VHF COMMUNICATIONS 1/94'~\: --- - --- --- - --"''--'''= = '''-'''= -'=
Dr.lng. Jochen Jirmann , DB 1 NV
Intermodulation Properties ofSwitching Diodes
" •• - ••• • Ig .?O M ~z
jour of commercially available highfrequency switching diodes was meltsured in a second investigation. Th eresults were obtained ustng resourceswhich were sHU almost on an ama teur level, and should thus no t heput down to the "dB scales", T hecomparison between the variousdiodc types is actually more important than the absolute values.
I;:':" .~ t.... ' m...- fI "", 8010A
,m~ ' '",,:'""''' ~ '-(r~:(r
Some attempts 10 improve the intermodulation properties of short-wavereceivers were described in (1). 11 wasdemonstrated there that the mainreason for the mod erate intcrmodula lion properties of many shortwave receivers should he looked forin the usc of unsuitahle switchingdiodes for the switching of the inputhand pass filters. Following numerousenquiries, the intcrmodulation bchav-
T• • t n I
IMIOi(i""'I'1P" (1 -1.T• • , n , I~, w l, tt -Po [j.;or d
OB 1NV I "'",, L;HP Thinkj,t 1- o 5011H ,
Fig.1: Measuring Rig for the measurement of the IntermoduletlonCharacteristics of Switching Diodes
Measurement Curvewithout Diode (1M2measurement).Circle shows locationor IntermodutattonProductSA: centre 25 Mill;
\: -30 dBm
~~~~~~--------(~VHF COMMUNICATIONS 1/94
I.THE MEASURING RIG
The previous expe riments. using anIC765 from OM Hercher, DL8MX. haddemonstrated that the critical leve labove which audible intcrrnodulationsarise should be 5011ght at an aer ialvo ltage of about IOOmY. 'I11is corresponds to an output of -6dHm. Bydefinition, an Sl signal has a level of-121dBm, so that the measuring rigmust process a dynamic range of 115dBto detect weak 1M products. 'Ibis is justabout possib le using commercia l measuring technology of the most expensivekind. In order to obta in usable resu ltswith amateur resources, meas urementswere carried out only at selected fixed
frequenc ies, and the frequency diagramwas drawn up in such a way thatharmonics from the test transmitter canbe separated from the 1M productssought. With some filters. a measurement dynamic range could be usable atabout 9OJB. The measuring rig issketched in Fig.I.
Two test transmitters act as signalsources, a Singer SC' 1000 and aHewlett -Packard R640A, the outputs ofwhich are combined by means of apower adder (Mini Circui ts I'SC2-1).The lest trausmiuer power is ImW orOdBm. A simple low-pass filter madc ofproprietary chokes (Siemens MCC.0.82 " H) with a limiting frequency of15 MHz reduces the inherent intennodulation of the test transmitter to below-100dBm. The low-pass filter is fol-
VHFCOMMUNICATIONS 1/94(fl ------ - --- -"-'-"-""'''''''''=-''''''''-!!.'''S~ e ~ ," ~ l e"4h. e , or oe I NV, Uer OlOn T,IS "C~ ~ 1 . ~ c . 9:
e r . " "' Cr ~C I Ii~ H,r.<J .t ." t 13: P"e ; ll oll
- 30 dB m
lowed by the test diode, which is biasedwith adjustab le DC . A high-pass with alimiting frequency of 20 MIIz rel ievesthe spectrum analyser (home-made bythe author) of the strong carrier wavesignals from the test transmiucr. Theanalyser was set to an average frequency of app . 25 MHz and 105 MHz/d iv.
For the measurement of total frequencies (second-order inrennodulation), the test transm itters operated at15 and 12 MHz. The second-ordermind product at 27 MHz can thuseasily he separated from the testtransmitter harmonics at 24 and30 Mllz.
To mea sure third-order intcrmod ulation,the test transmitters were set to 15.5 Hzand (. MHz and (he mixed product wasmeasured at 25 MHz.
To check the measuring ng. the diodewas short-circuited. foig.2 shows theanalyser screen print-out . The two testrransmiuer signals can be recognised(here 12 MHz and 15 MIV.), togetherwith their harmonic s at 24. 30. 36 and45 MHz. TIle reference level at the topedge of the screen was ·30d l3m here. sothat the reduction of inherent intennodulat ion products could be estimated atbetter than 85dB. A circle marks theposition of the 1M product to beexpected.
VHF COMMUNICATIONS 1/94
[".~ . 10
1 ·~l:1i: 'DB1NV co." ., "••."....
Fig:.4: Second-orde r Inter mod ulattonProducts plotted a~ainsl DiodeCurrent
In this range, measuring frequencies of12 and 15 MHz were used. 'th e diodeDC was varied from ZmA to 20rnA.The lest diodes used were a IN414X, aISS53 (from an IC765). a BA379. aBARIZ-l and a 81\244. A typica l 1Mspectrum ca n be seen in Fig.S. Here a8A379 was operated at a low current of2mA. The intennodulation signal canbe clearly recognised at 27 MHz, with alevel of -6OdBm between the firstharmonics of the test transmitter. Theintermodulation intervals measured for
various diodes ale plotted against thediode OC in Fig.4. As can he seen, thefirst round in the 1M contest goes to the8A379 from Siemen s, followed by theBAR12-1 and the 15553. The goodcut-off results from the 15553 universaldiode are surprising. But since thediodes removed from the lC765 carriedno type description. it might perhaps beconceivable thaI lCOM had secretlyused improved diodes here . It isn' t clea rfrom the parts list. It can clearly beseen how important a sufficiently highlevel of IX~ through the diodes is, sinceat current levels below lOmA theiuterruodu lati on products increasegreatly .
In this measurement range, the testtransmitters were tuned to 6 and Is.sMHz and the 1M product was evaluatedat 25 MHz . The diode rx: was alte redhere at only two values, 2mA and ."l mA,and the same diodes were used as inSect ion 2. Fig.5 shows the inherentinterference spectrum for the measuringrig. Ptg.c the 1M spectrum for aIN4l48 misused as a switching diodewith a diode current of SmA . Thereference line is at -IOdBm and the 1Minterval for third -order products isabout 2OdB! Fig.7 further shows theintermodulation intervals measured forthe various diodes. As can be seen, theBA379 gives the best results here too,followed by the HARI2-I, whereas the15S53 falls off markedly .
VHF COMMUNICATIONS 1/94(4'------------'-"'--=""""="""~i~e . ,~ua"" l .... o.. C~ ; "v. ~ . -"on ' . ' 8 ' O~ ~ I . ~ = . ~:~" £ ' . Io.. . <' ~ ' h,M. e ' ~ " l 'i ~ 0 " . 11 ::: 0 11
,I , ,
\. - 40 d8m I: : , : !
,II l ;
:\I --U1 !, Ih ;., I • ,i ;
Fig.S:Measure ment Cur vewit hout Diode (1M 3measurement).Circle shows locallon ofIntermodulalionProduct
SUMMARY OF RESULTSSO FAR
The measurement rcsuns listed essentially show four things:
Good, repeatable inrermcdulatio n interv als ca n be obtained only throughthe use of "correct" PIN diodes. butthey have their price. Miniaturere lays arc even better, but moreexpensive and bigger.
Universal diodes misused as highfrequency switches can yield verygood results (5553) or catastrophica lly poor result s (1N4 148).Moreover. it can not be calculatedwhat effect variations in themanufacturing parameters will have
(different production lines, differentproduction methods).
The relatively good cut-off resultsobtained in pract ise from the apparatus fitted with tuner switchingdiodes is not consistent with thepoor measurement results from the0 1\244 .
The existing apparatus should alsobe improved or re-constructed inorder to check whether sufficientOC is flowing through the diodes.An attem pt should he made to set avalue of about 20rnA by altering. theprotective resistors. It can be concluded from the results that the maincause of intermodulat ion interference in short-wave amateurreceivers should be sought in thearea of the high-frequency inputswitching diodes.
VHFCOMMUNICATIONS 1/94'-'-"-'====-"-""----- - ------ 'I,. \Fig.6:Measurement Curve:1M3 Spec tr um of aIN4148 wilh SmADiode Current
Sp•• t ",l . "" ~ , .. \ o~ OB I NV, V. " i o" 1.l a ' 0.' a ' . ~ " . B ~
G<- . h .<lru"' ~ p ' ~ , c' . e \ ",t In 0" .1 /,0;1
, ,\, , ,,I
- 10 dBm,i, ,
, , , ,, ,,I
, ,, ,
!:I [ I I
1 Il l i'\! ~~ ," I 'I,\ii iIjI lIM \~ 1 1 1, :v-' I II
; ~ I, • , ,
But since ove r-modulated coil s withferr omagnetic cores can also generateintermodu lation effects, the same meas uring rig was used to classify inductivecomponents.
5.INT ERMODULATIONS ININDUCTANCES
Here both the aperiodic case (coil aschoke) and the resonance case wereinvestigated. In the latter case, the coilwas brought into series resonance witha high-qua lity foil trimmer at 15 MHzand subjected to 12 and 15 Mllzmeasurement frequencies. Coils or series resonance ci rcuits were inserted intothe measu remen t circuit instead of thediodes. The following observati onswere made here:
With a choke effect, intennodulationproducts above -II OdAm werc notdetected either for rod core microchokes from the Siemens MCC ra l1 ge or for Neosid and TOKO readymade co ils selected at random. On lythe "VK200" six -bore core chokefrom Valvo or Philips Components,a favourite with VHF Communications readers, yie lded an 1M leve lof between 85 and 95d Hm,depending on the ferrit e materia l. Arx:: level of 50mA did not influencethe readings for any choke.
In resonance mode, the Neosid andTOKO ready-made coils, togetherwith some very sma ll ferri te ringco res, came up with 1M levels of- I00 ( 0 - J05dllm. The Siemens chokes stayed the co urse emazing tywell. Thei r intermodulation could beplaced at around - I IOdBm, Some
%..:;:,~,~""~,'%..~,~,~,~"-100 DB 1NV Li•• , 0 1. J1u... ~. ...
1:: In this connection. we might recall theband-pass filters with ring core coilspublicised many years ago by VE3TP,which were not exactly cheap to construct . but on the other hand have solvedevery receiver 1M problem so far. Thisstatement shows that. in spite of statements to the contrary from the industryand from a few. probably unquali fied,"specialists". it is possible to producereceiver input components which canm eet today's requirements in relation tosensitivity and high-level signa l strength . Since in our hobby we don 'tneed to worry about tenths of a peM y.like industri al manufacturers. we canobtain results which arc some orders ofmagnitude better for a slightly incre ase d cost!
3. Chokes in the filter structure, e.g. onthe operating voltage feed. arelargely uncrit ical. as long as they donot resonate.
2. In compact rigs. rod core chokes,such as the Siemens MCC. can beco nsidered as alternatives .
BARll - '
: - 10
Fig.7: Third-order InlermodulalioDProd ucts plott ed against 20lAlind SmA Diode Curre nts
: - 60·,"
Amidon ring cores, suitable for short-wave usc and of various sizes.were practi cally free from intcnnodulation.
The following des ign lips can thus bederived, some of whic h are in any casenot ne w, but which have prohahlyfallen into oblivion in Japan:
I. Input filters ef fectively resistant 10
1M can be produced only usingsufficiently large iron powder ringcores as inductances. They offer thebest compromise between the spacerequire ment and the level controllability ,
The author hopes that this acco unt ofhis measurements will start peopl ethinking about experi ments of theirown, and would be pleased to receivereport s of their experiences.
(I ) Dr. lng. LJirmenn. DBJ NVWilrried Hercher, DL8MX:Improvem ent in IntennodulationBehaviour of Modem Short -waveAmateur ReceiversVIIF Comm. 1/1993 pp. 38 - 43