A Shock To The System

I never thought that I would beso surprised by a receiver as tonot know how to startdescribing it, but I’m sittinghere wondering where tobegin, and how to tell you ofthe experience. The receiver inquestion is a brand newproduct from WiNRADiOCommunications. in Australia,and is marketed under theirWiNRADiO brand name with amodel number G-303i. As youwill know, WiNRADiOCommunications haveproduced a series of receiversunder the WiNRADIO nameover the last few years, butthese have included widebandmonitoring devices coveringfrequencies up to 1 or 2GHz,

and mechanically configured asplug-in cards to fit PCexpansion slots, that is to saythey sit inside the PC itself.Early personal computers wererenowned for being generatorsof high frequency electricalnoise, and you could usuallydetect a computer’s presenceby having a typical h.f. receiveranywhere within 50 to 100metres of the culprit. With theadvent of the European EMCDirective the situation haschanged, and there is no doubtthat modern computers fromreputable manufacturers aremuch less noisy than those ofeven a few years ago. Still it’s abrave manufacturer who willput a sensitive receiver actuallyinside a PC, but WiNRADiOCommunications have persisted

with development work ontightly screened enclosures andproper r.f. shielding with theresult that thecomputer/receiver marriage isnow a little more blessed thanbefore. So what makes the G-303i different, and why was Isurprised?

Proper ReceiverThe first thing to know is thatthe G-303i is not a widebanddevice but a ‘proper’ h.f.receiver specified to cover 9kHzto 30MHz, and equipped withall the facilities which we havecome to expect from a modernreceiver. Some lateral thinkingby WiNRADiO Communicationshas resulted in their analysis ofa typical receiver as consisting

of three separate sections; ther.f. and i.f. conversionelements, the demodulationand signal analysis elements,nowadays often carried out bydigital signal processing(d.s.p.), and a processorcontrolled management systemwhich drives the whole receiverincluding the frequencysynthesis for frequencyconversion and any front paneldisplays and so on. WiNRADiOCommunications haveessentially pointed out thatanyone with a semi-decentcomputer already has theprocessing power, and, with areasonable sound card fitted,also has the d.s.p. system, sothe only missing element of areceiver is the r.f. and i.f.conversion system, and that is

REGULAR NEWS FEATURE BROADCAST PROJECT SPECIAL COMPETITION QSL REVIEW BOOKS SUBS PROMO

WiNRADiOG-303i - PC Card HF receiver System

UK’s First from the UK’s First!

John Wilson brings us an in-depth look at thenew G-303i receiver from WiNRADiO. This firstoffering from a new series of receivers uses acomputer sound card to perform the final i.f.and demodulation functions. Coupled to thebenefit of a very user friendly display, this isone desirable radio.

24 Short Wave Magazine, February 2003

24,25,28,29,30 Win Radio 8/1/03 1:29 pm Page 24

precisely what the G-303ihardware consists of. The cardwhich you plug into the PCcontains r.f. front-end filtering,a mixer to convert to a first i.f.of 45MHz, some gaincontrolled stages at i.f., asecond conversion down to atypically d.s.p. second i.f. of12kHz and then the 12kHzbaseband comes out of thecard ready for processing in thecomputer’s existing sound card.It’s a very neat concept, and bydoing all signal processingunder software control it ispossible to carry out almost anyfunction you might need.WiNRADiO Communications’other strength is anacknowledged expertise ininnovative software, and theyhave really excelled themselvesin this department with the G-303i. Let me tell you about it.

What You GetThe G-303i was supplied forthis review already fitted to anIBM NetVista computerrunning a 2GHz Pentium 4processor and having 256MB ofRAM and a fairly whizzy soundcard. The monitor was a veryattractive IBM flat panel unitwhich I immediately wantedfor myself! This fits in well withthe recommended systemspecified in the G-303i manual,with a minimum systemrequiring at least a 500MHzPentium III with 64MB of RAM,20MB of free drive space and aSoundBlaster 16-bit full duplexsound card. There are severalwarnings in the manual textabout the whole system‘freezing’ should the G-303i befitted to a slower PC than thisminimum specification, and Iknow from previous experiencewith digital test instrumentsthat this can be a real problemif you try to compromise. Theright course of action is to gofor the fastest and bestequipped PC you can find (orafford). Firing up the G-303i

software displayed a screenshowing a receiver front panel,but with some features notnormally found on aconventional receiver. Thenecessary frequency displayand tuning knob were obvious,but the bottom left handcorner was occupied by aspectrum display centred onthe receiver tuned frequencyand extending for 10kHz oneach side. Overlaid on this wasa representation of the i.f.bandwidth currently in use,which changed as differentmodes and bandwidths wereselected, and which clearlyshowed, for example, that myfavourite Radio Five Live wasoccupying rather more of thespectrum than could beaccommodated within thestandard 6kHz a.m. bandwidthof the receiver. The displaylooks exactly like any of thoseyou may have seen in myreviews taken with a Rohde &Schwarz FSA analyser, and I dida quick check by connecting myFSA to the 12kHz output fromthe G-303i receiver, whereuponI could see that the on-screenG-303i display was identical.The only difference is that theFSA originally cost about£40,000 whereas the G-303icost - well, I’ll surprise youlater.

Tuning controls arecomprehensive, with aconventional looking mainknob above which are left andright arrows associated with a‘step size’ display and twomore arrows which run thesteps at 10 times the selectedrate. The step sizes available,selected by tiny up and downarrows range from 1Hz to100kHz in a sensible selectionbut for the omission of the all-important 9kHz step forEuropean long and mediumwave broadcast bands.However, the step size can beset to anything you might wantby simply entering the stepvalue into the step size box

using the PC keypad, so 9kHzcan be entered and used,although this is lost if you thenselect another ‘standard’ stepand try to come back to 9kHz -it ain’t there any more. A finalneat trick with the step size isthat if you place the mousedriven cursor between the upand down buttons, a sliderappears, and by dragging thisup and down the screen, themouse tunes the receiver upand down in frequency. Butthat’s not all.

TuningThe main tuning knob is‘rotated’ by placing the mousecursor at the top or bottom ofthe knob and clicking, whichthen tunes the receiver up ordown in frequency. At first thetuning step (500Hz) seems verycoarse, but by pressing the ‘Alt’key on the PC keyboard thesteps are reduced to 50Hz,whilst the ‘Shift’ or ‘Ctrl’ keyswill increase the step size to5kHz for rapid frequencyexcursions. I personally foundthis cumbersome in use, andmuch preferred to use the‘step’ facilities for tuning. Youcan also turn the main tuningknob up or down in frequencyby using the cursor up/downkeys on the PC keyboard, butthere is a curious anomaly herebecause although the tuningknob rotates, and the receiverchanges frequency, thespectrum and frequencydisplays both ‘freeze’ so youdon’t know what frequencyyou have tuned to until yourelease the cursor key. Not atall sensible, bearing in mindthat the other tuning methodsshow the frequency andspectrum changing as youtune. The same ‘freezing’occurs when using the ‘PgUP’and ‘PgDown’ keys to singlestep the tuned frequency fromthe ‘step size’ facility. But that’snot all.

Below each digit of the

frequency display are up/downarrows with the centre ‘slider’facility, and since the displayshows frequency down to 1Hzyou can move frequency at anyrate you choose, even tocomplete end to end frequencychanges using the 10MHzslider. But that’s not all. In thebottom right hand corner ofthe front panel is acombination display showing arow of yellow squares, eachone representing increasingtuning step size moving up ordown in frequency from zero,and a quasi-analogue displaygiving a variable tuning steprate as you drag the mousecursor along it. Choosing eitherof these methods from themouse will tune the receivercontinuously up or down in thechosen frequency steps untilyou release the mouse key. AndI think that’s all, but what anarray of facilities from which tochoose. Surely no-one can failto find a preferred method oftuning from this selection. But,actually, I forgot to mentionthat we old fashionedoperators can simply key in anywanted frequency from the PCkeyboard. Below 1MHz youneed to terminate the entrywith ‘k’ for kHz, but it’spossible to be on a newfrequency in a fewmilliseconds, and that’s oftenfaster than the operator of theservice to which you arelistening.

Before leaving thefrequency display, anotherfeature is the window belowthe frequency in which a textmessage appears telling youwhat service is allocated to thefrequency displayed. This canbe a band of frequencies, suchas an amateur band, or a singlefrequency such as RAF Volmet.The text can be changed bygoing to a text file held in thesoftware and typing inwhatever you need, or, asWiNRADiO Communicationssuggest, update the whole lot

REGULAR NEWS FEATURE BROADCAST PROJECT SPECIAL COMPETITION QSL REVIEW BOOKS SUBS PROMO

Short Wave Magazine, February 2003 25

24,25,28,29,30 Win Radio 8/1/03 1:29 pm Page 25

from an Excel spreadsheet. Icontented myself by correctingsome of the entries foramateur band allocations fromAmerican to European bands,just to see how it worked. Itworked well, and it wasfascinating to see theallocation descriptionschanging as the receiver wastuned around.

RSSIBefore going on to thedemodulation modes provided,let me tell you about the signalstrength metering. The ‘S’meter is an analoguerepresentation of a real meter,but unlike the average receiverhas three selectable ranges,calibrated in dBm, ‘S’ units andmicrovolts. In addition to theanalogue display, the signallevel is also displayed innumerical form at the bottomof the meter scale, and whenreading ‘S’ units or microvoltsthe dBm scale is still shownbelow the analogue scale, soonce again WiNRADiOCommunications have tried toplease everyone - even me!Associated in a way with thesignal level is a very neat all-mode squelch system in whichthe fully adjustable squelchsetting is displayed as a redband on the signal meter,showing the exact level in dBmto which the squelch is set. Thisis extremely clear andunambiguous, and very easy touse. Not only that, thecalibration in all modes is veryclose, although it is clear thatthe meter units change in1dBm steps which leads tosome peculiar conversionswhen in ‘S’ units or microvolts,with signals either being S3 orS4 but never in between. Nomatter, the metering isamongst the best I’ve seenoutside of professionalmeasuring receivers.

DemodulationAt this point you need to knowthat the G-303i can be suppliedwith ‘standard’ demodulatorsoftware, or the ‘Professional’demodulator which gives arange of additional andextremely useful functions. Thestandard unit provides for a.m.,with a bandwidth of 6kHz,a.m.n. with a bandwidth of4kHz, a.m.s. giving synchronousa.m. with the 4kHz bandwidth,l.s.b. and u.s.b. with

bandwidths of 2.5kHz, c.w.with a bandwidth of 500Hz,and three f.m. modes withbandwidths of 3kHz, 6kHz and12kHz. Remember that becausedemodulation and filtering isdone at the final i.f. of 12kHz,these bandwidth filters aredigitally derived and may ormay not equate to anequivalent conventional filterused in a more conventionalanalogue receiver. Theprofessional demodulatorextends the bandwidthsavailable and also adds a veryeasy to use continuousbandwidth adjustment usingthe previously mentioned‘slider’ control. This is a veryuseful feature and allows youto exactly match the filterbandwidth to the incomingsignal whilst taking account ofadjacent channel noise. Hard todescribe, but once experiencednever to be forgotten.Additionally, the professionaldemodulator adds d.s.b.(double sideband) and i.s.b.(independent sideband) modesto those available as standard.Neither demodulator optionincludes a notch filter or passband tuning.

The professionaldemodulator is definitelyworth including in thepurchase of the G-303i becauseit has a comprehensive userdefinable series of parametersfor the i.f. filters and a.g.c.settings, together with afascinating graphic display ofeach demodulator systemaccording to the modeselected, with the facility to‘connect’ an on-screenspectrum display to each partof the demodulator to showhow it is all working. Even thisis then extended to includereal-time SINAD and Totalharmonic distortionmeasurement on the incomingsignal. The user manual is very

helpful in leading the userthrough the various options,and if you really get in a tangleyou can restore the normaldefault settings at the touch ofan on-screen button. It really isall so easy to understand anduse.

As if that were not enough,clicking on a yellow arrowalongside the power switchdrops down a spectrumanalyser display below themain receiver panel, on whichyou can enter start and stopfrequencies and, frequencymeasuring increments. Startinga frequency scan can be doneover frequencies having norelationship to the frequencyto which the receiver is tuned,so you can be listening to astation on 900kHz and carryout a band scan from 9 to12MHz then the receiver willrevert back to the originalstation. The receiver is mutedduring the scan process but ittakes only a short time tocomplete. When the scan isdisplayed, the mouse pointercan be used to select anydisplayed frequency, and themeasured level is shown indBm. Clicking the mouseinstantly tunes the receiver tothe selected frequency, andsliding the mouse drags thereceiver frequency to matchthe mouse movement. There isonly one drawback to thesystem, which is that themeasuring bandwidth is fixedand very broad, so that if yousweep across say, 9 to 9.5MHz,instead of a series of sharpsignal peaks, you have a broadwavy line with veryindeterminate peaks. Thefacility is fine for sweepingseveral MHz, but not forlooking at a single broadcast oramateur band. There seems tobe no way of changing themeasuring bandwidth, and it isnot related to the selected

bandwidth in the mainreceiver.

Harnessing The PowerMy description of the receiverfacilities ends with the a.g.c.system, because I must get onand tell you how the G-303icame out under the testspotlight. The r.f. section of theG-303i is, as I have said,contained on the plug-in card,and can be considered as afrequency changer. This sectionof the receiver system isbasically analogue, and as Ihave seen previously in somed.s.p. receivers, the use ofdigitally derived a.g.c. has hadits drawbacks. The G-303idesigners have included ana.g.c. system within theanalogue section of thereceiver, and this is accessedfrom the on-screen front panelwhere you can select decaytimes of slow, medium andfast, together with an a.g.c. offfunction. These a.g.c. settingsare independent of the digitaldemodulation, but in bothstandard and professionaldemodulators there is a furthera.g.c. system which works onthe demodulated audio. In thestandard demodulator this a.f.a.g.c. has a single timeconstant, but the professionaldemodulator allows the user toset the attack and decay timesfor each of the three a.g.c.settings. Using this facility givesgreat insight into the effects ofinappropriate attack times onvarious types of signal, and is avaluable facility. All the morereason to opt for theprofessional demodulator.

So there we have a generaloverview of a verycomprehensive receiver from auser’s viewpoint, but thismerely demonstrates the skill(and it is a very real skill)inherent in the WiNRADiO

28 Short Wave Magazine, February 2003

REGULAR NEWS FEATURE BROADCAST PROJECT SPECIAL COMPETITION QSL REVIEW BOOKS SUBS PROMO

24,25,28,29,30 Win Radio 8/1/03 1:29 pm Page 28

REGULAR NEWS FEATURE BROADCAST PROJECT SPECIAL COMPETITION QSL REVIEW BOOKS SUBS PROMO

Short Wave Magazine, February 2003 29

Communications softwaredevelopment staff. What wehave to remember however, isthat this is an h.f. receiver, solet’s proceed on to the realtesting and see if the facilitiesand features live up to theirpromises. By the way, I haven’tforgotten the memoryfacilities, but will treat them asa PC feature.

A Matter OfCleanlinessOne thing which has plaguedmany receivers is thegeneration of internal spurioussignals, and with the G-303iliving inside a computer I madethe search for these unwantedsignals my first test. The jobwas made really easy becauseall I had to do was terminatethe antenna connector in ascreened 50Ω load and set thereceiver to sweep its entirefrequency range in 100Hz stepswith the squelch set to its mostsensitive setting. I could thenlet the receiver carry on whilst Idid something else, secure inthe knowledge that if aspurious signal was found thereceiver would stop until I

manually re-started it afternoting the frequency and levelof the unwanted signal. Howmany signals did I find?Unbelievably only one, at afrequency of 20MHz and at aninsignificant level of less than0.3µV. I was so astonished thatI did the whole test again, butthe result was the same. As faras I can remember, I have neverfound any receiver, analogueor digital which had suchcleanliness, and the G-303i hasset a new standard for othersto try and emulate.

Since I was on the trail ofsignal cleanliness I decided tocheck the G-303i synthesiserperformance by carrying outmy standard reciprocal mixingtests. You are by now allfamiliar with this test, with theresults expressed as the phasenoise in dBc/Hz (the noise in atheoretical 1Hz bandwidthextrapolated from a realmeasurement in a chosenreceiver bandwidth). At 5kHzfrom the on-channel frequencythe phase noise was -96dBc/Hz,at 10kHz -123dBc/Hz, at 20kHz-133dBc/Hz, at 50kHz -141dBc/Hz and at 100kHz -148dBc/Hz. To put this intocontext, the performance at100kHz is close to that of acrystal oscillator, and it is onlyat 5kHz that the phase noiserises. For a further check Imeasured my faithful AR7030,and whilst this was 20dB betterat 5kHz (-116dBc/Hz against -96dBc/Hz), the performance ofthe two receivers was equallymatched for all the otherspacings checked. Anothergood result for the G-303i.

More MeasurementsMeasured sensitivity for s.s.b.

12dB SINAD was substantiallyflat at -122dBm from 20 to3MHz, falling by some 2dB by30MHz and down to asurprising -113dBm by 1MHz, achange of 9dB. Detailedinvestigation revealed that at950kHz the sensitivity was -117dBm but at 951kHz it fellby 11dB to -106dBm. Taking alook at the return loss of the G-303i antenna socket showedthat there were some bandpass input filters with changeover frequencies of (what asurprise) 950/951kHz,1.8/1.801MHz and at7.3/7.301MHz. It would seemthat the filter for 951kHz to1.8MHz has much more lossthan the other front-endfilters, but whether this isnormal or a fault I am unableto say. The receiver publishedspecification gives the lowestreceive frequency as 9kHz, andchecking sensitivity in thisrange showed a gradual rolloff starting at 200kHz with asensitivity of -115dBm down to-92.5dBm at 20kHz. All thesemeasurements were taken withboth r.f. and a.f. a.g.c. systemoperating; without the a.f.a.g.c. the figures would haveshown even more variation.However, it has to be said thatRugby at 60kHz was rompingin, although 16kHz GBR neverdid show much life.

One thing struck meimmediately about the h.f.sensitivity and that was, basedon my experience with lots ofother receivers, the sensitivitywas higher than that necessaryin a receiver of this type, andthat the high sensitivity wouldprobably reveal effects on ther.f. intermodulationperformance. Using theaccepted technique of feedingin two r.f. signals, 20kHz apart,and raising their combinedlevel until an intermodulationproduct appeared and equalledthe minimum discernible signal(MDS) of the receiver gave mea third order intercept point of+3dBm. Repeating the testusing the ‘big signal’ methodused for professional receiversgave a slightly better thirdorder figure of +4dBm, with adynamic range of 92dB.

Second orderintermodulation performancedepended on the positioningof the two test signals relativeto the band pass filtering in thefront-end of the receiver, butwith the test signals in bandwith the IM product I measured

the second order intercept at+26dBm with a dynamic rangeof 80dB, whilst with the testsignals out of band relative tothe IM product I measured+36dBm with a dynamic rangeof 85dB. I have to make acomparison with anotherreceiver, if only to check thatmy test configuration is correct,and I chose my regular AR7030.This returned figures of thirdorder intercept point of+32.5dBm with a dynamicrange of 103dB, and a secondorder intercept point of+65dBm with a dynamic rangeof 92dB, and it should be notedthat the AR7030 has no inputfiltering to assist the secondorder performance. It’s alsoonly fair to say that the AR7030is a more expensive receiverthan the G-303i.

As a final comment, there isa growing belief based onextensive work by well knownauthorities on the thornysubject of h.f. receiverperformance, that it is thereciprocal mixing parameterwhich is paramount indetermining what constitutes agood or bad receiver, and it istrue that the intermodulationcharacteristics can be improvedin real life operation by thesimple expedient of puttingsome attenuation between theantenna and the non linearstages of a receiver. WiNRADiOCommunications have includedjust such a switchableattenuator in the G-303i andwhilst at 18dB it may be toomuch for my preference,switching it in will seriouslyimprove the third orderintercept point albeit at thecost of raw sensitivity, but theG-303i is, as I said, verysensitive, and is already aheadin the reciprocal mixing area.

One of the most revealingtests I normally carry outconcerns the behaviour of thea.g.c. system of a receiver, andit often reveals the reasons forthe odd clicks and pops onehears at the beginning ofspeech syllables and also at thestart of a c.w. dot or dash.Basically what I do is tocombine a steady low level r.f.signal with a stepped increase,usually between an equivalentof S4 and S9 on a standardmeter. The length of theincrease is usually set to about250ms and I look at the audiooutput of the receiver duringthe length of the step increase.All receivers utilising digital

24,25,28,29,30 Win Radio 8/1/03 1:29 pm Page 29

signal processing take a finitetime to carry out theprocessing function and thisleads to a measurable timedelay between application of asignal at the antenna and theaudio output appearing at theother end of the chain. Sincethe r.f. step has been appliedbefore the digital system hasdone its job, a fully digitala.g.c. system has a hard timecoping with the overloadwhich occurs before the a.g.c.control has been applied, andthis manifests itself as a loud‘click’ in the audio output. Oneoutstanding example of thisoccurs in the Collins 95S-1which goes into a wild‘squawk’ under these overloadconditions. WiNRADiOCommunications haveovercome some of theseproblems by having a real timehardware a.g.c. system in thereceiver card itself, and this isthe ‘r.f. a.g.c.’ shown on thefront panel. However, therealso has to be an ‘a.f. a.g.c.’within the d.s.p. system andthis is also selectable from theG-303i panel. Let’s see how thereceiver performed.

A Picture Paints...The receiver’s audio responseduring a 250ms input signalstep, with the r.f. a.g.c. on andthe a.f. a.g.c. off can be seenin the plot Fig. 1. The stepincrease can be seen and isclean. although the audiooutput level changes despitethe r.f. a.g.c. You can clearlysee the processing delay ofmore than 100ms betweenapplication of the signal at theantenna and the audioappearing at the output of thesound card. Switching the a.f.a.g.c. on to add audio controlcauses a dramatic change asseen in Fig. 2 where the a.fa.g.c. has brought up the lowlevel r.f. signal to the start ofthe increase, and then the r.f.a.g.c. has taken over to keepthe audio output more or lessconstant, but at the transitionfrom low level to high levelsignal there is a loud andobvious ‘click’ in the audio.Switching to the professionaldemodulator with both r.f.and a.f. a.g.c. switched onshows in Fig. 3 just how muchbetter the ‘pro’ softwareworks The transition from lowlevel to high level input signaltakes place with hardly anychange in the audio output,

but look at the ‘click’. For thistest I set the a.f. a.g.c. attacktime to 5ms, but thenrepeated the test with theattack time set to 30ms. As youcan see in Fig. 4 this resultedin rather more than a ‘click’ atthe transition, and is a goodillustration of why it is usuallynecessary to have as fast ana.g.c. attack as possible ons.s.b. speech or data signals.That deals with s.s.b. and c.w.signals, but what about a.m.?The audio output changeusing the standarddemodulator with only r.f.a.g.c. switched on is there tosee in Fig. 5. Note the clicknow at both start and finish ofthe level change, whereas withthe ‘pro’ demodulator and ther.f. and a.f. a.g.c. both on, Fig. 6 shows the much bettercontrol offered, although the‘click’ is still prominent.

What does this mean inpractice? Well, on signalsbelow about S9 (-73dBm) theaudio clicks are not much of aproblem, but on strongersignals the clicks are quiteaudible and make listening abit uneasy, particularly onstrong c.w. I realise that wec.w. listeners are a dying(literally) breed, but the clicksare noticeable on reasonablystrong a.m. broadcast stations- but the G-303i is still muchbetter than a Collins 95S-1, soyou pays your money andtakes your choice. Did Imention strong broadcaststations? What was the G-303ilike on my favourite900/909/918kHz tests?

No Monkey ChatterUsing either demodulator, thed.s.p. worked well, and I wasable to flick between the threetest frequencies using theup/down buttons after settingthe step size to 9kHz. Therewas a little sideband splatterfrom 909kHz but the twocontinental stations on eachside were clear and easy tohear, so no problems there.Going a bit further with the‘pro’ demodulator, I foundthat tuning to something likeFive Live and sliding thevariable bandwidth controluntil all the station’s sidebandswere within the filterpassband (and of course youcan see this all happening onthe spectrum display) youcould get really good audioout of the receiver, whilst on

Fig. 1: Thereceiver’s audioresponse during

a 250ms inputsignal step,with the r.f.

a.g.c. on andthe a.f. a.g.c. off

can be seen inthe plot.

Fig. 2:Switching thea.f. a.g.c. on to

add audiocontrol causes

a dramaticchange.

Fig. 3:Switching to

the professionaldemodulatorwith both r.f.and a.f a.g.c.switched on

shows just howmuch better the

‘pro’ softwareworks.

Fig. 4: For this test I

set the a.f. a.g.c.attack time to5ms, but thenrepeated thetest with the

attack time setto 30ms.

Fig. 5: The audio

output changeusing thestandard

demodulatorwith only r.f.

a.g.c. switchedon.

Fig. 6: The ‘pro’

demodulatorand the r.f. anda.f. a.g.c. both

on, shows muchbetter control,

although the‘click’ is still

prominent.Short Wave Magazine, February 2003

REGULAR NEWS FEATURE BROADCAST PROJECT SPECIAL COMPETITION QSL REVIEW BOOKS SUBS PROMO

24,25,28,29,30 Win Radio 8/1/03 1:29 pm Page 30

REGULAR NEWS FEATURE BROADCAST PROJECT SPECIAL COMPETITION QSL REVIEW BOOKS SUBS PROMO

weaker a.m. stations in theshort wave bands the slidingbandwidth enabled me to closedown until I got the best out ofwhat signal was available. Atruly excellent facility, but notso good when investigating the500Hz c.w. filter. Why not?

Listening to c.w. at thebottom end of 40 metres, Irealised that I could still hearsignals which were well outsidethe selected and displayed500Hz passband, so I did somemeasurements on the truebandwidth of the filter. Withthe standard demodulator,500Hz bandwidth, a.f. a.g.c.off, r.f. a.g.c. on, mode set toc.w., I applied an S9 signal atthe centre of the filterpassband to give an 800Hzaudio output. I then tuned thegenerator to each side of thefilter and measured thebandwidth between the pointsat which the audio output fellby 6dB. The true bandwidthwas 1.24kHz, not the 500Hzstated. I then switched the a.f.a.g.c. on and repeated the testto find that because of thelevelling effect of the a.f a.g.c.,the true bandwidth was now2.42kHz rather than the stated500Hz. I think that this shouldreceive some attention,because it makes the narrowc.w. filtering a waste of time.With the ‘pro’ demodulatorwhich gives total control overthe selected bandwidth, slidingthe bandwidth on c.w. from1kHz down to its narrowestsetting did absolutely nothingto the signals on each side ofthe displayed passband, whichremained clearly audiblethroughout. This being analarming discovery, I checkedthe other modes, but in all ofthese the filter passbands werecorrect, and attenuatedadjacent signals properly, so it’sa mystery. (John acted on MilanHudecek’s suggestions re: thesampling rate setting andcommented that the result wasa transformation of c.w. filteringfrom useless to superb with the -6dB points being 390Hz - Ed.)

ConclusionsI listened to all kinds ofmaterial during my all too brieftime with the G-303i and didnot find anything I couldn’thear. The performance in thel.f. beacon band around350kHz is worthy of mention,not only because of the factthat I could actually hear somany beacons, but that I couldsee them all spread out on thespectrum screen below thefrequency readout, and couldwatch the keying take place.Narrow band CB f.m. wasexcellent, as was s.s.b. and c.w.,with the caveat about the c.w.bandwidth. Listening to a.m.with the ‘pro’ demodulatorwas excellent, but I have to saythat the a.m.s. (synchronous)came out of lock quite easilyand did a little ‘squeak’ whenrestoring, but I found that withthe spot on accuracy of thel.s.b. and u.s.b. filtering it wasactually better to use thesemodes on weak and fadinga.m. signals.

The antenna connector onthe G-303i card is an SMA; abrilliant connector for highfrequency performance butvery small, so the G-303i comeswith an SMA to BNC adapter. Iwould recommend using ashort length of coaxial cablewith an SMA on one end and afree BNC socket on the other,because I have seen severalSMA/BNC solid adapters shearoff at the socket, and replacingthe SMA socket on the G-303iwill not be an easy job.WiNRADiO Communicationsplease note.

You may ask “What antennadid I use for my listeningtests?”, and you probablyguessed that it was an activeloop. To some extent thisexplains the terrific beaconband reception, and of coursethe loop was located some 20metres away from the dreadedPC, which could have helpedthe rejection of PC generatednoise. The G-303i is normallysupplied with a ‘get you going’

wire antenna, which wasmissing from the unit suppliedto me, so I made one accordingto the description in the usermanual. I can tell you thatthere was only a trace of noisefrom the computer monitor,which is a sign of the advancesmade by PC manufacturers, butthe signal levels were obviouslywell down, so I connected theG-303i to a 15 metre long wirevia 10 metres of coaxial cable.Again the noise level from thePC was very low, but the loop,as always, beats the livingdaylights out of lumps of wireantenna, so my advice is clear!

The user handbook is clearand easy to understand, andhas a detailed appendix on theuse and setting up of theprofessional demodulator sothat the user can investigatevarious software configurations- this is almost a hobby in itself.There is a WiNRADIO web sitewhich gives lots of additionalinformation, and becauseWiNRADiO Communicationsare so into informationtechnology the facilities existfor downloading additionalapplications for the G-303i;indeed just such an application(for a signal level monitor anddata store) was E-mailed to meduring the review, and I wasable to install it and have yetanother well though outfacility available to me.

I note that this is one of thelongest reviews I have everwritten, and I still haven’tcovered some of the aspects ofthis interesting new receiver.It’s difficult to place it in aparticular market positionbecause in order to use it, youneed to have a fairly decentcomputer system which couldcost you anything from £500 to£1000. If, as many of you do,have an existing PC of theminimum specification thenthe G-303i is definitely for youbecause the suggested sellingprice is around £300 for thestandard unit (see, I said Iwould surprise you) andaround £350 for the unit fitted

with the professionaldemodulator. I consider thatthis is the one to go forbecause of the hugelyextended capability of the ‘pro’demodulator. Starting with anempty desk and a full wallet,the G-303i and PC could bemore expensive than manyother stand-alone receivers,but let me put it this way; if Ihad to choose between aCollins 95S-1 and the G-303i(ignoring the obvious fact thatthe 95S-1 tunes to 2GHz), Iwould take the G-303i. That’s astatement I thought I wouldnever make.

And, sorry folks, you willhave to take it as read that thememory facilities are brilliant,but there are no more pages totell you about it.

SWM

In the UK the WiNRADiO G-303i is available from Falcon Equipment and Systems.Tel: (01684) 295807. E-mail: winradio@sda-falcon.co.uk For more details on theWiNRADiO products, news of new developments, distributors in your home territoryand software downloads visit: www.winradio.comThe G-303i costs £299 fitted with the standard demodulator. The additionalProfessional Demodulator costs £99 as an upgrade. A G-303i and professionaldemodulator bundle costs £350. All prices are plus VAT at 17.5%, P&P is £15.

Manufacturer’sResponseWe sent a copy of John’s findingsto WiNRADiO’s Milan Hudecek, hiscomments are as follows:

Excellent review, we are indeedthrilled. There are only a fewminor things we believe maydeserve some clarification.

Regarding the c.w. i.f.bandwidth issue, the selectivityof the filter depends on thefilter lengths, and this influenceis more pronounced thenarrower the bandwidth gets. Iwould suggest to adjust the c.w.filters to the maximum availablelengths (255), which providesthe sharpest filters. Admittedly,this feature is easy to overlookamongst all those sliders andedit boxes!

The filter length setting iscompromise between the filterselectivity and the CPUresources available. The IBMNetVista we supplied has plentyof excess power, so all filterlengths can be comfortably setto the max. For the StandardDemodulator, we have usedcompromise filter lengths toallow the software to run evenon a very slow, beginners’ PCs.

There is in fact yet one moretuning method - as if therewere not enough of them - thewheel on a wheeled mouse! :-)Easy to overlook, but somecustomers love this feature inour receivers.

Many thanks to John for havingnoticed the lack of 9kHz stepsize preset, and the anomalywith keyboard tuning. Bothhave been fixed and the newsoftware is now available fordownload.Short Wave Magazine, February 2003

24,25,28,29,30 Win Radio 8/1/03 2:32 pm Page 31