VHF-COMMUNICATIONS 3/84

Harald Fleckner, DC8 UG

A Noise Generatorwith Defined Noise Powerfor Applications up in the Microwave Range

In recent years, several construction al articleshave been published for noise generator s (1),(2), (3), fo r a noise measuring system (4), aswell CIS informati on regarding this measu ringtechnology (5), (6)

connection to an automatic noise measur­ing system , and gives further informationregarding operational experience.

1.NQISEDIQDE

", (01'1<."'''('', .o...ro"•• • t OJ)

Fig. 1: Case and dimensions of the noise diodeBAT 31

,,),10

)00

The noise diode used is a BAT 31, which ismanu tacturec by PhilipslMuila rd in a mic ro­wave case type SO 86 (Figure 1). This case is

The following article describes the con­struction of the source and provides tech­nical specifications of the noise diode, aswell as describing the construction of apower supply with switching amplifier for

The sem iconductor noise generators describ ­ed in (1) and (2) use the base- emitter d iode 01an RF-tm nsis\or operated in the backwarddirection. The avalanche er-ect required forwideband noise is not very prevalent due tothe low breakdown voltages of the RF­transistors BFR 34a and BFR 96 of < 5 V. It ismainly a tunnel effect that appears here ,which means that the noise spectrum dropsoli cons ide rably towards higher freq uencies(> 1 GH z), (7).

The described noise generator operateswith a silicon avalanche diode type BAT 31which provides a wideband noise spec­trum from < 10 Hz to > 18 GHz with a typi ­cal noise power (ENR == Excess NoiseRatio) of > 34 dB. This allows manual orautomatic noise measurements to be madeup to the satellite TV-band of 12.4 GHz withsufficient accuracy even for radio ama ­teurs.

137

VHF-COMMUNICATIONS 3/84

Fig. 2:Noise power as a f unction of frequency whenInstall ed In a 50 () system1SmA noise diode b ios

Abso lut e error l or 50nsystem Im peda nce

2,0 5.0Freouenc y (G Hz)

o

I

""'"z ' 01;\2 38•2 36• ....-r-........1-•'g 34

• 32' 'O"---",,,-~~~""",';c~~...,!;;-­III " 0•u<W

known from the varacto r and Gun n dio de tech­no logy

The no ise power as a function of frequency ina 50 D system is shown in Figure 2. The fluc­tuat ions of lhe ENR in the amateur bands inexcess of t GHz amount to ± 0.5 dB, whichcorresponds to an absolute value of 1 dBwhen referred to a nois e power of 35 dB: tfus is

a very low value with respect to the large fre­quency range,

Fig ure 3 gives the ENR as a function of d iod ecu rrent, and of the frequency. One will seecl early th at a current of 13 - 17 mA is requi redfor wideband opera tion . The breakdown volt­age is between 17 and 22 V.

nc.aNoise powe l as a function ofavalan che cu rrent w ith frequencyas parameter when installed in a50 o system

toNoi5e d iode bi" 5 curren t (mA l

l GHz

,•6o

Reco m me"d..d brosrange for brood bondoperat io n up t o 12A GHz

138

VHF-COMMUl\ICATIOI\ S 3184 - - --- - - - - -'.\

c

5tf lpl lf'e,--

Curr@1l" SQl';@ e- DC- 3k:k lf1g e- 50FL ~ A t t enuato rP8 002BA T 31

" Q:>14d B

"~ cu tcm YSWR <12

0

S,, ';hi 'Q p"~.. . "o~ ·,Oe BUG

L. ,~.,... I- 36 Ylle ~2~2 N n19A ~2OC ~ 00

nJl.JL oft

--- - - ----- - _0 0 _9V f ,o ~ 00 '" '1" ""

Fig. 4 : B lo ck diagram of fhe noise source and us power supply wit h switching ampli fie r

2.CIRCUIT

When used as noise sou rce In swncneoope­ration (manual or autc-n anc}, the componentshould exhibit a sou rce imped ance of 50 natannro es.

In c roe r to obtain this. the diode is nu.tt into a50 n coaxial or stripl ine system and terminat­ed with a wicebanc ette nuarcr of > '4 dB.VSWR < 1.2 (see Figure 4 ). The value ot thereturn loss has a decisive enect on the qua lityof the meas c-ement (see Sect ion 5) . In th is ci r­cuit. It is exclusively cetermmed by the at­tenuator. A varco of 20 dBwith a VSWR :;; 1.15up to 4 GHl and VSWR :;; 1 2~ tram4 - 12.4 GHz nas been found to bu sufficie ntlygood (Weinschel, HP, Radiall, or Greer-panHeme-mace uttenuatcrs are not recom mend­co here' With an attenuauon at 20 dB, theE:NR Will amount to app roximatoly '5 dB andits rreqcencv-oepencent uuctuanons Will nowbe determined by the diode and auenuatorGood coa xial atteouato rs have a tolerance of:; 0,5 dB.

Mosl norse measunnq systems are calib ratedlor use wilh an EXCQSS Noise Ratio ENR 0115,2 dB Thi s means that the source is virtuallycompatible,

In the author's prototype. the diode operatesin a 50 n stripline system as shown in Figures5 and 7. The switch or operating voltage is fed

to the stnpfme via three. series-co nnected150 n metal-film chip resistors. The position.dimensions. and quality 01 the restsiors havean enect on the wideband characteristics o fthe decoup ling .

Tho diod e current is adjusted with the aid of aFET·constant current source, which deter­mine s the noise power. Two par allel-connect­eo ATC·tOOchip capacitors decoup le the swlt-

h AT{: DO7 : SOJl

r~~if 0- u

h 1';oOfI-lh ip

IJATJ l

0De aUG OOS,

""

.---1 820p

Fig . ~ : SOn strl pllrle syslem wit h DC-YOlt3<,j('blo cking et the ou tpul connecl O1. andw iOOb '..>dcho ke in lhe supply line

139

VI-tF-eOM MUNICATIONS 3184

Fig_ 6: Power supply with swit ch ing ampli fi er and current source

ncU UG

Fig. 7: layou ' 01, 0.79 mm RTldumid boa rd

140

3.CONSTRUCTION

The diode system comprises mainly a double­coated. PTFE-board (RT-duroid 0.79mm) ontowhich a 50 0 line has been provided , Thecoupling ot the diode to the line must be madeextremely accurately, since it determines thevalue of the available noise power at the upperIrequency limit.

Further details regarding this are given inFigu re 8. The diode mount comprises a gold­plated Ma-screw with which the diode iscoupled to the stripline on the ground-side oftheboard via a Ma-nut. In order to achieve this,the line is provided with a 1.5 mm hole at theend. which is carefully extended to 2 mm ore­meter.The centrical position of the nut is cue­remeec by use of a dummy instead of thenoise diode (such as a defective varactor orGunn diode), After this, a small gold plate isprepared with a width of 23 rnm. lengthacoroc. 3 mm. and a 1.6 mm hole. and isplaced over the dummy and soldered to the

noise iig ure measuring system described in(II) , or with older noise figure measuringsystems [eq . tho GM 551 A)

AICU'

,,'SJRC ~io

J61

ching voltage from the attenuator. The switch­ing voltaqe should amounl to at least 33 V,which is generated most favorably using aseperaie power supply_ Figure 6 shows thecncun diagram of me power supply withswitching ampliher and current source- If theamplifie r input is provided with a modulatorpulse01, lor instance,9 V.500 Hz squarewavo­voltage, a current and voltage-controllablepulse ot the same frequency and sufficientpowor will be available at the output. Thisallows the source to be used together with the

VHF-COMMUNICATIONS 3184

Co,,:acts

Smell geld o'a:e w;th 1.6mrn hole ;I'.' _ 2.3 mm; I _ 3-4 m-n soldered to str ipline

~'/~~~~'7~,/,::,----- Copper coatingr PTFE (RTlduroid) 0.79 mm

-------- Cor r er coannq

BAT 31

-fF=='--- M4 con withhole of 1.6 mm (I;a ..:3 1.6 mmimi" 2.5 m m d" up

Fig.8:In stallation detailso f the no ise diode

BAT31

PTrE-~a rd with ~ mm hotA

Go ld-r imed M4-nut soldered into pj~c ~

ecro-ororor M4 bott wit ll llo l~

stnptme. The smal l gold plate can be obtainedtram a connect ion strip of a power trans istorThe dummy is now rernoveo, and the hole iswidened from the rear side thro ugh theM4-nut with the aid 01 a 2 .1 mm dr ill , so thatone can guarantee that the diode is coupled tothe strip line with the aid 01the lower cathodeconnection via the gold plate, With a carefu lconst ruction, one will achieve a good stnpline­coaxial transition

Alte r this, the PC-board is solde red into ametal box, The DC-switching voltage is nowfed in via a BNC-conne ctor. and the striplineshould be terminated with the lowest disconti ­nuity with an SMA or Nconnector (preferablyN-plu g). One sho uld only use preci sion N­con nector s (clean thread, gold p in, maybeusing a smaliliange). If one expects to obtainreproducible measur ing results, it is importantthat no movable coa xial con nections are pro­vided. For this reason, BNC sho uld not beused!

Final ly, the chip capacitors, chip resistors, andthe choke with the prec ision resisto r. and thetwo disk capacitors on the ground side, areinstalled.

On insertin g the diode, check for correct pola-

rity (blocked operatio n) using an ra-rneter.Fig ure 9 shows the author's prototype fromboth sidesThe power supp ly with switching amp lifier andco nstant current source can be accommodat­ed on the epoxy PC-board show n in Figure 10.fj manual operat ion is to be used exclusively,one wil l not require tho switching amplifier,which means il can be deleted from the board

4.OPERATION

Check the power suppl y before con nect ing it10the noise sautee. This is achieved by replac­ing the noise diode by a 20 V zener diode andconnect ing it via a dropp er resisto r of 470 0 tothe power supply. At an out put voltage of 36 Vfrom the stabilizer L 200, it should be poss ibleto adjust a current of 15 mA with the aid of thetrimmer potentiometer of the cur rent sou rce.CAUTION' T 2 mu st block!

Furthermore, the switching amp lif ier shouldbe driven with the mod ulator pulse from thenoise meas uring system, and the waveform at

141

=--------~~-- - _ _ _ _ _ _ _ -'V"H"F,~C"O"M~MCU~N,ICATIONS 3184

Fig .9:On the upper side of t henoise sou rce (w ith outattenuator), one w ill se eth e b locking capacito rs .the th ree ch ip resistors,and the buil t·i n diode.On the lo wer side , on ecan see the diode mo untand t he precis ion resistorso ldered int o p lacebetween the two diskcap ac itors.

the output checked with the aid of an oscillo­scope.

It everything is working correctly, the sourcecan be connected to the power supply. Thecurren t value should be aligned using thetrimmer of the current source. This can bechecked by measuring the voltage dropacross the precision resistor R" == 100 Q. Pre­l im inary, it is suff icient to adjust the cur rent to15 rnA ~ 1.5 V voltage drop. In the case of apulsed diode, one ca n measure th is best withthe aid of a RMS-voltmeler, or oscill oscope.

For relative noise measurements, or compari ­son measurements for opti mizing the receivesens it iv ity of amplifiers etc., it is not necessaryto r th e ENR to be ca librat ed exactly. However,for abso lute measurements, it is abso lutely

142

necessary for the source and attenuator to becompared to a cal ibrated no ise source overthe who le frequency range of interest.

5,OPERATIONAL EXPERIENCE

The au thor has been using a BAT at-noisesou rce w ith great success for more than threeyears. It was firstl y used in manual operationfor measuring the Y-facto r, and later used inconjunction with a noise measu ring system ,typ e 551 A, manufactured by Genera l Micro­wave

VHF-COMMUNICATIONS 3/84

,i'~'''i7 OOjJ 1 6r'~Vr-,

Fig. 10, Co mpo nent locat io ns on t he po wer supply /switching ~mp l jficr board DCa UG 006

For abso lute measurements, it is necessary toknow the ENR for the amateur radio bands.Comparisons to All-norse systems using Hj> .

sources showed that the source desc ribedhere, used in conjunction with a 20 dB at­tenua to r, exhib ited the follow ing ENR-values:

- i44l 1296 2320 3456 5760 10368 MHz144 432-

15.8 I 154 15.2 15.2 15.0 16.9 14.0dB-

Sutticient noise power is also available in theshortwave range. II is even possible to carryout alignment for maximum sensit ivity in the80 m band . However, exact noise power valueswere not determined .

The principle of noise measurements is not tobe desc ribed here, since suffic ient articleshave been publ ished regarding this subject(1), (2), (6) .

However, the author would like to give someinformation regard ing the measuring accu­racy, espec ially in conjunction with automaticnoise measuring systems.

Amon g experts, there is no doubt that all auto­matic noise measurements have a considera­ble uncertainty, espec ially in the range below3 dB The following errors are added to the

ttuctuatio ns of the source-ENR as a funct ion offrequency:

1. Accuracy of the noise metere go AIL 7514 ± 0.15 dB, GM 551 A ±0.25 dB, HP 340 A ± 050 dB, all in thelowest range .

2 Temperature error± 0.1 dB with a ± 10degree dev iation from290 Kelvin

3. Image noise powerup to + 3 dB, this can be avoided by usingsuitable circui try

4 Mismatch sou rce - fest objectThis error is not considered carefullyenough during noise measurements in therange below 3 dB. For th is reason, Figure11 shows this street graphically.

143

VHf-COMMUNICATIQNS 3/84

I_I SWR uf UUT . 1 2 Ibl SWfl of UUT = 1.50.'

,..Sou,~ SWH

~~, .zr,

"U SU

I~:::~~~~~~~~~~~uu.t.a1.25

_ 0.3

a,ae n r•

~ "i o.r

•, . G,?,$oo ro" SWR

Sou, e. ~WIl

1.25

/./ 1.2././ 1.15

U

ue

•••c.a

aa ,.,!;; .,,! ",! _ 0.1

a• - 0.2;

c.a

- 0 .4 _ 0 .4

- 0 '''0 2 ~~-:O-I"!1C:':'-o"C. -,0,'-'",0~"'a,u,'" No-- F........ CdBI

Idl SWfl of uu r - ".0

~ Sour'"SWR

~~l , 2b::::::========="

u,u

0.'

"

u.•

- 0 8

- 0 2U

_ 0,41.15

- 0.6 "1.25- 0.6

(c ) SWIl of UUT . 2.0', 0

ua

nsa Sou", " SWR s0 c.• Eg 1,25 ,

"• " 1.1!> ;;

Iu c

" - - - - - iU- 0> 1.15 !• t.z a, - 0,4 1.?!> ,, •

- 1.0 . ---L-J

o 2 4 6 8 10 12 14 16 18 roMeawr«l,.,.,., FiQutl' C<lBI

- 1 0 ----'- -0:-",-0,-,'o 2 4 6 8 10 Q 14 16 18 W

M<r_."" N",... ~•.,.,. " (d 8 )

Fig , 11; Ma.imum measuring eflOI a s a function 01th e me.suled n o ise hgure, as we ll as 01the VSWR·Ylllu.. sQfthe nQise SQurc e an d test o b tet1 (UUT. Hewlel1·P.tk....d).

Example: In the case of a determined noisefigure of 2 dB, a source·VSWR of 1.2, and atest object-YSWR of 2.0, the measuringuncertainty will be in the order of ± 0 5 dB!

As can be seen further, the retom loss of

' 44

Ihe noise source cannot be large enough inorder to provide tolerable measurements inthe range below 2 dB, especially when thetest objects posse ss a high stand ing waveratio.

VHF-COMMUNICATIONS 3184

If these errors are added in the correct sign,one will arr ive in a tolerance range that makesthe decimal oa-veroee seem rid iculous Evenif one assumes that those can appe ar withopp osite signs, and the actual accura cy of themeasur ing system is greater than the tole­rance value. one should note that diffe rencesof 05 dB must be accep ted, whe n using thesame preamplifie r at differ ent times with di ffe­rent measuring set-ups on the same measur­ing system.

This just is to demonstrate how questionablemost noise figure values under 1 dB are it theyare not measured using an exacnv calibratedhot/cold standard, but have been autom au­cauy measured , and tnat temns-cr-a-oa­values can be ignored.

To summarize, it can be said that the possib i­l ity of compariso n measurements in the SHF­range represents an important aid for thehome const ruc1or and cr itical amateu r consu­mer, especially consideri ng the incre asingact ivity. Absolute values should be of secondimportanc e to radio amateu rs, since the opti­mization of every receive system must bemade including the antenna , and Ihis IS

achieved best using solar noise(8)

6.REFERENCES

(1) Michael tnbrtcnt . DB 2GMA Noise Generator for VHF and UHFVHF COMM UNICATIONS 13.Ed ition 1/1982, Pages 38- 43

(2) Or A. Waxweiler:Rauschgenerator mit detinierterRauschleistungcq-DL 12/1981, Pages 585- 587

(3) H. Fleckner:No ise Generator with BAT31Dubus Info 3/1983. Page s 188 - 189

(4) Martin Dohlus:A Home-Made Autornanc Noise-PiqureMeasu ring SystemVHF COMMUNICATIONS 15.Ed ition 111983, Pages 2- 11VHF CO MMU NICATIONS 15.Edition 2/1983. Pages 66 -83

(5) Gannaway/Hol mes :Some Pitfa lls in Noisa-FlqureMeasurementsVHF COMM UNICATIONS 14.Editio n 1/1982. Pages 44 - 48

(6) B Stein:Automatic Noise-Fiq ure Measu rementsFacts and FancyHam Radio Magazine 1978.August. Pages 40 - 54

(7) Unger/Schultz:Elektrc niscbe aao erernenteund Netzwerke IVioweg-Verlag Braunschweig,ISBN 3-5281-3505- 0

(8) G. Hoch. DL6WU:Determining the Sensihvilyot Receive Systems with the Aidof Solar NoiseVHF COM MUNICATION S 12,Edition 2/1980, Pages 66-72

It is nCN ' possible lor you to order magaz ines. kits etc. using yourVISA Credit Card !To do so. please stale your crec nt-caro number and the validity date,and sign yOur orce r.

Yours - UKW-BERICHTEIVHF COMMUN ICATIONS

145