters.If your friend wants to connect to your PTC­II mailbox, he has 16 different frequenci es hecan try, depending on band conditions.

No spec ial software is req uired for rigcont rol. The PTC -I1 comes preco nfigured towork with Yaes u, Kenw ood , or ICOM rigs.You simply "tell " the PTC-I1which brand youown.

DPSK requi res exceptional frequency sta­bility . The PTC -I1 includes automatic offsetand drift compensation using a tracking algo­rithm that follows the received signal.

Although the manual only mentions itbriefl y, the PTC-I1 has the abi lity to vary youroutput power automatica lly dur ingPACTOR-I1 contacts. If conditions are good ,the PTC -I1will red uce the audio level appliedto your transceiver, thereby reducing your

output. If conditions go to pot , audio levelsautomatically increase to push your radio upto full power. The default setting forthis com­mand is OFF, so you must switch it on your­self.

Speaking of transmit audio , some trans ­ceivers offer audio line-input ports on thei rrear panels (these are often labeled AFSK).If the PTC- II doesn ' t seem to prov ide enoughaudio through this por t to drive your rig to fulloutput, no problem! Just use the PSKA com ­mand at your keyboard to adjust the transmitaudio level.

Finally, the DSP architecture of thePTC-I1 provides extraordinary flexibility forthe future . SCS plans to offer free firmwareupdates that will allow the PTC-I1 to operateon FAX and SSTV. All you' d have to do isplace the software in your computer and imple­ment the PTC-I1's UPDATE command.

Is PACTOR-II forYou?

If you demand high-performanc e HF digi­tal co mmunicat ion regardless of cost,PACTOR-I1 may be your dream come true.There is no ques tion that PACTOR-I1 offerssuperb performance, and its implementationin the PTC -I1 is outstanding. Is PACTOR-I1better than the other contenders for the high­performance market-CLOVER or G-TOR?PACTOR-I1 proponents would answe r "yes ,"but CLOVER and G-TOR discip les wouldargue otherw ise, depe nding on speci fic con­dit ions. Monitor the Interne t e-mail "reflec­tors" and you' ll be treated to an ongoing battleamong those who champion these modes .

In the mean time, the average ham has yetto emb race PACTOR-I1, CLOVER or G­TOR in signifi cant numbers.They seem com­fortable with PACTOR -I, espe cia lly since itprov ides excellent performance in control­lers costing less than $300. As one fellowcommented on the air during one of my tests,"PACTOR-I already gives me the ability toswap text at rates faster than I can read ortype . Isn 't tha t enough?" Perhaps, but thebenefits ofP ACTOR-II go beyond speed.Therobust nature of PACTOR-I1 make s globalcommunicat ion possible for low-power sta­tion s, or station s with less-than-optimal an­tennas, or both.

Regardless of its technical merits, the fateof PACTOR-I1 is likely to be decided in themarketplace. If PACTOR-I1 is to gain wideacceptance, hams will have to be conv incedthat the benefit s are worth the substantia l in­vestment.

Manufac turer: Spec ial CommunicationsSystems GmbH, Roentge nstrasse 36, D-63454Hanau, Germany; tel 49-6181-23368; [email protected] ; WWW http://www.scs­ptc.com. Manufactu rer's sug- geste d retailprice, $950; SCS RCU remote control ampli­fier unit, $ I95. (PacComm Packet Radio Sys­tems Inc, Tampa, Florida, manufactures anearly identica l version of the PTC-I1 in theUS, under a license from SCS, for $995. Call800-486 -7388 or e-ma il ptc@ paccomm.com .)

Page 52: Packet- Speed- More Speed

Packet Perspective----------------- - ------ - - - - - ----Stan Horzepa, WA 1LOU"

Geminids Packet Meteor-Scatter Test Results

NotesV-vertical antennaH-horizontal antenna; the following number indicates beam headingO-no packets were receivedDX-distance in miles between the two stationsPkts-number of packets detected

December 14, 1995, Gem in ids Packet Meteor-Scatter Test6 meters (50.62 MHz)Stations Reporting Stations ReceivedCall Sign Grid Ant Time Call Sign Grid OX PktsN2HMM FN21se H 300 0N02G FN21 V 0WD4KPD FM15ml H 235 KE4ROC EM64rg 545 2WD4KPD FM15ml H 235 WMNTF EM81 445 59N9QLW 1406 KE4ROC EM64rg 491 1VA3MW FM03 H 225 02 me ters (145.79 MHz)Stations Reporting Stations ReceivedCall Sign Grid Ant Time Call Sign Grid OX PktsKBnx V 0W1AW FN31jt H 225 0WA1LOU FN3 1mp H 225 0KA2JAK V 0N2NRD FM291q H 270 0457 K4ZQX EM75ic 648W21CZ FN02x a 1242 K4ZQX EM75ic 619 2WB2NGZ 0W3ADO H 0KB4LCI FM08xi V 0KD4LCR 0W5SR DM95cf H 315 0KE6AFE 0KC7CO V 0AA8SF V 0KA8JMW H 0WE8W EN85 2145 undecodable short burns receivedWD0GNK EN34rb H 135 0

The Geminids packet meteor-scatter testwas conducted on December 14, 1995. Thiswas the second test resulting from my pro­posal s to expand the horizons of packet­so to speak-by trying something truly dif­ferent . Compared to ordinary packetoperating, you can't get more different thanbouncing your signals off the fiery trails ofmeteors!

The first test was conducted during thepeak of the Perseids meteor shower inAugust 1995. Participation was limited thenbecau se the test was planned on the spur ofthe moment. Word was spread via theInternet, but not in print. Tho se of you an­tic ipating a test announcement in PacketPerspective had to wait until the second testwas announced in December.

With the greater advance publicity, Ihoped for an improved turnout for theDecember test. Mother Nature playedhavoc with our plans , however. The turnoutwas less than expected, and the results werenot as encouraging as the August totals .(Weak-signal meteor-scatter operators alsoreport a rath er mediocre Gemi nids event.)

On 6 meters, the best meteor-scatterpropagation was achieved by Dave Gay tko,WD4KPD, in North Carolina, who receivedpackets from Ed Myszka, KE4ROC, inAlabama, 545 miles away . KE4ROC wasalso heard by Joe Fadden, N9QLW, in Chi­cago, a 491 -mile path. For sheer duration,it 's hard to bea t the 59 packets thatWD4KPD received from WA4NTF over a445-mile path .

The resu lts on 2 meters were moreimpressive. The longest-di stance reportcame from Jim Mollica, N2NRD, in NewJersey, who copied packets from Tom Cash,K4ZQX , in Chattanooga, 648 miles away .Syd Chiswell, W2ICZ, in Buffalo alsoreceived K4ZQX's packets, a 619-milepath .

There was some confusion when stationsreceived what they thought were meteor­scattered packets, only to find out that theso-called DX was coming from APRS HFgateways within their tropo spheric earshot.For example, many stations received this ora similar packet from KF0ZH:KF0ZH >APRS ,GATE * ,W ID E / V: @

141518z4447.06NI09329.48W _202/0001T025/ROOO/POOO/U-II

If you missed the asteri sk next to GATE,you thought you hit meteor-scatter pay dirt.' One Glen AveWolcott , CT 06716-1442e-mail: [email protected]

However, those who spotted the asteriskrealized that the packet was relayed by anHF gateway station , not by a shooting star.

SoapboxI saw several short burns at 2 145Z dur ­

ing the second and third quarter-minutetransmit periods, but not fast enough for thesoftware to decode. Otherwise, only localsand no monitored packets during the24-hour period (WE8W). Much to my dis ­may, I discovered that our 2-meter preampfailed several hours into the Geminids test.I wonder if anyone received our beacons?(WB8IMY, op at WIAW). I ran all nightfrom FM03 on 50.620 MHz and didn't heara thing beaming at 225 °. I suspect the snowand freezing rain didn't help (VA3MW).145.79 MHz is totally crow ded here in thePhilly-to-New-York area . Had a blast! Isuggest readers grab MSSOF42F.ZIP fromftp .funet.fi in the pub/ham/vhf-work di-

rectory . Exce llen t program for meteor scat­ter work. The AZ/EL beam heading char tsalone are wort h the dow nload! (N2NRD).Wou ld like to do some more of this . I havea 2400-baud AFSK modem that may be in­teresting to try, but I think that everyonewho is serious with this mode needs to con­ver t to PSK modems (WD4KPD). Nothing,nada, zip. A 40-foot dis h with 25 kW ERPpointed fro m Maryland toward Chicagoshowed no receive hits. OSCAR arraypointed at Atlanta and Florida showe d nohits (WB4APR, op at W3ADO) . Sorry toreport that I heard nothing this time. I didn 'thave my beam ready in time. Will try agai n(KB4LCI) . Tha nks for your efforts gett ingthis going. I do n't do a lot of VHF wor k andthis has bee n my on ly ex perie nce withmeteor sca tter. It's fun ! The ice storm did anumber on my antenna (SWR of 5: I i), so Isaw nothing overnight (AA0SM, op atWD0GNK).

Operat ing 1-43

Page 53: Packet- Speed- More Speed

On-air Measurements of HF Data ThroughputResults and Reflections

Ken Wickwire (KB UY)[email protected]

KBUY@WAIPHY

1. Introduction

For the past year a team of colleagues and 11 have been collecting and analyzing data onthe throughput and other characteristics of various ARQ protocols available to hams andcommercial users for HF work. This activity was motivated by discussions (especiallyamong hams) about the relative merits of the new HF digital modes, such as PacTOR,GTOR, CLOVER II, CLOVER-2000 and PacTOR II. Since the discussions oftencentered on throughput in various conditions, and we were already running several of theprotocols, we decided to see for ourselves. This paper describes our assessment approachand measurement campaign, gives a summary of our main conclusions, and lists somefindings worth noting before protocol choices are made and protocol performance iscompared. The paper treats the packet and TOR modes in detail. More extensive reportson CLOVER II, NOS TCP/IP and the ALE orderwire will appear elsewhere.

2. Our Approach to Throughput Measurement

The randomly varying HF "channel"; that is, the combination of propagation conditions(fading, dominant ionospheric layer, etc.), and propagated and local noise, is generallyagreed to be the worst radio channel. Over the past 20 years, powerful DSP techniqueshave been developed to tame this wild conduit and put it to work for data transmission,even when it resists being used for voice traffic. These techniques are now embodied in asurprisingly large and growing number of data transmission protocols whose performanceis often impressive by HF standards. What people mean when they say (or write in anadvertisement) that one of these protocols is better than another is not always clear,however.I

HF data transmission protocols can be divided for general discussion into two categories:those with automatic repeat request (ARQ) and those without. In some cases, the latterare called forward error correction (FEe) protocols, because they use FEC but not ARQto control errors. ARQ protocols, which are almost always combined with FEC inmodern systems, generally deliver error-free data, although there is no guarantee that thedata will be delivered quickly. Since many users (especially military and governmentalusers, and operators of forwarding stations) demand error-free transmission, ARQprotocols have come to dominate technical discussions of late. For ARQ protocols, thedefinition of throughput, for example, is relatively straightforward; for protocols withoutARQ, which can deliver erroneous data, the concept of HF throughput is more difficu lt todefine . For these reasons we have decided to concentrate on ARQ protocols in ourassessments.

There are three basic ways to assess the throughput (and most other kinds ofperformance) of an HF data transmission protocol. First, you can try to devise a

1See the Acknowledgments and reference list at the end of the paper.2Although we recognize this shortcoming. we sometimes suffer from it ourselves.

1-44 Chapter 1

Page 54: Packet- Speed- More Speed

mathematical model of what happens during data transmission with it, convert the model,if necessary, into computer code, and run the code (or your pencil) to assess (i.e., predict)performance. While this is sometimes advocated as the best approach by those too lazy,poor or otherwise constrained to try other approaches, it frequently producesunconvincing or incomprehensible results. Many believe that the modeling approach isbest suited to the design stage of a new protocol rather than to the performanceassessment stage.

Second, you can connect a pair of working systems through an analog or digital HFchannel simulator, which can be set up to accurately produce various levels of some ofthe main phenomena of the HF channel, like multipath spread, fading and noise (usuallyGaussian). Using a channel simulator with real hardware and software producesstatistically repeatable results and allows reasonable-if not necessarily convincing­comparisons of different systems operating in so-called "standard channels," namely theones whose statistics are programmed into the channel simulator. A channel simulatorcannot, however, reproduce the statistical variations in transmission quality that occur ona real HF channel; it can't faithfully reproduce those caused by non-Gaussian (e.g.,impulse) noise, intermittent and random interference by man-made signals with variouswaveforms, day-night transitions, and polar and equatorial propagation anomalies.

The third approach is through on-air measurements. This has the advantage that anyonemeasurement is in a sense completely realistic and convincing, but the disadvantage thatthe conditions in which the measurement was taken are not generally repeatable. Thismeans that producing statistically convincing assessments with this approach requiresthat a large number of measurements be made (resulting in a large sample-size) and thatattention be paid to realistic and representative path lengths, power levels, antennas,diurnal variations and the spreads (variances) of performance statistics. This takes timeand a lot of cooperation from several outlying stations.

We believe that a combination of channel simulator and on-air measurements leads to themost convincing assessment of ARQ performance in the HF bands.' The simulatorcreates repeatable channel extremes, while properly conducted on-air measurementscomprise channel conditions the simulator hasn't been (or can 't be) set for. This paperdiscusses a measurement campaign we've pursued in that belief for the past year. Weshould note that although our results allow an informative comparison of the throughputsof the protocols we've treated, the past year's measurements need to be continued tocover all seasons with all protocols and a wider range of sunspot numbers.

For our on-air measurements we try to write software that allows tests to be runautomatically, so that the mistakes that we all tend to make during manual time-recordingand data-entry can be avoided. (Sometimes-as with CLOVER and NOS TCP/IPimplementations,-protocols come with their own interface software, and we use theexisting software capabilities. That leads to some manual data logging.) So far, oursoftware has been written in C for the Macintosh operating system, but it would work(with different I/O calls) on different operating systems.

With our software, we always measure file transfer time from the start of character-by­character uploading of a file to the sending modem to the time that a "message saved" (orequivalent) sent by the receiving station arrives via the sending modem to the test

30f course, this is only true when the two approaches produce results that agree, at least qualitatively. Forsome simulator results that agree qualitatively with our measurements, see Refs . 1,2 and 3.

Operating 1-45

Page 55: Packet- Speed- More Speed

program", Waiting for a "message saved" makes the transfer times a little longer thanthey would be if a human operator (or program) at the receiving station recorded theerror-free arrival of the file. That in turn makes the throughputs slightly lower (i.e., moreconservative) than they would be otherwise, but that is a small and reasonable price topay for measurements that require only one program and no operator intervention duringtests.

In the next section we'll describe the philosophy behind our choice of equipment and filetypes for most of the measurements made during our campaign.

3. OUf Philosophy for Choosing Equipment to Use and Files to Be Sent

In choosing equipment (e.g., the KAM for PacTOR assessment) for our tests, we havebeen motivated not only by expediency (we have KAMs), but by the view thatassessments of most interest to the most (prospective) operators are those of "common"operating setups; that is, ones widely available at competitive prices, and ones that offer awide choice of operating modes and good technical support. While it is no doubt truethat some implementations of PacTOR, for example, may have higher throughput thanothers because they use AID quantization of bit energy or more advanced filtering, theyare probably not in wide enough use to be part of a "common" operating setup .Nevertheless, if we had the time and money to buy and test all possible pairings ofimplementations of a particular protocol, we would gladly do it, since performance of the"best" or the "official" implementation is obviously of interest. In the meantime, weunselfishly invite others to fill in the gaps left by our work.

Likewise, we have chosen at this stage ASCII English text files of various sizes tocompare the transfer capabilities of protocols. With due respect to the many whoprobably send text files written in other languages, we believe that sending such filesrepresents a "common" application of the HF ARQ protocols described below. It shouldbe borne in mind that languages other than English and German, and files with a non­standard distribution of characters (e.g., all upper-case characters), may benefit very littlefrom the Huffman text compression used in current PacTOR implementations.

When a protocol like PacTOR, GTOR or CLOVER II comes with defaults for some of its"protocol-tuning" parameters (e.g., GTTRIES and GTUP for GTOR and BIAS forCLOVER II), we have used these defaults. This has been based on the belief that acommon setup would not have these parameters changed. (Optimal tuning of suchparameters is an area that should be looked at, however, and a few operators haverecently started to do so.) For packet, on the other hand, we consider good values ofPACLEN and MAXFRAMES to be highly dependent on channel conditions, and wejuggled these values frequently to increase throughput in our tests (see below) .

Finally, our philosophy says that if a protocol or common implementation offers datacompression, then it should be used (if there's a choice) unless we think it mightseriously expand a file (see the section below on data compression). This means that inthe case of PacTOR we used Huffman compression and in the case of CLOVER II, weused the "PKLIB" compression (probably a Liv-Zempel-Welch variant) offered by thestandard (i.e., "common") P38 terminal software provided by HAL with the modem.

4That is, we don't include linking and "negotiation" times in our throughput calculations. Others may viewthese times as legitimate components of transfer times.

1-46 Chapter 1

Page 56: Packet- Speed- More Speed

In the next section we'll discuss some of the protocols we have assessed in the past yearof our campaign along with some more advanced ones we may get to later.

4. ARQ Protocols Developed for UF Use and Their Throughput

Table 1 at the end of the paper lists most of the ARQ protocols that are in common use onthe air". With .the exception of the last three.all are used in amateur work, and the lastthree, developed originally for military use, will probably enter the amateur world insome form in the next few years.

The table classifies each protocol according to itsmodulation scheme, signali ngbandwidth, forward error correction capability, ARQ scheme, channel rate, characterformat, compression capability and measured throughput for "standard" (mainly lower­case) English text files. The throughputs with a O-symbol beside them have beenmeasured as part of our campaign, with enough samples for statistical significance incurrent sunspot conditions. The other throughputs are based on channel simulatormeasurements. The measured throughputs for packet and the TOR modes are from anaggregate of short near-vertical-incidence skywave (NVIS) and longer one-hop skywave(OHS) paths. For CLOVER II and the ALE engineering orderwire, the throughputs arefrom only NVIS paths. (We expect to begin OHS tests with CLOVER II this summer,and to publish ALE, NOS TCP/IP and CLOVER results this year.)

It should be kept in mind that in agreement with our measurement philosophy, for ourpacket and TOR throughput measurements we have used Kantronics KAMs withfirmware version 7.1 or higher. Other PacTOR implementations than the KAM's mayyield higher or lower throughputs than ours. Note also that we have used the HAL P38for all of our CLOVER II measurements; more expensive models, like the PCI-4000 ,have the computing power to select a 16-symbol signaling set, and may produce higherthroughputs.

5. Differences Between NVIS and OUS Throughput for TOR and Packet

NVIS throughput is generally lower than throughput over "standard" one-hop skywave(OHS) paths; that is, fairly long paths on which fading (and resulting inter-symbolinterference) is relatively slight, and average signal-to-noise ratios are comparativelyhigh. In fact, one-hop skywave measurements paint a relatively optimistic picture ofwhat operators can expect in day-to-day communications over HF.

However, some protocols appear to improve more than others when you go from NVIS toOHS operations. Tables 2 and 3 below (reprinted from recent papers listed in theReferences) give NVIS and OHS throughput and other statistics for AMTOR, PacTOR,OTOR and packet. (Recall that for packet, we juggled PACLEN and MAXFRAMES toincrease throughput.)

Throughputs in the tables are in characters/sec and times are in seconds. The first columngives the average throughput and its standard deviation, the average throughput per Hertz,the standard deviation of the mean throughput and the maximum observed throughput.The second column gives the number of links and the mean and standard deviation of the

5A recent newsgroup FAQ on signalling formats lists a number of ARQ protocols in use in Europe, the CISand Asia that we never heard of, so we may bemisleading our readers with this statement. Most of theseprotocols may be rather old and inefficient, like AMTOR, but we can't be sure.

Ope rating 1-47

Page 57: Packet- Speed- More Speed

"link time." The third column gives the number of"negotiation times" and the mean andstandard deviation of the negotiation time. Link time is the time (in seconds) betweensending the link command and receipt by the program of the "LINKED TO" notification.Negotiation time is the time between sending the link command and the start of message­file transfer. In most cases there are fewer negotiation than link times because we startedmeasuring the former part way through the campaign. The fourth and fifth columns givethe means and standard deviations of the transfer time and the number of transferredcharacters.

The standard deviation of the mean (equal to the standard deviation of the throughputdivided by the square root of the sample size) is an assessment of the variability of themean itself (which has its own statistical variability). The sd_mean's in the tables suggestthat our sample sizes are big enough to give us pretty high confidence that if we collectedmany more throughput measurements under roughly the same conditions, we would notget average throughputs that differed from the ones above by more than about a characterper second.

To calculate the average throughputs per Hertz [E (tput 1Hz ) ] , we divided the averagethroughput by the average signaling bandwidth. We calculated the latter using theformula for "necessary telegraphy bandwidth" (from the 1992 Dept. of Commerce RFManagement Handbook) BW =baud rate + 1.2 x shift, where shift for most of our TORand packet tests was 200 Hz. For AMTOR, the baud rate is of course 100; for PacTOR,GTOR and packet, we used the rough average of the baud rates chosen automatically inthe PacTOR and GTOR modes and manually in packet. Our estimates of these averagebaud rates were 150 (PacTOR) , 200 (GTOR) and 200/300 (NVIS/OHS packet). Theresulting average bandwidths were AMTOR: 340 Hz, PacTOR: 390 Hz, GTOR: 440 Hzand NVIS/OHS packet: 440/540 Hz.

The majority of NVIS measurements were at 3.606 MHz LSB, with some at 7.085 MHzLSB and 1.815 MHz LSB. They were made during the winter over all daylight hours andalso in the evening, after dark; a few were made in the middle of the night. Interferenceusually prevented throughput tests from about six to ten in the evening (2300Z-0300Z) on3.606 and 7.085 MHz.

Most of the OHS measurements were at 10.141 MHz LSB. About 20% were taken at3.640 MHz, 14.075 MHz, 14.123 MHz or 18.075 MHz, all LSB. These measurementswere made during the winter and spring over all daylight hours and also in the evening,after dark. However, interference often prevented throughput tests from about six to tenin the evening (2300Z-0300Z) on 3.640 MHz. The NVIS and OHS tests covered roughlythe six-month period from November, 1995, to April , 1996.

All measurements were made using transmitter output of around 100 watts, and allstations generally used sloping longwires or dipoles. These setups can be viewed asembodying average station capabilities. NVIS paths (in New England) were from 30 to200 miles long and OHS paths (on the east coast and from New England to the midwest)were from 400 to 1200 miles long.

In discussing the TOR and packet results let's start with some observations on NVIS andOHS communications quality in general. First of all, note that we haven't collected dataon the fraction of tries in each mode that we were successful in linking, "negotiating" andtransferring a file. However, we have found that over OHS paths, the three TOR modesand packet can get files through in the absence of strong interference on most tries duringthe day. This is in contrast with our NVIS results, which showed that except during the

1-48 Chapter 1

Page 58: Packet- Speed- More Speed

mid-morning and mid-afternoon "windows," packet and AMTOR had transferprobabilities well below one.

Under difficult conditions (especially those on NVIS paths leading to marginal SNRs)PacTOR occasionally out-performed GTOR in terms of throughput, although GTOR hashigher average throughput. This seems to confirm the rumor that GTOR needs highSNRs for high performance. However, this "role-reversal" happened much lessfrequently over OHS than over NVIS paths.

In the early evening on both NVIS and OHS paths, there was sometimes increasedinterference on the frequencies we used. During these periods of interference it was rareto see a file transferred. (An automatic link establishment (ALE) system, such asprescribed in MIL-STD-188-14IA, could probably have found a frequency withoutinterference.)

Table 2. Statistical Summary of NVIS Throughput Data

Mode E(thruput) No links No neg tms Etxfer tm) E(No_char)sd(thruput) E(lnk_tm) E(neg_tm) sd(xfer_tm) sd (No_ch r)E(tput/Hz) sdO_tm) sd(neg_tm)sd_mn(tput)max tput

AMTOR 5.20 cps 226 70 473.5 s 2358.11.13 cps 3.02 s 82.4 s 234.0 s 974.70.015 cps/Hz 3.16 s 30.1 s0.08 cps6.33 cps

PacTOR 17.83 cps 344 95 146.1 s 2452.75.50 cps 5.44s 38.7 s 90.0 s 1110.10.046 cps/Hz 8.39s 22.7 s0.30 cps25.10 cps

GTOR 23.52 cps 335 76 120.0 s 2531.710.06 cps 5.54 s 58.7 s 95.8 s 1580.30.053 cps/Hz 10.30 s 30.9 s0.55 cps44.12 cps

packet 5.68 cps 197 119 556.7 s 2484.93.53 cps 8.73 s 102.7 s 367.6 s 1043.10.014 cps/Hz 10.48 s 66.9 s0.25 cps17.34 cps

Turning to particulars, you can see that AMTOR and PacTOR average throughputs don 'tdiffer much on OHS and NVIS paths, although there is a slight tendency toward greaterstatistical variation (as measured by standard deviations) on NVIS paths. This similarityof average throughputs may explain why you don't hear much about differences betweenperformance on long and short paths in these two modes.

The big story is the differences between GTOR and packet performance on long and shortpaths. Average GTOR throughput on OHS paths was almost 50% higher on OHS paths

Operating 1-49

Page 59: Packet- Speed- More Speed

than on NVIS ones (32 char/s vs 23 char/s). This may reflect the presence of consistentlyhigher signal -to-noise ratios (SNRs) on OHS paths, since GTOR is said to thrive on highSNRs and suffer more than the other modes on low ones.

Packet throughput was two-and-a-half times higher on long paths than on NVIS ones(16 char/s vs 6 char/s). Some of this difference may have been caused by the fact that werestricted all our NVIS tests with packet to 200-baud operation. Although we based thisrestriction on observations of performance, it's possible that a more aggressive choice ofbaud rate on packet during the mid-morning.and mid-afternoon "NVIS windows" couldhave raised NVIS packet throughput somewhat. However, this does not explain all of theimproved performance, whose source must be the better OHS channel (fewer packet biterrors).

Another striking difference appeared in the average packet negotiation times(OHS : 35 s, NVIS : 103 s). (Recall that negotiation time is the difference between thetime a connection request is sent and the time file transfer starts .) This difference inaverage negotiation times apparently reflects the fact that the negotiation process for apacket BBS upload, which involves transmission of frames of various sizes, exposespackets at 200 baud much higher bit-error rates on NVIS paths than 300-baudnegotiations over OHS paths .

Table 3. Statistical Summary of OUS Throughput Data

Mode E(thruput) No links No_neg_tms E(xfer tm) E(No_char)sd(thruput) E(lnk_tm) E(neg tm) sd(xfer_tm) sd(No_chr)E(tput/Uz) sd(l_tm) sd(neg_tm)sd_mn(tput)max tput

AMTOR 5.70 cps 104 92 543.2 s 3009.60.80 cps 2.62 s 69.7 s 109.8 s 98.10.017 cps/Hz 3.81 s 15.2 s0.08 cps6.33 cps

PacTOR 20.19 cps 153 139 176.1 s 3058.85.49 cps 4.70 s 40.8 s 105.2 s 308.50.052 cps/Hz 6.53 s 29.4 s0.44 cps25.00 cps

GTOR 32.30 cps 158 144 119.9 s 3126.69.88 cps 4.44 s 50.9 s 102.6 s 501.40.073 cps/Hz 7.96 s 21.7 s0.79 cps44.12 cps

packet 15.67 cps 108 108 221.9 s 2975.04.58 cps 6.46s 34.6 s 141.4 s 259.80.029 cps/Hz 8.50 s 17.7 s0.44 cps24.59 cps

Maximum observable TOR throughputs were about the same for NVIS and OHS paths,although, as mentioned above, individual measurements came closer to their maxima

1-50 Chapter 1

Page 60: Packet- Speed- More Speed

more often on the long then on the short paths. On packet, we achieved maximum OHSthroughput of about 25 char/s vs about 17 char/s for NVIS.

6. Discussion of Packet Results

Our packet experiments over both NVIS and OHS paths have led to surprising results inview of what we have read on newsgroup discussions and elsewhere. For example, overOHS links we have consistently achieved average packet throughputs two to three timeshigher than average AMTOR throughputs, although not quite as high as PacTOR, andonly about half of the OTOR average (see Table 3 above). The parameters we haveadjusted to do this are PACLEN, MAXFRAMES, FRACK, SLOTTIME, RESPTIMEand PERSIST, and we have done all our OHS file transfers at 300 baud. (Since we havetried to choose frequencies and times where there is little interference, we have setPERSIST very high and FRACK, SLOTTIME and RESPTIME low for aggressive use ofthe channel.)

These high packet throughputs have been achieved, however, only during the day, and bymeans of very frequent, manual, changes of PACLEN and MAXFRAMES. Furthermore,we have managed to find frequencies that were by and large free of significantinterference from other signals (this appears to rule out most of the 20m band). Forexample, we have often been able to transfer files over both NVIS and OHS paths withcombinations like PACLEN = 100 and MAXFRAMES =5 in the absence of contention,which may be a revelation to some hams who have tried HF packet.

As a general rule, as packet begins to work in the morning on our links, values ofPACLEN/MAXFRAMES around 40/1 work best. From mid-morning till late afternoon,combinations like 80-100/4-7 often lead to high throughput. As the bands begin todeteriorate, it's back to near 40/1. PACLENs greater than about 120 bytes usually suffertoo many bit errors on our NVIS and OHS links to be worth trying.

After about 5 PM iocal time during the winter, throughput rapidly falls, and for most ofthe evening, getting files through in any mode was difficult. r.we got some NVIStransfers through in the middle of the night during the winter, but we didn't try any OHStransfers in the middle of the night.) On our links, trying a lower ham frequency in theevening usually led to increased interference, against which none of the modes did a greatjob.

Our experience with HF packet on OHS and NVIS links has convinced us that anadaptive protocol that adjusted HBAUD, PACLEN and MAXFRAMES using feedbackon throughput could go a long way toward polishing HF packet's tarnished reputation.However, with much better systems now available at reasonable prices, it is probably nolonger worth developing such a protocol.

7. The Effects on Throughput of Data Compression and File Type

Three of the protocols we have assessed over the air provide one or more types ofoptional or hard-coded data compression: PacTOR has (optional) Huffman compression,OTOR has hard-coded Huffman and run-length compression and CLOVER II with theHAL interface software has one or more hard-coded compression techniques from the so­called "PKLIB" suite. Other and future protocols may also include one or more

Operating 1-51

Page 61: Packet- Speed- More Speed

compression capabilities". Of course, even when a protocol doesn't include compression,the user is free to compress his files off-line before he sends them, provided that theprotocol can handle the compressed format and the receiving station can de-compress thefiles. (For an introduction to data compression see Ref. 7.) As mentioned above, wealmost always choose the Huffman option in PacTOR transfers of English text files.

How much a file gets reduced by a compression technique is strongly affected by thefile's type and the technique's approach, so that the user must have some understandingof the interplay of the two if he wants to use compression for high throughput.

In general, the closer the distribution of a file 's ASCII characters to the distribution ofcharacters in "typical" English (or other language to which the Huffman code has beentailored) text, the more Huffman will compress the file. The more repeated contiguouscharacters or bytes (e.g., spaces) in the file, the more run-length coding will compress it.The more repetitions of byte-pairs in a file ("an," "th" in "the," etc.), the more so-calledMarkov coding (multi-level Huffman) will compress it. The more repeated byte stringsin the file, the more "dictionary-based" methods like Lempel-Ziv-Welch (LZW)compression will squeeze it7. Finally, the bigger the monochromatic patches (e.g., bigexpanses of white background) in a graphics file, the smaller a graphics compressiontechnique (like those used in JPEG and for GIF files) will make it.

These facts means that if you send a text file consisting of a high proportion of upper-casecharacters with PacTOR, you won't get much benefit from Huffman, which relies (inmost PacTOR implementations) on a fixed text-character distribution in which certainlower-case characters (like "e") occur with relatively high frequency. Likewise, if youcompress a file off-line (e.g., zip it), you produce a compressed (8-bit, or binary) file thatlooks a lot like a pseudo-random string of bytes. If you then apply a built-in compressorlike one from the PKLIB suite used in the P38 CLOVER software from HAL, you willfind that the "compressed" file is actually a bit larger than the zip-file. (Of course, this isall right if the zip-ing did a good job.)

On the other hand, if you try to send an uncompressed executable (".exe") image as abinary file with CLOVER II and the HAL software, you'll find that the already pseudo­random structure of most executable (binary) files is likewise expanded rather thancompressed by PKLIB. To get efficient transfer by CLOVER II, you should compressexecutables off-line before submitting them to the HAL P38 terminal software.

Graphics files (not yet the main focus of our throughput experiments) are another story.If they're GIF or JPEG files, they're generally already compressed, so CLOVER andmost other compressors won't make them any smaller". PICT and BMP files, on theother hand, are not compressed, and often have big monochromatic chunks, so that the

6The TAC02 protocol suite developed by the DOD for transmission of battlefield-situation graphics overHF has data compression as an integral part.7For this reason, it is not a good idea to compare throughput for (cooked-up) files that consist of repeatedsections (for example, those made by repeatedly pasting a section to the end of the file). LZW willgenerally compress such a file by much more than 50%, whereas Huffman will only compress it by asmuch as it compresses the first section. The resulting comparison is therefore probably unfair to theHuffman, since in many cases one would send just the first section of such a file with the advice that it is tobe repeated N times at the receiver for whatever reason.8The popular shareware EXPRESS terminal program that also runs the P38 and other HAL CLOVERhardware offers built-in compression of files and tailored compression and transmission of graphics images .Since EXPRESS doesn't (yet) fall under our definition of a "common" implementation ("comes with themodem"), we don't cover it here.

1-52 Chapter 1

Page 62: Packet- Speed- More Speed

PKLIB compressor(s) in the HAL software usually make them a lot smaller, withcorrespondingly higher throughput.

So far in our NVIS experiments with CLOVER II using both compressed anduncompressed files we have found that compression plays a crucial role in the relativelyhigh average throughput (above 40 characters/s) we report in Table 1. (Recall that thisaverage applies to compressed English text files, and that OHS transfers are not includedin our CLOVER data.) The average CLOVER II throughput over NVIS paths foruncompressed files is only around 25 char ./s, which is about the same as the GTORthroughput of text files" .

As we pointed out in Section 3, we have not generally sent off-line-compressed files forthroughput comparison, so as not to penalize unfairly common implementations ofprotocols (like AMTOR and standard AX.25 packet) that can't easily handle binary files.The field of throughput comparison using compression techniques that aren't part of"common implementations" is a wide open and important one .

8. Concluding Remarks

One of the conclusions we've reached in our throughput assessments is that hams andothers need to separate long from short distance paths when they compare HF throughputperformance of ARQ modes, especially the amateur GTOR and packet modes. Somemoderation of opinion on HF packet performance may also be called for.

For HF data transfer, data compression plays a crucial role in increasing throughput, andit should always be used when it significantly lowers file or message size and thereceiving station is equipped to handle decompression. .

We hope that our throughput data will further clarify discussions of the HF digital modes.Our results should put throughput measurements of PacTOR II and other HF data­transmission systems in perspective. We have plans to report someday on theperformance of some of those newer modes, and encourage those already in a position todo so to publish their measurements.

Acknowledgments.

I'm grateful to Mike Bernock (KB IPZ), Dennis Gabler (KB5HVN), Doug Hall(KF4KL), Richard Harrison (NT2Z) and Bob Levreault (WlIMM) for handlingnumerous requests to put their stations on the air in various modes on variousfrequencies, and for regular mailbox cleanings.

Mike Bernock and Bob Levreault made a number of useful comments on the text of thepaper.

References

1. Young, T., et aI., "A Preview of HF Packet Radio Modem Protocol Performance," 13thARRL Digital Communications Conference, Bloomington, Minn. 1994.

9As already mentioned, GTOR applies Huffman and run-length compression to all transferred files.

Operating 1-53

Page 63: Packet- Speed- More Speed

2. Reynolds, P., "HF channel Simulator Tests of Clover," QEX, Dec. 1994.

3. Riley, T., et al., "A Comparison of HF Digital Protocols," QST, July 1996.

4. "HF Throughput Truths," in Packet Perspectives, QST. Feb. 1996.

5. Wickwire, K., et al., "On-air Measurements of HF TOR and Packet Throughput, Part I:Near-Vertical-Incidence-Skywave Paths," Digital Journal, March 1996.

6. Wickwire, K., "On-air Measurements of HF TOR and Packet Throughput, Part II: One­hop Skywave Paths," QEX, June 1996.

7. Nelson, M. and J-L Gailly, The Data Compression Book, 2nd Ed., M&T Books 1996.

1-54 Chapter 1

Page 64: Packet- Speed- More Speed

By Phillip Nichols, KC8DQF

Portable PacketsTurn a tiny "palmtop"computer into adynamite packet station!

Ihave been interes ted in computers since the "o ld days" whenI dabbl ed with a Co mmodore Vic-20 and my parent ' s Atari400. Back when the Mac intosh was still an Apple lIe! I keptpace wit h the changing technology while I upgraded my com -

puter skills. In recent yea rs I' ve relied on my noteboo k PC and aHew lett-Packard HP-200LX "palmtop" com puter.

Act ually, my notebook PC has take n a few trips to the comp uterdoctor. At one point it was absent for two months! It was during thistim e that I beca me more interes ted in explor ing the capabilities of myHP-200LX. I also learned of a fascinati ng hobby: Amate ur Radio .

After four weeks of study and the inevitable exa m, I was re­warded with my Technician Plus ticket. A few month s later I earnedmy Advanced. I was eager to see what I could do with this newhobby. Before long, I discovered packet.

What's Packet?In case you're a stranger to the term as hams use it, packet is a

meth od of conn ecti ng computer sys tems and netwo rks by radio­typica lly using FM transceivers. Many hams use packet to con nectto bulleti n board sys te ms (BBSs) where they dow nload files andexc hange elec tronic mail (e-mail) . Oth ers use packet to help themhun t OX and gather contes t points (OX PacketCluster networks),while some enjoy tracking movi ng objects on computer -gen eratedmaps (APRS-the Auto matic Packet Repor ting System ).

Bes ides the radio and a computer , all you need to get started is atermi nal node contro ller (TNC) . Think of a TNC as a radio modemand you' ll understand how it wor ks. A TNC takes data fro m yourcomputer and turns it into audio tones that the radio can transmit. ATNC also takes audio tones from the radio and converts them to datafor the com puter. In addit ion to the conversi on task, TNCs asse mbledata into the proper packet forma ts, check for erro rs in receivedpackets (and request rep lacements if necessary) and perform manyother functions.

TNCs can be stand-a lone dev ices-little boxes that sit next toyou r radio and computer. But TNCs can also be created in softwarewith ju st a tin y external modem to convert tones to data and viceversa. Software TNCs have the adva ntage ofrequiri ng little externalhardware. Tha t makes them idea l for portable packet operating!

Creating a Palmtop Portable StationPortability is important to me. I live with the most restrict ive

antenna requirements you ca n imagine: no outdoor anten nas al­lowed ! Like many "i ndoor hams," I hang anten nas in my livingroom . (I 'm blessed to be marr ied to a forg iving wife!)

I started to wonder if I co uld use my palmtop for packet. Wit h apalmtop and an FM rig, I could log on from anywhere-even frominside a shopping mall or at work.

The key ingredient turned out to be the BP-2 by Tigertronics. Itcomes wit h Bayco m 1.4 software and a modem (a sma ll 2-inch­square custom analog/digital modem based on the old Bell-202 stan­dard). The Baycom software makes a comp uter behave like a TNC .In compute rese we say it "emulates" a TNC. Bayc om is avai lablefree on various land line BBSs, CompuServe, and on the WorldWide Web.

The complete palmtop packet station is smaller than mylicense plate! From left to right, my HP-200LX palmtopcomputer , BP-2 modem and HTX-202 transceiver.

Palmtop Packet ResourcesTige rtron icsPO Box 5210Gran ts Pass , OR 97527te I800-8BAYPAC , or 541- 474-6700fax 541-474-6703;World Wide We b http://www.tigertronics.com

The Palmtop Paper(a newsletter devoted to palmtop com puting)Thaddeus Co mput ing57 E Broa dwayFairfield, IA 52556tel 800 -373-6114

Othe r palmtop software on the Web at: ftp:/Ieddie.mit.edulpub/hp951xlNEW/ includ es:

REAL95.ZIP-A satellite-tracking program . Slow, but effective .

GRIO.zIP-A grid-square locator

ARS-LOG.zIP- A logging program data ba s e for the HP-200 .

GE044CGA.ZIP-Graphical display of day/night illumination ofthe Earth as we ll a s the twilight "gray line ." Can a lso plotcontacts by la titude an d longitude .

I con nected my BP-2 modem to my Radio Shack HTX -202 hand ­held according to the BP-2 instructions, keep ing the cables as shortas possible. It worked like a charm!

The FutureI'm planning to operate a land -mobile packet station while I'm

taki ng vacat ion trips this summer. I'd also like to try aeron auticalmobile packet fro m a pri vate airplane j ust to see how it would work .Of course, a portable packet station like mine is ideal for public­service work . You could, for exam ple, assist disaster re lief by relay­ing emergency traffic . The possibilities are endless!

If you 're living in a restricted env ironment, or if you are in­trig ued by the idea of portable packet operating, get your hands ona palm top and give it a try!

e-mail [email protected] cket [email protected] http://www.gatecom.com/-maccabeus/hampage.html

IQ5T<-1

Operating 1-55

Page 65: Packet- Speed- More Speed

Software For The BP-2 and BP-2MFrom The Tigertronics Web Page

SOFTWARE.DOC 3/24/97

To help you get started with your BP-2M Modem, we have put together thislist of compatible shareware and commercial programs.

In this document we will tell you who makes the program, what modes theyhave to offer, and where you can get your copy! We will tell you how toconfigure each program for use with your BP-2M and make comments aboutconfiguration or operation . At the end of this document y ou will find a listof incompatible programs and the reasons why each of them will not work withthe BP-2M.

Tigertronics in no way endorses or supports any of the programs listedin this document. We have compiled this list only to aid you in locatingprograms that are known to work with the BayPac MultiMode. As y ou shouldexpect, some programs will work better than others. Many work better orworse depending on y o u r specific equipment configuration. We recommend thaty ou download and evaluate several programs to determine which wo rks best iny o u r particular application . This list represents only a sampling of thesoftware that is available. We will make an ongoing effort to refine andupdate this list as we become aware of suitable programs. The latest releaseof this list and many of the programs mentioned, can be downloaded from ourInternet Web Site at: http:/ /www.tigertronics.com

We would appreciate any comments y ou may have about programs that y ou areusing with the BayPac. We welcome all comments - Pro or Con! We would liketo add y ou r comments to the list, for all to s hare (if appropriate!).

INTRODUCTION

Before we get started, here a re a few notes on how this information isput together . The programs in t h i s document are listed in alphabetical orderby p rogram name. If the program is from a commercial source then thecompany's name will be listed. If the p rogram is shareware then y ou · will findthe word "Shareware" instead .

To find out where you can get your copy of the program, look for the"Source:", followed by one (or several) numbers. These numbers reference anInternet Web Page, a telephone BBS number , or the authors address (Sources arelisted at the end of this document) .

The "Comments :" line wi l l give y ou the information that y ou need toconfigure the program, along with any tips and suggestions that might behelpful.

The "Min Requirements :" line tells y ou what the computer hardwarerequirements are for each program (as per the programs documentation) .

The last thing we need to mention is that t he BP- 2M has many dif fe r ent"Modes " which allow it t o operate wi t h t he v a r i ous programs . The numbe r tha ty ou see following the l ine "BayPac Mode : " indicates whic h Mo de y ou must s elec tin the BPMODE program to be compat ible .

AEA ACARS

Source: 1

1-56 Chapter 1

Adva n c e d Electronics App l i cat i on s (AEA)

BayPac Mode : 3, 6

Page 66: Packet- Speed- More Speed

Comments:

Mi n Requi r ements :

supports:

AEA FAX III- - _ :...- - - - - - '-

Source : 1

Comments:

Min Requi rements:

Supp orts:

BAYCOM

Source: 2 , 1 9

Comments:

Min Re quire ments:

Supports :

Outstanding performance. Long range reception ofACARS signals i s possible do to the height of aircraftabove t h e earth.

ACARS (Aircraft Communications and Reporting System)

Advanced Electronics Applications (AEA)

BayPa c Mode : 3, 6

If you want to use a comm port other t han COMl, bes u r e to specify the 'comm port when y ou s t a r t thep r og r am : FAX /2 for COM2 , FAX / 3 for COM3 , etc .

IBM compatible XT, CGA, DOS 2 .1

CW , FEC (Sitor-B, Amt or-B & NavTex) . 4 5 - 1 00 bau dRTTY (Ba u do t , ASCII 7 & ASCII 8) 45 -100 baudFAX (288, 352 & 576 IOC) 60, 90 , 120 , & 24 0 LPM

The BayCom Team

BayPac Mode : I , 2

BayCom is probably the most popular packet terminalsoftware out there . In our testing, it consistentlyout p e r fo rms a l l other TNC emulat i ng s oftwa r e .Ba y Com is available as a shareware p roduct or as ac ommerc ial product . Version 1 .4 is the latest(and the last) SHAREWARE release . Ba yCom vl . 6 is thecurrent c ommercial version and it offers man y newfea t ures . I n a dd i tion to printed a n d bound manuals ,BayCom v l . 6 supports comm p o r t s 3 and 4, keyboardmacros, and 9600 baud operation (requires 9600 baudmo de m). Imp r oved security during remote operations ,n ew t r a n sfer protocols , and a screen saver are just afew of the enhancements .

I BM c omp a tible XT o r better, EGA or better, COMI orCOM2, DOS 3.2 or l ater .

Both Baycom v l. 4 and v l . 6 support 300 and 1200 baudpacket. BayCom vl .6 a lso supports 9600 baud operation(requ ires 96 00 baud modem). Baycom vl .4 and v l . 6s upport b inary (not . the YAPP protocol) and ASCII fi letransfer . Ba ycom vl .6 also automatically saves 7Plusencoded files to d i s k (decoding of the file s is NOTdo n e by Ba y Com) . Both v e r s i on s o f BayCom supp o r tr emote operations, and both can be set up to operatea s a mailbox or mini-BBS.

EZSSTV. ZIP

Source : 3, 19, 2025

Easy SSTV v 3 . 0

BayPac Mode: 3 , 6

Shareware

Comments: EZSSTV is a reduced-functionality v e r s i on of thepopular Pasokon TV. Version 3.0 NOW TRANSMITS,displays over 16 mil l ion colors, and has t h r e eadditional modes! This program is mouse driven

Operating 1-57.

Page 67: Packet- Speed- More Speed

Min Re qui r eme n t s :

supports :

and includes a built -in paint program for creatingo r editing your images. EZSSTV is easy to use, andit offers great Slow Scan performance !

IBM Compa tible 386DX, 640K, VGA, COM1 or COM2 only

SSTV Robot 36, Ma r t i n 1, Scottie 1 and Scottie DXWraase 120 and 180 sec.

FBB515 .Z IP FBB v 5 .1 5 Shareware

Source: 4, 1 9 , 23

Comments:

Min Requirements :

Supports :

BayPac Mode : 1 , 2

FBB is a popular packet radio BBS and Server programt h a t is identical in operation to the WORLI or WA7MBLBBS systems . Due to i ts many options and features ,the installat i on a n d s e t up of FBB is more complex tha nothe r packet prog r a ms .

IBM compatible XT, 640K, Mono CGAUses the TFPCX a x .2 5 driver (see comments un d e rTFPCX21 0.EXE) .

300 and 1200 baud packet . Gateways, chat , andconferencing . YAPP and . ASCII file transfer . Atelephone interface is also supported for thosewho want a "landline " connec tion!

GP161B .ZIP

Source: 5, 19 , 24

Graphics Packet v 1 . 61

BayPac Mode : 1, 2

Sh a r e wa r e

Comments :

Min Requirements:

Supports:

Graphics Packet i s a mou s e - d r i v e n packet t e r minalprogram . It c ome s with an ASCII text editor, anda screen saver. A BBS function i s also built i n ,but we were unable to test it due to a lack of Englishdocumentation .

IBM compatible XT , EGAUs e s t h e TFPCX a X.25 driver (see comments un d e rTFPCX210 .EXE) .

300 and 1200 baud packet . ASCII and 7Plus filetransfer .

GSHPC1 2 . ZI P GSHPC v1.2 Shareware

Source: 6, 19 , 20

Comments :

Min Requirements:

t-58 Chapter 1

BayPac Mode: 3, 6

GSHPC offers good SSTV performance, and is easy touse once y ou become famil iar wi th t h e program.' Theprogram has separate transmit and receive wi n dows ,and it supports BMP a n d TIFF images wi t h up to 16million colors! A paint program is a l s o built in .Select these options wh e n configuring GSHPC:

For PTT control use "Com x RTS- pin" .For demodulator u s e "Com- x DSR-pin" .Fo r modulator use "Co m_x-TxD-pin".

IB M compatible 386DX, 640 k, 1MB VGA

Page 68: Packet- Speed- More Speed

supports: SSTV Mart i n 1 , 2, 3, & 4Scottie I, 2, 3, 4 , DXRob o t Color 1 2 , 24 , 36, & 72Robot BW 8 , 12 , 24 , & 36sSC- 1 8s, 1 6s , & 32sSC- 2 3 0s, 6 0s , 120s , 180s

HAMCOM31,EXE Ha mComm v3 .1 Shareware

Source: 7 , 19

Comments :

Mi n Requirements :

Supp o r t s :

BayPac Mode: 3 ~r 4 f o r CW, 3 fo r all other modes,or 6 (rx only)

This vers i on of HamComm has many n e w f eatu res a ndimprovements. There is now a tuning indicatoron the main s c r e e n so that you do not have tochange screens to make a tuning adjustment !A new PACTOR listen mode has also b een added (re quiresregistrat ion) . with its great performance , andcontinuous improvements, its easy to see why HamComm isone of the most popular multimode programs .

IBM c ompatible XT , 640k, CGA, DOS 3 .0

CW , AMTOR ARQ and FEC 4 5 -2 00 baudRTTY (Baudot, ASCI I 7 & ASCI I 8) 4 5 - 2 00 b audPACTOR listen mode for registered users .

I NTCOM32. ZIP

Source : 8, 19

INTCOM v 3 .2

BayPac Mode : 3

Shareware

Comments :

Min Requirement:

Supports:

INTCOM is a simply mu l t i mode p rogram t h a t performswe l l and is easy to use . Except for an onscreen tuningscope (which is v e ry useful ) , y ou won ' t find any other"Bells and Whistles" .

IBM compatible 286 , Hercules or VGA , COM1 or COM2

CW, RTTY Baudot 4 5 , 50, 75 & 100 baud, ASCI I 110 b a u dSitar FEC Mod e B

JVFAX71A .EXE

Source : 9, 19, 2 0

JVFAX v 7 .1a

BayPac Mode : 3, 6

Sha reware

Comme n ts :

Min Requi r ements :

JV FAXis probably t he mo s t popu l ar program f o rSSTV a nd WE FAX operation. It performs very we l l i nboth tra nsmit and receive, and i t is e a sy to use!Wh e n configuri ng JVFAX , select "HamComm"f o r the demodulat or, and "Serial Audio"fo r t he modula t o r. Also , mak e s ure you h a v e thecorrect base address and IRQ for y ou r serial port!

IB M compa t i ble XT , VGA , DOS 3 . 0

Suppor ts: SSTV

FAX

Martin 1 and 2Scottie 1, 2, a n d DXRobot · 72c , BW 8s , 16s, and 32sWR24/128, WR48/128, WR48 /2 56, a nd WR96 /2 56WR12 0 and WR180Wef ax 288 , We f ax 576, Ms a t CHI, and Msat CH2Ha mColor , Ham 288b , Metr SN, a nd Me tr NSNOAA NS, NOAA SN, Color 240 , and H28 8 /120

Operating 1-59

Page 69: Packet- Speed- More Speed

NOAALLSN and NOAALLNS

MSCAN20 2 .ZIP

Source : 10, 19, 26

Micro Scan v 2 . 02

BayPac Mode: 3, 6

Shareware

Comments :

Min Requirements:

Micro Scan is a feature packed SSTV and FAX program .Unlike most programs, Micro Scan allows y ou to loador edit a picture at the same t ime that you arereceiving or transmitting ! Mico Scan s upports a mous e,and it comes with a bu il t in paint program, and s p e c t r umd isplay to aid in tuning . A c onnection t o the PCspeaker is required for transmitting , and theunregistered shareware v e r s i on only supports thosemodes marked with an asterisk.

IBM compatible 386DX, SVGA, COM1 or COM2 onl y

Supports : SSTV

FAX

Mart in M1* a nd M2Scottie Sl*, S2, and DXBW SSTV 7.2 *, 8*, 16*, and 32Wraase 24*, 48* , and 96BW 60 , 90, 120*, 180, and 240 LPMColour 120, 180, 24 0* , and 360 LPM

MUBAY102.ZIP

So urce: 11, 19 , 22

MUlti,-user BayCom v 1 .02

BayPac Mode : 1 , 2

Shareware

Comments:

Min Requirements :

Supports:

This is a Multi -user /multitasking packet radioprogram that al lows multiple operators to accessd ifferent areas of the s ys tem at the same time!Ins t a l lat ion of this p rogram i s a little mor ec omplex than t erminal type programs, but i t isfairl y easy to get it up and running . The PersonalMa i l System (PMS) is disabled in the sharewarev e r s i on .

IBM compatible XT, 640k, Mono, DOS 3.0Uses the TFPCX a x .2 5 driver (see comments underTFPCX2 10.EXE) .

300 and 1200 baud packet. UUencode /uUdecode of files.YAPP, 7Plus and ASCII file transfer. A scriptlanguage , file /notepad editor and directory browserare also included .

PC HF FAX

Source: 14

PC HF FACSIMILE v7 .0

BayPac Mode: 5, 6

Software Sys tems Consulting

Comme n t s:

Min Requirements :

Suppor t s :

1-60 Chapter 1

Professionally written , documented, and suppor tedsoftware. Perfor ms v ery we l l in all modes. Transmitfunctions in almost all modes , including WEFAX!

IBM compatible XT , 640K , CGA, DOS 2 . 1

CW , RTTY (Baudo t, ASCI I 7 & ASCII 8) 4 5 - 1 00 baudARQ , FEC, NavTex 75 and 100 baudRaw Hex 4 5 - 1 00 baudFax 48 0 HAMWEFAX 2 0 - 24 0 LPMPRESS FAX 20- 240 LPM and 4 8 0 HAM

Page 70: Packet- Speed- More Speed

PC SSTV

Sou r c e : 14

PC SSTV v 5 . 2

BayPac Mode : 5, 6

Sof tware Systems Consulting

Comments :

Min Requirements :

Supports:

Profess i onally wri t ten and wel l s uppo r t ed s o ft wa r e .Supports mos t SSTV mode s and o f f ers the use r theability to define n e w (custom) modes .

IBM compatible 2~6, 640K , VGA , DOS 2 . 1

SS TV 2 , 24 , 36, and 72 sec Color8, 12, 24 , and 36 sec BWScottie 1, and 2Martin 1, and 2AVT 90, 94, and 12 5

PKTMON12. ZI P

Source: 12, 19

Packet Monitor v 1 . 2

BayPac Mode: 3, 6

Shareware

Comments :

Min Requirements:

Supports:

This is a receive only p rog r a m used for monitoringHF and VHF packet radio t r a f f i c . PKTMON logs r e c eiv e ddata into separate f iles based on s ource and dest inat i on a ddAn on-screen tuning indicator aids in tuning.

IBM compatible 286

300 and 12 00 baud packet

PROSKAN .ZIP

Source : 13, 19 , 2021

Pro Scan v 3. 01

BayPac Mod e : 3, 6

Shareware

Comment s :

Min Requirements :

PROSKAN al lows you to receive a nd transmit SSTV images .FAX mode s are a l so prov i ded, but they are for recept iononly . PROSKAN is easy to conf igure, it supports amouse , and it has a built-in paint program for editingor creating images . The unregis tered s hareware v e r s i onof PROSKAN disables or limits a lot of the programsbetter fea tures .

IBM compatible 386DX, 64 0k, SVGA, DOS 4 .0

Supports : SSTV

FAX

Ma r t in 1 & 2 , Scottie 1, 2 , DX, & DX2BW 24, & 36 , AVT 24 , 90 , & 94, J-12 0WEFAX (576 IOC ) 40 - 240 LPM

SP650A.EXE

Source: 15 , 1 9

Super Packet v 6. 5a

BayPac Mode: 1 , 2

Share ware

Comments :

Min Requirements :

This is the l a s t shareware release of Super Packet.Installation and se t up of this p rogram is complex ,however it is easy t o use once y ou get it up andrunning .

IBM compatibl e XT o r At a ri ST , 640K, Mon oUses t he TFPCX ax.25 driver (see comments under TFPCX21 0 .EXE

Operating 1-61

Page 71: Packet- Speed- More Speed

supports: 300 and 1200 baud packet .and ASCII file transfer.and keyboard macros.

BBS capability wi t h 7PLUSRemote operations, scripts ,

TFX28.LZH TFX v 2 . 8 Shareware

Source : 19

Comments:

Min Requirements:

Supports:

BayPac Mode : I, 2

TFX is a new RAM resident (TSR) program that receivesand decodes packet transmissions. This program issimilar to TFPCX in that it is used as the front endto other programs (like Graphics Packet, Super Packet,e t c . ). The TFX28.LZH archive file ·contains severaldifferent driver p rograms . The file TFX PORT.COMsupports BayPac 1200 baud modems, wh i l e the fileTFX PAR. COM supports the Tigertronics BP-96 (9600 baud)modem .

IBM Compatible PC

300, 1200, and 9600 Baud packet

TFPCX210.EXE TFPCX v2.1 Shareware

Source : 16 , 19

Comments:

Min Requirements :

Supports:

BayPac Mode: I, 2

TFPCX is a RAM resident (TSR) program that receivesand decodes packet transmissions. TFPCX is almostalways used as the front end to other programs (likeGraphics Packet, Super Packet, etc .) , however it can beset up to act as a VERY SIMPLE stand alone terminalprogram. Detailed instructions fo r setting up TFPCXcan be found in the documentation files that accompanythe program . Programs employing this driver work verywell, but perhaps not quite as well as BayCom.

8MHz IBM compatible XT, 640k

300 and 1200 baud packet. Graphics Packet, Multi-userBayCom, Super Packet, TOP, and TSTHOST are just a fewprograms that use TFPCX .

TOP151 .ZIP

Source : 17, 19 , 22

The Other Packet v 1. 51

BayPac Mode: I, 2

Shareware

Comments :

Min Requirements :

The Other Packet (TOP) terminal program . TOP supportsup to 10 operational channels, and provides keyboardmacros and a screen saver.

IBM compatible XT , 640 KUses the TFPCX a x .25 driver (see comments under TFPCX210.EXE

Supports: 300 and 1200 baud packet.file transfer protocol .

7Plus , binary and ASCII

TSTHOST .EXE TSTHOST v1.4 2 Shareware

Source: 18, 19 BayPac Mod e : I, 2

1-62 Chapte r 1

Page 72: Packet- Speed- More Speed

Comments:

Requires:

Supports:

TSTHOST is a packet BBS/Server program that is loadedwith features! Up to eight communications channels aresupported, and all channels may be active with file transferor Personal Mail System (PMS) activity at the same time!

IBM compatible XT with 640K, EGA or VGAUses the TFPCX ax.25 driver (see comments under TFPCX210.EXE

300 and 1200 baud packet. 7PLUS, YAPP and ASCII filetransfer. Macro commands.

WINTNC11.EXE

Source: 11, 19 , 22

Win TNC v1.01

BayPac Mode: 1, 2

Shareware

Comments:

Supports:

Packet radio for Windows 3.1, Windows For Work Groups,and Windows 95! WINTNC features simple installation,on-line configuration of install parameters, andon-line help. When configuring WINTNC, under"Port Configuration", set the parameter "Carrier Sense"to "2" . Also, the author of WINTNC reports thatperformance may be improved by setting the WALKSTEPDIVparameter to 64. Interim upgrades to WINTNC are beingreleased as improvements are made. Check the WINTNC sourcesmentioned for these upgrades. The file WTNC101F.EXE is thecurrent upgrade to WINTNC .

300 and 1200 baud packet. Personal Mail System.

NOTES: 1) BayPac Mode.#6 is receive ONLY, but works with almost all multimodeprograms.

2) Only the shareware version of BayCom (v1.4) is available on the ARRLBBS.

SOURCES

1) Advanced Elect ApplicationsP.O . Box C21602006 196th Street SWLynnwood, WA 98036

Voice (206) 774-5554Fax (206) 775-2340Compuserve 76702,1013

2)

3)

BayCom v1.6Tigertronics, Inc.400 Daily LaneP.O. Box 5210Grants Pass, OR 97527

EZSSTV .ZIPJohn Langner WB20SZ115 Stedman St. # EChelmsford, MA 01824-1823

VoiceFax

Voice

(541) 474-6700(541) 474-6703

(508) 256-6907

4) FBB515.ZIPJean Paul Roubelat6, rue George 'SAND31120 ROQUETTESFRANCE

5) GP161B .ZIP

Operating 1-63

Page 73: Packet- Speed- More Speed

6)

7)

Ulf SaranDH1DAEVe i t - St oB- St r. 36570 76 Siegen

GSHPC12 .ZIPG. Sza b a do s - Ha nn DL4SAWAm Zun dhu t l e 7a76228 Karlsruhe / Germany

HAMCOM31 . EXEW. F. Schr oederAu g s bu r g e r Weg 63D-33102 PaderbornGermany

Vo i c e 07 21 / 47-53-19Fax 0721 / 47-53-19

Email schroeder [email protected]

8 ) INTCOM32. ZIPP . M. Haringsma (PA3BYZ)dr. Hattinkst r . 138563 AC Wi jckelThe Netherlands

9) JV FAX71A . ZIPEb e r hard BackeshoffDK8JVObschwarzbach 4 0aD-40822 MettmannGermany

10)

11 )

MSCAN2 02 . ZI PCombiTechMorelstraat 603235 EL RockanjeThe Netherlands

or North Ame r i c a n agent:

HAMVision NAP .O. Box 5073Salisbury, NC 28147-7806

MUBAY102.Z IPWI NTNC11 . ZI PJon We l c h50 Quarrydale RoadSutton In Ashfield,NottsNG1 7 4DRGr e a t Britain

Voice +311814-4252Fax +3 11814-4252

Voice (704) 636-3308Fax (704) 636-3308

Email jon@g7jjf .demon.co.uk

1 2 ) PKTMON1 .ZIP:rawel JalochaRynek Kl e p a r sk i 14 /14aPL- 31150 Krakow

13) PROSKAN .ZIPMa yna r d A. Philbroo k jr .520 Pleasant St.Willimantic , CT 06 226 Vo i c e (203) 4 56 - 116 7

14) Software Syste ms Consulting61 5 S. EI Camino RealSan Cl e me n t e , CA 92672

Vo i ceFaxBBS

(714) 4 98 - 5 784(714) 498 -0568(619 ) 259 -555 4

15)

1-64 Chapter 1

SP650A.EXEOrf e o DiBiase, NS8M64 0 E. 8th St .Salem, OH 444 6 0 Email ba350@y fn. ysu.edu

Page 74: Packet- Speed- More Speed

16) TFPCX21 0 .EXERene StangeO.-Grotewohl - Ring 341 5344 Strausberg

17 ) TOP151.ZIPF5 NZEBe rnard MEBS20 Rue Des Roses67120 DUPPIGHEIMTEL/FAX 885 07120

18) TSTHOST.ZIPI K1GKJMario TravaglinoVia Tries t e, 1028069 Trecate (No) - ITALY

1 9 ) ARRL BBS (860) 594-0306

I n t e r n e t links

20 . ft p : / / ftp. f u n e t . fi / p ub / ha m/ f a x sstv /21 . http:/ /www.mindport .net / -jamie722. ht tp : //members.aol .com/g7jjf /index.html23. http://www .vectorbd .com/bfd/f6fbb /24 . http:/ /www.vectorbd .com/bfd/packterm/2 5 . h t tp : //www .ultranet.com/-sstv /26 . h t tp://www.pi .net/-ctech/

NOTE : Ftp links are a vailable for most of these programs on the downloadp a g e o f our web s ite . (http: / / www.tigertronics .com)

SHAREWARE - DOING YOUR PART!

Most of the programs on this list are Shareware. Simpl y put, Sharewaremeans t hat you a r e welcome t o download a n d use the program at no charge for alimited trial period. Af t e r the e valuation period , if the program meet s y ou rnee ds and e xpec t ations and y ou p l a n t o continue using it, y ou are asked tos e n d a smal l regi s tration fee t o the author .

Some of t h e s e programs represen t "man years" of effor.t and rival theircommercial counterparts . By marketing these programs as shareware , theauthors are able to bring you outstanding products at rock bottom prices.But , Shareware onl y works if y ou honor y ou r part of the agreement .Tigertronics encourages y ou t o register t h e software y ou choose to keep andsupport the authors in their work. Register today!

INCOMPATIBLE PROGRAMS

AMTOR .EXE

ATFAX .ZIP

EMBAYCOM.ZIP

L2 PCX .EXE

Older v e r s i on of TOR306 .ZIP (See TOR306.ZIP)

Requires special interface.

Re qui.r e s modified "HainComm" interface. This program a l lowsa "HamComm" interface to r un BayCom software . This programis of little use since the BP-2M runs BayCom directly.

Supports serial FSK modem only .

Operating 1-65

Page 75: Packet- Speed- More Speed

PAC020 .ZIP

PCTOR.ZIP

PD- 20 3.ZIPt·

PROCW .ZIP

RTTY12G. EXE

SSTVFAX2. ZIP

TOR306. ZIP

TTY3 7 .ZIP

VESTER M. ZIP

WXMAN2 .ZIP

W95SSTV.ZIP

SSTVBL. ZIP

1-66 Chapter 1

Re qui r e s spe cial i nterface .

Older vers i on of TOR306. ZIP (See TOR3 06 . ZI P )

Re quires special interf ace .

Requires exter nal tone decoder on receiv e.

Requires spec ial interf a ce .

Requi res s pecial interface .

Requi r es special interface .

Requires special interface.

Complex setup/operation -'We couldn't get it to run .

Sound Ca rd only .

Sound Card only .

Sound Card only

Page 76: Packet- Speed- More Speed
Page 77: Packet- Speed- More Speed

Chapter 2Theory/Design

Page 78: Packet- Speed- More Speed

"9600-Ready" Radios:Ready or Not?As VHF/UHF FM transceiver manufacturers take the plunge andoffer new radios that claim to be ready for 9600-baud packetoperation with little or no modification, users are wondering if theyreally work. The ARRL Lab takes an objective look.By Jon Bloom, KE3Z

ARRL Senior Engineer

BER Test Box

BER Test Box

an AFSK signal at 9600 baud that will fitwithin the voice frequency range. Instead ,the binary data signal is encoded and fil­tered , then applied dire ctl y to the FM trans­mitt er, producing a true frequency-shift ­keyed (FSK) signa l. The dem odulatedsign al from the recei ver is detected bysimply determining whether the signalvoltag e is above or belo w zero- tha t is,whether the current level represent s a 1 bitor a 0 bit. The resulti ng data stream is thendecoded to recover the bin ary data origi ­nall y sent.

The first 9600-baud packet sys tem wasdeveloped by Steve Good e, K9NG. 4 In hisscheme , the bin ary data from the TN C wasfi rst passed through a "sc rambler" circuit.Despit e its name, scrambling has nothingto do with encryption or hidin g the dat a.It ' s simply a means of encod ing the data toensure that the numb er of I bit s is essen­tiall y equal to the number of 0 bit s in thetransmitted data stream . Doing this elimi­nates any de offset in the signal. For ex­ample, if we sent an unscrambl ed signalconsisting of a number of bits at +1 V fol­lowe d by a single bit at - I V, with this

/

/"/'

/"/'

3028

26242220ro3 18

o 16~ 14Vi

1210864

2o

-1 30 - 126 - 122 - 118 - 114 - 110 -106 -102 -9 8 - 9 4 - 90Receiver Input Level (dBm )

RFSignal

include measur ement s made usin g thesetechniques.

Background

Packet operation at 9600 baud imp osesdifferent requirements on the radio thandoes 1200-baud oper ation because the sig­nals applied to the radio are very different.At 1200 baud , the binary dat a to be sent isused to generate an audio frequency-shift­keyed (AFSK) signal in the ter mina l nodecontroll er (TNC). Thi s shifting audi o tone,cent ered at 1700 Hz, is the signal applie d tothe tran smitter' s audio input. From the re­ceiver, the audio signal is applied to theTNC, which dem odulates the AFSK signalback into binary data. The benefit of thisscheme is that the sig na l fit s neatl y into thefrequ enc y range of norm al speech, so it canbe sent through voice-grade radio s.

9600-baud packet uses a different ap­pro ach because it isn 't feasibl e to generate

ReceiverUnderTest

(A)

AudioOut

AudioIn

T he increasin g popularit y of 9600­baud packet, sparked by the ex ist­ence of several 9600-baud satell ites

and by the fru str ation man y packet usersfeel with 1200-baud terrestri al pack et, hasinduced several equipment manufacturersto include 9600-baud cap ability in theirnew VHF and UHF transce ivers . As withall aspects of radio performance , it ' s likel ythat not all rad ios perform equally well at9600 baud . To discover what differencesthere are in the 9600-baud performance ofthe se radi os, the ARRL Lab has de velopedtechniques to test radio performance with9600-baud signa ls . In thi s article, wepre sent a brief description of those tech­niques, along with test result s for radiosthat have rece ntl y appeared in Produ ctReview. 1.2 A companion article in a recentiss ue of QEX provides the full technicaldetail s of the test setup and softw are ."Future reviews of 9600-baud radios will

Figure 1-Bit error rate (BER) test setups used to test receivers(A) and transmitters (B).

Figure 2-Measured SINAD versus input leve l of the BER testbox mixer/demodulator.

Theory/Design 2-1

Page 79: Packet- Speed- More Speed

sequence repea ted a number of times, thesig nal wo uld have a substant ial de co m­ponen t. Differen t data patt ern s wo uldca use dif feren t de compo ne nts to bepresent. Sc rambling eliminates this dcco mpo nent, allowing us to use ac (capac i­tive ) coupling at var ious points in our sys­tem sin ce we don' t need to preserve a dcpar t of the tran smitted signal. It also meansthe system can tolerate a small frequencydifference between the transmitter and re­ce iver-a fre quency differenc e that showsup as a constant de offset.

K9NG's modem simp ly filte red thescrambled signal to e liminate harmon icsof the square -edged data sig nal, the n ap­pli ed the filtered sig nal to the transmitter.But filte ring the sig nal also causes eachtransmitted bit to spread out in time, over­lapp ing the nearby bits . As long as the sig­nal is transmitted , propagated and receivedwithout significant disto rt ion (amplitudeor phase error ), thi s spread ing of the bitsisn 't a problem . But if distorti on is present,different parts of the transmitted bit s getdistorted differentl y, and the rece ived sig­nal no longer has the same shape as when itwas tra nsm itted. This can ca use pro blemsfor the recei ving 9600 -baud modem .

To address this problem, James Miller,G3RUH, de veloped a 9600-baud modemdesign that, while compatibl e with theK9NG modem, generates a signa l that isless susc eptible to di stortion .> In theG3 RUH modem, the scrambled data streamis used to ge nerate bi t pulses that have araised-cosine spec trum. Such a pulse is atits max imum amplitude at the center of thebi t and, while the bit overlaps prece dingand following bits, its amplitude is alwayszero at the ce nters of those bits ." Th atmeans that eve n if the sig nal is somewhatdisto rted, the part of a bit that ove rlapsnearb y bit s will still probabl y be near zeroamplitude at the centers of tho se bits . Un­less the di storti on is severe, each tran smit­ted bit should still be larg ely unaffected bythe ove rlapped part s of the other bits.

G3 RU H also recogni zed that it is pos-

sib le to co mpensate for am plitude an dphase distor tio n if the character of tha t dis­tortion is known. For example, if you knowthe sig nal will pass throu gh a receiver thatattenuates hig h frequencies, you can boostthe ampli tude of the high frequencies in thetransmitted signal. His mean s of providin gco mpensation was to ge nerate the tran sm it­ted puls es using a RO M loo k-up tabl e, withse tta ble jumper s to se lect the pa rt icularpulse shape fro m the ROM. In a point-to­poin t link, where a single tran smitter andreceiver are used , this is an effective tec h­nique, allowi ng correction of the co mbineddistor tion of the transmitter and receiver.But in a system where a signa l may be re­ceived by one of a number of rece ivers, onlythe transmitter disto rtio n can be co rrec tedthi s way.

The G3R UH sys tem is the de facto sta n­dard 9600-baud packet mod em , so wetested the se radios for com patibility withG3RUH-type signals . These signa ls have aspectrum that is co nsta nt in amplitude fro mnearly de to 2400 Hz, then gra dually de­creases, reac hing - 6 dB at 4800 Hz anddim inishing to zero at 7200 Hz.

So me manu factur ers are now sell ing9600-bau d TNCs that use sing le-c hip960 0-ba ud mod em s. These operate usin gGaussian minimum-shift key ing (GMS K)waveforms , wh ich are similar-but notide ntical-to the signals used by G3RUHmodems. The two signal types are eno ughalike that G3 RUH modems and GM SKmodem s are int erop erable, alth ough with asma ll performance pe na lty co mpa red tousing iden tical modems at both ends of thelink. In ge neral, the results you' ll get froma given 9600-ba ud radio wi ll be similarwith either the G3 RUH or GMSK modem s,but spec ific radi os may affect one type ofsigna l more than another.

Testing Radio Perform ance

Th e best wa y to characterize the perfor­mance of a digit al communication link is toge t right to the bottom line: How acc ura telydo the bi ts ge t through? Th e meas ure of this

is the bit error rat e (BER). The BER stateshow many of the bits sent thro ugh a sys temwill be received incorrectly , on average.For example, iffor eac h 1000 bi ts transmit­ted thro ugh the system, one bi t is receivedincorr ectl y, the BER is III 000. For co nve ­nience, becau se the BER numbers are small(we hope !), they are usuall y writte n usingsc ientific notati on . In thi s exa mple, theBE R is IxI0-3.

We ca n relate a given BE R to real-worl dperformance . In AX .25 packet, a single badbit in a packet will ca use the packet to beunusab le. That mea ns that if yo ur ave ragepacket is, say , 1000 bit s in lengt h (a reason ­able number), and the BER is Ix 10-3 (oneerror in eac h 1000 bi ts), the ave rage num­ber of bad bit s in eac h packet will be I , andall of the pack et s wi ll be unu sabl e. Ofcourse, the BER only gi ves an avera ge ra teof errors. In reality, some of the packetswill have two or more bad bit s; some willhave none . So some of the packets will ar­ri ve int act. . .but not man y ! On the othe rhand, if the BER is Ixl O-s (one erro r in100,000 bits) the re wi ll be, on average. onebad bi t in ev ery 100 packets. You might noteve n notice that !

In a we ll-designed sys te m. the majorfactor that ca uses bit errors is noise. As thesignal ge ts weaker, the signal-to-noise ra­tio drop s and more of the transmitted bitsare corr upted by no ise , re sul tin g in im ­proper recept ion of the bit s. Any distorti onadded by the tran smitter and recei vercau ses even more bits to be recei ved erro­neou sly. It is th is distortio n added by theradio that we wa nt to measur e.

The test setups we used are shown inFigure I . Fo r recei ver testin g. the 9600­baud test signa l is used to modul ate aMarconi 204 1 signa l generator. which pro­vides distortion-free modul at ion of the sig­nal. In tra nsmitter testin g. transmi tted sig­nals from the radio being te sted are mixeddown to a low IF and dem odulated usin g awide -ba nd. low-distort ion dem odul atorbuilt by the ARRL Lab and described in thecom pan ion QEX artic le.

10-1 10- 1

10- 2 10- 2

0:: 0::w wm

10-3m

10- 3

10- 4 10-4

4 5 6 7 8 9 10 11 12 13 14 15 16

SINAD (dB)- 2 - 1 a 1

Frequency Offset (kHz)2 3 4

Figure 3-SER versus SINAD for the SER test box mixer/demod ulator.

2-2 Chapter 2

Figure 4-SER versus frequency offset for the SER test boxmixer/demodu lator .

Page 80: Packet- Speed- More Speed

1.5 2 2.5 3 3.5 4 4.5 5Deviat ion (kHz)

5.5 6

Frequenc y: 146.00 MHz (C120BDA)430.25 MHz (IC-B20H)

Deviation : + 3 kHz

1O-5 L----'--- -'-----''------'--_ -'------'_ ---'---_ -'--------'-_ --'-----'L----'--------'3 4 5 6 7 B 9 10 11 12 13 14 15 16

SINAD (dB)

Figure 5-BER vers us deviatio n for the BER test box mixer/demodulator .

Figure 6-BER versus SINAD for the ICOM IC-820H andStandard C1208DA receivers .

Th e BER test bo x generates a pseudo­random stream of bit s using the same algo­rithm imp lemented by the scrambler cir­cuit in the K9NG and G3RUH modems.Thi s data strea m is formed int o a G3RUH­type signal which is applied to the signa lgenerator (for rec ei ver testing) or the tran s­mitter under test (for transmitter testing).After passing through the system and beingdemodulated, the signal returns to the BERte st bo x. The test bo x insp ect s the incom­ing signal and compares it to the signal thatwas sent. To accomplish thi s, the test engi­neer adj usts a delay value in the BER testbox that matches the delay through the sys­tem. If there are no errors induced by thetest system, the returned bit will always bethe same (0 or I) as the bit se nt.

Counting Errors

Bit errors are by nature random . Thatmeans that if you run the same number ofbit s through the same sy stem in two differ­en t test s, you will likel y ge t different re­sults . But the average rate of errors shouldbe co ns istent. That means that if you sendenough bit s through the sys tem, you 'll getrep eatable results . So, how many bits doyou need to se nd? It turn s out that you needto send eno ugh bits to get a ce rtain numberof erro rs. How many errors yo u need to ge tdep end s on how co nf ident yo u want to beof the results . If you get 100 bit errors, youcan be 99 % sure that the measured BER iswithin a factor of about 1.3 of the value youwo uld ge t if yo u se nt an inf ini te number ofbits thr ough the sys tem (the " true" BER ). Ifyo u ge t only 10 errors, you ca n be 95% surethat the BER is within a facto r of 2 of thetrue BER .? Our test technique then , is tosend up to I million bit s th rough the sys ­tem . Th at let s us measur e BE Rs of IxIO -4( 100 er rors) or more with a high degree ofcon fidence and BER s of Ix I0-5 ( 10 er rors)with reasonable co nfidence . The BE R testbox se nds bit s through the sys tem 100,000at a time , and it always se nds at lea st

300,000, but if it gets more th an 100 errorsbefore reaching I milli on bit s, it stops thereand calculates the BER using the numberof bit s sent.

Tests Perfo rmed

To find out what effect the radios haveon the signal, we need to test them usingsignals that are somewhat noi sy. That way ,the err ors added by radio distortion becomequite apparent. We wanted to mea sure eachrecei ver at spec ific signal-to-noise rat ios.Since each unit exhibits a unique sens itiv­ity , a particular signal-to-noise ratio willoccur at a different input power level foreach receiver. To standardi ze the signal-to­noi se ratio, we first measured the SINADof the recei ver at its data output using aI-kH z tone and 3-kHz deviation . We sweptthe input pow er level from - 130 dBm to-90 dBm , resulting in a curve like that ofFigure 2. (Figures 2 through 5 are mea sure­ment s of the BER test box' s mixer/d emodu­lator, which act s like a receiver. Thi s cir­cuitry is used to test tr ansmitters, andtestin g the BER test box as though it is arecei ver shows us the performance of thetransmitter test setup. ) From that curve forany particular radio we can es tablish theinput sig nal level needed to produ ce a par­ticul ar SINA D.

For receive rs , we per formed three swe ptBER tests and one sing le-sta te test. Fir st ,we mea sured the BER using a 3-kHz peakdeviati on and a signal at the nominal oper­ating frequ ency of the recei ver. The inputpower level was stepped to measure BE R atSINA Ds from 3 to 16 dB . Thi s produces acurve like that of Figure 3. Next , we set theinput level to that needed for a 12-dBSINA D and swept the signa l frequ enc yfrom 4 kHz below the nominal operatingfrequency to 4 kHz above, producing acurve like that of Figure 4. Thi s tests thereceiver' s tolerance offreq uency erro r. Ourfin al swept mea surement wa s made at theoperating frequ enc y but with the peak de-

viation stepped from I kHz to 6 kHz , to testthe recei ver's tol erance of over- and under­deviated signals (Figure 5). Our final testwas to measure the BER on-frequency with3-kHz de viation and a -50 dBm input sig­nallevel. Thi s is a strong signal that shouldresult in an ex tremely low BER.

Tr ansmitter testing was somewhat sim­pler. Usin g the Lab-built mixer/demodula­tor in the BER test box , we attenuated thetran smitter output to get a 12-dB SIN ADfrom the test box mixer/demodulator andmeasur ed the BER using I million bits. Wethen increased the tran smitter signal intothe mixer by 30 dB and measured the BERagain. The latter test should result in essen­tiall y no bit errors if the transmitter is dis­tortion-free .

As yo u' ll see , the results of our test s arenot parti cularly enco urag ing . For thi s rea­son, we decided to mea sure the amp litudeand pha se responses of all of the radios , tosee if we could find out why some radiosproduced such poor BER res ults . The sys ­tem used to make the respon se measur e­ment s has been described previou sl y inQEX.8

To provide a final check of our testing,we spot chec ked so me of our mea surementswith a G3R UH modem , usin g its BER testmode to se nd and receive the data signals .The results showed a small, con sisten t im­provement in measured BER with theG3RUH modem compared to our test set.Thi s is acco unted for by the input low-pa ssfilter of the G3 RU H modem . It cuts of f at alower frequency than the fi lte r in our tes tset, which removes so me of the noise. Thesig nal-to- noise rat io- and the BER-istherefore improved slig htly .

Test Results

We tes ted fo ur previously rev iewed2-mete r FM radios : an ICOM IC-2 8 IH,Kenwoo d TM-25 1A, Standard C 1208DAand a Yaes u FT-2500M. The FT-2500M re­quires modification for 9600-baud opera-

Theory/Design 2-3

Page 81: Packet- Speed- More Speed

tion, as detailed in its manual. (A modifica­tion kit is available from Yaesu.) We alsotested an ICOM IC-820H 144/430-MHzmultimode transceiver. And finally, we in­cluded a TEKK KS-900 70-cm data radio,which has not been reviewed previously inQST. Thi s unit is a crystal-contro lled, 2-Wtransceiver desig ned specifically for FSKdata operation, not voice. (TEK K primarilysupplies radios to commercial acco unts andgenerally perfers not to sell direct to indi­viduals. TEKK will gladly refer prospecti veAmateur Radio customers to its retai l deal­ers. You can contact TEKK at 226 North­west Pkwy, Kansas City , MO 64150; 816­746- 1098 or fax 816-746-1093.)

Receiv er Test Results

The test result s for all of these radios areshown in Tab le I , in the form at we ' ll use toreport BE R measurements in fut ure re­views. For receivers, we 've selected threesigna l level s at which to report the BERexhibited by each recei ver. The 12-dBSINAD BER is used as a referen ce for othermeasurements and is also a level at whicheven an ideal recei ver will exhibit a BERof greater than IxIO -4. Th is co lumn ofTable I shows gross differences in weak­signa l perform ance between receive rs. The16-dB SINAD level is one at which exce l­lent receivers should show a BER of lessthan Ix I0-5. As Table I shows , some re­ceivers do show such perform ance, somedon ' t. Finall y, a measurement of BER witha-50 dBm input signal level should resultin a BER of less than Ix I0-5 from any re­ceiver that is even marginall y usable, andall of the rece ivers tested met thi s criter ion.

One should be carefu l about making toomuch of relati vely small differen ces be­tween rec ei ver s. For example, Figur e 6shows the full swept BER versus SINADfor two of the tested receivers : the ICOMIC-820H and Standard C 1208DA models .These two unit s show the greates t differ­enc e in their receiver BER per forman ceaga inst noi se, with the IC-820H having the

better pe rformance, but the di fferenceworks out to be equal to only a few dB ofSINAD. So, it would be wrong to think thatthe performance of the IC-820H receiverwill be dramaticall y different fro m that ofthe C 1208DA in pract ice. Wh at these testsdon't show, however, is the effect of re­ceiv ing a signa l that is already slight ly dis­torted. In ge nera l, the BER ofthe C 1208DAwill degrade more fro m a distorted inputsig nal than will that of the IC-820H.

Not show n in Table I are the result s ofour tests of frequency and deviation toler­ance. Norm all y, we will report only sig ­ni fic ant resu lts. For exa mple, Figure 7shows the BER vers us frequ enc y for theIC- 820H and the C 1208DA. Of all theradios tested, the IC-820H was the mosttolerant of frequency error. Thi s to leranceca n be helpful when you are trying to workthe 9600-baud packet sa tellites , with theirco nstantly changing, Doppler-shifteddownlink frequen cy. The C 1208DA, on theot her hand , can acce pt onl y a rel ati vel ysmall frequ ency error before performancedegrades substantially. Thi s, too, has itspositi ve side, as it indic ates that the band­width of the C 1208DA receiver is relativelynarro w, leadin g to improved adjacent­channel rejection .

All of the se rad ios not only tolerate sub­stantial overdeviati on , they perform betterat higher deviations. Fig ure 8 shows theBER vers us dev iatio n of the Yaes uFT-2500M (after modification), which isrepresent ative in that the BER decrea ses(improves) as deviation is increased. All ofthe radios showed substantia lly improvedperformance at 5-kHz deviation comparedto the usual 3-kHz deviation. Th is reall yshouldn' t be a surprise . Consider that inFM voice communication, we use a devia­tion of 5 kHz to carry voice signals thathave a frequen cy cont ent of up to about3 kHz. In 9600- baud packet , our spec ­trum goes up to 7.2 kHz (albeit at fairlylow levels tow ard the top of the spectr um)and the de via tion is only 3 kHz. Our mod u-

lat ion index is there fore much lower for9600-baud packet than for voice . Seeingthis, you might be tempted to crank up thedeviati on to get better perform ance, butthat' s not genera lly a good idea . For onething, the tolerance of frequency erro r isdepend ent on the deviation being 3 kHz .Increasing the deviat ion redu ces the toler ­ance of frequency error. More important isthe band width of the tran smitted signal.The main reason we use 3-kH z deviationis so that our signals will stay within a20-kHz-wid e channel. If you increase thedeviation , your tran smitted signal will"s lop ove r" into adjacent channels. A bet­ter fix if your system isn 't performing asyou' d like is to install a better antenn a or toincrease transmitter power. However, forpoint-to-point network backbone link srunning in wide channels (such as the100-kHz-wid e channels at 70 ern), wherethe frequency offset can be controlled sincethere are a limited numb er of stations usingthe link , it may well make sense to .increa sethe deviat ion to 4 or 4.5 kHz.

Th e bott om line for these receivers isthat they all are capable of recei ving 9600­baud pac ket signa ls. There are perfor­mance di ffer ences, however , and thesedifferenc es may become more significa ntwhen distorted, rea l-world signals are be­ing recei ved.

One thin g about the IC-820H should benoted . The imp edance of its data output isvery high-nearl y 50 kn . Th is meansthat the modem or TNC connected to theIC-820H must have a high- imped ance in­put. It es pecia lly must have an impedanc ethat is constant across the frequency ran geofthe 9600-baud audio signal. The G3RUHmodem does not have a constant imped­ance acros s this range. At low freq uencies,its input imp edance is near 100 kn, but itsimpedance dec reases at higher frequencies.For example, its impedance at 4 kHz isabout 55000-)45000 n. With this differ ­ence in impedance acros s the frequencyrange, the effec t of the load is to distort the

- 2 -1 0 1 2Frequ ency Offset (k Hz)

- 3lO- SL _ -L_ _ l-_---.L__.L-_-L_ _ ..L._----:_----'

- 4

Figure 7-Frequency tolera nce curves for the ICOM IC-820Hand the Standard C1208 DA receivers.

Figure 8-BER versus deviation for the Yaes u FT-2500M(modified) rece iver.

2-4 Chapter 2

Page 82: Packet- Speed- More Speed

Table 1Measured Performance of 9600-Baud Radios

144/430 -MHz multimode radio :ICOM IC-820H -113.0

Model

2-meter FM radio:ICOM IC-281HKenwood TM-251AStandard C1208DAYaesu FT-2500M

440-MHz data-on ly radio :TEKK KS-900

12-dB SINADlevel (dBm)

-113.2- 114.3-108.8- 113.3

- 110.4

Rece iverBER at 12-dB BER at 16-dB BER atSINAD SINAD -50 dBm

4.8x10-4 <1x10-5 <1x10-5

3.6x10-4 1.3x10-5 <1x10-5

3.9x10-4 1.7x10-5 <1x10-5

8.6x10-4 <1x10-5 <1x10-5

2.7x10-4 <1x10-5 <1x10-5

2.6x10-4 <1x10-5 <1x10-5

TransmitterBER at 12-dB BER at 12-dBSINAD SINAD +30 dB

7.4x10-3 1.1x10-3

2.2x10-3 2.9x10-43.7x10-3 7.0x10-44.9x10-3 3.8x10-4

5.1x10-4 <1x10-5

3.8x10-4 <1x10-5

The Bottom Line:How Good is Good Enough?

The values in Table 1 tel l the story of how each of the tested rad ios per­forms. But what , ultimately, do thes e numbers mean in terms of on-the-airperformance? That's ha rd to say, specifically. Wha t we can say is that lowe rnumbers mean better performa nce. (Be sure to take the exponent of 10 intoaccount whe n comparing values.) The strong-signal numbers (at -50 dBm forrecei vers a nd at 12-dB SINAD plus 30 dB for tra ns mitte rs ) are perhaps the bestindicators of the general suitability of the rad io for 9600-baud operation. Inthese columns, a rad io should be able to exh ibit a BER of 1x10- 5 or less. AsTable 1 shows, all of the receivers we te s ted do so . But on ly two of the trans­mitters produce this low BER. All of the other transmitters produce a BER that isa t least 10 times higher . In sid e-by-side comparisons over actual radio links,these high-BER tra ns mitters a re likely to perform substantially less well than thetransmitters that achieved a BER of less than 1x10-5 .

signal comin g from the IC-820H . Thi s canbe see n in Figure 9, which shows the BERvers us noise of the IC-820H both with andwithout the G3RUH modem loadin g thereceiver output. The effec t isn' t large in thiscase, but it is present. This sugges ts thatyou should check the output impedance ofthe radio and the input impedance of themodem or TNC before connec ting themtoge ther. If the rad io' s output impedance ishigher than the TN C' s input impedance,you may want to add a simple op-arnp orone-transistor buffer between the radio andthe modem.

Transmitte r Test Results

Although there are some differences inreceiver performance, it is in transmitterperformance that we see huge differencesamong the radios tes ted. Our measurementswere made with both noisy ( 12-dB SINAD)and strong (l2-dB SINAD plus 30 dB) sig­nals. As a refere nce, we measured the BERat these levels using a signal generator asthe "transmitter." The near-p erfe ct signalfrom the genera tor produced a BER of3.33 x I0 -4 at 12-dB SINAD , and no bi terr ors (BER < Ix I0-5) at a leve l 30 dB

greater. Each ofthe transmitt ers tested gavehigher BERs than the reference signal gen­era tor, as you would expect. The numbersare shown in Table 1. Remarkable amongthese were the IC-820H mult imode radioand the KS-900 data radio. Their transmit­ter per formance was nearly as good as thatof the refere nce signal generator. None ofthe 2-meter FM transmitters achieve d a lowBER, eve n with a strong signal! This is

quite troublin g, if not entirely unexpect ed.The main difficulty is in the low-fre­

quenc y response of the transmitt ers. Un­like voice signals, 9600-baud packet sig­nals conta in significa nt co mpo nents atfrequencies below 100 Hz. This poses aprobl em for a synthesized tran smitter, asexp lained in the sidebar, "Making Tr ans­mitters Speak 9600 Baud ."

While we can understand the difficul-

- 50o 200 400 600 800 1000 1200 1400 1600 1800

Frequency (Hz)

Frequency : 430.25 MHZ~

10-1 f-.-. J I 1 1 1 1 1 IOeviati on : ±3 kHz

10- 2 1 1 I 1 1 ~ 1 1 1 1 1 1 1 Im~

'" "~ 10- 3 1 1 1 1 1 1 1 ""'-{ 1 1 1 I

-c:ea.

IC-8 20H ,~ Ewith Modem -c

10- 4 1 1 1 I I I 1 1 I'\.~ 1 IIC- 820HNo Load

10- 5 1 1 1 1 1 1 I I 1 1 \I .... -l3 4 5 6 7 8 9 10 11 12 13 14 15 16

SINAO (dB)

o- 5

- 10

-1 5

- 20

- 25

- 30

- 35

- 40

- 45

1\ ~-- ~

IC-2 81HI

Figure 9-BER versus SINAD for the IC-820H, both without aload on the receiver data output and with a load consisting of aG3RUH modem .

Figure 10-Amplitude res ponse (to 1800 Hz) of the IC-281HandIC-820H transmitters.

The ory/Design 2-5

Page 83: Packet- Speed- More Speed

Making Transmitters Speak 9600 BaudProperly modulat ing an FM transmitter with a

G3RUH-type signal requires a transmitter modulationresponse that is flat in amplitude and linear in phasefrom a very low frequency (near dc) up through about7200 Hz. This is easy to do if the oscillator beingmodulated is a free-running VCO or VXO, but can bequite difficult if the oscillator is embedded in a phase­locked loop (PLL). And these days, the frequency­controll ing oscillator in most transceive rs is part of aPLL.

The problem arises because the PLL will try to"correct" any offset between the oscillator's operat ingfrequency and the programmed frequency of the PLL. APLL contains a low-pass loop filter, and only thosefrequency components of the modulating signal that fallwithin the loop filter will be affected by the PLL. There­fore, some of the modulati ng signal goes throug huntouched and some doesn't. The result, when thesignal is received at the other end of the link, is adistorted signal that may be unusable by the receivingTNC.

Another approach to modulating the PLL is tomodulate the reference oscillator. In this case, the PLLforces the VCO to track the change in the reference'.This time, only those frequency components that fallwithin the loop filter find their way onto the transmittedsignal. That's no better. It might seem that a simplechange to the PLL's low-pass loop filter could solve thisproblem, but that will lead to problems with loop lock-up

time, capture range or noise performance. In somecases, loop instability might even result.

The obvious solution to this problem is to apply thelow frequenc ies to the reference oscillator and the highfrequencies directly to the VCO.'This two-point modula­tion scheme can work, but you have to be very carefulhow you split the modulat ing signa l into its high- andlow-frequency components so that the signal isn'tdistorted in the splitting . It isn't easy .

The ideal solution would be to modulate a VXO, thenmix, or heterodyne, that signal with the output of anunmodulated synthesizer to produce the signal to betransm itted. This requires additional circuitry, of course.But consider an all-mode (CW/SSB/FM) radio like theIC-820H. To generate the SSB signal , such a radioalready generates a fixed-frequency signal and thenheterody nes it up to the output frequency. So theadditional complexity needed for good 9600-baudmodulation is already largely prese nt, Of course, youmay not want all-mode capability. If you are interestedonly in packet operation, the cost of the additionalcircuitry that an all-mode radio uses to do CW and SSBmay be prohibitive.

The radios we tested for this article aren't the onlyones on the market, of course . Elsewhere in this issue(Product Review) we report on an Azden PCS-9600DFM/packet transceiver, and other manufacturers arebringing out new data-radio desiigns quickly .,

;

ties of the manufacturers who are trying toadd 9600-baud capability to their voicerad ios, our understanding doesn 't changethe fact that the transmitter performance ofmost of these radios makes it difficult tosuccessfully use 9600-baud packet. This isparticularly true when the station at theother end is using a recei ver that exhibitsmargi nal performance.

Figure 10 shows the amplitude responseof two of the tested transmitters, theIC-820H and the IC-281 H. The lowest fre­quency point plotted on the graph is 22 Hz.The IC-820H is essentially flat from 22 Hzup, while the IC-281 H significantly booststhe lower frequencies . Of equal concern isthe variation in group delay with frequency(not shown). Below 100 Hz, the IC-281 Hdisplays a large variation in group delay,which is not unusual when large changesin amplitude response are pre sent. TheIC-820H performs much better in this re­gard, and this performance is mirrored inits outstanding transmitter BER numbers .

The G3RUH modem provides a wayof precompensating the modulating pulses,so it is possible that better transmitter per ­formance can be obtained from some ofthe se radios by selecting an appropriate

2-6 Chapter 2

pulse shape from the ROM in the modem.This requires some experimentation. TheG3RUH modem generates a test signal youcan use for thi s purpose in its BER testmode . By receiving the transmitter' s signalwith a known good receiver, you can selectfrom among the avail able pulse shapes toget the one that deli vers the best perfor­mance. But some newer 9600-baud TNCsdon 't have this facility. TNCs built usingsingle-chip 9600 -baud modems depend onproper performance of the transmitter.

Conclusion

In future reviews of 9600-ready radios ,we will present data in the form of Table I.From the three receiver BER values andtwo tran smitter BER values for each radioyou can get a good idea of the relati ve per­formance of these products . We'll only re­port in QST on frequency and deviationtolerance results if the measurements showsomething unique about the radio beingtested .

The readiness of the crop of 9600-readyradios we tested is mixed. Of the radiostested, only the KS-900 data radio and theIC-820H multi mode transcei ver showedboth transmitter and recei ver performance

that will result in efficient packet opera­tion at 9600 baud. Of the 2-meter FMtransceivers, receiver performance was ac­cept able in all cases-if not always out­standing-but transmitter performancewas disappointing.

Notes"steve Ford, WB8IMY, "QST Compares:

2-Meter FM Mobile Transceivers," QST, Jan1995, pp 70-76.

2Steve Ford, WB8 IMY, "ICOM IC-820H VHF/UHF Multimode Transceiver," QS T, Mar1995, pp 80-83 .

3Jon Bloom, KE3Z, "Measuring 9600 -Ba udRadio BER Performance," QEX, Mar 1995 ,pp 16-23 .

4Steve Goode, K9NG, "Modifying theHamtronics FM-5 for 9600 bps Packet Op­eration," Fourth ARRL Amateur Radio Com­puter Networking Conference, (Newington:ARRl,1985) .

5James Mille r, G3RU H, "9600 Baud PacketRadio Modem Design," 7th Computer Net­working Conference, (Newington : ARR l,1988) .

6leon W. Couch III, Digital and Analog Com ­municstion Systems, (New York: Macm illan,1993), pp 179-180 .

7Miche l C. Jeruchim, Philip Ba laban andK. Sam Shanmugan, Simulation of Commu­nication Syste ms, (New York: Plenum Press,1992), pp 498-501.

8Jon Bloom, KE3Z , "Meas uri ng SystemResponse using DSP," QEX, Feb 1995,pp 11-23.

Page 84: Packet- Speed- More Speed

By Tim RileYj Dennis Bodson, W4PWFj Stephen Riemanj and Teresa G. Sparkman

AComparison ofHF Digital Protocols

The world of HF digital communication is a confusingmix of protocols. Which offer the best performance?

Figure 1-Block diagram of the HF modem test setup.

Level and/or Channelimpedance Simulator#1

Modem#1 converter, Modem#2if necessary~put Output....

Output - - -~ h -Audio .... . Audio

Input

Input Output- Simulators require: I -L,...J o dBm, 600n , balanced signals I L,..J__ _ __ _ _ _ L_

-I I

ChannelI

Level and/or I SerialSimulator #2 impedanceconverter, I ~ Protocol - I--

it necessary I Analyzerl{

~utput Inpu~

P-I

t..... - t- - - I

I

GPI B Control EIA.232

486 Test Controller(IEEE . 88)

EIA 232/ISABus0 [ EIA 232/1SA BusLabViewSoftware

l1lirecent years our HF dig italsubbands have become Towe rsof Babel with several protocol sco mpeting for dom inance. The

familiar list includes AMTOR, G-TOR,CLOVER, PACTOR and PACTOR ILl Eachmode has its enthusiastic supporters, butwhich ones are most efficie nt when it comesto transferrin g information?

The task of evaluating HF digital protocolsis more than just an academic exercise. Ourgovernment has a stake in determining theanswers because HF digital operators have tra­ditionally supplied a large reservoir of backupcommunications services and operators dur­ing emergencies. In the interest of tapping thisresource, the National Communications Sys­tem (NCS) sponsored the National Telecom­munication and Information Administration(NTIA) to test various HF modems and estab­lish a technical baseline of standardized per­formance data on HF modem protocols. TheFederal Emergency Management Agency(FEMA) will use those test results to deter­mine which protocols could be used to serve asan interchange with the amateur community inthe event of a national emergency. This articlereflects the resultsof tests conducted at NTIA' sInstitute for Telecommunication Sciences(ITS) laboratories in Boulder, Colorado.

Solving the Propagation ProblemMany over- the-air tests have been con­

ducted on vario us HF modem/protocol com­binations. Atmos pheric conditions vary somuch, however, that it's impossible to drawsolid conclusions. Over-the-air test resultscan change from day to day, or even hour tohour. To obtai n scientifically valid informa­tion, you need stable cond itions. Unfortu­nately, Mother Nature refuses to guaranteethe stabil ity of HF propagat ion!

So, we turn to the laboratory, where it'spossible to create elaborate simulations of HFpropagation paths. The engineers at ITS havedeveloped an automated test bed for use inconducting controlled laboratory testing ofmodems and their protocols. Using this testbed, ITS subjec ted the protocols to a repeat-

able set of simulated propagation paths over awide range of signal-to-noise ratios (S/Ns).2

Data transfers were performed at variousSINs for each of the five protocols in theirAutomatic Repeat Request (ARQ) mode. Allfiles received were checked for erro rs to de­termine the validity of the tests. The through­put, a measure of the data transfer rate, wasmeasured for each protocol under simulatedionospheric conditions.

To conduct our tests we obtained HFmodems manufactured by Kantronics, HALCommunications, Advanced Electronic Ap­plications and SCS. However, specific mod­els are not identified in this article. You' llsee them referred to as modems A through Donly. The intention is to evaluate protocols,not hardware. It is important to note that theuse of specific hardware does not imply arecom mendat ion or endorsement by the Na­tional Telecommunications and InformationAdministration, nor does it imply that theequipment used is necessarily the best avail­able for this application.

Test ConditionsFor each test, two identical modems (see

Table I), controlled and monitored by one

80486-based PC, were physically connectedto each other by their audio-out/ audio-inports through two ident ical HF ionosphericchannel simulators (Figure I). Although themodems were both connected to the same Pc,the operat ion of one was entirely indepen­dent of the other.

The modems were linked to the computerthrough the PC' s serial communication ports(COM1and COM2), except for the CLOVERmodem, which plugged into the PC' s ISA bus.

Table 1HF Protocols TestedNote: The AMTOR and PACTOR protocolswere test ed on two modems to dete rmine ifthere were implementation differences in theresu lts. The other three protocols wereimplemented by unique modems . Modemsare ide ntified s imply as A, S, C and D.Protoco l Modem Computer InterfaceAMTOR A and S Serial portCLOVER C PC bus (plug-in card )G-TOR D Serial portPACTOR A and S Ser ial portPACTOR II A Se rial port

The orylD esign 2-7

Page 85: Packet- Speed- More Speed

Which Parameters ?Because of the wide variety of protocols

and the mann er in which they are imple­mented , we didn 't have enough time to test allpossib le parameter combinations. Becausethe goa l of the tests was to determine the suit­ability of the modem/ protocol comb inati onsfor use during eme rgency situations, we choseto test the prot ocols in their most robust mode.This meant testin g only the ARQ mode. Be­cause the ARQ mode is inhere nt ly error -free,there is no need to calc ulate error perfor­mance . Several of the protocols implementedalgorithms for choos ing modulation schemes .Otherwise , when manu al sett ings were re­quir ed, op timum settings were chose n base don the manu facturer ' s recommendat ions.

Seven or Eight Bits?Because data can be transmitted in either

7 or 8-bit form, the type of test data usedaffects the thro ughput de livered by eac h pro­tocol. Binary data and mixed -case ASCIIdata req uire a higher ove rhead when trans­mitted by a 7-bit protocol, thereby loweringthe effective thro ughp ut. FEMA supp lied anumbe r of files it viewed as being "ty pical"traffic . These included both straig ht text andbinary (wo rd-proc essor format) content. Tominimize the adva ntage 8-bit prot ocols haveove r 7-b it protocols, stra ight text fi les werechose n over binary files. When test ing theAMTOR protocol , which transfers data as

operat ion. The fi rst was a test control fil e ,contai ning a ser ies of values represe nting thesettings of the two simulators, the sett ings ofthe two modems and the file to be tra nsmittedduri ng eac h test. Result s of the test were wri t­ten to the second, a master log file.

For each tes t, LabView performed the fo l­lowin g tasks :

I . Rea d the desired values of simulatorsetti ngs, modem/pro toco l modes and param­ete rs, and the file to be transmi tted fro m thetest contro l fi le .

2. Set the sim ulators to the desired values .3. Initialized both modems to the desired

mode.4. Esta blis hed a link between the two

modems .5. Star ted a timer.6. Loaded the buffer of the tra nsmitting

modem with the tes t file.7. Monitored the trans mission and rec ep­

tio n of the fil e.8. Stopped the timer when the end-of-file

was received .9. Co mpared the received data to the

or iginal fi le to de termine the success of thetest.

10. Calculated the thro ughput and ap­pended the test infor mation to the master logfile.

11. Flushed the modems' buffe rs in prepa­ration for the next test.

LabView genera ted a front-panel display(Fig ure 2) that allowed the opera tor to moni­tor the progre ss of the test. Thi s "panel" alsosupplied controls to permit the ope rator torun tests man ually.

r.I1Rle ~)

Unk'04 .)

><Ie. In !'rag r... 0 )

eStop Time

Thruput

Current Teet

Byt.. Sem

Byte. Rec:evled

SlIrtTIme

o

o~

contro lled thro ugh the use of the NationalIns trument's prog ram Lab View (ve rs ion3.1.1) , a Windows -based applica tion capableof controlling, moni toring and measuringlaboratory tests.

Two files were used to auto mate the test

InltflUzed 2

Call Sign Z

I 12<41

con SIt" 1

22 19S07.2US 12Z ..,~ar;

23HAl. 06121i9S012Z3<!01 ::3:00 \'&.116 0<

It"lf

Uh,," au "11'" 'I.-II. 11 lit Ii I •• ".11 I It til" '10 I III • "" ,1,," It Ill, ' IUIIt·, "Iu. I.. I un.' •• I I'" 1.,1 111111.11 11"~ '" t III ,II. I 1.-.l"onnUhH.)tlun' 'fllin. 1""l.'IIIIII'1 11-"" I., till" 11111 • In '-),.11'" 41,"Villi .....1.•1.1. 11111 "'"1" 1,10.,11111' .11 10111 "II' 1111'" 1111 1'11I 'I I,·,.

111_, "II'" 11".1 Ill,' If IUl till IIII' 1111"II ••llllllhil II '''11 I" 1111111 lull

1"11I" lit IJ 1111. II rnu-. "uUI It I 'I '"1 .111' "UI • t" III II ••111.1 '

11'/1'

l'fIl"1 ,III t'nI'I'III'flIlI ' , 111 11 .1' .111111111.1111 "'1111 • I'hl,t"'11 k,"'11lnpr-rat m 'll''-''O''.I 1:11111.,1 101'1." Iltlt-: 1(1 "ll dl .. 1 1"1

• uJll1lJIIlI. •11 tun- rn rtn , l"lJ"1dlll'l II.' ,.. III I Iu- tUII ••11 ,')0:1'til' lu,·

uul .....J••I I ••• flll 11""')1' (1I1'lJ.Ulllt 1111111 1I1t1 Illt'lI 0..1 j'lI, lJ 'I"I,

r ent r-.. ll'J ')1"" ,tl,·.t hv rrllft t.1I till" 1111. t .·I.I ..... llIi, ••1 l'U, l.t Imlo"I'lll

"nl,,'lj"no:v II' 1),11'1 ~ tl 11111'" f'ulIl.tt I 11'1111 HIII"11I , 111 Iff rr , .11I,1

A serial protocol analyzer was used to moni­tor the ser ial port connection to the secon d(receiving) modem. A program running on theanalyzer determined whether the test hadtimed out (no data had been transferred for apredetermined period of time). The test was

6

'Sc00 5Q)

~~Q)

ti~ 4ttls:~ ----+-- CCIR Good"5 ------ CCIR Poorc.s:OJ 3::les:I-

28 10 12 14 16 18 20

Signal to Noise Ratio (dB)

6

'Sc00 5Q)

~~

~~ 4ttls:~ ----+-- CCIR Good"5c. ------ CCIR Poors:

3OJ::les:I-

28 10 12 14 16 18 20

Signalto Noise Ratio (dB)

Figure 3-AMTOR protocol, modems A and S , CCIR Good/Poor

2-8 Chapter 2

Figure 2-LabView generates this "contro l panel" on the PC monitor duri ng test ing.

Page 86: Packet- Speed- More Speed

40

2.0

MultipathDelay (ms)0.5

353025

Fading FrequencySpread (Hz)0.1

1.0

simulators ) were made to verify the modemsetup and determine the optima l throughput.Tab le 4 pres ents the clear -chan nel meanthrou ghput for eac h pro tocol and mod em .Since implementations vary , one prot oc olca n have sig nifica nt differenc es in throu gh ­put between di fferent mod els of modem s.With thi s in mind , a protocol was tested onmulti ple mod ems whenever possible. Thiswas the case with the AMTOR and PACTORprotocols only.

Th e clear -ch ann el throughput was usedto det ermine the limits of testin g. While theARQ mod e will theoretically pass data erro r­free , it also can require an unlimited num ber

Theory/Design 2-9

20

Signa l to Noise Ratio (dB)

15

Signal to Noise Ratio (dB)

CCIR Good

CC IR Poor

SIN (dB)

oto +20 (to +40for CLOVER)

o to +20 (to +40for CLOVER)

10

45

CC IR Good40 CCIR Poor

35

~~ 30

~.'!l 25~2!

20~

~a.s: 15goes:f- lO

00 10 12 14 16 18 20

50

40

~~~ 30..U!"2!~

I 20

er=

10

05

Test Limits and Time ConstraintsMea surement s ove r a clear channel (mo­

de ms connec ted back-to-back wit hout the

Figu re 5-G-TO R protocol, CCIR Good/Poor

CCIR Poor

message . Commonly used unit s of throu ghputare characters per second and words perminute.The throughput, in characters per sec­ond, refle cts the amount of time needed totransmit a file under the conditions of noise,mult ipath propagat ion and fadin g. The char­acters-per-second rate is used because thisre lates to the actual tim e need ed to transmit afile, and it includes the transmission ove rhead.Onl y tests in which the data was recei ved er­ror-free were inc luded in the result s.

CCIR Good

Table 2Simulated Ionospheric Conditions Test Matrix for the Two-Path WattersonModelIonospheric Condition

Figure 4-CLOVER protocol, CC IR GoodlPoor

Baud RateBaud ra te was another factor in modem

performance. Th e different baud rate sused for each protocol are summa rized inTa ble 3. Often the modems availab le forte sting dictated wh at baud rate could be usedwith a particu lar protocol. Most protocol sha ve a baud ra te at which th ey are co m­monl y used or are legally permitted to beused, while so me protocols automatica llyvary the ba ud rate, depending on conditions(as reflected by the received error perfor­manc e) . Some also va ry the baud rate as afun cti on of the modulation scheme . In allcases, the high est baud ra te is lim ited byFC C bandwidth re striction s. Test s wereperformed at the se typical baud ra tes, o r theprotocol was allowed to automatica lly se­lect the rat e.

The throu ghput, while related to baud rate,does not linearl y correspond to it. Variatio nsin overhead sizes (due to the use of error-de­tection and cor rection schemes) , rep eatedtran smission s, and variations in the numberof bits representing a sy mbol make signifi­cant differenc es in the throu ghput of a file or

Simula ted Propagation ConditionsThe selected pro pagation co ndi t ion s,

based on the Watterson model , were "CCIRPoor" and "CCIR Good" (see Tab le 2). TheSIN ranged from 0 dB to +20 dB (to +40 dBfor CLOVER). Bec ause the HF channelsimu lators are band -limited to a 5-kH z band ­width, thi s was the reference bandwidth formeasuring the SIN as record ed in the log file .

Compress ionAll of the HF modems we tested used som e

form of data compression , ranging from real ­tim e Huffm an compression to optimize dLempel -Zi v co mpres sion performed prior totran sm ission . In keeping with ITS ' s goa l ofchoosing an opt im um mode , compressionwas enabled thro ughout testing. However,this causes a sens itivi ty to the siz e of the dat afile, as well as to the type of dat a the fil econtains. Huffman compression operates ona small amount of buffered dat a, and is rela­tively inse nsi tive to file size and large datapatt erns. The CLOVER protocol uses com­pression software licensed fro m PKWARE,whi ch operates on the data prior to tran smi s­sion. The PKWARE softw are impleme ntsseve ra l compression schemes and, since itoperates on the fi le in its entirety , can resultin a more efficient and sop hist ica ted co m­pression. The refore, it is mor e effecti ve withlarger file s. In lig ht of the se differences, alarg e file of 15,183 bytes, which should favorneither compressi on method , was chosenfrom the selection supplied by FEM A. Whenco mpress ion is used, the data being tran smit­ted between the two mod em s is no longerstrictl y text , but includes nontext, binary val­ues, regardless of the file ' s ori ginal co ntent.

7-bit characters usin g the Baudot characterset, an all-uppercase vers ion of the test filewas used. For the other protocols, which use8-bit characters, a mi xed -case version of thesame file was used .

Page 87: Packet- Speed- More Speed

Figure 6-PACTOR protocol, modems A an d B, CCI R Good/Poor

Signal to Noise Ratio (dB)

Over-the-Air Test ComparisonIn 199 3, FEMA ran a series of ove r-the ­

air tests as pa rt of their prog ram to evalua tenew HF rad io produ cts .v' Sin ce FEMA an dITS tested two of the sa me mod em s and th reeof the sa me prot ocol s, an atte mpt ca n be madeto correlat e the results of both te sts.

FEMA chos e frequ encies characterized aseither good or poor , based upon the link qual ­ity analysi s (LQA) sco re supplied by an Au­tom ati c Link Es tablishment (A LE) trans­ce iver, when such a score was ava ilable.Wh en not ava ilable, RF output power wasadj us ted to simulate ap prop ri ate cha nnelconditions. Th ey estimated that a 10-d BSIN was the tran sition point bet ween goodand poo r conditions. Tes ts were co nduc tedacross a net wor k of thr ee nod es at FEMAfac ilities in Berryvi lle , Virgi nia ; Den ver,Co lorado ; and Urbana, Ill inoi s; on frequen­cies authorized for FEMA use. A co mpari­so n of the FEMA test s to ITS' s laboratorytest s is shown in Table 5.

The LQ A-ba sed channel qual ity designa­tor s of "Good" and "Poo r" do not corre spondto the CCI R definitions for path co nditions .Conse quently, you ca n't make a direct co m­pariso n of data . Because the 10-dB break­poin t between LQA Goo d and LQA Poor isonly an estimate, it ' s diff icult to split up thelaboratory measurem ent to appro ximate theLQA condit ions. The refo re, a genera lizedcompari son between the range of throughputsmeasured in the lab to the number s obta inedin the field is made, subject to these caveats .

Th e SITOR protocol was not tested in thelab because SITO R is a military ver sion ofthe AMTOR protocol. Whi le some di ffer­ences in the two protocols may ex ist, ITSdecided to drop SITOR from the tests, aga indue to tim e co nstraints. However, the SITORfie ld res ults are reason abl y cl ose to th eAMTOR lab res ults . PAC TOR II had notbee n intro duced at the time of the 1993 tests .

In general, higher th rou ghputs were re­corded in the laboratory tests than in the fieldtest s. Th e laborato ry tests a llowed fo r veryhigh SINs tha t would reflect optimum (if notunreal istic! ) rea l-wor ld conditions and werenot affec ted by the rapid ly cha ng ing ion o­spher ic co ndi tio ns that cha lle nge real HF

need ed to develop the dri ver s and initia liza ­tion software nece ssary for each protocol!modem combination .

ResultsThe data co llec ted during these tes ts is

summarized in the plot s in Fig ures 3 through7. All plot s displ ay values of thr ou ghput (incharact ers/second ) vers us SIN (in dB).

The plot s show the me an va lue ofth roughput for each SIN measured . Th e goa lwas to transmit a fil e three tim es for eachSIN value. During deve lopmen t of the testproc edure, more th an thr ee passes weremade at certa in settings. However, at lowSINs, less tha n three passes are inc lude d inthe plots du e to errors and ab orted tests .Again, the lack of tim e pre clude d us fromrunning a test unti l thr ee good completionswere accomp lished .

20

20

18

18

CCIR Good

CCIR Poor

16

16

14

14

Modem B

12

12

tions, it is of littl e value during emerge ncysitua tio ns, when rapid and reliable transmi s­sion of informati on is cri tical.

For an example of how time-consumingthese tests are, con sider the worst case : theAMTOR prot ocol. A suite of test s consist s of126 file transfers (2 1transfer s covering SINsof 0 dB to +20 dB in I-d B increme nts , for twoCCI R co nditions , with three attempts eac h).At best, a throu ghput of six characters perseco nd needed 42 minutes to transfer a15, I83-byte file, but as the SIN decreased ,the tran sfer time increased. Six days wererequired to complete the AMTOR protocolsuite of test s! Thi s does not include the time

10

10

Table 4Clear Channel Mean ThroughputProtocol Clear Channe l (optimal)

Throughput (char/sec)AMTOR (modem A) 6 .13AMTOR (modem B) 6 .26CLOVER 45.60 (robus t bias) ,

56 .87 (normal) , 69.05 (fast)44.6925 .4615 .67

133 .53

G-TORPACTOR (mode m A)PACTOR (mod em B)PACTOR II

8

8

6

6

Test BaudRates

100AUTOAUTO200AUTO

4

Signal to Noise Ratio (dB)

4

-- CCIR Good--.- CCIR Poor

Modem A

2

2

10092 to -550100,200,300100, 200200 to 800

25

'0c

200oCD

~CD 15<3~ctl.cSo 10:;a.s:Cl:::J 5es:f-

00

25

'0c 200oCD

.Ii!~CD 15<3~ctl.cSo 10:;a.s:Cl:::J 5es:f-

00

of ret ransmi ssions dur ing poor signal condi­tion s. Eve n und er moderate co nditions, oc ­cas ional retries and assoc iat ed hand shak ingwill extend transmission time. As the trans­mission channe l is degraded, the test time(time to pass a sing le data fi le) lengthens. Atso me po int, the tran smi ssion ti me (a ndthroughput) drops belo w a usabl e, pra cticalleve l. Due to the limited time available toperform all of the desired tests, some bre ak­point had to be se t to restrict unnecessar ytesting. We decided to end testing when thethro ughp ut had dropped be low 10% of itsclear-channel value. Whi le lower thro ugh­put may be usab le in non emergency situa-

2-10 Chapter 2

AMTORCLOVERG-TORPACTORPACTOR II

Table 3Baud Rates TestedProtocol Typical Baud

Rates

Page 88: Packet- Speed- More Speed

Signal to Noise Ratio (dB)

Figure 7- PACTOR II protocol , CC IR Goodl Poor

Notes1No doubt you'll notice tha t HF packet is miss ­

ing from the lis t. Although it is popular on VHFand above , packet is often a poor choice forHF communicat ion. This protoco l is not par­ticularly robust. It requ ires exce llent signalsat both ends of the path and also stable con­ditions for reasonable efficiency. With this inmind, HF packet was eliminated from consid­eratio n as a "se rious" conten de r.

2Two narrowband , Watterso n model HF propa­gation chan ne l simulato rs created the iono­spheric propagation conditions for each test.Two degraded condit ions were used : CCIRGood and CCIR Poor.

3H. F. Wetze l, "HF Modem Over-the-Air Pe r­formance Test Report, Globe Link Corpora­tion HF Modem Mode l GLC-1OOOA," FederalEmergency Management Agency , Was hing­ton , DC.

4H. F. Wetze l, "HF Modem Ove r-the-Air Pe r­formance Test Report, HALComm unicationsPCI-4000 HF Modem," Federal EmergencyManagement Agency, Washing ton , DC.

IQ!n-.-1

Theory/Design 2-11

Dennis Bodson , W4PWF, ARRL Roanoke Divi­sion Vice Director, is employed by the NationalCommunications Sys tem, 701 South CourtHouse Rd,Arlin gton, VA 22204-21 98. Tim Riley,Stephen Rieman and Teresa Sparkman are em­ployed at the Institute fo r TelecommunicationSciences, 325 Broadway, Boulder, CO 80303.

facturer was shipping un its . Con seq uently,multiple testing was performed only on thePACTOR and AMTOR protocols.

The throughputs of PACTOR andAM TOR were me asured on two modemsfrom different man ufacturers . As can be seenin the plots in Figures 3 and 6, there is a dif­fere nce in the two impl ementations of eachprotocol. In the ca se of PACTOR, the differ­ence is significant; the thro ughput of modemA is ne arly twice that of modem Beventhough the test conditions were identical forboth modems. No attempt was made to de ter ­mine why thi s differe nce exi sted, but it ra isesquestions regarding interoperability of mo­dem s from di fferent manufacturers using acommon protocol.

In testi ng the two imp lementat ions of theAMTOR pro tocol s, functional differencesemerged during the test development phase.One modem appeared to tran smit onlytho se charac ters contained in the BAUDOTcharacter se t. Undef ined characters wererecei ved as a qu estion mark . T he othermod em could tran smi t add itional (punctua­tion ) ch aracters, apparently using the char­act er positions the BA UDOT set listed as un­defined. This co uld cause prob lems whentryin g to communicate between the se twomod ems over a mixed-modem link . WhileAMTOR is not an app ropriate protocol foruse in transferring binary da ta or heavil yformatted text , the incorrect tran sfer of cer­tain punct uation characters co uld cor rupt themea ning of a text message.

Obviously, mo re extensive testing mus t bedone be fore a single protocol can be recom­mended for use by FEMA. Before that is done ,the overall requirements, includi ng the rangeof channel co nditions and the characteristicsof the traffic (content, size and freq uenc y oftran smission), mus t be better defined .

range of 0 to +40 dB , at the insistence of themanu facturer. A ll of the oth er protocol s weretested over a ra nge of 0 to +20 dB .

Similar results appear for the PACTOR IIprotocol , which also uses phase modulation ,alth ou gh the se lection of pha se leve l(DB PSK, DQPSK, 8-DPSK and 16-DPSK) isdone manu all y, and not automatically as wi ththe CLOV ER prot ocol. For testing, the high­est tran sfer rate (co rresponding to the high estphase level modu lation [16-DPSK]) was cho­sen. In both CLOVER and PACTOR II, thehighest th rou ghput measured und er CCIRGood conditions is nearl y half of the cle ar­chann el throughput, and roll s off qu icker fordecreasin g SINs than doe s the throu ghput ofprotocols that do not use phase modulat ionschemes . However , the throughput of the setwo pro tocols is an order of magnitude grea terthan the "bas ic" AMTOR protocol.

An attempt was made to de termine if thethroughput of a parti cular protoc ol was spe ­cific to the manu facturer ' s imp leme ntation,or whether it was represe ntative of all mo­dems usin g that protocol. Of the fiv e proto­cols tested, PACTOR and AMTOR are themost popul ar amo ng amate ur HF di gital op­era tors . CL OVER and G-TOR are propri ­eta ry protoc ols impl em ented by their re­spect ive deve lopers only . PACT OR II isbeginn ing to appear in modem s produced bya couple of ma nufacturers. But at the tim ethese tes ts were per formed, only one manu-

90

80

70 CCIA Goodu CCIR Poorcog 60

~~ 50~~

40.!<.

} 30

~I- 20

10

00 10 12 14 16 18 20

ConclusionsIn the case of CLOVER, the resul ts show

the obvious effect s of pha se modulat ion,automatic modulation selec tion, and the useof a sophisticated compressio n sche me.Throughput is notabl y depend ent on file sizeand , although not tested, content. Co mparedto its clea r-c ha nnel perfor mance, the use ofphase modulatio n demonstrates a suscepti­bility to multi path distort ion . As the through­put var ies, CLOVER change s phase modula­tion sche mes (B PSM at low rates, thro ughQP SM, 8PSM , 8P2A, 16PSM, to l 6P4A athigh ra tes ) wh ic h sho ws up as points of dis­co ntinuity in the plo tted dat a, notabl y at the12, 23 and 35-c har acterlseco nd points.

CLOVER al so uses a Reed-Solom on er­ror-correc t io n a lgor ithm, the effic iency ofwhich ca n be se t to thr ee level s (Rob ust,where 60% of the tra nsmi tte d block con ta insdata ; Normal, 75 %; and Fas t, 90%). The Fas tbias se tt ing , which produced the high estthro ughput, was used during testi ng. Basedon the test dat a, CLOVER see ms to be bestsuited for tr ansmitting lar ge file s. No te thatthe CLOVER prot ocol was tested over a SIN

operation s. Ho wever , for providin g a usefulperformance base line of protocol per for­ma nce , the laboratory tests ar e preferred .They can be rep eated at any tim e of day oryea r, en suring that the condi tio ns will be thesa me for all test s.

Table 5FEMA Over-the-air Performance Test ResultsProtocol/mode Compressed File size Channel Over-the-air ITS Lab

Data (bytes) Quality Throughput Throughput(char/sec) (char/sec)

SITOR ARQ No 650 Good 4.42 3.9-5.9Poor 4 .27

PACTOR ARQ No 650 Good 9.70 Not measuredPo or 5 .70

PACTO R ARQ Yes 650 Goo d 8.80 2.5-25 .0Poor Not measured

CLOVER ARQ Yes 3238 Goo d 28.03 7.0-40 .0Poor 16.83

Page 89: Packet- Speed- More Speed

Measuring 9600-BaudRadio BER Performance

DSP techniques make testing aG3RUH-compatible radio easy.

By Jon Bloom, KE3Z

One ofthejobs oftheARRL Labis to test the performance ofequipment sold to amateurs.

With the new crop of9600-baud radioscoming out, we had to develop a tech­nique for testing their performance.The best way to test the performanceof a radio used for digital communica­tion is, by far, to test the bit-error rate(BER) that the radio provides undervarious conditions. BER is a concep­tually simple metric that answers thequestion: How many of the bits getthrough correctly when a data streamis passed through the system?

The system we developed uses aTexas Instruments DSP Starter Kit(DSK) board that includes a

225 Main StreetNew ington, CT 06111emai l: jb [email protected] Chapter 2

TMS320C26 processor. Figs 1 and 2show the circuitry of the BER test boxbuilt in the ARRL Lab. The DSP gen­erates a test signal that is passedthrough the system under test . Forreceiver testing, the signal modulatesa low-noise FM signal generator thatfeeds the radio being tested. Thedemodulated output of the radio goesto the DSP input so that the demodu­lated signal can be compared to thetransmitted signal. To test transmit­ters, the DSP output test signal isapplied to the modulation input ofthetransmitter under test. The transmit­ter RF output is attenuated to a lowlevel, then applied to the Lab-builttest box where it is mixed with anunmodulated signal from a signalgenerator. The resulting IF signal isdemodulated by a low-distortion de­modulator, and the demodulated sig­nal is routed to the DSP input for com -

parison with the generated signal.

G3RUH Signals

The 9600-baud system used foramateur packet radio, both terrestri­ally and via the UoSat packet satel­lites, uses signals handled by theG3RUH modem design. In this system,the binary data stream coming out of aTNC is first "scrambled" to remove anyde component of the signal. Scram­bling, which has nothing to do withencryption or data hiding, is simplyencoding that ensures that, on aver­age, there are as many J-bits in thedata stream as there are O-bits. Theaverage voltage of the data signal isthus constant. The scrambled datasignal is then used to generate shapedpulses. The shaped-pulse signal iswhat is applied to the modulationinput of the FM transmitter.

On reception, the shaped-pulse

Page 90: Packet- Speed- More Speed

Fig 1-Diagram of the SER test connections to the TI DSK.

signal is filte r ed a n d limit ed, to re ­cover t he scramble d data stream. Aclock-recovery circuit genera tes aclock signal that is sy nc hronou s withthe incoming data. Using this clock,the modem descramb les the data

s tream, with t he end result being abinary dat a signal id en t ica l to thetransmitted da t a signal.

Since our test system h as to gener­ate and receive signals like those of theG3RUH modem, some discussion of

t he pulse shaping u sed is necessary.The general problem to be solved whense n ding digit al data using F SK is t olimi t t he bandwidth of the baseband(dat a ) signal before applying it to themodulator. But you have to be carefulh ow you do that.

When we limit the bandwidth of apulsed signal, we necessarily stretchthe pulses in time . Tha t is , a signalthat is bandlimited cannot also be timelimi t ed . Th at means t hat in limit ingthe ba ndwidth , we cause each pulse tooverla p adjacent pulses. In t heor y,each pu lse overlaps all of the otherpulses, but as you go further away intime fr om a par ticula r pulse, itsamplitude gets smalle r an d smaller.

So, we need to find some way ofallowi ng the pul ses to ove rl ap theirn eighbor s wit ho ut interference. Oneapproach to doing so is t o shape the

OSPAud ioOutput

OSK AudioInput (Fig 2)

AudioInput

AudiaOutput

TI OSK(TMS320 C26)

SyncOutput

TR

SerialI/O

JP4-22

10 k

HostComputer

47

U2NE5534

Buff erAmplifier

0.47

I;AudioOutp ut

OSK Audio Input(Fig 1)

r+;! I-+:""-....-W\r--I+ 5 V

lO I' F16 V

5OUT

,...--.---......---/ +5 V

O.O~-5V

3 4

0.4~

0.1

AGNO 6

L.Sh V-

U6MF4 7

;J:.,0.1

GAIN

> :....-_8-1IN V+

2 U54049

Audio t"'6=-------_-'VI.I\r-"-lOutput

4700

O.~

10k

Lim it er { 10Decoupiing 11

-5V

IFDecou pling

,.--'---,

UlNE604

GND

1N914

1N914

10 k

0.1

,...----'-'1 Limi t er

;h0"

L---.._.jf....---.:.:'2, LimiterInput

0.47

Cloc kInput

(500 kHz)

~. 1

133St.l

L3

,...-__.-__--..::8"'1 QuadratureInput

InputAudio Selec t

L2

>-.........JV'V\,----, Oemod

680-kHzLow- Pass Fil t er

20 k

2200

U3SBL-l

5 k

LOInput

(+7dBm)

Fig 2-Schematic diagram of the SER test box circu it.L1-1.2-mH To ko 10R S f ix ed inductor. L2 , L3- 41 turns #28 enam w ire on a

(Digi-Key part TK 4401-ND. ) T-50-1 toroid core.

Th eory/Design 2-13

Page 91: Packet- Speed- More Speed

at one-half the baud rate.! It contin­ues rolling off with this cosine curveuntil it reaches zero. In the G3RUHsystem, the spectrum begins rolling offa t 2400 Hz, reaches t he -6-dB point a t4800 Hz and reaches zero at 7200 Hz .

In the time domain, the pulse shapet hat results from such a spectrum hasa maximum at the center of the bitperiod, then decreases in a mpli t ude aswe move away from the center of thebit. The pulse signal goes negative,passing through zero at the center ofthe preceding and following bits . As wego farther away in time from the bitcenter, the signal alternates betweenpositive and negative values , al wayspassing through zero at the center ofeach bit. There are other spectra thathave pulse shapes that reach zero atthe cen t er of all the other pulses, bu tthe benefit of the raised-cosine spec­trum is that the pulse amplitude fallsoff rapidly with time . This is impor­tant because any amplitude or phasedistortion present in the system islikely to cause the zero-crossing pointsof the pulse to shift in t ime , causinglSI. Since the amplitude of the pulse issmall nea r the center of other pulses,the potent ial for h armful lSI is alsosmall .

Since each bit of the shaped-pulsesignal now extends across multiple bitperiods-both preceding and follow ingbits periods-we must take this in toaccount in generating our signal. An d,since this is DSP, what we are gener­ating is a sa mp led version of the sig­n al. What we m ust end up with at anysample is a signal that comprises acomponent from the cu r rent bit, pre­ceding bi ts and following bits. Theo­r et icall y , we need com pone nts from allof t he bits in the data stream, but theamplitude of each pulse falls off sorapidly that only a few successive bitsneed be u sed to generate any givensample . In this system, we ch ose toinclude components of the current bitand the four preceding and four follow ­ing bits.

Note that only I -bits contribute tothe signal; O-bits generate no pulse. Ifwe were to send a continuous streamofO -bits, we 'd get no pulses at all . (Ofcourse, the scrambler circuit ensuresthat will never occur in a real trans­mitted data str eam.) So, for eachsample we ne ed to add up the contri­bution of the curren t bi t an d the con­tributions of ei gh t other bits, some ofwhich may be O-bits that make no con­tribution. The con t r ibu t ion of a par­ticular preceding or follo wing I-bit is

S.J

- 0.001

. 6 .1271. 10. 4

7.4531·1 0'4

0 .00 19

0 .0016

2.9393.10' 4

3.8545 '10' 4

0.0036

0 .0075

0.0054

- 0 .0086

- 0 .0311

0.0452

- 0.0269

0 .0393

0 .1453

0 .2571

0.329

# bits: b c 9

(' n - 1 \

= 0 ,1 .. floor ~_. )

\ 2

n =36

Time values

Impulse response function

other pulses contributes any ampli­tude to the signal at that time.

The spectrum of one pulse that hasthese characteristics is straightfor­ward. It is flat from 0 Hz out to somechosen frequency, then rolls off with acosine-shaped curve, r each ing -6 dB

Baud rate: R = 9600

f sn = ·_··b

R

scale = 1.8110 1

Normalization of filtercoefficients

= S.1

Impulse response of filte r

Im pulse Res ponse of Filter

2

2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 3436

( max( h ) - mint h )

0.9

0.8

0.7

0.6

h . 0 .51

0.4

0.3

0.2

0.1

0

- 0.1

- 0 .20

he(t) = (~n( 2 ' 7[ ' f o · t)\ .1 COS (2 ' 7['!~~.\ 2 ·7[·f o·t I l 1-(H tl·t )2

1t d = ..- Sampling period

f s

= 0 ,1 .. n -1

scale

WRITEPR N( IMPULSE)

User-Defined Parameters:Sampling rate: f s = 38400

System bandwidth: B = 7200

Calculated parameters:

fo =~ f 1 =2·f o -B

f tl = B - f o

Fig 3-This Mathcad worksheet calculates the impulse response of the FIR filterused to generate the shaped-pu lse signal.

pulses so that, while each pulse doe soverlap its neighbors, the amplitude ofthat pulse is zero at the cent er of eachof the ot her pulses . Th a t way, we cansample the signal at the center of eachpulse period and see only the signalfrom the current pulse; none of the

2-14 Chapter 2

Page 92: Packet- Speed- More Speed

the amplitude of its pulse at thepresent time. To keep things simple,we use a sampling rate that is a mul­tiple of the baud rate-and thus amultiple of the pulse rate.

The method used to implement thispulse forming is an FIR filter. Theimpulse response of the filter issimply the samples that comprise asingle shaped pulse, extending overnine bit periods. Our sampling rate isfour times the bit rate , so the impulseresponse is 4x9 =36 samples long. AMathcad 5.0 worksheet, shown in Fig3, calculates the impulse response.

If we feed a single sample of ampli­tude 1 into the filter, preceded andfollowed by O-amplitude samples, theresulting output will be a single copyof our shaped pulse, as shown in theMathcad graph in Fig 3. To generateour data stream, we input the presentdata value (1 or 0) at one sample, fol­low it with three samples of 0, theninput the next data value. At any time,the FIR filter contains zeroes in allexcept (possibly) 9 locations, repre­senting the 9 successive data bits. Theoutput ofthe filter at any sample timecomprises components of 9 datapulses, which is what we want for ourshaped-pulse signal.

Now that we know how to generatethe shaped-pulse signal from a datastream, we need to think about how togenerate the data stream itself. Wewant to generate a data stream thatmimics the scrambled signal of the

G3RUH modem. The scrambler in themodem is a tapped shift register withfeedback. The logic equation for thiscircuit is:

y = Xo Etl x12 Etl x17

wherey is the output ofthe scrambler,xQ is the current input bit, x12 is the12th previous input bit and x17 is the17th previous input bit. Generatingthe test data stream is done by mak­ing Xo always a 1, implementing a shiftregister in software, and calculating yfor each new data bit-once every foursamples. Note that this is essentiallythe same signal generated by theG3RUH modem in its BER test mode.

Counting Bit Errors

Having generated a test signal, wenow need to consider how to comparethe signal coming back from the sys­tem under test to the signal we trans­mitted. In passing through the testedsystem, the signal will be delayed bysome amount, and the amount of delaywill vary from one system to another.What we need to do is determine whatthe system delay is, then rememberwhat our transmitted signal was thatfar back in the past in order to com­pare it to the samples of the receivedsignal. But we are only interested inthe value of the received signal at onetime during each bit period: the centerof the bit.

The DSP software handles this needusing a two-step delay process. First,the operator will tell the DSP system

how many samples of delay there arebetween the transmitted and receivedsignals. Of course, the system delay isnot likely to be kind enough to let thecenter of the received bits fall rightonto one of our samples-there areonly four samples per bit, after all. So,the second step is to adjust the phaseof the DSP's transmit and receivesample clocks. The combination ofthese two techniques lets us adjust theDSP's delay with fine resolution.

This calibration procedure is per­formed using a dual-channel oscillo­scope. One channel is connected to thesync signal from the DSK, the otherchannel is used to display the demodu­lated signal that is being fed intothe DSK input. The operator firstcommands the DSK to generate a cali­bration signal. The DSK does so bypassing a single I -bit through the FIRfilter described previously, followed byO-bits. The I -bit is repeated every 72samples, resulting in a single shapedpulse every 2 ms, The oscilloscope istriggered on the sync signal and set for0 .2 us/div. The operator then sendscommands to the DSK to alter thenumber of samples of delay betweenthe transmitted and received signals.The sync signal, a short pulse, is out­put during the sample the DSK be­lieves to be the center of the receivedbit. Thus the operator simply adjuststhe delay value, which causes thereceived signal trace on the oscillo­scope to move relative to the sync

Fig 4-The first-phase ca libration signal. When properlycalibrated, the calibration s ignal (lower trace) is aligned withthe sync pulse (upper trace) as shown . The oscilloscope isset for 0.2 us/div, The sync pulse is very narrow and hard tosee in this photograph but shows up well on the oscilloscopescreen.

Fig 5-Second- phase calibration uses the SER test signal andsync pulse to produce an eye pattern. The oscilloscope is setfor 10 f,[s/div. The center of the eye is adjusted to lead thesync pulse by 14 us as shown.

Theory/Design 2-15

Page 93: Packet- Speed- More Speed

Table 1-DBERT Commands

pulse. When the sync pulse is alignedwith the shaped pulse of the receivedsignal (Fig 4), the DSK is at the propersample-delay value.

To achieve final calibration, theDSK clock phases must be adjusted totake into account the part of the sys­tem delay that is less than one sampleperiod long. This is done by command­ing the DSK to output its test signaland switching the oscilloscope to10 us/div. Now the oscilloscope dis­plays a sync pulse at the right side ofthe screen, along with the eye patternof the received data (Fig 5). The propersampling point of the bits is the pointwhere the eye is most "open." How­ever, there is a 14-/ls delay in the DSKsystem between the actual samplingpoint and the sync pulse . Thus, theoperator commands the DSK to stepits clock phase until the center of theeye leads the sync pulse by 14 us. Thisis not a hugely critical setting-a few­microsecond error isn't normally de ­tectable in the measured BER. Thedelay does have to be in the ballpark,however.

There is one more detail to considerduring calibration. Depending on thesystem being tested, the received sig­nal may be inverted from the transmit­ted signal; the positive-going pulseswe sent may now be negative-going. Ifthis is the case, the DSK will count allthe correctly received bits as bad andall the bad bits as correct. So, the DSKsupports a user command to invert thesense of the received data. The polar­ity ofthe received signal is most easilyseen during the first calibration phase(Fig 4), and if the pulse is seen to beinver t ed then, the DSK's "invert" com -

Comma nd

IC45+

Ao1NQSVT

Retu rn value

NoneNoneError count (2 words)Error count (2 words)NoneNoneNoneNoneNoneNoneNoneSINAD value (4 words)Input voltage (1 word)None

mand should be issued. That willcause no difference in the displayedsignal, but the DSK will know "whichway is up."

DSP Program OperationThe DSP program used to perform

BER measurements is called DBERT.The object file, DBERT.DSK, is down­loaded into the DSK from the host com­puter. Once DBERT is running, it com­municates with the host computer viathe DSK's serial port. The serial I/Ocommunication mechanism was de­scribed in a previous article .s Whenserial I/O is occurring, the DSP inter­rupts must be disabled. This keeps theDSP from executing its signal-process­ing operations during serial I/O . Forthat reason, DSK operation is con ­trolled by the host computer using acommand-response sequence: Thehost sends a command to the DSK, theDSK executes that command-includ­ing any needed signal processing­and reports completion of the com­mand back to the host, if needed. Eachcommand to the DSK from the host isa single ASCII character. Some com­mands result in the DSK needing tosend data back to the host . In thatcase, once the command is completedby the DSK, it sends the data back tothe host via the serial interface. Table1 lists the commands supported by t heDBERT program.

Because the DSK has to stop its sig­nal I/O while communicating with thehost, there is a potential problem intaking a measurement. When the DSKbegins generating the test signal, it ispossible that the system being testedwill exhibit a response to the start-up

Description

Generates test BER signalGenerates ca libration signalPerforms 1O,OOO-bit BER testPerforms 1OO,OOO-bit BER testIncrement sample delayDecrement sample delayAdvance clo ck phaseNo rmal data po larityInverted data polarityGenerate 1-kHz sine waveQuiet (DSK output = 0 V)Take SINAD measurementReport input voltage valueGenerate test signal (4800-Hz sine wave)

transient from the DSK that invali­dates the result. So, the BER test sig­nal is started and the DSK waits20,000 samples (about a half-second)before starting to sample the receivedsignal.

We wanted to be able to test at con­si stent signal-to-noise ratios in orderto establish reference levels for com­paring different radios to one another.Since each unit will exhibit a uniquesensitivity, we needed some way ofadjusting the input power level to geta fixed output signal-to-noise ratio .The solution was to include a SINADmeasurement function in the DBERTprogram. When the DSK receives theSINAD command from the host, it gen­erates a I-kHz sine wave . About a half­second after it begins generating thesine wave, the SINAD measurementsoftware begins sampling the inputsignal. It then takes 8192 samples formeasurement. Each measured sampleis squared and added to a running sum(the signal -plus -noise value) . Thesamples are also run through a narrowI-kHz IIR notch filter that removesthe I-kHz test signal, leaving only thenoise. The output samples from thisfilter are also squared and added to a(sepa ra t e) running sum (th e noisevalue) . After 8192 samples have beenprocessed, the two 32-bit sum valuesare returned to the host computer. Thehost can then calculate the SINAD byexpressing the ratio ofthe signal-plus­noise value to the noise value in dB . Amore detailed description of this tech­nique was published in an earlier QEXarticle .s

The analog I/O of the DSK is per­formed by a TLC32040 integrated ana­log subsystem. This chip includes al4-bit D/A, 14 -bit AID, sample-clockgenerator and input and output pro­grammable switched-capacitor filters(SCF). The TLC32040 is driven by a10-MHz clock produced by the proces­sor chip. The frequency of this clock,in combination with the program­mable dividers in the analog chip, de­termines the available sampling rates.The design of the BER test softwarecalls for a 38,400 sample-per-second(sps ) rate . Unfortunately, this exactrate isn't possible with the 10-MHzclock. The nearest available rate isapproximately 38,461.5 sps . Thistranslates to a 0.16% error in thespeed of the test signals, which isnegligible in the context of BER mea­surements. However, this error is suf­ficient to make the test signal unread­able by a G3RUH modem since the

2-16 Chapter 2

Page 94: Packet- Speed- More Speed

clock-recovery loop in that de sign hasa very narrow lock range. I discoveredthis ea rly on in the development ofthissystem . To check that the problem wasin fact the speed difference a n d notsome probl em wit h the signal I wasgenerating, I temporarily remo ved the40 -MHz master oscillator on the DSKboard- from which the 10-MHz clockis derived-and connected a si gnalgenera tor to the clock input through aSchmitt trigger inverter. By settingthe signal generator t o generate ex­actly the clock fr equency needed to geta 9600-baud sign al, I foun d t hat theG3RUH modem was quite happy t oconsider my BER test signal a properone.

Another problem with theTLC32040 is that the input SCFis aband-pass filter that normally cuts offat about 300 Hz. While the filter cut­offfrequency is programmable , it can'tbe set anywhere near the very low fre­quency needed for this application. So,the DBERT program configures theTLC32040 to operate without its input

filter . This requires that a n externalantialiasing filter be added. As shownin Fig 2, and described in more detaillater, an SCF that cuts off at 10 kHzwas added to the BER test box t o fillt h is need .

The final char ac teristic of theTLC 32040 to note is that it includes noin t er n al sinx/x correction-not sur­prising in a chip with a programmablesample rate. Fortunately , with a sam­pling r at e of 38400 Hz and sign als ofonly up to 7200 Hz, the sinx/x roll-offat the upper end of the signal spectrumis only about Y2 dB . Still , this seem sworth correcting. The TLC32040 datashee t describes a first-order IIR filterthat can be used to perform this cor­rection. The Mathcad worksheet of Fig6 shows the calculation of the coeffi­cients of this sinx/x correction filter,which is implemented in the DBERTprogram. Fig 6 also shows the pr e­dicted out put response both before andaft er the correction, as well as theestim ated group delay of the correc­t ion filter .

As noted, the calibration procedurerequires a sync pulse. The DSK has nouncommitted out put signa l lines us­abl e for outputting a pulse, so DBERTou tputs the sync pulse on the ser ialdat a output line. Since t he sync pulseis less than a microsecond long, thiswon't usually bother the ser ial I/O chipofthe host computer. (We did fin d onecomputer that was occasionally con­fused by the pres ence of the syncpulse. ) The RS-232 signal is availableon one of the DSK's expans ion connec­tors. A 10-kQ resistor connected to thispoint brings the sync signal out to thefr ont panel of the BER test box. Thepresence of the resistor protects theRS-232 line from accidental shortingor application of another signal.

Host Application Software

We used two different programs.One, which we won't cover in detailhere, manages the Lab's computer­controlled signal generator forstepped BER measurements at vari­ous signal level s an d frequencies . The

Computation of Corrected System Response

2.51=-'__

--- ---------------_.--............,."

Frequency Response

Group Delay

0.5 1 1.5 2 2.5 3 3.5 ~ 4.5 5 5.5 6 6.5 7

f (kHz)

1.25

1.75

2.25

2.75

-0. 1

dB -0.15

- 0.2

0.25

0.3

' 0.35

0.4

0.45

-0.50

0.1 ,-.--.--r--r-,--.--.----.-.-,,--.-.----.--.0.05

-6.939. 10- 17 ,--~-~-,-,- - - - - - - - - - - - - - -

-0.05

Upper frequency of passband

Amplitude correction at fl

Nonnalized lower passband frequency

Amplitude correction at fn

Sampling frequency f ' 4800

f n ; 0.125 Nonnalized upper frequency

Resulting first-order IIR filter: y[n];v,(l -v,)x[n)+v,y[n·l)

Given

A' ( l - 2,pl'cos (2 ' '''f n ) . P12

) - P22' ( 1 _ P 1)2. 0

A L·( l - 2·P 1·cos (2 '''' f L) . P12

) - P22' ( 1 - P 1)2 . 0

IV1\ , Find ( P 1 ,P 2)

Iv2 /

v I ; ' 0.10496 16 v 2 = 1

(

f 's in 1t' - - \

str : , .~ M(nf".-f s

A _ ( rt-f n )2 A ; 1.053029 3sin (no f~)

Compute need ed first-order IIR coefficients

P 1 dP2 ,2

f s ' 384 00

f , ~n f s

P t f') - ar g ( M (f)o 0.5 1 1.5 2 2.5 3 3 5 4 4.5 5 5.5 6 6.5 7

f (kHz)

f x ' 1 ,70 .. 7000 '"x 2' '' ,70 ·2 '" .. 7000 ·2 ·"

Fig 6-Correction of sinxlx roll-off is performed with a first-order IIR filter having the coefficients calculated by this Mathcadworksheet.

Theory/Design 2-17

Page 95: Packet- Speed- More Speed

other program, BERT.EXE, is a C pro­gram that communicates with theDSK under operator control. Commu­nication with the DSK is normallyperformed at 19,200 baud, althoughslower speeds can be used. To ensurethat no serial overruns occur, BERTuses interrupt-driven serial I/O .

The BERT program is quitestraightforward. It accepts keyboardcommands from the operator andsends them to the DSK. The commandset is the same as that of DBERT­BERT just passes these commandsthrough to the DSK-with a couple ofadditions. Also, BERT "knows" aboutthose DSK commands that generate aresponse from DBERT. When such acommand is given, BERT waits for theresponse from the DSK, then convertsthe incoming serial bytes to binaryvalues. In the case ofBER test values(commands 4 and 5), it prints out thereported number of errors and theBER, calculating the latter based onthe number of bits the DSK wascommanded to use in its test. For theSINAD command (8) , BERT computesand prints the SINAD and the distor­tion percentage using the values re­turned by DBERT.

The DBERT program can performBER tests using 10,000 samples or100,000 samples. The BERT programalso provides a 1-million sample BERtest (the 6 command), which it per­forms by commanding the DSK toperform a 100,000-sample BER test 10times . BERT sums the values fromeach of these tests to get the finalresult. As each BER test result isreturned from the DSK, BERT printsthe running total of samples, errorsand BER. The L command performsthe same function, except that it stopsif 100 or more total errors have beenreported.

If the operator selects the calibrate(C), idle (1), sine-wave (N) or quiet (Q)commands, BERT remembers the se­lected command. Whenever a BER orSINAD test command is performed,BERT waits for that command tocomplete, then sends the appropriatecommand to place DBERT in the mostrecently selected mode from amongthose listed.

BER Test Box Hardware

The BER test box contains severalop-amp amplifiers used to keep theinput signals wit h in the range of theTLC32040 analog input. The box alsocontains a mixer and a demodulator,shown in Fig 2, designed by ARRL Lab

2-18 Chapter 2

Engineer Zack Lau, KH6CP. To testtransmitters, the transmitter outputis attenuated with a high-power at­tenuator down to a level that theSBL-1 mixer, U3, can handle. The LOinput of the SBL-1 is driven by a +7­dBm signal from a signal generator setto a frequency 373-kHz below or abovethe transmitter frequency. The outputfrom the SBL-1 passes through a low­pass filter that removes the sum fre­quency, leaving only the 373-kHzdifference frequency. This signal isapplied to an NE604 FM IF chip thatcontains a limiter, IF amplifier andquadrature FM demodulator. Thefrequency at which the demodulatoroperates is set by L1 and its associatedsilver-mica capacitor. When Zack builtthis circuit, the components he usedjust happened to fall at 373 kHz. Ifyoureproduce the circuit, it's likely thatyour copy will work at a slightly differ­ent frequency. This shouldn't presenta problem; the exact operating fre­quency isn't critical.

Since the NE604 operates from asingle 5-V supply, its output is a posi­tive voltage . This signal is fed intoU4A, which amplifies the signal andalso removes the positive offset, sothat the result is zero volts when theinput signal is at 373 kHz. U4B ampli­fies this demodulated signal or an ex­ternal input signal (usually the outputof a receiver being tested) with adjust­able gain. The signal is finally filteredin an MF4 4th-order Butterworth low­pass SCF, U6, that acts as theantialiasing filter for the DSP input.The cut-off frequency of U6 is deter­mined by its input clock, with the clockbeing 50 times the cutofffrequency . Inour test set, this clock is supplied byan external function generator, but itcould as easily be provided by a crystaloscillator and divider chain.

Interpreting BER Measurements

Bit errors are (or should be) duemainly to corruption of the signal bynoise . Thus, they should be random.That being the case, if you make thesame measurement several times, youare likely to get different results eachtime . But the more bits you sendthrough the system, the more the ef­fect of noise is averaged and the moreconsistent the measurements will be .That raises the question: How manybits do you need to use to get a validresult? The answer to that questiondepends on what you mean by "a validresult ." The more bits you use, thecloser your measurement will be to the

"true" BER you would get if you sentan infinite number of bits through thesystem. Since sending an infinitenumber of bits through the system is alittle, well, impractical, we need tocome up with some guidelines forselecting a finite number of bits to use.

Unfortunately, in order to do that wehave to make some assumptions aboutthe character of the noise in thesystem. Fortunately, the assumptionswe make hold up pretty well for realsystems . (There's nothing new aboutthis; we make assumptions about thecharacter of system noise all the time.For example, when we relate noise tosystem bandwidth we often assumethat the noise is uniformly distributedacross the spectrum of interest.)

What we find is that with a givennumber of bits in the sample set, anda given number of errors within thosebits, we can establish a confidenceinterval. The confidence interval tellsus how likely it is that our measure­ment is within some specified factor ofthe "true" BER. For example, we mightfind that our measurement gives us an80% confidence that the value iswithin a factor of 3 of the true BER.The end result is that we can get asgood a measurement as we want if weare willing to wait for enough bits togo through the system. Of course, at alow BER we don't get many errors, sowe need to send a lot of bits!

What's interesting about the confi­dence interval is that it depends on thenumber of errors detected. Supposeyou made two BER measurements. Iffor the second measurement youdouble the number of bits sent and getdouble the number of errors, you endup with the same BER but a higherconfidence. On the other hand, if youdouble the number of bits but measurethe same number of errors (you'remeasuring at a lower noise level, forinstance), you get a lower BER but thesame confidence interval. What thatmeans is that all we need to do is toensure that we have at least the num­ber of errors needed to establish thedesired confidence interval.

There are two useful sets of numberswe 've used here in the ARRL Lab forour BER testing. Ifyou measure 10 biterrors, you are 95% sure that you arewithin a factor of about 2 of the trueBER. And ifyou get 100 bit errors , youare 99 % sure that you are within afactor of about 1.3 of the true BER.4 Ifyou look at a curve ofBER versus sig­nal-to-noise ratio, you'll find that afactor -of-2 difference in BER occurs

Page 96: Packet- Speed- More Speed

with a fr action of a dB change in sig­nal-to -noise ratio . That suggests thata measurement that results in 10 biter r ors is a pretty good one , an d ameasurement t h at results in 100 biter ro r s is a very good me asurement .That's why t he L command exis ts inthe BERT program. It pops 100 ,000bits a t a time through t he system ,stopping wh en 100 or more err ors havebeen reported-because continuing tom easure more bit errors is a waste oftime- or when 1 million bits have beense nt . Usi ng a million bits means thata BER of 1x10-4 (100 bit er rors) orworse can be me asured wit h great ac­curacy, an d a BER of 1x10 -5 (10 bit er­ro r s) can be me asured with decenta ccur acy. Of course, if you have timeon your hands you ca n mea sure a BERof 1x10- 6 by sending 10 million bits

through the system . (That takes over17 minutes at 9600 baud!)

ConclusionThe system described here has been

used to measure a number of 9600­baud radios. Some of the test resultswill be presented in an upcoming QS Tarticle, and future QST r eviews of9600-baud radios will include mea­surements made using this sys tem.The system has proven to be effectiveand easy to use. I should add that wespot-checked the results obtained withthis system by measuring BER using aG3RUH modem. It gave BER resultsthat were consistently slightly betterbecause its input filter cuts off at alower frequency than the SCF in theBER test box , improving the signal-to ­noise ratio .

The software for this system, includ­ing the sour ce code, is availa ble fordownloading from the ARRL BBS(203-666-0578 ) and via the Internetusing anonymous FTP fromftp .cs.buffalo.edu. The fil e name isQEXBERT.ZIP.

Notes"Couch, L. W., Digital and Ana log Comm u­

nication Systems (New York : Macmi llan,1993), p 179 .

2Bloom , J., KE3Z , "Measuring SystemRespo nse with DSP," QEX, February1995, pp 11-23.

3Bloom, J., KE3Z, "Measuring SINAD UsingDSP", QEX, June 1993, pp 9-13 .

4Jeruchim, M. C., Balaban, P. andShanmugan, K. S., Simulat ion of Commu­nication Systems (New York: PlenumPress , 1992), pp 492-501.

TheorylDesign 2-19

Page 97: Packet- Speed- More Speed

The WA4DSY 56 KILOBAUD RF MODEMA Major Redesign

By Dale A. Heatherington, WA4DSY

Abstract

In /987/ designed a 56 kilobaud RF modem which was sold in kitform by GRAPES, the Georgia Radio AmateurPacket Enthusiast Society. This paper describes how the WA4DSY 56 kilobaud RF modem was radically redesigned tolower cost, reduce size, and improve reliability, manufacturability and useability. The reader is refered to the ARRLpublication Proceedings ofthe 6th Computer Networking Conference, page 68 for details on the original design.

Overview

The original modem was implemented on 3 PCboards. It required both plus and minus 5 volts for themodem and 12 volts for the external transverter. Thepurchaser of the kit had to fabricate his own enclosureand obtain a suitable power supply. There were no sta­tus indicators. Only those hams with above averagehome brewing skills would attempt to build the unit.However, once built, the modem performance and reli­ability were quite good. Several high speed networkshave been successfully built using these modems.

Unfortunately the large amount of skilled laborrequired to build a modem kit and to some extent, thecost have limited the wide spread adoption of thesemodems for high speed networking. I have redesignedthe modem to address these issues.

The new design implements the modem on a single4 layer printed circuit board powered from a 12 voltpower supply. The PC board measures about 7 incheson each side. Signals produced by the new design areidentical with the original and the new modems willinteroperate with the old ones.

Most of the modem functions are implementeddigitally in a Xilinx (tm) Field Programmable GateArray (FPGA). The bandwidth limited MSK signal isgenerated digitally at 448 kHz to eliminate the need foranalog double balanced modulators and a 90 degreephase shifter. This signal is up converted to the 10 meterband at I milliwatt. The converter is synthesized over a2 mhz' range (28-30 mhz). An external transverter con­verts the signal to the 222 or 430 mhz band. The delaybetween RTS and signal out is quite low, about 20microseconds or I baud interval.

Note that this is a true modem which converts datato RF, unlike the G3RUH and K9NG designs which arebaseband signal processors and don't do any modulatingor demodultating.

The receiver is implemented with a single chipdevice and is synthesizer tuned from 28 to 30 MHz. Aquadrature detector is used for FM demodulation. Thedemodulated signal is sliced using a circuit similar to the

2-20 Chapter 2

one in the original design which automatically adjuststhe slicing level. The signal is then fed into the FPGAwhere clock recovery and data carrier detection are donedigitally. The delay between receiving a signal andcarrier detect indication is about 3 milliseconds.

The user interface has been greatly improved. TenLEDs indicate received signal level. Other LEDs indi­cate Request To Send, Data Carrier Detect and Ready.The data interface is dual mode. A single switch selectsCMOS or RS422 modes. The signals are presented on aDB15 connector wired to mate with the Ottawa PI2Packet Interface Card. Other devices such as TNCs canbe connected by wiring an appropriate cable and con­nectors.

Unlike the original design which allowed the user toreconfigure the modem for different baud rates, the datarate of the new design is fixed. Major changes arerequired to both the RF and FPGA circuits for use at anyother baud rate. Part of the reason is because the first IFof 448 KHz must be 8 (or some power of 2) times thebaud rate. Also, the receiver chip is being operated closeto its maximum rate at 56KB.

Data Coding

All data coding is done in the FPGA chip. The chipis clocked at 14.31818 MHz. The transmit clock isobtained by dividing by 256. The exact baud rate isactually 55.9304 kilobaud, the same as the originaldesign. The transmit clock signal is sent to the user onthe RS422 interface. The user's transmit data source isexpected to use the rising edge to clock out each bit. Themodem samples data bits on the falling edge of theclock.

The transmit data stream is scrambled using thesame shift register configuration as the original modem,a 17 bit register with feedback taps at stages 5 and 17.This is not compatible with K9NG and G3RUH 9600bps modems. For more details on the scrambling sys­tem, see the section titled "Descrambler" below. Afterscrambling , the data enters the digital state machinewhere both NRZ to NRZI conversion and RF waveformtable lookup operations are performed.

Page 98: Packet- Speed- More Speed

Modulator

This modem has no physical modulator. AIl RFwaveforms are stored in EPROM . A digital statemachine fetches the appropriate waveform segment fromEPROM in response to the current data bit to be trans­mitted. 32 samples of the stored digitized waveformsegment are read from the EPROM and sent to a Digitalto Analog Converter (DAC) during each baud interval.The carrier frequency is exactly 8 times the baud rate topermit the splicing of different waveform segmentstogether without phase discontinuties . The result is a448 kHz bandwidth limited MSK signal. Unlike theprevious version of this modem, there are no modulatorrelated adjustment controls. The signal always has per­fect phase shift and deviation characteristics . The signalis identical to the one produced by the originalWA4DSY 56 KB modem.

Transmitted signal characteristics

Modulation is MSKBandwidth is 70 kHz at 26 DB down3.5 DB amplitude variation14 kHz FM Deviation90 degree phase shift per baud

Top trace: Raw TX signal from 8 bit DAC

Lower Trace : 56KB TX Clock

UpconverterandlF

The 448 kHz MSK signal shown in the photo aboveis first filtered with an 80 kHz wide 3 section LCbandpass filter to remove digital sampling noise. It's

then mixed with 10.245 MHz and converted to 10.693MHz. The 10.693 MHz signal is passed through two10.7 MHz (180 kHz BW) ceramic filters to remove thelocal osciIlator and unwanted lower sideband (10.245and 9.797 MHz). The undesired frequencies are reducedby at least 90 db.

The desired 10.693 MHz signal is then mixed with aYCO signal in the 39 MHz range. The lower sidebandof the mixer output (28-30 MHz) is selected with a twosection LC bandpass filter. Both conversions are donewith NE602 frequency converter chips.

The 29 MHz signal is amplified 30 db by an MMICchip and sent out to the users transverter on a BNC con­nector. The output level is adjustable from - I0 DBM to+5 DBM.

The local osciIlators are running at all times toassure instant response to the user's "Request to Send"control signal. The total delay is less than 20 microsec­onds from RTS to RF data signal out. Contrast this to theoriginal design which required up to 6 miIliseconds tostart the transmitter.

Receiver

The receiver uses a Motorola MCI3135 chip for allRF signal processing. The received signal in the 28-30MHz range from the user supplied transverter is filteredby a two section bandpass filter before entering thereceiver chip. The first local osciIlator is a YCO in the39-4 I MHz range controIled by a synthesizer. The sig­nal is mixed with the YCO to convert it down to 10.693

Receiver Eye pattern

MHz. The signal is bandpass filtered with a single 180kHz wide ceramic filter before being mixed with 10.245MHz and converted down to 448 kHz. A 60 kHz wideLC bandpass filter provides both selectivity and deem­phasis . Frequency modulation is recovered with aquadrature detector.

Frequency Synthesizer

I used a Motorola MCI45162 synthesizer chip forthis design. It is programmed seriaIly with a three wireinterface. It has completely separate reference countersand divide by N counters for transmit and receive . The

TheorylDesign 2-21

Page 99: Packet- Speed- More Speed

reference oscillator running at 10.245 MHz also drivesboth the receive down-converter and the transmit up­converter.

The YCOs are designed to cover 39 to 41 MHz,10.693 MHz above the 29 MHz IF frequencies. Thereceive YCO is included in the MC13135 receiver chip.The transmit YCO is a Colpits transistor oscillator withan emitter follower output buffer.

Since there is no microprocessor in this modem, Ineeded a way to generate the data to program the syn­thesizer. I had the good fortune of having extra space inthe EPROM; so I put the frequency data there. Up to 8different bit patterns for independent TX and RX fre­quencies can be stored in the EPROM. The first thingthe FPGA does after loading its configuration data isread one of the selected frequency bit strings into thesynthesizer chip. Three switches code for the 8 fre­quencies. Any 8 frequencies can be programmed intothe EPROM using a simple program written in C. Theprogram will be supplied with the modem. If the userwants a custom set of frequencies, he must have accessto an EPROM programmer or order a custom EPROMfrom the dealer . Still, this is superior to the originaldesign which used custom crystals often requiring a 6 to8 week wait and costing 10 to 15 dollars each.

Gated Tracking Data Slicer

Before the recovered signal can be used it must beprocessed to determine the state of the received bit, 1 orO. This is done with an analog comparator chip. Itsthreshold is set exactly halfway between the voltagelevel of a " 1" and a "0" . It outputs a "1'~ if the input ishigher than the threshold and a "0" if it's lower. Thereis a problem when the carrier frequency of the incomingsignal changes. The voltage levels of the ones and zeroschange so the threshold is no longer exactly half waybetween them. This causes an increase in errors. Onecommon solution, which doesn't work very well, is toAC couple the output of the demodulator to the detector.This is fine if the short and long term average of thenumber of ones and zeros is equal. This ideal conditioncannot be guaranteed even if a scrambler is used. Amuch better solution is to put some intelligence in thedetector so that it averages the voltage level of the onesseparately from the average of the zeros and then sub­tracts the two averages to obtain the ideal thresholdlevel. This circuit doesn't care about the ratio of ones tozeros as long as there is a reasonable number of each. Ascrambler is used to make sure there is a reasonablenumber of both ones and zeros. The circuit will com­pute the correct threshold if the input signal carrierfrequency is anywhere within the expected range of theones and zeros, in this case plus or minus 14 kHz. Thedata slicer used in this implementation is gated with therecovered clock. It only sees voltage levels near thecenter of a baud interval. A leaky sample and holdtechnique is used to grab the middle 1 microsecond ofeach bit. There is little variation in the peak levels frombit to bit thus reducing unwanted fluctuations in the slic­ing level.

2-22 Chapter 2

Clock Recovery

Clock recovery is done digitally in the FPGA.There are no adjustments such as YCO center frequencyas in the original design. The phase of a 3 moduluscounter is compared with the data zero crossing times.The counter is driven by a 3.579545 MHz clock. Thebaud rate for each modulus is listed below.

Modulus Baud Rate

63 56.818 (fast)64 55.930 (on time)65 55.069 (slow)

The counter can divide by 63, 64 or 65. The divideby 64 setting produces a 56 kHz clock. If the zerocrossing was late relative to the counters terminal countthen the counter is counting too fast. The countermodulus is set higher so it will be earlier next time. Ifthe zero crossing is early the modulus is set lower so itwill be later next time. If no zero crossing is detectedthe modulus is set to "on time" so the clock won't driftduring strings of ones or zeros. This scheme onlyintroduces about 0.5 microseconds of clock jitter (3%).

Data Carrier Detector

The data carrier detector is also implemented digi­tally in the FPGA. There are no adjustments. Two gatesare used to separate data zero crossings which fall withinplus or minus 12 1/2% (in sync) of the terminal count ofthe 3 modulus counter described above from the zerocrossings which fall outside this range (out of sync). Ifthe clock recovery circuit is phase locked, all zerocrossings should fall within the 25% "in sync" window.This is true even at low signal to noise ratios. The "insync" zero crossings cause a 5 bit counter to increment.The "out of sync" zero crossings cause the counter todecrement. The "carrier detected" flip flop is set whenthe counter reaches maximum count (31). The flip flopis reset when the counter reaches minimum count (0).The counter is designed not to overflow. It has "stops"at count 0 and 31. Carrier detect occurs when the clockrecovery circuit has acquired phase lock and 31 more "insync" zero crossings have occurred relative to "out ofsync" zero crossings. This takes about 3 milliseconds orabout half the time of the original 56KB modem withless falsing. Measurements show solid carrier detecteven when the bit error rate is as high as 6%. Randomnoise can't assert carrier detect because the zero cross­ings have random timings and will occur with equalprobably at any point in the baud interval. Since 75% ofthis interval is devoted to decrementing the 5 bit counter,it will quickly go to zero and reset the carrier detect flipflop. Periodic waveforms that are harmonically relatedto the 56 kHz clock frequency will trigger carrier detectif the clock recovery circuit phase locks to it.

NRZI to NRZ conversion

NRZ is a data signaling format in which zeros arerepresented by a certain voltage level and ones byanother. NRZI is a signaling format in which zeros arerepresented by a change in voltage level while ones are

Page 100: Packet- Speed- More Speed

indicated by no change . NRZI coded data is not affectedby inverting the data voltage levels or the mark/spacefrequencies in the case of FSK. This modem convertsthe incoming NRZI data to NRZ data with a simple cir­cuit consisting of a "D" Flip Flop and XOR gate. Thesecomponents are in the FPGA chip.

Descrambler

A self synchronizing data scrambler was used in thismodem for two reasons. First, it makes the data streamlook like a random stream of ones and zeros regardlessof the data being transmitted. This characteristic makesthe tracking data slicer and clock recovery circuits workbetter. Second, it makes the RF spectrum look andsound like band limited white noise. In other words, theRF energy is spread evenly over the modems bandwidthand shows no single frequency lines regardless of thedata being transmitted. Any potential interference toneighboring channels is limited to an increase in thenoise floor instead of squeaks, squawks, and otherobnoxious noises. This type of scrambling is alsocommonly used in high speed synchronous modems fortelephone use.

The hardware to implement the scrambler anddescrambler is very simple. It consists of a 17 bit shiftregister and two XOR gates, also known as a LinearFeedback Shift Register (LFSR). Each transmitted bit isthe result of the exclusive ORing of the current data bitwith the bits transmitted 5 and 17 bits times before. Todescramble the data, it is only necessary to exclusive ORthe current received bit with the previous 5th and 17thbits. If the data consist of all ones, the scrambler willproduce a pseudorandom sequence of bits that willrepeat after 131,071 clock pulses or every 2.34 secondsat 56 kilobaud.

This linear feedback shift register scramblingscheme does not violate the FCC prohibition againstcodes and ciphers because its purpose is to "facilitatecommunication" and the algorithm is publicly available.

8 Bit FIFO and Bit Repeater Mode

To allow this modem to be used as a full duplex bitrepeater I have included a first in-first out (FIFO) bufferand logic circuitry to route the received data bits back tothe transmitter. When repeater mode is enabled, the data .carrier detect signal will assert request to send. A 2second watch dog timer prevents transmitter lockup.The FIFO buffer is 8 bits long. To allow for both plusand minus "bit slip", the FIFO does not start sendingdata until it's half full. The transmitter then pulls bitsout of the FIFO at it's fixed clock rate while the receiverinserts bits into the FIFO at the receiver clock rate. TheFIFO is reset when data carrier detect (DCD) drops.With a data rate difference between incoming and out­going data of 0.01% a packet of 40,000 bits (5000 bytes)can be retransmitted before the FIFO buffer overflows.Since this is much larger than typical AX25 packets, Idon't think this restriction will be a problem. Keep inmind that DCD must drop between packets to reset theFIFO.

A single internal switch enables the bit repeatermode. The external data I/O connector remains active toallow communication with a local computer.

Watchdog Timer

A 2 second watchdog timer is built in and may bebypassed with a switch setting. The timer is reset whenRTS is false and begins timing when RTS is true. PTTwill be turned off if RTS is not removed after about 2seconds. A resistor and capacitor set the time-out value.This is the only analog timing circuit in the modem.

Remote Control

CkarData

Scrambler

Scr.lI!nbblD.~

17BitShiftReai~n

TnnsmisJioo Media

Descrarnbler Computers used in amateur packet networks areoften located with the TNCs, modems and radios ininaccessible places such as mountain tops. When thecomputer software crashes, which it often does, the con­trol operator doesn't want to have to go to the distant siteto push the reset button. A remote control reset functionis built into this modem.

Normally open relay contacts are available forwhatever use may come to mind. The contacts closewhen the modem decodes several milliseconds of apseudorandom bit pattern sent to it from another modemin convenient reach of the control operator. A push but­ton on the rear panel causes the modem to send thespecial sequence.

Scrambler Block Diagram

Note: G3RUH and K9NG scramblers use 17 bitshift registers tapped at stages 17 and 12. The sequenceproduced is not maximum length.

The sequence generator is an 8 bit linear feedbackshift register with user specified taps. The tap locationsare specified along with the frequency data in theEPROM. Each RXffX frequency pair may have a dif­ferent code assigned. Each set of taps produces a uniquecode. The receiving modem must see at least 256 bits of

Theory/Design 2-23

Page 101: Packet- Speed- More Speed

Pin assignments

Power: The power input connector is a common2.1 mm round DC power jack used in many other con­sumer electronic devices . Positive voltage is supplied onthe center pin. A 12 volt 2.5 amp external switchingpower supply runs the modem and transverter.

Data: A female DB15 connector is used for thedata interface . The pin assignments are the same as theOttawa PI2 card. A single switch on the PC boardchanges the electrical standard from balanced RS422 tounbalanced CMOS. The RS422 interface is based on the26LS32 and 26LS31 chips. The CMOS interface uses a74HC244 chip.

the pattern before it will start to close the relay. Thesequence must continue for another several millisecondsto allow the relay to close. A single incorrect bit in thesequence will reset the decoder so that another correctsequence of 256 bits are needed to cause relay closure .The front panel "ready" LED will change from green tored when the code is being received.

Signal Level Display

Ten LEDs on the front panel indicate relative signalstrength. The RSSI signal from the MC13135 receiverchip drives an LM3914 linear bar graph display control­ler chip. I have found this a most welcome feature. I'veused it to map signal coverage areas by first setting mystation to "ping" the local 56KB packet switch every 2seconds; then with only a modem, transverter andantenna in the car, I can get a good idea of how well thepacket switch covers various areas by watching the sig­nallevel LEDs and the DCD LED (and the road).

Tune-up and Test Aids

Since only half the EPROM storage was used forthe main FPGA configuration code, state and waveformtables, I provided a switch which allows the modem to"boot up" using the other half of the EPROM. In theother half is an FPGA configuration for a direct digitalfrequency synthesizer which is used to sweep themodems tuned circuits . It sweeps a 200 KHz range cen­tered on 448 KHz. A square wave is also generated onthe receive clock output line for scope sync. The risingedge coincides with the 448 KHz center frequency.Adjustment of the filters for proper response shaperequires only a 30 MHz dual trace scope. The scopemust be adjusted so the rising edge of the "receiveclock" square wave is centered and exactly one completecycle is displayed. The other channel can then be usedto probe various points in the modem to observe thefrequency response envelope calibrated to 20 KHz perhorizontal division. The receive filter can be checked ifthe an attenuator is placed between the TX and RX BNCconnectors. The transmitter becomes the sweep genera­tor.

A push button on the rear panel will activate thetransmitter and send scrambled marks. The 2 secondwatchdog timer is automatically bypassed to allowtransverter tune-up or power measurement.

Interfaces

Transverter: Power and PTT (Push To Talk)transverter signals are provided on a 5 pin DIN connec­tor . The remote control relay contacts are also on thisconnector. BNC connectors are provided for the 29MHz transverter IF signals. Pin assignments are as fol­lows:

1. PTT2. Relay contact3 Ground4. Relay contact5. +12 volts @ 2 Amps

2-24 Chapter 2

RS422

1. No connection2. + Receive Clock3. + Receive Data4. + Transmit Clock5. + Carrier detect6. + Transmit Data7. + Request to Send8. Mode Select9. Ground10. - Receive Clock11. - Receive Data12. - Transmit Clock13. - Carrier Detect14. - Transmit Data15. - Request to Send

CMOS

1. No connection2. Receive Clock3. Receive Data4. Transmit Clock5. Carrier detect6. Transmit Data7. Request to Send8. Mode Select9. Ground

10..15 No connection

(out)(out)(out)(out, low true)(in)(in, low true)(Not Used)

(out)(out)(out)(out, high true)(in)(in, high true)

(out)(out)(out)(out, low true)(in)(in, low true)(Not Used)

Page 102: Packet- Speed- More Speed

Internal Option Switch Functions

Switch Off

1. RS422 CMOS2. Normal RX Mute Disable3. Normal Repeater Enable4. Normal Scrambler Disable5. Normal Key Transmitter6. Normal Tune up7. Normal Watchdog Disable8. Frequency Select 29. Frequency Select 110. Frequency Select 0

Performance

Due to deadline and and other time constraints, Iwas unable to do a bit error rate test on the latest PCboard revision. The previous version needed about 2 dbmore signal to achieve the same bit error rate as theoriginal WA4DSY 56KB modem. This was a receiverproblem related to excessive wideband digital noise get­ting into the receiver RF stages. This will be resolvedbefore production.

Performance with off frequency signals seems to beat least as good as the original design, degrading onlyabout 1 db with a 5 KHz frequency offset.

The response time from RTS to DCD has beenmeasured at about 3 milliseconds. I'm using aTXDELA Y value in NOS of 5 ms and haven't encoun­tered any problems. This is much faster than the originaldesign which required a TXDELA Y setting of 15 ms.

The transmitter spectral bandwidth is about thesame as the original design.

Applications

This modem can find uses in several areas. .Whenthe original modem was introduced in 1987, the com­puting power available to the average ham was quitelimited and had problems keeping up with 56KB data.Today (1995) the average ham can afford a 66 MHz 486machine. Multitasking operating systems such as OS/2and Linux running on these machines allow hams to setup their own TCPIP Web sites on the air. Applicationssuch as a Web server are useless at 1200 baud. For thisreason, I believe this modem has as much potential foruse as a user LAN modem as it does for point-to-pointlinks. Assuming the built in full duplex bit repeaterworks as expected, I hope to see many 56KB FDX userLANs spring up around the world. They would workjust like an FM voice repeater except for the 70 KHzbandwidth requinnent. There is one such LAN inOttawa, Ontario, Canada.

We plan to put up a full duplex 56KB MetropolitanArea Network on 222.400 (input) and 223.85 MHz(output) here in the Atlanta, Georgia area using a 56KBmodem, a receive converter, transverter and a Sinclairduplexer. Users, of course, only need a modem andtransverter .

The U.S. now has a 1 MHz wide band (219 to 220MHz) for "point-to-point fixed digital message forward­ing". The band is divided into ten 100 KHz channels.This modem is ideal for that service.

Sales and Marketing

By now you're probably wondering how to get oneof these modems. You can't, at least not right now. Themodem is still in development. However, I'm negotiat­ing with a well known manufacturer of packet radioequipment to produce and sell this modem. I hope to seethem advertised for sale late in 1995 or early 1996.GRAPES will also be involved in modem sales.

Transmited spectrumHorizontal : 50 KHz/div isionVertical : 10 DB/division

Transmited spectrumHorizontal: 20 KHz/divisionVertical: 10 DB/division

A Prototype Modem

Editor's Note: They are now available from:

PacComm Packet Radio Systems Inc.

4413 North Hesperides Street

Tampa, FL 33614-7618

Phone: 800-486-7388 (Orders)

813-874-2980

Fax: 813-872-8696

Email: [email protected]

Web Site: http://www.janrix.com/paccomm/

Theory/Design 2-25

Page 103: Packet- Speed- More Speed

l\)I

l\)0'1

LIMITERFo = 448 KHZBW= 70 KHZ

MIXERFo = 10.693 MHZBW= 180 KHZ

MIXER

... a ... ... a ... ... a <, ... Quadra ture -... Pass ... ... Pass ... .. Pass y ... Detec tor

.4 ~ ..~

Freq . 10.245Synthesizer

MHZ39 - 41 MHz

4TRACKING .. NRZ I ... DESCRAMBLE .. DATA OUTDATA .... TO NRZ ... XJl 17+XJl5+1 ....

SLICER

t .4~ ... CODE ... REMOTE CONTROL.... DETECTOR .... RELAY CONTACTS

l... CLOCK .. CLOCK OUT.... RECOVERY • ...... 8 BIT FIFO ... TO TRANSMITTERJII"'" ... FOR BIT REPEATER

~,

~DATA + ...

CARRIER CARRIER DETECT..DETECTOR

Fo = 29 MHZTRANSVERTER BW= 2 MHZRX SIGNALl\)

o:::TIII"t:l-CD....

56 KB Modem Receiver Block Diagra m

Page 104: Packet- Speed- More Speed

10.245 MHz

•Mixer

.. .. _- \ I •.... a,..,..

Fa = 448 KHzBW = 80 KHz

I ~ / BANn\Pfo\;:);:) ,

EPROM32K x 8

56KB TX Dat a

~ Dig ital StateI ~

8 Bit

.... Machine DAC

56 KHz Cla ck

Clock64x

Baud Rate

Fa = 10.693 MHz Fa = 29 MHz 30 DB AmplifierBW= 180 KHz Mixer BW = 2 MHz

~ / ~~~~ \ I ~ ~ / ~~~I\] [> •29 MHzIF Output1MW

-I::TCDo...-e-cCDIII

cO":::::J

Freq .Synthes izer39 -41MHz 56 KB Modem Transmitter

Block Diagram

I\)

N.....,

Page 105: Packet- Speed- More Speed

I\)I

I\)(Xl

I--

>(/)o

~

~ziiiw

~:1: 1---ew

"-c'"'"'"e;::- 12~~~o

I~

~00-

j~

t= ' 3

IR: V

Sh oet 1 o f 4Drawn bv DAH

~

v

D

2 N44 0 1VCC

4lt.RS

WA4DSY 56 KB MODEM

L2

. 12

9

Numbe r

REPEATER ENABLE

SCRAMBLER DISABLE

RX MUTE DISABLE

SEND REM RESET CODE

KEY TRANSMITTER

C'

B

Rem ot e Cont ro l Relay

li t Ie

Dat e 3· 2 2 - 9 5

Filena m e 56KMOOEM.S0 1

Size

53

~5 2

...I...

~ 7e,~

I D54: 10

" ~ FREO SELECT 0

5 4 :912 -0 ' FREQ SELECT 1

54 :8 ;13 -o~ FREO SELECT 2

C'

C '

I I I I

T

1

5 4 ' 3

~~19 54 :2~~

5 4 :4,--_-'-' !-7--<~

54 :5,--__.J'..!.._~ KEY TRANSMITTER

RSO

4 .7 K

LAST REV. 6- 18 · 95

44 8 KHZ

~

TX SIGNAl..

.WlH DQW.'/?11 TH ENS""llH DATA

c

c

~

IC6 4 .7UF I R8

1~4 '/+ I 4 7 0 K

7 • ~ 2 SEC WATCHDOG TIME

WATCHDOG DISABLE

R2

4 .7K

U2 .U4 ,U5 PO\\l:RED FROM THIS LINE

F}C2 2

1 1UF lO I UF

I

BASE BAND RX SIGNAL

1121314 15 1' 17 1,

j Iii:IJIlI: I ~ I,

' I'l' l' I'I' l' 1O~ 543 2 1 0

rtdbd-,"),41· ROATA 3

76 MO- ATRIG ~

1 5 AI . £!2. v o _ _ 12KJ · VIS- II 0 PWADN

110 .05 4 RESET ~ ~ ~ ~

v o 5 113 TClKIN. 1/ 0 110 50

I/O '"71 A:2. v o 110 '"71 A:) - II O LIO <1712 M . VO 110 4 '. .. A!i. II 0 v o 045

~ :: ~ ~ ~ ~~ :~51 M - 1I0 0 110 "0

1 1 Ali . v a (f) 110 as10 AID - vo U 1,10 3 .

• Al , . vo X 110 37.. A12 . II a vc 352 A13 -1I 0 v o 3 0 DCD-

:i ::::~~ C3 ~ ~ ~ : =~~..n _ fr: II 0 ~ ~ TXO55 INI1:. 11 0 110 25 TXCl K

7 o. P I/ O 24, II 0 I/O 23

71 ' 0 _ '0 " AXCLX > ~.. ADY/BUSY· ' 0 '0 ac I =I/O 110 3.. ' 0 ' 0 " s rnlln OCX ~ ,:~ CS I. va va ~ ~ ( i . I

I I I I I I II !...£. II 0 165 11 ~~ . va ~~ 1 5

5 7 1l1. , . BeLKIN- II all 0 14r-t-

QI(;l I l

B

72 DO. DIN-7 0 01 . 1/0

~~" D" ' D&5 03 . II 0

62 [)4 - 1106 0 0 5 . v o5. [)G - va56 07 . v a

eo~

""~

"oa:0­w

U4

~I--!-

1 1 Q I12 0 213 Q3

" oa' 6 as17 oal' Q1

" oaTM5 2 7 C2 5 6 J

2 .7KA./V'vR1 2

B I

VCC

T NOTE:

R l I~

VCC..,..

vee I .t 1>---o---ll-Tu ne Up 54 :6

C6 1

. 1 uF

.12

T

14 I

T

C57

~

R1 3

.1 u F

4701<R1 8

470 K

"""'o OUT

I

Cf)

L.--J C4

1 4. 3 18 18 MH Z r. 0 I U ~

n n

1:7£ 8FT T luF

±f2~.7UF

Tt .eklng Oa 11 Slic . ,

Y'~Y2 7

Y3 'Y4 3

LM7805 crn

U3 :B

_. I

U l

7 4HC2 40

A

A

,I

C'

I

0 5~,

74 HC 4 0S3

R1 5

"1 K

~-~~ .A "

"'"DOW-. I L

I I J 99

'"'18 V

5 1

VCC~

"~". ,..C ...

cor-

C'J l

RX DEMeO

12 VOLT PO\\l:R INPUT

4

3

'7VCC

READY PTT DCD

0 1w 0 3

~~2~a:

@~DlZw a: ~~ "w "a:

o 2

0 ec O ~M,..a: ~ a: .. ~ ON a: ,..

21N

zw

C ww a:a: o

" "'" '"+ +

~

~

I\)

o:::TDl

"'lJ-CD....

Page 106: Packet- Speed- More Speed

I-

o

10.7 MHZ

180 KHZ BW

TOKO- CFSK

.-

I

2N4 401

~~~

-~o on- ... ", > -a: '<t a:'-lt)

cEMIFILTER

59 L22

101UF

. 1 UF

I

C5

~I U8

~OUT' A~

. 1uF

vec

T 'T' "" m448 KHZ TO 10 .693 MHZ UP CONVERTER , I

I~C9

B

. ;;~ll? i 2 I IN2

Tl'8

L2

2 20 0PF 2%

n5 60UH

.3 GND • \I~ C11 esc- a esc- ,ka> -r- NE60 2

I

Ban dpass f ilterFo • 448 KHZBW . 80 KHZ

'0 1500PF 2%

L1

L3 L4

A

2

1 I 1 3 I

VCC

EMIFILTER

3

f-

f-

IRe:

. 12'11

Sheet 2 of 4

Drawn bv DAH-

10 t,4ETER IF OUT

D

56KB Modem Transm it IFI it ie

Date 3· 23 - 95Filen am e 56KBMODEM.S02

I

LAST REV. 6- 18- 95

L6

c

4 .7 uH ~C15 ll'"

6BpF L8 ~Tu

47 uH

Ba ndpass filter

Fo • 29 MHZ

BW = 2 MHZ

C21

30 DB AMPLIFIER

c1J,W' .ra

.01 Uf

L9

I I I I I ~

...L

T.0 1uF

L-

I

COPYRIGHT (Cl 1995 DALE A. HEATHERINGTON WA4D5Y

r-

f-

'0

51'S INumba r

WvR3 0

-&

"' ~o'" ...a: " U9 C14

, ::: : :::: 'Oi1 UF

3 GNO • V '

• 7 -osc. B osc- E L7 ....'5-l NE60 2

;;T .47 uH

330PF

I rx FREQ-,.--

330 PF 7

B

C2 5

C3 0

l C2 0

101uF

2N4401

~34." lK

~

0 4r.:::

C2BLr­10PF

'01

27 pF

R3 3

4.7 K

0 5

C2 6~2N440 1

R2 9

150 K

a>~o'" ona: _

C24

1/1\

330PF

L10 L..--.j( I I

39 MHZ Tran sm it veo

A

1N544 4A

06

R2 7

lOOK

C23

.0 1uF

C27

R31

150K

.1 uF

.--

TXPD OUT

Tl'1

L...-

4

3

f-

I-

-I~

CDo...~cCDCIl

(Q:J

I\).I\)to

Page 107: Packet- Speed- More Speed

I\)I

(,)o

2

3

J-

J-

J-

6

>­U)o

~

§za:w

~wI

'"w;i.o

'"'"'"

~a:>­a­Do

ASS I

o

RX DEMOO

I

VCC

EYETP3

~~;;;? a:

'&

is

OP ou'f1---lL

op

I:

c

oEMoooUT

EMIFILTER

F7

I;

R3 74 7K l C3 2

-r­15PF

RSSI

TP4

iC45 C46, "T'

~.1 UF J OOOPF

C5 1-L

J 1UFIF

TP5

QUAo

L11

3 .9 m H

R3 8

39 0

J-

MC13 13 5

2MXIN

~

LIM

IN

~ ~~

r:iliJ'=~

C31

'1LJJ'1 O.

1 1 VM M CX X CI 0N U1 T

U1 2

.H

R3 9

I

TP2RX VCO

10.7 MHZ IF FILTER

,MXIN2

SYNTH LOC K

B

VC

1 1 0 2 2L L 0 L L0 0 U V 0 0

• E T C E •C

~

C52

10P{

/I

330PF C4!l.L

33 PL

~VO- A

C4 7

Z L15

, ' 47UW

Ii

R3 6

TVCC

C60 .1 UF\I

R51.II/V'33 0K

r-t-

";1 r14:t :i~-'

..LC5 0-r-

68p F

I', LOCK

TP6

o

u,

"o

~

R35' 0 • .. ..... I 'VV'v

15 0 K

voo~

L1 4

,4 7 uH

L12

4 .7u H

2ND LO

L13

.4 7 uH

'&

~3 6u.C.

1Xl

L.':'.-

29 MHZ BAND PASS FILTER

1\

C49

3 3 0 pF

C4 1 8 2 pF

"

10 .245 MHZ

J4

A

---10Y1

C3 5

'&

SYNTH DATA.

3 4

TRANSVERTER RECEIVE INPUT

''"'ffl2 1rl 3 8 39

IJlJl )

3

J-

J-

I-

I\)

o::rQ)

't:l-CD...

~

l C5 6-r-

56KB MODEM RECEIVER4

A 1 B

RXVCC""'l:.­

C53

O. l uFl L18

100uH

1500 PF 2 %

1

~ 2 20 0PF 2 %

448 KHZ IF FILTER

c

15 0 0PF 2 %

LAST REV. 6, 18, 95

1

lltl e

S llS INumber

Dat e 4 · 19 · 95Filenam e 5 6KMODEM.S03

o

-fR:VDraw n bv DAHSh e e l 3 o f

Page 108: Packet- Speed- More Speed

-'lD l lRXD

-'l O12 TXClK

De D-

-'l O14

~-'l O15

~-'l O16

D

><Il0

~z12eza:w

~:I:

..:w~0

:::'"uI-

MOOE· ~ 12a:>Q.

0U

DB 15 PINS ( RS422)

1 NC

2 10 RX CLOCK

3 11 RX DATA

4 12 TX CLOCK

5 13 • CARRIER DETECT

6 14 TX DATA 137 15 • RTS

8 • MODE ( CMOS ONLY)

9 GND

~

J P l

Not e: The fu nc t ion 01 -, MODE " is

undef ine d at this tim e .

Tho sig na l is supplie d 10 t he

Xilinx FPGA on an unused pin (25 ) .

It

J 5,-- ---------, u.

tv"t'V'\'Y\(Y'o(Y"'(YwY'ti:'H-H-H-H-H-H+-t-t' 0

II rr

_17161514131211

~

~~:

~~~~~~

r-t--t r-t-

~

74HC244

c

2 6 lS32

U1 3

~Al ~A2 ,

~ ,7 4 HC244

R40

VCC

~

U1 6 :B

19 G

" ~1::1 Al Yl 715 sa Y2 51 1 A3 Y3 ~

VCC-r-

TP7IRX ClK

TXD

R1'5-

B

C6 2

J1UF

VCC

A

U14LM3 9l4N.---­

LEO 10R HI

LED I

2 1 C AA 1 H45 7 LED ,I--llREF OUT I 13

33 0 l ED 1

LED 'REF ADJ

LED 5

R LD ~'6S"0· I I l EO 4 17

RSSI \l LEO 3SIG IN ,.

LED 2

LEO 1MOOE

3

BAR GRAPH SIGNAL SffiENGlH DISPLAY

4 .7k

c

. Rev

A

Shee t 4 of 4

Dra wn bv DAH56KMODEM.SQ4

D

Jille

Date 4- 18 · 95

SI'S INum ber

il

LAST REV. 6 - 18· 9 ... Filena me

R4 8 10 0

VCC

R4 7 TOPEN = RS42 2

CLOSED = TIl.

-~~

-&

B

-I~

CD 140...'<-C

CDUlIe 1 A:::l

NW......

Page 109: Packet- Speed- More Speed

Data Radio Standard Test Methods

Burton Lang, VE2BMQDonald Rotolo, N2IRZ

North East Digital AssociationPO Box 563, Manchester NH 03105

AbstractThe Data Radio Standard Test Methods document is introduced and explained. Thedocument consists of a number of standardized test methods, written in a clear, step-by-stepformat. Each test method is designed to be easy to perform, yet yield meaningful results.The rationale and organization of the DRSTM are discussed, including the proposed DRSTMdatabase of measurement data.

IntroductionWith the ever-increasing demands of modern packet networks upon the radios being used, there existsno standardized group of test procedures for these radios, or even a list of which radio characteristicsare important, and why. If we are to maximize the efficiency of the shared radio networkingenvironment we use, then we must have a better understanding of the significant characteristics of theradios being used. In an effort to resolve this apparent lack of information , VE2BMQ began writingthe Data Radio Standard Test Methods in August 1994. In October 1994, the Board of the NorthEast Digital Association endorsed this effort as beneficial to advancing the radio art, and encouragedother organizations to do so as well . In November 1994, the Board of the Radio AmateurTelecommunications Society also endorsed the DRSTM . In this paper , these efforts will besummarized.

Why do we need this?Recently, a test report of radios advertised as "9600 Baud ready" appeared in QST [May 1995, pp24-29] . The author, Jon Bloom KE3Z, provided a detailed analysis of the bandwidth considerationsfor proper operation at 9600 baud, also briefly mentioning deviation and its effects. He then providedtest results of the Bit Error Rate (BER) for a few radios, a critical yardstick of data radioperformance. However, the BER is measured in a continuous data stream, without the normallyfound pauses between groups of packets. These pauses, which are set by TXDeiay (transmit delay)can significantly affect the overall performance of a packet radio link.

2-32 Chapter 2

Page 110: Packet- Speed- More Speed

As an example, compare two commonly used 70 em radios: The TEKK T-Net Mini (Model KS-900L)and the ICOM IC-449A. The TEKK has a RXITX turnaround time of less than 12 milliseconds,while the ICOM has been measured at over 410 milliseconds. At 9600 baud, this translates to a lossof data capacity on the order of 80%. From the table below, we can see that a 2400 baud link with aTXDeiay of 40 mS has a higher throughput capacity than a 56 kb link with a TXDeiay of 500 mS!

Baud Byte Throughput (bps) given a TXDeiay of:Rate Time OmS 40mS 250mS 350mS 500mS

1200 6.67mS 127 121 99 91 812400 3.33mS 254 233 163 143 1214800 1.67mS 506 431 241 199 1589600 .833 1015 750 316 248 18756k .i04 8131 2124 435 316 223

[Reprinted from NED A 1994 Annual, pg .81 . Assumptions: 230 Byte data per 256 byte transmission, 16 byte acknowledgement, bothtransmitters have same TXD]

The point is that one parameter that was not even measured in the test has a huge influence upon theperformance of a data radio in a real packet network. If such a serious omission can be found in atechnically competent journal such as QST, it can be said that most amateurs have an incompleteunderstanding of the parameters affecting data radio performance.

OrganizationThe Data Radio Standard Test Methods (DRSTM) is organized much like the published procedures oflarge standards organizations, such as the ASTM, IEEE or SAE. Each test method first defines theparameter to be measured, then explains its importance to data transmission. A detailed, step-by-steptest procedure is then provided, along with set-up diagrams and a standardized form for recording andinterpreting the test results. This ensures that all test results, whatever their source, are as reliable aspossible . Efforts are taken to use commonly available test equipment wherever possible. This allowsas many people as possible to perform the tests . The tests themselves are kept as simple as possible,while still yielding meaningful results, which allows a wide range of radio equipment to be tested.While we believe that many radio users would not actually perform measurements, just knowing howa particular parameter is measured can offer insight to its effects.

In addition, the DRSTM manual contains a detailed Glossary, a thorough explanation of the databasestructure, and performance requirements for all test equipment.

The scope and significance of each test that has been written so far is summarized below:

DRSTM-ol Transmitter Power-on Time Delay

Scope: This test procedure is intended to measure the time from the start of the transmitter keying(push-To-Talk or PIT) line becoming active until the RF output has risen to 90% of its final value.

Significance: The time delay that a radio transmitter's power output has when it is keyed can rangefrom microseconds to hundreds of milliseconds. A transmitter that is keyed but not putting out RFpower can create a serious 'hidden transmitter' problem. In addition, in the case of a synthesized

TheorylDesign 2-33

Page 111: Packet- Speed- More Speed

transmitter, a significant difference between the Transmitter Power-on Time Delay and theTransmitter Power On-frequency Time Delay (see DRSTM-02) would indicate the possibility ofserious interference to users on other nearby frequencies .

DRSTM-02 Transmitter Power On-frequency Time Delay, Frequency Stability Time Delay,and Modulation Stability Time Delay

Scope: This test procedure will measure three Data Radio characteristics:1. Time from PIT becoming active to the RF output appearing within the designated passband of atest receiver.2. Time delay until the transmitter output frequency has stabilized to within 5% of the channelbandwidth relative to its final stable frequency, or until the PLL loop tone or any other extraneoussignal has decayed to less than 10% of the normal system modulation level, whichever is longer.3. The time delay until the transmitter's modulation envelope has reached 90% of its final stablevalue.

Significance: The time delay until a transmitter's output frequency appears within its intendedchannel can range between microseconds and hundreds of milliseconds . Any difference between thePower Output Time Delay (see DRSTM -Ol) and the Power Output On-frequency Time Delay wouldindicate the possibility of serious interference to users on nearby frequencies . Any substantial delayin the appearance of the transmitter output signal in the receivers of other users on the same channelcan result in a serious 'hidden transmitter' problem.

The additional delay of waiting for the frequency to settle down, or signals that have a significantamount of non-data modulation (such as the damped oscillation of a PLL oscillator feedback loopimmediately following lockup) can have a serious effect on decoding the data signal. This couldrequire a much longer TXDeiay setting for usable operation.

Similarly, the time delay for the modulation envelope to settle down to its final stable value can alsoaffect proper data recovery at the beginning of a data transmission. This rare condition has beenobserved in certain radios such as the Motorola MOCOM 35, where a large RC time constant in thereactance modulator retards the modulation envelope by several hundred milliseconds .

DRSTM-03 Transmitter Short-term Frequency Stability

Scope: This test procedure is intended to measure the transmitter and receiver frequency changes thatoccur when the radio is exposed to extreme temperature and supply voltage conditions.

Significance: Excessive changes of either the transmitter or receiver channel frequency can causefailure or degradation of packet radio links, if the change forces the signal beyond the passband of thereceived at either end. This information is important to designers of packet network facilities whereequipment may be housed in unheated enclosures , as well as builders of emergency response networksthat must not fail under extreme conditions .

It must be recognized that the quartz crystal used as the primary channel element or as a reference fora PLL is a major factor in frequency stability. Amateur radio users often use crystals from suppliersthat may not follow the manufacturer's specifications. testing radios using such crystals can lead tomore appropriate selection of crystals for particular radios .

2-34 Chapter 2

Page 112: Packet- Speed- More Speed

DRSTM-04 Receiver Detector/Frequency/Squelch Recovery Time Delay

Scope: This test procedure measures:1. The time between when PIT is released until a signal is received at the detector output2. The time between when PIT is released until the receiver center frequency is within 5% of itsdesign bandwidth relative to its final stable receive frequency, or until the PLL loop tone or otherextraneous signal has decayed to within 10% of the normal demodulated data signal amplitude,whichever is longer3. The time between when PIT is released until the receiver squelch circuit (and all precedingcircuits) begin to pass a demodulated signal.

Significance: A useful data signal cannot be recovered from a received signal until the detector andall RF/IF/LO circuits ahead of it have recovered following a period of transmission.

Radios which take a long time for their frequency to stabilize, or that have considerable 'non-data'signals (such as the damped oscillation of a PLL oscillator feedback loop immediately followinglockup) can have a serious effect on decoding the data signal. This could require a much longerTXDeiay setting for usable operation.

Radio squelch circuits which take a long time to open following a transmit period would presentdecoding difficulties when the TNC or modem is connected after the squelch circuit (such as at thespeaker jack). This would also require a longer setting for TXDelay .

DRSTM-05 Receiver Squelch Turn-on Time Delay

Scope: This test procedure measures the time between the start of a test signal until the squelchcircuit in the radio opens and passes demodulated audio.

Significance: When using a TNC or modem connected after the squelch circuit, the time it takes forthe squelch circuit to open and pass audio affects operation. If there is a significant time delay inopening the squelch, the TNC could decide to transmit before it realized the channel was occupied.This would cause a 'hidden transmitter' like problem .

DRSTM-08 Receiver Output Level, Impedance, Demodulated Frequency Slope (De-emphasis) and Demodulation 6dB Cutoff Bandwidth.

Scope: This test procedure measures :1. The voltage level of the demodulated data signal output, per unit modulation level.2. The impedance at the point where the demodulated audio output signal is connected.3. The amount of audio output signal variation with frequency (De-emphasis slope) measured wherethe output is connected, expressed as dB per octave.4. The audio frequency at which the demodulated signal drops to one-half the voltage on both thehigher and lower sides of the normal operating frequency range.

Significance: The level and impedance of the demodulated audio signal from a data radio receiver isuseful information when designing data systems and interfacing TNCs and modems. The .de-emphasisresponse of a receiver affects TNC or modem operation. Most TNCs and modems require a flatfrequency response while most radios offer some pre- and de-emphasis to improve voice quality. Foroptimum performance, these responses should be matched. The high side frequency response helpsdetermine the maximum usable bit rate, while the low side frequency response is important withcertain direct FM modulation systems (such as the G3RUH modem).

Editors Note:the latest information for this project may be found on the Web at:http://www.rocler.qc.ca/burtldrstm.html.

Theory/Design 2-35

Page 113: Packet- Speed- More Speed

DRSTM-09 Transmitter Modulation Drive Requirements, Input Impedance, ImpedanceResponse Slope, Pre-emphasis Slope, Maximum Deviation and ModulationFrequency Capability.

Scope: This test procedure measures :1. The voltage level of the modulation signal per unit of transmitter deviation.2. The input. impedance of the modulation circuit .3. The response characteristics of the modulator input impedance, as it varies with audio frequency.4. The variation in the deviation when the modulating frequency is changed (pre-emphasis) .5. The maximum deviation capability without distortion.6. The highest modulating frequency which does not reduce the first modulation sideband by morethan 20dB.

Significance: The level and impedance of the modulator is useful when designing data systems andinterfacing TNCs and modems. The modulator frequency response characteristic should be matchedwith the receiver response to obtain an overall 'flat' system response. A non-flat system responsewill distort data signals. The impedance variation of the modulator with frequency can also affect themodulator's frequency response characteristic. The maximum modulation (deviation) capabilitywithout distortion is useful when designing data systems and interfacing TNCs and modems. Themaximum modulating frequency is useful in determining the maximum capabilities of a higher-speeddata link, as well as in avoiding the radiation of wide sidebands produced by computer clock noise,etc. FM transmitters with direct connections to their modulators have been found radiating 10 MHzwide sidebands, caused by leakage of the TNCs 5 MHz CPU clock.

DatabaseAn integral part of the DRSTM concept is a database containing the measurement results obtained bythe DRSTM users . Each DRSTM enables the user to make the same measurements, consistently, andprovides a form on which to record all measurements. It is well known that radios have differingcharacteristics for many parameters, even radios of the exact same type and of consecutive serialnumbers . It is anticipated that, by provid ing an international clearinghouse and database of allmeasurements, that the amateur community would be better served . this may also provide anincentive for radio manufacturers to either publish the data-relevant parameters, or at least designradios with data transmission in mind. While no firm plans have been developed, it is expected thatthe database would be available on-line in some fashion, with free access for all.

ConclusionThe establishment of standards for various radio characteristics having significance in datatransmission will eliminate much of the confusion and misinformation in the area today. Thesestandard test methods, used to measure the performance of data radios, could be used by anyonehaving reasonable experience with common electronic test equipment. The international databasewould disseminate this collected data and, as manufacturers noticed that certain radios were unsuitablefor data use, convince radio designers to modify their designs to acommodate data transmission. Inthis manner, the authors hope to improve the radio art.

It is the author's hope that other knowledgeable persons would step forward and offer their expertisein either writing standard methods, suggesting new tests, evaluating existing tests, or performing testsand disseminating their findings. At this time, these standard tests should be considered tentative . Ifyou can help in any way, please contact the authors .

2-36 Chapter 2

Page 114: Packet- Speed- More Speed

javAPRSImplementation of the APRS Protocols in Java

Steve Dimse [email protected] http://www.bridge.net/-sdimse

This paper describes my implementation of the Automatic Position Reporting System (APRS)protocols in the computer language Java. APRS is one of the most innovative uses of ham radio in recentyears. javAPRS extends the usefulness of APRS to the internet.

Java Basics

Java was designed by Sun Microsystems as a language to promote the use of distributedcomputing resources over the internet. Based on the C language, with an object programming philosophydrawn from Smalltalk and Lisp, platform independence, and built-in security, it is a language uniquelysuited to network programming.

The principle drawback to the use of Java at present is slow execution speed . Unlike traditionalprograms, which are compiled into machine specific object code before distribution, Java is compileddown to byte code which is then interpreted on the local machine, often executing at 1O-25~ of the speedof native code. There is reason to be hopeful, however. Several companies are working on 'just-in-time'compilers, which convert byte code to native code on the local machine just prior to execution. Also, thehorsepower of today 's computers makes even the interpreted version run at an acceptable speed in mostapplications.

The leading use of Java at present is to write programs, called applets, which are run within aWorld Wide Web (WWW) browser, such as Netscape. javAPRS can work in this fashion, or as a standalone program . Security restrictions limit some of the interesting possibilities for javAPRS as an applet,such as using separate servers for map and APRS data, or connecting to multiple other computers to plotdata from multiple LANs.

javAPRS Design

In the basic design of javAPRS, I have tried to create a system which can be used now by peoplewithout programming knowledge to add APRS data to their web pages. In addition, using the objectoriented programming (OOP) features of Java, the system is designed to be easily extended by other Javaprogrammers. This sort of extension does not require access to source code of javAPRS. The interfacesused by other programs will be posted in my web pages as they are finalized. Source code will not befreely available, but I will consider requests for the source code on a specific basis. The remainder of thispaper discusses the use of javAPRS in web page creation . Programmers interested in extending javAPRSshould contact me directly for more info.

javAPRS Applet Parameters

The basic syntax to call an applet in hypertext markup language (HTML) is:

<APPLET codebase = "javAPRS /" CODE="j avAPRS .class " WIDTH=400 HEIGHT =300>

Theory/Design 2-37

Page 115: Packet- Speed- More Speed

This executes the applet "javAPRS.class" in the directory "javAPRS" (relative to the H1ML file the call isfound in). It allocates a space of the size indicated in the browser window. Other HTML commands, suchas <CENTER> work just as they would for other graphic elements such as images. The lines afterAPPLET tag consist of a series of parameters which are passed to the applet to control its behavior. Eachparameter has a default value, usually the most commonly chosen option, and if the default parameter isdesired, it need not be declared .

Map Parameters

At this time, javAPRS understands two kinds of maps . One or the other map must be used, or theapplet will not run. The map files are stored in a subfolder "/maps" relative to the "codebase" named in theapplet call.

<PARAM name = "dosMap " va l ue = "anymap .map "><PARAM name = "g i fMap " value = "a nymap .map" >

Maps can be automatically or manually scaled.

<PARAM name = "au t oSca l e " value = "t rue "> (Default t rue )

This will cause the map to be scaled to fit the window the applet is presently running in. If autoscaling isnot used, then the following three parameters may be used to set the magnification and offset of the map.For now, the only way to figure out the value of these parameters is trial and error.

<PARAM name = "scale" value = "2 . 0" > (Def au l t 1. 0 )<PARAM name = "o f f s e t X" value "100 "> (De f au l t 0 )<PARAM name = "offse t Y" value = "100" > (De f au l t 0 )

Two options work only with dosMaps, namely:

<PARAM name = "showMapLabels" va lue = "true "> (de f au l t true )<PARAM name = "sh owAl IMapLabels" va l ue = "t r"-:e" > (d e f au l t fa lse )

which will show either all map labels or those designated at the present scale.

Data Parameters

Data to be displayed by javAPRS is one of three types, either NMEA (only RMC and GGA arerecognized at present), TNC data (raw data from a TNC which is the MacAPRS log file format), and HSTfiles produced by dosAPRS. Any or all of the three types of data may be displayed, but only one file ofeach type can be used.

<PARAM name = "NMEAf i l e " v a l ue = "NMEA .data "><PARAM name = "TNCf il e" value = "ko4hd .da t a "><PARAM name = "HSTf il e" va lue = "ma r a t hon . da t a ">

The way the data is displayed is affected by several parameters:

<PARAM name = "displayVectors " value = "t rue "> (de f au l t true )

This will draw vectors for course and speed info if present in a position report.

<PARAM name '= "showCa l lsigns " va lue = "t rue" >

2-38 Chapter 2

(de f au l t true )

Page 116: Packet- Speed- More Speed

This prints a callsign next to each position report.

<PARAM name = II s t a t i o::.Li s t ll va l ue = Ut ru e ll >

javAPRS can keep a station list for the stations that are 'heard' in the data stream. At the end of a filereadin , the contents are dumped to the java console (select the option "Open Java Console" in Netscape. Itwill also speed up the redrawing once the data have been read, only the last position of each station will beplotted.

If stationList is true, then this option will show the name of each new station as it is heard . If the homestation has been specified, the bearing and distance to the station will also be displayed.

There are a number of other, less important parameters that are available to fine tune the display to suit theuser. Please refer to my web pages for more details.

jayAPRS Sample Applications

The javAPRSclasses that exist now allow a person without programming skills to create a WorldWide Web page containing an APRS map, and display a file containing positions of various APRSstations, objects, and track plots. More complete instructions may be found on my web site. Here are threeexamples, with the HTML code used to call the applet, and the URL's to reference them.

Theory/Design 2-39

Page 117: Packet- Speed- More Speed

HST file replay(h t t p://www . br i dg e . net/- s d i mse/marathon. h t ml)

<APPLET codebase = "JavAPR2/" CODE= "]av?PFS.::: l as2 " W:::YCH=S 'X HEI':;HT =::' 50><PARAM name "dosMap" va lue = "washdc .map" ><PARAM name "HSTf i l e " va l ue = "mar a t hon .hs t" ><PARAM name "s l eep" value = "50"><PARAM name "s t a t i onl i s t" value = "f a l s e" ><PARAM name "showSt a t i onNames" value = "f a lse" ><PARAM name "copyr i gh t Top" va Lae = "f a l s e "><PARAM name "s ca l e" value = "1 . 6" ><PARAM name "of f s e t x " value "620" ><PARAM name "of f s e t y" va lue "310" ></.ll.PPLET>

2-40 Chapter 2

Page 118: Packet- Speed- More Speed

GIF files for maps(h t t p: / / www . b r i dge . ne t / - s d i ms e / gi f map . h t ml )

<APPLET codebase = " j av APRS/ " CODE= " javAPRS. c l ass " W:::DT!-i=50C' HEIGHT =:'50><PARAM name "gifMap " value = "cudj oe .gif "><PARAM name "gi fMapLeft" value = " 81. 558 " >< PARAM name "g i fMapTop" value = "24 .7 "><PARAM name "gifMapPPDh" v alue = "3800 "><PARAM name "gifMapPPDv " v alue = "3900" ><PARAM name "sleep" value = "700 "><PARAM name "LLfile" value = "boa t t r i p . ll" ><PARAM name " s t at i on l i st" valu e = " f al se" ><PARb,M name "c opy r i gh t Top " value = " fa lse ">

< /APPLET>

Theory/Design 2-41

Page 119: Packet- Speed- More Speed

2-42

Display of TNC data capture(http: / /www.bridge . net /-sdimse /trip.html )

<APPLET cod ebase = "Ja vAPRS/" CODE="J3vlI.PRS.class " W:urH=500 HE::::;HT =::'5 0><PARAM name "dosMap" value = "usa .map" ><PARAM name "TNCf i l e" va lue = "::nc .dat a" ><PARAM name "dr awvect ors " va lue = "t ru e" ><PARAM name "home I D" value = "K0 4HD-9" ></AI:PLET>

:r:::~,,:::::.::,,: :!'t::::: ::: ::::,., : ::,...:::::::: : : ::::: ::::::::::::li'ffi

Chapter 2

Page 120: Packet- Speed- More Speed

Future directions

The samples above are static data, and although there is some degree of animation while the data isplotted, they could be done just as easily with screen captures and OIF files. Soon I hope to have thesystem able to access real time data over the net. This is where javAPRS can fill a unique role. Thepossibility exists to incorporate internet connectivity directly into the standalone APRS programs currentlyin use, allowing web users to see the data obtained at various APRS stations across the country in realtime. Also, it would be possible to write a server program, that could be connected to many differentAPRS sites and share their data, creating a nationwide APRS network, no longer limited to 300 baud andthe vagarities of HF propagation. I plan to pursue these plans, and hope to have some results to share bythe 1997 DCC, if not sooner.

Editors Note:

For current information on Java try these sites:

http://www.bridge.netl-sdimse/javAPRS.html

http://www.bridge.netl-sdimse/javAPRSprog.html

Theory/Design 2-43

Page 121: Packet- Speed- More Speed

On-air Measurements of MIL-STD-188-141A ALEData Text Message Throughput over Short Links

Ken [email protected]

For the past six months a colleague WlIMM and I have conducted automatedmeasurements of throughput when ASCII text files are sent over short or "tactical" HFpath s using the ALE Data Text Message (DTM) engineering orderwire (EOW). This is apreview of our results.

Tactical HF links generally use either surfacewave or near-vertical-incidence skywave(NVIS). Surfacewave works out to about 50 miles and NVIS to about 300 miles.Multipath, D-layer absorption and interference usually affect NVIS, which usually haslower throughput than communications over "standard" (i.e., long) one-hop skywavepaths.

ALE systems' ability to measure channel quality, and use it to make choices of goodchannels, now offers improved performance over tactical paths. ALE systems employ aslow but robust waveform that uses interleaving and two kinds of forward errorcorrection (FEC) to combat the fading, noise and interference of HF channels.

ALE standards prescribe three engineering orderwire protocols for half-duplex datatransfer: the Automatic Message Display (AMD) mode, the Data Text Message (DTM)mode and the Data Block Message (DBM) mode. DTMs can transfer ASCn text usingthe ALE waveform and an ARQ protocol.

We have carried out more than 200 measurements of throughput (in char/s and char/s/Hz)using DTMs on a 35-mile path . We have used 125-watt Harris RF5022 ALE radios,which implement DTMs of constant size (300 bytes) and "memory ARQ," in which up tosix erroneous repeats of a message segment sent as a DTM are stored and compared whennecessary in an attempt to construct an error-free segment. The RF5022's ALE firmwaresegments messages longer than 300 bytes into 300-byte DTMs. These D'I'Mssegrnentsare ACKed one at a time . Experiments suggest that messages 300 to 1000 characterslong produce the highest throughput consistent with shortest run time.

Our two stations used broadband sloping longwires that allowed the radios to drive theantennas without tuners . The antennas have both vertical and horizontal components, sothat they can launch both surfacewave.and NVIS signals.

The ALE modems were programmed to try frequencies between 2 and 16 MHz.(IONCAP runs suggested that any link above 8 or 9 MHz probably used surfacewaves,which were chosen frequently at night, when interference was heavy.) The tests coveredseven months from September, 1995, when the average sunspot number was near thebottom of its cycle.

Our measurements were automated by two C-programs. The first runs throughput tests .At the start of a test, and between DTM transfers, the receiving station is scanning the setof programmed frequencies and will stop upon hearing a call. If a link occurs, the callingand receiving stations negotiate a DTM transfer and the sender begins sending the DTM.

The program usually starts by performing a link quality assessment (LQA) exchange,which gives both stations up-to-date info on channel quality. After the exchange, thecalling radio links with the receiving one. When the calling radio informs the program

2-44 Chapter 2

Page 122: Packet- Speed- More Speed

that a link has been established, the program commands the local radio to list the channelnumber and the corresponding LQA scores for later analysis.

When the LQA scores for the linking channel have been logged, the program prepares thelocal radio to send an orderwire. It then uploads a standard English ASCII text file fortransfer. After sending commands that require a particular response, the program invokesa function called chkresponse () that scans the serial-port input from the local radio forthe appropriate response ("LINKED", "MESSAGE RECEIVED", etc.) .

As the program runs, timers set by the computer's clock measure "link time" and"message transfer time." Message transfer time is the time between start of character-by­character uploading of the file to the local radio and receipt of the MESSAGERECEIVED notification. Transfer time does not include link time. The programcalculates throughput by dividing the number of characters sent by the message transfertime. Since the message transfer time includes the few seconds needed for the receivingstation to send the MESSAGE RECEIVED frame, the throughput measurements areslightly pessimistic.

The throughput-measuring program writes its results to an archive file. The archive filestores appended, time-stamped data in abbreviated format that can be analyzed off-line bya statistics program. Here is abbreviated statistical output for all the tests run up to 6 May1996:

nOYLE_linJr.s = 235E (l ink_t1~e_ALE) ~ 25 .l6 S , sd (link_t imeY LE) = 21 .42 sE(ni:1.11:;fec~imeJ.LE) = LJ9 . C) s . :3d( trclns f er_tml,,:_ALE) = CJ8 .7 sE ([l(,-file_cr;ars~::";.E) ;:. 770 .4 , s<:1 ( n(;_f i le_cha~-:::J\LE ) = 615 . 1E(cput_ALE) = ::; . ::1 cps , sc1(tLJut __~E) = 1. 16 cps , sd(mean_tput_AI..,E) = 0 .08 cpsn1ax_t hrup.lt_.~E ;:. G. 60 '~rB , E(thnlpltt_ALE/Hz) -t; 0 . 003 cps/Hz

Link in~l h i ~~t . o:;rr(lm :ChcUlneJ L (;·: .394 "lHz ) :CllC\lllld 2 (2 . 82/\ "1J-Iz ) :Channe l .:; (3 .16(: >!Hz) :Chc'tlme 1 1 (4 . Sbe; /-'!Hz) :Chcillr.<-.l :J (", . 031 MHz) :Channel 6 \6 .8 7:) t·1Hz) :Chunn..:>"). 8 (7 .GSll I>lHz) :Channel 9 ( '! . .~ ()5 MHz) ::~hannel lO (1 0 .3 .~ll /;!Hz l::;:hann,,·l 11 c:.U. ';23 /;!Hz) :c:tk um e J. L ' (1 3 . 692 /;!Hz) :

Channe- I 13 (lS . 4H7 MHz i :

1 1 il lk71) link.s29 1InksR I i nks56 Links4 l inks1 link7 links3 linksSlinks4') J. i nks6 l inh~

EO and sdt) stand for the expectation (average) and standard deviation of a measurement.About two-third s of a set of measurements will be within one standard deviation of theirmean and over 90% will be within two. The sd (mean_ t pu t _ ALE ) here suggests that oursample sizes are big enough to give us high confidence that if we collected morethroughput measurements under roughly the same conditions, we would not get averagethroughputs that differed from the one above by more than a tenth of a character-persecond. One should keep in mind that our "conditions" correspond to winter and springoperations at low sunspot numbers. Average throughput in other seasons and at muchhigher sunspot numbers will probably be different.

To calculate the average throughputs per Hertz [E (thruput_ALE /Hz)], we divided theaverage throughput by the ALE signaling bandwidth. For the latter we used the formula

Theory/Design 2-45

Page 123: Packet- Speed- More Speed

for "necessary telegraphy bandwidth" (from the 1992 Dept. of Commerce RFManagement Handbook):

where R (= 375 bits/s) is the channel rate, NT (= 8) is the number of MFSK tones in theALE waveform, f= (= 2500 Hz) is the highest tone and fmin (= 750 Hz) is the lowesttone. FOI: these values, the signaling bandwidth is 1875 Hz. At the end of its output, thestatistics program prints histogram values of the frequencies chosen for linking by thesending ALE radio (more on these below).

Because of the relatively low channel rate used by the ALE waveform (375 bits/s or 125symbols/s), and the high overhead used to provide the waveform's robust forward errorcorrection (about five-sixths of the channel rate), average DTM throughput (about 5char/s) is modest. What distinguishes ALE DTM transfers from ASCII transfers usingseveral other ARQ protocols, like the AMTOR, PacTOR, GTOR and AX.25 packetprotocols, is the fact that on tactical links, DTM transfers are much more frequentlysuccessful, especially at night. The main reason for this is ALE's ability to look foranother frequency for linking if the previously tried frequency fails. File transfers weresuccessful on roughly 90% of the automated attempts.

Our ALE systems have probably often linked on surfacewave and possibly E-layerfrequencies rather than on NVIS frequencies, which lie near the bottom of the HF band.Surfacewave frequencies appear to have been chosen often at night. (Most traditionalpropagation prediction programs do not suggest surfacewave frequencies for night timeor any other operation.) One reason for avoiding NVIS communications at night is thatthere is almost always more interference on NVIS frequencies at night than during theday. For short-range communications, it is important to use antennas that can launchboth NVIS and surfacewave signals; that is, antennas with both vertical and horizontalcomponents.

The relatively small standard deviation of DTM throughput reflects the DTM protocol'srestricted ability to adapt to changing conditions. The low variability of throughput overshort paths and the reliability of transfer are also reflected in the fairly small differencebetween average and maximum (6.6 cps) observed throughput.

Average link time was about 26 seconds, with a standard deviation of about 21 s. Thisstandard deviation implies that establishing a link required at least two attempts fair lyoften during our tests. (A single successful link handshake takes about 20 seconds.)

The histogram data produced by the statistics program are graphed below. The histogramshows that most DTMs were transferred at 2.824,3 .166,5.031 and 13.692 Mhz. Themiddle two frequencies were chosen mostly during the day and probably supported onlyNVIS. The 2.824 MHz frequency appears to support both surfacewave and NVIS.(2.394 MHz may have had antenna-matching problems that caused its poor performance.)The transfers on 13.692 MHz were mostly at night and were probably by surfacewave. Itis unlikely that an inexperienced communicator would have tried 13 MHz for night-timeoperation over this link.

2-46 Chapter 2

Page 124: Packet- Speed- More Speed

70

60

(/) 50~c

2 400~

Q) 30.0E::J 20z

10

0-.::t -.::t (0 L.O ..- I"- ('l') L.O L.O ('l') co N r-,m N (0 (0 ('l') co L.O co 0 ('l') N m co('l') co L.O 0 (0 L.O r-...: ('l') 0 L.O (0 -.::tN N ('l') -.-i L.O r-, m ..- 0 ('l') L.O

..- ..- ..-

Frequency (MHz)

These on-air results suggest that although the ALE DTM engineering orderwire mode isrelatively slow, on tactical links it can perform ASCII file transfers more reliably thanseveral other ARQ systems in current use.

Acknowledgments.

I am grateful to John Morgan of the Harris Corporation for providing the Harris RF5022radio systems with which these measurements were made, and to Bob Levre ault(W 1IMM) for maintaining one of the ALE stations used for the tests .

Theory/Design 2-47

Page 125: Packet- Speed- More Speed

CLOVER - The Technology Grows and Matures

Lessons Learned and Pleasant Surprises

Bill Henry, K9GWT

The "CLOVER" waveform idea was first presented to radio amateurs by Ray Petit in theJuly, 1990 issue of QEX. His narrow-bandwidth mode that required custom radiotransmitter and receiver design soon evolved into a more universal waveform for use withany HF SSB transceiver - called "CLOVER-II".

Development of CLOVER-II was not as simple or as fast as we had expected. It seemedlike we were frequently "re -inventing the wheel" and development schedules were re­written every week. After a few false starts and a couple "wrong turns", CLOVER-IIwas finally shipped and put on-the-air in late 1992. Since then, thousands of CLOVERmodems have found their way into ham shacks around the world. CLOVER-II is nowavailable in several different versions of the PCI-4000, the P38, and the DSP-4100modems as well as in customized systems for commercial and government customers.CLOVER has now been used in every conceivable HF radio application. This includesham radio, ship traffic, aircraft communications, bank data transfer, computer file transfer,image transmission, and even digital voice. Virtually anything that can be digitized andstored in a PC has been sent somewhere via HF radio and CLOVER, often at datathroughput rates that match or exceed those we see on 1200 baud VHF packet radio .

Early CLOVER Lessons:

CLOVER started with a great burst of enthusiasm. We had a bunch of new ideas and a"hundred or so" ways to implement each one. The new DSP architecture promisedunheard of design freedom. This was our first product where virtually every detail was setby software. We soon became quite familiar with the phrases: "It's only software." ­usually followed by - "It will take how long to make that change?" We demonstrated the"gas laws" often, especially the one that states "gas expands to fill the available volume" .Qur version is "Software expands to fill available memory space and consume allprocessor time".

CLOVER itself experienced many evolutional changes . The first version had a bandwidthof 100 Hz, one tone and used only one phase shift modulation mode . Duringdevelopment, the CLOVER waveform soon expanded to four tones, a bandwidth of 500Hz, and a total of 160 different modulation modes - a real "knob-twister's paradise"!Coming from the HF packet world, we also intended to fix some of the problems we'dseen with AX.25 on 20 Meters. The CLOVER ARQ protocol therefore includes selectiveblock repeat, in-block error correction coding, bi-directional data flow (no "OVER"command), and adaptive modulation control. While some or all of these ideas had been

2-48 Chapter 2

Page 126: Packet- Speed- More Speed

used previously, CLOVER was the first to combine the multi-level, multi-modemodulation waveform with an ARQ protocol specifically designed for use over HF radiolinks. No doubt about it, CLOVER is a complicated mode .

Predicting software completion turned out to be very risky business. The first targetrelease date was "April, 1991". This quickly slipped to "fall, 1991" to "April, 1992" andfinally to "November, 1992 (actually the last day ofNovember, 1992). AND - we weren'tdone! During the first 6 months of CLOVER's commercial life, we issued 8 no-costsoftware up-grades. We've since provided almost 50 software changes to CLOVER andDSP software, the most recent occurring this summer - July, 1996.

While the hardware schematic and block diagram have been very stable, parts cost andavailability have been continuing problems. Price and delivery promises made in 1990 bysuppliers have proven to be "optimistic dreams". This is particularly true of the MotorolaDSP components. The cost of these parts remains high and periodically they become"non-deliverable". In early 1995, frustrated with this high cost and unpredictable delivery,we redesigned the modem to use the much less expensive and more available TMS320C25processor and related parts. The P38 modem is the result, the first and so far only DSPmodem with a list price under $400 (it usually sells in the "mid-$300 range) .

Finally, CLOVER has experienced the typical problems of a pioneering mode . Correctlytuning the radio receiver to exactly match a CLOVER signal takes practice. Virtually allradios made since 1990 can meet the ±10 Hz tuning increment requirement, but you needa "soft touch" on the knob and patience to make small corrections and then wait. Oncelearned, tuning a CLOVER signal is easy - BUT - it's not like any other mode you've everused! Likewise, tuning the transmitter for no ALC and less than "Max-Smoke" outputtakes personal discipline, particularly hard for us old-time RTTY types. These are notnew problems in HF radio. The early SSB pioneers faced the very same problems in the1950's. Tuning-in an SSB signal was a lot harder than tuning an AM signal. 40 yearslater, we think nothing of tuning an SSB signal- "No big deal"! Likewise, we all soonlearned that adjusting a linear amplifier was a lot different than tuning the AM finals for"cherry-red plates". Inaccurate tuning and incorrect transmitter adjustment severely limitCLOVER performance and have frequently been the underlying problem behind "it .doesn't work" complaints. After 5 years, CLOVER is finally tuning the comer where weare getting comfortable with using it and can also say "no big deal" about these problems.

Marine CLOVER:

Ships at sea have used the "SITOR" mode for HF data communications for 30 years .Special frequency channels are allocated for ship-to-shore Narrow Bandwidth DirectPrinting Telegraph (NBDP) use. "Paired frequencies" are used with ships transmitting onone set of channels and shore stations on another set. Within the "ship" or "shore" sub­bands, the channels are spaced exactly 500 Hz apart . Particularly in the case of shorestation allocations, every 500 Hz wide channel is in active use, usually by very powerfultransmitters (1 kW to 10 kW). It is therefore very important that adjacent channel

Theory/Design 2-49

Page 127: Packet- Speed- More Speed

interference be prevented. HAL has teamed with Globe Wireless of HalfMoon Bay,California to develop a special version ofCLOVER that is tailored to meet ship-to-shorerequirements. While the original CLOVER-II waveform has a frequency spectra that isexactly 500 Hz wide (@ -50 dB), this leaves no "guard band" between adjacent channelsfor tuning error. A marine version was therefore developed that has a -50 dB bandwidthof400 Hz, well within the FCC Part 80 limits for use on NBDP HF channels. We call thisversion "CLOVER-400". The spectra-ofCLOVER-400 and the FCC limits are shown inFigure 1.

Globe Wireless has created a "Global Radio Network" of public coast stationsdistributed around the world . All public coast stations are tied into the global network viawhatever common carrier service is most cost effective for that location (wire line,satellite, Internet, etc.). Regardless of its location at sea, a ship can establish HF radiocontact with one or more network coast stations. When not actively sending data, eachship receiver constantly scans the frequency list of Globe stations, listening to traffic orSITOR "free signals", and logging signal quality data for each station heard. At any giventime, the shipboard computer therefore "knows" which frequency and station is optimumfor communications . Each ship acts as a "passive sounder", obtaining "LQA" data (LinkQuality Assessment) for each usable coast station without the need to transmit.

All coast stations are equipped to use either SITOR or CLOVER-400 on NBDP HFchannels. To maintain compatibility with older vessels, communications start in SITORmode and then switch to CLOVER-400 when the shore station recognizes a CLOVER­equipped ship. Use of CLOVER not only increases the speed at which data is delivered, italso provides error-corrected 8-bit data transfer of any computer file, be it text, data,image, or executable software. In fact, new software for the CLOVER modem on eachship is passed via CLOVER on HF radio - the system upgrades itself. All this isaccomplished at a fraction of the cost the user would otherwise have to pay for satellitecommunications .

Voice Bandwidth CLOVER:

While amateur data modes emphasize narrow bandwidth (500 Hz or less) to conservelimited spectrum, commercial and military HF allocations are virtually all for "voicebandwidth" channels (ship-to-shore excepted). The U.S . Civil Air Patrol (CAP) has madefull use of the four "tone channels" of CLOVER-II to expand the capacity of their limitednumber ofHF channel allocations. The CAP technique is to share an HF voice channelbetween four non-interfering and independent CLOVER ARQ links. This applicationtakes advantage of CLOVER's exceptionally high stop-band suppression . Selection of theCLOVER-II "tone channel" is a user-set feature included in all PCI-4000 and DSP-41 00modems (not available in the P38 modem). The CAP CLOVER "multiplex" ofa voicechannel is shown in Figure 2.

2-50 Chapter 2

Page 128: Packet- Speed- More Speed

CLOVER-400Frequency Spectra

II!

100)Hz--

~J,f\ '\A!::::T v ~CLOVER-400

-t------+----~__+_-__++-_+_-4_+_-_+----!-.-._----

O.-------.-------,.-----:---.,..........,:--~----,--------,-------,

-1 0 -t-----+-----+---+-If--_+_~+-+--__+----__!_----___j

FCC Pa 1 80.211 (f) \

(Q Limits <. I~ -20 <, ! I~ ~_!.... 50DHZ·~__·__·_ll -25dB

.~ 1 \1

a. -30-t------+-----t---+--+---+-----i!-----i~-------lE -~ -

3200I

27002200

I iI

III I I I I I II I

25 )OHz -+---+------+---II~

I I

1700Frequency (Hz)

700

r----·-.--.-----.---.--

1--

Figure 1.

CIVIL AIR PATROLCLOVER-II Multiplex

!\tt0!\IV\!\ /\/\f ~!\IV\!\'IY 1 i¥ IYY 1i

ARQ Link # ARQ Link #3

--

(Q -20+----+-----:t-----t-----il-----!---- --1"'0'--"

a.>-g -30-t-- --it-----+tt------+t+------+tt-- - ----H------1~

a.E~ -40-+- - - -t+------H+----+t+-----++t-----H-----1

30002500I I I I I I I I I

1000 1500 2000Frequency (Hz)

-60 I I I I I I

o 500

- 5 0 -t-- - ----I-+-- - - H+- - - +++- - - +--H- - - --+-f--- - --1

Figure 2.

Theory/Design 2-51

Page 129: Packet- Speed- More Speed

CLOVER technology has also been expanded to gain higher data throughput by usingmore tones and a higher symbol rate. "CLOVER-2000" uses 8 tones, a symbol rate of62.5 Baud, occupies a bandwidth of2000 Hz, and has a data throughput rate that isapproximately 4 times that of CLOVER-II - up to 250 bytes/sec. (2000 bits/sec). Thespectra ofCLOVER-2000 is shown in Figure 3, typical throughput characteristics inFigure 4, and ARQ timing information in Figure 5.

On-the-air testing of CLOVER-2000 has shown that it is very robust and often providesbetter communications than simple scaling ofCLOVER-II performance might predict.Over an 800 mile path on 10 MHz at mid-day, CLOVER-2000 has reliably provided datatransfer at an average rate of 1000 bits-per-second, and frequently running at 1500 to2000 bits/sec for 5 to 10 minute periods. The wider bandwidth and faster symbol rate ofCLOVER-2000 (compared to CLOVER-II or CLOVER-400) do not cause excessiveblock failure or repeats. In fact, the short ARQ frame of5.5 seconds makes CLOVER­2000 very responsive to changes in the ionosphere.

CLOVER Modems:

There are now three modem products that support one or more versions of CLOVER ­the PCI-4000, the P38, and the PSP-4100. There have been 4 major versions of the PCI­4000, two of which are still in current production. The "standard" PCI-4000, firstintroduced in 1993, will support CLOVER-II and FSK modes (RTTY, AMTOR, Pactor)but cannot be used with CLOVER-400 or CLOVER-2000. CLOVER-400 capablemodems (called "GL-4000") are only available from Globe Wireless . Use of CLOVER­2000 (or CLOVER-400) requires a faster 68000 processor and additional memory . Thishardware version is called the "PCI-4000 Plus". All versions of the PCI-4000 use on­board DSP and 68000 software that is upload via the PC data bus. Software up-dates canbe quickly and easily obtained from the TECHLINE BBS.

A low cost DSP modem was introduced in 1995 to provide CLOVER technology to radioamateurs at minimum cost. As noted earlier, it was necessary to change the DSP systemfrom the Motorola DSP56001 to the TI TMS320C25 (plus related components) . The TI'G25 is a "previous generation" device and does not have the speed or convenientinstruction set of the 56001. Fitting CLOVER-II into the 'C25 was a real "squeeze". Asmight be expected, this required a few trade-offs in performance and reduction of some ofthe deluxe features of the "high horsepower" PCI-4000. In particular, the P38 does nothave processing power to receive the highest two amplitude modulation modes - 8P2Aand 16P4A. The P38 will, however, communicate via CLOVER-II with any other P38,PCI-4000, or DSP-4100 equipped station. Like the PCI-4000, DSP and 68000 softwarefor the P38 is loaded via the PC bus and up-dates are readily available from TECHLINE.

2-52 Chapter 2

Page 130: Packet- Speed- More Speed

CLOVER-2000Frequency Spectra

-1O-+-- - --++---I-+--+++--+-+--++f---\-f-- ..H-+-~

-20-+-----i-----t-----t---- -+-----'i--- ---i

co~ -30-+----~---~---~---~ -~~

0.. -40-+----4-1-----~---~---~----+t-----j

E~

-50 -+-----I--I-----~---~----J-----+-+----j

-60-+----"-------I------+-----f-----t------t--j-----j

300025001000 1500 2000Frequency (Hz)

500-70 -+-_r__T-...--r--I---r--r--,-----r-~"'T'"'""""""1r__T'"----r-~__r__T'""""""'T~_+___r_r__T~_+_......___r__T-,--l

o

Figure 3.

Theory/Design 2-53

Page 131: Packet- Speed- More Speed

CLOVER-2000 THROUGHPUTBias Comparison

4035305 10 15 20 25SIN (dB, 3000 Hz Noise BW)

o-5

I , I, FAST BIAS

TI i

II

IIr ... "'-" '-" '-'~" '-"'-"'-" '-" T" -"'- " '- "'-' "

II INOR iL BIASI/ ( f I ROBUI3T BIAS I

~/-··-···-···-·· ·l/-···jI iI I

- I I !.. I V ! I I- I J ..._..._.....-/ I !

! f/ / I I- .I I

I- I I- Ii~7 I I.. I"~

! I

if I-.1'" .1.. !

- If! I- I

I I I I I I I I I I I I I I I I I I I I I I I I I I TTTo-10

--. 250oQ)(/)-(/)Q)+-' 200>-..c..........+-'::Ja...c 150rn::J0lo....

..c

.....·0 100Q)+-'oQ)lo....lo....

0 50oI

lo....

0lo....lo....

W

Figure 4.

2-54 Chapter 2

Page 132: Packet- Speed- More Speed

I-­(/)<[

u.

CLO VER-2000 MULTI-BLOC K AR Q DATA FRAME

R Ref. Se que nc e 0.01 6 sec .DEL AY DELAY CCB = BPSM/17 /60 0.272 sec.

G No-Signat Gop 0.01 6 sec.

DBn Da t a Block vor. sec.

ROBU ST BIA S (6 0%) BYTES/ MAX BLOCK BL KS/ ARQ FRAME THRU-PUTRATE MOD BLOC K FRAME ERRORS TIME FRAME TIME BYT ES/S EC

165 16P4A 255 900 300 0.68 8 sec 6 5.440 sec 165.4

110 8P2A 255 600 200 1.024 sec 4 5.440 sec 1l0.3

83 8PSM 255 450 150 1.360 s ec 3 5.440 sec 82.7

55 QP SM 255 300 100 2.048 sec 2 5.440 sec 55 .1

28 BPSM 255 150 50 4.080 sec 1 5.440 sec 27.6

NORMA L BIAS <75%) BYTES; MA X BLOCK BLKS; ARQ FRAME THRU-FUTRA TE MOD BLOC K FRAME ERR ORS TIME FRAME TIME BYTES/SEC

207 16P4A 255 11 28 186 0.688 sec 6 5.440 sec 207.4

138 8P2A 255 752 124 1.024 s ec 4 5.440 s ec 138.2

104 8PSM 255 564 93 1.360 sec 3 5.440 sec 103.7

69 QPSM 255 376 62 2.048 sec 2 5.440 s ec 69.1

35 BPSM 255 188 31 4.080 sec 1 5.440 s e c 34 .6

PAST BIAS (9 0%) BYT ES/ MAX BLOCK BLKS/ ARQ FRAME THRU-PUTRAT E MOD BLOCK FRAME ERRORS TIME FRAME TIME BYT ES/SEC

24 9 16P4A 255 1356 72 0.688 sec 6 5.44 0 s ec 249.3

166 8P2A 255 904 48 1.024 sec 4 5.440 s e c 166.2

125 8PSM 255 678 36 1.360 s e c 3 5.44 0 sec 124 .6

83 QPS M 255 452 24 2.048 sec 2 5.440 s ec 83 .1

42 BPSM 255 226 12 4.080 sec 1 5.440 s e c 41.5

Figure 5.

Theory/Design 2-55

Page 133: Packet- Speed- More Speed

The DSP-4100 modem was introduced in late 1995. Unlike the PCI-4000 and P38, theDSP-4100 is not a plug-in card for a PC. Rather, the DSP-4100 operates from 12 VDC,handles data and commands via an RS-232 serial I/O port, and has status lamps on thefront panel. In short, the DSP-41 00 looks and acts much like a standard phone linemodem, but it is used with HF radio systems. The DSP-41 00 is proving to be the mostpopular configuration for commercial use of CLOVER, particularly in applications wherebattery power and/or lap-top computers must be used. The DSP-4100 uses non-volatileFlash ROM which also may be upgraded by loading new DSP or 68000 software via theserial port. Thus, software in all three DSP modems can be easily upgraded withoutopening a cabinet and replacing EPROM IC's.

In The Future:

CLOVER continues to grow and improve. Present software gives performance andfeatures that were not available when CLOVER started. The "it's only software"development problem is finally working in our favor. Within limits, virtually everyCLOVER modem sold can be upgraded in the field at little cost to the user. Customversions of the hardware, terminal software, and CLOVER itself have been and continueto be developed to meet the needs of its users . CLOVER-2000 and the DSP-41 00 havegreatly extended the horizon for use of CLOVER technology. CLOVER modemhardware and software is already being installed inside some models ofHF transceivers. Inthese systems, there are no outward signs that "CLOVER is being used unless you tum upthe receiver volume control and hear the distinctive 4- or 8-tone "CLOVER twitter" .

The future direction of CLOVER technology - waveform, protocol, software, andhardware - will be determined by the user. CLOVER provides the "missing link" thatmakes HF radio an attractive cost-effective alternative to satellite or wire-line datacommunications.

2-56 Chapter 2

Page 134: Packet- Speed- More Speed

CLOVER BmLIOGRAPHY

Townsend, Jay : "The HAL Conununications P38 HF Radio DSP Modem " (New Product Review), CQ, October, 1995 , p. 30, CQPublishing. Hicksville, NY .

Vinson, Glen : "HAL P38 vs AEA PK-232", Digital Journal, September, 1995 , pp 12 & 13, International Digital Radio Assoc iation("IRDA"; fonncrly American Digital Radio Society. "ADRS"), Box 2550, Goldenrod, FL. 32733-2550.

Ford, Steve: "Product Review - HAL Conununications P38 HF Modem", QST, August, 1995 , pp 71-73, American Radio Relay League(ARRL), Newington, CT .

Blegen, Hal: "The HAL P38 DSP", Digital Journal (RTTY Journal), August, 1995, pp 8 & 22, International Digital Radio Association("IRDA"; fonnerly American Digital Radio Society- "ADRS"), Box 2550, Goldenrod, FL. 32733-2550.

Schulz, Fred : "P38, die neue CLOVER-Platine" SWISS ARTG Bulletin (6/95), pp 14-17, SWISS Amateur Radio Teleprinter Group, c/oArturo Dielet, HB9MIR (Secretary), Blauenwig 8, 4335 Laufenburg. Switzerland (in German).

Walder, Stephan: "Neu bei del" SWISS·ARTG: P38 von HAL" , SWISS ARTG Bulletin (4/95), pp s & 13, SWISS Amateur RadioTeleprinter Group, clo Arturo Dielet, HB9MIR (Secretary), Blauenwig 8, 4335 Laufenburg. Switzerland (in German).

Hollingworth, Jack ; "The New HF Data Mode : CLOVER II", RADIO COMMUNICATIONS, November, 1993 (pp 52-54) , December,1993 (pp 62-64), January, 1994 (pp 68-70), RSGB, Lambda House, Cranborne Road, Potters Bar , Herts . EN63JE England

Schilling. H.J., and Paul Williams: "Clover ein neus digitales Funk-Verfahren fur Kurwelle ", CQ DL, Jun e, 1993, pp 38 5-3S7, DeutschenArnateur-Radio-Clubs (DARC), Postfach I I 55,3507 Bau nata l, Germany (in German),

Healy , Jam es W.: "HAL Conununications PCI-4000 CLO VER·I1 Data Controller" (Product Review), QST, May, 1993 , pp 71·73,American Radio Relay League (ARRL), Newington, CT .

Townsend, Jay : "CLOVER - PCI-4000", RTTY Journal, April , 1993 (pp 3-4) May/June, 1993 (p 20),1904 Carolton Lane , Fallbrook,CA 92028-4614.

Walder, Stephan: "CLOVER · Eine neue Betriebsart", SWISS ARTG Bulletin (1/93), pp 5-12, SWISS Amateur Rad io Tel eprinterGroup, clo Arturo Diet let, HB9MIR (Secretary), Blauenwig 8, 4335 Laufenburg. Switzerland (in Gennan).

Schulz, Fred : "CLOVER ist eingetroffen", SWISS ARTG Bulletin (1/93), pp 13·15, SWISS Amateur Radio Teleprinter Group, c/o ArturoDietlet, HB9MIR (Secretary), Blauenwig 8, 4335 Laufenburg. Switzerland (in Gennan).

HAL Conununications, PCI-4000 CLOVER-II Data Modem Reference Manual and PC-CLOVER Operator's Manual, November, 1992 ,HAL Conununications Corp ., Urbana, IL (S25 .00) .

Henry, George W., Ray C. Petit : "CLOVER · Transmision de datos rapida en HF", CQ (Spanish Language edition), November., 1992 , pp45-48, CQ Publishing. Hicksville, NY (in Spanish).

Petit, Ray C.: "The CLO VER-II Conununication Protocol- Te chnical Overvi ew", ARRL 11th Computer Network ing Conferenc eProceedings (October, 1992), American Radio Relay League (ARRL), Newington, CT .

Henry, George W., Ray C. Petit : "HF Radio Data Conununications: CW to CLOVER", Conununications Quarterly, Sprin g. 1992, pp 11­24; CQ Publishing. Hicksville, NY.

Henry, George W., Ray C. Petit: "CLOVER: Fast Data on HF Radio" , CQ, May, 1992 , pp. 40-44; CQ Publi shing. Hicksville, NY.

HorzepaStan (ed); George W. Henry & Ray C. Petit: "CLOVER Development Continues", "Gateway", QE X, March, 1992 , pp. 12-14 ,American Radio Relay League (ARRL), Newington, CT .

Henry, George W., Ray C. Petit : "CLOVER Status Report" , RTTY Journal, January, 1992, pp. 8.9, Fountain Valley, CA

Henry, George W., Ray C. Petit : "Digital Conununications for HF Radio- AMTOR & CLOVER"; paper presented at Amateur RadioDigital Conununication Seminar, Sl Louis, Mo., October 26, 1991.

Petit, Ray C.: "CLOVER-II: A Techn ical Overv iew", ARRL 10th Computer Networking Conference Proceedings (19 91), pp. 125-129;American Radio Relay League (ARRL), Newi ngton, CT .

Petit, Ray C.: "CLOVER is Here" , RTTY Journal, Fountain Valley , CA; January, 1991, pp. 16-18; February, 1991 , pp. 12-13 ; March,1991 , JIll. 16-17; April , 1991, p 10.

Pet it, Ray C.: "The CLOVERLEAF Performance-Oriented HF Data Conununication System" , QEX, July, 1990, pp. 9-12; reprin ted inARRL/CRRL 9th Computer Networking Conference Proceedin gs (1990), pp. 191-194 ; American Radi o Rela y Leagu e (AR RL),Newington, CT .

Theory/Design 2-57

Page 135: Packet- Speed- More Speed

Chapter 3Networking

Page 136: Packet- Speed- More Speed

Packet PerspectiveConducted By Sta n Horzepa , WA1LO U

One Glen Ave, Wolcott , CT 06716-1442

Internet: stan .horzepa @corp .gdc.com CompuServe: 70645,247

RMNC/FlexNet:The Network of Choice in Western EuropeEric Bertrem, F5PJE , has read articles inPacke t Perspective dealing with NET/ROM,TCP/ IP, TexNet , ROSE and other networkingsystems used in the US, but has never readanyth ing here about RMNC/Fle xNet, which isused throughout Europe. This month Ericdescribes how RMNC/FlexNet works and howto use it.

The HardwareThe Rhe in Mai n Networ k Controller

(RMNC) is the hardware portion of the sys ­tem and Flex Net is the software portion; 95 %of the software is wri tte n in C and the remain­der, the low-l evel I/O, is writte n in 6809 as­sem bly language. Develop ment of the sys tembegan in Ge rmany in 1987 and came to fru i­tion when the fir st RM NC/FlexNet digipe aterca me on the air in 1988.

Ty pically, the contro ller uses the fo llow ­ing configura tion of equi pme nt:

Reset II card that provides I/O for rem otecontrols, clo ck sig nals for the other ca rds andresets fo r card fa ilures.

Solomas tercard for contro l ofand upload­ing software to the Slave cards.

Slave cards, one for eac h radio link, thatrun up to 115,200 baud s usin g 4, 8, 12 orl 6- MHz CPU clo cks. The lat est vers ion,RMNC 3, includes the fo llow ing option s: aKISS/RS -232 interface to link the digi pea terto any PBBS capable of usin g KISS , a 1200­baud modem with digi tal echo of the receivedframes (s imilar to the bit regenerator opt ionof TA PR 9600 modems) and 9600-baud FSKDF9 IC (G3R UH-compatible) modem withdigit al echo, if needed .

ANETs are fully compatible French ver­sio ns of the RMNC cards tha t were devel­oped by F5CAU and F6BNY. Each ANETscar d generates its own clock signal s, there­fore , the Reset II is unnecessary unless youneed its I/O por t.

All the card s plu g into the bus card ( 16car ds maximu m at eac h dig ipea ter) and ca nbe connected easil y to external modems op­erating fro m 4800 to 38,400 bauds (ma ny mo­dems are ava ilable in Ge rmany for abo ut $90eac h) . Usi ng such high-speed mod ems meansusing dedicated transceivers. The firs t ded i­cate d links in Germany began on 70 em, butthe ba nd qui ckly became ove rcrowded, soDF9 IC design ed a 23- cm hal f- and full­duplex transcei ver , Interlink I and InterlinkII , res pective ly . Th ey are FSK-rea dy withI to 20 watt s output.

Today, most German links are on 1.2 GH zat 9600 or 19,200 bauds. The network is effi­cie nt and provid es fa st respon se to down­load s, uploads and even ke yboard- to -key ­board rea l-time chats. I have a QSO everyday wit h a friend 700 km fro m hom e through

se ven digipeaters and often do wnl oad newversions of German softw are from a PBB S1000 km from here with no link failures .

Th e Softw ar eFirst used in Ge rmany, FlexNet qui ckly

spread into Switzerland and then to France,where we previ ously had unr eliable networkswith slow (l 200 -baud ) links. FlexNet is alsoin operat ion in Austria, Belgiu m, Hun garyand Holl and . Ma nua ls are avai lable inFrench , Ge rman and, soon, English. Her e ' show it works (the foll owin g examples havebeen translated into Eng lish).

Af ter I connect to my local RMNC/FlexNet digip eater, I rece ive the follow ingmessage:(1) CONNECTED to F6BIG-7 RMNC/FlexNetV3,3a*Mt Semnoz (1700m) JN35BT 17 km SWAnnecy (74)(H)elp =commands - (L)inks - (A) =news ­(Q)u it* 'c f6big-S' => bbs Annecy ** 'c hb9iac-S' => dx-cluster Geneva *Please, only ONE connect at a time, thanksto all the users!Users access on 70 cm = 433.775 MHzThis channel will change to 9600-baudG3RUH in the near future. Get the equip­ment!

Next, I invoke the P (for Parameters) co m­mand to get the per tinent param eters of thedigip eat er. In respon se, the digipeater sendsme the follow ing information for eac h of itsports :po id td qso usrtifr rifr txkby rxkby qty modelinks

"po" is the port numb er (one Slave cardper port )

" id" the SS ID used (no SSID mea ns theport is linked to another digipeater)

"qso" and "usr" represe nt the num ber ofQSOs and users on the port .

" ti fr" and "rifr" are the number of Ifra mes sent and rec ei ved within the last 10min ute s.

"txk by" and "rxkby" are the numb er ofkilobytes sent and recei ved during the last 10minutes.

" qty" refers to the quality of the lin k."mode" indicate s the speed of eac h port and

the CPU clock used by the card on that port(+ for 8 MHz, ! for 12 MHz, # for 16 MHz) .

Other paramet ers inc lude "d" for full­duplex, " t" for exte rnal transmit clock, "r"for external receive clock and "z" for the NRZmode . The time to reach a neighborin gdigip eater (in 100 ms steps) is prov ided .

On FlexNe t, user ports have SSIDs (forexa mp le, SSIDs 2, 7 and 6 on F6B IG-2) .The se help remo te stations to choose the cor-

rect "output" on each node, in conjunctio nwith the MH function which records the last200 sta tio ns heard on each port. If my in­tended connec tio n is recorded in the dest ina­tio ns list, tha t record is used to make the con­nection.

Nex t, I inv oke the D (for Destinations)command:

D DB0ZDF

This asks the sys te m if DB 0ZDF is aknow n dest inati on , that is, is DB 0ZD F re­corded in the destin ations list? If it is, I' llreceive the ava ilable use r SSIDs and tim eneeded to reach it (the T param et er). Th esys tem responds with

*** DB0ZDF (0-12) T = 319

which indicates that DB0ZDF is a kno wndestinat ion , so I don' t have to specify the pathto connect with DB0ZD F; all it takes is CDB0Z DF and the sys tem responds withlink setup...*** route : HB9X HB9EAS F6KDL DB00RTDB0CPU DB0AAI DB0ZDF*** connected to DB0ZDFRMNC/FlexNet V3.3c Duplex-DigipeaterDB0ZDF Mainz JN49cx.QRG R 64 SieheA = News C = Conference modeI = Information H = Help

The system's response to my P (Param­eters) co mma nd indi cates that this digipeateris one of the best in Germ any. Run by manyham s on the RM NC/F lexNet team , speed shere are high (up to 19,200 baud s), usin gInterlink 23-c m tran sceivers and- mos t ofthe time-ante nna dish es to impro ve linkquali ty witho ut needin g much power.

Time to DisconnectTh is was only a quick tour of FlexNet.

There are many other user and SysOp com ­mands available, and the system is simple andits links are quite reli able .

Rece ntly, the RMNClFlexNet team devel­oped a new concept called "PClFlexNet." Thesoftware is the same, but the hardware is simplyany 386-based computer with a minimum of540 kilobytes of RAM and a IO-megabyte harddisk. A BayCom USCC plug-in card provide stwo, four or eig ht radio ports. PClFl exNet alsosuppor ts Ethern et cards , BayCom modems,KISS links and soon all the other SCC cards. Inthe near future, it will have an IP router and a"ping-pong" confere nce mode (a network thatallows you to chat with friends far away fromyour location by simply connecting to yourlocal digipeater).

If you need more information or you haveany question s, please contact me via packe t aiF5PJE @ F6BIG.FRHA.FRA.EU or via theIntern et at f5pje @stdin.gate link.fr.ne t

Networking 3·1

Page 137: Packet- Speed- More Speed

AVAILABILITY OF SEVENTY 9600 BAUDPACKET CHANNELS ON TWO METERS

By Bob Bruninga WB4APR@ WB3V.MD

Unbelievable? Not really! With the advent of the latest 9600 baud packet radio modems, thereis an un-exploited mechanism for opening up dozens of half duplex data channels without ANY impacton existing voice and data bandplans . Read this proposal thoroughly before jumping to any conclusions.First you must consider two apparently un-related facts:

TINY PACKETS: With the new 9600 baud packet modems, not only is there potential for higherspeeds, but the packets are also 8 times shorter than conventional packets. This means that each packetoccupies the channel for less than 1/10 of a second at a time. If another signal appears on the datachannel, then the presence of the new signal can be detected in 0.1 second .

AVAILABLE CHANNELS: Throughout the two meter band, in every corner of the country, there arealmost 70 FM channels assigned to single VOICE receivers that, in general, only use their singlefrequency about 1% to 40% of the time. If a mechanism could be designed to permit these singlereceivers to continue to use their frequencies on a primary basis, at any time, with priority access andcontrol, then the rest of the time (60% to 99%) they could use their channel for moving digitial data .This mechanism could more than double the bandwidth available in the two meter band! (I used 2meters only to get your attention. One hundred twenty channels on the 440 band is the actual targetbecause of the availability of2 Watt 9600 baud UHF data radios for under $150 each .)

VOICEIDATA CHANNEL SHARING: In the past, attempts to share voice and packet have all failed,not because it is a bad idea, but because it has not been done properly. Under a very unique set ofconditions, however, voice and data can easily share a narrowband FM channel, IF:

1) VOICE has priority at ALL TIMES and all voice operations are completely transparent tothe voice user

2) DATA can never interrupt or attempt to use a busy channel3) VOICE users can pre-empt/interrupt data instantly at ANY time4) VOICE users do NOT hear packets or in any way are encumbered by shared use.5) NO MODIFICATIONS TO ANY voice radio is required

In other words, voice users do not even know that the channel is shared with data. To make data use ofa channel work on a secondary basis (behind the scenes) with NO impact on voice usage, there areadditional requirements:

1) There must be ONLY one voice RECEIVER on the channel, and it MUST be able to hearEVERY voice user on the channel.

2) This single voice receiver and its conventional COR or un-modified squelch circuit must havetotal control over channel use.

3-2 Chapter 3

Page 138: Packet- Speed- More Speed

THE BEST KEPT SECRET: Think for a moment about the input frequency of a typical voice repeater.There is only one receiver listening, and it can hear EVERYONE that desires to use the repeater. If therepeater RECEIVER does not hear anyone using the channel, then IT alone can decide to use thechannel itself for data! Now if the data is transmitted in 0.1 second bursts, FROM THIS SITE (on theinput channel), with a pause in-between to listen for voice users, then no one will be denied access tothe voice repeater for any longer than 0.1 second! Also, while the repeater is transmitting data on itsinput frequency, nothing is being transmitted to anoy users on the output! In this manner, we canTRANSMIT data FROM this repeater site at 9600 baud on the input frequency WHENEVER THEREPEATER IS NOT OTHERWISE BEING USED FOR VOICE .

MAKING A DATA LINK: Now combine one such voice repeater with another operating in the samemanner, and you have a two-way data link between these two sites. Not only is this a FULL DUPLEXchannel, but it also operates with NO HIDDEN transmitters, and NO CONTENTION, because there isonly ONE transmitter on each such channel. From a data perspective, each repeater site is a data nodethat transmits on ONE assigned frequency and can LISTEN on as many additional channels as desired .This is an ideal multi-node backbone network for passing traffic point-to-point over long distances!

HARDWARE: To avoid even momentary delays during normal repeater conversations, the externalcarrier detect of the modem is not only driven by the COR of the repeater receiver, but also by theHANG-TIME of the repeater transmitter as an indication that the repeater is engaged in a conversation.This way, data will only be transferred after the repeater has been unused by voice users for a while.

FORWARDING: Obviously, this type of channel will NOT support any traffic where an impatienthuman is on one end. This channel, however, is ideal for off-hour BULK forwarding between multi­channellevel-4 nodes under computer control.

BAUDRATE: Although I suggested off the shelf 9600 baud radios, the use of 15 KHz repeaterchannels in most areas of the country might require slowing to 4800 baud to be certain that all energy iscontained within 12 KHz. Also, the $139 TEKK 2 watt UHF data radio was actually designedcommercially for 4800 baud, and performs excsllently at that rate . Also, to avoid co- channelinterference, operating at power levels above 2 watts is probably not advisable. Fortunately, 9600 bauddata sounds just like white noise, and usually will NOT open the squelch of another repeater anyway.Finally, to experiment with this concept, you can actually try conventional 1200 baud during off hours .You can limit packets to a single frame (about.8 second instead of the usual 3 to 5 seconds), by settingMAXFRAME to 1 instead of the default 7. By adding an additional hang-time timer, so that the DATAmode is only activated when the repeater is unused for over 3 minutes, a new voice in-the-night willprobably not even notice a maximum of 0.8 second delay.

SYSTEM LINKS: Remember, ONLY THE REPEATER ITSELF can transmit on its own INPUTfrequency. The way to build a network is then to have each such repeater node TRANSMIT on its owninput frequency and similarly LISTEN to the other nodes transmitting on THEIR input frequencies . Toget data from your basement BBS, simply have it LISTEN to the repeater node on the input frequencyand the BBS then TRANSMITS on its OWN unique frequency. The repeater node then listens on thisunique frequency . If you are willing to cut throughput in half in order to save on hardware costs, youcan let your BBS transmit (at low power) on the same frequency as some distant repeater node (its

Networking 3-3

Page 139: Packet- Speed- More Speed

voice input freq) . This causes a hidden transmitter problem at your repeater which will hear both yourBBS and the remote repeater, but with the tremendous bandwidth available, this is probably not aproblem. The following diagram shows a typical arrangement for the nodes located at a pair of146.34/94 and 147.81/21 voice repeaters.

I T146 .34I·

/1\1 R147 .81

T.81 \ --------1-1-1R.34 \ / \ 1 NODE 1

/ \ --------I

1 BBS 1

T147.81

I/1\

R146 .34 I1-1 -1--------- / T.34

1 NODE 1 / \ / R.81-------- / \

I

1 BBS 1

Notice that by pairing up a 146 MHz repeater with a 147 MHz repeater, you get at least 1.6 MHzspacing between the two digital frequencies and about 1 MHz between the digitial and the Voicerepeater output (which of course is NEVER transmitting when the digital is in use.

Also notice, that to save dual receivers at the NODE sites, we are cheating a PURE network design byallowing the BBS's to also transmit on the same frequecy as the DISTANT repeater node . This mustbe done with a beam or low power so that the distant repeater CANNOT hear nor even detect thepresence of the BBS. Otherwise, the BBS would KEY UP the distant VOICE repeater!

CONCLUSION: Even if only 30% of the voice repeaters begin to share their channels, this could openup over 600 Kbytes PER SECOND of additional digital forwarding capacity on 2m and 70 em! Whynot?

3-4 Chapter 3

Page 140: Packet- Speed- More Speed

Building a Packet Networkby Karl Medcalf WK5M

FIGURE 1

extending the range of communication. Inorder to use these digipeaters, however, theuser had to know the call sign and/or aliasof each digipeater from the beginning to theend of the route, and also had to know whatorder the digipeaters had to be used.

As we look at figure 1, we see how a usermight connect to a distant station using thesedigipeaters . This system, although it wasused successfully, had many limitations. Inorder for this to work, each station in the linkhad to be able to communicate directly withits neighboring station, and since the AX.25protocol places a limit of 8 relay stations, thislimits the potential range. In addition, eachstation in the link must be turned on andoperating - this could be hit and miss sincesome people tend to turn their systems offwhen they aren't operating.

Another defect with this system is that oncethe link is established through these relaystations, failure of anyone station wouldeffectively'cause the complete failure of thelink. This failure could be due to naturalcauses (lightning strike, power failure) orthrough human causes (the user turned offhis TNC or changed frequencies on his radio).

Although figure 1 shows that there may bemore than one path from point A to point B,the packet system has no means to select the"best" or most reliable path, nor is it capableof detecting failures and routing the dataaround the failure. With the release by Soft­ware 2000 of the NET/ROM firmware, a newera of packet radio began.

o

o 0

o 0A

o User stations

All stations on 145.01

PurposeThis paper will present various viewpoints onnetwork construction, and does not intend toimply that anyone concept is superior to anyother. It is intended to provide node operators(current and future) with ideas for considera­tion to help improve the existing system.

We'll first present the network as it existsin many areas, mostly at 1200 baud on bothuser ports and backbone ports, and then pre­sent ideas for expanding and modifying theexisting systems to provide higher through­put and reliability.

BackgroundSince the beginning of amateur packet radio,users have tried to push the limits. This hastaken many forms: how far can I get, howmuch data can I pass, how fast can I go.

In 1987, Software 2000, Inc. developedthe NET/ROMTM code, which replaced theEPROM in TAPR clone TNCs, in an attemptto improve the packet situation. This codeprovided the first attempt to build a net­work using amateur packet radio. Much ofthe current network system throughout theworld is based on this code, and new imple­mentations continue to arrive on the scene.

Network ConceptsPacket radio networks are intended to pro­vide users with the necessary means to passdata from point A to point B with little re­gard for the details involved. As such the net­work node operators must be aware of theimpact that any changes and additions tothe network may cause.

When amateur packet radio began, eachpacket station was virtually an independ~nt

entity. Users could connect to nearby sta~lOns

and pass data unencumbered, but long dis­tances could not be spanned easily. Everypacket station had the ability to be a digi­peater, a digital relay statio? that could re­transmit whatever packets It heard, thus

Networking 3-5

Page 141: Packet- Speed- More Speed

The Beginning of Packet NetworksBy replacing the EPROM firmware in therelay stations, these stations became intelli­gent for routing data from point A to point B.The new firmware converts these TNCs fromsimple digipeaters into Network "nodes".The nodes broadcast a simple packet on aperiodic basis, typically once each hour, andthus other nodes recognize the presence ofneighboring nodes automatically. Each nodemaintains a table of all of the other nodes itcan hear (its neighbor nodes), and broadcastsa list of these nodes. Each node would listento the neighboring node broadcasts, and thuslearn of distant nodes - nodes which cannotbe heard directly, but which the neighborcan hear. For instance, if Node A can talkto Node B, and Node B can talk to Node C,then Node A can talk to Node C (by relay­ing through Node B).

The initial network was built using existing1200 baud TNCs connected to 2-meter radios,and operating on the same frequency thatpacket operators had been using for some5 years - 145.01 MHz. This provided a rela­tively inexpensive means to improve thepacket system - no new investment in TNCs,radios, or other hardware was required ­only a new EPROM. As we look at figure 2,we see how the network may have looked inthe beginning. All stations were operatingon a single frequency, and all were operatingat 1200 baud.

Problems quickly arose with this system.Since all nodes and all users were on thesame frequency, the frequency quicklybecame overloaded with data. Remember,each node broadcasts its list of other known

oA

o User stations

o Nodes

All stations on 145.01

3-6 Chapter 3

nodes periodically and the users were stilltrying to use the same frequency. Addition­a~ly, ma~y nodes were installed at relativelyhigh altitudes enabling them to talk greaterdistances, but the additional height added to~he hidden station problem. A user attempt­mg to connect to a node may suffer collisionsfrom a neighboring node many miles away.

A Better WayTo overcome some of these problems, nodeoperators began building node "stacks" - twoor more nodes at the same physical locationconnected to each other through the serialport of the TNCs. These node stacks wouldhave one node dedicated to "user" access, andthe second node, on a different frequency, fornode-to-node or "backbone" communication.This scheme allowed the data to pass withlittle contention over the backbone, and alsoreduced the hidden station problem sincelocal users were no longer sharing the fre­quency with distant nodes. As more usersand more nodes began to appear, node opera­tors extended the node stacks with additionalTNCs. The additional nodes were on differentfrequencies or bands, and in some cases oper­ating at higher speeds.

High speed backbone systems are seen asone of the keys to success of a network sys­tem. Unfortunately, at that time, 9600 baudpacket radio was in its infancy, and therewere no commercially available radios capa­?le of ?perating these speeds. Node operators,m their quest for higher speeds, modifiedexisting radios to permit the higher speeds.Today, 9600 baud is rapidly being introducedinto networks - partly because the currentsystem is severely overloaded, and partlybecause the major radio manufacturers nowproduce radios capable of operating 9600baud without modification.

One Network Plan~igure 3 shows one possible plan for develop­mg a network. Each node in the system con­sists of at least 2 ports - one for the LocalArea Network (LAN) to provide user access,and a second for the backbone. The user ac­cess port is a 1200 baud port, while the back­bone port typically runs at higher speeds -

Page 142: Packet- Speed- More Speed

FIGURE 3

oo

o

o

o

o

o

oIIIII User f::: 145.07 0L _

o

oo

o

o

o

o

o

,- - - -- - - ---

I--- - I

o User stat ionsI

0 0 I

I 0 0 I

BNode stacksI 0 f::: 430.55 0

I

I 0 f::: 145.09I

I-=.J LANs

I 00 I

1 User f ::: 145.09I

----------- __ __ __ 1

FIGURE 4 io - 0--- - - -- - -- ----- - ,

1----- -- - - - 0-- -- - ---- 0 0 I

I 0 1 I0-"0

0 0,

I 0I I 0 / 8 I

I 0 I I o -: 0 a l l I

I 0 I I <, / 0 I

I 0 0 <, /

I f::: 430.55 I<, / I

/

I 0 f ::: 145.01f::: 145.07 1

I 0 0 I I f ::: 430.55 0 1

I I I 0 0I

I 0 0 0 I 1 0 0 0 I

I I I 0 1

IUser f ::: 145.01

I User f ::: 145.07 1

- --- -- - - - --- - - I ---- --- -- - - --___ __ _ _ _ __ J

- - -- - --- ----I

- --- - - - -

I 0 0--- ,

0 0 I

I 0 00 0 I

o User stationsI 0 0 I

I f::: 430.55 0 I

D NodesI f:::145.09 -------- - - -0 0 I

I 00 I

BNode stacksI 0 0 0 0 0

I

I 0 I

II LANsI User f ::: 145.09

0 0 0 0 IL ______ I--- - - -- --------- - - ---

Networking 3-7

Page 143: Packet- Speed- More Speed

FIGURE 5

requires that each node be equipped withdirectional antennas, and must be capable oftransmitting and receiving on at least twofrequencies for the backbone, plus whateveruser frequencies are required. Obviously thiscan quickly become a cost-limiting factor, butin heavily populated areas the cost may beshared by many users.

00 :1

1

1

1

I

1= 430 .551= 223.401=145.09

o

o

o

, -- - - - - -- -

I 0 0 0 ::0 1= 430.95 0 :: 0 1= 145.01 0 1

: 0 000 :---- -- --J

o

, -

:01

1

II1

,---- - --

00 0 :0 1

o :0 1

1

,---- - - - - - -: 0 0 0 0 :1 1= 430.55 0 :10 1= 145.01 0 1: 000 0 1I I

o User stations

§Node stacks

IJ LANs

101

1 1= 223.401 1= 430.951 1= 145.07: 0

Network LoadingUsers generally do not present a heavy loadto the network system. Since users typicallytype slowly, and send short messages to thestation they are communicating with, thehigh-speed backbone is very capable of han­dling several users without problems. How­ever, Bulletin Board Systems (BBSs), DXPacket Cluster Systems, Conference Bridges

Another possibilityThe previous plan can have some drawbacks .With the backbone nodes being capable ofhearing very distant nodes, there is the pos­sibility of collisions between the nodes whichcould severely impact throughput. To over­come this problem, some areas are usingstaggered frequencies and/or full-duplex be­tween the nodes. Figure 5 shows a possibleconfiguration using this scheme. This scheme

9600 baud or higher. Adjacent area LANsoperate on different frequencies so they donot interfere with each other. The two portsare connected together at the node to allowusers from one LAN to connect to users inanother LAN through the backbone.

In this example, the user access port shouldhave its antenna at a relatively low height,perhaps 20 or 30 feet, so that all users inthe LAN can reliably connect to the node,but also so that all users in the LAN areclose enough that they can hear each other.This reduces the possibility of collisionsbetween users accessing the system. Thebackbone port antenna is generally locatedat a fairly high spot, providing the ability toconnect to very distant nodes with few inter­mediate nodes required.

Frequently, users in one LAN may wish totalk to users within the same LAN, andtherefore do not require the services of thenetwork. It may well be advisable, therefore,to provide a "direct connect" frequency withinthe LAN for these purposes. Typically thisfrequency would not have nodes linked to thebackbone system.

In some areas, due to available resources,it may not be feasible to have the LANcover a small enough area that all users canhear each other. This would normally occurin a lightly populated area where the userbase cannot afford to install many dual-portnodes. In these cases, the backbone link maybe provided by a single node with two ports,and additional low-speed nodes installed onthe LAN frequency to increase the LAN cov­erage area. Figure 4 shows such a system,expanding figure 3 with additional low-speednodes.

3-8 Chapter 3

Page 144: Packet- Speed- More Speed

and the like present a very different situationto the network.

BBSs provide users with a tremendousamount of information - regardless of whatyou may think of the value of that informa­tion. This data arrives at the BBS fromanother BBS and so on, which means thatthe data has to travel through the networkat some point. Within a LAN, users accessthe BBS to retrieve information, but theremay be only one BBS in that LAN. Thereforethe BBS must use the backbone to send datato other BBSs. This presents a couple of ques­tions to consider: 1) should users access theBBS on the LAN frequency, or should a sepa­rate frequency be assigned for user access tothe BBS services, and 2) should the BBS for­ward traffic through the network via the lowspeed (user) access port of the node, or shouldthe BBS have direct access to the backbonefrequency.

There are no concrete answers to these ques­tions, they must be decided on a local basis.Things to consider include the channel load­ing caused by users accessing the BBS, thecost of the equipment required, available fre­quencies in the area, and perhaps other"political" factors.

Node OwnershipWhen networks first started, all that wasrequired was a simple TNC, some new firm­ware, one radio, and a two-meter antenna.The overall investment to set up a node wasrelatively small. Those who built the networkalready had a TNC, and the additional in­vestment for a used radio, small antenna,and cheap power supply was no problem.However,.as the network grew, node stacksstarted appearing, and the cost of multi-portnodes grows quickly. For example, a singletwo-port node might consist of

1 - 1200 baud packet TNC $1301- 9600 baud packet TNC $2001 - 2-meter radio $3001 - 2-meter antenna $501 - 70-cm radio $3501 - 70-cm antenna $752 - power supplies $150

Feed line $250

This brings a simple two-port node to a mini­mum cost around $1500. Not very many indi­viduals will spend this amount to provide ahilltop node, and if there is a need for a 3- or4-port node, the cost becomes prohibitive. Ifyou are lucky enough to find a person willingto provide such a node for your system, you'revery fortunate, indeed. Consider, however,that private ownership of the nodes comeswith potential problems: what if the ownerdies, or gets tired of packet, or needs the cashfrom selling the gear, or moves? You may sud­denly find a hole in your network that youcan't easily replace.

Perhaps a better solution is for packet groupsto sponsor nodes within the area. In suchcases, the packet group owns and operatesthe nodes, so one member of the group mov­ing or tiring of the mode doesn't disrupt thenetwork.

Network ParametersSo now that you've built a network, howdoes it automatically select the best pathfrom point A to point B? The answer lies inthe parameter settings within each node.Network nodes listen for the node broadcastfrom all neighboring nodes, assign a "quality"to each neighbor, and calculate a quality toeach destination the neighbor has reported.These quality figures are then used in therouting of packet data to its final destination.

There are probably as many theories aboutsetting network parameters as there are net­work node operators. This paper describesone such idea that is being used in theKansas network and seems to provide reli­able node operation.

Quality figures can range from 0 to 255, with255 being "perfect" and 0 being totally un­usable. As we set the quality to a neighbor,there are several items to consider. First,what speed is the link to this neighbor. A9600 baud link is better than 1200 baud, but56 KB is even better. If the backbone (link)frequency is on 2-meters and users are alsoallowed on this frequency, this is not as goodas a 2-meter link that is closed to users. Per­haps the link is on 70-cm, which typically hasfewer users and therefore should be a higher

Networking 3-9

Page 145: Packet- Speed- More Speed

protocol, so your network may consist of amixture of the various types of nodes. Someof these may offer features not found inothers, but the basic networking protocolremains constant.

In the past, installing a node meant that youhad to remove the standard user firmware(EPROM) in your TNC and replace it withone that was specifically written for network­ing. This meant that you gave up the "pri ­vate" use of your TNC and dedicated it tobeing a node. In addition, this required you toburn your own EPROM after modifying cer­tain bytes in the image to contain your call­sign, node alias, and your chosen parameters.

70140

95

BandOpening

B

F

NormalConditions220/189220/189/140189/120189/120162/104

..I..I..I..I..I

Qualityfrom Node Ato Node:

BCDEF

FIGURE 6

A

o Nodes

--- 70-em, 19,200 baud, 220 quality

- - - 2-m, 1200 baud, 140 quality

---- 2-m heard during band opening

_ ••- 70-cm heard during band opening

Using this scheme, during band openings,the distant nodes that are heard are assigneda relatively low quality, causing the node tocontinue to use a more reliable path when­ever possible. Although this will tend to limitthe number of nodes in the nodes list, a userattempting to connect to one of the distantnodes will probably succeed.

quality. What is the distance between nodes,and how reliable are the connections betweenthese stations (don't forget to account forvarying weather conditions here). A link maybe 100% reliable during the winter monthswhen there are no leaves on the trees, andfail miserably when the leaves come out.Rain and high humidity can also affect thereliability of a link.

Given the above, we developed the followingguidelines (this is only a portion of the guide­lines):

1. A 70-cm link with high-reliability, nousers, and operating at 19,200 baud willbe assigned a quality of 220.

2. A 70-cm link with high-reliability, no usersand operating at 9600 baud is assigned a200 quality.

3. A 2-meter link with high-reliability, usersallowed and operating at 1200 baud isassigned a 140 quality.

These high-reliability neighbors are "lockedin" at the quality shown, and the node is con­figured so that other nodes heard on thesefrequencies are assigned 1/2 the quality ofthe locked in nodes. Figure 6 shows thisscheme and the resulting quality for eachnode.

Building A NetworkSince the release of NET/ROM in the late1980's, there have been many derivatives ofthe networking protocol. The first of thesewas TheNet, introduced by NORD><LINK.Other derivatives that followed were theG8BPQ code, developed by John Wiseman,TheNet Plus, various flavors of TheNet X1-J,most TCP/IP NOS programs, and nowKantronics has introduced K-Net™. All ofthese systems use the same networking

3·10 Chapter 3

Page 146: Packet- Speed- More Speed

To create a two-port node, you neededtwo TNCs, each equipped with the specialEPROM (but each EPROM needed its owncallsign and alias). These two TNCs wereconnected together through their serial ports.As you expanded to a three- or four-portnode, more call signs/aliases were required,and connecting the serial ports together re­quired a somewhat complex diode matrix toavoid data collisions on the serial ports. Thenumber of diodes required is determined bythe formula 2*N*(N-1) where N is the num­ber ofTNCs being connected together. Fora three-port node, that's 12 diodes and for afour-port node you need 24 diodes .

With the release of Kantronics K-Net forthe KPC-9612, users now have the abilityto build a two-port node with only oneTNC, providing both a 1200 baud user­access port and a 9600 (or even 19,200)baud backbone port. Since both ports residein the KPC-9612, the node requires only onecall sign and one alias. The unique featuresof the K-Net in the KPC -9612 don't stop herethough - the TNC can still function as anend-user TNC. The PBBS in the KPC-9612is still usable (there's a BBS command onthe K-Net node) and a user can still run mul­tiple connects, a KA-Node™, and even oper­ate Host mode with specialized software.

Configuring a K-Net node is unlike othernodes - all the necessary commands areavailable at the command prompt. There isno need to patch the EPROM with the call­sign and alias as other nodes require. Couplethis with the remote access capability andyou can place the K-Net/KPC-9612 on aremote hilltop. When the current trusteemoves, an authorized user can connectremotely and change the call sign and alias ofthe node without making a trip to the site. Inaddition, the K-Net supports the NET/ROMinterface over the RS-232 serial port, so itcan be easily connected to an existing nodestack using other brands ofTNCs.

Building a four-port node with two 1200 baudports and two 9600 baud ports would simplyrequire two KPC-9612s with the K-Net firm­ware, and an interconnecting RS-232 cablewith only three wires - ground, TXD, andRXD (TXD and RXD must be crossed). If you

only require one high-speed port but needtwo 1200 baud user ports, that can be accom­plished using a KPC-9612 and a KPC-3 withthe optional K-Net firmware EPROMS.

Building or expanding network systems isnot extremely difficult, but for the networkto operate smoothly and provide maximumbenefit, it requires cooperation and coordi­nation between the node operators, BBS sys­tem operators, and other network serviceproviders.

DAMA - Another approachThere has been some discussion and pre­vious papers presented on DAMA (DemandAssigned Multiple Access). While thissounds like a completely different network­ing scheme, it really isn't. The basic premiseof DAMA is twofold: 1) Prevent collisions be­tween stations using the node, and 2) Allowadditional time for stations sending moredata than for those sending little data.

A DAMA node (called a DAMA MASTER) isactually a LAN node with at least one link toa backbone network. A single DAMA mastercan support up to 16 TNCs (thus 16 radios),perhaps providing two or three backbone linkfrequencies, a couple of user frequencies,some frequencies for BBS systems, and so on.Each LAN has a DAMA master node, and theadjacent DAMA masters are linked together.This inter-node link uses NET/ROM protocol,exactly like the networks just discussed.

Within a LAN, users connect to the DAMAmaster. Through special coding in the AX.25frame, the user's TNC is placed into a"DAMA SLAVE" made. From this point on,the user's TNC will not transmit any packetsuntil the master has instructed it to do so bysending a poll frame to the slave. In this'manner, no two slave TNCs will transmit atthe same time, eliminating collisions.

The demand access feature of DAMA comesthrough intelligence built in to the masternode. When the master polls a user, the usermust respond - even if it's only an ack saying"I have no data to send". After polling stationA, the master polls station B, then C, and soon. Any station that does not have data to

Networking 3-11

Page 147: Packet- Speed- More Speed

send has its priority reduced. The masterthen polls those stations with higher prioritymore frequently than those with low priority.As soon as a low priority station responds tothe poll by sending data, the master bumpsthat user's priority back up to the maximum.

The DAMA system may provide an excellentalternative in areas where LANs are verylarge, resulting in many LAN users beingunable to hear all other users. The majordrawback to the DAMA system, at this point,is that the DAMA master station requires acomputer at the node site to run the specialsoftware. In addition, multiple frequencyoperation requires one TNC for each fre­quency with special DAMA firmwareinstalled.

Users accessing DAMA systems should beusing special DAMA slave TNCs. The DAMAsoftware as distributed contains EPROMimages for TNC-2 clones, enabling them to beDAMA slaves. These EPROMS use a slightlymodified version of the WA8DED Host modeto communicate to the computer, and somesoftware packages are readily available foruse with these. Among the more popular pro­grams in this category are Grafik Packet andESKAY.

3-12 Chapter 3

Kantronics will soon be releasing a newDAMA capable EPROM for some models oftheir TNCs. The new DAMA EPROM willallow users access to DAMA networks as aslave using any terminal program, dumbtermi nal, or even Kantronics Host mode.

ConclusionNetworking in the amateur packet world canbe considered to be still in its infancy. Thefirst NET/ROM firmware was releasedaround 1987, and many improvements havefollowed. Other networking systems havebeen developed (ROSE and TCP/IP) for ama­teur use, but to-date, none has been declaredthe clear "winner" for packet networking.

Debates rage about the "best" or "perfect"network, but if such a system alreadyexisted, world-wide VHF packet would bea reality instead of a dream.

K-Net and KA-Node are trademarks of KantronicsCo., Inc.

NET/ROM is a trademark of Software 2000, Inc .

Page 148: Packet- Speed- More Speed

The Puget Sound Amateur Radio TCP/IP Network

Steve Stroh N8GNJ

Puget Sound Amateur Radio TCP/IP Group14919 NE 163rd Street

Woodinville, WA 98072 USAe-mail : [email protected]

Abstract

The Puget Sound Amateur Radio TCP/IP Network (also known as WETNET, the WashingtonExperimenter's Tcp/ip NETwork), centered in the Seattle, Washington metropolitan area, has built anextremely functional packet radio network based on TCP/IP networking and cellular RF techniques. Thenetwork encompasses more than eighteen separate Local Area Networks, an estimated 200 users, four9600 baud bit regenerative repeaters, and a full time Internet gateway. This paper is intended to providean overview of an operational Amateur Radio TCP/IP network.

A Frequently Asked Question is "Why TCP/IP? Can't you do everything with regular packet radio thatyou can do with TCP/IP?" The answer is... somewhat, but not really. A key feature about AmateurRadio implementations ofTCP/IP is that its various capabilities are built in- you don't have to combinedissimilar systems to do mail forwarding, file transfers, provide Packet Bulletin Board System (PBBS)services, multi-connect chat sessions, multiple ports, etc.- "It 's (all) in there". A typical TCP/IP stationcan do:

• file transfers (including binary files) • electronic mail• keyboard to keyboard (chat, telnet) • automatic routing• finger (display short text files) • ping (test link integrity)• operate as a very capable Packet Bulletin Board System• operate as a very capable Net/ROM, TheNET, XU node, etc.• accept multiple connections from AX.25, Net/ROM, and other TCP/IP stations• access multiple ports, including modem, RS-232, terminals, and Ethernet connections

These operations are simultaneous, since the TCP/IP software is written to multitask. The capabilitiesoutlined above are only a subset of TCP/IP's capabilities.

Some Background on Amateur Radio TCP/IP

A complete discussion of all of the various TCPlIP utilities and capabilities is beyond the scope of thispaper , but a few deserve some discussion. One ofthe most interesting capabilities of Amateur RadioTCPlIP is e-mail. Since each TCP/IP station can send and receive e-mail , there cannot be a "chokepoint" in the network for message traffic- each TCP/IP user has the ability to send e-mail to any otheruser from their station. This was a much appreciated feature after having been through "PBBSForwarding Battles", where certain PBBS Sysops wouldn't forward to other PBBS Sysops. With theadded capability of a mailing list- being able to forward a single message to multiple recipients , e-mailmade PBBS ' , and their problems, almost irrelevant. TCP/IP users have the capability to participate inPBBS forwarding, and can "translate" PBBS messages to e-mail messages, and vice-versa.

Networking 3-13

Page 149: Packet- Speed- More Speed

Being able to add multiple ports easily is liberating. TCP/IP users can gateway between multiple ports,and this capability is used to provide redundant routes, and to provide network access to remote LocalArea Networks whose users cannot access other parts of the network directly.

A TCP/IP station can effectively service AX.25 users by providing the capabilities of a PBBS, a networknode, a "chat node" , and much more to AX.25 users.

Amateur Radio TC~/IP authors have traditionally (begun by Phil Karn KA9Q) made all source code,(written in C) freely available. If you want to add a new feature, or fix a bug, the source code isavailable, and several Puget Sound TCP/IP operators have made good use of this to add features and fix .bugs, and contribute back to the code pool.

Network Operating System (NOS) was the name that KA9Q gave to his "second generation" of AmateurRadio TCP/IP software (the first was named NET) . Since NOS was released, there have been numerousoffshoots ofNOS, including JNOS, TNOS, and others. In this paper, NOS is used interchangeably withthe term "Amateur Radio TCP/IP Software".

It is important to remember that "Amateur Radio" TCP/IP is completely interoperable with"commercial", or "wired" TCP/IP . TCP/IP was enhanced to make use of Amateur Radio just as it wasto make use of satellite links, other networking systems, and modems.

Networking

VHF/UHF antenna

oPC runs modified NOS or Linux(1) TNC wfTAPR 9600 modem w/bit regen (5) UHF duplexer(2) TNC w/1200 baud w/open squelch mod (6) VHF RF eqpt.(3) UHF land mobile radio, modified for repeater and 9600 baud(4) VHF land mobile, modified for 9600 baud (used for 1200)

Figure 1 - Diagram of a typical Puget Sound Network Switch site

3-14 Chapter 3

Page 150: Packet- Speed- More Speed

•Repeater User •»>: -Strfiplex User • •Dual-Ported User •Repeater User

220 1200 baudRepeater LAN

Internet

•Internet Gateway

•Dual-PortedUser

Simplex LAN

Network Switch2m SPX / UHF RPT

2m 9600 baudrepeater LAN

Dual-Ported User

Dual-Porte d User

•Repeater User

Simplex RF link Figure 2 - Puget Sound Amateur Radio NetworkRepeater User RF link

Wire Diagram of Potential Routes

Typically , the Switches do not offer any "user functions" and only handle routing and other networkfunctions . If a user function is enabled, it is typically routed to other systems in the network. This maychange as the Switches are converted to Linux .

Several LANs have been formed by multi-ported stations that access a repeater and a 2m simplexfrequency . Network users have gradually adopted the cellular telephone approach of using low powerand a low RF profile , and thus are able to reuse frequencies . Typically, the multi-ported station formsthe center of a simplex LAN. Any stations joining the LAN are encouraged to be within good DCDrange of the other stations on the LAN. This keeps overall performance on the LAN reasonably high,despite the jnevitable collisions and hidden transmitters resulting from simplex operation.

The network is linked full time to the Internet through a Network Switch at the premises of one of thelargest Internet Service Providers in the Seattle area. The primary use of the gateway is to allowparticipation in the global Converse Bridges . Amateur Radio TCPIIP users can also telnet, FTP, etc. toInternet hosts , as long as the user has a good connection to the network so the Internet system doesn'ttime-out from long delays.

The group makes extensive use of subnetting- the third "octet" of the Amateur Radio TCP/IP IP addressis specific to a particular LAN. For example, 44.24.103.xxx is an IP address for the 220 1200 baud

Networking 3-15

Page 151: Packet- Speed- More Speed

repeater, and 44.24.101.xxx is an IP address for the 147.60 simplex LAN. RIP (Routing InformationProtocol) is used to propagate routes throughout the network.

Operational

A typical user station consists of a 386 PC using a Tekk KS-900 UHF radio with a beam, an AEA PK96or MFJl270 TNC with TAPR 9600 baud modem . The favored 2m 9600 baud radio is a modified GEMVP. The TCP/IP software of choice is the JNOS.(WG7J). It is not uncommon to for the NOS PC tobe configured as a "router" for other PC's in the household via serial or Ethernet connections to a secondPC running "commercial" TCP/IP software, often with a graphical user interface- Windows/Winsock,OS/2 Warp, or Xwindows on Linux.

AX.25 users are welcome to connect to any TCP/IP station, both to browse around on that particularsystem or be routed to another system on the network. Some user stations and Switches activelyencourage AX.25 and telnet connections by having numerous information files, binary files, and BBSareas available,. The group, at present, hasn't placed a high priority on interconnection between AX.25PBBS forwarding and Net/ROM -type networking. There is limited connectivity between local PBBSsystems and TCP/IP users through a TCP/IP to PBBS gateway which transfers PBBS messages via e!mail. Personal PBBS messages are sent privately, and PBBS bulletins are sent via a mailing list.

Setup of TCP/IP for the first time has traditionally been a problem. The group has developed a JNOSsetup disk, which includes all executable and configuration files necessary to get JNOS up and running.It includes a customized setup program which prompts the user for IP address, callsign, and otherparticulars, and then customizes the AUTOEXEC.NOS startup file, creates directories, and installs filesas appropriate. From all accounts, the disk is working well in getting TCP/IP newcomers on the air.

The Puget Sound Amateur Radio TCP/IP Group

Early members of the group had been active in other area packet radio groups and had become burnedout from the formal duties of a group, such as holding office, constitutions, dues, newsletters, etc. Adeliberate, conscious decision was made that this group would remain informal in order to concentrateon technical and other innovative lack-of-management techniques. When funding has been needed forrepairs or new equipment, money is donated by members of the group. This arrangement has workedrelatively well, but the group did recently decide to get "slightly less disorganized" by requestingvolunteers for Keeper ofthe Notes, Keeper ofthe List ofProjects, and Keeper ofthe Shekels .

The group communicates primarily through a mailing list. Members of the group that wish to pose aquestion, comment on a previous posting, or share information send a single message to the list, and thatmessage is automatically propagated. If a user's connectivity is good, messages from the list can bedelivered in minutes. Mailing list functions have been automated using Majordomo mailing list serversoftware on a Linux system.

The group meets monthly on a weeknight evening. In keeping with its informal nature, meetings are"moderated" only to the point of "collision avoidance". There is usually plenty to discuss , and formalpresentations are rare. Any "burning issues" have typically been discussed on the mailing list prior to

3-16 Chapter 3

Page 152: Packet- Speed- More Speed

the meeting, and there are only a few "reports". The group has a small advertisement in the local"Computer Paper" which draws in a few lapsed hams and technically curious non-hams.

An ongoing effort is to write effective documentation. Those involved in the documentation effort areenthusiastic about the potential of using the World Wide Web (WWW, or Web) to distributeinformation. Articles can be distributed as they are created, and updated individually. Web pages, beingcomposed of ASCII text, can easily be distributed as e-mail, or printed from a Web browser. Using Webpages also enables other groups worldwide to easily access the group's information. The group willsupport Web pages on the Internet gateway (typical graphically rich Web pages), and a well connectedRF site in the network (primarily text Web pages). The group's current Internet Web page is located at:http://wetnet.ampr.org/

Future goals and projects• Enhancing the reliabi lity of the overall network is a high priority• Expand high speed TCP/IP links to other areas in the Pacific Northwest, especially Vancouver and

Victoria, British Columbia and Portland, Oregon.• Implement several 56 KBPS and higher links• Replace NOS with Linux at all Network Switches• Increased use of Linux and other TCP/IP capable systems• Contribute to the forthcoming TAPR TCP/IP book• Sponsor a TAPR Annual Meeting and/or an ARRL Digital Communication Conference• Accommodate increased use of PC's, Macs, Amigas, and Unix systems running their native

implementations of TCP/IP.• Aggressively test and implement new radio-based routing techniques as they are developed

Conclusion

TCP/IP works very well for Amateur Radio networking. It has been a satisfying learning experience toget it running and keep it functioning. The results are well worth the effort expended.

The group would enjoy hearing from Amateur Radio TCP/IP users around the country and the globe thathave implemented Internet access to their Amateur Radio Networks.

For those that are able, wouldn't it be more interesting to use our "radio" computers to participate inmailing list discussions such as NOSBBS, NETSIG, and others? There is no reason not to if yoursystem has adequate connectivity to the Internet.

Networking 3-17

Page 153: Packet- Speed- More Speed

References

Steve Stroh N8GNJ, [email protected] for questions on the Puget Sound Amateur Radio TCP/IPGroup,

Ken Koster N7IPB, [email protected] for questions on the Puget Sound Amateur Radio TCP/IPNetwork,

Jeremy Donimirski WA7YGB, [email protected] for questions on the US West NewVectorAmateur Radio Group and the Cellular Base Station systems .

The most complete book on Amateur Radio TCP/IP is NOSintro, subtitled TCP/IP of Packet Radio, Anintroduction to the KA9Q Network Operating System by Ian Wade G3NRW, copyright 1992 byDowermain Ltd., ISBN 1-897649-00-2. NOSintro is available from the ARRL .

A good FTP site for downloading Amateur Radio TCP/IP software from the Internet is:ftp.ucsd.edu/hamradio/packet/tcpip

The mailing list for Puget Sound Amateur Radio TCP/IP Network users is:[email protected]

My thanks to the users and builders of the Puget Sound Amateur Radio TCP/IP Network- it's not anetwork without users to use it. It's been quite a ride, and we're "finally having fun yet". I'd especiallylike to thank N7IPB, KD7NM, WA7QFR, N7NKJ, and WA7FUS for inviting me into the group.

3-18 Chapter 3

Page 154: Packet- Speed- More Speed
Page 155: Packet- Speed- More Speed

Chapter 4Construction

Page 156: Packet- Speed- More Speed

An Easy Path toPacket: the IMPA single-chip modem makes foran inexpensive packet package.By Tony Marchese , N2YMW

35 Shannon CrescentSpencerport, NY 14559 -9758

N2YMW 'IMP'P"c*t1 t Modem

Except as indicated, decimal values of C6capacitance are in microfarads (I-'F) ; 0.3~others are in picof arads ( pF);resistan ces are in ohm s; k= 1,OOO. 01,02,03* See t ext 2N3904

n.c.= not connect ed ~IC pins no t shown are unus ed.OB9 pin numbers are shown in parentheses.

E B C

Figure 1-Schematic of the IMP circui t. RSpart numbers in parentheses are RadioShack; OK part numxbers are Oigi-Key(Oigi-Key Corp , Box 677, Thief River Falls ,MN 56701-0677; tel 800-344-4539, 218­681-6674 ; fax 218-681-3880) equiva lentparts can be substituted. Unless otherwisespecified, resistors are ' /. -W, 5%-tolerancecarbon-composition or film units .C6-100 fl F, 16 V (RS 272-1029 ;

OK P6620)01-07-1N914, 1N4148 (RS 276-1122;

OK1N4148CT)01, 02, 03-2N3904A (MPS3904;

OK 2N3904)R9-1 OO-kQ trimmer potentiometer

(RS 271-284; OK 3296W-104 orOK 3296Y-104)

R11-User-selectable for radios thatcombine PTT and TXA lines; refer to yourtransceiver's user manual.

U1-TCM31 05 FSK modem (available fromJOR Microdevices, 1224 S 10th St, SanJose, CA 95112, tel 800-538-5000, 408­494-1400 ; fax: 408-494-1420)

U2-LM358, dual , low-power comparator(OK LM358AN)

U3-LM78L05 positive 5-V, 100-mA voltageregulator (OK AN78L05)

Y1-4.4336 MHz (OK X083 [HC-49/UAholder]; OK SE3406 [cylindrical holder])

Mise: enclosure 3' /.x2 '/8X1'/8(270-230) ;connectors for transceiver; OB25 or OB9connector as needed; 8-pin IC socket;16-pin IC socket

of the digital and analog signals. Q3 invert sthe CLK signal to VI ' s transmit/receivestandard select (TRS) input, which, in con­junction with bit-rate-select input s (TXRI)and (TXR2) at ground potenti al, configurethe TCM3105 as a 1200-b aud, Bell 202­comp atible modem . This configuration usesidentical tone s for the transmit and receivesignals, with 1200 Hz representing a mark(logic I) and 2200 Hz repre sentin g a space(logic 0), the appropriate frequencies for1200 baud , half-duplex packet operat ion .Amateur packet uses a half-duplex commu­nication mode, (ie , one direction at a time)beca use radio transcei vers don 't typically

...........--......-/+5 V

C5*100l-'F

16 VU3

78L05

~OUT GNO IN

1N914(3)

+------iPTI

Y1 C116 4.4336 MHz 20

C2 ~O Receive Audio'--_-=-j (from Headphone

Jack)t--'----i f-......--..---t RXA

......---IGNO

rial port and typically requires no externalpower suppl y (more about that later).

Circuit Operation

Refer to Figure 1. V1 performs all of thefiltering , timing, and conversion function srequired for modulation and demodulation

R868 k

7

06

05

R756 k

C90.01

FSKModem

TXO 2(3 ) I---+I--+---I

6

OTR20(4) ~""I--"

n.c.

R222k

+5 VH--'\/\/\r+-'\I\!\r--,

U1TCM3105N

R4

47 k 10 k

+5 V

07 R11N914 220 k

OTR 20( 4) I-...--'W\r..,

H ere' s a Bell 202-compatible, FSKmodem I've dubbed the IMP-anIntrodu ctory Modem for Packet.

The IMP can be used with any computerequipped with an RS-23 2-C serial port, ap­propriate communications softw are and atransceiver to acce ss 1200-baud packet ap­plic ation s such as PBBSs and satellites. TheIMP connects directly to the computer' s se-

Construction 4-1

Page 157: Packet- Speed- More Speed

Sign alfrom

Co mput erComputerModemCommonComp ute r

tran smit and receive simultaneo us ly . 0 I and0 2 limit the incoming audio signal to ap­proxim at el y 0. 65 V as the maximum signa lle vel for the TCM3 105' s Re ceive Ana loginput (RXA) is speci fied as 0.78 V.

Q2 and U2 co nvert the information sig­nals bet ween the ±3 to ±15 V, RS-232- Clevel s and the 0 to 5 V level required by theTCM 3105 . 04 through 0 6 logicall y OR thevo ltage from the RT S, DTR and T xD signa llin es to typi call y produce a 7 to 15-V depower supply. Th e combination of Q4 andC5 pro vid e a relativel y sta ble 5-V de sourcecapable of providing the lOrnA required bythe IMP .

Construction

No special co ns tructio n tec hniques arerequi red and a PC board is ava ilable. I Th ebo ard fits ne atl y in side a sma ll (31/ 4X21/ sxI lls-inch [HWD]) project ca se specified inthe part s list. I built my latest version of thismodem (i ncl ud ing ho me brewing the PCbo ard and lettering the enclosure) in twoevenings. You'll need a cable to connect themod em to yo ur tran sce iver. Refer to yo urtran sce ive r ' s documen tat ion to determinethe required co nnec to r type s and wiring .

Adj us tme nts

The co mpute r requires packet co mm uni­ca tio ns software . Th e IMP is compatib lewith a varie ty of T NC-Ie ss pack et so ftwaresuc h as BayCom , writte n by FlorianRadlherr, DL8MBT. BayCom is avai lablefro m the ARRL BBS and FTP site ,2 and fromman y shareware distributors.

Co nne ct yo ur tran scei ver and a co mputerequ ipped wi th an RS-232-C port to the IMP .Turn on the radi o and tun e the recei ver to anac ti ve packet frequency . Start the packetsoftware , but don 't att emp t to transmit .Wh ile running the software in the receivemode , pl ace a voltme ter across the modem ' sT xD and GN D pad s and verify that the volt­age supplied by the co mputer' s serial port isa minimum of +7.0 V. If the vo ltage is below7.0 V, yo u' ll need to power the IMP from anexternal 9-V battery or other power source .

The IMP is eas ily modified for 9-V bat­ter y operatio n. A 9- V batt ery co nnector, anSP ST switch and some hook-up wire are allyou 'll ne ed . So lder the po sitive batter y wireto one side of the swi tch. Remove the serialport ' s T xD wire from the PC board. So lder alen gth of wire to the swi tc h operating armand to the T xD pad . Complete the modi fi ca-

1A PC board for this project is available fromFAR Circuits , 18N640 Field Ct, Dundee , IL60118-9269 , tel 708-836-9148 (voice and fax).Price: $4.50 plus $1.50 shipping for up to fourboards . VISA and MasterCard acc epted with a$20 minimum orde r, or $2 service charge onorde rs of les s than $20. A PC-boa rd temp latepackage is available free of charge from theARRL. Send your request for the MARCHESEIMP MODEM TEMPLATE along with astamped business-size SASE to the TechnicalDepar tment Secreta ry, ARRL, 225 Main St,Newington, CT 06111-14 94.

2ARRLBBS: 860-594-0306;ARRLInternet FTP;oak .oakland.edu in the pub I hamradio I arrl lqat-binaries directory.

4-2 Chapter 4

Table 1RS-232-C Signals Used with the IMP0825 RX TX(089) RS232 Level LevelPin No. Signal (V) (V)2 (3) TxD +10 +104 (7) RTS -10 +105 (8) CTS D~a X7 (5) GND 0 0

20 (4) DTR +10 Data

Figure 2-An inside view of the IMP modem.

tion by so lde ring the batt er y co nnec to r 'snegat ive lead to a co nve nie nt gro und point.

Wh en the vo ltage at T xD is greate rthan 7.0 V, ver if y the output of U3 by placinga vo ltme te r ac ross U I pin s I an d 9. Yo ushould measure 4.5 to 5.5 V. (If yo u don' t,check the orie nta tio n of U3.) Yo u shouldalso see info rma tion app ear on the monitoras dat a is recei ved. Yo u may need to incr easethe recei ver 's vo lume control slightly ifnoth in g appea rs on the screen. The receivemod e should no w be opera tional.

Th e la st adj ustme nt optimize s themod em' s analog output level. Att ach adummy ante nna to the transceiver. Activatethe softwa re's tran smit mod e while mo nitor­ing the radio's output on anot her receiver.Adj ust R9 (TXADJUST) until you hear a cleansignal.' Th at ' s it ! Install the modem in itscas e, rep lace yo ur tran scei ver ' s antenna andyo u're rea dy fo r packe t operation!

I' ve successfu lly used the IMP to access

3The IMP modem was test ed in the ARRL Labusing a Kenwood TH-22 and ICOM IC-2AT,the latter having a PIT switch in series withthe mike line . When the mike connector wasplugged into the IC-2AT, the receive mode wasinterrupted . Making a small adjustment to R9brought the receiver bac k to life.

Notes+10 V (most of the time)Activates PTT dur ing transmitData from modem during receiveSignal groundData to mod e m during transmit

infor ma tion on various pac ket BBS s and ampat iently awaiti ng my firs t satelli te contact.I hope yo u ha ve as much fun with the IMP asI have had .

Finally , I wo uld like to than k Virg ilAn sley, KF2XT, for his inspiration, pa tie nceand Elmering .

Tony Marchese , N2YMW, was born in Rochester,New York, in /960. His interest in radio began in theearly I970s, but he didn 't get his Amateu r Radiolicense until Mar ch 1994. when he receive d hisTechnician plus HF license. Tony is currently refur­bishing an old Heathkit SB300/SB400 "twins " pairand hopes to be on HF soon.

Tony holds seve ral degrees including a BSc incomputer engineering and AAS degrees in biomedi ­cal engineering and electrical engineering technol­ogy. He is presently employed as the manager oftheBiomedical Engineering Departm ent at Rochest erGeneral Hospital and as an adjunct instructor fo rthe Computer Related Curricula and Electrical En­ginee ring Technology Departm ents of MonroeCommunit y College.

Tony has many interes ts, but the majorit y ofhis per sonal time is spent in fami ly activities­whether it be a trip to the zoo or a home-improve­ment proj ect-with his wife, Colleen, and theirtwo daughters, 5-year-old Sarah and l '/z-year-oldEmma. Tony enjoys designing and buildin g elec­tronic proj ects and is presently concentrating onthe development of devices fo r the physica llychallenged. IO§'fL.1

Page 158: Packet- Speed- More Speed

Radio-TNC Wiring DiagramsThe following information was graciously provid ed by Gloria Medcalf , KA5ZTX. She has done a fine job in compiling wirin g diagrams

for the most common TNCs and radios used for packet radio . To use these diagrams, wire pins with the same name s to each other. For example,the Tr ansmit Signal on the radio connector is wired to the Transmit Signal of the TNC connector. Pins are labeled with the descriptive namesused in Chapter 5 of Glo ria's book, What is YO llr TNC Doing ? Consult her book for informatio n not found here . Manu facturers have beenknow n to change thei r usage of pins, so it would be wise to compare these diagrams with those given in the produ ct manual. Damage mayoccur if a wrong connection is made .

TNCs

AEA, PK-232

AEA PK-232 Radio Connector

5-PTI 1 - Receive Signal

4 - Ground 2 - Transmit Signal

Female (wiring side)

AEA, PK-88

7 - External Carrier Detect

AEA PK-88 8-pin Mic Connector

8 - Receive Signal1 - Ground

2 - Transmit Signal

3-PTI

Female (wiring side)

5-pin DIN connector used on many popular TNCs including the DPK-2, DSP-2232, KPC-l,

MFJ-1270, MFJ-1274, PacComm 200 , PacComm 220, PK-87, PK-900, Tiny-2, TNC-2

TAPR TNC-2 5-pin DIN Connector

3-PTI

2 - Ground

1 - Transmit Signal

4 - Receive Signal

(some models)

Male (wiring side)

Construction 4-3

Page 159: Packet- Speed- More Speed

Notes:Pin 4 on the KPC-l has a separate connector the Received Signal (pin 4) . Pin 5, External Carri er Detect, is used on the PK­900, DSP-2232, DPK-2

Kantronics with DB-9 connector for KPC-9612, KPC-3, KPC-4, KPC-2400,KAM Plus, KAM

Kantronics VHF OB-9 Connector

3- PTT

5 - Receive Signal

9 - Ground

2 - External Carrier Detect

1 - Transmit Signal

6 - Ground

Male (wiring side)

Kantronics with DB-IS connector for KPC-96I2 and Data Engine

Kantronics OB-15 Connector (9600 Baud)

3 - Transmit Signal

Male (wiring side)

2 - Receive Signal

1 - PTT

11 - Ground

Kantronics HF 8-pin DIN connector for the KAM Plus and the KAM

Kantronics HF a-pin DIN Connector

8 - External Carrier Detect

3-PTT

5 - FSK out (for RTTY)

4-4 Chapter 4

Male (wiring side)

6 - Receive Signal

1 - Transmit Signal

4 - Key out (for CW)

2 - Ground

Page 160: Packet- Speed- More Speed

Radios

8-pin Mic Connector for Alinco and Kenwood radios

Kenwood a-pin Mic Connector

1 - Transmit Signal

8 - Ground

6 - Receive Signal

(provided on some models)(not wired if speaker plug is used)

2-PTI

Female (wiring side)

Note s: Pin 6 may not be sued on some modes that use the speaker plug.

Speaker Plug

Speaker Plug (typical)

3.5 mm mono plug(to speaker jack)

Receive Signal

Ground--- ----IH'----- lWsleeve

Construction 4-5

Page 161: Packet- Speed- More Speed

9600 Data Jack and PTT Jack for real panel connection to Alinco DR610T and DR150T

Alinco 9600 baud rear connections - DR61OT

Receive Signalring

Transmit Signal

Ground

PTI Signal *sleeve

tip

2.5 mm stereo plug(to Data jack)

Ground *--------10 Yoo,---lhDsleeve 3.5 mm mono plug

(to PTI jack SP1)

*Models like the DR150T use the same 2.5 plug wiring , but havePTI and Ground wired to the front-panel 8-pin mic connector.

Note s:The DRI50T uses the 2.5 mm plug but PTT and Ground are wired to the front-panel 8-pin mic connector. The radio mayhave to be in the packet mode for the PTT to be activ ated on the plug. Some Alinco manuals have the PTT wiring information printedincorrectly.

Separate Mic and Speaker Plugs for Alinco and ICOM IC-02AT and IC-2AT and newer

models

ICOM 2AT Style Hand-Held Radios (and newer models)

2.5 mm stereo plug(to mic jack)

PTI

Transmit Signal0.1 I-lf

___~-mv3.9 KQ

Ground

Receive Signal

-----~ lhDsleeve 3.5 mm mono plug

(to speaker jack)

Notes: Values of the resistor and capa citor are not crit ical. Alinco radio s may requ ire a 15K n resistor instead of the 3.9 Kn asshown .

4-6 Chapter 4

Page 162: Packet- Speed- More Speed

Single Mic and Speaker Plug for Alinco and ICOM IC-W2A and newer models

ICOM W2A Style Hand-Held Radios (and newer mode ls)

Transmit Signal

PTT

Receive Signal

Ground

3.9 KQ *tip ring

sleeve 2.5 mm stereo plug(to speaker/mic jack)

*Radios made by other manufacturers that use this configuration ,. may require different resistor values .

ICOM /Yaesu with Transformer

TNC Hand-Held Rad io(ICOM,Yaes u)

PTT

Transmit Signal

Ground

Transmit SignalGround--2Receive Signal Receive Signal

Notes : Alinco Hts may require a 15 KQ resistor. A transformer is used instead of the resistor and capacitor for ICOM and Yaesuradios.

6-pin Mini DIN for Azden and Kenwood radios

6-p in Mini DIN Rear Panel Data Connector

6 - External Carrier Detect 5 - Receive Signal (1200 Baud)

4 - Receive Signal (9600 Baud)

2 - Ground

3-PTT

1 - Transmit Signal

Male (wiring side)

Construction 4-7

Page 163: Packet- Speed- More Speed

Note s:Some radios may require you to select Pin 4 or Pin 5 for the Received Signal. This is sometimes done throu gh a menuselection . Other rad ios may alway s have both pin s active so that no selection is nece ssar y. Both speeds can be rece ived on Pin 4,however, 1200 baud may have more noise if Pin 4 is used instead of Pin 5. The audio at Pin 5 may be unsquelched. The audio atPin 4 is unsquelched. Many 1200 baud TNCs accept squelched audio as the default mode. If onl y unsquelched audio is available,the TNC will always be receiving noise and must be able to detect packet signals out of the noise. This is often accomplished witha TNC command for carri er dete ct mode. Unsquelched is the standard for 9600 baud operation.

Mic Schematic for Drake TR-7 and Kenwood TR-7400

Typical Schemat ic of Microphone (4-pin mic connector)

Pin 1 - Transmit SignalPin2 -PTI

o~ PTI Switch

=

Female (wiring side)

Typical Speaker Plug

Typical Connection for Receive Signal to Speaker Plug

mono plug(to speaker jack)

Ground

Receive Signal

------1~'--_lWsleeve

ICOM 8-pin Mic Connector

Icom 8-pin Mic Connector

1 - Transmit Signal 8 - Receive Signal (some models)

(not wired if speaker plug is used)

6 - Ground

5 -PTI

Female (wiring side)

4-8 Chapter 4

Page 164: Packet- Speed- More Speed

Speaker Plug (typical)

3.5 mm mono plug(to speaker jack)

Ground

Receive Signal

--------1~'----------JlhDsleeve

Notes :Pin 8 is not used if the external Speaker Plug is used .

ICOM 8-pin Modular (RJ-45) Mic Connector

Icorn a-pin Modular Mic Connector (RJ-45)

Pins are numbered 1 to 8 from left to right when looking atconnector with the cable towards you, the locking mechanismdown , and the connector pins facing up.

4 -PTI6 - Transmit Signal

3 - Receive Signa l II 1 7 - Ground

• ~ I connector

~ pins

( cable

Notes :Pins are numb ered I to 8 from left to right when lookin g at co nnector with the cabl e toward s you, the locking mechanismdown, and the connector pin facing up.

Construction 4-9

Page 165: Packet- Speed- More Speed

ICOM Radios with 8-pin accessory connector

Icom a-pin DIN Connector - ACC (1)

6 - External Carrier Detect

3-PTT

5 - Receive Signal

1 - FSK (some models)

4 - Transmit Signal

2 - Ground

Male (wiring side)

ICOM Radios with 13-pin accessory connecto r

Icom 13-pin DIN Connector - ACC

13 - External Carrier Detect

10 - FSK11 - Transmit Signal

2 - Ground

12 - Receive Signal

3-PTT

Female (wiring side)

Not es: If you use FSK with the IC-706, see the manual for the sett ing the radi o' s tone , shift, and keyin g polarity. The pin s on thisconnec tor are very close together and are very difficult to solder. To make it easier, remove the unne eded pins for the conn ector beforeso lder ing .

4-10 Chapter 4

Page 166: Packet- Speed- More Speed

Kenwood 8-pin Modular (RJ-45) Mic connector

Kenwood a-pin Modular Mic Connector (RJ-45)

Pins are numbered 1 to 8 from leftto right when looking atconnector with thecable towards you, thelocking mechanismdown, and the connector pins facing up.

3 - Transmit Signal

2 - Receive Signal

( cable

Notes:Pins are numbered I to 8 from left to right when looking at connector with the cable toward s you, the locking mechan ismdown and the conn ector pins facing up. The Receive Signal is unsquelched. Many TNCs are defaulted to accept squelched audio .With unsquelc hed audio, the TNC will always be receiving noise and must be able to detect packet signals out of the noise. Thi s isoften accomplished with a TNC command for carrier detect.

Construction 4-11

Page 167: Packet- Speed- More Speed

Kenwood 13-pin DIN accessory connector, ACCY 2

Kenwood 13-pin DIN Connector - ACCY2

9 -PTT

13 - PTT-- - - - - - -I-- - - ~

early models I II

early models • Ionly ~ I

II

2- FSK

late models

11 - Transmit Signal

4 - Ground

3 - Receive Signal

Female (wiring side)

Notes : Pin 2 is used for FSK on late model radios. Later modes do not need Pin 13 connecte d. Ear ly models have separate PTTpins for completing the circu it and for muti ng the mic. Both pins must be connected with a small-s ignal diode (1N9l 4 or equi valent )as shown for ear ly models. Check your manu al. the pins on this connec tor are very close toge ther and are diffic ult to solder. To makeit easier, remove unneeded pins for the connec tor before solder ing.

Kenwood 6-p in Mic con nector

Kenwood 6-pin Mic Connector

1 - Transmit Signal

2 -PTT6 - Ground

Female (wiring side)

4-12 Chapter 4

Page 168: Packet- Speed- More Speed

Kenwood Mic and Speaker plugs for the TH-78 and TR-2600

Kenwood 2600 Hand-Held Radios (and newer models)

0.1 I1f

PTT

Transmit Signal----------I II----~ ring

----~L---[(t:Vsleeve 3.5 mm stereo plug

(to mic jack)

2.5 mm mono plug(to speaker jack)

Ground

Receive Signal

------------1~I----wsleeve

Kenwood with Transformer

TNC

Transmit Signal

Hand-Held Radio(Kenwood)

Transmit Signal

Ground Ground

PTT PTT

Receive Signal Receive Signal

Notes: You may use a transform er instead of the blocking ca paci tor.

Construction 4-13

Page 169: Packet- Speed- More Speed

Radio Shack radio s with 8-pin Modular mic connector

Radio Shack 8-pin Modular Mic Connector (RJ-45)

Pins are numbered 1 to 8 from left to right when looking atconnector with the cable towards you, the locking mechanismdown, and the connector pins facing up.

2 - Ground5 - Transm it Signal

I II 6 -PTI

• ~ I connector

~ pms

( cable

Notes :Pin s are number I to 8 fro m left to right when looking at connector with the cab le towa rds you, the locking mechanism downand the connecto r pin facing up.

Yaesu and Radio Shack Mic and Speaker plugs for FT-ll, FT-530, FT-727, HTX-202, HX-404

Yaesu FT-727 Hand-Held Radios (and newer mode ls)

Transmit Signal

PTI

0.1 uf

___~f-------rw2.2 KQ 2.5 mm mono plug

(to mic jack)

Receive Signa l

Ground----~ Wsleeve 3.5 mm mono plug

(to spea ker jack)

4-14 Chapter 4

Page 170: Packet- Speed- More Speed

ICOM IYaesu with Transformer

TNC Hand-Held Radio(ICOM,Yaesu)

PTT

Transmit Signal

Ground

Transmit SignalGround-CL.--~ __Receive Signal Receive Signal

Notes :You may use a transformer instead of the blocking capac itor and resis tor.

Yaesu radio with 8-pin Mic connector

Yaesu 8-pin Mic Connector

8 - Transmit Signal 7 - Ground

Female (wiring side)

6-PTI

4 - Receive Signal (some models)

(not wired if speaker plug is used)

Yaesu 5-pin DIN for FT-990 and FT-IOOO when using AFSK

Yaesu 5-pin DIN Rear Panel Packet Connector

3- PTT

5 - External Carrier Detect

2 - Ground

1 - Transmit Signal

4 - Receive Signal

Male (wiring side)

Notes: Th e rad io ' s Packet dip swi tches , whic h are located und er the top access panel, need to be set to the tones used by the TNC.The radio's rece ive fi lters are centered on the tones se lec ted with the Packet dip switches. Most TNCs use 2 100/2310 Hz. For packetthe KAM and KAM Plus use the tones set by their MARK and SPACE comma nds ; for all other digi ta l modes, they use the tones setby their MARK, SPACE and SH IFT co mmands.

The "P acket FSK Tone Pairs" char t in some Yaesu manu als is pr inted incorrectl y . (Although this chart is called FSK tone, thetransmission meth od used is Audio or AFSK. Th e char t sho uld read;

Construction 4-15

Page 171: Packet- Speed- More Speed

Dip Switches 10701127023

160011800ONON

TNC Tone Pair20 25/2225 2 11012310ON OFFOFF ON

OFFOFF

OFF is to the left and On is to the right when fac ing the radio .Th is conne cto r requires high level dri ve which ca n be adj usted in the TNC or with the radios Mic co ntro l for prop er ALC operation.

Yaesu 4-pin DIN for FT-990 and FT-IOOO when using RTTY FSK operation

Yaesu 4-pin DIN Rear Panel RTTY Connector

4 - Groun d

3 -PTT

1 - FSK

2 - Receive Signal

Male (wir ing side)

Note s: The radio ' s RTTY dip switches , which are located under the top acc ess panel, need to be set for the shi ft and low tonefreq uency used by the TNC. Shift is normally 170 Hz. Low tone frequency is normall y 2125 Hz , except in Europe where 1275 Hzis the norm.

The radi o ' s RTTY Tone slide, which is located under the top access panel, need s to be set correc tly. Th is is es pecially nece ssaryfor AMTOR and RTTY/ASCII mode s. Wiring the FSK pin s bet ween the radio and the TNC creates a simple cont act -swit ch circui t.Th e TNC open ca n clo ses the sw itch. The radio ' s RTT Y Tone slide tell s the radio whether to transmit a high tone (space) or a lowton e (mark ) when the switch is opened (and the opp osit e when closed). Check your TNC man ual to see if the TNC open s or clo sesthe switch to send a mark . If the circuit is ope n whe n the TNC send a mark, the radio 's Tone slide switch should be se t to REV . SomeTNCs have an FSK invert command that can be used instead of changing the radi o' s RTTY Tone slide .

Th is co nnec tor may be used for any two-t one digit al mode, such as AMTOR, G-tor , 300 baud Packet , Pactor , and RTTY/ASCII.Wh en you select the RTT Y mod e (on the fro nt of the radi o), the radio uses the FSK pin on thi s co nnec tor to gene rate the transmittedsignal.

Yaesu Data IN/OUT plu g for the FT-5100

Yaesu FT-5100 DATA IN/OUT Plug

Receive Signal

Transmit Signal

PTT

Ground

4-16 Chapter 4

2.7 KQ

ring

3.5 mm stereo plug(to rear panelDATA IN/OUT jack)

sleeve

Page 172: Packet- Speed- More Speed

ICOM IYaesu with Transformer

TNC

PIT

Hand-Held Radio(ICOM, Yaesu)

Transmit Signal

Ground

Transmit SignalGround-c_~ _Receive Signal Receive Signal

Notes: Th is data ja ck requires a high outp ut drive level fro m the TNC. A transformer may be used instead of the res istor andcapacitor.

Construction 4-17

Page 173: Packet- Speed- More Speed

Introducing the Ottawa PI2 (Packet Interface 2) CardThe PI2 ca rd is a PC-comp atible synchro nous interface card for high- speed packet radi o. Th e PI2 is the successor to the original

Otta wa PI ca rd, firs t int roduced in late 1990 and now wide ly used in packet rad io networkin g applic atio ns, in a variety of PC ' sranging from XT's to 486's . Th e PI2 is a two-port interface card whi ch offers a superior alternative to Terminal Node Controllers,since it will not become obsolete when you wantt o move up to speeds beyond 9600 bit s per seco nd. The high -speed port of the PI2is designed for operation at spee ds up to at least 250 kilo bits per seco nd. A 1200 bps modem for the low-speed port can be installedon the card , allow ing packet operation by simply adding a radio . In shor t, thi s ca rd offers the packet radio experimente r highperfo rman ce at a reasonable cost.

Th e PI2 card is produ ced by the Packet Working Group of the Ott awa Amateur Radi o Club, Inc ., a nonprofit , volunteer orga ni­zatio n. All proceeds from the sale of the ca rds are used for the developm ent of amateur packet networking facilit ies.

TECHNICAL SPECIFICATIONS

85C 30 CMO S SCC chip (7. 3 MHz clock )Two half-duplex port s:

The 'N port: High-speed DMAThe 'B ' port: Low-speed interrupt-driven

High- speed port interface opti ons:Buffered TTL (standard) fo r dir ect connectio n to modems, e.g., GR APES 56 kb/ s modem or Kantroni cs D4-1 0RS-422 (user-ins talled option)

Low- speed port is uncommitted . User can either install on-board TCM3 105 1200bps modem, or install header for external TTLconnections

On-board modem has state machine DCD for 'open squelch' opera tionHigh-speed port has pro visio n for 32X bit rate clock output and rx data line gating, for G3RUH-ty pe modemsOn-board timer chip for reliable CP U-independent timingDriver software, including source code, for KA9Q NOS included (backwards compatible with the original PI card)Compatible with PC, PC/ XT , and PC/AT type system sSmall size (6.25" length) and low power consumption 0

Pre-c onfigured software and detailed docum ent ati on included

The PI2 card and its predecessor have been exten sively tested with the GRAPES 56 kb/s modem and KA9Q NOS, on various XT­and AT-class sys tems. Bench tests indicate that the high-speed port can support speeds of more than 460 kb/s .

Support is available via the Internet and Usenet.( Contac t: [email protected]).

SUMMARY OF DIFFERENCES BETWEEN THE PI2 AND PI CARDS

A mech anical fit problem has been fixed (olde r PI card wo uld not fit in some PCs unless the bra cket was rem oved ).The PI2 has a CMOS SCC (85C30) instead of NMO S, and run s at twic e the clo ck speed (7.3 vs 3.6 MH z). Th e CMOS SCC has

several improvements over the NMOS chip, including relaxed timing constraints which should result in fewer problems related tononstandard PC bus timing . One result of the higher clock speed is that it extends the maximum bit rate for internal SCC clockrecovery to 57.6 kb/ s from 28.8 kb/ s (note: maximum bit rate wi th ext ernal clo cking is much high er than thi s).

Th e PI2 has a CMOS timer (82C54) instead of the NMOS 825 3, and it run s at twice the clock rate. Th e PI2 high-speed port ' sconnections to the out side world now have TTL buffers (the PI only had direct connections to the SCC chip).

The PI2 high-speed port has provi sion for RS-422 interfacing (PI was TTL-only).The PI2 can be populated with an on-board 1200 bps modem , including state machine DCD cir cu it and watchd og tim er, dr iven

by the low-speed port. Th e PI only had provision for external connec tions to thi s port (PI2 still has this too). A complete part s kitfor the modem is available.

Th e gate oscilla tor circuit in the PI has been replaced with a canned oscillator in the PI2.The RX Data line is now qualified by DCD , to prevent unn ece ssar y interrupts caused by noi se .Th e PI2 has pro vision fo r a 32X bit rate transmit clock output on the high-spe ed port, which can be used to run the state machine

in G3RUH-type modems. The PI software dr iver has been modifi ed to suppo rt thi s mode of clocking .The PI software dri ver has been enh anced to allow different interface nam es to be att ached to the low- and high- speed port s. The

new dr iver is com patibl e with the old card .Th e documentation has been ex tensively re-wr itten and improved .Th e PI2 is slightly larger (abo ut 1.25" longer ) than the PI.

ORDERING DETAIL S

Th e PI2 ca rd is available onl y full y assembled and tested (modem not included), for US $ 125, or CD N $ 155, plu s shipping. Aco mplete parts kit for the on-board modem is avail abl e for US $30 or CDN $40 (doc umentatio n on the modem is included with eve ryPI2 card, so you can acquire yo ur ow n parts if you wis h). For orders from North Amer ica , shipping is by air parcel post. Fordestinations outs ide North America, shipping could be by surface mail. If you require fas ter delivery or shipping by other means,

4-18 Chapter 4

Page 174: Packet- Speed- More Speed

please inquire before ordering.The package includes KA9Q NOS (JNOS version) configured for the PI2 card , driver sourc e, and installation documentation with

schematic. In addition to the built-in driver for NOS , a packet driver (AX.25 class) compliant with the FTP Softw are Inc. specifi­cation is included in the package . The original PI card has been supported in NOS for several years now, and new relea ses of thedri ver s will be readil y ava ilable from the usual Intern et ftp sites (e.g., ucsd.edu, ftp.ece.orst. edu, and hydra.carleton.ca).

The low-speed modem kit contains a complete set of part s (including modem chip, crystal, and state mach ine ROM ) and detai ledinsta llation instructions. Assemb ly and tune -up shou ld take about an hour. The PI2 documentation also includes instru ction s oninstalling the RS-422 line drivers and recei vers.

Send this order form to:

Att n: Packet Working GroupOttawa Ama teur Radio Club , Inc .P.O. Box 8873Ottawa, ONK IG 312CANADA

Name : Call sign : _Address:City : Prov./State: . _ZIP/Postal Code: Cou ntry : _Telephone: E-ma il addres s: _

Item Quantity Price Total

PI2 card (assembled & tested ): __x US$125 (CDN$ 155) =On-b oard 1200 bps modem kit: x US$30 (CDN$40) =Add $ 10 (US$ or CDN $, fla t rate per order) for shipping:To tal amount enclo sed :

Plea se make mone y-order or check (sorry , no credit cards) payable to:"Ottawa Ama teur Radio Club" .

Construction 4-19

Page 175: Packet- Speed- More Speed

Modifications for the Alinco DR-1200

MODIFICATION TO HELP WITH INTERMOD PROBLEMS WITH THE ALINCO 1200 DATARADIO

CAUTION:

This radio employs static sensitive devices. Suitable means should be emp loyed to assure

a static-free environment prior to beginning work on this radio .

1. Remove power and antenna from rig

2. Remove bottom cover

3. Locate c-19, left side front of board (labeled, near q3).

4. Remove c-19 : suck up the solder and carefu lly pry away from board wh ile still warm.

5. Save the capacitor in case you got the wrong part!

6. Reassemble the radio, no retuning necessary.

This capacito r connects 01 (back to back diodes) to the first IF line. 01 is supposed to be a "limit er" . Unknown whyAlinco chose to put in such a limiter, as the integrated IF chip IC-1, an MC3357, has a built-in limiter. Also, the KenwoodTM-231 which uses nearly the same layout and circuit , doesn't have such a circuit. While the intermod hasn't gone awayentirely (I 'm not done with it yet!) , it has diminished enough to make the rig usable on packet AND voice now.

73 de Bill, KOZL@WOLJF.#NECO .CO.NOAM

ALINCO DR-1200 CONSTANT TNC RX AUDIO MOD

Pin 6 on the DR-1200 is connected to the speaker audio lead inside the rig to allow you to

use a sing le cable to connect your TNC . The problem with this is that you have to leave the

audio level up a little for the TNC to decode properly. Most people can only

stand so much packet racket! Here's a solution :

CAUTION:

The following procedure involves desoldering smd (surface mount devices). If you are not

qualified to work on smd devices do not have the proper equipment, do not attempt this

mod! Seek professional technical assistance. The orig inator of this message is not

responsible for any loss, involving the performance of this modification.

Caution: this radio employs static sensitive devices . Suitable means should be employed to

assure a static-free environment prior to beginning work on this radio.

1. Remove power and antenna from the DR-1200

2. Remove both top and bottom covers

3. Remove vol, sq, and vfo knobs

4-20 Chapter 4

Page 176: Packet- Speed- More Speed

4. Remove vfo and mic plug nuts. Use needle nose as spanner, being carefully not to

damage the threads.

5. Carefully remove the front plastic cover. Pry the four tangs gently!

6. Remove the three small phillips screws holding the control board to the chassis. Gently

swing control board up to expose the back of the board.

7. Locate the pink wire connecting pin 6 of the mic jack on the control unit to the underside

of the main board. Desolder the pink wire from the >main< board end and resolder that end

of the pink wire to the "hot" side of the volume control. Identify the volume control by looking

at the front panel that you have removed. The hot side is the terminal on the volume control

(there are three) that is closest to the cpu chip .

8. Reassemble the radio.

NOTES: Some TNCS have a low input impedance on the RX audio line. The KPC-1 is 10

ohms, the KPC-2 is 600 ohms, but the PK-232 is 10Kohms. The lower impedance TNCS

may "kill" the receive audio when you plug them in. The cure is to remove the offending

load resistor in the TNC! If you do this, leave one end of the resistor connected for the next

owner to reinstall, if needed. If you run regular 8 ohm audio into a TNC that has had the

load resistor cut out, and don't have a speaker connected also, you may damage your

transceiver's audio output stage. Caveats aside, the benefit is that now you can turn the

audio all the way down and still get good copy . And, you can turn it up to monitor the

channel without reaching around and fumbling with a speaker plug!

73 de BILL, Golden, CO

KOZL@WOLJF.#NECO.CO.NOAM

MODIFICATION TO EXPAND TUNING RANGE 'OF THE DR·1200T TO 130·170 MHZ.

CAUTION:

If you transmit outside the designed limits of 144-148 MHz and/or the mars freqs, you may

be risking your final pa transistors! The originator of this message is not responsible for any

loss or damage or loss of use or safety if you decide to attempt this modification of your

equipment.

(Horrible that we have to put disclaimers on something like this, but there 's always one sue­

happy jerk who refuses to be responsible for himself...sigh)

This simple mod will expand the 1200 's tuning coverage and allow you to copy weather

radio traffic on 162 MHz. Be sure to follow the manual's instructions on tuning, since you

can end up with 350 and 880 MHz displayed!

Construction 4-21

Page 177: Packet- Speed- More Speed

1. Remove power and antenna from the DR-1200

2. Remove the top cover.

3. With the front of the radio facing you , you will see a loop of yellow wire on the left, just

behind the front panel. Cut this wire and tape the ends.

4. Reassemble the radio.

The radio will now tune from 130-170 MHz .

Caution:

This radio is not intended for commercial use or authorized for use in other then the

amateur, races or mars services! Don't be a bootlegger!

-73- AB5S

4-22 Chapter 4

Page 178: Packet- Speed- More Speed

GMSK Da1a Products

AX384 & AX576

High Speed Packet Radio ControllersDatalAX384JA - Provisional Information C> 1996 GMS K Data Products

.!.\)~). <]tilJh c5~"drp",ck.., ";Radio Conlrolk>r S;;l- - --- :~ ; --~'ll! . Uri Sc.1Ll1P ""'-'...- ....._ ec., Sh lllUOO tx:J) _ PIT

Feature List

• Radio port speeds of 4800, 9600, f 9200 and 38400 bls with AX384 or 7200, 14400,28800, 57600 bls with AX576

• GMSK Radio Modem operates full duplex at all the radio speeds above with nocomponent changes

· RS232 port speeds from 9600 to 38400 bls (AX384) or 115200 bls (AX576)• TAPR TNC-2 EPROM compatible including NET/ROM & 64k EPROM's such as

TheNet XU, and ROSE.· A Real Time Bit Repeater can be enabled from software. This can provide a contention

free LAN in its coverage area. An intelligent FIFO buffer is also included allowintransmissions of very long frames without bit under/over-runs. Other TNC functions areunaffected allowing simultaneous use e.g. as a Node - TheNet Xl J

• 'Set-up' software in ROM allows modem configuration to be modified on screen viasimple terminal or enhanced Windows" software program (supplied)

• Live link Bit error rate measurements enable easy setup of data links• Full Morse Ident. as per UK. license regs., regardless of TNC software fitted. Can be

disabled for use elsewhere!.· 96k ROM space allows in addition to ' Set-up and Kiss ROM', 1 x 32k EPROM, 1 x 64k

EPROM or 2 x 32k EPROM images to be fitted.· 128k RAM can be fitted in place of normal 32k if required• Radio control signals PTT & Mute can be set active high or low from Set-up software• 10 Mhz Z80 Processor ensures no lost or missed frames due to software errors.• Bi-Phase Data Coding can be selected in place of data randomizer to allow simple

interfacing to most types of voice radios (Includes FM I PM crystalled or synthesized)• RS485 Interface option allows multiple TNC 's to be connected in 'Node Stack' with

simple 4 wire cable. No more diode matrices!!

Construction 4-23

Page 179: Packet- Speed- More Speed

Description

The AX384 and AX576 are High performance packet data controllers designed for AmateurRadio users . Using the very latest in VLSI and RISC Processor technology these units set nestandards for Packet Radio TNC 's . With radio operating speeds at a maximum of 38400 blsor 57600 bls respectively the AX384 and AX576 enable packet radio users to step up fromthe 1970's technology of 1200 bls to the techniques of the 90's and beyond. The AX384 andAX576 are equally suited for use as High speed BBS ports, Network Nodes, or for the Userthat wants the very best from the packet radio network.

By maintaining software compatibility with the TAPR TNC-2 these TNC 's support the mostpopular operating software such as TAPR v1.19 , NETIROM, and WA8DED 'Host Mode ' .Even 64k EPROM's such as TheNet XU and ROSE can be fitted with only a simple linkchange. 2, 32K EPROM images can be put into I 64k EPROM and each run by simple linkselection. These ROM's can be fitted in addition to the TNC operating ROM which issupplied. Both models can be fitted with either 32k or 128k RAM as required.

To ensure rapid and error free connections with the host computer or terminal the RS232 portsspeeds can be set to 9600 , 19200, and 38400 b/s. The AX576 additionally offers 57600 and115200 b/s.

The full duplex GMSK radio modem included can provide excellent performance in the mostdemanding conditions. When used at 9600 bls it is compatible with other modem designs,however it is capable of much more. In the AX384 the radio modem can be set to 4800, 9600 ,19200 or 38400 bls without component changes. For the AX576 the speeds are 7200 , 14400,28800 and 57600 b/s . Most types of high speed TNC use a data-randomizer developed byK9NG in their modem section. The AX384 and AX576 include this randomizer as a defaultoption. However the user can also select an alternative method of data coding known as 'Bi­Phase coding'. The use of Bi-phase coding lowers the data rate possible in a channel butmakes interfacing to typical voice radios much easier. These radios can by crystal controlledsynthesized and FM or PM. When using ' Bi-phase coding' the data rates possible are 2400bls in 12.5 khz channels, or 4800 bls in 25 khz for the AX384 or 3600 bls and 7200 bls forthe AX576. In ' Bi-Phase coding ' each data bit ' l ' is represented as '00' or '11' and a ' 0' isrepresented as '01 ' or ' 10' , in this way there are no long strings of ' 1' or ' 0' sent over the airThis ensures that the data can be sent in the normal speech audio band-width withoutexcessive low frequency content.

A unique feature of the AX384 and AX576 is the inclusion of a ' Real Time Bit Repeater' .This allows the user to install a contention free' LAN in an area very simply. The operation ofthe repeater is as follows. The repeating station is equipped with an AX384 or AX576 TNCand a Full Duplex radio . This radio has split transmit and receive frequencies. The stationswishing to use the repeater use split frequency half duplex radios . This is the same situationas with a voice repeater. As soon as one of the user stations starts to transmit data the repeaterkeys its transmitter and starts to relay the input data. As soon as the other user stations detectthat the repeater is sending they are inhibited from transmitting. Thus all contention to accessthe repeater is avoided. Since the data is resent in real-time this is unlike a normal Node,where the packet must be fully received before it is forwarded to its end destination. The bitrepeater in the AX384 and AX576 includes full data bit rate clock regene ration to avoidexcessive clock jitter on the repeated data . It also includes an extending FIFO buffer to ensurebits cannot overrun or underrun where the bit rate clocks of the user stat ions and the repeaterare slightly different.

4-24 Chapter 4

Page 180: Packet- Speed- More Speed

The main operating software ROM supplied includes a 'KISS mode' driver for the TNC anda 'Set-up' program. For many users such as those running G8BPQ Node, KA9Q TCPIIP, andLinux" ax25 software this is all that is required. The 'Set-up' program allows the user tosimply configure some TNC and modem functions. The 'Set-up' program may be driven bya simple terminal program or a Windows" based program (supplied with the TNC's). Thescreen photograph below shows the simple terminal based software.

The 'Set-up' mode is activated by holding in the 'Set-up' button while switching on the TNC.It also allows various test signals to generated by the radio modem to aid the user get the bestperformance from the connected radio equipment. When used on a radio link with a AX384or AX576 at each end the users may do direct bit error rate measurements on the link to allow' tuning' of the RF equipment for best performance. The 'Set-up' program enables a on/offtone keyed Morse code identification signal to be sent every 29 minutes. This meets the needsof the UK license but can also be useful in situations of potential interference problems. Theother features which can be configured by the 'Set-up' mode are shownin the 'screen shot'above . These include the sense of the radio interface signals PTT and 'Mute' which can beactive high or low.

A 10MHz Z80 Processor is fitted as standard, This fast processor ensures there are nodropped or lost frames even at the highest operating speeds.

An optional RS485 interface allows Multiple TNC 'Node' stacks to be connected withoutdiode matrices using simple twisted pair cables . This RS485 interface can be operated at9600, 19200, 38400, and additionally on the AX576, 57600, and 115200 bls, for the verybest in multiple node stations.

As can be seen the AX384 and AX576 are 'state of the art' TNC's yet are available in bothkit and built versions. Full user documentation is included in electronic form along with aWindows" based program to control the 'Set-up' mode. The construction is from highquality materials including a RFI screen coated stylish grey casing. The styling is designed tomatch well with modem office and computer equipment. Finally the PSU is extensivelyfiltered and smoothed to ensure correct operation of the TNC even in the presence of strongRF fields . The AX384 and AX576 are 'future proof' by design and construction!

Construction 4-25

Page 181: Packet- Speed- More Speed

Operating Specifications

Power Supply:

Current Consumption:

Radio Bit Rates: AX384

AX576

+9 to + 15 volts via power connector

200mA (Typical maximum).

4800, 9600, 19200, 38400 b/s

7.200, 14400,28800,57600 b/s

Radio Interface: Receive audio,

Transmit audio,

Press to Talk (PTT) .

Mute Input.

Transmit audio output impedance 10k Ohms (max) .

Transmit audio output level 0 Volts (min .) to 5 Volts (max) .

Receive audio input level 50 mVolts (min.) to 5 Volts (max)

Receive audio input impedance lOOk Ohms (min.).

Computer Interface

RS232C interface with the following operating speeds

AX384 9600, 19200, 38400 b/s

AX576 9600, 19200, 57600, 115200 b/s

RS485 interface also capable of working at the above speeds (optional)

Further Information

To find out more on these TNC products or other ax25 Packet Radio products, or to place anorder please contact OMSK Data Products .

Thank you very much for your interest in the AX384 and AX576. Please watch out for furtherinformation on new OMSK Data Products developments for ax25 Amateur Packet Radio

OMSK Data Products80 Colne Road,Halstead,EssexC092HPEngland

4-26 Chapter 4

Fax: +44-(0)-1787472290Email : [email protected]

Page 182: Packet- Speed- More Speed

W31WI/TAPR TAC-2(Totally Accurate Clock) Project

March, 1997

TAPR is actively working with Tom Clark, W3IWI, in developing the TAC-2 kit. The TAC-2 isthe follow-on to Tom's original TAC project. The goal is to have a finished kit complete withparts and documentation in second quarter of 1997. Information on the TAC-2 project'sprogression will be documented here and on the TAPR-BB Iistserver.

Project Status

March 15th, 1997

Testing continues and a final production board is nearlycompleted. Documentation is being completed and will bereviewed by between 3 and 5 people. More information will bepresented when the final production kit cost and availability isknow. The kit is looking very good. The pictures to the right are(top - component side of the TAC-2) and (underneath - Side view '----'-==

of the TAC-2 with Motorola Oncore-VP mountedunderneath).

January 1997

The project team has the rev-b boards in hand and are building.When docs are written and testing is completed with this round, we willbe asking for Beta testers. Details on how to particpate in thebeta-testing will be announced in the Winter TAPR PSR (PacketStatus Register) and on the TAPR-BB e-mail list. It is expected that

Construction 4-27

Page 183: Packet- Speed- More Speed

a small group of 5-10 people will be asked to particpate in thebeta-test ing.

November 1996

The project team is reviewing the design of the second alpha board. The secondalpha-board will be put into the board shop (3-4 week turn) once everyone feels thatit is ready . It is not expected that a beta run will be done until the Spring .

October 1996

The first alpha boards arrive. Alpha testers installed componentsand found that they work ! Documentation begins . Project teams decidesto add functionality and addit ional components to board. This requires achange in the board layout and a second alpha board run.

September 1996

Purchase of the alpha PCB authorized by the TAPR BoD.

Totally Accurate Clock (TAC)

Excerpts from TOTALLY ACCURATE CLOCK ANNOUNCEMENT, Tom Clark, NASAlGSFC(February 2, 1995)

The "TAC" name is supposed to invoke a smile on your face . Many of you rememberHeathkit's "Most Accurate Clock" (a \IIJ\I\N receiver) and I see advertisements for VLF clocks(\llJ\l\NB in USA, DCF77 in Europe) that still use the "Most Accurate Clock" name in theiradvert ising. Since the "TAC" is 3-4 orders of magnitude better than the "Most Accurate Clock"units, the "Totally" name seems warranted (also, TAC are my initials and this was begun as ahome project!).

The TAC project began when I was on sabbatical at Onsala when Bernt Ronnang got me anearly prototype of a Motoro la PVT-6 OEM GPS receiver . In that incarnation, the PVT-6 waspretty disappointing. When I got back home, I had Motorola update the internal firmware andfound that its personality had changed completely -- it was now very precise, but it had about a500 nsec bias. I contacted a friend at Motorola who was involved in the PVT-6 software andhe told me that tests at USNO had uncovered the same error. I was added to the "beta" group,got my initial prototype updated with the latest firmware and began more detailed testing .What I then found was that the PVT-6 receiver had the best timing performance I have seen inany small GPS receiver. With a small amount of care in setting it up, it now gives 50 nsec orbetter RMS timing precision and biases appear to be < 20 nsec.

The TAC project now involves both hardware and software. Let me briefly describe both toyou .

HARDWARE:

The core of the TAC consists of a GPS. Several are going to be supported in the TAC-2design (Motorola ONCORE, Garmin GPS-20, Trimble SK-8) The circuit board allows thesevarious GPS to be mounted to the board .

The TAC-2 adds a number of desirable features:

4-28 Chapter 4

Page 184: Packet- Speed- More Speed

~j The 1 .~ ~S output signals are buffered through a 74AC04 gate to improve the drivecapabi lities and to act as a "fuse" to prevent damage to the receiver in case of anoperator goof. Up to three independent buffered outputs are provided, and the buffers willdrive about +2 to +2.5 volts into a 50 ohm termination. The normal logic polarity is positivegoing at the epoch time, but this can be inverted if desired by some simple jumpers.

.: Up to three open collector 1PPS signals are also available, negative going . Normallythese would be used to drive display LEDs, but they can be used for other purposes.

.: The add-on board includes an RS232 driver that provides 1 PPS time synchron ization toan attached computer. The 1 PPS signal is normally connected to the computer's DCDinput.

::-.: The RS232 I/O to the computer is buffered and isolated from the receiver to act as a"fuse" to prevent damage to the expensive receiver in case of an operator goof. AnRS232 OR-gate is supplied for the receiver input signal to allow RTCM SC104 Differentia lGPS signals to be fed to the GPS receiver .

SOFTWARE:

A program called SHOWTIME displays the current time in BIG DIGITS you can see fromacross the room. In addition to just showing the UTC time, it includes a display of the date,day-of-week, day-of-year, local and Greenwich Mean Sidereal times, JD and MJD, and eventhe current GPS week. You can enable audible "\N\NV-like" time ticks to assist you in settingthe formatter (or your wrist-watch). You can have the software automatically reset the PC'sinternal clock with about 25 msec accuracy. All the time display updates and audible tickshappen synchronously with the GPS 1 PPS signal because the PC reads the tick on its DCDline.

SHOWTIME allows you enter timing offsets and handles all the arithmetic for you. It allows youto make easy corrections for time delays in cables and the instrument and it tells you (with 1nsec resolut ion) the actual epoch of the 1PPS tick and it gives you an estimate of the accuracyof the tick. SHOWTIME gives you a nice display of which satellites you using and whichsatellites are above the horizon . This includes a bar-graph "S- meter" for each of the GPSsatellites currently in lock which are updated once per second.

The software lets you change operating modes (timing vs. position, elevation masks, sate lliteselection criteria , etc) easily and when you are running in position-determination mode, it willhandle all the position averaging tasks for you. At any time, you can save the currentconfiguration (positions, timing offsets, receiver modes, etc) to a disk file and restore thatconfiguration at a later time.

At this time SHOWTIME runs stand-al one on a separate MS-DOS PC (but it does seem to runOK in the DesqView multitasker). Once you have set parameters into the GPS receiver, thePC operations can be terminated.

Construction 4-29

Page 185: Packet- Speed- More Speed

Visit Tom's TAC site

at: ftp://aleph .gsfc.nasa .gov/GPS/totally.accurate.clock

or

ftp://bootes.gsfc.nasa .gov/GPS/totally.accurate.c1ock/

Tucson Amateur Packet Radio8987-309 E Tanque Verde Rd #337 * Tucson, Az * 85732Office (817) 383-0000 * Fax (817) 566-2544Internet e-mail: tap [email protected]

-_.- - -_ ....- ....... ~--- -~--

Index New Find Home

.~~~~~~~~- _._ __._ __ _.._-_ .- _._-------_.__.__.

__ people have accessed this page since November 20th, 1996.

Greg Jones, WD5IVD, [email protected] (www)

(This page last modified on ;Har<:ii13. 1997)

Copyright 1996, Tucson Amateur Packet Radio Corp .All subsequent TAPR pages are copyright 1996, Tucson Amateur Packet Radio Corp.

4-30 Chapter 4

Page 186: Packet- Speed- More Speed

Emulat ing the "To t ally Accurate Cloc k "Tom Cla rk, W3IWI

Ap r i l 14 , 1 99 6

-= -=-=-=- =-=-= -=-=-=-=-=-=- =- = -=-=-=-=-=-=-= -= -= -= -= -= - = -= -

A number o f peopl e have e xpressed an i n t e re s t i n making clones o f my "TotallyAccu ra t e Cl o c k " (TAC) . This note wi l l documen t ways yo u can fabricat e as i mple , bu t funct ional vers i on of the TAC wi t h little more than a n e xtra wirein the cable.

HARDWARE DESCRIPTION:Th e core o f the TAC is a comme r c ia l OEM GPS receiver board . The boards thatare supported are

1 . Mot orola ONCORE BASIC and BASI C EVAL models2. Mo t o rola ONCORE VP and VP EVAL3. Motorola ONCORE XT4 . Garmi n GPS- 20 (and probably GPS- 2 5 , but i t hasn't been t e s t e d )

To f unct i o n as a TAC, I requi re tha t the r e c e ive r has

a . Comp u t e r serial I /O - at least with NMEA messages a t 4 8 00 baud (and forthe ONCORE receivers , Motorola 's p r opr ie tary binary protocol at 9600ba ud ) .

b . A 1 Pu lse per s e c ond ou t pu t nominall y synch ron i z e d with the UTC second .For the Motorola receivers, this means tha t e ither Option A (1 PPSOu t p u t ) or Option I (RAI M + 1PPS ) mus t be ins t a l l e d ; the EVALdeveloper's kits have ~hese options standard . The Garmin GPS- 20 receiverhas 1 PPS as standard feature . The Moto ro l a 1 PPS signal is 0. 2 secondsin duration and the Garmin 1 PPS i s 0 . 1 sec onds . Both h a v e "TTL" level1 PPS signal s which rise zero t o -+5 vo l t s a t the nominal timing epoch .

The TAC proj e ct uses stock receivers with one important addition - the 1 PPSsig na l i s a vailable for prec ise epo ch t i mi ng by the user (usual ly on a BNCconnec t o r) and it is also sent to t he at tac he d comp u t e r. I have adopted theconv e n t i o n that the 1 PPS signal to the compu t er wi l l be on the RS232 DCDhandshaking line . The TAC-to -computer interface r equires 4 wires (preferablyshielded) :

FUNCTION

GPS TXD f r om GPSGPS RXD to GPS1PPS from GPS (DCD)Si g n a l GroundCable Shield

RS23 2 wi t h25 -pin conn e c t o r

pin #3pin #2pin #8pin #7pin #1

RS232 withwith 9 -pin connector

pin #2pin #3pin #1pin # 5pin #5

The GPS receivers also require t h e user to provide e xternal DC power :

* The Motorola ONCORE BASIC and XT (a nd the VP EVAL models in a me t al box)al~ have DC-to -DC converters and power regulators so that the user cansupply any s upply voltage between +10VDC and about +3 0VDC. In this noteI wi l l ca l l this +1 2VDC .

* The ONCORE VP and Garmin receivers r equire the user to provide regulated+5VDC power and will be damage d if t he wr o ng vo l t a g e is applied. Thiscan b e easily generated b y a 7805 - type regulator chip .

Al l the GPS receivers have a multi-pin connec tor to supply power , prov idecompu ter I /O connections, and delive r t he 1 PPS signal to the user. TheMo torola BASIC and VP receivers use a 2x5 10-pin connector with pins on 0 .1"c e n t ers ; the user is wa r n e d that the BASIC and VP series of receivers usedifferent connections and a receive r ma y be damaged if wiring for the wrongr ece ive r type is used! The Mo t oro l a XT and VP EVAL receivers are housed in anextruded a luminum box and has a DB- 9 conn e c t o r . The Garmin receivers use asing l e - r ow 12-pin connector .

Construction 4-31

Page 187: Packet- Speed- More Speed

In a dd i t i o n to the power+I /O connec t or, all the receivers ha v e a coaxialcable connector for an external antenna . +5VDC powe r i s supplied on thisconnector fo r an antenna mounted RF preamplifier. Al l the receivers use asmall MCX - s er i e s connectors, e xcept fo r the metal- cased Motoro l a XT/VP EVALmodels which us e a BNC connector .

Al l the rec e ivers excep t for the Mo torol a VP ha v e RS232 vo l t a g e levels fo rtheir TXD /RXD s erial I /O . The RS232 signals 'idle' a t a vo l tage between -3and -10 vo l t s. They go to a vo l t a g e between +3 and +10 vo l t s when a data"on e " i s sen t. The Mo t o r o l a VP uses ' inve r ted TTL' l e vel s, i d ling a t +5V andgo ing t o zero wi t h data. Thus normal c~mputer RS2 32 vo l tages must be invertedand l e vel shifted if you a re using a VP- s e r i e s board -l e vel receiver.

Mo torola s ells a developer's v e r s i o n of the VP rece i ve r c a ll e d the VP EVAL .The VP EVAL ve r s i o n of the VP has an internal c ircuit board t ha t provides theRS2 32 l e vel c o nv e r s ion and a +12 -to -+ 5 vo l t DC- t o -DC power conver t er . Thes a me mechanical package is s o l d by Mo torola as the XT model , e xcept that theXT apparently uses the older ONCORE BASIC board.

- = - = - = - = - = - = - = - = - = - = - = - = - = - =- = - = - = - = - = - = - = - = - =- =- =- =-= - = - = -

SYNCHRONIZ ING THE COMPUTER:The receivers all transmit time, date a nd position data to a n attached computer v ia the RS232 link . For simplicity I have c hosen t o use the NMEA-018 3(National Marine Electronics As s o c i a t ion ) standar di zed message s at 4800 baud .Al l GPS r ece ive r s have to develop precise timing i n order t o perform theirnavigation t a s ks . Each second they perform a lot o f compu t a t i on s and thenoutput the t i me and navigation results to the user. Howe ver , the user reportis the l owest p riority task , s o the time -tag containe d i n the ASCI I text inthe NMEA message lags true UTC (Universal Time Coordinated ) time by a fractionof a second . This has the effect of the voice announcement on WWV/CHU/JJYrad i o being like "At the tone the time WAS xx : xx: xx" . So me u s e r s o f GPSreceivers have tried to determine (a real kludge, i n my op i n i o n ! ) ad hoc"latenes s" errors so that they could obtain more accu ra te t i me .

I c hose to adopt a more rigorous s cheme to achieve c ompu t e r s ynchroni zation .I determine d that the timing o f the 1 PPS signal gene ra t e d on the receiverbo a r ds was very g ood -- f or the Mo torol a receivers I can obtain 20 - 5 0 nano­second (nsec) accuracy and precision if I set up the r ece i ver properly, whichwas the g e nes i s of the TAC project . Th e Garmin GPS- 20 i s l e s s a c c u r a t e with-1 mic r o s eco n d (usec) performance -- still v e ry i mpr essiv e .

To achieve c omp u t e r synchronization, I decided to s e nd the ha rdware -generated1 PPS signal to the computer using the Carrier Detec t (DCD) RS232 handshakingwire . In software I would get the date/time from the NMEA test message,increment to the next second, and then wait for the DCD hardware signal toarrive . As the 1 PPS signal is received , the software app l ies the time thatwas calculated as appropriate to the next second . Thus the computer 's time isaccurate to the level of the latency of the recognition of the DCD hardware 1PPS signal.

Although the RS232 specifications call for signal l e v e l s of -3 to -10v (low)and +3 t o +10v (high), I have found that most mode rn PCs work reliably whendriven wi t h ' TTL level' signals (Ov and +3 to +5v) .

THE ' REAL' TAC DESIGN :My initial TAC hardware design was designed around the or i g i na l MotorolaONCORE BAS I C (t h e n called PVT-6 ) rece ive r . I desig ned a small circui t boardto provide a n umber of interface tasks :

* Isolate d , fast r ise-time buffers t o p rov ide clean 1 PPS s i gnal s t o theuse r

* RS2 32 line d r i v er for the DCD 1PPS signal* RS232 i s olators to protect the rec e iver f r om user e r rors ('idi o t fuses ')* An L- b a nd preamp l i f ier to imp rov e rece iver RF p erformanc e (a n d provide an

' RF id i o t fu s e')

4-32 Chapter 4

Page 188: Packet- Speed- More Speed

* The abil ity to re -bias the antenna-r,ounted preamplifier wi t h a v o l t a g eo the r than +SV

* Battery backup for the receiver's time-of-day clock and BBRAM

At NASA/G SFC we produc e d more than 50 such TACs wh ich a r e i n daily u s e atgeodetic a nd a s t ronomica l obs e r va tories a round the wo r l d. Thei r pe r forma ncei s documen ted i n the man y data f iles on my ftp ser ver a t the URL

ftp : / /al e p h . gs fc . nasa .gov/ GPS/ t o ta l ly. a c c u rate . c loc k /

On a l e p h yo u wi l l f i n d a b lock diagram of the TACs we produced In the fil e

ftp : / /a leph .gsfc .nasa .gov/GPS /tota l l y .accurate .clock/tac -bl o k . gi f

a nd yo u wi l l find a photo of the TAC at Onsala Space Observator y (Swe den) in

f t p :/ / a l e p h.gs fc . n a s a . gov/GPS/ t o t a l ly . a c c u r a t e . c l o c k / tac -foto .g i f

The original Motorola ONCORE BASIC receiver used in the TACs is sti llava i l a b l e , but Motorola is phasing it out for the lower cost and hig he rp erformance (with 8 ins tead o f the original 6 channels) ONCORE VP .

In addition to the Motoro l a receivers, I have developed the first p ro to t ypeo f the Ga rmin-based "TAC Lite" as a lower cost alternative for t he radioa mateur community . I a m now developing a new TAC interface board tha t wi l lsupport any of these r ece ive r s , but it is not y e t a vailable . It wi l l probablybe made ava i l a b l e thr u the Tu cson Amateur Packet Radio (TAPR) a mateu r R&Dgroup . TAPR has already announced the avai lability of the Garmin GPS- 2 0rece i vers for the TAC Lite design . Information on the TAPR-relatedde velopments can be f ound on TAPR's WWW page at URL:

http : / /www.tapr.org /

My a leph file server also serves as the repository for TAC support so ftwa reand documentation . In parti cular, you may want to fetch the mo s t currentv e r s i on of my SHOWTIME cont ro l le r/d i s p l ay software and its do c umen t a t i o n .With v e r s i o n 3 . 10 o f SHOWTI ME i s included minimalis t support f o r t he GarminGPS -20 'TAC Lite' .

TAC HARDWARE EMULATION:Until the new TAC interface boards are a vailable , I have drawn up s ome simpl e"TAC Emulation" sketches for each of the rece ivers is included in t h i se mu l a t e . zip distribution - the files are emulate.ps (in Postscript ) andemulate .gif (in .gif format). The following d iscussion refers t o t he s eschematics .

Al l 4 v e r s i o n s shown in emulate.ps and emulate .gif have in common

* They assume yo u have +1 2VDC power* The RS232 connections shown assume a standard IBM-PC 9-pin connec t o r* They a l l l ack a 'proper' RS232 driver for the DCD 1 PPS connection and

assume y o u r PC wi l l wo r k wi t h TTL levels.* They show a 1 PPS BNC conn e c t o r for yo u r use in parallel with the DCD

s i g n a l . The receiver 's 1PPS output drive is limited , so the ri s e time sare not a s fast as I 'd like to see .

An optional battery i s shown . I highly recommend that y ou use a ba t t ery tohelp the r e ce ive r ' wake up' in a 'smart' mode. Some receiver boa r ds a l r eadyhave a bat te r install e d; if thi s i s the case for you, then omit t he optionb a ttery , diode and res i sto r. Ea ch of the receivers s eems to wo r k OK withbattery vol t a g e s from +3 .5v t o +9v. The NASA/GSFC p r oduced TACs u se a 9v Ni Cdr e char gable batter y a n d I have also used disposable 9v alkalin e batte rie swi t hou t t he r es i s t o r /diode c ha r g e r . If y our receiver has a battery buil t -in ,it is p robably a 3.6v unit.

I f you need to pro v ide a bat te ry , y ou could use a 3.6v cordless t e leph onebattery o r a 9v NiCd battery (either obtained at Radio Shack) . You wi l l ne ed

Construction 4-33

Page 189: Packet- Speed- More Speed

a Silicon diode like a lN4001 to prevent the batter from d ischarging whenpower is off.

You now n e e d to compute the proper c harge current limi t i ng resistor . A fu llycharged NiCd cell ends up a t a vo l t a g e o f 1.4v. A " 3 . 6v ba t t e ry " has 3 c e l l s ,so its float vo l t a g e is 4 .2 v . A " 9v battery" has 6 c e l l s , so i t fl oats at8 .4v. The diode wi l l have a drop of -0. 6v across it . You wi l l want tot r i c k l e - c ha r g e the battery wi t h 5 t o 10 rna . So for a " 9v" battery beingc h a r g e d from 13 .8v (not an untypical "12 vo l t " supply) , the s erie s bat t erycharge current limiting resistor wou l d be :

R [ (1 3 . 8v supply) (8 .4v battery) - (O.6v diode)) / (0 . 005 t o 0 .01 a mps )4 80 to 960 ohms ( I ' d use 68 0 ohms )

or f o r a " 3. 6v" battery c ha r g i ng f rom +5V (ou t o f the 78 05 r e g u lator) :

R [ (5. 0v supp l y) (4 .2v battery)4 0 to 80 ohms (I'd u s e 68 o hms )

(O.6v diode)) / (0 . 005 t o 0 .0 1 amps )

The most complicated of the e mulator circuits is that for the ONCORE VP,since the RS2 32 logic levels must be inverted . I n the new TAC interfa c e boa r d ,I p lan to use a MAX232 /LT l1 81 RS232 leve l converter chip . Since the MAX 2 32 /LTl181 has t wo TX and t wo RX l e vel converters , if y ou us e it, y ou migh t aswe l l buffer the DCD lPPS l ine also. You could probably use a CMOS inve rterchip (like a 74HC04 ) a l s o .

Tom Clark

4-34 Chapter 4

Page 190: Packet- Speed- More Speed

ARRL MEMBERS

This proof of purchase

may be used as a $1.50

cred it on your next ARRL

purchase. Limit one

coupon per new membership, renewal or

publication ordered from ARRL Headquarters.

No other coupon may be used with this

coupo n. Validate by entering your member­

ship number from your QST label below:

Page 191: Packet- Speed- More Speed

PACKETSPEED

PROOF OFPURCHASE