JohnSwallow’spingercircuit–aforayinto1940sand50selectronicarchaeology

JohnGould(wjg@noc.ac.uk)andGwynGriffiths(gwyn@autonomousanalytics.com)

1. BackgroundOverrecentyearsJGmadeanumberofvisitstothehistoricship,Medusa,nowbasedatGosport.HMSMedusawasaharbourdefencemotorlaunch(HDML),awoodenhulledvesselbuiltandlaunchedinPoole,Dorsetin1943.http://www.hmsmedusa.org.uk/.HMSMedusaendedherworkinglifeasoneoftheAdmiraltyHydrographicDepartment’sinshoresurveyvessels(1953-1968)commandedatvarioustimesbyRogerMorriswhobecameHydrographeroftheNavyandbyDesmondScottwhobecameExecutiveSecretaryoftheIntergovernmentalOceanographicCommissionofUNESCO.MyinterestwasarousedbyMedusa’sroleinD-Day.TheinvasionfleethadtobeguidedthroughgapsthathadbeensweptintheminefieldlaiddownthecentreoftheEnglishChannel.ThesegapsweremarkedbeforeD-Daybyacousticbeaconsanchoredtotheseabed(OperationEnthrone)andMedusa’srolewastohomeinononeofthebeaconsandsitoveritasavisiblemarkertoguidetheinvasionfleet–abitlikethingsthatweoceanographersdo,butundermuchmoredangerousconditions.Sothatsetmethinking:Whatkindofabeaconwasit?And,Didthe10kHzpingersthatweused(startingwithJohnSwallow’sfloatsin1954/5andTonyLaughton’sdeep-seacameras)oweanythingintheirdesigntotheWWIIbeacon?TheMedusawebsitedescribesthebeaconasanFH830andaGooglesearchrevealedthatitwasdesignedbyaCanadian,GeorgeWhalley(1915-1983),amemberoftheRoyalCanadianNavyVolunteerReservewho,postwar,wasbestknownasabroadcasterandpoet.(http://archives.algomau.ca/gwp/node/62).MyfirstenquirywaswithNigelGodsellwhomaintainsawebsitesimilartoOceansWormleybutdevotedtotheworkoftheAdmiraltyResearchLaboratoryinthehopethathemayhaveknownofalinkbetweentheWWIIpingersandtheNIOoneswhenGroupWwasatTeddington.Hedidnotknowofonebutwehadaninterestingdiscussionaboutnickelscrolltransducers.IsenthimaphotoofthetransduceronJohnSwallow’sfloatandhesentmeaphotoofonethatwasmuchlarger.(Figure1)

Figure1.LeftJohnSwallowdeployingoneofhisfloatsintheIrmingerSeain1962(CourtesyWHOIarchives)RightA6’diameternickelscrolltransducerfromARLbeingbroughtaboardRRSDiscoveryII.(CourtesyNigelGodsell).

2

SowhatofFH830?IlaterdiscoveredthatthehandbookanddesigndetailsareheldintheNationalArchivesandobtainedscansandsimilarcopiesfromAlanWatsonattheMedusaTrust.

Figure2SchematicoftheFH830acousticbeaconandthecoverofalaterAdmiraltyhandbook(CourtesyofNationalArchives,Kew)WhatwasimmediatelyclearwasthattherewasnoreallinkbetweenFH380andtheNIOdesign.FH830usedaquartzcrystaltransducerworkingat20kHzandlinkedtoaconicalprojectortodirecttheacousticsignalupwards.AtvarioustimesIinvolvedGwyninmyFH830discussionsandthissparkedhisinterest.Hewondered;CouldhebuildreplicasoftheFH830andSwallowcircuits?andHowwouldtheyperform?Hereishispartofthestory2. Dr.JohnSwallow's1955acousticpingerJohnSwallow'sseminal1955paper1onhisneutralbuoyancyfloatcontainsthecircuitdiagramofitsacousticpinger.ThecircuitwasdesignedbyRickHubbard,atechnicalassistantwithinGroupWatTeddington(Figure3).HubbarddidnotmovedowntoWormley2.ThecircuitisreproducedhereasaninsetinFigure4whichalsoshowsaphotographofthecomponentlayoutandtheacoustictransducer.Thereisanelegantsimplicityinitsdesign,summarisedinasinglesentenceinSwallow’spaper,"Thetransmitterconsistsofanickelscrollresonantat10kc/s,woundtoroidallyandenergizedbydischargingacapacitorthroughaflashtube".

Figure 3 Admiralty Research Laboratory Group W 1952. Rick Hubbard, back row left alongside D.W. (Dick) Privett, later NIO Librarian, In front are Henry Charnock (L) and George Deacon. NIO/IOS Directors. (Courtesy NOL archives)

1 Swallow, J.C., 1955. A neutral-buoyancy float for measuring deep currents. Deep Sea Research, 3: 74-81. 2 Our thanks to Brian McCartney for this information, 16 October 2018.

3

Thetypeofflashtubeinthiscircuithadahistorygoingbacktothe1920swiththeinventionofthegas-filledthyratronasahighcurrenttriggeredswitch.Thedevicewasmadesmaller,morerobust,andlessexpensiveduetotheworkofDr.Harold"Doc"EdgertonatMITintheearly1930s.Hisfascinatingdiariesarealsoavailableonline;forinstanceon6May1931hedescribesamains-poweredcircuitforaflashtubestroboscopeusingacircuit3nottoodissimilartothatofHubbard.Histubebecameknownasthestrobotron.Edgertonalsomadeseminalcontributionstothedevelopmentofsonar,andcofoundedthecompanyEdgerton,Germeshausen,andGrierthatbecameEG&Gin1947.EG&G'sMarineInstrumentssubsidiarydevelopedawidesuiteofoceanographicequipmentincludingsidescansonars,acousticreleasesandvectoraveragingcurrentmetersthatwereusedatWormley.Let'slookatthepingercircuitindetail,drawingonJohnSwallow'snotesfromhisdiary4andillustratedwithoscilloscopemeasurementsmadeonareplicaconstructedin2018,

Figure4.Thesixelectroniccomponentsandthenickelscrolltransducerthatmadeupthe1955acousticpingerusedinJohnSwallow'sneutral-buoyancyfloattogetherwiththecircuitdiagram.Source:NationalOceanographicLibrary4152(Circuit),4160(Components).

Figure5.ThecircuitofFigure4,withoutthetransducer,asconstructedin2018usingthesametypeofflashtubeandsimilarcomponentstotheoriginal.Whileperiodresistorshavebeenusedthecapacitorsaremodern,as1950scapacitorsofthistypewillhaveahighleakagecurrent.2.1HowthepingerworksThe360Vbatterychargesthe8µFreservoircapacitorviaa10kWcurrent-limitingresistorinabout0.2s,themaximumcurrentdrawnfromthebatterybeingabout36mA.ResistorRandtimingcapacitorCdeterminethepingrepetitionrate.TheNSP1flashtubeisacold-cathode"Neostron"withfourelectrodes5.Itpassesnocurrentuntilthevoltageatthetriggerelectrode(connectedtoRC)risestoatubeandage-dependentvalueofbetween80and130V.Attheinstantoftriggeringanarcformsfromthepositiveanodeatthetopofthetubetothecathodenearthebase.Thepeakcurrentcanbe200A.Inpracticethecurrentislimitedbytheresistanceofthewiring,theinternalresistanceofthereservoircapacitorandtheinductanceofthetransducer.Typically,thecurrentpulsemaylastfor10µs.

3 See https://edgerton-digital-collections.org/notebooks/03 on pages 88 to 90 4 Available at https://viewer.soton.ac.uk/nol/fullscreen/2417/1/ 5 Details and data sheet available at: http://www.r-type.org/exhib/aaa0473.htm

4

Thecurrentpulse,flowingthroughthecoilwoundthroughthenickelring,inducesamagneticfieldthatcausestheringtochangeitscircumference-nickelbeingamagnetostrictivematerial.Asimpleanalogyisthatoftappingabelloraglass-theimpulsegivesrisetoa"ringing"attheresonantfrequencyofthephysicalitem-beitabell,aglassorinthiscasetheringofnickel.Followingtheimpulse,thedurationoftheringingisdeterminedbytheamountofdamping,beitwithinthenickelring,fromthewirewoundoverthenickelorfromthesurroundingwater.Thetransmissionfrequencyof10kHzandthepulsedurationarethereforedeterminedbydimensionsandphysicalpropertiesandnotbytheelectronics.2.2JohnSwallow'sSeptember1955testsontheelectronicsSwallow'sdiarydescribesaseriesoftestsonanumberofpingerspriortoadeployment.Hisnotesshowacompleteunderstandingofthecircuitandlikelysourcesofproblems,notsurprisinggivenhistraininginelectronicswhileattheUniversityofCambridgein1941–43,followedbytrainingintherepairandmaintenanceofradioequipmentattheAdmiraltySignalsEstablishment,LytheHill,Haslemerein1943–44andhistourofdutyattachedtotheRoyalNavyinCeylonin1944–466.Heexaminedthe:

• Optimumnumberofturnsonthetransducer:Throughexperimenttheoptimumnumberofturnsonthetransducerwasfoundtobe30frommeasuringthesignalreceivedonasimilartransduceroverashortpathinwater,havingnoticedthathisearly,peculiar,resultswereaffectedbyairbubblestrappedonthetransducerface.• Lifetimeoftheflashtube:Swallowcheckedthatthe~144,000flashesoverafourdaydeploymentatoneevery2-3secondswasaverysmallfraction(0.27%)oftheexpectedlifeoftheflashtube,givenas300hoursat50pulsespersecondinthedatasheet.However,theprojectedlifewasnotbeingachievedinpractice.Ofsixpingershetestedon3September1955threehadproblemswiththeirflashtubes:nottriggering,givingacontinuousfaintglowandtakinglongerthancalculatedtotrigger,andnotstarting"exceptwhentapped".Whilethecircuitaspublished(Figure4)showedtheflashtubeasthecommercialFerrantiNSP1,Swallowactuallyusedthemilitaryequivalent,theCV2207.Itisthisdatasheetthathastheexpectedlifetime.However,itdoesnotcontainawarning(presentinthe1957NSP1datasheet5)onthepermanentdamagethatcouldoccurifthetubewasrunwithtoolowadischargecurrent(of5Aorless).Lowcurrentsresultinacontinuousglowratherthanaflash,leadingtodamagefromexcessiveheatdissipation.Possiblysomeofhistubeshadbeenrunwithtoolowacurrentatsometimeintheirlife.• '0V'connectionofthetimingcapacitor:By5Septemberanothertubehadstoppedflashing.Swallowdeterminedthatonecausecouldbea"voltagesurge"causedbyhavingthetimingcapacitorCconnectedtothe'0V'endofthetransducercoil,asinthecircuitdiagram.Hemodifiedthecircuitonallthepingersbymovingthetimingcapacitor'0V'connectiontothecathodeoftheflashtube.Allwereworkingfineintothefollowingdayandtherearenofurthernotesonproblemswiththeflashtubes.• Batterylife:The360Vsupplywasbuiltfrom24SiemensS123hearingaidbatteries,eachbatteryweighing13/8ounces8,foratotalbatteryweightofabout1.07kg.From26Octoberto5November1955Swallowhadtwopingersontesttodeterminethelifeoftheirbatteries.Oneunithadan8µFreservoircapacitorandbeganat29pulsesperminute,theotherhada4µFcapacitorandbeganat19pulsesperminute.Figure6showshisresultsforUnitNo.12withthe8µFcapacitorstartingwithfresh"BatteryNo.2".Thegraphalsoshowsourcalculationforthecumulativeenergyextracted,comprisingtheenergystoredinthecapacitorsandthatdissipatedintheseriesresistorsduringcharging;the85.5kJisequivalent

6 From typed notes of a conversation between John Swallow and Margaret Deacon 30 November 1994. 7 Data sheet available at: http://www.r-type.org/pdfs/cv220.pdf 8 https://viewer.soton.ac.uk/nol/fullscreen/2417/20/

5

to135mAh.Swallowusedthesefigurestoestimatethatthebatterieswouldgiveapingerwitha4µFcapacitorstartingat20pulsesperminutealifetimeofabout10days.

•

Figure6.VoltageandpulsesperminuteduringabatterylifetimetestbyJohnSwallowduringOctober-November1955togetherwithourcalculationofcumulativeenergyextractedandexpectedpulserate.

• Variationinpulserate:Oneofthelimitationswiththissimplecircuitisthattheinevitabledecayofthebatteryvoltagehasasubstantialeffectonthepulserate.Assumingafixedtriggervoltagevt,andabatteryvoltageofvbthetimebetweenpulsesisgivenby: -ln(((vb-vt)/vb)^CR)withvtat110VthecalculatedchangeinpulserateinFigure6closelymatchesthatobservedbySwallow.Laterpingersusedcrystal-controlledpulsetimingcircuitsthatenabledvisualpulse-to-pulsecorrelationonwet-paperrecorderssuchastheMufax,impossiblewiththissimplecircuit.

2.3Costofthe1955pingerandtoday'sequivalentThetotalcostofcomponentsforasimplepingerfloatin1955wasabout£17.JohnSwallow'sdiarygavethecostsofindividualparts,asshowninthistable.We'veaddedthecostsin2017usingafigureforinflationtogetherwithtoday'sactualpriceswhereavailableforeachitem.Theflashtubeisstillavailableas"NewOldStock"fromspecialistsuppliers,andisarealbargaincomparedwiththe1955cost.Theotherelectroniccomponentsareeitherasimilarpriceorcheaperinrealtermswhilethealuminiumtubesaremoreexpensiveandthebatteriesespeciallyso.Forthebatterieswe'velistedexactequivalentsthatareavailabletodayfromsuppliersthatmakereplicas.Amodern,cheapersolutionwouldbetousestandardalkalinecellsina12-voltpackanduseasolid-stateDC-to-DCstepupconvertertogivethe360V.

2.4WaveformsasseenonthemodernreplicaThesethreescreencapturesfromanoscilloscopeforthemodernreplicaillustratetheworkingoftheSwallowpingerasdescribedinthetextabove.Thefirsttwocoveratimespanof2.4seconds.

6

Figure7showstherapiddischargeofthe4µFreservoircapacitorfollowedbyrechargingto300Vinabout100ms.Figure8isthevoltageattheflashtube'striggerelectrode.Thesharpdecreasetozerohappensastheflashfires,thevoltagethenrisesascapacitorCchargesthroughtimingresistorRuntilitreachesthetriggervoltageatwhichthecyclerepeats.Heretheintervalis660ms,shorterthanSwallowwouldhaveused.Figure9representsthecurrentthroughtheflashtube;thetracecovers24µsandshowsacomplexwaveformwiththemaindischargehappening3µsaftertheinitialtrigger.Itisthislargecurrentpulse,withitsveryfastleadingedge,thatexcitesthemagnetostrictivetransduceratitsownresonantfrequency.

Figure7.Voltagewaveformatthereservoircapacitor.200msand100Vperdivision.

Figure8Voltage waveform at the trigger electrode. 200ms and 50V per division.

Figure9.Currentwaveformattheflashtubeoutput.2µsperdivision,currentscalenotcalibrated..

7

3. GeorgeWhalley'sWWIIFH830acousticpingerTheacousticpingerwithintheFH830sub-surfaceMarkerBuoydesignedbyGeorgeWhalleyattheAdmiraltyintheearlyyearsofWorldWarIIwasdifferentineveryrespectfromJohnSwallow's1955pinger.

Figure 10. FH380 Mk 1 acoustic beacon (L) and it associated transmitter unit. (From B=National Archives Courtesy of the Medusa Trust) Thefollowingtablesetsouthoweachcircuitfulfilledtheessentialfunctionsofanacousticpingerandthecomponenttechnologies.MostoftheinformationinthistableontheFH830pingerwasobtainedfromdocumentADM277/28intheUKNationalArchives9.

Function Whalley,WWIIFH830 HubbardandSwallow,1955Transmissionfrequency

Vacuumtube(valve)inductor-capacitortunedcircuit.Nominalfrequency20kHz.

Mechanicalresonanceofnickelstripwoundasatightscrolltoformaring.Nominalfrequency10kHz.

Pinginterval Chargingofcapacitorsviaaresistordrivenbythetransmissionsignalwithswitch-selectedperiods."Slightlyaffected"bythebatteryvoltage.

Chargingofacapacitorviaaresistorfromthehighvoltagebattery.Intervalincreasesmarkedlyasbatteryvoltagedecays.

Pingduration Setbycapacitors,valvegridcurrentandthecut-offvoltageandmutualconductanceoftheoscillatorvalve,typically100ms.

Setbythemechanicaldampingoftheoscillationsofthenickelscroll,typicallyafewmilliseconds.

Acoustictransducer

Piezoelectric,quartzcrystal.Highimpedance,about1Mohminparallelwith295pF,voltagedriven,about1400Vpeak.

Magnetostrictive,nickelscroll.Lowimpedanceofafewohms,currentdriven,about50Apeak.

Activedevices Twodirectlyheatedvacuumtubes(valves)requiring4vfilamentbatteries,oneastheoscillator,andtheotherastheoutputamplifier.

Singlecoldcathodeflashtuberequiringnofilamentsupply.

Circuitcomplexity Twenty-twocomponents,twospeciallywoundautotransformersintheMk1pinger,excludingtimercircuitry.

Fivecomponents,nonespeciallymade.

9 "ASDIC beacon buoy. Includes 6 photographs depicting: Asdic beacon buoys: FH 830, ...", catalogue entry at http://discovery.nationalarchives.gov.uk/details/r/C527364

8

Maximumdepth 25fathoms(about46metres) Testedto4500metres.Endurance Upto85days About10days.Batteries Lowvoltage:MK1FourVarleyV80

2Vleadacidaccumulatorsinseriesparallel,weightabout24kgtotal10.MkIIandIIIusedsixSiemensNsizedrycellsinseriesparallel.Highvoltage:ThreeEverReadyWinner120Vdrybatteriesinseries,weight9.9kgtotal.Totalbatteryweightabout34kg.

Highvoltage:24SiemensS123hearingaidbatteries.Totalweightabout1.07kg.

Totalweightofbuoy/floatinair.

278lb(126kg) About10kg.

3.1Howthepingerworks-oscillatorThepingercircuitinFigure11hasbeentranscribedfromthatinFigure4ofthe"ExtractfromtheF.H.830Mk.IOperatingandMaintenanceInstructionBook"intheNationalArchives9.TheresonantfrequencyoftappedcoilL2withfixedcapacitorC10andvariablecapacitorC2setsthetransmissionfrequency.C2providesavariationofabout200Hzaround20kHz.OscillationismaintainedbytheamplificationprovidedbyV1withthefeedbackprovidedbycouplingcapacitorC8.

Figure11DiagramofthepingerpartoftheFH830buoycircuit,redrawnfromtheoriginal.InaconventionaloscillatorthelowerconnectionofbiasresistorsR4andR3wouldbedirectlytothecommonnegativeofthefilamentandhighvoltagesupply(whichwewillcallground).However,heretheconnectionisviatheresistorchainR1andR5-R8,inparallelwithcapacitorsC5andC6.Thisarrangementsetsboththetransmissionpulsedurationandthepulseinterval.WhenthecircuitisswitchedonthelowerconnectionsofR3andR4areatgroundvoltage,asthecircuitoscillatesthegridandfilamentoftheeachvalveactasdiodes,rectifyingpartoftheACwaveform,andanegativeDCvoltagebuildsupatthegrid.ThepotentialdifferenceacrossR4drivesacurrent10 No details were given, and none could be found, for the Varley accumulators. To give an indication of weight and capacity dimensions from the drawings were scaled by the buoy outside diameter of 16", to give a size of about 5"x5"x7.25". The closest 2V accumulator capacity and weight, typical of period, were then obtained from http://www.valve-radio.co.uk/literature/oldham-accumulators-and-batteries/ High voltage battery details from https://www.radiomuseum.org/r/ever_winner_120.html

9

throughit,chargingthecapacitorsC5andC6,withasmallfractionthroughthehighvalueresistorchainR1,R5-8.WithatimeconstantdeterminedbytheproductofR3inparallelwithR4,andC5inparallelwithC6,thevoltageattheV1gridbecomesmorenegativeuntilitreachesthecut-offvoltageandthetransmissionpulseends.Cut-offisthenegativegridvoltageatwhichnocurrentflowsthroughthevalve.ThedischargepathforC5andC6isthroughtheresistorchainR1,R5-R8,whichsetsthepulseinterval.Oscillationsdonotimmediatelyrestartasthecapacitorsdischargeasthegainofthevalvewillbeverylowclosetothecut-offpointandagreatergainisneededtostartoscillationscomparedwithmaintainingoscillationoncestarted.TheamountofgainneededismainlydeterminedbytheturnsratiobetweentapsL2BandL2DandL2BandL2A,L2BbeingatACgroundduetoC3;thehandbookdoesnotincludethisdetail.WithR3=51kWandR4=100kWandC5=2µFandC6=1µFthetimeconstantCR,andhencetheapproximatepulseduration,is102ms.TheoperatorcansetthepulseintervalviaaplugconnectiontotheendsofR5-8,whichgavetimeconstantsof1.35,2.88,5.88and13.8seconds.However,asV1beginstooscillateafterabout0.34ofatimeconstanttheapproximatepulseintervalswere0.5,1,2and4seconds.3.2Howthepingerworks-amplifierTappingL1CprovidesanimpedancematchfromtheoscillatortunedcircuittothegridoftheamplifierV2viacouplingcapacitorC7.TheamplifiedsignalatV2anodeistappedintothetunedautotransformerL1atL1B,withthehighimpedancequartztransducerconnectedattapL2CviacouplingcapacitorC9.VariablecapacitorC1isinparallelwiththequartztransducerandservestotuneL1toparallelresonance.Somewhatconfusinglythecircuitdiagramandthetextrefertothequartztransducerasthe"oscillator",which,atleastthesedays,suggeststhatthequartztransducersetstheoscillationfrequency.Thatisnotso,L2andthecapacitorsinparallelsetthefrequency.BoththeamplifierandoscillatorvalvesaretypeP410,directlyheatedaudiooutputtriodes11thatwereintroducedin1929.WhiletheDCvoltageattheanodeofV1isreducedbytheproductofR2andthecurrentflowingthroughit,thefullbatteryvoltage,initially360V,ispresentattheanodeofV2.Thisfarexceedsthe"absolutemaximum"of150VintheP410'sdatasheetandshowstherobustnessofthese1920striodes.Overall,thispingerhasaclevermethodofprovidingapulsedtransmissionwithauser-settablepulseintervalinacompactdesignthatusesstandardcomponents,anditwouldhavebeencommonpracticetowindinductorsforspecificrequirements.

3.3WaveformsasseenonamodernfunctionalreplicaAfunctionalreplicaoftheFH830circuitinFigure11hasbeenconstructedwithcomponentsofvaryingvintage,Figure9.Whilecapacitorsfromthe1940sareavailabletheywouldnotbeappropriateforuseinthiscircuitduetoinevitableleakagecurrentsexceeding1µAthatwouldaffectthetimingcircuitsandthevalvebiasing.Resistorsfromthe1940shavebeenused.P410valvesarenowveryscarce,andwhenavailabletheycostabout£100each.Valveswith2Vfilamentsweremorecommonandarelessexpensive.ThisreplicausestwoPM2audiooutputtriodesdatingfromthelate1920s.TheFH830handbookdoesnotprovidedetailsofthetwoautotransformersotherthantheinductanceofthecompletewindings,15.5mHfortheoscillatorand160mHfortheoutput.Here,thecoilshavebeenwoundonferritecoresthatwereintroducedin194912,withthetappingpointscalculatedtomatchthecircuitimpedancesand,fortheoscillator,toprovidesufficientfeedbackgiventheamplificationavailablefromthePM2valve.

11 See http://www.r-type.org/exhib/aaa1263.htm 12 The 1949 example in Figure 2 at https://ieeexplore.ieee.org/document/4490128

10

Figure12Thefunctionalreplica,constructedtotheoriginalcircuitdiagrambutwithcomponentsofvaryingvintage.

Figure13Uppertrace:Outputvoltageacrossa440k�dummyloadshowingapulsedurationofabout60msandapulseintervalofabout460ms(Y-axis:200Vperdivision,X-axis:50msperdivision).Lowertrace:VoltageatthejunctionofR4andC5;themeanlevelis-21V(Y-axis:1Vperdivision).TheuppertraceinFigure13showsapulseoutputof800Vpeaktopeakacrossa440k�dummyloadatasupplyvoltageof120V.Giventhevalvesare90yearsoldtheyhavenotbeenpushedto360VasintheFH830.Thepulselengthat60msisshorterthanthe100msinthehandbook.However,thefollowingTableshowsthatpulselengthisdependentonsupplyvoltage,andcouldconceivablyreach100msat360V.Thepulseintervalisalsoshorterthaninthehandbook,herethesettingwasfor1secondbuttheintervalwasabout455msat120V.Thereislessofavariationforthepulseintervalwiththeappliedvoltage,farlessthanfortheSwallowcircuit,ausefulfeature.However,pulselengthandintervaldodependonindividualvalvecharacteristics,swappingtheoscillatorandamplifiervalveshalvedbothtimes.ThelowertraceinFigure13showsthevoltageatthejunctionofR4andC5wherethemeanlevelwas-21V.HerethegridisdrivenmorenegativeduringthepulseasC5andC6chargeupuntilthevalvecutsoffandtransmissionstops.Thispointthenbecomeslessnegative,andwhenthehysteresistorestartoscillationisovercome,andthegainincreases,thecircuitbeginstooscillate.TheslowriseandgradualdecaythatarecharacteristicofthiscircuitareshownmoreclearlyintheexpandedviewinFigure14.

11

Figure14:Expandedtimetracefortheoutputpulseshowingtheslowinitialriseandthegradualdecay(X-axis:10msperdivision).

Highvoltagesupply(V)

Pulseinterval(ms)

Pulselength(ms)

Peakvoltage(V)

45 365 40 14460 410 45 20090 435 50 304120 455 60 408150 430 70 464175 430 75 520

4. PostscriptandatributeThisisasfaraswehavebeenabletoprogress.Themajorstumblingblockhasbeenthatwehavebeenunabletolocatetransducersofthetypesusein1944andfrom1955onwardstotesttheacousticoutputofthesepingers.ThoseofuswhoworkedwithSwallow’sfloatsinthe1950sand60scanvouchforthechallengeoflocatinghisfloatsevenwhenworkingundergoodacousticconditionsandfromaquietresearchship.ItincreasesouradmirationfortheanonymousASDICoperatoraboardHMSMedusaonD-Daywhoensuredthatthebeaconwaslocatedandthatthevesselremainedonstation!Hemusthavebeenoneofthemostimportantunsungheroesoftheday.AcknowledgementWearegratefultotheNationalOceanographicLibraryforpermissiontoquotefromJohnSwallow's1955diaryandtoreproducethetwophotographsNOL4152andNOL4160.