Jan/Feb 2000 1

R

L

R

A

David Sumner, K1ZZPublisher

Doug Smith, KF6DXEditor

Robert Schetgen, KU7GManaging Editor

Lori WeinbergAssistant Editor

Zack Lau, W1VTContributing Editor

Production DepartmentMark J. Wilson, K1ROPublications Manager

Michelle Bloom, WB1ENTProduction Supervisor

Sue FaganGraphic Design Supervisor

David Pingree, N1NASTechnical Illustrator

Joe SheaProduction Assistant

Advertising Information Contact:John Bee, N1GNV, Advertising Manager

860-594-0207 direct860-594-0200 ARRL860-594-0259 fax

Circulation DepartmentDebra Jahnke, ManagerKathy Capodicasa, N1GZO, Deputy ManagerCathy Stepina, QEX Circulation

Offices225 Main St, Newington, CT 06111-1494 USATelephone: 860-594-0200Telex: 650215-5052 MCIFax: 860-594-0259 (24 hour direct line)e-mail:qex@arrl.org

Subscription rate for 6 issues:In the US: ARRL Member $22,nonmember $34;US, Canada and Mexico by First Class Mail:ARRL member $35, nonmember $47;Elsewhere by Surface Mail (4-8 week delivery):ARRL member $27,nonmember $39;Elsewhere by Airmail: ARRL member $55,nonmember $67.

Members are asked to include their membershipcontrol number or a label from their QST wrapperwhen applying.

Jan/Feb 2000 QEX Advertising IndexAlmost All Digital Electronics: 64American Radio Relay League: Cov III, Cov IVCommunications Specialists, Inc: 15

Crestone Technical Books: Cov IINo Walls Inc: 56Tucson Amateur Packet Radio Corp: 46Z Domain Technologies, Inc: 22

About the CoverMaking RF from 2 to 250-MHzSee Steve Hageman’sarticle on p 3

Features3 Build this 250-MHz Synthesized Signal Source

By Steve Hageman

16 Effects of Boom and Element Diameters onYagi Element Lengths at 144, 432 and 1296 MHzBy Guy Fletcher, VK2KU

23 A Scanner Controller for Varactor-Tuned ReceiversBy Thomas K. Duncan, KG4CUY

31 ARRL Technical Awards

32 Automotive RFI EliminationBy Stuart G. Downs, P.E., WA6PDP

37 A Homebrew Logic AnalyzerBy Larry Cicchinelli, K3PTO

47 A High-Performance Homebrew Transceiver: Part 4By Mark Mandelkern, K5AM

Columns57 RF By Zack Lau, W1VT

62 Letters to the Editor

61 Next Issue in QEX

In order to ensure prompt delivery, we ask thatyou periodically check the address informationon your mailing label. If you find any inaccura-cies, please contact the Circulation Departmentimmediately. Thank you for your assistance.

QEX (ISSN: 0886-8093) is published bimonthlyin January, March, May, July, September, andNovember by the American Radio RelayLeague, 225 Main Street, Newington CT 06111-1494. Yearly subscription rate to ARRL mem-bers is $22; nonmembers $34. Other rates arelisted below. Periodicals postage paid atHartford, CT and at additional mailing offices.POSTMASTER: Form 3579 requested.Send address changes to: QEX, 225 Main St,Newington, CT 06111-1494Issue No 198

Copyright ©1999 by the American Radio RelayLeague Inc. For permission to quote or reprintmaterial from QEX or any ARRL publication, senda written request including the issue date (or booktitle), article, page numbers and a description ofwhere you intend to use the reprinted material.Send the request to the office of the PublicationsManager (permission@arrl.org)

2 Jan/Feb 2000

THE AMERICAN RADIORELAY LEAGUEThe American Radio Relay League, Inc, is anoncommercial association of radio amateurs,organized for the promotion of interests in AmateurRadio communication and experimentation, forthe establishment of networks to providecommunications in the event of disasters or otheremergencies, for the advancement of radio artand of the public welfare, for the representationof the radio amateur in legislative matters, andfor the maintenance of fraternalism and a highstandard of conduct.

ARRL is an incorporated association withoutcapital stock chartered under the laws of thestate of Connecticut, and is an exempt organiza-tion under Section 501(c)(3) of the InternalRevenue Code of 1986. Its affairs are governedby a Board of Directors, whose voting membersare elected every two years by the generalmembership. The officers are elected orappointed by the Directors. The League isnoncommercial, and no one who could gainfinancially from the shaping of its affairs iseligible for membership on its Board.

“Of, by, and for the radio amateur, ”ARRLnumbers within its ranks the vast majority ofactive amateurs in the nation and has a proudhistory of achievement as the standard-bearer inamateur affairs.

A bona fide interest in Amateur Radio is theonly essential qualification of membership; anAmateur Radio license is not a prerequisite,although full voting membership is granted onlyto licensed amateurs in the US.

Membership inquiries and general corres-pondence should be addressed to theadministrative headquarters at 225 Main Street,Newington, CT 06111 USA.

Telephone: 860-594-0200Telex: 650215-5052 MCIMCIMAIL (electronic mail system) ID: 215-5052FAX: 860-594-0259 (24-hour direct line)

Officers

President: RODNEY STAFFORD, W6ROD5155 Shadow Estates, San Jose, CA 95135

Executive Vice President: DAVID SUMNER, K1ZZ

The purpose of QEX is to:

1) provide a medium for the exchange of ideasand information among Amateur Radio experi-menters,

2) document advanced technical work in theAmateur Radio field, and

3) support efforts to advance the state of theAmateur Radio art.

All correspondence concerning QEX should beaddressed to the American Radio Relay League,225 Main Street, Newington, CT 06111 USA.Envelopes containing manuscripts and letters forpublication in QEX should be marked Editor, QEX.

Both theoretical and practical technical articlesare welcomed. Manuscripts should be submittedon IBM or Mac format 3.5-inch diskette in word-processor format, if possible. We can redraw anyfigures as long as their content is clear. Photosshould be glossy, color or black-and-white printsof at least the size they are to appear in QEX.Further information for authors can be found onthe Web at www.arrl.org/qex/ or by e-mail toqex@arrl.org.

Any opinions expressed in QEX are those ofthe authors, not necessarily those of the Editor orthe League. While we strive to ensure all materialis technically correct, authors are expected todefend their own assertions. Products mentionedare included for your information only; noendorsement is implied. Readers are cautioned toverify the availability of products before sendingmoney to vendors.

R

L

R

A Empirically SpeakingBy the end of this year, the press will

have raised more hullabaloo over thenew millennium than over all othertopics combined. For experimenters,2000 marks the passage of AmateurRadio into a period of unprecedentedchallenge and change.

Soon, we will celebrate with our 200thissue! Since I have a good excuse, I’mactively seeking updates on older ar-ticles. You authors may like this chanceto give us a few paragraphs on what’schanged in your designs, in your think-ing or in your lives. I know work contin-ues on many of your projects afterpublication. Too often, we don’t get tosee the subsequent refinements.

You may have ideas about how toimprove circuits or software that yousaw in QEX. Perhaps you don’t thinkyour idea justifies an entire article orletter to the editor. Maybe you havean update on the availability of aproduct, an overdue correction or justa note to say “Gee, look how thingshave changed!” I think we shouldmake space for this kind of thing, es-pecially because it gives us a chanceto look back at where we’ve been.

Now is the time to ask “Where elseare we going with QEX?” Well, we’veseen numerous comments that youlike what you’re seeing thus far. Yourthoughtful letters have sustainedsome interesting discussion. The flowof articles is up and we are coveringmany important themes in communi-cations. We want to broaden ourperspective by including material rep-resenting a growing range of interestsand skill levels.

Technology’s accelerating pace,however, forces experimenters tospend more time researching possibili-ties for designs, which leaves less timeto write about what we actually built.Scarcity of good published informa-tion, in turn, makes it tougher for thenext person to do his or her research.Major update cycles for referenceworks, such as The ARRL Handbook,are rapidly getting shorter.

Since QEX articles are usuallypublished well within a year of theirsubmission, we fill gaps by providingtimely exposure for projects, some ofwhich may migrate to the Handbook.We look forward to helping documentsome exciting new applications as wehead into a new era.

In This QEX

Our examination of frequency syn-thesis techniques continues with SteveHageman’s 250-MHz generator. Theavailability of high-performance cir-cuit building blocks has put this typeof project within reach of amateurs.Steve employs computer control forflexibility and simplicity. Purity andstability issues are addressed in thedesign.

Guy Fletcher, VK2KU, has per-formed some fastidious measure-ments on VHF, UHF and microwaveYagis with some interesting results.He shows how the data support hisformulation of “Guy’s Rule” for therelation between boom and elementdiameters and element length. Themodel removes some of the surprisesand much of the time involved inaccurately designing antennas inthose frequency ranges. I suspect thetheory behind it also comes into playmore often than not for other types ofantenna structures. Thanks to ourfriends at WIA for their assistance.

Thomas Duncan, KG4CUY, showshow to build frequency-control func-tions around a VCO—almost any VCO.He illustrates important PLL and con-trol-system features.

Stu Downs, WA6PDP, went to greatlengths to eliminate all his automotiveRFI on the HF bands. Armed with alaudable understanding of conductedand radiated emissions, he set aboutisolating and curing his problems, oneby one, until his receiver was quiet.His revelations should help frustratedhams and automobile manufacturersalike.

As opposed to RF, digital circuits tendto behave in binary fashion: They eitherwork or they don’t! Troubleshootingthem can involve observing a myriad ofhigh-speed signals. Larry Cicchinelli,K3PTO, brings us a tool of significancein an increasingly digital world: aPC-driven logic analyzer. Again, the per-sonal computer proves itself a matchlesscontrol system.

We get down to baseband with MarkMandelkern, K5AM’s home-brew trans-ceiver. Mark shows special interest innoise limiting in this penultimate seg-ment of his series. In RF, Zack Lau,W1VT, explores the use of ladder line atUHF—73, Doug Smith, KF6DX, kf6dx@arrl.org

Jan/Feb 2000 3

You can build this high-quality signal generator.PC control and prefab parts make it relatively easy tobuild and operate. An off-the-shelf doubler extendsits range (fourth harmonic) to –16 dBm at 1 GHz!

By Steve Hageman

9532 Camelot DrWindsor, CA 95492shageman@sonic.net

Build this 250-MHzSynthesized Signal Source

1Notes appear on page 14.

Ifind it extremely useful to havea general-purpose RF sourceavailable when I want to quickly

breadboard a receiver. With a stablesource, I can first concentrate on a newreceiver’s RF and IF portions and usethe source as a temporary LO. Second-ly, when the receiver’s LO is finished,the source can be used as the RF inputto verify performance. Finally, as ageneral-purpose troubleshooting tool,it sure is nice to have any frequencyand amplitude you want right at yourfingertips, when you want it.

The “wireless revolution” comes toour aid in this project. All the circuitryused in this project is the result of

manufacturers’ pushing the state ofthe art to provide highly integratedsolutions for the wireless industry. It’sa great time to be interested in radioand homebrewing!

How the Source WorksThe CW signal generator, as I have

designed it, (see Fig 1) is one of a classof synthesizers employing “coherentdirect synthesis.”1 Coherent means thatall the frequencies are derived from onemaster oscillator—the 40-MHz clock onthe Numerically Controlled Oscillator(NCO) board in this design. Thesynthesizer is direct because it usesfrequencies generated by PLL circuitsin a single mixer to generate the output.

Coherent synthesis is advantageous

because it tends to eliminate frequency-drift problems. In fact, if the referenceoscillator in this synthesizer drifts, theoutput frequency drifts in exactly thesame direction and at the same rate.When multiple frequency referencesare used, this may not be the case.2

The synthesized source describedhere uses many excellent ideaspresented previously.3, 4 ,5 These arecombined to make a 2-250 MHz,programmable VHF source.

In Rhode’s 1994 article (seeReference 2) he described using anNCO as a “micro-adjustable” frequen-cy reference to a PLL circuit. Thistopology has the advantage of allow-ing the PLL to operate at a highreference frequency, thus allowing ahigh loop bandwidth (BW), whilehaving sub-hertz channel spacings.

4 Jan/Feb 2000

Fig 1 (see right)—As this block diagramof the synthesized source shows, thedesign is partitioned into five functionalblocks (the PLLs share the same circuitdesign). Each of the RF blocks ishoused in a die-cast metal enclosure tosuppress cross-talk between modulesand contain any RF radiation.

While Reference 1 describes manyother methods of achieving smallchannel spacing, I have adopted thisapproach as the simplest way ofachieving the disparate design goals ofa high reference frequency and smallstep size.

Looking at the block diagram shownin Fig 1, the NCO board (also Fig 2)contains the 40-MHz master oscillator,the NCO circuit and a divide-by-fourcircuit that generates a separate10-MHz reference. The 32-bit NCO, aHarris HSP45102 device, yields a stepsize of:

d Hzf . = • =40 10

20 009

6

32 (Eq 1)

The output of the NCO is set for10.7 MHz. As per Rhode’s assessmentof the design problem, using a commonIF allows the use of off-the-shelf, low-cost filters. I chose a ceramic filterbecause it provides more than ade-quate sideband suppression and itsimpedance is such that I could matchit with simple resistors while keepingplenty of signal level to drive the PLL’sreference input.

I also considered 455 kHz. This wouldhave had the advantage of lower distor-tion from the NCO, but the availableceramic filters need to be matched inthe 1 to 3-kΩ range. This would havecomplicated the matching circuitry tokeep the losses at an acceptable level.

The filtered NCO output drives thePLL 0 board (Fig 3). This PLL designuses the readily available, highlyintegrated Motorola MC145191. TheMC145191 is a low-power, 1-GHz PLLthat contains an internal referencedivider, N + A feedback divider andcharge-pump phase detector. Thereference, or “R” divider is used to setthe PLL’s channel spacing. Here, the10.7 MHz of the NCO is divided by 107to get a reference frequency of100 kHz. When the value of the N + Acounter is changed by one, the outputof the VCO steps by 100 kHz. So theexact NCO frequency is used tointerpolate between the PLL’s basic100-kHz channel spacing.

Mathematically, this works out to thetotal “frequency gain” of the circuit(750 MHz / 10.7 MHz = 70). For a stepsize of 1 Hz, the NCO must have afrequency resolution of no greater than1 Hz / 70 = 0.014 Hz. As shown above,this requirement is easily met by theNCO circuit used.

The output of PLL 0 is an adjustable500 to 748-MHz signal that is used bythe mixer board to synthesize thesource’s 2 to 250-MHz output. PLL 1 issimilar to PLL 0. In fact, it uses the

same PC board and parts. PLL 1 pro-duces a fixed frequency of 750 MHz,however. The input to PLL 1 is fixed at10 MHz (divided from the 40-MHzcrystal oscillator) and the R divider isset to 100 to yield a 100-kHz referencefrequency to the phase detector (sameas PLL 0).

The PLL outputs are used as inputsto the Mixer board. The Mixer boardlooks much like a standard down-converting receiver because that’sessentially what it is.

The 750-MHz PLL 1 signal is used asfixed-frequency local oscillator (LO)drive to a mixer, while the 500 to748-MHz PLL 0 signal is used as the RFinput. Both PLLs are low-pass filteredto remove any harmonics. As we will seelater, this greatly improves the spur-ious performance of the synthesizeroverall.

The filtered PLL signals are routedto a modular doubly balanced mixer(Fig 4). The desired output from themixer is the difference frequency of 2to 250 MHz. To ensure that the mixeroperates with low intermodulationdistortion, its IF output is broadband-terminated at 50 Ω with a Hewlett-Packard MSA-1105, 1.3-GHz MMICamplifier. The MSA-1105 provides areasonable input match up to 3 GHzand provides about 12 dB of gain. Themixer’s sum-frequency output isfiltered by a 270-MHz low-pass filterat the output of U4 as shown in Fig 4.

After several more stages of amplifi-cation, the RF signal is fed through aPIN-diode attenuator, final amplifierand directional coupler. The direc-tional coupler is used to separate theforward output power from any that isreflected, so that the true forwardpower is leveled correctly. This stan-dard scheme has been used for decadesin leveling the power of RF sources(see Reference 4).

The forward-coupled path is detectedin the leveling loop by an AnalogDevices AD8307 broadband log ampli-fier to form a dc voltage that is propor-tional to the output power in dBm. Thisdc voltage is compared to the referencevoltage produced by a digital-to-analogconverter (DAC) on the CPU board andthe PIN-diode attenuator is adjusted bythe leveling loop until the equilibriumis achieved. By changing the leveling-DAC output voltage, the output level ofthe synthesizer may be changed overthe range of –20 to +15 dBm. A direc-tional coupler is used because it separ-ates the true output power of thesynthesizer from any reflections, thusincreasing the accuracy of the output

level in the face of possible wildlyvarying termination mismatches (seeReference 4).

The CPU board (Fig 5) contains aMicrochip Technology PIC16C63,single-chip microprocessor. The PIC isused as a smart UART; that is, itreceives command packets from a PCover a 19.2-kbps RS-232 port and exec-utes instructions to set the hardware orshift serial data out to the PLLs, NCOand leveling DAC. The PIC operates asa “smart-logic” device that can bequickly reprogrammed during develop-ment. It essentially replaces dozens ofdiscrete logic chips with just onepackage. The CPU also monitors thelock status of the PLLs and function ofthe leveling loop, then sends thisinformation back to the PC.

As shown in Fig 5, the RS-232interface includes a “link port.” Thisconnection allows up to four sources toshare one RS-232 port. The address-select switch must be set to the desiredaddress for each source on the linkport. When a command is sent to asource, only the source with the properaddress acts on that command.

The PIC does no significant calcu-lations; it merely acts as go-betweenfor the PC and the actual hardware. Asdescribed later, the PC program doesall the calculations and sends binaryinformation to the PIC via the RS-232serial port. The PIC converts thisbinary information to a bit stream thatsets the PLLs and other componentsto desired states.

Here’s the bottom line of this wholescheme: By using a fixed frequency of750 MHz for the LO and mixing thiswith a variable frequency of 500 to748 MHz, we can produce a 2 to250 MHz output. Conventional VCOscannot be made to tune much morethan an octave. By using a heterodynetechnique, we can make a synthesizerthat tunes over seven octaves in asingle band.

Some Design SpecificsOne thing I’ve found while working

with PLLs is that it is very easy to getthe frequency you want. The toughpart is not getting the frequencies youdon’t want! In fact, until one learns to

Jan/Feb 2000 5

6 Jan/Feb 2000

Fig 2—The NCO circuit is used to derive the master-clock references for the rest of the source. The NCO-driven, 10.7-MHzoutput is used as a super-fine adjustable reference for the PLL 0 board. PLL 1 and the PIC microprocessor are driven by fixed10-MHz clocks. Unless otherwise specified, use 1/4 W, 5%-tolerance carbon composition or film resistors. All capacitors are50-V X7R ceramics unless otherwise noted.

J1-J3—SMB connector, PC board mount(Digi-Key #J648-ND)L1-L3—Panasonic EMI filters (Digi-Key#P9807CT-ND)U1—Harris HSP45102 NCO (Allied#HSP45102PC-40)

U2—Harris CA3338AE DAC (Allied#CA3338AE)U3—40 MHz clock oscillator (Digi-Key#XC279-ND)

U4—CD74HCT74E (Digi-Key#CD74HCT74E-ND)U5—LM2940CT 5 V regulator (Digi-Key#LM2940CT-5.0-ND)

measure phase noise and discretespurious frequencies, life goes on quitemerrily.

The frequencies you don’t want arediscrete spurs—such as power-line-related, PLL-reference feed-through—and mixer-intermodulation spurs,as well as the PM noise commonlyreferred to as phase noise.

Reference feed-through is addressedby using a low PLL bandwidth suchthat the reference frequencies are

easily filtered out due to their muchhigher frequencies. Power-line-relatedspurs fall into two main categories:

1. 120-Hz ripple: Since most linearsupplies are full-wave rectified, 120 Hzis the main interfering frequencyoutput of the power supply. This ripple,if inadequately filtered, will show up asmodulation on the output of the VCO. A60-Hz subharmonic may also appearbecause the rectifier diodes do notconduct the same on each half cycle,

although when this does happen, theamplitude is usually 10-40 dB below the120-Hz component. 120-Hz contam-ination is predominately a problem ofconduction through the power andground wires.

2. Magnetic field coupling: Thefrequencies most commonly seen are60 Hz and 180 Hz. 60 Hz is the funda-mental frequency of the magnetic fieldin the transformer. When a magnetic-core transformer is driven relatively

Jan/Feb 2000 7

hard, the magnetic field waveformbegins to look like a square wave nearsaturation. This leads to the appear-ance of odd harmonics. So 60 Hz,180 Hz (third harmonic) and 300 Hz(fifth harmonic) tend to show up onVCOs that are placed too close to power-line transformers.

The other common discrete spursthat show up in synthesizer circuits areunwanted mixing products and feed-through of various clocks. Unwantedmixing spurs are best remedied byremoving spurious frequencies beforemixing. I have done this here by low-pass filtering the PLL signals, bycareful frequency planning and byusing the proper drive levels to themixer to limit spurious responsescaused by overdrive, especially on themixer’s RF input.

Phase noise is a different sort ofproblem. Instead of discrete-frequencyleakage, it is best thought of as “fre-quency smearing.” On a spectrumanalyzer, phase noise looks like anincrease in the noise floor around asignal rather than a discrete spur.Phase-noise energy is spread acrossmany frequencies and decreases inenergy with increasing offset from thecarrier.

Every component in this synthesizerpotentially adds phase noise to thedesired signal. In oscillator circuits,two basic tenets help improve thephase-noise performance:

• Use a high-Q resonator.• Use lots of power in the resonator

circuit (see Reference 5).The master 40-MHz reference clock is

a crystal-oscillator design and, as onewould expect because of its high Q, ithas very little phase noise: better than–100 dBc / Hz at 1 kHz offset from thecarrier. Given the block diagram andthe reference’s phase noise, a calcu-lation of the absolute-best phase noiseat the output of the signal generator ispossible. Since the reference oscillatoris at 40 MHz and the output frequencyof the PLL is 750 MHz, the frequencymultiplication is 750 / 40 = 18.75. Thephase noise increases by 20 dB for every10 times the frequency is multiplied.

For a multiplication factor of 18.75,the phase noise will increase by at least:

20 18 75 25log . ( ) = dBc/Hz increase (Eq 2)

So at the 1-kHz offset, we can expect abest-case phase noise of –100 + 25 =–75 dBc / Hz.

A frequency divider reduces thephase noise of a signal by a similaramount; ie, –20 log N dB, where N is thecounter’s division ratio. The action of

the NCO is that of a divider also,reducing the phase noise by its divisionratio. Since the output frequency of theNCO is 10.7 MHz and the input clock is40 MHz, the phase noise reduction is:

2040

10 712log

.

≈ dBc/Hz (Eq 3)

It’s interesting that although wecontinue to divide the reference fre-quency in the PLLs to 100 kHz, thephase noise of the synthesizer overallwill not decrease because of this divi-sion. This is because we need a fre-quency higher than the reference inthe first place, so we eventually needto multiply the reference up, and thePLLs used here do so. They multiplythe 100-kHz reference up to the orderof 750 MHz, or 7500 times.

Division has limits, however. TheCMOS dividers used in the MC145191have a reported phase-noise floor ofapproximately –160 dBc / Hz at largeoffsets; below 10 Hz, the noise floor ofthe dividers also tends to increase (seeReference 1). So in reality, infinitelydividing a noisy reference can produceno better than the divider’s phase-noisefloor, approximately –160 dBc/Hz. Thisis low, but when multiplied up again by7500 (a 77 dBc / Hz increase), the phasenoise increases again.

This concept is more intuitive if weconsider an analog example: If a signalis divided in amplitude by 1000, thenamplified by 1000, it is pretty obviousthat the output signal-to-noise ratio(S/N) will degrade from that of theoriginal signal. The power divider has anoise figure equal to its attenuation andthe amplifier has its own noise figure;hence, we add noise figure to the signalat every stage, decreasing its S/N. Inthe PLL-synthesizer’s case, this in-crease in noise figure at every stage actsto increase the phase noise of thesynthesizer.

Addition or subtraction of frequency—as in the mixer portion of thiscircuit—does not decrease phase noise;rather, the phase noise increases by thesquare root of the sum of the squares ofthe original signals being mixed.Therefore, if one source having lots ofnoise is mixed with a quiet one, then thetotal noise is essentially just the noisysource’s contribution. Of course, adoubly balanced mixer such as the oneused here also has loss and addsapproximately 7 dB to the compositenoise figure.

Now you should see what I mean bythe “ignorance is bliss” statementabove. Every signal path has theopportunity to worsen the phase noiseof the original signal.

In my circuit, as in most PLL-basedsynthesizers, the greatest phase-noisecontribution is that of the VCO. I useda commercial VCO for several reasons.First, if I used world-class VCO circuitslike those Rhode described in Refer-ence 2, the component count would haveskyrocketed beyond what was practicalfor this project. Second, we can get avery good performance-to-price ratioand save building time by using amodular VCO. We must pay attentionto PLL loop design and take care thatthe rest of the signal paths to not addunduly to the overall phase noise.

A major part of the PLL board’scircuitry (Fig 3) is contained in theMotorola MC145191 IC. This devicecontributes very little phase noise, andits use of a current-output charge pumphelps to limit noise. This is so becausethe output essentially goes to a high-impedance state when locked and onlysmall correction pulses are added fromtime to time to maintain stability.

The charge pump and error amplifiercan have significant impact on phasenoise for several reasons, includingnoise in bias currents, input-referrednoise and loop-filter design. This isbecause any unwanted signal on theVCO-tune line will show up asmodulation on the VCO output. TheVCO used has a tuning sensitivity ofabout 25 MHz / V. So if there’s amicrovolt of extra noise on the tuningline, output modulation of 25 Hz will beseen. If this extra microvolt is randomin nature and has a moderately widebandwidth, the effective phase-noisefloor of the entire circuit will suffer.

I made one concession to greatlyimprove performance in the PLLcircuit; I use an industry-workhorse,low-noise operational amplifier: theOP-27.6 The OP-27 has noise at least20 times lower than most commonlyavailable, single-supply op amps onthe market. This required the additionof a negative-bias supply (thus compli-cating the power-supply circuitry),but I felt the increased performancewas worth the cost.

You hear of PLL circuits having abandwidth. This bandwidth is set bythe RC frequency-shaping networksbetween the phase-detector output andthe VCO tuning-voltage input; it isrequired to be within certain ranges bythe laws of circuit design and practic-ally. Bandwidths less than 300 Hz, orso, tend to suffer from microphonics.On the other hand, the maximum band-width is on the order of 5 to 10 timesless than the PLL reference frequency.Since the phase detector employs a

8 Jan/Feb 2000

Fig 3

Jan/Feb 2000 9

Fig 3 (see left)—The same circuitry isused for PLL 0 and PLL 1 boards. Theheart of each PLL board is the MotorolaMC145191. A Mini-Circuits VCO is usedbecause of its high performance andbecause use of a commercial VCOminimizes assembly time. Unlessotherwise specified, use 1/4 W, 5%-tolerance carbon composition or filmresistors. All capacitors are 50-V X7Rceramics unless otherwise noted.“Poly” indicates polyester capacitors.J1, J2—SMB connector, PC boardmount (Digi-Key J648-ND)L1-L4—Panasonic EMI filters (Digi-KeyP9807CT-ND)U1—MC145191 PLL (Newark#MC145191F)U2—OP-27 op amp (Newark #OP-27GP)U3—465-765 MHz VCO module (Mini-Circuits Labs #POS-765)U4—HP MSA-1105 MMIC (Newark #MSA-1105)U5—LM2931AZ 5.0 V regulator (Digi-Key#LM2931AZ-5.0-ND)U6—LM78L15ACZ 15 V regulator (Digi-Key #LM78L15ACZ-ND)

sampling process, the Nyquist band-width limits also apply. Between thesetwo extremes are the design tradeoffsthat can consume many days of experi-mentation and analysis (see References1, 2 and 5). Large bandwidth is desi-rable for fast switching, but less band-width is needed to limit referencefrequency feed-through.

Below the crossover frequency of acarefully designed PLL, the loop canbe made to reduce phase noise by anamount equal to the loop gain ofcircuit. This can be used to advantagebecause it will help to clean up a noisyVCO. The limit on how clean the VCOcan be made is set by the character-istics of the reference frequency.Above the loop bandwidth, the outputnoise is dominated by the VCO noisealone because the loop is not able tocorrect for perturbations.

Usually, phase noise is traded offagainst settling-time and reference-spur requirements. I chose a loop band-width of approximately 1 kHz, sincethis met several design goals simul-taneously. The multiplied reference-frequency phase noise crosses the VCOphase noise at approximately thisfrequency. A 1-kHz crossover frequencyallows for a PLL settling time of approx-imately three milliseconds (to 0.1%).The PLL susceptibility to microphon-ics with a 1-kHz bandwidth is reducedand the 100-kHz reference is prettyeasy to filter with passive parts withoutseriously effect-ing the loop’s gain andphase-margin performance.

Power-Supply DesignAt offsets below 500 Hz, the power-

supply circuitry is critical. I used someoff-the-shelf toroidal transformers (seeFig 6) in the power supply becausethese units significantly reduce any 60,180 and 300-Hz magnetic fields withoutadding any special µ-metal shielding.

Within a toroidal-coil structure, themagnetic field is theoretically con-tained completely inside the core. Thisis not the case with transformers usingE-I laminations, where the field “blows”out at the corners. In addition, the useof a distributed-power architecture—where each circuit has its own localregulators—serves to limit conductednoise in the sensitive PLL and outputcircuits. Remember: A microvolt at theVCO input pin produces 25-Hz FM onthe output, so conducted and radiatednoise must be contained before it getsto any of those circuits. Putting each RFcircuit block in its own die-cast, metalbox serves to shield any spurioussignals entering or exiting to any of theother circuit blocks, too.

PerformanceThe source’s phase-noise perfor-

mance is shown in Fig 7. Note thatrespectable phase noise of better than –91 dBc / Hz is achieved at 10 kHz offsetfrom the carrier. This should be morethan acceptable for most homebrewingtest needs. Fig 8 shows the close-inspurious performance. The referencefeed-through is better than –79 dBc.

Harmonic performance is one area inwhich a substantial design tradeoff wasmade in the name of simplicity. In acommercial synthesizer, sub-octavefilters might have been used to reducethe harmonics generated by the out-putamplifier chain. This would have re-quired much more complexity than wasdeemed practical for this project. Thissource has one band-limiting filter inthe forward signal path. The 270-MHzlow-pass filter I used strips off the sumfrequencies of the mixer and helps toreduce the RF and LO feed-through ofthe mixer, but does little to improveharmonics below 150 MHz. Even so, atan output power of 0 dBm, harmonicsare typically below –45 dBc from20 MHz to 250 MHz. Below 20 MHz, thePIN-diode attenuator used on the mixerboard adds most of the distortion.7 Atan output frequency of 2 MHz, theharmonics rise to –30 dBc. If improvedharmonic performance is needed for aspecial situation, Mini Circuits BNCfilters8 can be added to the output asrequired.

Non-harmonic spurious responsesalso result when signals are mixed. Thefrequency plan of this design was con-

strained by the availability of suitableVCOs and the desire to make it reason-ably repeatable by a homebrewer.9Even with these restrictions, the fre-quency plan produces very few “cross-ing spurs.” Crossing spurs occur where(nF1 + mF2) coincides with the desiredoutput.

The worst crossing spur happens at250 MHz, where a rather large third-order and two smaller seventh- andeighth-order (LO to RF) spurs cross thedesired output. The next-largest spurcrossing happens at 187.5 MHz, wherefifth- and ninth-order spurs cross thedesired signal. A much less problematiccrossing occurs at 150 MHz, where aseventh-order spur crosses. The finalspur crossing, at 125 MHz, is a seventh-order spur that (at least with myspectrum analyzer) is in the noise level(ie –80 dBc).

A spur-avoidance technique isapplied in the PC control program (seebelow) that basically shifts the LOfrequency down to 740 MHz within1 MHz either side of the spur crossingsat 250, 187.5 and 150 MHz. The RF issimilarly shifted to keep the spursaway from the carrier by at least4 MHz. This is perhaps reasonable formost general uses, but if the sourcewere used as a LO in a receiver, wewould also have to avoid the imagefrequencies of the receiver’s mixer.Since I can’t envision every possibleuse of the source, I can’t take care ofthese crossings for every possible casewithout significantly increasing thecontrol program’s complexity. Thecontrol program has the capability ofdisabling its spur-avoidance behaviorif needed, however. Refer to Table 1 fora summary of typical performance.

Software for the SourceSoftware support is provided in two

forms. For stand-alone applications,a 32-bit, Windows-95 program10 isprovided to drive the source. With thisprogram, the frequency and outputamplitude may be set quickly to anydesired value by means of the Windowsgraphical user interface (GUI).

For computerized test applications, Ihave also provided what is known as an“Active-X” control. This 32-bit controlis much like a dynamic linked library(DLL). Many Windows-95-compliantprograms may use Active-X controls,including Word and Excel. Program-ming languages such as Visual Basicand the specific-test generationprogram HP VEE11 can also use Active-X controls. Basically, the Active-Xcontrol is a standard library format

10 Jan/Feb 2000

Fig 4 (see right)—The Mixer board iswhere the source comes together. ThePLL-module outputs are difference-mixed to provide an output of 2 to 250MHz. A leveling circuit keeps theoutput level fixed regardless of loadmismatch. Unless otherwise specified,use 1/4 W, 5%-tolerance carboncomposition or film resistors. Allcapacitors are 50-V X7R ceramicsunless otherwise noted. “Poly”indicates polyester capacitors.J1, J4—SMB connector, PC boardmount (Digi-Key J648-ND)L1-L2—Panasonic EMI filters (Digi-KeyP9807CT-ND)L3—330 nH 100 mA RFC (Digi-KeyM9R33-ND)U1, U3—780 MHz low-pass filter (Mini-Circuits Labs #PLP-850)U2—10-1000 MHz mixer (Mini-CircuitsLabs #SBL-1Z)U4, U6, U7, U9—HP MSA-1105 MMIC(Newark #MSA-1105)U5—270 MHz low-pass filter (Mini-Circuits Labs #PLP-300)U8—2-400 MHz variable attenuator(Mini-Circuits Labs #SYAS-1)U10—250 MHz coupler module (Mini-Circuits Labs #PDC-20-3BD)U11—AD8307AN logarithmic amplifier(Newark # AD8307AN)U12—LM662N op amp (Digi-Key#LMC662N-ND)U13—LM2940CT 15 V regulator (Digi-Key #LM2940CT-15.0-ND)U14—LM2931AZ 5.0 V regulator (Digi-Key #LM2931AZ-5.0-ND)

that encapsulates the functionality ofthe source into a single well definedcomponent that many Windows pro-grams can plug into and use.

All of the software described inclu-ding the Visual Basic 5 source codeand the firmware for the PIC may bedownloaded from the ARRL’s Website.12

Source ConstructionIf you have never worked with

surface mount technology (SMT) butwish to learn with a relatively simpleproject, then look no further. Thesource’s circuitry has been designedand optimized to use the absolute-minimum number and variety of SMTparts. In fact, only eight different SMTparts are required for construction!Three of these are ICs; the rest areresistors and ceramic capacitors. Theshipping charges for the passive compo-nents is greater than the part costs!

Recent editions of The ARRLHandbook and recent QST articles13

provide describe SMT-assembly tech-niques and—with a small solderingiron tip—it’s not difficult at all. In fact,now that I’m used to the technology, Ifind prototype building with SMT partsfaster and easier than using leaded

components. SMT really is somethingto be learned and used—not avoided!

The synthesizer circuitry is parti-tioned exactly as is the block diagram.All SMT components mount on thetrace side of the circuit boards14 asshown in Figs 9 and 10. This allowseasy construction of the SMT partsfirst, followed by the addition of thethrough-hole parts.

Each circuit board is sized to fit in adie-cast aluminum box, obtained fromJameco. The RF connectors (marked“Jx” on the schematic) are all on oneside of each PC board, so that the boxmay be drilled and the PC board slippedinside as shown in Fig 11. The use ofdie-cast boxes greatly reduces radia-tion coupling between circuit sections.

The dc and programming connec-tionsmay be made by passing insu-latedwires through holes in the box. As longas holes are kept smaller than 1/4 inch,feed-through capacitors are not needed.The only precaution I took with the PLLswas to pass the power leads through aRadio Shack snap-on common-modefilter (#273-105). It is a reasonableprecaution to put one of these filters onthe RS-232 connection and the powercord also, just to be sure that no “nasties”get into or out of the source.

The NCO and Mixer PC boards havelarge, TO-220-packaged voltage regu-lators on one side of the board. Thetabs should be attached to the die-castbox for extra heat sinking. The TO-220tabs need no insulation as they are atground potential.

The RF connections between mod-ules use SMB, snap-on RF connec-tors. The mating sockets (Digi-Key#ARF1235-ND) are very easy to assem-ble using RG-174 subminiature coax. Ifind these SMB cables are easier to puttogether than BNCs, which never seemto come out right for me.

The Mini-Circuits modules are sup-plied in small, steel-pin-mount pack-ages. After the basic functionality of thecircuit is tested, these packages shouldbe tack-soldered to the circuit board’stopside ground plane on both longedges. This helps to reduce the induc-tance between the package and circuitground and improves shielding.

I also took the precaution of mount-ing the PLL modules using double-sided foam tape and rubber grommets.This helps damp any vibration fromother pieces of test equipment.

If you are going to use the source asan RF input to a receiver, you shouldalso take some special precautions in

Table 1—Typical Performance Summary

Frequency Range 2-250 MHzFrequency Resolution 1 HzAmplitude Range (calibrated) –15 to +15 dBmAmplitude Resolution 0.1 dBMaximum output +17 dBmAmplitude Accuracy ±0.5 dB over full frequency and

power rangeFrequency Accuracy Error Correctable to 0 Hz frequency drift

per hour (after warmup): 0.0002 %Frequency Stability 0.001 % over range of 15-35°CPhase Noise –64 dBc/Hz @ 1 kHz

–90 dBc/Hz @ 10 kHz–110 dBc/Hz @ 100 kHz

Harmonics: 2 MHz to 20 MHz < –30 dBc20 MHz to 250 MHz < –45 dBc

Spurious: 2 to 220 Mhz < –65 dBc220 to 250 MHz < –35 dBc

LO and RF feed-through < –55 dBmClock feed-through (fundamental and harmonics) < –85 dBmOutput Match (SWR)

2 MHz to 7 MHz < 2:17 MHz to 250 MHz < 1.3:1

Note: Output at 100 MHz, 0 dBm output power and temperature of 25°C, unless otherwise specified.

Jan/Feb 2000 11

12 Jan/Feb 2000

Fig 5

Jan/Feb 2000 13

Fig 6—The power supply uses toroidal transformers to prevent radiated magneticfields from modulating the source output. As a further noise-reduction technique,all of the critical RF modules contain their own local linear voltage regulators toprevent cross-talk. Unless otherwise specified, use 1/4 W, 1%-tolerance carboncomposition or film resistors.

Fig 7(see right)—The phase noise of thecompleted source is a respectable –91dBc/Hz at a 10 kHz offset from thecarrier.

Fig 5 (see left)—The PIC16C63microprocessor acts as a “smartUART,” receiving commands from thePC via an RS-232 port and setting thesource hardware appropriately. The RS-232 connection operates at a speedy19.2 kbaud. Unless otherwise specified,use 1/4 W, 5%-tolerance carboncomposition or film resistors.L1-L2—Panasonic EMI filters (Digi-KeyP9807CT-ND)S1, S2—Two-position DIP switch (Digi-Key #GH1102-ND)U1—PIC16C63 programmed, see text(Digi-Key #PIC16C63-20ND)U2—MAX232CPE RS-232 ASIC (Digi-Key#MAX232CPE-ND)U3—AD7243BN (Newark # AD7243BN)U4—LM78L15ACZ 15 V regulator (Digi-Key #LM78L15ACZ-ND)U5—LM2931AZ 5.0 V regulator (Digi-Key#LM2931AZ-5.0-ND)

R3—200 Ω single-turn trimpot, 5%T1—120/240-V primary, 14-V center-tapsecondary, 1.07 A (manufactured byTalema Inc, Digi-Key #TE70050-ND)T2—120/240-V primary, 14-V center-tapsecondary, 0.5 A (manufactured byTalema Inc, Digi-Key #TE70030-ND)

U1—LM2941CT 17-V regulator (Digi-Key#LM2941CT-ND)U2—LM79L05ACZ 5 V regulator (Digi-Key #NJM79L05A-ND)U3—4-A, 100 PIV bridge rectifier(RadioShack #276-1171)

the housing of the completed unit toachieve minimum RF leakage. AsFig 11 shows, the chassis I used for theconstruction of the source is paintedwith a nice, scratch-resistant finish.While pretty, it does not allow thechassis to be sealed, in an RF sense.For minimum leakage, all the joints ofthe chassis must be sanded to baremetal, so that each piece makes goodcontact with the others. Anotheroption is to build a sub-chassis fromscrap copper-clad PC board materialand place all the RF modules in thisbox, then place this sub-box inside thepainted chassis. The power supply, asshown in Fig 12, is constructed bread-board-style on a piece of Vector “perf”board because of its simplicity.

CalibrationNotice that there is not a single

adjustment in this entire design—well,okay, except for the power supply.There are only two error sources ofconcern in the design. The first is theaccuracy of the frequency reference. A0.005% (50 ppm) error in the 40-MHzreference translates directly to a0.005% error in the source’s outputfrequency. This is very easy to correctin software. If an accurate measure-ment can be made of source frequencyusing, say, a frequency counter, thenthe error is corrected with the PCcontrol program.

The other important error source isthe output-leveling circuit. Since thecoupler and log amplifier are subjectto gain and offset errors, some means

14 Jan/Feb 2000

Fig 8—The close-in spurious performance of the source shows a referencerejection of –80 dBc. The span for this display is ±125 kHz from the carrier.

Fig 9—The PC boards are constructed with the through-hole components mountedto the top of the board. This is a view of the Mixer board component side.

of correction is needed. This is doneusing a DAC on the CPU board. TheDAC output range and resolutionexceed those needed in the circuit. Theprocedure is to set a level with the PCcontrol program and measure theoutput with a power meter, spectrumanalyzer or scope (in order of decrea-sing accuracy). The frequency used forleveling-circuit power correction is10 MHz, making a low-frequency scopemeasurement practical.

A linear-regression, straight-linecurve is then fit to the data and inputto the PC control program. This linearequation is then used to determine theproper DAC output corresponding topower levels from –20 to +15 dBm.

Other leveling-circuit errors arecaused by the coupler and log-amplifieramplitude-versus-frequency errors.These errors are accounted for in amanner similar to the above. At10-MHz frequency intervals, the poweris measured with the output set to0 dBm. A simple correction curve is fitto the data.

These two leveling correctionsprovide better than 0.5-dB outputaccuracy over the entire frequency andoutput-power ranges of the source. Ifyou don’t have the equipment avail-able to calibrate the source, I willsupply default data that should typi-cally provide better than 1.5 dBaccuracy in the leveling circuits. TheWindows-95 control program alsoallows a pure power offset to be addedto the calculated powers. This is usefulto correct for the gain or loss of externalcircuits connected to the source.

The Most Popular QuestionIf this project is like every other I have

done, the most popular question will be:“Can I extend the frequency to 500 MHz?”Well, as in all projects, design goals mustbe traded against complexity, perfor-mance and practicality. 250 MHz waschosen because it is a reasonable upperlimit of what can be assembled bread-board-style by a reasonably skilled RFbuilder. Getting back to the questionhowever, the answer is: “Yes, thefrequency can be extended, but not in thesense of getting higher-frequency VCOsor mixers.” An external frequency doub-ler (Mini Circuits Model FD-2) can beused on the output of the generator toget to 500 MHz, with certain distinctdrawbacks. Since the doubler conver-sion loss is 13 dB, a maximum output ofonly +2 dBm at 500 MHz is possible. Thefourth harmonic from this doubler is only18 dB below the second, however, so a1000-MHz signal can be generated at

about –16 dBm. The phase noise de-grades by 6 dB per doubling, but theresulting performance is still veryrespectable.15

Notes1V. Manassewitsch, Frequency Synthesizers

(New York: John Wiley and Sons, 1987).2U. Rohde, “Key Components of Modern

Receiver Design,” Parts 1 and 2, QST,May pp 29-32, and June pp 27-31, 1994.

3Qualcomm is assigned a patent on this ap-proach (US Patent Number 4,965,533).Many books and articles reference the tech-nique but not the patent (Rhode mentionsboth and a prior use in a Rhode & Schwartzsynthesized source). The Qualcomm Website has information on this technique andmany other interesting products at www.qualcomm.com. [My research reveals useof the DDS-driven PLL as early as 1984. It’slikely QEX will soon publish an article writ-ten by one of the first investigators of this

Jan/Feb 2000 15

Fig 11—All major circuit blocks in the source are designedto be mounted in low-cost, die-cast enclosures fromJameco. Here is a view of a completed NCO board with someadded shielding (on the right side) around the 10.7-MHzfilter to reduce any clock feed-through to the 10.7 MHz PLLreference signal.

Fig 12—What all your hard work can produce—your own VHFsynthesized source. I used a new Bud chassis to house mysource, but a surplus enclosure can work equally well.

Fig 10—The SMT components are attached to the solder sideof the PC boards. This is a view of the Mixer board solderside showing SMT construction.

technology. I think it will shed some light onthings.—Ed.]

4S. Adam, Microwave Theory And Applica-tions (New Jersey: Prentice-Hall, 1969).

5R. Rhea, Oscillator Design and ComputerSimulation, Second Ed. (New York:McGraw-Hill, 1995).

6The OP-27 data sheet is available fromAnalog Devices at www.analog.com. It isinteresting to note that Analog Deviceshas recently introduced some low-noise,single-supply op amps, but their availabil-ity to homebrewers is limited at this time.

7The PIN-diode attenuator and the low-fre-quency performance of the mixer/amplifierchain are what set the low-frequency limitof this design. The basic topology (withmore complex circuitry) can theoreticallyproduce an output to dc, limited only bythe residual FM of the PLL.

8Mini Circuits Labs, PO Box 350166,Brooklyn, NY 11235-0003; tel 800-654-7949, 718-934-4500, fax 718-332-4661;http://www.minicircuits.com/.

9This design can be built by an experiencedRF builder with little difficulty. In fact, thatis how this project started. To see a photoof my original breadboard, visit my Website at www.sonic.net/~shageman.

10The control program requires Windows 95,98 or NT with one 16650-controlled serialport available for communication and 3MB of available hard-disk space.

11HP VEE is a graphical test-generation pro-gram that can use Active-X components.Visit www.tmo.hp.com and search for“VEE.”

12You can download this package from theARRL Web site at www.arrl.org/files/qex/. It includes complete parts lists. Lookfor 0100HAGE.ZIP. A programmed PIC isavailable from the author for $30.00 US,postpaid in the US and Canada. Add $5 tocover postage and handling to Europe,$15 to South America. Send payment as acheck or money order, no credit cards.

13S. Ulbing, “Surface Mount Technology—You Can Work with It!” Parts 1, 2 and 3,QST, April-June, 1999.

14FAR Circuits, 18N640 Field Ct, Dundee, IL60118-9269; tel 847-836-9148 (Voicemail), fax: 847-836-9148 (same as voicemail); farcir@ais.net; http://www.cl.ais.net/farcir/.

15I maintain a FAQ page at www.sonic.net/~shageman. Be sure to check there forupdated information on this project andothers.

16 Jan/Feb 2000

Want to build some VHF/UHF/Microwave Yagis? Someup-front measurements can cut the time required for

tuning. Use this method to determine boom-correctionfactors for Yagi elements mounted through the middle of—and in good electrical contact with—a metal boom.

By Guy Fletcher, VK2KU

12 Sassafras Gully RdSpringwood, NSW 2777Australiaguy@ics.mq.edu.au

Effects of Boom and ElementDiameters on Yagi ElementLengths at 144, 432 and

1296 MHz

Editor’s note: Since this articleoriginally appeared in Amateur Radiomagazine of the WIA in March 1999,corrections and clarifications havebeen provided by the author.

My experiments were performed atfrequencies of 144.2, 432.2 and 1296.2MHz and provide data for boomdiameters of up to 0.08 λ. They alsoexplore the effect of element diameter.The results show clearly that thecorrection depends not only on boomdiameter but also on element diameterand element length.

The effect of the boom on eachelement may be represented by a neg-ative reactance at the center of the

element. A simple empirical formulafor this reactance agrees well withall the experimental data and allowscorrection for any combination of boomdiameter and element diameter.These results are given in the form ofa universal graph.

The observed dependence onelement length is intrinsic to themodel of boom reactance and leads toa correction that tapers as the elementlength decreases. This may be ade-quately represented in practice by asimple power-law modification of thevalue for a standard element length of0.42 λ taken from the graph.

The use of tapered corrections forthe different element lengths (ratherthan a single fixed correction) hasbeen applied to examples of practicalYagis. The difference is negligible at144 MHz and small at 432 MHz. At

1296 MHz, however, where boomdiameters may be relatively large (interms of wavelength—Ed.), the use ofa fixed correction appears to changethe performance parameters of theantenna quite significantly.

BackgroundBoom correction factors are

discussed by Günter Hoch, DL6WU, inthe VHF/UHF DX Book (edited by IanWhite, G3SEK) and other similarreferences.1 Günter’s corrections may1I. White, G3SEK, Ed., VHF/UHF DX Book.

This book is available from your localARRL dealer or directly from the ARRL as#5668. Mail orders to Pub Sales Dept,ARRL, 225 Main St, Newington, CT06111-1494. You can call us toll-free at tel888-277-5289; fax your order to 860-594-0303; or send e-mail to pubsales@arrl.org. Check out the full ARRL publicationsline at http://www.arrl.org/catalog.

Jan/Feb 2000 17

be embodied in a formula developed byIan:

C

B

B B= 25.195 229

2

λ λ

– (Eq 1)

where C is the correction and B is theboom diameter, both in millimeters.This formula is not valid for boomdiameters greater than 0.055 λ,although diameters of up to 16 mm(0.07 λ) are common at 1296 MHz.Ian’s formula is plotted in Fig 1. Itincludes no dependence on elementdiameter or length. Also, the curve isassumed to pass through the origin,although there is no real reason toexpect this. C is obviously zero when Bis zero, but the ratio C/B need not bezero to make C zero.

There seem to be no data availablefor larger boom diameters, which isperhaps why some amateurs haveremarked on the difficulty of matchingantennas correctly at 1296 MHz. Theexperiments to measure boom correc-tions are in fact quite straightforward,so I decided to make some simplemeasurements. The scope of the projectexpanded rapidly as the unexpectednature of the results appeared.

Theory and ModelThe complex voltage reflection

coefficient (ρ) represents the magni-tude and relative phase of the ratio ofthe reflected voltage wave to theforward voltage wave at a load. In theseexperiments, the reflected power ( PR )and the forward power ( PF ) weremeasured rather than the voltages:

ρ =

P

PR

F

1

2(Eq 2)

The voltage standing wave ratio(SWR), σ, is:

( )( )|-|

|+|=ρρσ

11 (Eq 3)

Any element of a Yagi antenna hasenergy stored in the fields surroundingit. Near the element center, the currentis large and the voltage small; near theends, the current is small and thevoltage large. If the element passesthrough a larger conductive boom at itscenter, the skin effect forces the currentto flow around the outside of the boominstead of directly along the element’ssurface. This reduces the volume of themagnetic field around the element andtherefore reduces the energy storedthere. Since the stored energy is direc-tly proportional to the self-inductance(L) of the element, the effect of the boom

Fig 1—Plot of G3SEK’s formula for boom correction.

Fig 2—Data for frequency 144.2 MHz, B = 32.0 mm, d = 6.35 mm.

is to contribute a negative reactance tothe element’s impedance, Z. This nega-tive reactance contribution increasesin magnitude as the boom diameterincreases.

For thicker elements, the volume ofthe magnetic field is reduced anyway,because the field is limited to the regionoutside the element. Thus, there is lessfield volume for the boom to remove, sothe effect of thicker elements will be toreduce the magnitude of the boom’seffect, hence also reducing the correc-tion required. [See the sidebar“Plumber’s Delight Meets EM Theory”for another way of looking at this.—Ed.]

The element-plus-boom can berestored (approximately) to its origi-nal electrical state by lengthening the

element so as to contribute a positivereactance to offset the boom effect.This is the boom correction. BrianBeezley, K6STI, writes in the hand-book to his Yagi design and analysisprogram YO6 that elements of differ-ent diameter are electrically equiva-lent when the phase angles of thecomplex self-impedances are the same.This differs from simply equating theimaginary components (the reactiveparts) of Z.

The ExperimentsThirteen experimental measure-

ments were made with the boom (B)and element (d) diameters shown inTable 1, as limited by available mat-erials. The signal source was a Yaesu

18 Jan/Feb 2000

FT-736R transceiver, delivering 25 Won 144 MHz and 432 MHz and 10 W on1296 MHz.

Forward and reflected powers weremeasured with a Bird 43 wattmeterusing different plug-in elements forforward and reflected power. Therelative precision for the reflectedpower was about 0.02 W on 144 MHz,0.04 W on 432 MHz and 0.01 W on1296 MHz. The absolute measure-ment accuracy was not nearly as goodas this, but the experiments consistedessentially of comparing differentantennas to obtain the same reflectedpower, so calibration errors are not asimportant as reading precision.

For each frequency and boom dia-meter, a simple three-element Yagi wasdesigned using YO6 and constructed ona dry wooden boom (usually rectang-ular). The feed impedance was around25 Ω and T-matching was used with aconventional 4:1 balun. In each case,the elements were cut to the expectedlength; then the T-bars and the lengthof the driven element (DE) were ad-justed for zero reflected power with nometal boom sleeve in place. The metalboom for the director (D1) was an exactsliding fit over the wooden boom andextended about half of the distanceback toward the DE and a similardistance forward. Elements werepinned in place with self-tappingscrews—which made no observabledifference to any reading—to ensuregood electrical contact between ele-ments and boom. This arrangementguaranteed that the director could berepeatedly removed and replaced inexactly the same position.

To avoid ground effects, each an-tenna was mounted to radiate verti-cally upward. With the boom sleeve inplace and the director cut deliberatelylong, the forward and reflected powerswere recorded for each director length(L1), as the length was systematicallyreduced by small amounts until thereflected power was near zero. (Some-times the measurements were contin-ued well beyond this point.) The boomsleeve was then removed and theprocess repeated over a similar rangeof reflected powers. The reflectioncoefficient (ρ, equal to the square rootof the power reflection coefficient) wasplotted against L1. The expectation wasthat two parallel curves would result,their separation being the desired boomcorrection. In fact, the curves were notquite parallel!

Element lengths were measured witha steel ruler on 144 and 432 MHz to aprecision of about 0.2 mm and with dial

Fig 3—Data for frequency 432.2 MHz, B = 20.2 mm, d = 4.76 mm.

Fig 4—Data for frequency 1296.2 MHz, B = 16.2 mm, d = 4.76 mm.

Table 1—Thirteen experimental situations

144.2 MHz: B = 32.0 mm, d = 4.76 mm, 6.35 mm432.2 MHz: B = 16.2 mm, d = 2.40 mm, 3.18 mm, 4.76 mm

B = 20.2 mm, d = 2.40 mm, 3.18 mm, 4.76 mm, 6.35 mm1296.2 MHz: B = 16.2 mm, d = 1.60 mm, 2.40 mm, 3.18 mm, 4.76 mm.

calipers on 1296 MHz to a precision of0.01 mm. These two methods are notequivalent, in that the ruler measuresa length averaged by eye over the endfaces, whereas the calipers measurebetween the high points on each endface. However since all measurementsin any one experiment were madeconsistently, the accuracy of theexperimental boom correction factors

found from a length difference shouldapproach twice the appropriate pre-cision above. The smoothness of theraw data curves supports this belief.

The Results of the ExperimentsFigs 2, 3 and 4 are typical of the 13

graphs obtained for reflection coeffic-ient ρ as a function of director length(L1) with and without a metal boom

Jan/Feb 2000 19

sleeve. Careful study of these and othergraphs shows that the boom correction,measured by the separation of the twocurves, decreases slightly as thedirector length is reduced. This findingis not very surprising, but is significantbecause such a dependence has notpreviously been suggested.

From each graph, the directorlengths with and without the boomsleeve were tabulated at several valuesof reflection coefficient ρ (eg, 0.1, 0.15,0.2, 0.25 and 0.3) and a set of boomcorrections was found. For each of themore than 50 pairs of director lengths,YO6 was used to find the complexelement impedance, Z. The programactually requires a reflector to bepresent, so this was placed 100 metersbehind the driven element, where itwould have no discernible effect. (Theuse of a particular program such asYO6 to find element impedance is opento some criticism. This important pointwill be discussed below.)

The impedances from each pair ofdirector lengths were used to find thenegative reactance (X) contributed bythe boom. This is best illustrated by anexample. The results from Fig 3 forρ = 0.1 are reproduced in Table 2.

The boom correction is here 9.15 mm.The X in the column for impedance withthe boom sleeve present represents theunknown contribution of the boom to Z.The value of X was found by equatingthe phase angle (φ) of Z with andwithout the boom sleeve, so that the twosituations are electrically equivalent.This gives X = –j18.50 Ω. Originally, thecomparison was made by simply equa-ting the imaginary components of Z, butthis procedure led to model curves thatdid not converge in the way actuallyobserved, so the phase-angle methodwas adopted. The values of X found inthis way were reasonably consistentover the whole range of ρ and wereaveraged.

Finally, the value of X was used topredict boom corrections over a widerrange of element lengths typical of along Yagi by reversing the procedure.For the example in Table 2, thecalculated boom corrections range from7 mm (for the shortest director) to12 mm for the reflector. This showsclearly the variation of boomcorrections to be expected over thelength of such a Yagi and the errorsintroduced by using a fixed boomcorrection for all elements. A table ofsuch calculated boom corrections(which include the experimental valuesas a subset) was generated for all 13experiments.

Table 2 —Results taken from Fig 3 for ρ = 0.1

No Boom Sleeve With Boom Sleeve

Director Length (mm) 299.45 308.60Impedance Z (Ω) 51.5 – j35.7 55.5 – j20.0+X

Fig 5—Dependence of inductance L on B and d.

Fig 6—Boom corrections C/B for elements of length 0.42 λ and for variousd/B values.

Each of these 13 tables of calculatedcorrections was plotted against direc-tor length L1, and they were all foundto fit closely to a simple power-lawrelationship. The optimum value ofthe power varied slightly across theexperiments, but a satisfactory fit forall the data was given by:

C k L= ( )1 1 8. (Eq 4)or

C

C

L

L0

1

0

1 8

=

.(Eq 5)

where C0 and L0 correspond to somestandard director length. For variousreasons, this standard element lengthwas chosen to be 0.42 λ, and the finalgraph presented in Fig 7 correspondsto this standard length.

20 Jan/Feb 2000

Effect of Boom andElement Diameters

In spite of trying many differentplots, it has not proved possible torepresent the dependence of thelength correction (C) on elementdiameter (d) in any simple way. Thisis not entirely surprising because ofthe complexity of the effect on elementimpedance of varying the tip length tocompensate for the effect of the boom.It has proved very helpful to break theproblem into two separate parts:

1. The effect of the boom on theelement impedance: As explainedabove, this can be represented as apure negative reactance, the value ofwhich depends also to a lesser extenton the element diameter.

2. The increase in length required tocompensate for this reactance: so as torestore the original phase to the ele-ment impedance.

The boom reactances from the 13different experiments have been con-verted to inductance (L) and plotted inFig 5 as L/B versus d/B. The induc-tance is plotted as a positive quantityfor convenience, but remember that itcontributes negatively to Z.

The graph of L/B versus d/B shows aremarkable linear relationship:

Fig 7—Boom corrections C/B for elements of length 0.42 λ. B/λ marked for each curve.

Seven Simple Steps1. Calculate the wavelength (in millimeters) from:

λ = 299792.5f

(Eq 8)

L

B

d

B=

0 5994 0 999. – . (Eq 6)

Only one point (at 432.2 MHz)departs appreciably from the line ofbest fit. Having due regard to theaccuracy of the data, this relationshipis most simply expressed as:

L B d= 0 6 1 0. – . (Eq 7)In this simple and elegant expression

(Guy’s Rule!), L is the value of thenegative reactance contributed by theboom/element combination to the ele-

ment impedance Z. For the values ofthe constants as presented, L is innanohenries, while B and d are inmillimeters. With this rule, thereactance of any boom and elementcombination can be predicted withreasonable confidence.

Calculation of Boom CorrectionsIt is straightforward—but not

particularly convenient—to use theinductance value given by my rule tocalculate a value for the boom

where f is in megahertz: Eg, f = 1296.2 MHz, λ = 231.3 mm.2. Choose a boom diameter B and element diameter d, both in millimeters. Eg,

B = 16.2 mm, d = 3.18 mm.3. Calculate the ratios B/λ and d/B. Eg, B/λ = 0.070, d/B = 0.196.4. Refer to Fig 7. Draw a vertical line corresponding to the value of d/B and

read off the value of C/B from the appropriate curve. Interpolate between thecurves as necessary. Eg, C/B = 0.645.

5. Calculate C (in mm) from C/B by multiplying by B. This is the boom correctionfor an element of length equal to the standard length, L0. Eg, C0 = 10.4 mm.

6. Calculate the standard length L0 from L0 = 0.42 λ. Eg, L0 = 97.1 mm.7. Calculate the correction C for any element of length L from Eq 5. Eg, for

L = 90.0 mm, C = 9.1 mm.

Jan/Feb 2000 21

Plumber’s Delight Meets EM TheoryGuy’s theory about why Yagi elements appear electri-

cally shorter when attached to a conductive boom proveddifficult for me to understand at first. The following repre-sents my perception of the effect as clarified by him. Ihope it helps you, too.

Current flowing in a conductor such as a Yagi elementproduces a magnetic field aligned everywhere at rightangles to the direction of current flow. The shape of mag-netic field lines satisfying this condition is a circle concen-tric to the element, as shown in Fig A.

The presence of a conductive boom interferes with thedevelopment of the magnetic field near the boom,

correction C for any combination ofboom diameter, element diameter andelement length. This involves usingYO6, first to find the complex impe-dance Z of the uncorrected elementlength and hence its phase φ. Then by aprocess of trial and error to find a newelement length which, when combinedwith the negative reactance contrib-uted by the boom, has the same phase.

Instead, for the standard elementlength of 0.42 λ, boom corrections havebeen calculated over a wide range ofboom diameters (B) and elementdiameters (d) covering all the sizeslikely to be met in practice. The resultsmay be plotted in the form C/B versusB/λ, as in Fig 6, with separate curvesfor different element diameters.Alternatively, C/B may be plottedversus d/B, as in Fig 7, with separatecurves for different boom diameters.Other possibilities include using d/λ inplace of d/B.

Fig 6 may be compared directly withFig 1, based on the G3SEK formula.The curve shapes in Fig 6 are gener-ally similar to that in Fig 1, but it isapparent that the intercepts on thevertical axis of Fig 6 are well abovezero for all values of d/B. The curves inFig 6 also intersect, making it hard to

Magnetic FieldLines

Boom

Element

Current = I e

(A)

thereby affecting the current flow there. See Fig B. Ex-periments verify the electrical shortening of the element.Using knowledge of how EM fields interact with matter, itseems to me this theory could be developed to predict theexact magnitude of the effect. What properties of theboom material are included in the solution? What geom-etries minimize the effect?

I do not find much mention or analysis of this idea inthe literature. It appears to merit further considerationin modeling of antennas, especially at UHF and micro-wave frequencies. What do you think?—Doug Smith,KF6DX.

use in practice. The reason for theseintersecting curves is clearer in Fig 7.In general, as the element diameterincreases, the boom correction factorC/B decreases as expected from thediscussion earlier. In the case of thickbooms however, the boom correctionfactor also falls for very thin elements,for which the reactive component islarge. This is, in fact, a consequence ofthe use of a standard element of fixedlength rather than fixed phase. Theuse of a standard length is much easierto use in practice, but leads to curvesthat intersect when boom diameter isused as the horizontal coordinate.

For the practical prediction of boomcorrections, Fig 7 is significantlyeasier to use than Fig 6 because thevarious curves are well separated andgenerally less inclined.

Practical Significance of Length-Dependent Boom Corrections

The detailed results described in thisarticle are novel in that they lead toboom correction factors that depend notonly on the boom diameter, but also onelement diameter and length. It isreasonable to wonder whether this hasany real practical significance whencompared with the simpler system of a

fixed correction factor presently inwidespread use. If the correctionsshould indeed taper from larger valuesfor longer directors and the reflector tosmaller values for the shorter directors,then the effect of using a single, fixedcorrection is to apply a correction thatis too small for the longer elements andtoo big for the shorter ones.

This can be easily simulated in anantenna analysis program such as YO6by adding the fixed correction to everyelement and then subtracting thetapered corrections. Such simulationslead to the conclusion that—at144 MHz—the difference between thetwo approaches is negligible. This is notat all surprising since the correctionsare a small fraction of the elementlengths. At 432 MHz, small differencesare apparent, but do not appear verysignificant. At 1296 MHz, however, thefixed and tapered corrections differ byconsiderably more than acceptable con-struction tolerances. The predictedantenna properties also differ signific-antly, with some loss of gain when afixed correction is used and majordifferences in the feed impedance. Thisis consistent with the matching diffi-culties previously experienced at1296 MHz by some amateurs.

22 Jan/Feb 2000

Several local amateurs have nowconstructed long Yagis for 1296 MHzusing the VK2KU tapered corrections,and in each case, they have reportedthat matching the Yagi proved quitestraightforward.

ConclusionsThe raw data graphs—such as Figs

2, 3, and 4—appear to show unequi-vocally that boom corrections dependnot only on the boom diameter (as afraction of wavelength), but also onelement diameter and elementlength. The dependence on elementdiameter may not be very startling,but the dependence on element lengthappears to be a novel idea that wasinitially unexpected.

The formula for calculating thenegative reactance contributed by theboom/element combination is also new,but it fits the experimental data verywell. This rule is the key to calculatingboom corrections for any combinationof boom and element diameter. It maywell be that the use of a differentcomputer program for finding elementimpedance would lead to somewhatdifferent values for the negativereactance, and so to slightly differentconstants in the formula. When theprocedure is reversed, however, andthe same program is used to find thecorrections in other situations, suchdifferences between programs shouldlargely be eliminated. In effect, thecomputer modeling is used to inter-polate between boom correction factorsthat were found directly by experi-ment. Thus, I believe that the graphicalresults as presented in Fig 7 aresubstantially independent of thecomputer modeling and represent aclose approximation to the truth.

The length dependence of the correc-tions appears to be best described by apower law of order 1.8, though this valuedoes not seem to be very critical. The fixedboom corrections commonly usedelsewhere extend up to a boom diameterof 0.055 λ. The experiments described inthis article extend this range up to 0.070λ and calculations have been carried out

up to 0.080 λ, thus covering the importantrange of booms thicker than 12 mm at1296 MHz.

AcknowledgementsI particularly thank Gordon

McDonald, VK2ZAB, and Ian White,G3SEK, for their encouragement andadvice in a project that threatened toget out of hand, growing rapidly froma planned single measurement at1296 MHz into a comprehensivesurvey over three frequency bands,three boom diameters and fiveelement diameters.

Guy Fletcher was a senior lecturer(associate professor) in Physics (electro-magnetism and optics) at MacquarieUniversity in Sydney until his retire-ment in 1997. He holds a masters degreefrom Cambridge, UK, and a doctoratefrom Macquarie.

Guy’s previous call signs are G3LNX(UK 1957-1967) and VK2BBF (1967-1998). His amateur interests includeVHF/UHF DX on 144, 432 and 1296MHz, mainly SSB, some CW. He alsostudies propagation modes on thosefrequencies, including tropo’, sporadicE, ducting and aircraft scatter. He isalso active in antenna design and con-struction.

Jan/Feb 2000 23

Forget about optical encoders—here’s a controller boardthat provides an easy, push-button frequency control and

scanning interface for varactor-tuned receivers.

By Thomas K. Duncan, KG4CUY

1107 Alta Vista AveHuntsville, AL 35801thomas.k.duncan@cpmx.saic.com

A Scanner Controller forVaractor-Tuned Receivers

1Notes appear on page 30.

Microcontrollers appear inmany construction projectsin current Amateur Radio

literature, but their usefulness is notlimited to newly constructed equip-ment. In this project, which served asmy introduction to embedded micro-controllers in radios, I grafted ascanner controller into a single-con-version, varactor-tuned receiver.

With increasing activity on 10meters, I wanted a radio to periodicallymonitor certain haunts for upper-sideband voice activity. My store-bought scanner that covers thesefrequencies handles only AM and FM.Besides, I was looking for an excuse toget my feet wet with simple micro-controllers. A 10-meter receiver built inthe early months of increasing solaractivity kept whispering “Use me!”every time it was switched on. I wantedto maintain that radio in workingorder in case this project was a completefailure. The original design—in particular, the board layout—was therefore modified slightly toadd the necessary hooks for the

controller. These modifications arequite simple; it should be possible toadd such a controller to any voltage-tuned receiver.

The Microchip PIC product line waschosen for ease of programming: Aprogrammer could be bought or builtat reasonable price. Developmentsoftware was widely available at littleor no cost, and certain chips areFLASH programmable, so no EPROMeraser was needed. A 16-characterLCD with an on-board controller anda set of five momentary-contact push-buttons (FREQUENCY UP, FREQUENCYDOWN, SCAN, SELECT CHANNEL andPROGRAM CHANNEL) comprise the newtuning mechanism. Recent QSTarticles1, 2, 3 aimed me toward thesedesign decisions.

DescriptionAs the push-button functions imply,

the unit operates as a minimal-func-tion scanner:

• Manual tuning takes place via thefrequency up/down buttons. Tuningspeed is increased by a factor of eighteach time the SELECT CHANNEL buttonis pressed while holding FREQUENCY

UP or FREQUENCY DOWN. The display

tracks frequency while tuning.• The current frequency may be

stored into the current-channelEEPROM space by pressing thePROGRAM CHANNEL button.

• The current channel is incre-mented by pressing SELECT CHANNEL,thereby tuning the receiver to thevalue currently stored for thatchannel. There are 10 channels.

• To scan continuously through allchannels, press SCAN. Scanningpauses when the received-signal linegoes high and stops if it remains highfor five seconds.

In addition, a calibration mode maybe entered by pressing FREQUENCY UPand FREQUENCY DOWN simultaneously.In this mode, the frequency countermay be adjusted to accommodate theexact frequency of the ceramic reson-ator or crystal used in the clockoscillator. Similarly, the IF-offsetmode allows the IF to be set so that thedisplay shows the received frequencyrather than the LO frequency, whichthe counter actually measures.

Some microcontrollers have built-inD/A converters that might generatetuning voltages. They are generallylimited to 8-10 bits of resolution. It wasmy desire to cover 28.0 to 29.2 MHz at100 Hz or better resolution, thus

24 Jan/Feb 2000

requiring at least (1,200,000/100) =12,000 steps, or 14 bits. Because Iwanted to use FLASH-programmabledevices, on which I/O pins were scarceuntil recently,4 a serial interface to anyexternal D/A converter was desirable.Serial D/As with the required reso-lution were rare in DIP packages, so Ichose an Analog Devices AD8402 dualRDAC, or “digital potentiometer,” tofeed an op amp that develops thetuning voltage. Each RDAC section has8-bits of resolution, multiplied in the opamp to provide the desired tuninggranularity. A CMOS rail-to-rail opamp, the National LMC662CN, is usedto provide adequate tuning-voltageswing.

With outputs driving the LCD andRDAC, and inputs from push-buttons,a frequency counter and received-signal line, the 13 I/O pins of one 16F84would not suffice. Two 16F84s aretherefore used, communicating witheach other via two lines: One line is agate pulse for the frequency counter,and the other carries unidirectionalserial I/O. One PIC services the displayand provides the gate pulse. The othermeasures frequency, detects thereceived-signal input and handles thepush-button inputs.

A block diagram of the controllerappears in Fig 1. While most of theimportant work takes place in PIC1, theinterface to the Optrex LCD module is notentirely trivial. It takes much code toconvert the LO frequency from the rather Fig 1—Controller block diagram.

Fig 2—Receiver modifications.

obscure units sent over by PIC1 intodisplayable digits at the input frequency.This division of labor also allows PIC2 toserve as an additional timer used to gatePIC1’s frequency-counter softwareduring the 16-ms counting periods (each16F84 has one timer).

Modifications to theOriginal Receiver

While the original radio already had aLO buffer amplifier to drive another PICdesign (one of Almost All DigitalElectronics’ DFD-series displays),5 ano-ther single-stage FET buffer was added

Jan/Feb 2000 25

at the controller end since the radio andcontroller would be on two separateboards, connected by a length of coax. Theschematic in Fig 2 shows the tuningvoltage and LO-sampling connections.

The received-signal input requirednew circuitry. The original radio didn’thave AGC, so the spare section of anaudio preamplifier, LM1458, and anFET were used to generate about +4 Vin the presence of audio and less than+1.5 V with no audio. In one of its earlyversions, the controller supported onlymanual recall of stored frequencies, sothe received-signal input wasn’trequired. The control signal is onlyneeded to stop scanning when areceived signal is found. The schematicof Fig 3 shows the received-signalcircuitry added to the original radio.

Controller Hardware DetailsFig 4 is the complete controller sche-

matic. The PIC has internal pull-upresistors on the five push-buttonports, so the push buttons need onlyshort these pins to ground. An 8-MHzceramic resonator is used as the clockfor each PIC. Inexpensive 8-MHz crys-tals could be used as well, but theyrequire capacitors between each end ofthe crystal and ground. The 39-kΩ/0.47-µF network on each MCLR inputguards the PICs against false startswhen power is applied. In retrospect,these pins could probably have beenconnected directly to +5 V without vio-lating the power rise-time rules. These

Fig 3—Received-signal indicator circuit.

and all other 16F84 details are avail-able in documentation available onMicrochip’s Web site.

The 32 Ω trimpot R2 in the tuning-voltage circuit is used to adjust formonotonically decreasing tuning vol-tage with increasing RDAC setting.Tuning voltage is adjusted in twoparts: One section of the RDAC drivesthe non-inverting op-amp input, andthe other part drives the invertinginput. The pot adjusts the range of theleast-significant RDAC driving theinverting input so that the tuningdirection does not reverse at either thelow or high ends. This adjustment ismost easily made by connecting afrequency counter with at least 10-Hzresolution to the controller’s LO inputand pressing the FREQUENCY UP orFREQUENCY DOWN button momentarilyseveral hundred times. If when tuningthe counter momentarily reversesdirection (while pressing FREQUENCY

UP the counter goes down), adjust thetrimpot and try the offending range offrequencies again. If the button can bepressed at least 256 times (ensuringthe complete 8-bit range of the leastsignificant RDAC has been traversed)and no frequency reversal occurs, theadjustment is correct.

Controller SoftwareMicrochip provides PIC software-

development tools at no charge on itsWeb site, www.microchip.com. TheMPLAB package is an integrated

development environment that runsunder Windows: I used MPLAB 3.99under Windows NT 4.0. While it ispossible to program PICs in C andBASIC, I chose to stay very close to thehardware by using assembly language.The instruction set is simple andstraightforward, and Microchip sup-plies many support routines written inthe assembly languages of the variousPIC products.

The two PICs comprising the con-troller perform the major tasks shownin Table 1. Some of the tasks deservespecial explanation.

Measuring FrequencyThe LO frequency applied to PIC 1

feeds a prescaler that divides theinput frequency by 16. This, in turn,drives an eight-bit counter whichinterrupts PIC 1 when it rolls overfrom a full count of $FF to $00 [apreceding “$” indicates a hexadecimalnumber—Ed.]. When PIC 1 needs tomeasure frequency, it asks PIC 2 for agate pulse, which is 16 ms long. Whenthe gate pulse rises from its quiescentlow value to high, PIC 1 enables thetimer interrupt, clears the timer, andwaits for the gate pulse to go back tolow. During this wait loop, timerinterrupts occur. On every interrupt,PIC 1 increases the 16-bit frequencycount by 256 by incrementing thehigh-order byte by one. When the gatepulse falls to low, PIC 1 disables inter-rupts and moves the current counter

26 Jan/Feb 2000

Fig 4

Jan/Feb 2000 27

value—equal to the number of pre-scaler counts since the last inter-rupt—into the low-order byte of the16-bit frequency count.

The 16-bit frequency count, CLO,therefore represents the input fre-quency divided by the prescaler setting(16) and by the quantity (1 s) / (16 ms)= 62.5, for a net division by 1000, whichrepresents the LO frequency in kilo-hertz. This number is sent to PIC 2.

Converting LO Frequencyto Tuned Frequency

Depending on the number of CPUcycles in the gate-pulse timing routineand PIC 2’s clock frequency, the gatepulse may not be exactly 16 ms long, soan adjustment is provided. PIC 2 cal-culates the tuned frequency as:

F FC GPM

tuned IFLO= +

( )( )104 (Eq 1)

Where GPM is the gate pulse multi-plier, around 10,000. Its value isstored in the PIC 2 EEPROM, and itmay be adjusted by commands fromPIC 1 to PIC 2. Similarly, the additive

Fig 4 (See left)—Complete controllerschematic diagram. Unless otherwisespecified, use 1/4 W, 5%-tolerance carboncomposition or film resistors.Parenthetical references show (partmanufacturer, source and source partnumber). Equivalent parts may besubstituted for those shown.DS1—16×1 character LCD module (Optrex, Digikey #DMC-16117AN)Q1—J309 FETU1—AD8402-10 dual 10-kΩ RDAC (Analog Devices, Allied #AD8402AN10)U2, U3—PIC 16F84, 10 MHz (Microchip, Digikey #PIC16F84-10/P-ND). PICs must be programmed to function.U4—LMC662 dual CMOS rail-to-rail op amp (National, Allied #LMC662CN)U5—78L05 +5-V regulatorX1, X2—8 MHz ceramic resonator with capacitors (Murata, Allied #CST8.00MTW

Table 1—PIC Tasks

PIC 1 Tasks

1. Measure LO frequency by asking PIC 2 for a gate pulse and counting-timer overflows driven by the LO signal while the gate pulse is high2. When the SELECT CHANNEL button is pressed, increment the channel, retrieve its frequency from EEPROM, set the LO to that frequency and send the channel number and frequency to PIC 2 for display3. Respond to the PROGRAM CHANNEL button by storing the current frequency in the currently selected channel’s EEPROM slot4. Respond to the FREQUENCY UP and FREQUENCY DOWN buttons by adjusting RDAC settings accordingly5. When the SCAN button is pressed, continuously step through channels 0 through 9, pausing five seconds on each channel until the received-signal line is active more than momentarily or any button is pressed, then stop scanning6. Send gate-pulse-multiplier calibration messages (increment multiplier, decrement multiplier and store multiplier) to PIC 2 when in the gate-pulse-calibration mode7. Send IF-offset-calibration messages (increment IF, decrement IF and store IF) to PIC 2 when in the IF-offset-calibration mode.

PIC 2 Tasks

1. Generate frequency-counter gate pulses;2. Convert PIC 1 LO counts to tuned frequency and display it on the LCD;3. Display the current channel number;4. Adjust and store the gate-pulse multiplier in response to commands from PIC 1;5. Adjust and store the IF offset in response to commands from PIC 1.

Table 2—Serial Command Structure

Command Command Byte # Argument Bytes Example

Display raw decimal number N 2 Nxx, where xx is a 16-bit binary number.Convert LO counts to tuned frequency and display F 2 Fxx, where xx is a 16-bit binary number.Convert LO counts to LO frequency and display L 2 Lxx, where xx is a 16-bit binary number.Display channel number C 1 Cx, where x is an 8-bit binary number

between 0 and 9.Send gate pulse P 0 PBlank display B 0 BDisplay gate pulse multiplier value S 0 SIncrement gate pulse multiplier value by 1 G 0 GDecrement gate pulse multiplier value by 1 g 0 gRewrite gate pulse multiplier to EEPROM W 0 WDisplay text string T Null-terminated eg, THELLO<0> displays HELLO

term FIF accounts for the intermediatefrequency. It is stored in the PIC 2EEPROM and is adjustable.

There are no multiply and divide in-structions in the PIC instruction set,but Microchip provides routines ontheir Web side for both fixed- and float-ing-point operations. Fixed-point rou-tines6 are used in the frequency calcu-lation, with the multiplication donebefore division to provide the greatestprecision.

Setting FrequencyPIC 1 stores a channel’s CLO in

EEPROM when the PROGRAM CHANNELbutton is pressed. The RDAC setting

corresponding to this frequency is notstored. Further, the relationship be-tween RDAC setting and frequencydepends on oscillator stability. Inorder to set the LO to a particular fre-quency, PIC 1 does a binary searchacross the LO tuning range. Begin-ning with the extreme RDAC settings($0000 to $FFFF), the frequency ateach extreme is measured. The RDACsetting at the midpoint is calculated,and the frequency measured at thatpoint. If the desired frequency lies atthe midpoint, tuning is complete. Ifthe desired frequency lies above themidpoint, the low extreme is set tothe midpoint; otherwise, the high

28 Jan/Feb 2000

Fig 5

Jan/Feb 2000 29

extreme is set to the midpoint. This op-eration converges on the desired fre-quency provided tuning is monotonicwith respect to RDAC setting asdetailed above.

Serial ProtocolSerial I/O routines are adapted from

Fig 5 (See left)—PIC programmerschematic diagram. Unless otherwisespecified, use 1/4 W, 5%-tolerancecarbon composition or film resistors.Equivalent parts may be substituted forthose shown.K1, K2—5 V DPDT relay, 5-V coilY1—4-MHz ceramic resonator with capacitors

Fig 6—(A) PIC controller etching pattern, full size 3.87×6.07 inches. (B) PIC programmer etching pattern, full size 2.92×6.21 inches.

(B)

(A)

a Microchip application note.7 Commu-nication is unidirectional, with PIC 1issuing commands. Except for the send-gate-pulse command, PIC 2 does notindicate completion of a command, soPIC 1 must allow sufficient time for thecommand to complete. Table 2 des-cribes the various serial commands.

PIC ProgrammingMPLAB produces the Intel-format

object code files necessary to programthe PICs, but a programmer is required.Programmers are available at reason-able cost from various companies, butthey are also easily built. I chose to

modify a programmer circuit given onthe University of Nottingham’s Micro-chip Web site maintained by SteveMarchant.8 This site covers the pro-grammer hardware (PROG84) andseveral software packages to drivethe programmer. I chose MassimoVeneziano’s PPF84, which runs underDOS and Windows 95. The schematicand PC-board layout for my variant ofSteve’s programmer are shown in Figs5 and 6. Images of the etching patternsare available on the QEX Web site.9

OperationThe controller enters its normal

30 Jan/Feb 2000

Table 3—Operating Displays and Actions

Operating Mode(Display Tuned Frequency and Channel Number)Buttons Pressed Action

FREQUENCY UP Increment frequencyFREQUENCY DOWN Decrement frequencySELECT CHANNEL Go to the next channel and tune to

its stored frequency.PROGRAM CHANNEL Store the tuned frequency for the

current channel in EEPROMSCAN Scan through all channelsFREQUENCY UP + FREQUENCY DOWN Enter gate pulse calibration modeFREQUENCY UP + SCAN Increase frequency incrementFREQUENCY DOWN + SCAN Increase frequency decrement

Gate-Pulse-Calibration Mode Functions(Display Gate-Pulse Multiplier and LO Frequency)Buttons Pressed Action

FREQUENCY UP Increment GPMFREQUENCY DOWN Decrement GPMSELECT CHANNEL Enter IF Offset Calibration ModePROGRAM CHANNEL Store GPM in EEPROMSCAN Return to operating mode

IF Offset Calibration Mode Functions(Display IF and LO Frequencies)Buttons Pressed Action

FREQUENCY UP Increment IF OffsetFREQUENCY DOWN Decrement IF OffsetPROGRAM CHANNEL Store IF Offset in EEPROMSCAN Return to operating mode

operating mode on power-up. Tuningproceeds very slowly, one RDAC countat a time, when either FREQUENCY UPor FREQUENCY DOWN is pressed. Toincrease the tuning speed (ie, RDACincrement and decrement), momen-tarily press SCAN while holdingFREQUENCY UP or FREQUENCY DOWN inplace. Tuning speed is indicated by asingle digit on the LCD; it appearsonly while tuning. Table 3 shows theoperating modes of the controller, theinformation displayed and actionstaken for commands given in eachmode.

Using the Controllerwith Other Receivers

The controller should work with mostHF voltage-tuned receivers. Operationat higher frequencies is limited by thePIC prescaler; operation at much lowerfrequencies probably requires a higher-valued drain choke in the LO buffer.With a 16-bit tuning resolution andlimited tuning voltage swing, thecontroller is probably useful only withsingle-band receivers.

There are certain assumptions inthe software that may need to bemodified for your particular radio:1. The LO is assumed to tune on the

low side such that IF = RF – LO. Forhigh-side tuning, the PIC 2 codemust change to subtract the IFoffset, rather than add it. Fordirect-conversion receivers, the IFoffset is set to zero.

2. The displayed frequency resolutionis 1 kHz, which is adequate for mypurposes in the 10-meter band.Resolution may be increased byincreasing the width of the gatepulse generated by PIC 2, decrea-sing PIC 1’s prescaler ratio, or both,and suitably adjusting the formulathat calculates tuned frequency.

3. It may be convenient to generate areceived-signal indicator that goeslow when a signal is received. If so,the received-signal-detection logicneeds to be reversed.

4. If your application differs frommine, the default values for thegate-pulse multiplier (10,075) andIF (10,916 kHz) should be changedin the PIC 2 code to make initialadjustments easier.

Construction TipsWith two PIC clocks generating

8-MHz square waves and other digitalactivity depending on which buttonsare being pushed, the controller boardgenerates quite a bit of noise. It isimpossible to shield the board and

interconnecting wiring too thoroughly.If you are adding the controller to anexisting receiver, I recommend housingthe two in separate shielded enclo-sures. My receiver and controller aremounted in the same enclosure, and Iam still working on shielding.

If you will be making any changes tothe software, make sure to plug thePICs into 18-pin DIP sockets thatmove with the PIC. There will be onesocket on the controller board, and oneattached to the PIC until softwarework is complete. Tearing off socketpins during repeated insertions andextractions is much less expensivethan tearing off PIC pins. The DIPsockets fit just as well as IC pins intothe zero-insertion-force socket on theprogrammer.

Notes1J. Hansen, W2FS, “Using PIC Microcon-

trollers in Amateur Radio Projects,” QST,Oct 1998, pp 34-40.

2S. Hageman, “A Synthesized 2-Meter FMReceiver with PC Control,” QST, Feb1999, pp 35-40.

3S. Hageman, “PIC Development on a Shoe-string,” QST, Mar 1999, pp 49-51.

4The newer PIC 16F87X series of FLASH-programmable microcontrollers alleviatesthis problem.

5N. Heckt, “A PIC-Based Digital FrequencyDisplay,” QST, May 1997, pp 36-38.

6F. Testa, AN617: “Fixed-Point Routines,”Microchip Technology, Inc, 1996.

7S. D’Souza, AN593: “Serial Port RoutinesWithout Using Timer0,” Microchip Tech-nology, Inc, 1997.

8www.ccc.nottingham.ac.uk/~cczsteve/pic84.html

9You can download this package from theARRL Web http://www.arrl.org/files/qex/. Look for 0001DUNC.ZIP.

Thomas has been a radio amateurfor only a few months, having afterabout 30 years of procrastinating fi-nally developed his code skill enoughto get a Tech Plus license. His creden-tials include BSEE and MSEE degreesfrom Georgia Tech and a First ClassRadiotelephone license issued in 1969,despite which he has never been em-ployed as an electrical engineer or tech-nician. For the past 23 years, he hasbeen a computer programmer for sci-entific and engineering systems.

Jan/Feb 2000 31

ARRL Technical Awards

Nominations are now open for an ex-citing slate of national ARRL awards:• ARRL Technical Service Award• ARRL Technical Innovation Award• ARRL Microwave Development Award

The ARRL Board of Directors hascreated a program of technical awardsto recognize service, innovation andmicrowave development. In takingthis action, the Board answered inter-national, federal and organizationmotivations: International regula-tions recognize the “technical investi-gations” conducted by the AmateurService. The FCC recognizes theamateur’s “proven ability to contrib-ute to the advancement of the radioart,” the need to promote technicalskills and expand the “existing reser-voir . . . of trained operators, techni-cians and electronics experts.” TheLeague’s purposes include promotingexperimentation and education in thefield of electronic communication, re-search, development and dissemina-tion of technical, educational and sci-entific information.

All of these purposes are fundamen-tal to the amateur service and theydeserve recognition. Hence, theBoard’s initiative to encourage andreward achievement in technicalareas. The ARRL technical awards en-courage and tangibly reward ama-teurs who have given outstandingtechnical service. The awards providean opportunity to publicly recognizeamateur’s service, Amateur Radio andits many benefits for society.

ARRL Technical Service AwardThe annual Technical Service

Award is given to a licensed radioamateur whose service to the amateurcommunity and/or society is exem-plary in the area of Amateur Radiotechnical activities. These include,but are not limited to:

• Leadership or participation intechnically-oriented organizationalaffairs at the local or national level.

• Service as an official ARRL tech-nical volunteer: Technical Advisor,Technical Coordinator, TechnicalSpecialist.

• Communication of technical educa-tion and achievements with othersthrough articles in Amateur Radio litera-ture or presentations at club meetings,hamfests and conventions—this includesencouraging others to do the same.

• Promotion of technical advancesand experimentation at VHF/UHF,with specialized modes and work withenthusiasts in these fields.

• Service as a technical advisor toclubs sponsoring classes to obtain orupgrade amateur licenses.

• Service as a technical advisor toAmateur Radio service providers, gov-ernment and relief agencies establishingemergency-communication networks.

• Aid to amateurs needing special-ized technical advice by referral toappropriate sources.

• Aid local clubs to develop RFI/TVIcommittees and render technical as-sistance to them as needed.

• Aid to local technical-programcommittees to arrange suitable pro-grams for ARRL hamfests and con-ventions.

Each Technical Service Award win-ner receives an engraved plaque andtravel expenses to attend an ARRLconvention for the formal award pre-sentation.

ARRL TechnicalInnovation Award

The amateur community has wit-nessed great technological changesover the past 75 years. Amateurs havebeen at the heart of many advances inthe radio art. Enduring ARRL policyencourages amateurs to remain at theforefront of technological advance-ment. The ARRL’s annual TechnicalInnovation Award is granted to alicensed radio amateur whose accom-plishments and contributions aremost exemplary in the areas of techni-cal research, development and appli-cation of new ideas and future systemsof Amateur Radio. These include, butare not limited to:

• Promotion and development ofhigher-speed modems and improvedpacket radio protocols.

• Promotion of personal computersin Amateur Radio applications.

• Activities to increase efficient useof the amateur spectrum.

• Work to alleviate long-standingtechnical problems, such as antennarestriction, competition for spectrumand EMC.

• Digital voice experimentation.• Amateur satellite development

work to improve portable and mobilecommunication.

• Improvements in solar, natural

and alternative power sources for fieldapplications.

• Practical compressed video trans-mission systems.

• Spread-spectrum technology andapplications.

• Technology development to assistdisabled amateurs.

Each Technical Innovation Awardwinner receives a cash award of $500,an engraved plaque and travel ex-penses to attend an ARRL conventionfor the formal presentation.

ARRL MicrowaveDevelopment Award

The microwave and millimeterbands are great frontiers for ama-teurs. They present amateurs with avast test area for development of bothnew and traditional modes. The an-nual ARRL Microwave DevelopmentAward is given to an amateur indi-vidual or group whose accomplish-ments and contributions are exem-plary in the area of microwave devel-opment. That is, research andapplication of new or refined activityin the amateur microwave bands at1 GHz and above. This includes adap-tation of new modes in both terrestrialand satellite techniques.

Each Microwave DevelopmentAward winner receives an engravedplaque and travel expenses to attendan ARRL convention for the formalpresentation.

Nominate Now!Count yourself among those who

know that technical advancement isnot a lost ideal in the amateur commu-nity. Now is the time to nominate yourcolleagues for one or all of the awardsdescribed above!

Formal nominations may be madeby any ARRL member. Submit supportinformation along with the nomina-tion document, including endorse-ments of ARRL affiliated clubs andLeague officials. Nominations shouldthoroughly document the nominee’srecord of technical service during theprevious calendar year. Informationconcerning the character of the nomi-nee should be as complete as possible.The ARRL’s Amateur Radio Technol-ogy Working Group will serve as theaward panel, review the nominationsand select the winners.

Send nominations to: ARRL TechnicalAwards, 225 Main St, Newington, CT06111. Nominations must be received atHeadquarters by March 31, 2000. Sendany questions to Headquarters or e-mailjwolfgang@arrl.org.

32 Jan/Feb 2000

Modern vehicles are RF-noisy environments.Come learn how to identify and silence

your mobile noise sources.

By Stuart G. Downs, P.E., WA6PDPARRL Technical Specialist

10983 Shy Bird LnSan Diego, CA 92128-4017stu.downs@trw.com

Automotive RFI Elimination

[Editor’s note: Readers should checkwith their automobile manufacturers’EMC departments before performingany modifications that may voidvehicle warranties.]

Automotive RFI can drive hamscrazy and I was no exception. Evenwith a good noise blanker, RFI makesit very difficult—if not impossible—tohear weak stations. It is also difficultto identify noise sources using amobile HF radio as a noise-proximitytester, that is, to obtain noise signa-tures by picking up other noisyvehicles, power lines, traffic lights,etc. It is theoretically possible to elimi-nate all automobile RFI noise.

Automotive BackgroundAutomobile manufacturers have mini-

mized conducted and radiated ignition,electric-motor and generator noise pro-blems by adding inductive/resistivespark-plug wires, resistive spark plugsand judiciously placed capacitors. Thiswas done primarily to reduce noise in theAM and FM broad-cast bands. The theorybehind this is quite simple. By placingresistive and inductive distributed ele-ments in a series circuit, voltage risetimes are increased and current isdecreased. In addition, a capacitor placed

between power and ground provides alow-impedance return path for low-frequency noise. The problem is thecapacitor looks inductive (lead reac-tance) as the frequency gets higher. Atsome frequency, the component self-resonates. Rise time is related to thespectral content by the following approx-imation:

(occupied BW)(rise time)

ft

≈ 0.35(Eq 1)

consider the following:

I =dq

dt(Eq 2)

Current flow (in amperes) is the rateof change of charge (in coulombs) perunit time. A constant dc current movescharge (electrons) through a wire and,for all practical purposes creates astatic (unchanging) magnetic field.When the current is time varying, say:

I t K t( ) = ( )sin ωwhere K is the peak magnitude andsin (wt) describes the function of time(ω = 2πf), things start to happen: Atravelling electromagnetic wave isproduced. Taking the time derivativeof Eq 3’s current yields:

2

2= ( )d q

dtK tω ωcos (Eq 4)

This equation represents acceleratingcharge that radiates electromagneticenergy. With dc current, there is no elec-tromagnetic radiation; with a time-vary-ing current, you get electromagnetic ra-diation. See the Appendix for proof.

Solutions to ignition impulse noisehave already been found and success-fully demonstrated in aircraft withinternal-combustion engines. TheVHF, AM aircraft band is especiallysusceptible to ignition noise. Aircraftmanufactures fixed these problems soaviators could use their AM radios.Next time you have a chance to look atan aircraft engine, observe the shield-

(Eq 3)If the rise time increases, the spec-

tral energy content decreases. Withcharge flow lessened by the increasedinductance and resistance, the magni-tude of the radiated electromagneticfield is also reduced. These noise-re-duction techniques were not intendedfor HF ham radios, but they do help alittle. The problem is that it’s notenough. Broadband noise is picked upby our HF rigs! Essentially, we drivearound with spark-gap transmittersunder our hoods, connected to ignition-wire antennas!

To help understand how things radi-ate, let’s look at things in mathemati-cal terms. The time rate of change ofvoltage (dv/dt) and current (di/dt) bothare reduced using the previously men-tioned methods. This is easily derivedfrom the first-order differential equa-tions for resistive or inductive spark-plug wire systems. Keep in mind thataccelerating charge radiates electro-magnetic energy. To understand this,

Jan/Feb 2000 33

ing on the ignition wires.

My Mobile RigAfter getting on HF again following

a long absence, I decided to install mynewly purchased Kenwood TS-570Dtransceiver into my 1995 Toyota 4WD4-Runner. After routing power andground wires directly from the batteryterminals and using a network ana-lyzer to tune my ham-stick antennas(courtesy of IDAFAB’s Glenn Borland,KF6VZK), I thought I was ready formy first QSO driving while hamming.I couldn’t have been more disap-pointed. The RFI level was betweenS-7 and S-9. Murphy had strickenagain. Oh boy—what to do?

Not long after this, back at home Iheard a mobile-7 station on Interstate10 east of Tucson working the Philip-pines. I just about fell out of my chair.Weeks later, I was on a one-day picnictrip to Anza-Borrego Desert StatePark. With the engine turned off, Iheard a station in Uganda on 10 metersand worked him! I never would havebeen able to hear him with the engineon: There would have been too muchRFI. I realized right then that I had toeliminate the noise. Elimination of RFInoise is not black magic, but it followstheory just like everything else. It usu-ally involves a multi-step process.

Conducted and Radiated NoiseTwo different types of noise need to

be considered in automobile RFI con-trol: conducted and radiated. Con-ducted noise is called that because itcomes through wires and other con-ductors directly from the noise source.Radiated noise enters the receiver atthe antenna via electromagnetic fieldsfrom the source. To effectively get ridof any noise source, one must firstunderstand if it is conducted or radi-ated and then determine where it’scoming from. If one has no theoreticalunderstanding, it’s all the more diffi-cult and one is left with the process of“Easter-egging.” That is, you mightget lucky and find the cause by addingall sorts of time-tried solutions.

The noise sources in my automobilewere determined to be in the ignitionsystem, which includes the spark plugs,wires, distributor and coil, dc motorsand a noisy oil-pressure sender.

Conducted-Noise TestsThe radio amateur needs to first

determine if the conducted noise ontheir vehicle’s 12-V power system isproducing noise in the their rig. Todetermine this, do the following:

Fig 1—Braid installed over spark-plug wires at distributor.

Fig 2—Filter cons installed on bracket.

1. Start the engine and switch therig on. This must be done without anantenna because you don’t want toconfuse conducted noise with radiatednoise that is picked up through theantenna. Make sure the air condition-ing, all fans, blowers, blinkers andwindshield wipers are turned off.

2. Turn your transceiver’s modeswitch to AM and listen on HF, thenincrease the engine speed a little.

3. Do you hear any popping, whin-ing, crackling or any other noise thatincreases with engine speed?

4. Turn the engine off. Now turn onthe fans, blowers and air conditioningone at a time.

5. Do you hear any whining?Did you notice any noise in steps 3 or

5? If you did, the noise you heard or saw

on the S meter is conducted throughyour dc bus. If no noise was heard, yourrig is not susceptible to the conducted12-V power noise. If you heard noise insteps 3 or 5 you will need to install alow-pass filter that is capable of operat-ing with your trans-mitter’s load cur-rent, typically about 20 A for a 100-WHF rig. With derating, the filter shouldbe able to pass around 25 A or so with alow voltage drop. Obviously, the filter’sdc resistance needs to be quite low oryou’ll get a significant voltage drop be-tween the battery and your rig.

The ideal filter for this applicationwould be of the L-C variety, containinga series inductor and shunt capacitor.This filter’s forward voltage transferfunction decreases noise on the outputas a function of frequency. Dc output

34 Jan/Feb 2000

current and voltage will not be affected.Do not use a pi filter here. Such a filtermay couple noise from the dc bus downthrough the first capacitor to ground,then up through the second capacitorinto your rig! The inductor side shouldbe attached to the battery and the in-ductor-capacitor node to the radio side.The best type of power filter for thisapplication is called a “filter con.”

FiltersTo reduce RFI, one must understand

filter basics. Filters can reduce oreliminate both conducted and radi-ated noise. Choosing the correct filterfor a particular application can be asignificant task in itself. Say, for in-stance, you want to get rid of noise ona power supply’s output. You thinkthat adding a capacitor will solve theproblem; but after placing the capaci-tor, you find it did no good! Amongother things, filter placement dependson source impedance. Noise from alow-impedance source cannot be re-duced by a capacitor alone. It is sim-ply a matter of voltage division:

out inload

source loadV V

Z

Z Z=

+

(Eq 5)

If Zsource is small, you can take it tozero. Then you’re left with:

out inV V=The point is to think about what is

happening in the circuit. The sourceimpedance may determine what filterto add. To get good attenuation from alow-impedance source, one needs toadd an L-C filter. Batteries are typi-cally low-impedance sources!

To understand this better, think ofa voltage source having zero outputimpedance in series with an imped-ance Z. Z can be resistive, inductive,or some combination of both; it repre-sents the opposition to current flow. Inthe case of the lone capacitor, it had noimpedance to work with. Therefore,the noise currents charged up the capquickly because its impedance waslarger than that of the source.

with a bracket, washer and nut.3. The parasitic reactances have al-

ready been minimized.4. The self-resonant frequencies of the

filters are high.5. They come in various current ratings.

Second-order filter cons typicallyattenuate noise at –40 dB/decade offrequency above the 3-dB point.

-3dBfLC

≈ 1

2π(Eq 7)

Filter cons are good up to a couplehundred megahertz. Typical series-inductance values for 25-A dc power-line filters would be in the range of 100to 200 µH, with a shunt capacitor ofabout 1 to 2 µF. The break-point fre-quencies of these filters are about15 kHz, or so. Essentially, this filterforms an ac voltage divider with thenoise source’s impedance, thus reduc-ing ac noise at its output—more aboutthis later. Smaller filter cons rated at5 A or so have less inductance.

This type of filter can be large de-pending upon the voltage and currentratings: The inductor must not satu-rate. It must support the full loadcurrent, which can be up to 25 A for a100-W transmitter—and don’t forgetderating. Inductor size is directly pro-portional to the amount current flowingthrough it. If the inductor saturates,you have only the wire resistance andlittle additional inductance. In thiscase, you would just have a first-orderfilter composed of nothing more than anR and C—not very good! Take the timeto select the right filter up front.

Make sure that the ground return onyour rig goes right to a solder lug on thebracket as close to the filter con as pos-sible. After the filter is installed, repeatthe above tests to see how effective it is.

Most transceivers have filters on thedc power line coming into the trans-ceiver. In my case, the KenwoodTS-570D did not have a conducted noiseproblem. I did not see any indication onthe S meter, nor did I hear any con-ducted noise. I also looked on an oscil-loscope connected across my batteryand noticed low-frequency ripple simi-lar to what one would see out of a full-wave bridge rectifier: about 100 mV(P-P) punctuated with high-frequency,damped sinusoids. It was obvious thatthe filter in the rig was adequate.

able on the S meter.2. Next, turn your mode switch to AM—

with the noise blanker off—andstart your engine. You’ll probablyhear a popping noise. Note theS-meter reading.

3. Increase the engine speed. Do younotice that the popping noise in-creases with engine speed? The pop-ping noise is impulse noise from theignition system; it is caused by thecurrent flowing in various parts ofthe engine. In another words, yourignition system is radiating, andyour antenna is picking it up. This isthe noise problem that you shouldprobably solve first.If you already have resistive spark

plugs and wires, the voltage andcurrent rates of change dv/dt and di/dtshould not be altered. Something elsemust be done. Grounding the hood willnot eliminate this noise. We mustattenuate the electromagnetic radi-ation without affecting dv/dt or di/dt,both of which are necessary for properengine performance.

Electromagnetic radiation of thissort can be greatly reduced. Maxwell’sequations—the four that describe allelectromagnetic waves—can yieldgreat insight to the behavior of fieldsat boundary conditions. One boundarycondition for E fields occurs as follows:The tangential E-field component of apropagating electromagnetic wavedecreases to zero at the surface of aperfect conductor (metal) and iscontinuous across the boundary. Thatis to say: The E field shrinks to a smallvalue across the skin depth of a metalshield, since there is no perfectconductor. The normal component ofthe H field passes through the metalboundary. The E-field attenuationtypically will be on the order of 60 dBor so. That’s a 1000-to-1 change! Togreatly reduce or eliminate ignitionnoise, therefore, we must shield theignition system—it’s the only way.Don’t worry; it’s a job, but not as hardas you might think.

In shielding my ignition system, Iwanted to ensure that I had good RFgrounds and low-impedance current-return paths. That is, I wanted to becertain that—while each spark-plugfires—the return current travels via ashield over the spark-plug wire. Thecurrent path from the coil through thespark plug wires to the spark plug andback to the coil should be as direct andnon-inductive as possible. Any otherground paths on an engine might lookinductive and present a virtually opencircuit at RF! The plan, therefore, was

Power Filter ConsPower filter cons are nice solutions

to ham radio automotive noise prob-lems for the following reasons:1. L-C (series-L, shunt C), C-L-C

(shunt-C, series-L, shunt-C, a pinetwork) and L-C-L (series-L,shunt-C, series-L, a T network) vari-eties are available at surplus andfrom a variety of manufacturers.

2. The filter components are hermeti-cally sealed in a metal tube; so theycan’t radiate and are easily mounted

Radiated-Noise Tests1. Switch on your rig with an antenna

connected and notice the ambientnoise level with the engine off. It isbetter to do radiated tests on dayswhen the ambient noise is not read-

(Eq 6)

Jan/Feb 2000 35

to use flexible braid as a low-impedanceground path.

Shielding My Toyota’s IgnitionSpark-Gap Transmitter

My ignition system was shielded inthe following manner. First, I hadcopper-tube spark plug covers made forall six spark plugs. They snap into placeover the base of the plugs (groundpotential) with a slight interference fit.I then soldered flexible braid to theoutside of each spark plug’s coppercover. In this way, I was able to snap

the cover over the plug at its base toprovide a good ground, at the same timeplugging the electrical-connection bootonto the spark plug in one motion. Thecopper spark-plug covers were made forme by Tooling Associates.1

There are two common spark-plug-base sizes. Toyota spark-plugs in a4-Runner are of the smaller-diameter. Toshield them, I used 1/2-inch flexible braid(Belden #8669)2 over each spark-plugwire all the way from the spark plug to

the distributor. To increase dielectricstrength, I wrapped the spark-plug wireswith insulated flexible tubing beforeinstalling the braid. With ground nowlocated so close to the wires, I sure didn’twant any high-voltage breakdown. I used3/4-inch braid (Belden #8662) to cover thelarger-diameter boots at the distributor.

Belden braid comes in 50-footlengths, so maybe a club or grouppurchase would be best. You can findthe braid in the Newark catalogue. Sofar, the silicone spark plug wires haveexhibited no high-voltage breakdown.

Appendix: Strength of EM Fields Radiated by Accelerating Charge

Fig A—The situation of Eqs 8 through12.

Fig B—The situation of Eq 13.

The following proof quantifies the field strengths of an electromagnetic waveas created by accelerating charge in a wire [or antenna element—Ed.]. Thisderivation begins with one of the four basic laws of electromagnetism knownas Maxwell’s equations. These laws were formulated as a set by Maxwell, butcertain parts of them were obtained directly from the work of others such asGauss, Ampère and Thomson.

The first law may be stated: “The work required to carry a unit magnetic polearound a closed path is equal to the total current passing through any surfacethat has the path for its boundary.” Vector calculus says the same thing moresuccinctly, here in integral form:

H dl I• = +∂∂∫ conduction

D

t

φ(Eq 8)

where the left-hand side represents a line integral around a closed path, dl isa vector element of length along that path, H is the magnetic-field vector andφD is the electric flux linking the path. See Fig A. The time rate of change of φDis written as a partial derivative to show that the path of integration is not mov-ing in space.

In a wire, there is no E field, so we are left with:

0

2

conduction

2

π

π

rH dl = I

rH

∫ •

=

(Eq 9)

where r is the distance from the wire. But, current is the time rate of change ofcharge, q :

conductionIqt

=∂∂ (Eq 10)

and so:

∂∂qt

2 rH= π (Eq 11)

Now give the charge some acceleration (changing velocity). Taking the timederivative of both sides:

2

2= 2∂

∂∂∂

qr

Htt

π (Eq 12)

This shows that an accelerating charge generates a magnetic field thatchanges with time. A changing magnetic field is related to the electric field byMaxwell’s second law, which may be stated: “The work required to carry a unitelectric charge around a closed path is equal to minus the time rate of changeof the magnetic flux through that path.” Again using integral form:

E dl =tB∫ • −

∂∂φ

(Eq 13)

where E is the electric-field vector and φB is the magnetic flux linking the pathof integration. See Fig B. We have an electromagnetic wave!

Maxwell’s equations may also be written in differential form. An excellentbook that explains this vector calculus stuff in understandable ways is H. M.Schey, Div, Grad, Curl and All That, 3rd edition, ISBN 0-393-96997-5.

1Notes appear on page 36.

36 Jan/Feb 2000

Second, the distributor was zinc“flame-sprayed” to create a volumetricmetal enclosure (shield) only a few milsthick, but still thick enough that the Efield exponentially dies off and meetsthe E-field boundary criteria. This couldalso have been done with the coil. FlameSpray, Inc of San Diego3 graciouslysandblasted and flame-sprayed thedistributor. The distributor had to bemasked so that the metal did not coverareas where its was not needed. Plasticparts are routinely sprayed for EMI/RFIprotection in military and commercialapplications, and the flame-spraypeople are in this business.

Taping of the distributor can be atask; otherwise high-voltage arcingcan occur. I taped the coil myself with1-inch-wide, 1-mil-thick copper tape,which was supplied by Chomerics ofWoburn, Massachusetts.4

Third, I used narrow hose clamps toconnect the ends of the braid to the wiresto the distributor and coil below theinsulating boots. It worked quite well.See Fig 1.

After completely shielding theignition wires, coil and distributor, Istudied the ignition circuitry forsneaky RF paths. Next, I put two filtercons between the +12-V power line,the igniter and the coil. I mountedthem on a bracket with a good RFground next to the coil and igniter. Thereason for this was simple: I did notwant broadband RF energy in the wirefrom the distributor to the coil to getinto my 12-V power line and radiate.

All coils and transformers are self-resonant and start to look capacitive—rather than inductive—at some freq-uency. The Toyota’s high-voltage ignitioncoil is no exception. Remember that allcurrent—even RF—wants to return to itssource. Why not give it a low-impedancepath by making the path short? Mini-mizing the current-loop size minimizesthe radiated energy, as well.

Ah, but what kind of filter conshould I select? In this case, an L-Cfilter con was the optimum choicebecause the output capacitor shuntsthe dc power at the coil and igniterright at its source, creating a veryshort return path for RF. The seriesinductance presents a large inductivereactance that keeps RF off the +12-Vpower line.

Judging from the wire gauge fromthe battery to the igniter and coil, thedc current must be on the order of acouple of amperes. Therefore, I used a5-A filter con that’s about the size of afat pencil and about 1-inch long. Its dcworking voltage is 50. See Fig 2.

Fuel-Pump NoiseWhen I finished the job, I started the

motor, switched the radio to AM andnoticed that the ignition noise wasgone! Now I heard another noise thathad been masked by the first noisesource: a whine that did not changewith engine speed. What was it? —Some kind of motor that wasradiating! Since all other motors andfans were turned off, it could only bethe fuel-pump motor inside the gastank. With the engine on, I pulled thefuse to the fuel system. The noisestopped—I had found it!

Again, I thought about EM theory. Iknew it was motor noise and thatevery motor has a rotor and a stator. Ithought there must be some sort ofbrushes making sparks. Since sparksmake broadband noise and rotors andstators can self-resonate, I theorizedthat the RF energy must be couplingcapacitively through the rotor orstator and looping back through mypower system. After some groundtesting, I found two ground returns forthe fuel pump system: one through thevehicle ground and one via a power-return wire! I wanted to put the filtercons right at the source, but that wasgoing to be tough unless I removed thegas tank—which I did not want to do.As it turned out, there was an accesspanel on top of the gas tank, accessiblethrough the back seat. Again, I wantedto short out any RF return currentsand short out the sneak path.

The place to do this was at the top ofthe gas tank. I mounted two 5-A filtercons on a bracket, in turn connected toa tab through a self-tapping screwmade for holding the gas-tank wires tothe top of the tank. Fuel-pump powerand return passed through both of thefilter cons. By now, you’re probablywondering what kind I used. I used alow-pass pi network. It worked! WhenI turned the engine back on, the noisefrom the fuel pump was gone!

+12-V bus. I saw some noise but thoughtit was too low in frequency to cause whatI was hearing. Obviously, the realsource was something that switched onshortly after engine start and changedafter decreasing engine speed. I thoughtit might be the sender in the fuel pumpand therefore pulled the gauge fuse,opening all sender circuits. The noisestopped! I then knew it was some sort ofsender, but which one?

Then it hit me: Oil pressure rises whenthe engine is first started, until it reachesits proper value and it changes againwith engine speed—some sort of rheo-stat. I pulled the connector from the oil-pressure sender and the noise stopped! Ilooked at my S meter. There was no noiseindication, and I could not hear anyvehicle noise! I removed the sender unitfrom the engine. With help from GlennBorland, KF6VZK, I cleaned the sensorbody of “chem” film and soldered a filtercon onto it. We used a high-power solder-ing iron for this because we wanted thejob done quickly to avoid damaging thesender. This requires a large, hot thermalmass. There was no good place to put abracket. Best was to mount the filter condirectly onto the sensor and that’s whatwe did.

Shake-Down TestSince I had put many hours into

solving my RFI problems, Janice andI decided to go to Anza-Borrego DesertState Park for a picnic and a hike. TheHF rig went along too, although it wasin the back seat. (Never put the rig onthe right-front seat with the girlfriendor wife along—it might find its wayonto the highway!)

With the engine on, there was nonoise indication on the S meter in theAM mode. I quickly turned to SSB. Icould hear all kinds of stations—evenweak ones; I worked several stationsand thought how nice it was to be ableto finally use my rig to its full potentialwithout limiting RFI. We hiked upPalm Canyon and were rewarded asighting of a bighorn sheep—a borrego.It was a great day!

Oil-Pressure-SenderRheostat Noise

Well, by now I had spent considerableeffort eliminating noise problems, butI wasn’t going to quit now! With theengine running for a while, I noticedthat I had no noise for a couple ofseconds—then all heck broke loose. Thenext noise did not whine but crackled.When I increased the engine speed,it stayed nearly constant. When Idecreased engine speed, it went awayfor a couple of seconds, then came back.I wondered what would cause this.

I took out my scope and looked on the

Notes1Contact Pete Bird, 15570 Corte Montanoso,

San Diego, CA 92127; tel 619-672-1949.2Belden Wire & Cable Co, PO Box 1980, Rich-

mond, IN 47374; tel 765-983-5200, fax 765-983-5257; http://www.belden.com/.

3Flame-Spray Inc, 4674 Alvarado CanyonRd, San Diego, CA 92120-4304; tel 619-283-2007; contact Gary Logan.

4Chomerics Division, Parker Hannifin Corp, 77Dragon Ct, Woburn, MA 01888-4014; tel 781-935-4850, fax 781-933-4318; mailbox@chomerics.com; http://www.chomerics.com/. Contact Dennis Hennigan, WA1HOG,extension 4140.

Jan/Feb 2000 37

Digital designers, experimenters and technicians needto look at signal sequences. Their tool of choice is a logic

analyzer. This homebrew adapter and software turna PC into an eight-channel analyzer suitable for

serious work and digital exploration.

By Larry Cicchinelli, K3PTO

15116 Bellmont AveAllen Park, MI 48101k3pto@arrl.net

A HomebrewLogic Analyzer

With the increasing impor-tance of digital communica-tion in Amateur Radio, this

project should be of interest to manyhams. The logic analyzer (LA) de-scribed in this article will meet theneeds of most home experimenters.The hardware design is such that in-dividual builders may make changesand still use the software I have devel-oped. The software is available asshareware. A free version is available;it uses no hardware other than yourprinter port (EPP mode).1

This LA is very similar to earlier com-mercial instruments. It takes a sampleevery clock cycle and stores the data fordisplay. The memory depth defines thenumber of samples, as well as the totalsample interval. In this case, thememory size is 131 kB. Therefore, witha sample period of 200 ns, the sampleinterval is approximately 26 ms.

Newer LAs have much faster sampleclocks—usually with periods of picosec-onds. Instead of storing data at eachsample period, they store data when-ever a state change is detected. Thetime of the change is stored along withthe data. This allows much greater flex-ibility in the data analysis, but alsorequires more memory, because thetime and value must be stored withhigh precision for each sample.

Some of the general specificationsare:

• 8 channels• 5-V logic only• Time resolution from 200 ns to

1 ms per sample• Windows 3.1 or Windows 95 com-

patibility• Enhanced parallel port required• User configuration via .INI file• Pre-Trigger capability of 4096

(212) samples• Memory depth of 131 kB (217 bytes)• Circuits built with HCMOS parts—

usually good to at least 25 MHz.The operation of each circuit is de-

scribed so that the reader might gainsome insight as to how they work, as wellas ideas for modification. A few possiblesuggestions for modifications are:

• Increase clock speed to 20 MHz toallow 50-ns sampling

• Change the Pre-Trigger memorysize

1You can download this package from theARRL Web http://www.arrl.org/files/qex/. Look for LA0999.ZIP. It includes acomplete set of documentation.

38 Jan/Feb 2000

• Allow more sampling-rate choices,including an external input

Here are a few notes on the circuitdescriptions. The suffix characters ofthe signal names have the followingmeanings:

PL = Pulse Low, the active level is apulse with a steady-state value of 5 V,but pulses low to 0 V for a short periodof time

PH = Pulse High, similar to above,but pulses high

L = Active level of 0 VH = Active level of 5 V

timing values are usually indicatedwith Hex notation ($FFF, for example)

I use this system because I do a lotof programming in C; only those sig-nals that go between the circuits havebeen named.

The strip menu has a DEBUG entry,which allows the builder to test thesystem. I did most of my hardware de-bugging using this facility. Several ofthe circuits can be tested as they arebuilt. Not all of the ICs need be in-stalled before testing begins. A sug-gested construction order is: Control,Time Base, Address Generator, Input,Trigger, RAM and Pre-Trigger. TheTime Base and Address Generator canbe built and tested independently ofthe Control circuit, if desired.

The HELP menu is not available atthe time of this writing. I hope to haveit ready by the time this article hits thestreets. It will contain much of the op-erational information in this article, aswell as details on the DEBUG options.

I am very interested in any feedbackthat anyone may have about the op-eration of this system. I will try to in-corporate any changes that enhanceits operation.

Theory of Circuit OperationFig 1 shows a block diagram of the

LA. Refer to it and the individual cir-cuit schematics as you read.

Address GeneratorThis circuit (Fig 2, built on a daugh-

ter board) is constructed using syn-chronous binary counters. As such,they generate a binary sequence creat-ing the address information for boththe RAM IC and the Pre-Trigger ad-dress. The advantage of a synchronouscounter is that all outputs change stateon the same clock edge. This insuresthat there is no delay accumulationfrom the least to the most-significantstages.

The operation of the circuit is as fol-lows: Until a trigger signal occurs,U404 and U405 are disabled, thus

allowing U401 through U403 to formthe address counter for the Pre-Trig-ger memory. U401 through U403 startcounting immediately following theAddrReset-PL signal. They will con-tinue to count until they reach $FFF(409510), at which point the count rollsover, and begins again at $000. Thiscontinues until a trigger signal occurs.At that time, both the Triggered-PLand the Triggered-H signals willoccur. The Triggered-PL will presetthe address counter to a value of$21000. Please note two significantfacts about this value: (1) the $20000bit, bit 17, is not used as an addressbit, and (2) the address is one countbeyond the Pre-Trigger range.

The Triggered-H (via HighAddrEna-H) signal enables the last two ICs inthe circuit. Eventually the count willreach $3FFFF and roll over to $40000.

This inhibits any further counting,because the AddrDone-L signal, de-rived from bit 17, is fed back to one ofthe enable inputs of U401, the first ICin the counter chain. The AddrDone-Hsignal is used to tell the PC that all thesamples have been stored.

When the system is sampling theinput data, the AddrClk-PH signal isderived from the Time Base circuit.When the data needs to be dumped tothe PC, the signal is derived from thePC via one of the control-port bits.

Note that the same signal is used toadvance the counter as well as to storethe data value into RAM. This is validbecause the RAM chip does not requireany data hold time to store the data.Therefore, by the time the addresslines have changed (specified as 30 ns,typical), the RAM is done with thedata for the address just ended.

Fig 1—Logic Analyzer block diagram.

Jan/Feb 2000 39

Fig 2—Address Generator schematic diagram.

When the system is not sampling in-put data, the clock signal for thecounter chain is derived from theStrobe. The AddrReset-PL signal isdelayed by an RC delay circuit inorder to ensure that the Load signal ofU405 is still low when the clock signalgoes high. This is required becausethe 74HC161 has a synchronous loadcircuit, meaning that the counterperforms its load operation only whilethe load input is low and when theclock goes high.

ControlThis section (see Fig 3) contains sev-

eral different functions:1. U104 is used to select whether the

PC needs to read from, or write to, thesystem. It generates four Read Enablesignals (only three of which are used)and four Write Pulses, which are usedto latch control data from the PC.

2. U101 and U102 latch the LA con-trol signals Time Base Select, SampleEnable and Address Reset

3. U105 controls the clock signalsfor the Address Generator and theRAM as well as the Count Enable forU404 and U405

4. U106A allows triggering to begin.This insures that the Trigger circuitwill not be enabled until the Pre-Trig-ger RAM is full.

InputRefer to Fig 4. The 74HC14 is a

Schmitt trigger. As such, it gives theLA some degree of immunity fromnoise on the input signals. These sig-nals are delivered to both the RAMand the Trigger circuits. Since theRAM data ports are bi-directional,this circuit also makes the RAM dataavailable to the PC.

Please notice that this input circuitis valid for 5-V logic only! If you needany other logic levels, you must condi-tion the signals before applying themto this circuit, or the ICs may be dam-aged. The input resistors (1 kΩ) givesome degree of protection, since thereare diodes within the 74HC14 that areconnected from the input to bothground and power. The resistors areplaced close to the input pins to mini-mize the effect of input (specified as10 pF, maximum) and wiring capaci-tances. There is also a set of pull-downresistors to ensure that the HCMOSinputs never float.

Pre-TriggerRefer to Fig 5. When a trigger signal

occurs, the value of the next addresswill be stored in the Pre-Trigger

address latch by the Triggered-H sig-nal. This value is made available tothe LA program so that it can deter-

mine the address—in the range$00000 to $00FFF (0 to 409510)—atwhich the trigger event occurred. The

40 Jan/Feb 2000

program can then display the datavalues in the correct sequence for thePre-Trigger display.

RAMThe RAM device (see Fig 6) was cho-

sen mainly because it is available in aDIP. This device has a minimum clock-pulse-width requirement of 60 ns.This does not allow the system tooperate at its fastest sample rate of10 MHz (100 ns). It will work at 5 MHz(200 ns), which is fast enough for mypresent requirements. A very simplecircuit modification would accommo-

Fig 3—Control circuit schematic diagram.

date another memory device having afaster response time. It may be assimple as just plugging it in—with thesignals routed appropriately.

When sampling the input data, theSampleClk-PH signal is applied to theR/W line. When reading the data backto the PC, the R/W signal is at a staticlevel generated by the control logic.

Time BaseWhen the system is sampling the in-

put data, this circuit (Fig 7, built on adaughter board) generates the sampleclock used to advance the address

counter, as well as the signal that writesthe data into the RAM. Although onlythree bits are currently implemented toselect the sample rate, there are five bitsavailable. This allows up to 32 selec-tions. The .INI file allows the user todefine the sample period for each imple-mented selection. The primary reasonfor allowing two more bits, is that mul-tiplier values of 1, 2 or 5 may be imple-mented. This gives sample periods thatare similar to those of an oscilloscope.

TriggerThe Trigger circuit (Fig 8, built on a

Jan/Feb 2000 41

daughter board) is used to generatethe signals that enable the system tobegin storing the input data beyondthe Pre-Trigger address range. It al-lows the user to specify the TriggerState for each of the eight input linesindividually. Any one of three valuesis allowed: 0, 1 or “don’t care.” Theoperation is such that when the sys-tem is enabled by the user, the Ad-

Fig 4—Input circuit schematic diagram.

dress Generator will generate onlythose addresses that are defined as thePre-Trigger range. When a triggerevent occurs, the Address Generatorwill still increment to the next ad-dress. This address is stored in thePre-Trigger latch circuit. The triggerevent will always be displayed atsample 409510. The next sample willbe stored at address $01000 (409610).

Table 1—Trigger States

Bit Trigger1 0 State

0 0 always0 1 never1 0 01 1 1

42 Jan/Feb 2000

See the timing diagram, Fig 9.The Trigger State selection is ac-

complished using two control bits foreach input. The least-significant bit,bit 0, is used to define the trigger in-put level as being either a “1” or a “0.”The most-significant bit, bit 1, is usedto enable or disable the input signal tothe comparator. Table 1 is a truthtable describing the operation. Pleasenote that the Trigger circuit is syn-chronous. This means that the triggerevent will only be detected at the ris-ing edge of the sample clock.

ConstructionI have not presented construction

details because individual buildersmay choose among several construc-tion methods. Nonetheless, here issome general information on the cir-cuit as I built it: Three of the circuitswere built on daughter boards thatplug into the main board via pin-stripheaders and sockets. I chose thismethod for two reasons. (1) The larg-est circuit board on hand was not largeenough to hold all the ICs. The one inthe parts list would also require theuse of daughter boards. Larger onesare available from the same vendor.(2) Those circuits that are on plug-inboards can be easily replaced in orderto experiment with modified designs.

Several construction details havebeen deliberately omitted, such as thepower supply, cable to the PC, probecable, mounting methods, etc. Build-ers should determine the best methodsfor their own needs.

LA Front Panel ControlsRefer to Fig 10 while reading these

topics.

Starting Sample NumberEnter the position of the desired

sample number at the left-hand edgeof the display window.

Ending Sample NumberEnter the position of the desired

sample number at the right-hand edgeof the display window. Note: The num-ber of samples viewed at any one timeis limited to the range: 5010 ≤ numberof samples ≤ 30,00010.

CursorMove the cursor to the desired

sample. The cursor may also be movedby using the mouse to click on a loca-tion within the display area.

The above controls allow the userto type the desired number into theappropriate input box. However, the

Fig 5—Pre-trigger circuit schematic diagram.

Fig 6—RAMcircuitschematicdiagram.

Jan/Feb 2000 43

entered value does not take effect untilthe ENTER key is pressed. This mode ofoperation was chosen in order to disal-low what might be illegal values gener-ated by partial entries. In order to givethe operator some visual feedback, thebackground color of the box changesuntil the ENTER key is pressed.

Slide UpMove the display window to the set of

samples that begin at 90% of the endingsample number, keeping a constant num-ber of samples in the display window.

Slide DownMove the display window to the set

of samples which begin at 10% of theending sample number, keeping a con-stant number of samples in the displaywindow.

Fig 7—Time Base schematic diagram.

Cursor Memory Check BoxesThese boxes allow the user to store

the current cursor location. As the cur-sor is moved, the Distance-to-Cursorvalues are updated. This allows theuser to make time-difference mea-surements.

DrawUnder most circumstances, the pro-

gram knows when to draw the wave-forms. However, the waveforms de-velop “holes” when the cursor ismoved. This control is then used toredraw the waveforms. I do not cur-rently know of any method withinVisual Basic that will allow the cursorto move within the display windowwithout leaving these holes. Warn-ing—The time it takes to display thewindow is directly proportional to the

number of active channels and thenumber of samples to be displayed. Itcan take several seconds to draw onesignal if there are 30,000 samples.This is processor dependent—I cur-rently have a 166-MHz ’686.

TriggerThis signals the hardware to ac-

quire another sample set.

Notes on the FreeVersion of the Program

Only two channels are enabled: D0and D1 of the printer port. No exter-nal hardware is supported. Triggeringoccurs after the Trigger control hasbeen pressed and a change is detectedin either or both signals.

Larry Cicchinelli, K3PTO, was firstlicensed in 1961. He received a BSEE

44 Jan/Feb 2000

Fig 8—Trigger circuit schematic diagram.

from Drexel Institute of Technology in1969 and a MS in engineering sciencefrom Penn State in 1981. Larry hasworked at Visteon Automotive Systems

(Ford Motor Co.) since 1967. Most of hiswork (the first 25 years) has been thedesign and development of automatictest equipment and automotive elec-

tronic subsystems. His other activitiesinclude church and reading. Larry alsowrote “A Dip Meter with Digital Dis-play,” QEX, Oct 1993, pp 19-21.

Jan/Feb 2000 45

Fig 9—Trigger-circuit timing diagram.

Fig 10—Logic Analyzer main screen capture.The symbols used in the diagram: “_” is 0 V, “=“ is 5 V, “X” is “don’t care;” “..”indicates time duration.

Appendix: EPP-Mode Printer PortUsage for the Logic Analyzer

Data Port (read and write): Base + 0bits 0 .. 7: PPDR-Bit0 .. 7 = pins 2.. 9

Status Port (read only): Base + 1bit 6: AddressDone-H = pin 10bit 7: Triggered-L = pin 11

Control Port (write only): Base + 2bit 0: strobe = pin 1bit 1: RegSel-0 = pin 14bit 2: RegSel-1 = pin 16bit 3: RegSel-2 = pin 17bit 5: Enable bi-directional port

(0 = write,1 = read)—this signal is notavailable on the printer connector.The design implements two sets ofhardware registers that allow the DataPort to be used to both read and write,based on the Register Select bits:Register 0 Write: Control

bit 0: SampleEna-L, SampleEna-Hbit 1: AddrReset-PL

Register 1 Write:TimeBaseSel-0 .. TimeBaseSel-4

Register 2 Write:TrigSel-00 .. TrigSel-31

Register 3 Write:TrigSel-40 .. TrigSel-71

Register 4 Read: not usedRegister 5 Read:

ReadReg5-L:stored data from RAM(strobe is used to advance the address)

Register 6 Read:ReadReg6-L:low eight bits of Pre-Trigger address

Register 7 Read:ReadReg7-L:bits 0 .. 3: upper four bits of

Pre-Trigger address

46 Jan/Feb 2000

Parts List

Vendors: M = Mouser; RS = Radio Shack; DK = DigiKey; J = Jameco

Part QtyVendor Part Number

74HC126 2 M 511-M74HC12674HC14 3 M 511-M74HC1474HC74 1 M 511-M74HC7474HC151 1 M 511-M74HC15174HC155 1 M 511-M74HC15574HC157 1 M 511-M74HC15774HC161 5 M 511-M74HC16174HC4518 2 M 511-M74HC451874HC273 6 M 511-M74HC27374HC540 1 M 511-M74HC54074HC541 3 M 511-M74HC54174HC688 1 M 511-M74HC68847-µF, 10-V capacitor 4 M 140-XRL10V470.2-µF capacitor 20 M 21RZ320PC Board, 4.5 × 8” 1 M 574-3677-6PC board 3 RS 276-168BPin-strip header, right angle 3 M 517-5111TNPin-strip header, straight 1 J SMH36Pins for SCH36 100 J FCH1

Part QtyVendor Part Number

9-Pin D subminiature, chassis male 1 DK A2043-ND9-pin D subminiature, female cable 1 M 156-13099-pin D subminiature hood 1 RS 276-153925-pin D subminiature, male chassis 1 M 571-74791228-pin socket 7 M 571-2640357314-pin socket 1 M 571-2640463316-pin socket 10 M 571-2640358320-pin socket 11 M 571-2640464332-pin socket 1 M 571-26440183Housing, 36 pin 3 J SCH36Resistor Net, 47 kΩ 1 M 266-47KResistors, assorted 1 RS 271-312SW, DPDT-2P 1 J 2197950’ wrapping wire 1 RS 278-501Box, Aluminum, 10×6×3.5 inches 1 DK L108-NDTest Clips, set of 6 2 DK 923848TC551001BPI (1MB RAM) 1 DK TC551001BPL-70-ND10 MHz oscillator 1 DK SE1208-ND

Notes1. Use of headers, housings and PC boards depend on the construction techniques chosen by the builder.2. Many parts can be obtained from several vendors

Jan/Feb 2000 47

Here is a complete description of the AF board.

By Mark Mandelkern, K5AM

5259 Singer RdLas Cruces, NM 88005k5am@roadrunner.com

A High-PerformanceHomebrew Transceiver:

Part 4

1Notes appear on page 56.

Designing and constructing theAF board of a radio is a rela-tively easy task. Op-amp

circuits are perfectly predictable,special features can be added with littleeffort; signal levels are high and easilymeasured with simple equipment. TheAF board in this radio contains the BFOamplifier, product detector, noise limit-er, speech amplifier, balanced modula-tor and CW sidetone oscillator.

Part 1 gave a general description ofthe K5AM homebrew transceiver,built for serious DX work and contestoperating.1 Parts 2 and 3 described theIF and RF boards.2, 3 This article gives

circuit details for the AF board. Ageneral description of this board wasgiven in Part 1; the board is shown herein Fig 1. The block diagram in Fig 2shows the arrangement of the variousstages.

FeaturesSome of the special features of the

AF board are:• High-pass transmit audio filter to eliminate hum.• 60-Hz receive audio filter to eliminate hum.• Electronic attenuators in all circuits with front-panel controls to avoid routing audio signals to the panel.• Class-A audio output stage with only 0.3% total harmonic distortion.• Noise limiter for reduction of atmospheric static.

Circuit Description:BFO and Carrier Amplifiers

Fig 3 gives the schematic for thestages used to provide LO injectionpower for the product detector andbalanced modulator and that providea carrier for CW operation. Specialeffort was made to prevent carrierleakage into the IF board during SSBoperation. Both a switched MOSFETand a diode switch are used to gate thecarrier. The total residual hum, noiseand carrier on the transmitted SSBsignal are more than 65 dB down.

Receiver CircuitsFig 4 is the schematic of the stages

used to process received audio signals.Signals arrive from the IF board atterminal σ9R1 and are fed directly tothe product detector. Measuring LO

48 Jan/Feb 2000

injection at a DBM (doubly balancedmixer) is difficult with simpleequipment. The diodes in the DBMcause distorted scope patterns at theLO port that are not easy to interpretas to power level. The 47-Ω resistor atthe product detector’s LO port helpsisolate the port and allows meaningfulreadings at the test point indicated.The 2-V(P-P) level specified at the testpoint is estimated to correspond to a+4 dBm LO injection level. This isappropriate, although the device israted at +7 dBm (see the discussion inPart 3).

A special low-noise op amp wasselected for the first AF amplifier. Carewas taken to install this low-noiseamplifier immediately adjacent to theproduct detector to avoid hum pick-upon a connecting cable. The gain of thelow-noise op amp is adjusted for 300 mV(P-P) at terminal σPD. The AM and FMdetector circuits are adjusted similarlyand the three JFET gates are used toselect the appropriate detector.

Noise LimiterAs hams painfully know, noise

blankers do not blank static caused byatmospheric thunderstorms or coronanoise produced by high-tension powerlines. In the eternal struggle to copy DXsignals (no matter how weak) undernoisy conditions (no matter how severe)hams have always been eager to try anyconceivable noise-reduction method.

In times past, when AM was theprimary mode for voice work on the ham

Fig 1—Top view of the AF board in the K5AM homebrew transceiver. At right, fromthe top, are the balanced modulator, the BFO amplifier, the carrier amplifier for CWand the product detector. The two-stage transmit high-pass AF filter is included inthe balanced-modulator compartment, eliminating any possibility of hum pick-upon a connecting cable. Similarly, the first receiver AF amplifier, a special low-noiseop amp, is included in the product-detector compartment. All other AF circuits arein the large compartment at the upper left; the lower left compartment is for theAM/FM circuits. (See Note 8.)

Fig 2 (see right)—AF board blockdiagram. Signals from the IF boardarrive at terminal σ9R1. After detection,amplification and filtering, the audiosignal travels from output terminal σAFto the audio-output module on the rearpanel. For transmitting SSB, there aredual speech inputs: a MIKE jack on thefront panel and a line-level jack on therear panel. The DSB output is at terminalσ9DSB leading to the IF board, the SSBfilters and RF speech clipper.Potentiometers labeled in all capitalletters are front-panel controls; othersare circuit-board trimpots for internaladjustment. An explanation of theterminal designators is given in Part 2,Table 1. The control lines are providedby the logic board.

bands, receivers had noise limiters.One common type was the shuntlimiter, a diode clipper applied to thedetected AF signal. In some radios, theclipping level was adjustable. Otherradios had an automatic noise limiterthat used the detected carrier—a dcsignal sometimes also used to drive theS-meter—to establish an appropriatebias level for the clipping diodes.

Modern commercial receivers haveno noise limiters. Many hams, espec-ially 160-meter DX hounds, have useddiodes on the headphone line to clip thenoise. This can be very effective. Adisadvantage is that the degree of clip-ping depends on the AF voltage level atthe headphone jack, the headphoneimpedance and the AF gain (AFG)setting selected by the operator. Themain disadvantage is that to obtain afair amount of clipping, the receiver AFoutput level must be much higher thannormal; this is likely to cause distortionin the receiver AF circuits, degradingsignal intelligibility.

These problems are avoided byputting the noise limiter inside theradio, ahead of the AF GAIN control.Clipping at low AF levels is a simplematter. With no AM carrier to set thebias level, we cannot expect an auto-matic noise limiter, but we do want anadjustable noise limiter. We need apanel control to set the clipping levelaccording to conditions. In Part 1,Fig 3, the NOISE LIMITER control is justbelow the meters, third knob from theleft. The limiter is quite effective,

especially with atmospheric static forweak low-band DX signals.

“Limiter” is a good term. It remindsus that the circuit cannot eliminatethe noise, as a good blanker can undercertain conditions, but that it onlylimits the noise to a level set by theoperator. This level is usually the peaklevel of the desired signal, so that thenoise and the signal are evenlymatched in a fair contest to register inthe operator’s brain.

Receiver sensitivity measurementsare based on the minimum discerniblesignal (MDS) specification. This meansthat the (S + N)/N is approximately3 dB. Experienced DX operators know,however, that this ratio is no “mini-mum signal”—it is “arm-chair copy,” orin other words, “loud.” DX operatorscan copy CW signals that are wellbelow the MDS. Thus, the limiter canturn barely detectable signals intoreasonably good copy.

Don’t expect the noise limiter toeliminate thunderstorms! Its effective-ness is only apparent on a narrorange of weak DX signals, those withina few decibels of the ambient noiselevel. When signals fade below thatlevel, nothing will help; when signalsbuild up a few decibels, a noise limiterwon’t be needed. Still, most newcountries added to an oper-ator’s 160-meter DXCC list involve signals withinthat narrow range.

The noise limiter is normally usedwith the AGC on. This is the safestoperating method (see Part 1, pp 21-22). Receiver gain is enough thatambient antenna noise on any bandactivates the AGC system. Thisinsures that the AF level into the noiselimiter is essentially constant. Theresulting headphone volume is thusalso constant. Signals do not “jump outthe noise,” as is often said aboutreceivers reported to be “quiet.” Suchreceivers merely have inadequate

Jan/Feb 2000 49

50 Jan/Feb 2000

Fig 3—BFO and carrier-amplifier schematic diagram. Except as noted, each resistor is a 1/4-W, carbon-film type. All trimpots areone-turn miniature types, such as Bourns type 3386; Digi-Key #3386F-nnn. The unmarked coupling and bypass capacitors areall 10 nF disc ceramic types. Also, each control and power terminal has a bypass capacitor that is not shown. Electrolyticcapacitors have 25-V ratings and values given in microfarads; those less than 100 µF are tantalum types. Capacitors labeled“s.m.” are silver micas, with values given in picofarads. The values of RF chokes (RFC) are in microhenries. Part-suppliercontact information appears in Notes 9, 10, 11 and 12.

The MOSFETs are all small-signal, VHF dual-gate types. Type 3N140 is used here, but any similar type may be substituted.Replacement-type NTE 221 is also available. The JFETs are J310s. Except where otherwise indicated, the diodes are small-signal silicon types, such as 1N4148; the bipolar transistors are type 2N2222A. Except as noted, the op amps are sections ofLM1458Ns. Unless shown otherwise, each op amp is powered by the +15 V and –15 V rails. Not shown is the filtering at each opamp power terminal. This varies with the application. The minimum is a 100-nF, monolithic ceramic bypass. More sensitive andlow-level circuits also use a 100-Ω isolation resistor and a 10-47 µF tantalum electrolytic at the terminal.

Potentiometers labeled in all-capital letters are front-panel controls; others are circuit-board trimpots for internal adjustment.The control lines are provided by the logic board. The signal levels indicated at various test points are scope readings.

L1—FT-37-61 ferrite toroidal core, 14 turns of #26 AWG enameled wire.

Fig 4—(see right) Receiver AF circuitschematic diagram. For general noteson the schematics, refer to the captionfor Fig 3. Capacitors labeled simply “1”are 1-µF monolithic ceramics (see Note10). The DBM used as a productdetector may be obtained in smallquantities directly from themanufacturer (see Note 12). Signallevels indicated on the diagram aregiven with the AF GAIN set for 2 Woutput with an 8-Ω resistive load at thespeaker jack.C—Polypropylene capacitor, 2.7 nF, 2%tolerance. Panasonic #ECQ-P1H272GZ,Digi-Key #P3272. Where 2C is indicated,use two of these in parallel; thisensures the closest match.C1—Polypropylene capacitor, 1 nF.Panasonic #ECQ-P1H102GZ, Digi-Key#P3102.C2—Polypropylene capacitor, 470 nF.Panasonic #ECQ-P1H474GZ, Digi-Key#P3474.

AGC systems; they are not necessarily“quiet” with regard to sensitivity ornoise figure. Listening for weaksignals in the noise on such receiversrequires one to set the gain too high,creating an aural hazard when astrong signal suddenly appears. Theflat AGC characteristic in this receiverallows the noise-limiter clipping levelto be easily set. If “noise-jumping”signals are desired at other times, theIF GAIN control may be used to raise theAGC threshold (see Part 1, pp 21-22).

The noise-limiter circuit is shown inFig 5. Op amp U1 raises the receiver AFlevel from 3 V to 14 V (P-P), while opamp U2 reduces it to the former levelafter the clipper. When clipping is used,however, the AFG control will need tobe increased. With both controls on the

C3-C4—Polypropylene capacitor, 220nF. Panasonic #ECQ-P1H224GZ, Digi-Key #P3224.R—Metal-film resistor, 1 MΩ, 1%tolerance. Digi-Key #1.00MXBK. WhereR/2 is indicated, use two of these inparallel; this ensures the closest match.R1—AFG potentiometer, 5 kΩ. This hasbeen replaced five times in the lasteight years. The current resident, a 2-Wwirewound type found at a hamfest,seems headed for a long life.U1-U2—Electronic attenuator, MotorolaMC3340P; discontinued. Available asreplacement component NTE #829,Mouser #526-NTE829 (Mouser). Adatasheet is available on the Web:http://www.mot-sps.com/books/dl128/pdf/mc3340rev6.pdf. Omitted from theschematic is the filtering at the +15-Vsupply terminal: a 100-Ω isolationresistor, a 100-nF monolithic ceramiccapacitor and a 47-µF tantalumelectrolytic capacitor at pin 8.

Jan/Feb 2000 51

52 Jan/Feb 2000

front panel, this is convenient; theclipping level is normally set accordingto conditions and not changed until thethunderstorm is over.

The clipping level is set by the NOISELIMITER panel control. Voltage followerU3 is required to provide a low-impe-dance negative bias source for thenegative-clipping diode. Inverter U4tracks U3 and provides reverse bias tothe positive-clipping diode of equalmagnitude but opposite polarity. Thisinsures that the positive and negativehalves of the signal waveform areclipped equally. When the panel controlis fully clockwise, the diodes have noreverse bias; each clips to 0.7 V. Thusthe 14-V (P-P) level allows a maximumclipping depth of 20 dB. With thecontrol fully counterclockwise, thediodes are completely reverse-biased,and the noise limiter is effectively off.

When noise limiting is needed forCW, the control is usually set nearmaximum; distortion on CW signals isof no consequence. For SSB, the fullsetting causes noticeable distortion,but signals are very intelligible. Forless distortion, retard the controlsomewhat. With 6 dB of clipping, thedistortion is barely noticeable. Atthose times when signals can be heardthrough the static without the limiter,use of the limiter makes listening farmore pleasant and reduces fatigue.Also, the limiter can be left at 3-6 dBat all times; it won’t noticeably effectthe signals, but will reduce the noisefrom key clicks and splatter.

When a DX signal is too weak to hearduring a static crash, the noise limiterstill serves a valuable function andmay enable an otherwise impossiblecontact. It saves the operator’s ears!This is both a long-term safety featureand a short-term operating aid. Anoperator’s hearing ability is depressedafter some minutes of listening to loudnoise. With the AFG high enough tohear a weak station between staticcrashes, one’s hearing threshold maybe degraded within ten minutes. Onthe other hand, if the gain is lowenough so that static crashes don’tcause desensitization, a weak stationwon’t be heard. The solution to thisdilemma is an adjustable noise limiter.

The noise-limiter schematic dia-gram is separated from the maindiagram so that it might be built andinstalled in other receivers. Detailsand pin-outs are given. If the receiverAF level at the point of insertion isother than that specified the gain ofthe input and output amplifiers may bechanged easily.

Fig 5—Noise-limiter schematic diagram. For general notes on the schematics, referto the caption for Fig 3. The 1-kΩ resistor at the output of U1 is needed to preventexcessive current drain on U1 when the diodes conduct. The gain-setting resistorsfor U4, while they might be of any (equal) value to set the gain at –1, should be ofno lower value than shown. The circuit of U4 establishes a virtual ground at theinverting input; a low-value input resistor would unduly load the potentiometercircuit, resulting in less available reverse bias. At the full counterclockwiseposition of the control, a reverse bias of 9 V is presented to the diodes. This isenough to cut off the diodes, since the peak AF voltage is 7 V in either direction.U—Quad op amp, LM348N. Each of the four op amps is shown separately. A fifthsymbol shows the power connections, but all is contained in a single, 14-pin DIPpackage. Note that while type LM324N is used in other parts of the radio, it is asingle-supply type vulnerable to crossover distortion. It is useful for dc-controlcircuits because it has a wider output voltage swing, but it should not be used forAF. The dual LM1458N and quad LM348N types used on the AF board are standardlow-distortion 741 types.

This noise limiter has been veryhelpful in DX work. One can onlywonder: “Why don’t all receivers havenoise limiters?”

DSP Loop for External FilterThe DSP loop was added only a few

years ago to accommodate a TimewaveModel DSP-9+ filter. The loop ispositioned before the AF GAIN control.This is done so that the AFG knob onthe radio can be used independently ofthe filter. This avoids the usual cum-bersome arrangement wherein theAFG on the radio must be continuallyreadjusted for the proper input leveland the AFG control on the externalfilter used to control headphonevolume. This is the main reason foradding the DSP loop. There are alsoother problems:• There is a 6-dB loss in the external filter between the line input and line output jacks.• Signal levels are low and hum can be introduced.• External devices installed on the speaker line can suffer from annoy- ing RFI during transmissions.

The solution is to raise the AF level20 dB for the loop and use an external

amplifier to compensate for the loss inthe filter. A simple 20-dB pad in theradio at the loop’s return jack restoresthe AF signal to the original level.

The external amplifier is shown inFig 6. It has controls to set the properlevels both to and from the filter.These need be set only once: The filtercan be switched in or out at any timewith no readjustment of receiver,amplifier or filter controls. This proce-dure eliminates the need for any modi-fications to the external DSP filter andallows adjustment for a variety ofdifferent external units. A switch onthe rear panel of the radio cuts the loopout of the circuit. This avoids the needfor a patch cable when the externalfilter is not connected. Whenconnected, the filter is switched in andout by its own controls.

Electronic AttenuatorsTo avoid routing audio signals to

front-panel controls and possible humpickup, electronic attenuators are usedfor the AFGAIN function and all otherfront-panel audio level controls. Theattenuators [MC3340Ps —Ed.] providea 60-dB range, with a pin 2 controlvoltage range of 3-6 V. The maximum

Jan/Feb 2000 53

Fig 6—External-DSP-loop-amplifier schematic diagram. For general notes on theschematics, refer to the caption for Fig 3. The amplifier may be housed in a smallaluminum box. Defense against RFI from the transmitting antenna is not critical,since the amplifier and filter are positioned before the receiver muting gate.The amplifier is powered by the same 13.6-V dc power supply used for the numerousoperating-bench auxiliary gadgets. The amplifier is fitted with a 6-foot shieldedpower cable terminated in a two-pin Jones plug. The 13.6-V dc fused distributionboxes on the operating bench contain two-pin Jones sockets. These connectors aresolid, reliable and easy to use. Since the shack has them only for this purpose, thissystem is safer than one that uses other common connectors, such as phono plugs.P1—Jones plug, two-pin, Cinch #P-302-CCT, Digi-Key #CJ102P. Mating socket, Cinch#S-302-AB, Digi-Key #CJ302S.

Fig 7—Sidetone-oscillator schematic diagram. For general notes on the schematics, refer to the caption for Fig 3.

C1-C3—Polypropylene capacitor, 47 nF.Panasonic #ECQ-P1H473GZ, Digi-Key#P3473.

U1—MC3340P. See caption for Fig 4.

voltage gain of the attenuators is four.The circuits are configured for 300-mV(P-P) input, with output varying from1.2 to 1200 mV (P-P). This same setupis used for the MIKE GAIN, SIDETONELEVEL and CW OFFSET SPOT LEVEL

circuits. (One attenuator for each cir-cuit.) Each of these circuits requiresonly a single, unshielded dc lead to thefront panel. Simplicity of wiring andavoidance of hum was well worth theextra trouble of installing the ICs.

I like the normal operating positionsof all controls to be between 10 and 11o’clock, with 40-dB of gain reduction atthe full counterclockwise position and20-dB increase when fully clockwise.Satisfying this particular demand wasmore difficult than expected—thecharacteristics of the devices variedfrom sample to sample. Each circuitrequires two trimpots to set the desiredcurve, and the adjustments interactgreatly. Alignment was a bit tedious,but the attenuator circuits have func-tioned perfectly and have required nofurther attention.

Muting GateAfter the attenuator is the JFET

muting gate, used for TR switching andsquelching. The deceptively innocent-looking capacitor, C2, in the sourcecircuit deserves attention. Since allother circuits in the radio are quite fast,this capacitor determines receiverrecovery time. Some delay is needed, orthe radio will squeal in protest for a fewmilliseconds at the end of each trans-mission, or after each dit or dah whenoperating CW QSK. Too much capacityhere will cause excessive delay anddestroy the radio’s break-in ability. On

the other hand, too little capacity willallow clicks to be heard in the head-phones. The divider for the blockingvoltage also affects the timing. Thecircuit shown here mutes the receiversmoothly and quietly. Stations can beheard breaking in while the radio istransmitting CW at 50 wpm.

60-Hz Notch FilterThe next stage is an adjustable-Q

60-Hz notch filter.4 It eliminates any

hum, whether from internal circuits orincluded on received signals. Thecomponent values specified result ina calculated notch frequency of58.9 Hz. The component manufac-turers’ tolerances result in a worst-case error of 1.7 Hz. This possible erroris one reason for using an adjustable-Q filter: to ensure that the notchincludes 60 Hz. The trimpot isadjusted to obtain a 3-dB band-rejectrange of about 55-65 Hz. Measured

54 Jan/Feb 2000

notch depth is 50 dB.The notch filter circuit is a modern

version of the traditional “twin-T”network. The network alone has a Q ofonly 0.3, but with feedback derived fromthe upper voltage-follower op amp, the Qcan be adjusted from 0.3 to more than 50.The trimpot determines the amount offeedback. The lower voltage follower isneeded to drive the network from a low-impedance source so the circuit is un-affected by the resistance of the trimpot.Notch depth depends on componentmatch (see Note 4).

The last stage is a summing op ampthat combines the signals from thedetectors, sidetone oscillator and CW-offset spot mixer on the RF board. Thegain of the summing-amplifier stage isadjusted to provide the correct audiosignal level at terminal σAF, whichleads to the AF output module on therear panel.

Sidetone OscillatorThe sidetone oscillator is shown in

Fig 7. Adjustments are provided forpitch, feedback and output level. Thefeedback control was originally in-tended to allow adjustment for theminimum amount needed to sustainoscillation, thus to obtain the purest-possible sine wave. However, a puresine wave may produce fatigue overlong operating periods. The feedbackcontrol may be used to introduce some

Fig 8—AF-output-module schematic diagram. For general notes on the schematics, refer to the caption for Fig 3. The 1-Ωresistor in the +18-V supply line to the output stage is a shunt for current metering during tests. A DMM set to the millivoltrange will read out in milliamps directly. These 1-Ω shunts are also included at many other points in the radio where a currentmeasurement is required. The complementary-output transistors are available at Hosfelt.C—Polypropylene capacitor, 1 nF, 2%tolerance. Panasonic #ECQ-P1H102GZ,Digi-Key #P3102. For 2C use two ofthese in parallel; this insures theclosest match.

R—Metal-film resistor, 37.5 kΩ, 1%tolerance. Digi-Key #37.5KXBK.

RFC1—RF choke, 10 µH. The chokemust have a dc resistance of 1 Ω orless.

harmonic distortion, which may resultin a more pleasant sound. Recall thata good violin produces many nice-sounding harmonics. When in eitherTUNE or PULSE TUNE mode, the side-tone oscillator is disabled.

AF Output ModuleThe AF-output module is mounted on

the transceiver rear panel, with theoutput transistors using the panel as aheat sink. The circuit is shown in Fig 8.Two stages of low-pass filtering areused, each with a 3-dB cut-off frequencyof 3000 Hz. The measured compositeresponse is –3 dB at 2500 Hz and –6 dBat 3000 Hz.

Each filter stage is a unity-gain,maximally-flat Butterworth filter with:

Q = 2

2(Eq 1)

in a voltage-follower configuration.5This configuration is the simplest toapply with respect to componentselection. It also involves the simplestformulas. One first chooses the cut-offfrequency, f (in Hz), and the capacity,C (in farads). Then the resistance R (inohms) is simply:

RfC

= 1

2 2 π(Eq 2)

response; it serves merely to protectthe op amp (see Note 5, p 126).

The output stage operates class A.The biasing is chosen for 50 mA idlingcurrent. Total harmonic distortion(including hum and noise) of the AF-output module measures 0.3% atnormal levels and 1.5% at the full 2-Woutput. In a test involving the entirereceiver, an RF signal on the 20-meterband produced total harmonic distor-tion (including hum and noise) of 0.6%at normal levels.

Transmit CircuitsFig 9 is the schematic of the stages

used for transmitting. The speechamplifier has two alternative inputs.The front-panel MIKE jack is rarelyused. I don’t like a clutter of cables onthe operating bench or protrudingplugs that impede control-knob access.I always use the rear panel SPEECHjack.6 This jack provides a high-level,600-Ω line input and accepts speechsignals from the station’s speech-distribution system and the digitalvoice keyer.7 Running this input at ahigh-level minimizes hum pick-up—often a problem when digital voicerecorders are used. The nominal inputlevels are 30 mV (P-P) at the front-panel MIKE jack and 300 mV (P-P) at therear-panel SPEECH jack. The invertingop-amp summing stage combines thetwo inputs with no interaction.

The 10-kΩ resistor at the non-inverting input does not affect the

Jan/Feb 2000 55

The JFET microphone amplifier islocated directly at the front panel on avery small circuit board attached to thejack. This minimizes hum pick-up. Asingle shielded lead runs from thisboard to the AF board. The microphoneamplifier will accept high-impedancemikes; for low-impedance mikes, anappropriate load resistor must beprovided externally. For my Heil HC-5mikes, I install a 2.2-kΩ resistor in themike plug. This arrangement providesthe greatest flexibility (see Note 7). TheRC network at the mike jack filters RF.Silver-mica and polypropylene capa-citors are used for their low-noisecharacteristics.

After the summing amplifier, there isan electronic attenuator for MIKE GAIN

control. It is adjusted as describedabove in the “Receiver Circuits” section.The electronic attenuator avoids theneed to route speech signals to the frontpanel, again minimizing hum pick-up.Capacitor C2 at pin 6 of the attenuatoris for high-frequency roll-off; it may bealtered as desired. The value given forC2 results in a 3-dB roll-off at 5 kHz.After the attenuator is a JFET gate.This, along with switched stages on theIF board, eliminates any possibility ofmodulation during CW or TUNE opera-tion. The next op amp drives the high-pass filter.

High-Pass FilterTwo stages of high-pass filtering are

used, each have a 3-dB cut-off

frequency of 300 Hz. The measuredcomposite response is –3 dB at 380 Hzand –50 dB at 60 Hz. Care was taken toinstall the high-pass filter immediatelyadjacent to the balanced modulator,thus eliminating any possibility of humpick-up on a connecting cable.

Each stage is a unity-gain, maxi-mally-flat Butterworth filter with

Q = 2

2(Eq 3)

Fig 9—Transmitter AF-circuit schematic diagram. For general notes on the schematics, refer to the caption for Fig 3. Thesignal levels after the mike-gain attenuator are given for the normal control-knob position of 11 o’clock (see Part 1, p 21 forinformation about microphone calibration). The DBM used as a balanced modulator may be obtained in small quantitiesdirectly from Mini Circuits.

C, C1—Polypropylene capacitor, 10 nF,2% tolerance. Panasonic #ECQ-P1H103GZ, Digi-Key #P3103.

C2—Polypropylene capacitor, 5.6 nF.Panasonic #ECQ-P1H562GZ, Digi-Key#P3562.

R—Metal-film resistor, 75 kΩ, 1%tolerance. Digi-Key #75.0KXBK. For R/2use two of these in parallel; this insuresthe closest match.U1—MC3340P. See caption for Fig 4.

in a voltage-follower configuration (seeNote 5). This configuration is thesimplest to apply in regard to compo-nent selection. It also involves thesimplest formulas. One first choosesthe cut-off frequency f (in Hz) and thecapacity C (in farads). Then the

56 Jan/Feb 2000

resistance R (in ohms) is simply:

Rf C

= 1

2π(Eq 4)

With a fixed value of C, the cut-offfrequency may be changed (within lim-its) by simply using the inverse relationbetween f and R. For example, with thevalue shown for C, a cut-off frequencyof f = 200 Hz can be obtained withR = 112.5 kΩ. The 10-kΩ resistor at thenon-inverting input does not affect theresponse; it simply protects the op ampas previously described. The 10-kΩfiltering resistor at the +7-V biassupply point also does not affect theresponse; the 47-µF capacitor providesa virtual AF ground for the network.

ConstructionThe AF board is built as shown in

Fig 1. The general method of construc-tion was described in Part 1, where theneed for careful shielding and leadfiltering was discussed. The power andcontrol leads are filtered as describedin Part 2. Most of the purely-AF cir-cuits are hand-wired on perf boards.Plain copper board and truly “ugly”construction is used for all circuitscontaining 9-MHz signals. The board’sunderside is shown in Fig 10.

Test and AlignmentNormal operating levels at various

points of the circuit are indicated on theschematic diagrams. The associatedtrimpots are adjusted to obtain thespecified oscilloscope readings. Thereare quite a few trimpots in the radio.When building a single unit, it is ofteneasier to install a trimpot than toindividually select a component. Thisminor expense is justified by saving ofhours of work in selecting and changingcomponents. This is especially truewhen a radio is designed and builtstage-by-stage. Often, a section is builtwithout full knowledge of what liesahead. Desired levels are apt to changein the light of final testing. None of thetrimpots has required readjustmentsince the radio was built. Exceptionsoccur when modifications areintroduced; then the trimpots becomeespecially useful.

SummaryThis article gives a complete descrip-

tion of the AF board in a high-perfor-mance homebrew transceiver. Double-balanced mixers are used for both theproduct detector and the balancedmodulator. Special filters are includedto avoid hum both in reception and inthe transmitted signal. As in the

remainder of the radio, flexibility andoperating convenience are prime designfactors.

Notes1M. Mandelkern, K5AM, “A High-Perfor-

mance Homebrew Transceiver: Part 1,”QEX, Mar/Apr 1999, pp 16-24.

2M. Mandelkern, K5AM, “A High-Perfor-mance Homebrew Transceiver: Part 2,”QEX, Sep/Oct 1999, pp 3-8.

3M. Mandelkern, K5AM, “A High-Perfor-mance Homebrew Transceiver: Part 3,”QEX, Nov/Dec 1999, pp 41-51.

4R. C. Dobkin, “High Q Notch Filter,” LinearBrief #LB-5, in Linear Applications Hand-book, National Semiconductor Corp,Santa Clara, California, 1980.

5W. Jung, IC Op-Amp Cookbook (Indianapo-lis: Howard W. Sams and Co, 1974),pp 331-333.

6Similarly, I don’t use the front-panel HEAD-PHONE jack, but only the rear-panelSPEAKER jack. On the operating bench isa 40-year-old speaker/headphone switchbox with a headphone-level control forequalization. This provides instant switch-ing without fussing with the headphone

Fig 10—Bottom view of the AF board. Effective filters are installed at each terminal andcoax cables are soldered directly to the double-sided circuit board (see the discussion inPart 2). To minimize connector troubles, the board is hard-wired to the radio; a 12-inch-long bundle of wires and cables allows the board to be easily lifted and serviced.

plug or readjusting the AFG.7M. Mandelkern, K5AM, “The AMSAFID: An

Automatic Microphone Switcher AmplifierFilter Integrator Distributor,” QST, Nov1995, pp 47-49.

8Sharp-eyed readers may notice that the AM/FM compartment is empty. This only provesthat this shack follows true ham-radio tradi-tion: Nothing is ever really finished.

9Digi-Key Corporation, 701 Brooks Ave S,PO Box 677, Thief River Falls, MN 56701-0677; tel 800-344-4539 (800-DIGI-KEY),fax 218-681-3380; http://www.digikey.com/.

10Hosfelt Electronics, 2700 Sunset Blvd,Steubenville, OH 43952; tel 800-524-6464, fax 800-524-5414; E-mail hosfelt@clover.net; http://www.hosfelt.com/.

11Mouser Electronics, 2401 Hwy 287 N,Mansfield, TX 76063, tel 800-346-6873,fax 817-483-0931; E-mail sales@mouser.com; http://www.mouser.com/.

12Mini Circuits Labs, PO Box 350166, Brook-lyn, NY 11235-0003; tel 800-654-7949,718-934-4500, fax 718-332-4661; http://www.minicircuits.com/.

See Part 5 in March/April 2000 QEX

Jan/Feb 2000 57

By Zack Lau, W1VT

225 Main StNewington, CT 06111-1494zlau@arrl.org

RF

1Notes appear on page 61.

Feeding Open-WireLine at VHF and UHF

It can be quite challenging to make ahigh-power balun that works well atVHF and UHF by scaling ferrite-coredesigns used at HF. The upper frequency response limit can be increasedby making the wires shorter, butsmaller cores must be used to get asufficient number of turns for effectivecoupling between the turns. Obvi-ously, smaller cores won’t handle muchpower. Thus, while ferrite cores maywork for lossy TV baluns, I wouldn’trecommend them for transmit appli-

cations at VHF and UHF. A loss of0.5 dB may not be significant in casualreceiving applications like cable TV,but it is quite substantial at the1000-W level—it is a loss of 10.9% or109 W!1

Fortunately, it is seldom necessaryto cover a wide frequency range—mostVHF antennas cover only one amateurband. Thus, one can use a narrow-bandwidth design, and avoid lossyferrite materials.

A balun that works well at VHF isshown in Fig 1, using a λ/2 of coax. Itis often used to feed Yagi antennas. Itis important to use an electrical λ/2and shorten the physical coax lengthby its velocity factor. (Sometimes,

people use a section of copper-jacketedsemi-rigid coax such as UT-141A toreduce the difficulty of weather-proofing.) This has a 4:1 ratio, typi-cally transforming a balanced 200-Ωload to an unbalanced 50-Ω load.However, it could also be used totransform a 300-Ω twin lead to 75-Ωcoax. The velocity factor is 0.70, soa 71.4-cm length has an electricallength of 102 cm. Solid-dielectriccables are often preferred; as foamor partial air dielectrics may showmore variation in measured velocityfactors.

A λ/2 of coax does two things—itreflects the input impedance to theoutput and introduces a 180° phaseshift. Thus, if the coax is terminatedin 100 Ω, it reflects the 100-Ω load as

58 Jan/Feb 2000

100 Ω, no matter what the impedanceof the coax. The two 100 Ω loads inparallel provide the desired 50-Ω inputimpedance. For a single frequency, theimpedance of the coax doesn’t mattervery much. However, the impedancedoes affect the bandwidth of the balun.For resistive terminations, raising thecoax impedance improved the band-width of the input match, as shown inFig 2. Thus, 75- or 100-Ω coax might beused to improve balun bandwidth.RG-62 and RG-133A are two types of93-Ω coax that are commerciallyavailable, the latter being moresuitable for high-power work.

Since RG-133A may be difficult tofind, I’ve made 100-Ω coax in the pastout of 0.5-inch brass tubing and #12copper wire.

The equation for the characteristicimpedance (Z0) of air-dielectric coax is

ZD

d0 138=

log (Eq 1)

Where D is the inside diameter ofthe outer conductor and d is the out-side diameter of the center conductor.Thus, with a wall thickness of 14 mils,

Z0 1380 5 2 0 014

0 0808106=

− ×( )

=log. .

.Ω

(Eq 2)I used thin Teflon spacers to hold the

center conductor in place. Thus, in-stead of searching for rarely manufac-tured RG-125/U or 150-Ω coax, youmight make your own from tubingand wire.

Typical line impedances for com-mercial twin-lead or open wire are 300and 450 Ω, not 200 Ω. Thus, a match-ing transformer is needed to go from200 Ω to the desired line impedance.Charles Emil Ruckstuhl, W1ZJD,showed that a λ/4 of 300-Ω twin-leadcould be used to match a 200-Ω balunto 450-Ω homebrew ladder-line.2 Heused a λ/2 balun of RG-213 to feed a500-foot run of ladder line on 2 meters.This is shown schematically in Fig 3.

Transmission-line impedances otherthan 450 Ω can also be matched with aλ/4 line section. The impedance of thesection is:

Z Z Z0 = ×out in(Eq 3)

For 300 Ω, the impedance is:

300 200 245× = Ω (Eq 4)This impedance is non-standard,

but this can be generated at home ormodified from commercially availabletransmission lines. Andrew Griffith,W4ULD, describes a technique formodifying ladder line with a propanetorch and a wooden jig.3 Andrew low-

Fig 1—A 4:1 balun using λ/2 of coax.

Figure 2—Raising the coax impedance to from 50 to 150 Ω improves the bandwidthof a 50:200 Ω λ/2 coaxial balun.

Fig 3—A 9:1 balun using λ/2 of coax and a 300-Ω λ/4 matching section.

ers the impedance by reducing thespacing and reforming the polyethyl-ene insulation.

Alternatively, the equation for openwire line is:

Zs

d0 2762=

log (Eq 5)

Where Z0 is the line impedances is the spacing between the centers ofthe lines d is the diameter of the wires

Or,

sd

Z

=

210

0276 (Eq 6)

Thus, if the impedance is 245 Ω and#12 wire (with a diameter of 80.8 mils)

Jan/Feb 2000 59

Table 1—Improved performance of a 9:1 balun using 75-Ω coaxial λ/2 balun (50TO450B) versus one using 50-Ωcoax (50TO450L). Both use a 300-Ω λ/4 matching stub to go from 200 to 450 Ω.

Compact Software - ARRL Radio Designer 1.5 14-JUL-99 11:54:58File: C:\ARD\BALUNQ5.CKTFreq MS11 MS21 MS31 PS21 PS31 MS11 MS21 MS31 PS21 PS31(MHz) (dB) (dB) (dB) (deg) (deg) (dB) (dB) (dB) (deg) (deg)

50TO 50TO 50TO 50TO 50TO 50TO 50TO 50TO 50TO 50TO450B 450B 450B 450B 450B 450L 450L 450L 450L 450L

120.000 –27.60 –3.08 –3.10 –66.6 134.9 –18.19 –3.45 –2.89 –59.7 135.6121.000 –28.06 –3.06 –3.11 –67.6 133.2 –18.68 –3.42 –2.91 –60.9 133.8122.000 –28.54 –3.06 –3.11 –68.5 131.5 –19.19 –3.38 –2.93 –62.1 132.0123.000 –29.03 –3.05 –3.12 –69.5 129.8 –19.72 –3.35 –2.94 –63.4 130.2124.000 –29.52 –3.04 –3.13 –70.5 128.1 –20.26 –3.32 –2.96 –64.6 128.5125.000 –30.03 –3.04 –3.13 –71.4 126.5 –20.81 –3.29 –2.97 –65.8 126.7126.000 –30.56 –3.03 –3.13 –72.4 124.8 –21.39 –3.27 –2.99 –67.0 124.9127.000 –31.10 –3.03 –3.13 –73.3 123.1 –21.99 –3.24 –3.00 –68.2 123.2128.000 –31.65 –3.03 –3.14 –74.3 121.3 –22.61 –3.22 –3.01 –69.4 121.4129.000 –32.23 –3.03 –3.14 –75.2 119.6 –23.25 –3.20 –3.03 –70.6 119.6130.000 –32.83 –3.03 –3.14 –76.1 117.9 –23.92 –3.19 –3.04 –71.8 117.9131.000 –33.46 –3.03 –3.14 –77.0 116.2 –24.63 –3.17 –3.05 –73.0 116.1132.000 –34.11 –3.03 –3.13 –77.9 114.5 –25.37 –3.15 –3.06 –74.1 114.4133.000 –34.80 –3.03 –3.13 –78.8 112.7 –26.14 –3.14 –3.06 –75.3 112.6134.000 –35.53 –3.03 –3.13 –79.7 111.0 –26.97 –3.13 –3.07 –76.5 110.9135.000 –36.30 –3.03 –3.13 –80.6 109.3 –27.86 –3.12 –3.08 –77.6 109.1136.000 –37.13 –3.03 –3.13 –81.5 107.5 –28.81 –3.11 –3.09 –78.8 107.4137.000 –38.03 –3.03 –3.13 –82.4 105.8 –29.84 –3.10 –3.09 –79.9 105.6138.000 –39.02 –3.04 –3.12 –83.2 104.0 –30.98 –3.09 –3.10 –81.0 103.9139.000 –40.10 –3.04 –3.12 –84.1 102.2 –32.26 –3.09 –3.10 –82.2 102.1140.000 –41.33 –3.04 –3.12 –85.0 100.5 –33.70 –3.08 –3.10 –83.3 100.4141.000 –42.74 –3.04 –3.12 –85.8 98.7 –35.40 –3.08 –3.11 –84.5 98.6142.000 –44.41 –3.04 –3.12 –86.7 96.9 –37.44 –3.07 –3.11 –85.6 96.9143.000 –46.45 –3.04 –3.12 –87.5 95.2 –40.00 –3.07 –3.11 –86.7 95.1144.000 –49.08 –3.04 –3.12 –88.4 93.4 –43.40 –3.07 –3.11 –87.8 93.4145.000 –52.79 –3.04 –3.12 –89.2 91.6 –47.76 –3.07 –3.12 –89.0 91.6146.000 –58.68 –3.04 –3.12 –90.1 89.9 –48.96 –3.07 –3.12 –90.1 89.9147.000 –61.69 –3.04 –3.12 –90.9 88.1 –44.78 –3.07 –3.12 –91.2 88.1148.000 –54.84 –3.04 –3.12 –91.8 86.3 –41.03 –3.07 –3.12 –92.4 86.4149.000 –50.34 –3.04 –3.12 –92.6 84.5 –38.24 –3.07 –3.11 –93.5 84.6150.000 –47.29 –3.04 –3.12 –93.5 82.8 –36.07 –3.07 –3.11 –94.6 82.9151.000 –44.98 –3.04 –3.12 –94.3 81.0 –34.28 –3.08 –3.11 –95.7 81.1152.000 –43.13 –3.04 –3.13 –95.2 79.2 –32.77 –3.08 –3.11 –96.9 79.4153.000 –41.57 –3.04 –3.13 –96.1 77.5 –31.46 –3.09 –3.10 –98.0 77.6154.000 –40.23 –3.03 –3.13 –96.9 75.7 –30.28 –3.09 –3.10 –99.2 75.9155.000 –39.04 –3.03 –3.13 –97.8 74.0 –29.23 –3.10 –3.10 –100.3 74.1156.000 –37.97 –3.03 –3.14 –98.7 72.2 –28.26 –3.11 –3.09 –101.4 72.4157.000 –37.00 –3.03 –3.14 –99.6 70.5 –27.36 –3.12 –3.08 –102.6 70.6158.000 –36.11 –3.03 –3.14 –100.4 68.7 –26.52 –3.13 –3.08 –103.8 68.8159.000 –35.27 –3.03 –3.14 –101.3 67.0 –25.73 –3.14 –3.07 –104.9 67.1160.000 –34.49 –3.03 –3.15 –102.2 65.3 –24.98 –3.16 –3.06 –106.1 65.3161.000 –33.76 –3.02 –3.15 –103.1 63.5 –24.27 –3.17 –3.05 –107.2 63.6162.000 –33.06 –3.02 –3.15 –104.1 61.8 –23.60 –3.19 –3.04 –108.4 61.8163.000 –32.40 –3.03 –3.15 –105.0 60.1 –22.95 –3.20 –3.03 –109.6 60.1164.000 –31.77 –3.03 –3.15 –105.9 58.4 –22.32 –3.22 –3.02 –110.8 58.3

is available; the required spacing is312 mils.

Baluns can also be modeled on acomputer—I’ve modeled the both the50 to 200 and 50 to 450 Ω baluns usingAmateur Radio Designer (ARD).4 Thissoftware has a three-port model. Theopen-wire line is modeled as two sepa-rate ports each having one half of theline impedance. Thus, a 450-Ω open-wire is modeled as two 225-Ω coaxiallines (see Fig 4). When creating theARD report form, don’t forget to clickon Terms... to set the proper termina-tions of 50, 225 and 225 Ω.

The performance of the balun can bedetermined by looking at the phase andpower relationships between the ports.Ideally, there would be exactly 3 dB ofloss from the input of the balun to each

Fig 4—Modeling a 450-Ω balanced line as two 225-Ω coaxial lines with ARD’s three-port model.

of the balanced outputs and a 180°phase difference between the trans-mission coefficients. Thus, if port 1 isthe input and ports 2 and 3 are the

outputs, you want to see 180° phasedifferences between PS21 and PS31. Itis erroneous to expect a 180° phase shiftfor PS32, the phase of the transmission

60 Jan/Feb 2000

*Model 2-meter 50-ohm to 450-ohm balun*Half wave balun and quarter wave matching section*Model of 100 ft of UT-141A semi-rigid coaxBLKCAB 1 2 Z=50 P=1200000MIL V=0.70 C1=.348913 C2=.024644141LINE:2POR 1 2END

*Model of 100 ft of 300 ohm transmitting*tubular parallel line.*Data from page 19.2 1999 ARRL HandbookBLKCAB 1 2 Z=300 P=1200IN V=0.80 C1=.105641 C2=.022346LINE:2POR 1 2END

*Verify that the attenuation constants*are the same when modeled as two 150 ohm*transmission linesblkcab 1 3 z=150 p=1200in v=0.80 c1=0.105641 c2=0.022346cab 2 4 z=150 p=1200in v=0.80 c1=0.105641 c2=0.022346trf 100 1 0 2 r1=50 r2=300trf 200 3 0 4 r1=50 r2=300300line:2por 100 200end

*Model 2-meter balun—50 ohms unbalanced to 200 ohms balanced*Note: ports 2 and 3 are 100 ohms with respect to groundBLKCAB 1 2 Z=50 P=.719M V=0.7 C1=.348913 C2=0.024644BALUN:3POR 1 1 2END

*****************************************************Look at the effect of coax impedance on a 4:1 balun

*Balun with 50 ohm coaxBLKCAB 1 2 Z=50 P=0.719M V=0.7 C1=.348913 C2=0.024644TRF 1 1 2 R1=200 R2=50BAL50:2POR 1 2END

*Balun with 75 ohm coaxBLKCAB 1 2 Z=75 P=0.719M V=0.7 C1=.348913 C2=0.024644TRF 1 1 2 R1=200 R2=50BAL75:2POR 1 2END

*Balun with 100 ohm coaxBLKCAB 1 2 Z=100 P=0.719M V=0.7 C1=.348913 C2=0.024644TRF 1 1 2 R1=200 R2=50BAL100:2POR 1 2END

*Balun with 150 ohm coaxBLKCAB 1 2 Z=150 P=0.719M V=0.7 C1=.348913 C2=0.024644TRF 1 1 2 R1=200 R2=50BAL150:2POR 1 2END*****************************************************

*Balun with 200 to 450 ohm transmission line transformer*Quarter wave of 300 ohm line as matching section*Zero loss caseblkcab 1 2 z=50 p=0.719m v=0.7 c1=0 c2=0trl 1 3 z=150 p=0.488m v=0.95trl 2 4 z=150 p=0.488m v=0.9550to450:3por 1 3 4end*

*Balun with 200 to 450 ohm transmission line transformer*Quarter wave of 300 ohm line as matching section*lossy line case with UT-141A 50 ohm semi-rigid balunblkcab 1 2 z=50 p=0.719m v=0.7 c1=0.34913 c2=0.02464CAB 1 3 Z=150 P=16.18IN V=0.80 C1=.105641 C2=.022346CAB 2 4 Z=150 P=16.18IN V=0.80 C1=.105641 C2=.02234650to450l:3por 1 3 4end

*Balun with 200 to 450 ohm transmission line transformer*Quarter wave of 300 ohm line as matching section*lossy line case with 75 ohm coax for the balun*loss set equal to that for UT-141Ablkcab 1 2 z=75 p=0.719m v=0.7 c1=0.34913 c2=0.02464CAB 1 3 Z=150 P=16.18IN V=0.80 C1=.105641 C2=.022346CAB 2 4 Z=150 P=16.18IN V=0.80 C1=.105641 C2=.02234650to450b:3por 1 3 4end

*Balun with 200 to 450 ohm transmission line transformer*Quarter wave of 300 ohm line as matching section*lossy line case with 100 ohm coax for the balun*loss set equal to that for UT-141Ablkcab 1 2 z=100 p=0.719m v=0.7 c1=0.34913 c2=0.02464CAB 1 3 Z=150 P=16.18IN V=0.80 C1=.105641 C2=.022346CAB 2 4 Z=150 P=16.18IN V=0.80 C1=.105641 C2=.02234650to450c:3por 1 3 4end

*Balun with 200 to 450 ohm transmission line transformer*Quarter wave of 300 ohm line as matching section*lossy line case with 150 ohm coax for the balun*loss set equal to that for UT-141Ablkcab 1 2 z=150 p=0.719m v=0.7 c1=0.34913 c2=0.02464CAB 1 3 Z=150 P=16.18IN V=0.80 C1=.105641 C2=.022346CAB 2 4 Z=150 P=16.18IN V=0.80 C1=.105641 C2=.02234650to450d:3por 1 3 4end

*Lengthen line lengths to optimize for*both 144 and 432 MHz.*Balun with 200 to 450 ohm transmission line transformer*Quarter wave of 300 ohm line as matching section*lossy line case with 75 ohm coax for the balun*loss set equal to that for UT-141Ablkcab 1 2 z=75 p=0.729m v=0.7 c1=0.34913 c2=0.02464CAB 1 3 Z=150 P=0.416m V=0.80 C1=.105641 C2=.022346CAB 2 4 Z=150 P=0.416m V=0.80 C1=.105641 C2=.02234650to450x:3por 1 3 4endfreq*step 120mhz 164mhz 1mhz*1mhz 10mhz 100mhz 1000mhzstep 140mhz 150mhz 1mhzstep 420mhz 450mhz 1mhzend

opt*Optimizer used to determine attenuation constants for 300 ohm twin

leadline f=1mhz ms21=-0.09dBline f=10mhz ms21=-0.3dBline f=100mhz ms21=-1.1dBline f=1000mhz ms21=-3.9dBend

Figure 5—The Amateur Radio Designer file used tomodel baluns.

Jan/Feb 2000 61

Table 2—Performance of a Dual-Band 432/144 MHz Balun

Compact Software—ARRL Radio Designer 1.5 15-JUL-99 09:37:30File: balunq5.ckt

Freq MS11 MS21 MS31 PS21 PS31 MS11 MS21(MHz) (dB) (dB) (dB) (deg) (deg) (dB) (dB)

50TO 50TO 50TO 50TO 50TO 50TO 50TO450B 450B 450B 450B 450B 450L 450L

140.000 –41.33 –3.04 –3.12 –85.0 100.5 –33.70 –3.08141.000 –42.74 –3.04 –3.12 –85.8 98.7 –35.40 –3.08142.000 –44.41 –3.04 –3.12 –86.7 96.9 –37.44 –3.07143.000 –46.45 –3.04 –3.12 –87.5 95.2 –40.00 –3.07144.000 –49.08 –3.04 –3.12 –88.4 93.4 –43.40 –3.07145.000 –52.79 –3.04 –3.12 –89.2 91.6 –47.76 –3.07146.000 –58.68 –3.04 –3.12 –90.1 89.9 –48.96 –3.07147.000 –61.69 –3.04 –3.12 –90.9 88.1 –44.78 –3.07148.000 –54.84 –3.04 –3.12 –91.8 86.3 –41.03 –3.07149.000 –50.34 –3.04 –3.12 –92.6 84.5 –38.24 –3.07150.000 –47.29 –3.04 –3.12 –93.5 82.8 –36.07 –3.07420.000 –32.28 –3.06 –3.22 105.5 –58.9 –22.70 –3.27421.000 –32.89 –3.06 –3.22 104.6 –60.6 –23.34 –3.25422.000 –33.53 –3.06 –3.22 103.7 –62.3 –24.00 –3.23423.000 –34.19 –3.06 –3.22 102.8 –64.1 –24.69 –3.21424.000 –34.89 –3.06 –3.22 101.9 –65.8 –25.42 –3.20425.000 –35.63 –3.06 –3.22 101.0 –67.5 –26.18 –3.19426.000 –36.41 –3.06 –3.22 100.1 –69.3 –26.99 –3.17427.000 –37.24 –3.06 –3.21 99.2 –71.0 –27.85 –3.16428.000 –38.14 –3.06 –3.21 98.3 –72.8 –28.78 –3.15429.000 –39.12 –3.06 –3.21 97.5 –74.5 –29.78 –3.14430.000 –40.19 –3.06 –3.21 96.6 –76.3 –30.87 –3.14431.000 –41.39 –3.07 –3.21 95.7 –78.0 –32.08 –3.13432.000 –42.76 –3.07 –3.21 94.9 –79.8 –33.43 –3.12433.000 –44.34 –3.07 –3.20 94.0 –81.6 –34.98 –3.12434.000 –46.23 –3.07 –3.20 93.2 –83.3 –36.78 –3.12435.000 –48.57 –3.07 –3.20 92.3 –85.1 –38.90 –3.11436.000 –51.57 –3.07 –3.20 91.4 –86.9 –41.37 –3.11437.000 –55.33 –3.07 –3.20 90.6 –88.7 –43.79 –3.11438.000 –57.45 –3.07 –3.20 89.8 –90.4 –44.59 –3.11439.000 –54.50 –3.07 –3.20 88.9 –92.2 –42.86 –3.11440.000 –50.84 –3.07 –3.20 88.1 –94.0 –40.31 –3.11441.000 –47.97 –3.07 –3.20 87.2 –95.7 –37.98 –3.11442.000 –45.72 –3.07 –3.21 86.4 –97.5 –35.99 –3.12443.000 –43.88 –3.07 –3.21 85.5 –99.3 –34.30 –3.12444.000 –42.33 –3.07 –3.21 84.6 –101.0 –32.84 –3.13445.000 –40.99 –3.06 –3.21 83.8 –102.8 –31.54 –3.13446.000 –39.80 –3.06 –3.21 82.9 –104.6 –30.38 –3.14447.000 –38.73 –3.06 –3.22 82.0 –106.3 –29.33 –3.15448.000 –37.75 –3.06 –3.22 81.2 –108.1 –28.36 –3.15449.000 –36.85 –3.06 –3.22 80.3 –109.8 –27.46 –3.16450.000 –36.01 –3.06 –3.22 79.4 –111.6 –26.62 –3.18

coefficient going from port 2 to port 3.Table 1 shows the results of modeling

the balun with 50- and 75-Ω coax. Sur-prisingly, simulations showed that the75-Ω coax worked better than the 50,100 and 150-Ω varieties, assuming thecoax loss are constant.

The balun will also operate on oddharmonics. Remember that the λ/2 linefor the balun must supply a 180° phaseshift. At odd harmonics, the coax adds(2×w) + 1 180° shifts, where w is a wholenumber. Alternatively, the coax addsw×360° plus an additional 180° phaseshift. Of course, the w×360° phaseshifts equal a 0° phase shift, so the neteffect of the longer line is still a 180°phase shift. In practice, the longer line

will narrow the bandwidth of the balun,but the balun may still be useful fornarrow-band work. Similarly, the coaxis still a multiple of λ/2, which isrequired to reflect the proper impe-dance between the ends of the coax.Table 2 shows the 144- and 432-MHzperformance of a 450:50 Ω balun usingλ/2 and λ/4 transmission lines cut for144 MHz. I used 75-Ω coax, as it offersa better bandwidth than 50-Ω coax.

No-Loss Transmission Line, Almost,” Pro-ceedings of the 19th Eastern VHF/UHFConference of the Eastern VHF/UHF So-ciety (Newington: ARRL, 1993), pp 65-75.

3A. Griffith, W4ULD, “The 1/3-WavelengthMultiband Dipole,” QST, September 1993,pp 33-35.

4ARRL products are available from your lo-cal ARRL dealer or directly from theARRL. Mail orders to Pub Sales Dept,ARRL, 225 Main St, Newington, CT06111-1494. You can call us toll-free at tel888-277-5289; fax your order to 860-594-0303; or send e-mail to pubsales@arrl.org. Check out the full ARRL publicationsline on the World Wide Web at http://www.arrl.org/catalog.

Mark Mandelkern, K5AM, brings usthe fifth and final segment on hishomebrew transceiver. As we wrap upthe series, the focus is on the controlsystem, PTOs, frequency counter andconstruction details. Mark is alsoworking on a piece for us about the“front-end” designs—the parts thatchange from band to band.

We begin a look at another high-per-formance homebrew rig with the firstpart of John Stephensen, KD6OZH’sseries on his ATR-2000. It’s a synthe-sized, 20-meter mono-bander designedwith due regard for actual conditionsfound on 14 MHz at John’s QTH. Hesets forth his goals and neatly explainsthe rationale behind his architecturalchoices. He gives circuit examples thataddress the needs for low phase noiseand high receiver dynamic range.Check it out.

Jim Kocsis, WA9PYH, has been ex-perimenting with 1691-MHz weather-satellite imaging and contributes hisdown-converter design. This synthe-sized unit may also cover severalAmateur Radio bands with slight modi-fications. Jim uses convenient50-Ω blocks, where possible. The resultis an efficient and compact design:There is only one alignment point!

Johan Van de Velde, ON4ANT, pre-sents his homebrew, multi-band Yagidesign. He examines tradeoffs in ele-ment spacing and boom length andweighs his computer-modeling resultsbefore beginning construction. If youwant to cover several HF bands withvery good performance, look at his in-terlaced designs.

Next Issue in QEX

Notes

1Percentage loss= 100 1 10 10−

−

loss

2C. Ruckstuhl, W1JZD,“A Simple Dirt-Cheap

62 Jan/Feb 2000

Letters to the EditorCenter-Loaded Whip AntennaLoss (Jul/Aug ’99)

I have studied the article “CenterLoaded Whip Antenna Loss” in theJuly/August ’99 QEX and question afew of the statements. In the para-graph below Table 1, you state “The‘lossy whip’ is just one of many typesof compact, but poor, antennas avail-able to the unsuspecting ham.” Ipresume you meant the combinationof low-Q loading coil and whip.

In several places, you refer to pianowire. I have a number of differentcommercial mobile antennas and theyall use stainless-steel whips, some ofwhich are machined to provide ataper. Piano wire has a completelydifferent material composition and aconstant diameter.

I do not understand the premise ofTable 1. I gather the idea is to showthat a coil with a Q of 33 is some 3 dBworse than some others, but we don’tknow the configuration of the othercoils, the length of the antenna or thelocation of the coil in the antenna.The last listing is apparently for awhip without a coil, but of whatdimension and for what frequency?

You have apparently ignored thelosses contributed by the vehicleground. Some references were made tolosses of 3 and 6 dB in the coil/reso-nator assemblies. I cannot believe thisis entirely true because 45 W (in mycase) of dissipation in the low-QHustler inductor would probably resultin degradation of insulation and plasticcovering. This has not been the case.The temperature rise is discernable,but not excessive by any means. I donot know of anyone who has destroyeda coil unless running high power on 160meters to a coil wound on PVC.

I do not dispute the possible reduc-tion in field strength with these low-Qconfigurations, but I don’t think all ofthe loss is in the coil. Some of the lossescan apparently be offset by increasingthe length of the base section. Raising orlowering the input resistance bychanging coil parameters does notnecessarily result in field strengthchanges. Is there such a thing as aninfinite-Q and a loss-less coil?

The article by Bruce Brown,W6TWW, (ARRL Antenna Compen-dium, Vol. 1) has been widely quotedin ARRL publications. Most of thesubsequent analyses have been based,I believe, on his conclusions. I am notknowledgeable enough to disagreewith Brown’s conclusions, but I have

some trouble with a couple of thestatements that seem to be taken asfact. On p. 109, for example: “Thus thecoil forces a much higher current intothe top section than would flow in theequivalent part of a full 90°-highantenna.” If this is so, then wouldn’t abase-loaded antenna coil “force” morecurrent in the top section?

From another view, Jerry Sevick,W2FMI, in a series of articles someyears ago concluded that 75% loadingwas somewhat better than mid-pointloading. His conclusions however, werebased on measurement of input resis-tance over a known ground system, notfield-strength measurements. It may bethat W2FMI’s conclusions are correct forantennas of 45° and larger, yet may nothold for antennas of 30° and less, whichBrown addressed.

Incidentally, Sevick appended some“Loading Coil Investigations” to hisBuilding and Using Baluns and Ununsbook. Among other things, hecompares Hustler coils for 80/40/20meters to homebrew coils and coilstock. While the Hustler coils, as weknow, suffer from pretty low Q, hisconclusions are curious: “On 20meters, even the Hustler whips areonly poorer by about 5 dB from theideal quarter-wave condition. Whenlooking at the other numbers, thequestion comes up as to whether it’sworth replacing the Hustler resonatorswith high-Q coils.” For 40 meters, histable shows an efficiency of 11.3% forthe Hustler and 21.5% for a larger-diameter B&W coil. On 20 meters, theHustler yields 35.5% and the B&W coil41%. That equates to near 3 dB on 40and almost nil on 20. His efficienciesare based on antenna input resistancechanges and not field strength, so theymay not be completely valid.

The AA6GL MOBILE programincluded with Compendium 4 is aninformative tool, but it is also basedlargely on the conclusions drawn byBrown. My own experience withHustler coils shows that the effi-ciency can apparently be improved byusing a longer base section. In arecent Florida QSO Party, I operatedwith Hustler coils on three bands (40/20/15) using the multi-coil adapterand an 8-ft base section. This puts thecoil at about 75% of the antennaheight. My contact total was about25% better this year than last, when Iused the small bug-catcher, a 6-ft baseand a 5-ft top. Band conditions, by theway, were much poorer this year.

Without confirming A/B field-strength tests, that proves nothingexcept that the low-Q Hustler config-urations are certainly competitive.Having three coils on one base is also avery nice feature. The arrangement isnot good for highway speeds, however.I do not favor operating while inmotion, from a safety standpoint.

Perhaps this fall I will set up a smalltest range and check out some of thesetheories, such as center versus 75%loading, high-Q versus medium- andlow-Q coils and any other possibilitiesthat occur to me.—Arlan Bowen,N4OO, 349 Persimmon Rd, Sopchoppy,FL 32358; abowen@nettally.com

Hi Arlan,

There is a simple technique todetermine the relative field from anywire antenna; it will show how toploading, center-coil loading and othermethods for changing current distri-bution affect antenna performance.The field from an antenna is directlyproportional to the area under itscurrent-distribution curve. You canplot the current as a function of posi-tion on the antenna. The area underthe curve is then described in ampere-feet or ampere-meters. You will findthat the area often remains nearlyconstant for a fixed-length antenna,regardless of the method used tochange the current distribution. Thismeans that top loading, for example,may not increase gain per se, althoughit will change the input impedance. Toploading with horizontal elements atopa very electrically short antenna doeshowever significantly increase the areaunder the curve.

Keep in mind that the RF currentreverses phase every half wave alonga wire, and this will create a negative(opposite phase) ampere-feet areaunder part of the curve that mustsubtracted from the positive ampere-feet area. In addition, if some of theantenna wires are oriented such thattheir current direction opposes that ofthe main radiator, the ampere-feetarea associated with these wiresmust also be subtracted.

My recent article on mobile-whipantennas assumed a good conductorbeneath the whip, because I have analuminum plate on the top of my van.Whip performance from installationsusing steel as the ground plane maybe different, in which case roof lossesmay be more significant than coillosses. Some vehicles have a plasticroof, in which case I highly reco-mmend that a large aluminumground plane be installed and care-fully bonded to the vehicle ground.This may also help to reduce EM field

Jan/Feb 2000 63

intensity within the vehicle, but thatwill vary from one instal-lation to thenext.—Grant Bingeman, KM5KG,1908 Paris Ave, Plano, TX, 75025;DrBingo@compuserve.com

On Antenna-Mounting Effects

Dear Doug,

Since about the middle 1970s, I havemaintained a particular interest in theperformance of antennas in their oper-ational environments.1 My studies haveincluded effects of the platform struc-ture that supports the antenna, such asautomobiles, ships and sailboats withaluminum masts and conductive rig-ging; the effects of the environment,such as high-rise buildings and high-voltage power lines; the effect of Yagiantennas on masts used to support half-slopers, or the mast itself as a shunt-fed,grounded radiator on the lower bands;and the effect of the support structureand towers on the performance of anten-nas they support, such as dipoles andloops.2, 3 I tell you this because I wish tocomment on some aspects of articles andremarks published recently in QEX.

Peter Madle, KE6RBV, (QEX, May/June 1999, pp 32-42) numerically mo-deled HF and VHF antennas on cars,campers and RVs. I have carried outsimilar studies, comparing experimentswith modeling.4 I have a particularinterest in HF coil-loaded antennas,previously on 4×4 SUVs (GMC Jimmyand Yukon trucks) and currently on aFord F-250 pick-up truck. I use a versionof D. K. Johnson, W6AAQ’s screwdriveran-tenna, as fabricated and modified byLarry Parker, VE3EDY. I have hisshort-coil version for 80 meters andshorter wavelengths and a long-coilversion for 160 meters. This is anexcellent, versatile antenna. For campapplication, I use a segmented 4.9-meter(16-ft) military-style whip (stainless-steel copper-coated) over the basesection with a tuning inductor, insteadof the 1.8-meter (6-ft) whip for mobileuse. I prefer to mount the antenna onthe left-rear bumper of the truck. Thislocation provides gain in the directionaway from the antenna toward theright-front corner of the vehicle. In acampground, one can always face thevehicle in the right direction. Partic-ularly for the lower bands, it is best tohave the full length of the truck workingfor you, providing enhance-ment in thedesired direction.

My main purpose here is to commenton modeling VHF antennas. In this case,we are not interested in the space-wavepatterns, which Madle has calculated forperfectly conducting and finite ground,except to ensure that we have an at-the-horizon lobe. For VHF, we are concerned

with the ground-wave pattern, whichgenerally is similar but can be different.In addition, to compare modeled perfor-mance with measured field strengths,we are definitely concerned withground-wave field strengths (at HF aswell as at VHF).

A final comment on VHF whipsmounted in the center of the roof: Onemight expect that this would be thedesired location, but it is notnecessarily. Currents induced on thebody of the vehicle can significantlycontrol the radiation pattern, partic-ularly if the vehicular structurehappens to exhibit a resonance for thefrequency in use. I have numericallymodeled (using NEC-4D and a de-tailed wire-grid model) short whipsmounted in the center of the vehicleroof over a wide range of frequencies(30-100 MHz, unpublished). I foundthat there are clear resonant effects,which can result in deep azimuthalnulls over narrow bands of frequen-cies. We much earlier found similareffects for HF marine whips onmetallic ships (see Reference 1).

To minimize the effect of the vehicleon the pattern at VHF, it might bebetter to minimize the currentsinduced on the vehicle as a result ofdirectly feeding the antenna againstthe vehicle. We can do this by using avertical-J or Slim-Jim antenna. In thiscase, from a structural point of view,the best place to mount the antenna ison a short stub mast attached to theleft (or right) rear bumper. Therefore, arear-bumper-mount 2-meter antennamight not be a waste of time.

On a different topic, Rudy Severns,N6LF (QEX, September/October ’99,p 59) has commented briefly on shunt-fed grounded verticals. He referenced a1937 classic paper by Morrison andSmith. Personally, I like Laport,5 anantenna book that overviews manydifferent types of antennas as well asvarious versions of shunt-fed groundedtowers. Broadcasters employ isolatedtowers, the impedance characteristics ofwhich are well known. In general, ama-teurs use a mast as a vertical radiatorfor the 160-, 80- or even the 40-meterband that also supports one or moreYagi antennas. John True, W4OQ’sarticle entitled “How to Design Shunt-Feed Systems for Grounded Radiators,”published in a 1975 ham radio articleand reprinted in Communications Quar-terly,6 has discussed the effect of this“top loading” on the resonant frequencyof the mast and hence its impedance,which is important knowledge if we aregoing to shunt feed the tower. I wellremember John, having corres-pondedwith him in past years. He is now asilent key. He did not have the tools to

numerically model the tower and Yagiantenna that we have now to determinethe effect that the Yagi might have onthe directional pattern of the shunt-fedmast. The effect can be significant.7

The Yagi’s elements can carryappreciable current and so are clearlyparts of the radiating system. If theYagi is large, for example a 40-meterYagi on a shunt-fed mast being oper-ated on the 80-meter band or a20-meter Yagi on a mast for 40meters, one can rotate the direction offire of the tower by rotating the Yagi!

Induced currents in support struc-tures that arise because the antenna isfed against the structure, or currentsinduced to flow by parasitic excitation,whether intentional or unintentional,8are an important consideration whenanalyzing the performance of antennas.These effects range from insignificant todominating and may control the direc-tion of fire. Re-radiation by structures inthe near field of the antenna is animportant consideration that is gener-ally ignored by most amateurs who writearticles about antennas. For example,we can read in our literature that such-and-such an antenna is decibels betterthan an antenna previously used, butthe evaluation was made with both thenew and reference antennas in closeproximity. Both antennas can contributeto the radiation field, no matter whichantenna is driven.—John S. (Jack)Belrose, VE2CV, 17 Rue de Tadoussac,Aylmer, QC J9J 1G1; john.belrose@crc.ca

References1J. S. Belrose, “Scale Modelling and Full-

Scale Measurement Techniques with Par-ticular Reference to Antennas in TheirOperational Environments,” AGARD LSNo. 131, The Performance of Antennas inTheir Operational Environments, August,1983, pp 2-1 to 2-25. Available: NTIS Ref.No. N84-12367.

2J. S. Belrose, “The Effect of SupportingStructures on Simple Wire Antennas,”QST, Dec 1982, pp 32-35.

3J. S. Belrose, “Loops for 80-m DX,” QEX,Aug 1997, pp 3-16.

4J. S. Belrose and L.Parker, “A Tunable AllBands Camp/Mobile Antenna (TABA): Ra-diation Characteristics determined by Ex-periment and Modelling,” CommunicationsQuarterly, Fall 1998, pp 47-57.

5E. A. Laport, Radio Antenna Engineering,McGraw-Hill, 1952, pp 104-111.

6J. True, “How to Design Shunt-Feed Systemsfor Grounded Vertical Radiators,” Communi-cations Quarterly, Fall 1991, pp 90-95.

7J. S. Belrose, “Comments on Gamma-Matching Grounded Towers,” Communica-tions Quarterly, Spring 1994, pp 74-79.

8J. S.Belrose, “Elevated Radial Wire Systemsfor Ground-Plane Type Antennas-+ Part 3: Amonopole antenna with parasitic wire ele-ments,” Communications Quarterly, Spring1999, pp 37-40.

64 Jan/Feb 2000

Jan/Feb 2000 65

66 Jan/Feb 2000