The CCDandreadout electronics for the OMC instrument on Integral

illuminatedversionof < 6e 2 rmsattheOMCreadoutrateof 328Kpixs 2 1 .TheOMCrowshifttimeof2.3 µ sgivesaframe transfertimeof < 2.4ms,sothatintegrationsofafewseconds willhaveframeshiftsmearof < 0.1%.TheCCDhasadump gateanddrainrunningalongsidethereadoutregister:thisfa- cilitates ushingtheCCDanddumpingunwantedareasina windowedimagingmode. Thetelescopeisdesignedtooperateinthephotometric V - bandforcomparisonwithstandardastronomicalcatalogues,a V -band lterintheopticalpathde nesthepassband( < 500to 600nmFWHM).TheCCDisback-illuminatedtomaximise QE,withastandardanti-re ectioncoatingtomatchthe lter passband.Figure2showsQEasafunctionofwavelengthfora CCDwiththiscoating. Fig.2.<br><br> CCDQEasafunctionofwavelength. Twotrade-o stobemadeinselectingdevicetypeareas follows. 1.Invertedmode.AnAIMO(AdvancedInvertedMode Operation)devicehasmuchlowerdarkcurrentandhence easescoolingrequirements.However,thepenaltyisalower full-wellcapacity(andhencedynamicrange)andgreater darksignalnon-uniformity(DSNU).Itwasdecidedtouse anon-AIMO(orNIMO)device.<br><br> 2.Anti-blooming.Thetrade-o hereisthatanti-blooming preventsthechargefrombrightstarsfromspreadingup anddownimagecolumns,saturatingtheseareas.However, anti-bloomingalsoreducesfull-wellcapacity.Giventhe relativeinfrequencyofbrightstarsandthesmallareaof theimagea ected,itwasdecidedtouseaCCDwithout anti-blooming. TheCCDispassivelycooledto < 2 80Cfortworeasons 3 Toreducedarkcurrenttoasu cientlylowleveltoallow longexposures.Darkcurrentasafunctionoftemperature forNIMOdevicesisgivenby I d = CT 3 exp[ 2 6400 / T ](1) where I d isthedarkcurrent C isanormalisationconstant T isthetemperature(Kelvin). At 2 80C,thepredicteddarkcurrentis < 4 × 10 2 2 e 2 pix 2 1 s 2 1 .Inprinciplethisallowsexposures upto < 10 6 s.<br><br> 3 Toreducethee ectsofradiationdamagetoacceptable levels. Atthistemperaturethenormalceramicdual-in-lineICpackage forthisdeviceisunsuitableasthefocalplane atnessrequire- mentwouldnotbemet.Thereforethesiliconismountedona custominvarmountingdisc,andconnectionismadeusinga exi-trackateachendofthechipwirebondedtothebondpads onthesilicon,asshowninFig.3,whichshowsanengineering modeloftheCCD. D.M.Waltonetal.:TheCCDandreadoutelectronicsfortheOMCinstrumenton Integral L277 Fig.3.<br><br> PhotographofCCDwith exi-tracksoninvarmount. Theotherendofeach eximateswithconnectorsonthe FPAelectronicsPCB.The exiprovidesanelectricalcon- nectionwithreasonablylowthermalconductivitybetweenthe CCDat < 2 80CandtheFPAeat < 0C.TheFPAehasathermo- staticheatertopreventitstemperaturedroppingbelow 2 40C inordernottoviolatethetemperaturerangeofsomeofthe components.Anotherfactorinthedesignofthefocalplane iscontamination.TheCCDat 2 80Cwillactasasurfacefor molecularcontaminationtocondenseon.TheFPAhasbeen designedtoallowheatingoftheCCDto < 20Ctore-evaporate thesecontaminants.ThustheCCD,mountingand exihadto bedesignedusinglowoutgassingmaterialsandbeabletowith- standtemperaturecyclingoverthisrange. 2.2.Biassuppliesandclocksequencer Biassuppliesarederivedfroma36Vsupplygeneratedinthe PCE(PowerConditioningElectronics,suppliedbyINTA)and aregeneratedusingstableanaloguecontrolcircuitry.Twoof thesupplies, V SS (theCCDsubstratebiassupply)and V RD (the outputstageresetdrain),canbecontrolledinordertoallow forin-orbit at-bandshiftscausedbyionisingradiation(see below).ThisisachievedbytheDPEloadingregistersinthe ROEusingtheLSL.EachoftheseregisterscontrolsaDAC, theoutputofwhichcontrolsthesupply.<br><br> TheclocksequencerconsistsofacardofHCMOSlogic circuitrywhichgeneratesalltheCCDclockingwaveformsand CDStimingwaveforms.Statemachinesgeneratetheindivid- ualtripletsforthethree-phaseclocking,whilecountersenable anddisablethestatemachinesinanappropriatesequence.The countersareloadedwithdefaultvaluesfromtheROEPROM atboot-up,butcanbechangedbytheDPE,againloadingROE registersusingtheLSL.Forexample,thedefaultintegration timeis10s,buttheDPEcancontrolthisintherange25ms to819.2s.Inprinciple,evenlongerintegrationtimescanbe achievedbytheDPEloadingregistersandsettingandmonitor- ing agsappropriately.Otherparameterswhichcanbeadjusted includewindowsizeandpositionandbinningfactor.Lesser usedparameterscanalsobechanged,suchastheframetrans- fercount.Thismakesthereadouthighly exible:forexample aTimeDelayandIntegration(TDI)modecouldbee ected; andforotherapplicationsoftheROE,di erentCCDtypescan beaccommodated.Themasterclockforthesequenceristhe 5MHzHSLclock.Thisgivesapixelperiodof3.05 µ sanda readouttimeforafullimageof < 3.5s. 2.3.Analoguesignalchain,ADCandDPEdata interface ThesignalfromeachoutputnodeoftheCCDisampli edin theFPAe rstbyalownoiseFET,thenalownoiseoperational ampli er(op-amp),thenalinedriverop-amp.Thiscombina- tionampli estheCCDoutputby × 16,raisingthe < 0.5Vmax- imumsignaltoasu cientlyhighleveltoallowtransmission fromtheFPAtotheROE.IntheROEthesignalisprocessed bytheCDS,isampli edbyanotherfactor1.25,andisthen digitised.The12-bitADChasabuilt-insample-and-holdam- pli er,afullrangeof10Vandaconversiontimeof800ns. TheROEhastwoother exibilitiesbuilt-in.<br><br> 3 Thefull-wellcapacityof < 120Ke 2 andthereadoutnoise of < 8e 2 rms(includinganalogueelectronicscontribution) meansthatthedynamicrangeoftheCCDisundersampled bytheADC,whichhasaconversionof < 30e 2 / DN.Toen- ableoperationwithresolutiondowntothereadoutnoise, the rststageoftheROEanalogueelectronicshasase- lectablegainof × 1(default)or × 6,selectablebyaregis- terbitloadedfromtheDPEusingtheLSL.The × 6range changestheconversionto < 5e 2 / DN,withanADCfull scaleof < 20Ke 2 . 3 Thereisalsoamodeinwhichtheampli er ltertimecon- stantsareincreasedbyafactor5andthereadoutslowed downcorrespondinglytofurtherreducereadoutnoise. TheADCoutputgoestoan8192wordFIFO,whichfeedsthe PISOandHSLcircuitry.TheFIFOeasestimingrequirements onthehandshakingbetweentheROEandDPE.<br><br> 3.Testing Testingonthegroundhasbeenasfollows. 3.1.CCDtestingbye2v Testingcarriedoutbye2vrelatedtoCCDmanufacturingpro- cesstestsandensuringthattheCCDsmeettherequirements forthemission.Particulartestsincluded: 3 Manufacturing:temperaturecycling,thermalshock,vibra- tion,geometricalmeasurements. 3 Performance:QE,darkcurrent,ampli erresponsivity 3 Quali cation:acceleratedlife,radiation.<br><br> 3.1.1.Radiationtesting RadiationtoleranceofCCDsiscriticalforspaceapplications. Testingconsistedofirradiatingtwodevicesfromthe ight batchwithCo 60 gammaraysatBrunelUniversity 9sCentrefor RadiationDamageStudies.Co 60 gammarayshaveameanen- ergyof < 1.25MeVandcausepredominantlyionisingdamage, L278D.M.Waltonetal.:TheCCDandreadoutelectronicsfortheOMCinstrumenton Integral althoughtheenergyissu cientlyhightoproducesomebulk damage(seee.g.Janesick2001).Thedevicesweretestedbe- foreirradiationandafter10,20and30krad(Si),corresponding toseveralyearsinorbit.Thedoseratewas < 9krad(Si) / hour, sothateachradiationincrementtook < 1hour. Ionisingradiationproduceselectron-hole(e-h)pairsinthe CCD.Forthosegeneratedintheinsulatinglayers,theelectrons aremobileandcandi useawayfromthegenerationsite,while theholesareleftbehind,buildingupanetpositivecharge, causing atbandvoltageshifts.Ifthedeviceisbiassed,there isanelectric eldacrossmanyoftheinsulatinglayerswhich separatesthee-hpairs,whereasthepairsaremorelikelytore- combineifthedeviceisunbiassed.Forthisreason, atband shiftsare < fourtimesgreaterifthedeviceisbiassedduring irradiation,comparedwithanunbiasseddevice.Forthesetests thedeviceswerebiassed,inordertoassesstheworstcase at- bandshifts.<br><br> Bulkdamageconsistsoftrapsinthebandstructureofthe siliconbetweenthevalenceandconductionbandsandcharge generationsites,createdbydisplacementofatomsinthelat- tice.Thetrapshavecharacteristictimescalesforcapturingand releasingelectrons,whichcanleadtochargetransfere ciency (CTE)degradation.Chargegenerationsitescanleadto chot pixels d. Principalresultswereasfollows. 3 Flatbandshifts:shiftsof125mVperkrad(Si)wereob- served.Thisisinaccordancewithotherresultsone2vde- vices(Robbins2000).<br><br> 3 Darkcurrent:foratypicalsubstratevoltage,thedark currentincreasedfrom < 0.01e 2 pix 2 1 s 2 1 at 2 85Cto < 0.04e 2 pix 2 1 s 2 1 after30krad(Si). 3 CTE:CTEwasmeasuredusingFe 55 X-rayeventsat < 2 100C.ParallelCTEfellfrom < 0.999995to < 0.99996. ReadoutregisterCTEwasessentiallyuna ected.<br><br> 3 QEandPixelResponseNon-Uniformity(PRNU):these werealsouna ected. 3 Readoutnoise:thistoowasuna ected. Theseresultsagreewellwithvaluesreportedintheliterature (e.g.seeRobbins2000).Inorbit,bulkdamagewillbecaused bysolarprotonsandcouldleadtoalargerlossofCTEper krad(Si).Theinstrumentstructureprovidesshieldingtoreduce thedose.<br><br> 3.2.TestingatMSSL TestingatMSSLconcentratedonthefollowingfactors. 3 CCDs.Mappingofdarkcurrent,brightanddarkdefects. 3 FPAe,ROE.Functionaltests,workmanshipinspections etc.,interfacingtoMSSLGSE(forROE / FPAe / CCDcon- trolanddataacquisition),interfacingtoESA-suppliedDPE simulator(inconjunctionwithUCD / DIAS) 3 CCDswithFPAe,ROE.Veri cationofcorrectimaging performance,responsivityandnoiseofCCD / FPAe / ROE combination.Forexample,Fig.4showstheresultsfrom alighttransfercurvetestusedtomeasuretheresponsivity andnoise.Thistestconsistsofmeasuringthesignaland Fig.4.<br><br> Lighttransfercurveusedtomeasureresponsivityandreadout noise. noiseoverarangeofsignalsfromzeroupto < halffull- well.ThetopleftplotshowstheleftchannelADCoutput (DN)asafunctionofintegrationtime,whiletheplotbelow hasthesamehorizontalscaleandplotstheresidualsfrom astraightline t.Thetoprightplotshowsnoisesquared (DN 2 )againstsignal(DN):noisesquaredisthesumofthe squaresofreadoutnoiseandsignalshotnoise,sothatthis plotgivesastraightlinetheslopeofwhichisthesystem conversionfactor,DN / e 2 ,andtheinterceptatzerosignal givesthesquareofreadoutnoise(foradetailedderivation seee.g.Janesick2001).Theplotbelowagainshowsthe residuals.Thebottomfourplotsfollowthesamesequence fortherightchannel. 3.3.InstrumentandS/Cleveltests ThreefullmodelsweredeliveredtoINTA: 3 EngineeringModel(EM).<br><br> 3 Quali cationModel(QM),laterrefurbishedasFlightSpare (FS). 3 FlightModel(FM). D.M.Waltonetal.:TheCCDandreadoutelectronicsfortheOMCinstrumenton Integral L279 Fig.5.<br><br> 512 × 512portionofOMC 9s rst-lightimage,showinggamma TrianguliAustralis. Atthislevel,testsincludedquali cationandacceptancetests suchasinstrumentvibration,andperformancetestingsuchas alignment,focusingoftheFPAwiththetelescopeandcalibra- tion,resultsofwhicharereportedinMas-Hesseetal.(2003). 3.4.In-orbitresults OMCwasthe rstinstrumenton Integral tobecommissioned.<br><br> IthasbeenoperatingsuccessfullysinceOctober2002and hasobserved > 20000targets.Figure5showsonequarter (512 × 512pixels)ofthe rstlightimageinfalsecolour. Anapproximatelylogarithmicintensityfunctionhasbeenused sothatthebrightnessindicatesthestellarmagnitudesrather than uxestogiveanindicationofthedensityofstarsvisible. Inthis10sintegration,abrightstarnearthecentresaturatesthe CCDandcausesbloomingoverseveralpixels.Thisisgamma TrianguliAustralis,withvisualmagnitude2.87.<br><br> OMC 9sstrengthisinacquiringlightcurvesthanksto Integral 9shighorbit,sothatlonguninterruptedobservations arepossible.ThusOMCisparticularlysuitedtoobserva- tionsoftime-variablesourcessuchasnovae,eclipsingbina- riesandAGN.MoredetailedresultsarereportedinMas-Hesse etal.(2003),e.g.lightcurvesofeclipsingbinaries. Acknowledgements. FundingforMSSL 9sdevelopmentofthereadout electronicsandGSEwasprovidedbytheUK 9sParticlePhysicsand AstronomyResearchCouncil(PPARC).Fundingforthebuildingof theEM,QM / FSandFMelectronicsandaportionoftheCCDpur- chasewasprovidedbyUCD-DIASthroughProdex,withtheremain- deroftheCCDpurchasecostprovidedbyINTA.Wewouldalsolike toacknowledgetheadvicereceivedfrome2vtechnologiesandmany helpfuldiscussions.Theauthorsthankthetechnicalandadministra- tivesupportsta atMSSLfortheirhardworkthroughouttheproject.<br><br> References Janesick,J.R.2001,Scienti cCharge-CoupledDevices,SPIE monograph;PM83 Mas-Hesse,J.M.,G´1menez,A.,Culhane,J.L.,etal.2003,A&A,411, L261 Mazy,E.,De se,J.M.,Plesseria,J.Y.,etal.2003,A&A,411,L269 Robbins,M.2000,TheRadiationDamagePerformanceofe2v technologiesCCDs,e2vTechnicalNoteS&C906 / 424

less