Single Photon Counting Camera - Micro Photon Devices · SPC3 is a single photon counting camera...

Micro Photon Devices S.r.l. – Italy All Rights Reserved

M I C R O P H O T O N D E V I C E S

SPC3

Single Photon Counting Camera

© Micro Photon Devices S.r.l. Via Stradivari, 4

39100 Bolzano (BZ), Italy email [email protected]

Phone +39 0471 051212 • Fax +39 0471 501524

SPC3 User Manual Version 1.1.0 - November 2015

SPC3:SinglePhotonCountingCamera64x32

v1.1.0 MicroPhotonDevicesS.r.l.–ItalyAllRightsReserved Page2

TableofContents INTRODUCTION 3

SINGLEPHOTONAVALANCHEDIODE 4

Furtherreadings 6

SPC3HARDWARECHARACTERISTICS 7

Electricalconnections 7

Cameramechanicaldimensions 10

SPC3OPERATION 11

Standardoperation 11

Gatedoperation 12

FLIMOperation 14FurtherReadings 18

Acquisitionparameters 18Integrationtime 18Dead-timetimeanddead-timecorrection 20Backgroundsubtraction 23

Additionalinformation 23CalculationoftheSnapsizeandtheContinuousAcquisitionDatarate 23Cameraauto-protection 24

SPC3SOFTWAREINTERFACE–VISUALSPC3(WINDOWSONLY) 25

Installation 25

SoftwareInterface 25Mainwindow:ControlPanel 26MainWindow:PlotsettingsandPlaybackControl 28MainWindow:ControlButtons 30MainWindow:AcquisitionParametersSummary 31CounterWindow 31FLIMWindow 32

SPC3-SDK,FILEFORMATSANDIMAGEJPLUGIN 33

SYSTEMREQUIREMENTS 37

COPYRIGHTANDDISCLAIMER 37



IntroductionSPC3 isasinglephotoncountingcamerabasedona2-D imagingarrayof64x32smartpixels.Eachpixel

comprisesasingle-photonavalanchediode(SPAD)detector,ananaloguefront-endandadigitalprocessing

electronics. This on-chip integrated device provides single-photon sensitivity, high electronic noise

immunity,and fast readoutspeed.The imagercanbeoperatedat full resolutionwithamaximumframe

rate of about 100.000 frames per secondwith negligible blind time (dead-time). Inside each pixel three

independentcountersare integrated.Eachoneof themcanbe independentlygatedthroughanexternal

signal; counter 1 can also be gated by an internally generated signal (e.g. for FLIM mode acquisition).

Counters2and3, instead, canalsocountdownward,with theirdirectioncontrolled throughanexternal

signal.

SPC3pixels featurehighphoton-detectionefficiency (PDE) in thevisibleandnear-UVspectral region,and

lowdark-countingrates,evenat roomtemperature.The imager iseasily integrated intocommonoptical

setups thanks to a Thorlabs® SM1 thread, a C-mount mechanical adapter and a high-speed USB 3.0

computerinterface.ThecameraisshowninFigure1.

ThecameradiffersfromconventionalCharge-CoupledDevices(CCD)orCMOSsensorsbecauseitperforms

a“fullydigital”acquisitionofthelightsignal.Eachpixeleffectivelycountsthenumberofphotonswhichare

detectedbythesensorduringtheacquisitiontime.

Figure1. SPC3:SinglephotonCountingCamera.



SinglePhotonAvalancheDiodeTheSPADisap–njunctionwhichisreverse-biasedwellaboveitsbreakdownvoltage.Underthisoperating

condition, theabsorptionofasinglephotongeneratesanelectron-holepair,which isacceleratedby the

highelectricfieldacrossthejunction.Theenergyofthechargecarriersiseventuallysufficienttotriggera

macroscopic avalanche current of few milliamperes through the device. The SPAD biasing electronics,

namely the Active Quenching Circuit (AQC), is integrated on the same silicon chip. The AQC senses the

avalanche current, quenches it and resets the diode to its initial state. The avalanche is quenched by

decreasing thevoltageacross theSPAD junctionbelowbreakdownforanadjustable time (hold-off), that

rangesfromfewtenstofewhundredsofnanoseconds.TheSPADbiasvoltageisthenrestoredtotheinitial

state,andthusthediodeisreadytodetectthenextphoton.Thetotaltime,whichisrequiredtorestorethe

initial state of the SPADafter the detectionof a photon, is nameddead-time.During thedead-time, no

photonscanbedetected.TheAQCprovides,additionally,avoltagepulsetoanassociatedcounter,whichis

integratedineachpixel.

ThemeasurementoflightsignalsbyaSPADhasseveraladvantagesconcerningthesignaltonoiseratio.No

analoguemeasurementofvoltageorcurrent isneeded,sincethedetectoracts likeadigital“Geiger-like”

counter. It follows thatnoelectronicnoise isaddedbyanalogue todigital convertersoramplifierswhile

measuringthesignal.Additionally,thedetectorislesssensitivetotheelectromagneticinterferenceorthe

electricalnoisegeneratedbyexternalequipment,differentlyfromchargecoupleddevices.

TheprimarynoisesourcesforSPADsare:

1) Dark counts. This noise source comprises all processes which can start an avalanche across the

SPAD junctionbutwhicharenotcausedbythedetectionofaphoton.Thetypicalsourceofdark

countsisthethermalcarriergenerationprocess.Darkcountsarethedominantnoisesourceforthe

SPADsoftheSPC3.

2) Afterpulsing. The detection of a photon can trigger, with a certain probability, an additional

detection eventwithin fewmicroseconds. This spurious event is caused by charge carriers, that

flew through the junction during the avalanche, remained trapped and are then subsequently

released. These carriers are accelerated by the high electric field across the junction andmight

trigger another avalanche like the photo-generated electron-hole pairs, when released after the

dead-time. The afterpulsing probability depends on the detector dead-time and it is usually

reduced to about 1-2% at the normal operating conditions. Longer dead-times decrease the

afterpulsingprobability.



3) Cross-talk.Duringtheavalancheprocess,photonsareemittedbyhotcarriers.Thesecanpropagate

through the silicon and trigger a detection event in SPADs which are placed in close proximity.

Practically,theCross-talkprobability isvery lowamongthepixelsoftheSPC3(about10-5),dueto

thelargedistancebetweentheSPADs.Itscontributiontothetotalnoiseismostlynegligible.

Since the read-out noise of CCD or CMOS cameras is normally slightly higher than one electron (per

integrationtime)andsincethedarkcountingratesoftheSPADpixelsoftheSPC3isaboutonehundredper

second,itfollowsthattheoptimaloperatingconditionforthecameraisatfastormoderateframe-rates.

Particularly, the dark-counts contribution is low with exposure times of about 0.1s. By reducing even

furthertheintegrationtime,thedarkcountnoisebecomesnegligible.

Due to thedead time, theSPADhasa linearworking range, i.e thenumberofdetectioneventswithina

defined time period (T) depends linearly on the illumination intensity, only at low ormoderate photon

fluxes.At largephotonfluxes thenumberofdetectedphotonsdeviates from linearityandsaturates toa

constantvalue,whichisthereciprocalofthedead-time(Tdead).Asamatteroffact,sincethediodeisofffor

aperiodequaltoTdeadeverytimeitdetectsaphoton,themaximumcountratehappenswhenthedetector

oscillateswithaperiodTdead.SupposingNimpthenumberofimpingingphotonsonaSPADactiveareaduring

anintegrationtimeT,Ncountedthephotonscountedbythedetector,andtakingintoaccountthePDE,Ncounted

isequaltoNimpreducedbythePDEandmultipliedbytheratiooftheactualintegrationtimedividedbythe

setintegrationtimeT:

𝑁!"#$%&' = 𝑁!"# ∗ 𝑃𝐷𝐸 ∗𝑇 − 𝑁!"#$%&' ∗ 𝑇!"#!

𝑇= 𝑁!"# ∗ 𝑃𝐷𝐸 ∗ (1 − 𝑁!"#$%&'

𝑇!"#!𝑇

)

𝑁!"#$%&' =𝑃𝐷𝐸 ∗ 𝑁!"#

1 + 𝑃𝐷𝐸 ∗ 𝑁!"# ∗𝑇!"#!𝑇

𝑒𝑞. 1

ByexpressingNimpasafunctionofNcountedoneobtainsalso:

𝑁!"# =1

𝑃𝐷𝐸∗

𝑁!"#$%&'

1 − 𝑁!"#$%&' ∗𝑇!"#!𝑇

𝑒𝑞. 2

Thetotalnumberofdetectedeventsisthussmallerthanthetotalnumberofphotonswhichcrossthep-n

junction,becauseofthesaturationeffectandthePDE.When𝑁!"# ∗ PDE =!

!!"#!,thenumberofcounted

eventsisalreadyreducedof50%.SinceaSPADisblindforTdeadaftereachtriggering,itisalsoclearthatina

timeperiodT,themaximumnumberofphotonsthatcanbecountedis:

𝑁!"#$%&',!"# =𝑇

𝑇!"#! 𝑒𝑞. 3



Figure2. Numberofmeasuredphotonsinanimageof1msexposureanddead-timeof100nsasafunctionofthenumberofincomingphotons(PDE=100%forsimplicity).

Let’ssupposeforexampleanexposuretimeT=1ms,aPDE=50%forsimplicityandadead-timeTdead=100

ns. Inthiscase,themaximumnumberofphotonsthatcanbecounted(ormoreingeneralthemaximum

numberofSPADtriggerings)is10’000.Inourexamplealso𝑁!"# ∗ PDE =!

!!"#!correspondsto10’000;this

means thatwhenaphoton fluxof 20’000’000photons/shit thedetector, during the integration timeT,

only20’000(Nimp)photonscrosstheactiveareabutonly10’000canbedetectedduethePDEof50%,and

finallyonly5’000eventsare,onaverage,counted(Ncounted)becauseofsaturation.See,forexample,Figure

2thatshowsthenumberofcountedphotonsasafunctionofthenumberoftheimpingingphotonsforan

exposuretimeT=1ms,adead-timeof100nsandaPDEof100%forsimplicity.

Furtherreadings

M.Ghioni,A.Gulinatti,I.Rech,F.Zappa,S.Cova,“ProgressinSiliconSingle-PhotonAvalancheDiodes”,IEEEJournalofSelectedTopicsinQuantumElectronics,2007,13,852-862.

D.Bronzi,F.Villa,S.Tisa,A.Tosi,F.Zappa,D.Durini,S.Weyers,W.Brockherde,"100000Frames/s64×32Single-PhotonDetectorArrayfor2-DImagingand3-DRanging”,IEEEJournalofSelectedTopicsinQuantumElectronics,2014,vol.20,no.6,pp.354-363.

M.Vitali,D.Bronzi,A.J.Krmpot,S.N.Nikolic,F.J.Schmitt,C.Junghans,S.Tisa,T.Friedrich,V.Vukojevic,L.Terenius, F. Zappa, R. Rigler, "A Single-Photon Avalanche Camera for Fluorescence Lifetime ImagingMicroscopyandCorrelationSpectroscopy", IEEE JournalofSelectedTopics inQuantumElectronics,2014,vol.20,no.6,pp.344-353.



SPC3hardwarecharacteristics

Electricalconnections

TheSPC3haspowerandUSBconnectionsononeside,andamulti-coaxialconnectorontheothersidefor

variouscontrolsignals,asshowninFigure1andFigure3.SPC3 isprovidedwithacableadapterfromthe

multi-coaxialconnectortostandardSMAconnectorsasshowninFigure4.Adescriptionofthe inputand

theoutputscanbefoundinTable1.TheElectricalspecificationsoftheinputandoutputsarereportedin

Table2.

Table1. SPC3InputsandOutputsdescription.

USB High-speedUSB3.0connector.ItisrecommendedtouseshortUSBcables(<2m).

+12VDC Jackforconnectingtheprovided+12Vpowersupply.

SYNCOUT Coaxialoutputwhichcandrive50Ohmterminatedtransmission lines.A50OhmDC load

termination is required. Electric pulses of 3.3V LVCMOS are generated at this output for

synchronizingthecamerawithanyexternaldevice.TheSPC3canbeconfiguredbytheuser

to output the internal 50MHz reference clock, synchronised with the internal 100MHz

master clock, for synchronizing any external instrument (such a laser) with the internal

generated gate. The same output can be configured also to generate a signal (pulse)

synchronous with the start of each hardware integration frame. The delay between the

actualstartoftheintegrationframeandtherisingedgeofthispulseisshowninFigure5.

For more details on gated operation, see the corresponding paragraph in the SPC3

Operationsection.

TRIGIN Thiscoaxialinputrequiresa3.3Vpulse(5Vtolerated)ofminimumdurationof40nstostart

thecameraacquisitionusingasignalfromanexternaldevice(e.g.ashutter,alaser,etc.).

Thefirstframestartsbetween20nsand40nsaftertheTRIGINrisingedge.SincetheTRIG

IN signal is asynchronous respect to the internal clock, such delay is of course not-

deterministic (unless it is synchedwith the SYNCOUT signal). The input impedance is 50

Ohm.SeeFigure5forreference.

GATEIN1

GATEIN2

GATEIN3

Active-low inputswith50OhmDC input impedance.Each inputcontrolsoneofthethree

in-pixel counters. When GATE IN is ‘1’, the incoming photons are not counted by the

measuring electronics. 3.3V LVCMOS signals are acceptedbut the inputs are5V tolerant.

Theminimumdurationof aGATE INpulse is 5 ns. The actual internal gate is delayedby

about15ns.

DIRECTION

Inputtoselectthecountingdirectionofcounters2and3.Theinputhasa50OhmDCinput

impedance; 3.3V LVCMOS signals are accepted but the input is 5V tolerant. A low value

meansUPcounting.DIRECTIONtimingsTBD.



Figure3. Cameraside-views:hardwareconnections.

Figure4. CamerawiththecableadapterandtheSMAoutputdescription.

Figure5. TimingofTRIGINandSYNCOUTsignalswithrespecttotheactualstartoftheacquisition.



Table2. ElectricalCharacteristicsoftheSPC3measuredatroomtemperature(25°C)

Symbol Parameter Signal Min Typ Max Unit

VIH Inputlogic-highvoltage(5Vtolerant) TRIG IN, GATE,DIRECTION 2.4 3.3 V

VIL Inputlogic-lowvoltage TRIG IN, GATE,DIRECTION 0 0.8 V

VOH Outputlogic-highvoltage(50Ωload) SYNCOUT 2.7 V

VOL Outputlogic-lowvoltage(50Ωload) SYNCOUT 0 V

tPI PropagationdelayfromTRIGINtoacquisitionstart TRIGIN 20 40 ns

tIN TRIGINpulseduration TRIGIN 40 ns

tPO PropagationdelayfromacquisitionstarttoSYNCOUT SYNCOUT 5 10 ns

tOUT SYNCOUTpulseduration(framesyncmode) SYNCOUT 50 ns

tir Inputrisetime(10%-90%) TRIGIN,GATE 5 ns

tif Inputfalltime(10%-90%) TRIGIN,GATE 5 ns

tor Outputrisetime SYNCOUT 1 ns

tof Outputfalltime SYNCOUT 1 ns

tgate GATEpulseduration GATE 5 ns

tgate_delay GATEdelay GATE 10 15 20 ns

tdir_delayDIRECTIONdelayXXX DIRECTION XX XX ns

- Internalreferenceclock-frequency SYNCOUT,internal 50 MHz

- Internalreferenceclock-duty SYNCOUT,internal 50 %



Cameramechanicaldimensions

Figure6. SPC3mechanicaldimensions.Alldimensionsinmm.



SPC3OperationThe SPC3 camera features three differentmodes of operation that can be optimised through the use of

severalworkingparameters,allcontrollablefromacomputerusingeitherthedeliveredSDKlibraryorthe

providedWindows®PCsoftware.Thefollowingparagraphsdescribeindetailallthesefeatures.

Standardoperation

Thecamera isoperated in itsstandardmodewheneachpixel is simplyused to countphotons inauser

defined integration time and the counters are not gated (see the next paragraph for details).When the

SPC3isoperatedinthismode,threedifferentacquisitionmodesarepossible:

• Live acquisition: single images from the camera are acquired and downloaded to the computer.

SinceinthismodeitisthePCthatcontinuouslyasksfornewimages,thetimebetweentwoimages

cannotbepreciseandstronglydependsonhowthePCsoftwarebehavesandonthePC’scurrent

workload.Thismodeshouldnotbeusedforactualscientificacquisition,butitisinsteadusefulfor

acquiringalowframerate“live”movie(i.e.real-time),forinstancewhenaligningthecamerainthe

optical setup or adjusting the acquisition parameters with the current illumination conditions.

Considering the practical limitations of the USB connectionwhen transferring small data blocks,

typical achievable frame-rates are in the range of 50-100fps (actual values depend on the

performance of both the PC hardware and the software employed). As a consequence, the PC

requestsofliveimagesshouldbelimitedtoamaximumof1every10msorlongertime.

• Snap: a sequenceof images isacquiredand stored into the internalmemoryof thecamera.The

dead period between subsequent frames (inter-frame dead-time) is less than 10 ns, and the

frames’timingisaccuratelydeterminedbytheinternalclock.Oncetherequirednumberofframes

ismeasured,theelectronic interfacetransfersthedatatothecomputer.The imagescanthenbe

displayedandsavedonthehard-drive.Thismodecannotbeusedforliveacquisition,sincealldata

will be downloaded to the PC only at the end of the entire snap, which can last few seconds.

Instead,thisapproachallowsdataacquisitionsatthemaximumframe-ratepossible,since itdoes

notdependontheactualthroughputoftheUSBlinkorPCsavespeed.Snapsizeislimitedbythe

capacityoftheinternalmemory(128MiB).Formoredetailsonhowtocalculatesnapsize,seethe

paragraphonCameraConfiguration. IfyouneedtoacquiremoredatayoushoulduseContinuous

Acquisitionmode,asdescribedbelow.

• ContinuousAcquisition: a continuousacquisitionofdatabetweenaSTARTcommandandaSTOP

commandfromthePCisperformed.Oncetheacquisitionisstarted,theinternalmemoryisusedas



a buffer, and data should be continuously read by a PC. In case the internal memory gets full

becausethePCfailstodownloadthedataswiftlyenough,datalossoccursandanerroris issued.

Themaximumachievable frame rate stronglydependsonPCperformance (speciallyon theHDD

writespeed).ThankstoUSB3.0itispossibletoacquirethefullarrayatthemaximumspeed(about

100kfps).However,inordertosaveallgenerateddataatthishighthroughputitisalsocrucialthat

theemployedmass-storageunitisveryfast(seetheparagraphonCameraConfigurationformore

detailsonhowtocalculateactualdatarates).SSDstorageunitsarethusrecommended.IfyourPC

cannotsustainsuchahighthroughputyouwillhavetoreducetheframerateortosaveonlyhalf

the array. If this is not possible you must resort to the Snap mode, at the expense of the

experimentduration.

In both Snap mode and Continuous Acquisition mode it is possible to trigger the actual start of the

acquisitionwithanexternalsignal.

Inallthreemodesitispossibletooutputapulsesynchronouswiththestartofeachintegrationtime.Itis

alsopossibletointernallysubtractabackgroundimagepreviouslystoredintothecamera.

Gatedoperation

Normally,thecountersofeachpixeloftheSPC3willjustcontinuouslycountallthepulsesgeneratedbythe

SPADs.Thisoperatingmodeiscalledfree-runninganditisthedefaultoperationmode.Anyway,eachofthe

three integrated binary counters which register the detected number of photons in each pixel can be

enabledordisabledbyagatesignal.Thisoperatingmodeiscalledtime-gating.

Thegatesignalforcounters2and3isprovidedonlybythetwocoaxial inputsGATEIN2andGATEIN3.

Thegatesignalforcounter1insteadisthelogicalORbetweentwoindependentdigitalsignals:GATEIN1,

i.e.oneofthecameracoaxialinputs,andSoftwaregate,generatedinternallybytheSPC3electronics,which

canbesetbytheuser throughtheprovidedsoftware.Eachof thethreeexternalGATE INsignalscanbe

periodicoraperiodicanditwillbeappliedasynchronouslyanddirectly(throughtheORgate)totheSPADs.

TheSoftwareGateisaperiodicdigitalsignal,whichissynchronizedwiththeinternalReferenceclock(20ns

period).TheusercansettheDelay, i.e.thetimeshiftbetweentherisingedgeoftheReferenceclockand

thehigh-to-lowtransitionofSoftwaregate,and theparameterWidth,whichdefines thedurationof the

“counter-enabled”Softwaregatepulse.Essentially,theSoftwaregategeneratesaperiodicgatesignalofa

givenWidthandDelayfromthereferenceclock.NotethattheWidthisthedurationoftheSoftwareGate’s

lowstateasshowninFigure7.

TheWidthoftheSoftwareGate,whichisgeneratedbythecontrolelectronics,canbevariedintherange

0÷100%of thereferenceclock, instepsof1%.Actually,duringa factorycalibration, theGatewidthhas



been internally discretized with 1000 points over the full range. In this way, whenever one of the 100

nominalselectablevalues isrequestedbytheuser,theSPC3cameraissettotheclosedpossibleone, i.e.

the one with the smallest possible error. Note that, however, values below 2ns could be unreliable.

ParticularlytheminimumachievableGatesignalisabout1.5ns.Theactualsetvaluecanbereportedtothe

userby the software.TheDelay canbevaried in the range -40%÷+40%of theSoftwareGateperiod, in

nominalstepsof0.1%,correspondingto20ps.Theactualgate-delaystepcanactuallyvarydependingon

theusedcamera,howeveraninternalcalibrationensuresthattheintegralnonlinearityindelaysettingis

always less than+/- 20ps.An initial offsetmayalsobepresent. Thisoffset isonpurposenot calibrated,

since it will sum up with other setup-depended offsets (e.g. from different cable lengths). The overall

systemoffsetcanbeevaluatedthroughopticalmeasurementsemployingashort-pulselaser(i.e.lessthan

100pspulsewidth).

WhentheSPC3isoperatedingatedmode,thesamethreeacquisitionmodesofthestandardoperationare

possible,thistimeinconjunctionwiththeuseofthethreegates.

Figure7. Hardwaregatingexample:Delay=5nsandWidth=5ns(bothof25%oftheReferenceClockperiod).ThereferenceclockhasbeensenttotheSYNCOUToutputofthecamera.Additionally,aGATEINsignalwas provided by the user. Only the photons, which are detected when the “Gate signal” is on, arecounted.Examplevalidforcounter1;forcounter2and3theexampleremainsthesamewithoutthesoftwaregate.Insuchcaseonlythe4thphotonwouldnotbetdetected.



FLIMOperation

Fluorescence Lifetime IMaging (FLIM) is an imaging technique based on the measurement of the

exponentialdecayrateofthefluorescencefromafluorescentsample,ratherthanofitsemissionintensity.

Thefluorescencelifetimeofamoleculeisameasurementoftherateofdecayoftheemission,whichisa

propertyoftheindividualsinglemolecule,andthusit isunaffectedbychangesinprobeconcentrationor

excitationintensity.FLIMcanbeusedforinstanceasanimagingtechniqueinconfocalmicroscopyandtwo-

photonexcitationmicroscopy.

AtypicalapplicationofFLIMisthestudyofFörsterResonanceEnergyTransfer (FRET), themechanismof

energy transfer between two chromophore molecules with the emission band of one overlapping the

absorptionbandoftheother.Whenthistwocomponentsareincloseproximity,thefirstone,calleddonor,

initiallyinitselectronicexcitedstatemaynon-radiatively(withoutemissionofthelight)transfertheenergy

to the second one, called the acceptor. The result is the quenching of the donor fluorescence, thus the

decreaseofthedonoremission lifetime.Theefficiencyoftheenergytransfer is inverselyproportionalto

the sixthpowerofdistancebetweendonorandacceptor,making theeffectnoticeableonlyatdistances

shorterthan10nm.FRETisusedtodeterminetheproximityinthenano-meterrangebetweenproteinsor

other elements of interest associatedwith suitable fluorophores labelled as donors and acceptors. Such

measurements are used as a research tool in fields such as biology and chemistry. Different ways of

performing FRET measurement exist. The obvious problem with the intensity-based steady-state FRET

measurementsisthattheemissionbandofthedonorextendsintotheemissionbandoftheacceptorand

the absorption band of the acceptor extends into the absorption band of the donor. Furthermore, the

concentrations of the donor and acceptor and the fraction of donor molecules linked to an acceptor

molecule are variable and unknown. All this makes the intensity-based FRET measurements requiring

calibrationwiththesamplescontainingonlydonorsandacceptors,ormeasurementsinwhichthesecond

step is the destruction of acceptor by photo-bleaching, obtaining the FRETmeasurement as the relative

increaseofdonor fluorescence intensity.TheFLIM-basedFRETmeasurementhas theadvantage that the

resultsareobtainedfromasinglelifetimemeasurementofthedonorandarenotaffectedbythechanges

of the sample concentration, excitation intensity and other factors that limit the intensity-based FRET

measurements.

FLIM imagesaremostoftenobtainedbyemploying theTimeCorrelatedSinglePhotonCounting (TCSPC)

techniqueandapointdetector(suchasaPMT),which isthenscanned inordertoreconstructthedecay

histogramforeachimagepixel.Theexponentialdecaymodel isfittedtoeachpixelhistograminorderto

determine the lifetime. The colour of each image pixel represents the determined lifetime, giving the

possibility to obtain images with contrast between materials with different decay rates even if they



fluoresceatthesamewavelength,or,ontheotherhand,identifytworegionofthesamematerialevenif

theyhavedifferentintensity.TheuseofanarrayofSPADshasmajoradvantagesforimagingapplications

since, as opposed to single pixel systems, it does not require any scan of the sample. A second major

advantageofthepresentedpixelstructureconcernstheshortdead-time,whichhasalowerlimitofabout

50ns,thusallowingthecountingofveryhighphotonfluxes.

TheembeddedshortgatecapabilityenablestheuseofSPC3asaFLIMcameraeventhoughtheemployed

SPAD imager chip is only counting the detected photons and it is not time-tagging them. In fact, it is

possibletoemploythegateinordertoimplementatime-gatedFLIMdetectionsystem.Intime-gatedFLIM

thedecayofthefluorescenceisreconstructedbyrepetitivelycountingthenumberofdetectedphotonsin

short timewindows thatareprogressively shiftedwith respect to theexcitation laserpulse,as shown in

Figure8.Inordertoperformthistypeofmeasurement,theSYNC_OUToutputformtheSPC3camerahasto

beconnectedtothetriggerinputofapulsedlasersourceabletogenerateopticalpulseswithwidthofat

mostfewhundredsofpicoseconds.Suchlaseristhenusedtoexcitethefluorescenceinthesampleunder

test.TheinternallygeneratedGatesignalactivatesthecountersafterthegenerationofthelaserpulsefor

thetimedefinedbygatewidth.Accordingly,thefluorescencedecaykineticsismeasuredbychanginggate

shift(position)overtime.Bothgateshiftandgatewidthhaveoptimalvaluesdependingonthelifetimeof

theexcitedstateofthefluorescentmoleculesandontheimagingframe-rate.

Inorder tospeed-upFLIMacquisitionsand increase the framerate, theSPC3includesanautomaticFLIM

modeinwhichtheinternalgatesignalisautomaticallygeneratedwithuser-definedstep-widthandshifted

oftheuser-desirednumberofsteps.

Figure8. OpticalwaveformreconstructionusingtheGated-FLIMapproachwiththeSPC3camera.



Each"FLIMacquisition" is thuscomposedbyasequenceofFLIMelementary (step) frames,one foreach

desiredgateposition(step),eachoneconsistingofanacquisitionwiththesameexposureparametersofa

standardacquisition.FLIMmodeacquisitionsarepossibleinSnapmodeandContinuousAcquisitionmode

whereasarealLivemodeinnotavailable,sinceeachFLIMacquisitioniscomposedbyseveralelementary

frames.However, theVisualSPC3 software (seebelow)offersa seamlesslyequivalent Livemodewhich is

internallyimplementedperformingrepetitiveSnapacquisitions.

Duetothetimerequiredfor internalgateshifting(which isalsodependentontheshiftparameters),the

timerequiredforafullFLIMacquisitionisnotsimplythesumoftheintegrationtimeofeachstepanditis,

instead,convenientlycalculatedbythecamerasoftwareandreportedtotheuser.Areferenceclockinside

thecameraguaranteesthateveryFLIMacquisitionstartsonlyafterthistimeiselapsed,thusensuringthe

isochronicityofFLIMacquisitions,andallowingtherecordingof“FLIMmovies”.

Exactly as for the gated operation, the gate signal is generated synchronized with the internal 50MHz

referenceclock.Itswidthcanbevariedintherange0÷100%ofthereferenceclockperiod(20ns),insteps

of1%,whilethegateshiftcanbevariedintherange-40%÷+40%ofthesameperiod,innominalstepsof

0.1%,correspondingto20ps.Ofcourse,inordertomaketheGated-FLIMwork,theexternallaserhastobe

synchronisedwiththegateandthus, incaseofFLIM,theSYNCOUToutputstheinternalreferenceclock.

Finally,asshowninFigure8,sincethereferenceclockperiodis20ns,thenominalmaximum“observation”

timewindowisalso20ns.Actually,sincethegateshiftrangeis-40%÷+40%,thegatecanbemovedfrom2

nsto18nsandthustheactualmaximumobservationwindowis16ns.Thereasonforthereducedrangeis

duetothefactthatforshiftslongerthan+/-40%ofperiod,theshiftaccuracyisnotsatisfactoryandthus

theSPC3FPGAdoesnotallowtheirselection.

Theactualwidthandphaseofthegatesignalobtainedforagivensetofparametersmaychangeamong

differentSPC3units,duetofabricationtolerancesoftheFPGA.That’swhythegatewidthiscalibratedby

MPDbeforedeliveringthecamera.

TheinitialpositionofthegateinaFLIMsequenceisalso internallycalibratedandhasanaccuracybetter

than +/- 20ps plus a constant offset (see also paragraph on Gated Operation). This shift-offset is not

calibratedsinceincaseofaFLIMexperimentalset-upwhatisreallyimportanttomeasureisthefulloffset

of thegatesignalwithrespecttothe laserpulse.Thisoffset isnotonlydeterminedbythe internal shift-

offsetdue to the camera, but also (andprobably largely) by the delays introducedby the used external

interconnectioncablesand the laser itself.Acalibrationof theactualgate shiftwith respect to the laser

excitation is therefore necessaryafter the camera ismounted into themeasurement setup. This can be

easilydonebydirectlyshiningthelaserontothearray,withoutinterposinganyfluorescentsampleorfilter,

andperformingafullmeasurementovertheentire20nsshiftrange.Byinspectingthemeasurementand



finding the laserpeak, it ispossible toevaluate theshiftbetween theactual laserpulseand the internal

reference clock. Actually it is not even important to measure it. What needs to be done is simply the

adjustmentofthecables’lengthfromtheSYNC_OUTofthecameratotheTRIG_INofthelaserinorderto

placethelaserpulseatthebeginningoftheFLIMobservationtimewindow.Inthiswaythefluorescence

decaywillbecentredinsidethe16nsFLIMwindow.Thiswouldalsopreventanyfoldingofthefluorescence

signalandtheassociatedmeasurementdistortion.Ofcoursesincethe firstandthe last2nsof theFLIM

windoware ignored,thelaserpulse isnotseenwhenitfallsthere. Ifthishappens,making1mlongerof

shorter the cable that connects the camera to the laserwillmake the laser pulse reappear in the FLIM

observationwindow. Such calibration could be useful also to zero any skew among pixels but, inmany

applications, like the FLIM, it is usually not needed; should it be, the SPC3user can contactMPDat the

followingemailaddressformoredetails:[email protected]

Finally,afactorycalibrationisperformedalsoontheactualaveragegate-shiftbin(i.e.minimumgateshift):

thisvalueisreportedtotheuserthroughtheprovidedsoftware,anditshouldbeusedasascalefactoron

thetimeaxisofthereconstructedFLIMwaveform.

FLIMdataacquiredcanbeeithersavedonamultipageTIFFwithembeddedacquisitionmetadataaccording

totheOME-TIFFformatorintheproprietarySPCFformat.Forbothformats,imagedataiscomposedbya

setofimagesfollowinga"FLIMfirst,timesecondscheme",i.e.withthefollowingframesequence:1stgate

shift of 1st FLIM measurement, 2nd gate shift of 1st FLIM measurement, ..., nth gate shift of 1st FLIM

measurement,1stgateshiftof2ndFLIMmeasurement,2ndgateshiftof2ndFLIMmeasurement,...,nthgate

shiftof2ndFLIMmeasurement,etc.

OME-TIFFfilecouldbeopenedwithanyimagereadercompatiblewithTIFFfile,sincemetadataaresaved

intotheImageDescriptiontaginXMLformat.InordertodecodeOME-TIFFmetadata,itispossibletouse

free OME-TIFF readers, such as OMERO or the Bio-Formats plugin for ImageJ. Formore details see the

OME-TIFF web site: http://www.openmicroscopy.org/site/support/ome-model/ome-tiff/. OME-TIFF

metadata include theModuloAlongT tag,which allows theprocessing of FLIMdatawith dedicated FLIM

softwares such as FLIMfit (see http://www.openmicroscopy.org/site/products/partner/flimfit). A tutorial

on basic usage of FLIMfit is available at http://help.openmicroscopy.org/flimfit.html. In order to import

OME-TIFF file generated by SPC3 camera, simply select “Load FLIM data…” from Filemenu. For further

supportonFLIMfitpleasecontactdirectlyitsdeveloper.

SPCF file are binary files composed by a headerwith acquisitionmetadata followed by raw image data,

containingthe8/16bitpixelvaluesinrow-majororder.SPCFfilecanbereadusingtheprovidedImageJ/Fiji

plugin. In addition, since the format is fully documented in SPC3-SDK documentation, users can directly

readdatafromtheirowncode.



FurtherReadings

1. JosephR.Lakowicz.PrinciplesofFluorescenceSpectroscopy3rdedition.Springer(2006).

2. J.Pawley,HandbookofBiologicalConfocalMicroscopy,3rdEdition.NewYork:PlenumPress,2006.

3. T.W.J.Gadella,Ed.,FRETandFLIMTechniques,Volume33.Amsterdam:ElsevierScience,2008.

4. S.P.Chan,Z.J.Fuller,J.N.Demas,andB.A.DeGraff,“Optimizedgatingschemeforrapidlifetime

determinationsofsingle-exponentialluminescencelifetimes.”AnalChem,vol.73,no.18,pp.4486–

4490,2001.

5. D.D.-U.Li,S.Ameer-Beg, J.Arlt,D.Tyndall,R.Walker,D.R.Matthews,V.Visitkul, J.Richardson,

and R. K. Henderson, “Time- domain fluorescence lifetime imaging techniques suitable for solid-

stateimagingsensorarrays.”Sensors,vol.12,no.5,pp.5650–5669,2012.

Acquisitionparameters

Integrationtime

Each pixel of the SPC3 camera integrates three 9-bit binary counters (two of them with up-down

capability),whichmeasurethenumberofphotonsdetectedduringaspecifiedHardwareIntegrationTime

(HIT). TheHIT is thus the time during which the impinging photons are counted without resetting the

integratedcounters,i.e.thetimefromwhenthesecountersareclearedandstartedtothetimewhenthey

arelatchedandread. ItcanrangefromtheHardwareReadoutTime(HRT)ofthearraytoabout0.65ms.

TheHardwareReadoutTimeofthearrayisproportionaltothenumberofcounters(Ncounters)inuseandit

alwayscorrespondstotheinverseoftheframe-rate.InotherwordstheHRTistheminimumtimeneeded

toreadthecountersinuseforallthepixels.HRTvalueisgivenby:

𝐻𝑎𝑟𝑑𝑤𝑎𝑟𝑒𝑅𝑒𝑎𝑑𝑜𝑢𝑡𝑇𝑖𝑚𝑒 = 𝑁!"#$%&'( ∗ (5 𝑛𝑠 ∗ 𝑁!"#$% + 160 𝑛𝑠)

Incaseofuseofasinglecounterperpixel,thistimeisequaltothetimeneededtoreadthe2048pixelsand

is10.40µs;for2countersitisequalto20.80µsandfor3countersitisequalto31.20µs.Itmustbealso

noted that even when saving only half the array, in order to reduce the mass storage bandwidth

requirement,thereadoutwillbecarriedoutfortheentireimagerandsotheHardwareReadoutTimewill

notbereduced.

It is alsopossible to increase the total integration timeby internally (inside the camera’s internal FPGA)

accumulating severalHardware Integration Time, in order to obtain the so called Frame Exposure Time.

This is a viable solution since the SPC3 has no readout noise, thus there is no penalty in repeating the



readout,apart foranegligibledead-time (or inter-framedead-time)of less than10nsbetweenadjacent

HardwareIntegrationTime.ThenumberofHardwareIntegrationTimesthatareaccumulatedtoobtaina

FrameExposureTimecanrangefrom1to65534.AbeneficialsideeffectofaccumulatingseveralHardware

IntegrationTimes insteadofusinga longerHardware IntegrationTime is thattheequivalentdynamicsof

the internal counter is increased. In fact, sinceeachHardware IntegrationTimecanaccommodateup to

511photondetections,2Hardware IntegrationTimecanhold1022detections,10Hardware Integration

Timecanhold5110andsoon.

Finally, it ispossibletoemploytheinternaldetectorgatingcapabilityofCounter1 inordertoachievean

EquivalentHardwareIntegrationTimesmallerthantheHardwareReadoutTime.Thisispossiblebygating-

offthedetectorduringaportionofHardwareIntegrationTime.EquivalentHardwareIntegrationTimecan

beasshortas10ns.Pleasenotehowever,thatusinganEquivalentHardwareIntegrationTimeshorterthan

theHardwareReadoutTimeresultsinacorrespondingincreaseintheinter-framedead-time,byanamount

equal to the difference between the Hardware Readout Time and the chosen Equivalent Hardware

IntegrationTime.Since inthisparticularworkingmodethegate iscontrolleddirectlybythecamera, it is

notrecommendedtoapplyanyexternalgatesignalortochangetheinternalgatesettings,eitherwiththe

PCsoftwareorwith thestandardgate-controlproceduresprovided in theSDK.Particularly thenewgate

settingswilloverwritethealreadyappliedonesthusinterferingwiththeEquivalentHardwareIntegration

Timegenerationprocess.

Inordertoconfigurethesethreeparameters,theSPC3canbesetintotwodifferentmodes.Innormalmode

the camera settings are constrained so that the Hardware Integration Time is equal to the Hardware

ReadoutTime, inorder toavoidorat least reducetoaminimumtheprobabilityof theoccurrenceofan

overflow of the integrated counters. In fact, the internal 9-bit binary counters can overflow at long

HardwareIntegrationTimesandthusinduceartefactsinthedetectedimage.

Asalready shown, themaximumnumberofphotonsNmax that canbecountedper second,givenadead

timeofTdead,is:

𝑁!"# =1

𝑇!"#!(𝑐𝑜𝑢𝑛𝑡𝑠/𝑠𝑒𝑐𝑜𝑛𝑑)

thusthemaximumnumberofphotonsthatcanbecountedduringaHardwareIntegrationTimeequaltoT

is:

𝑁!"#!!"# =1

𝑇!"#!𝑇

Forinstance,whenusingasinglecounter,Tisequalto10.40µsandwiththeminimumTdeadof50ns,the

maximumcountedphotonsis:



𝑁!"#!!"# =𝑇

𝑇!"#!=10.40 µ𝑠50 𝑛𝑠

≅ 208 < 511

thus,anyoverflow isavoided since511 is themaximumnumber that canbecountedbya9-bit counter

before overflowing. Instead, when using all the three counters, T is equal to 31.20µs and with the

minimumTdeadof50ns,themaximumnumberofphotonsthatcanbecountedis:

𝑁!"#!!"# =𝑇

𝑇!"#!=31.20 µ𝑠50 𝑛𝑠

≅ 624 > 511

andtheoverflowmayoccurwithverystronglight.

As a consequence, in order tominimise the probability of incurring in overflows, or even to completely

avoidtheminsomecases,acleverstrategyistoobtainlongFrameExposureTimesbyaccumulatingmany

Hardware IntegrationTimes,eachequal to theHardwareReadoutTime,directly in theSPC3.Due to the

short inter-framedead-timesmaller than10ns, thisoperatingmode inducesonlynegligible signal losses

anddoesnotincreasethenoise(sinceread-outnoiseisnotpresentintheSPC3Camera).

Inadvancedmode, bothHardware Integration Timeand Frame Exposure Time are user adjustable, thus

particularcaremustbetakenbytheuserinordertoavoidartefactsduetotheoverflowoftheintegrated

9-bit counters, which is no more avoided or minimized. In addition, also the ability to achieve shorter

EquivalentHardware Integration Time on Counter 1, by using the internal gate generation, is enabled.

InternalgateisautomaticallyenabledwhentheuserselectsaHardwareIntegrationTimeshorterthanthe

HardwareReadoutTime.Itisthenuptotheuserthecalculusoftheactualinter-framedead-time;forthis

purpose,sincetheHardwareReadoutTimeisequaltotheinverseoftheframe-rateinSnaporContinuous

Acquisitionmodes,itcouldbeusefultohavealookattheactualframe-ratebyenablingandcheckingthe

SyncOutoutput.Pleasealsonotethatwhenthisfeature isenabled,onlyCounter1willbeautomatically

gated,whereas Counters 2 and 3will have an actualHardware Integration Time equal to theHardware

ReadoutTime. Figure9 summarizes the relationshipbetweenHardware IntegrationTime,ReadoutTime,

FrameExposureTimeinthetwoworkingmodes.

Dead-timetimeanddead-timecorrection

SPADdead-timecanbeselectedbetweenaminimumvalueof50nsandamaximumvaluewhichdepends

on the specific camera but is never lower than 130 ns. There are about 60 discrete values that can be

assumedover the full range, thuswhenever a value is input by the user, the SPC3 camera is set to the

closestpossiblevalue,accordingtofactorycalibration.Theactualvaluecanbereportedtotheuserbythe

software.



Figure9. Integrationtimesandworkingmodes.

Thedead-time strongly affects the imagequality. It limits theafter-pulses generated ineach singlepixel

when set at high values, e.g. 150 ns, but it introduces a limitation on the total number of photons per

second thateachpixel candetect. This inducesanon-linearity,whenbright spots areobserved (see the

introductorysectionforfurtherdetails).

Itispossibletoimprovethequalityoftheimagebycorrectingforthenon-linearityintroducedbythedead-

time(seesection“Single-PhotonAvalancheDiode”).Infact,asimpleoperationasbackgroundsubtraction

might perform incorrectly when large numbers of photons per second are detected by the pixels. This

occurs because, for example, a cumulated flux of 10’000 photons (5’000 signal photons + 5’000 optical

backgroundphotons)mightsufferalargernon-linearityinthedetectionthantheacquisitionofonly5’000

opticalbackgroundphotons,aspredictedbyEq.1andshowninFigure2.Forsakeofclaritylet’sconsider,

asbefore,5’000photonsper integrationwindowTbothforthebackgroundandforthesignal.Let’salso

assumeT=1ms,adead-timeof100nsandaPDE=100%forsimplicity.Inthissituationiffollowsthatonly



3’333photonsarecountedwhenacquiringthe5’000background(BG)photonsandonly5’000arecounted

whenacquiringthecumulated10’000photonsofsignalandbackground(SG+BG),seeEq.1forreference.It

is thus clear that subtracting themeasured 3’333 BG photons from the 5’000 SG+BG photons is totally

incorrect and results in the over-subtraction of the backgroundwhich should have been 2’500 and not

3’333.Inanycase,thecorrectwayofproceedingistocorrect,bymeansofEq.2,forthedead-timeeffect

the actualmeasured number of photons and obtain the real number of impinging photons (10’000 and

5’000)forbothimagesandonlyAFTERperformthesubtraction.

Byenabling theembeddeddead-timecorrection, theeffectof thenon-linearity ismostly removedanda

betterandfasterprocessingofthedatabecomespossible.Figure10showstheimprovementoftheimage

qualityoftwo“white”frameswithautomaticbackgroundsubtractionenabled.Thereferenceimageshows

a noisier pattern compared to the dead-time corrected one. This is due to an “over-subtraction” of the

background,causedbydead-timeinducednon-linearity.

Theeffect isevenmorevisible in the intensityhistogramontherightofFigure10.Thehistogramof the

reference imagedeviates froman expectedGauss-like profile and shows a remarked shoulder for lower

numberofphotons(greencurve).Conversely,thecorrectedimageisclosertotheidealGaussdistribution

(bluecurve).

Itmustbeobservedthat thedead-timecorrection isonlyeffectiveunderamoderate illuminationof the

sensor.ForthisreasonwehavedecidedthattheSWimplementeddead-timecorrection,doescompensate

the non-linearity shown in Figure 2 but it does not in any case expand the counting range beyond the

natural limit imposed by the hold-off time, as shown in Figure 11. As a consequence, when a strong

illuminationisapplied,thecorrectedvalueswillinanycasesaturatetotheexpectedsaturationlevelgiven

bythecurrenthold-offtime.

Figure10. Improvement of the image quality by using the dead-time correction algorithm. The intensity

histogramsofthetwoimagesareplotontheright.



Figure11. Effectofthedead-timecorrectorinanimageof1msexposureanddead-timeof500nsasafunctionofthenumberofincomingphotons.

Backgroundsubtraction

SPC3 can perform in-camera background subtraction. For this purpose the user must have previously

acquiredadarkimage(i.e.withtheshutterclose)withthesameparametersoftheactualacquisition(i.e.

thesameintegrationtime,hold-offandgate),andthenhastoenablethebackgroundsubtraction.

Additionalinformation

CalculationoftheSnapsizeandtheContinuousAcquisitionDatarate

Integration Time and Half-Array Saving Mode, directly influences the data size of a single Snap mode

acquisitionandthedatarateinContinuousAcquisitionmode.

Snapmodedatasize,inbytes,isgivenby:

𝑆𝑛𝑎𝑝_𝑆𝑖𝑧𝑒 = 𝑁!"#$%_!"#$% ∗ 𝑁!"#!"# ∗𝐵𝑃𝑃8

where Nframes is the number of frames in the Snap and bit-per-pixel (BPP) is 8 if there is no internal

accumulation of several Hardware Integration Times and both dead-time correction and background



subtraction areoff, 16otherwise.Npixel_Saved is thenumberof savedpixel and currently canbe2048 (full

array)or1024(halfarray).

ContinuousAcquisitionmodedatarate,inbytes-per-second,isgivenby:

𝐶𝐴_𝐷𝑎𝑡𝑎𝑅𝑎𝑡𝑒 = 𝑁!"#$%_!"#$% ∗𝐵𝑃𝑃8

∗ 𝐹𝑟𝑎𝑚𝑒𝑅𝑎𝑡𝑒

where

𝐹𝑟𝑎𝑚𝑒𝑅𝑎𝑡𝑒 =1

𝐻𝑎𝑟𝑑𝑤𝑎𝑟𝑒_𝑅𝑒𝑎𝑑𝑜𝑢𝑡_𝑇𝑖𝑚𝑒 ∗ 𝑁!"#!

beingNsumsthenumberofinternalaccumulations.

Cameraauto-protection

TheSPC3hasaprotectionmechanism that shutsdown theSPADhighvoltagewhen toomuchcurrent is

flowing through the array, for example when really high photon fluxes are impinging on the camera.

Solution:Reducetheamountoflightthatfallsonthesensor,disconnectandreconnectthecameratothe

hostPCandrestartthesoftware.



SPC3Softwareinterface–VisualSPC3(Windowsonly)Thecameraisprovidedwithanacquisitionandcontrolsoftware,whichisrunningonMicrosoftWindows

operatingsystems:VisualSPC3.Thissoftwareprovidesadirectaccesstoallthecamerafunctionsanditisa

basic tool todisplay images. It is recommendedtouse theSoftwareDevelopmentKit foramore flexible

camera control. Due to the start-up time needed by the internal firmware to be fully working, the

VisualSPC3 should be started at least 15 seconds after having powered the SPC3. Failing to do so will

generateanerrorbythesoftware.Simplycloseandrestartthesoftware.

InstallationDouble-clickthedesiredinstaller(32-bitor64-bit)andfollowtheguidedinstallationprocedure.Pleasenote

thatdevicedriverfromOpalKelly,providedontheSPC3USB-Key,mustbeinstalledbeforeconnectingthe

cameratothePC.

SoftwareInterfaceTheVisualSPC3 interface is composedamainwindowdivided in several sectionsas showed in Figure12,

andaseparatewindowforeachactivecounter(seeFigure13).Anadditionalwindowwillalsopop-upwhen

FLIMmodeisenabled(seeFigure14).

Figure12. Screenshotofthemainwindowofthegraphicaluserinterface.



Mainwindow:ControlPanel

Thispanelallowssettingallthecamerahardwareparameters.

In the Exposure Parameters subsection it is possible to set the image integration parameters (exposure

time, number of frames for Snap mode and hardware binning). Either normal or advancedmodes are

available. There are four numeric fields (see Table 3 for an example of settings), which can be either

editableorautomaticallyadjustedbythesoftwaredependingoftheselectedmode,namely:

ü Frameexposuretime(Normalonly):integrationtimeforeachframerequiredbytheuser.Thepossible

values range between 10.40 µs and 2.40 ms. This value does not correspond to the Hardware

IntegrationTimeofthesensor.Longerexposuretimesareobtainedbybinningavariablenumberof

frames.

ü Hardware Integration time: physical integration time of the SPC3 sensor. This value is fixed to the

HardwareReadoutTimeinnormalmode.Inadvancedmode,itcanrangefrom10nsto655.25µs.

ü Hardware binning: number of frames of duration equal to Hardware Integration Time that are

summed-up before being stored in the computer memory. In Advanced mode this field can be

changedbytheuser, inNormalmodeit isadjustedbythesoftwaredependingonthechosenFrame

ExposureTime

ü Numberofframes:numberofframesthatareacquiredduringaSnapevent.Thisnumberrangesfrom

1to65534for16bitimagesandto131070for8bitimages.

Table3. PossibleconfigurationsoftheExposureParameterssubsection(NormalandAdvancedmode)

NormalAcquisitionControlBox

The duration of a frame is obtained by binning a sequence of frames ofshorter exposure times. In this example, a 52.00 µs Frame exposure isobtained by summing 5 frames of 10.40 µs. This acquisition is thenrepeated 1000 times and the data are transferred to the host computer.Practically,5000framesof10.40µsareacquiredandbinnedbyafactorof5. Thisacquisitionmodedoesnotdegrade the signal-to-noise ratioof theimagesbecausenoread-outnoiseispresent.

AdvancedAcquisitionControlBox

Theuserhasafullcontrolontheframeacquisition.In the example, each frame has an effective duration of 103.7 µs and nobinningoftheframesisperformed(Hardwarebinning=1).Theacquisitionis then repeated 1000 times and the data are transferred to the hostcomputer.Frameexposuresdownto10nscanbeobtained.WARNINGtheintegratedcounterscanoverflowatlongexposuretimes.



Inthissubsectionthereisalsoasetofcheckboxestofurthercustomizingtheacquisition.Thefull listand

theirexplanationcanbefoundinTable4.

Table4. ListofcheckboxesintheExposureParametersubsectionandtheirexplanation

Acquirehalfarray Only the central 32x32 portion of the array is acquired. Useful toreducerequiredbandwidth.

EnableCounter2

EnableCounter3

Enable theextra in-pixel counters.Whenacounter isenabledanewCounter window pop-ups, when a counter is disabled thecorrespondingwindowisclosed.

Signeddata

When at least Counter 2 is enabled, this checkbox indicates to thesoftwareifthedatafromCounter2and3hastobeconsideredsignedornot.Usefulwhenthesecountersareemployed inup/downmode.Note that, if negative values are expected, this checkbox MUST bechecked,otherwisedatawillbeincorrect.

Waitforexternaltrigger Thecamerawaits foranexternal triggerontheTRIG IN inputbeforestartingtheacquisitionofanimagesequence.

EnablebackgroundsubtractionThe acquired dark frame is subtracted by the output image. A darkframe acquisition must have been previously performed (see theControlButtonsparagraph).

Force8bitmode AvailableonlyinAdvancedmode.Dataaretruncatedto8bit.Usefultoreducerequiredbandwidth.

IntheSoftwareGatingsubsectionitispossibletoconfiguretheinternalgategeneration,whichappliesonly

toCounter1.Twogatedoperationmodearepossible,PeriodicGatemodeandFLIMmode.InPeriodicGate

modeaperiodicsignaliscontinuouslyappliedtoCounter1,asexplainedinTable5.

Table5. PeriodicGatingsectionconfigurationexplained.

EnableGate Enable/Disablethehardwaregating

DelayChange the delay of the gate signal respect to the internal reference clock. The delay value isexpressedasapercentageofthereferenceclockperiod(20ns).E.g.Delay=+10%è+2nsdelay

WidthDurationofthegate-onsignal.Thisisexpressedasapercentageoftheinternalreferenceclockperiod(20ns).E.g.Width=20%è4nswidth.

InFLIMmodetheinternalgateisautomaticallycontrolledinordertoperformagatedFLIMacquisitionwith

parametersexplainedinTable6.SeeparagraphonFLIMoperationforfurtherdetails.



Table6. FLIMmodesectionconfigurationexplained.

EnableFLIMmodeEnable/DisableFLIMmode.WhenFLIMmodeisenabledanadditionalFLIMwindowpop-ups(seeparagraphbelow).

GateWidthDurationofthegate-onsignal.Thisisexpressedasapercentageoftheinternalreferenceclockperiod(20ns).E.g.Width=20%è4nswidth.

FLIMsteps NumberofstepstobeperformedforeachFLIMacquisition.

FLIMshiftperstepGateshiftforeachFLIMstepsinthousandthsofreferenceperiod.E.g.Step=50‰è1nsstep

FLIMStartpositionStartpositionoftheofthefirstFLIMstepwithrespecttotheinternalreferenceclock.Thedelayvalueisexpressedasapercentageofthereferenceclockperiod(20ns).E.g.Delay=-10%è-2nsdelay

IntheDeadtimesubsectionitispossibletoadjustthehold-offtime(intherange50÷150ns)andenable

theembeddeddead-timecorrection.

In theSynchronizationOutputsubsection it ispossible toenable thegenerationofvoltagepulseson the

SYNCOUToutput.Twopossiblesynchronizationsignalsareavailable:

ü HITsync:a50nspulseatthebeginningofeachHardwareIntegrationTime

ü Ref.Clock:theinternal50MHzclockisreplicatedtothisoutput.

MainWindow:PlotsettingsandPlaybackControl

Theplotsettingsapplyonlytotheimagesdisplayedonthescreen.Nochangesareintroducedtothe

datainthememorybuffer.Therefore,theseparametersdonotaffectthesavedimages.Allthissettings

canbechangedinreal-timeduringdisplay.TheplotsettingscanbeadjustedasexplainedinTable7.

After a Snap has been acquired, it is possible to review the images by using the controls shown in

Table8.Itisalsopossibletomovealongthedatausingthescrollbar.Thecurrentframenumberand

FLIMstep(ifinFLIMmode)arealsoshown.



Table7. Plotsettingsandcontrolbuttons.

Automatically sets the black and white levels of the image according to the presentframe: theblack level is assigned to thepixelwith theminimumnumberofdetectedphotons intheframewhilethewhite level isassignedtotheoneswiththemaximumcount.Thissettingwillremainappliedalsoforthefollowingframes.Sinceitispossibleto have in the following frames pixels with values lower than black or higher thanwhite,theywillberespectivelyindicatedinblueorred.

Resetsblackandwhitelevelstodefaultvalues:0÷255for8-bitimages,and0÷65535for16-bitimages.

FlipX FliptheimagealongtheX-axis.

FlipY FliptheimagealongtheY-axis.

Rotate90°CW Rotatetheimageby90°clockwise.

KeepWindowsDocked Ifselected,movingthemainwindowwillalsomovetogetheralltheotherwindows.

Table8. Liveviewercontrolbuttons.

Gotothefirstacquiredframe

Gotothepreviousframe

Gotothenextframe

Gotothelastacquiredframe

Playtheimagesequencebackward

Continuous rewind.When the last (first) acquired frame isdisplayed, it restarts fromthefirst(last)frame

Playtheimagesequenceforward



MainWindow:ControlButtonsTable9. Cameracontrolbuttonsforstarting,recordingandsavingframeacquisitions.

Startthelive-modeacquisition.Oncetheacquisitionhasbeenstarted,thebuttonbecomesaredbox.Pressitagaintostoptheacquisition.InFLIMmodetheaveragecountsoverallFLIMstepsisvisualized.

Starts the acquisition of a sequence of frames (Snap mode). If no external trigger isrequested, the acquisition starts immediately and the button becomes a red box. If anexternaltriggerisrequired,thebuttonturnsfirstinanorangebox,theninaredoneonlywhen the trigger is detected and the acquisition really starts.When thebox is orange, aclickonthebuttonstopsthewaitingforatrigger.Whentheboxisred,theusermustwaitfortheendoftheacquisition.

Starts the continuous acquisition of frames. If no external trigger is requested, theacquisitionstarts immediatelyandthebuttonbecomesaredbox. Ifanexternal trigger isrequired,thebuttonturnsfirstinanorangebox,theninaredoneonlywhenthetriggerisdetected and the acquisition really starts. For interrupting the waiting for the externaltriggersignalortheacquisition,onceithasbeenstarted,pressthebutton.WARNING:Continuousmodeacquisitionmaygeneratehugeamountofdatainshorttime,dependingonacquisitionsettings.Check“Totalsaveddata” intheAcquisitionParametersSummaryboxfortheactualsizeofdatasavedontheHDD.

Performstheacquisitionofabackgroundframe.Oncethe light ispreventedtoreachthedetector, only dark counts are measured. A dark frame is generated by averaging thenumber of frames acquired. This frame is then sent to the acquisition electronics toperformhardwarebackgroundsubtraction.Theacquisitionofthebackgroundframemustbe performedusing the sameparameters thatwill be used for the actualmeasurement.Anychangeintheacquisitionparameters,includingthegateparametersandthedead-timecorrectionsettings,REQUIRESanewacquisitionofthebackgroundframe.Thebackgroundframeisnotstoredinsidethecameraandislostwhenthecameraisswitchedoff.

SavestheacquireddatabyaSnapintoaSPC3formatfileorspecialSPC3FLIMformat.

SavesthedataacquiredbyaSnapintoamultipageTIFFfile.ThefileiscompliantwiththeOME-TIFF specification, which allows embedding in the file metadata all the acquisitionparametersforthespecificmeasurement,aswellasFLIMmodeparameters.OME-TIFFfilecanbeopenedwithanyimagereadercompatiblewithTIFFfile,sincemetadataaresavedintotheImageDescriptiontaginXMLformat.InordertodecodeOME-TIFFmetadata,itispossibletousefreeOME-TIFFreader,suchasOMEROortheBio-FormatspluginforImageJ.FormoredetailsseetheOME-TIFFwebsite:http://www.openmicroscopy.org/site/support/ome-model/ome-tiff/.OME-TIFFmetadataforFLIMmodeacquisitionincludetheModuloAlongTtag,whichallowsthe processing of FLIM data with dedicated FLIM software such as FLIMfit (seehttp://www.openmicroscopy.org/site/products/partner/flimfit).

ExitsVisualSPC3



MainWindow:AcquisitionParametersSummary

In this section, the information on the acquisition, based on the actual parameters, is reported. The

explanationofalltheshownparametersisreportedinTable10.

Table10. ExplanationoftheparametersshownintheAcquisitionParametersSummarypanel.

Framerate ThisistheactualframerateofaSnaporContinuousAcquisition

TotalFrameTimeThis is the actual frame exposure time (i.e. the realHardware IntegrationTime,multipliedbythenumberofinternalsums)

Totalacquisitiontime ThisisthetotaltimerequiredtoconcludetheSnapacquisitionofNframes

Interframedead-timeThisistheactualinter-framedead-time.Normallyitis<10nslong,butmayincrease in Advanced mode if very short Hardware Integration Times areselected.

Bitperpixel Numberofbitperpixel(8or16)

Snapfilesize FiledimensionforaSnapacquisitionwiththecurrentparameters

Totalsaveddata(onlyvisibleduringContinuousAcquisition)

Thisistheactualsizeofdatasofarsavedtothedisk.Thenumberisupdatedinreal-timeaftereverywriteoperation.

CounterWindow

VisualSPC3willshowuptothreewindows(seeFigure13)showingthemapoftheaccumulatedcounts in

eachcounter.WindowforCounter1isalwaysvisible,windowsforCounter2and3willpop-upifselected

intheExposureParameterssubsectionoftheControlPanel. InadditiontosettingsshowninTable7, it is

possibletomanuallyadjusttheblackandwhitelevelsoftheimageanditscontrastbyusingthecontrolson

the color-bar at the left of the image.After a Liveor Snapmodeacquisition is stopped, it is possible to

directlyreadthenumberofphotonsdetectedbyeachpixelbysimplymovingthemousepointeroverthe

image(availableonlyifwindowsizeissettothedefaultvalue.Check“Keepwindowsdocked”torevertto

defaultsize).

InFLIMmodeonlyCounter1windowisavailable.InLivemodetheaveragecountsoverallFLIMstepsare

visualized,whereas inSnapmodecounts foreach individualFLIMstepareshown.For thesakeofclarity

let’sconsideraSnapof10FLIMacquisitions(camerainFLIMmode)andlet’ssupposethateachacquisition

is composed by 50 steps (Gate Shifts, i.e. ‘Gate positions’): a total of 500 frameswill be thus shown in

Counter1window,followinga"FLIMfirst,timesecondscheme".Withsuchscheme,theframesequence

wouldfollowthisordering:1stgateshiftof1stFLIMmeasurement,2ndgateshiftof1stFLIMmeasurement,

...,nthgateshiftof1stFLIMmeasurement,1stgateshiftof2ndFLIMmeasurement,2ndgateshiftof2ndFLIM

measurement,...,nthgateshiftof2ndFLIMmeasurement,etc.



Figure13. ScreenshotoftheCounterwindow.

FLIMWindow

WhenFLIMmodeisenabled,anadditionalFLIMwindowpop-ups,asshowninError!Referencesourcenot

found..Thisgraphrepresentsthecountedphotonsineachstep(GateShift)ofthecurrentFLIMacquisition

foraselectedpixel,anditisthustheTCSPChistogramthatreconstructsthewaveformunderinvestigation

convolutedwith the IRFof the system.The IRF is roughlya rectangular shapewithawidthequal to the

Gatewidth.Tochangewhichpixelisvisualized,simplyclickonthedesiredpixelinCounter1window.The

graphisupdatedinrealtimeinLivemode.InSnapmodeitshowstheentireFLIMwaveformcorresponding

tothecurrentlyvisualizedFLIMacquisition(i.e.itwillnotchangeif,inCounter1,adifferentFLIMstepfrom

thesameFLIMacquisition isvisualized). It ispossibletochoosebetweenlinearandlogarithmicscalefor

the Y axis. TheAccumulate checkbox, if selected,will cause the histogram to accumulate counts among

frames.

Figure14. ScreenshotoftheFLIMwindow.Inthiscase,theacquisitionistheIRFofthesystemwhenGatewidthissetto2ns,and800gatestepsspanning16nsareacquired.



SPC3-SDK,FileFormatsandImageJPluginSPC3 isprovidedwithaSoftwareDevelopmentKit (SDK),composedbyashared libraryand fewexample

projects.TheSDKallowstheuseracquiredatafromtheSPC3andtochangeoperatingparameterdirectly

fromtheirownapplication.ItisavailableforWindows,MacOSXandLinuxoperatingsystems,bothin32-

and64-bitversions.More informationonavailable functionscanbe found in the relateddocumentation

includedintheUSBkey.

Acquired data from Snap acquisitions can be saved both in OME-TIFF format and in aMPD proprietary

binaryfileformat.DatafromContinuousacquisitioncanonlybesavedinMPDproprietaryformat.

OME-TIFFfilecouldbeopenedwithanyimagereadercompatiblewithTIFFfile,sincemetadataaresaved

intotheImageDescriptiontaginXMLformat.InordertodecodeOME-TIFFmetadata,itispossibletouse

free OME-TIFF readers, such as OMERO or the Bio-Formats plugin for ImageJ. Formore details see the

OME-TIFFwebsite:http://www.openmicroscopy.org/site/support/ome-model/ome-tiff/.

ProprietaryformatfilesproducedbySPC3cameraarebasedonthesamebinarystructure,butdifferentfile

extensions are used for differentiate among data. Photon counting acquisitions are stored in .spc3 files

whereasFLIMmodeacquisitionsaresavedin.spcffiles.Inaddition,eachcameraisprovidedwithan.spce

file containingmeasured photondetection efficiency. All these files contain binary data, composedby a

headerwithacquisitionmetadata followedby raw imagedatawhichcontain the8/16bitpixel values in

row-majororder.Thebyteorderislittle-endianforthe16bitdata.Rawdatahavethefollowingmeaning:

.spc3files: data represents photon counts in each pixels during each Exposure Time, with acquisition

parameters reported in theheader. Incasemorecountersareused,dataare interlaced, i.e.

thesequenceof frames isthefollowing:1st frameof1stcounter,1st frameof2ndcounter,1st

frameof3rdcounter,2ndframeof1stcounter,etc.

.spcffiles: data represents photon counts in each pixels during each Exposure Time, with acquisition

parametersreportedintheheader.Imagesfollowa"FLIMfirst,timesecondscheme",1stgate

shiftof1stFLIMmeasurement,2ndgateshiftof1stFLIMmeasurement, ...,nthgateshiftof1st

FLIM measurement, 1st gate shift of 2nd FLIM measurement, 2nd gate shift of 2nd FLIM

measurement,...,nthgateshiftof2ndFLIMmeasurement,etc.

.spcefiles: data representmeasuredPhotonDetectionEfficiency foreachpixel.Values range from0 to

10000,i.e.a32.5%PDEwillbeexpressedas3250.Eachframerepresentsawavelength,whose

rangeandsteparereportedintheheader.



The header is composed by a signature of 8 bytes (0x4d5044ff03000001, with byte 0 set to 4d), and a

metadata sectionof 1024bytes, as reported in Table 11.Note thatmultibyte numerical fields are little-

endian.

Table11. HeaderstructureofSPC3proprietaryfileformats.spc3,.spcf,.spce.

Byteoffset Numberofbytes Description 0 10 UniquecameraID(string) 10 32 SPC3serialnumber(string)42 2 Firmwareversion(1.11savedas111) 44 1 Firmwarecustomversion(standard=0)45 20 Acquisitiondata&time(string)65 35 Unused100 1 Numberofrows101 1 Numberofcolumns102 1 Bitperpixel 103 1 Countersinuse104 2 Hardwareintegrationtime(multiplesof10ns)106 2 Summedframes108 1 Deadtimecorrectionenabled 109 1 Internalgateduty-cycle(0-100%) 110 2 Holdofftime(ns)112 1 Backgroundsubtractionenabled113 1 Dataforcounters1and2aresigned114 4 Numberofframesinthefile118 1 Imageisaveraged119 1 Counterwhichisaveraged 120 2 Numberofaveragedimages 122 78 Unused200 1 FLIMenabled201 2 FLIMshift%203 2 FLIMsteps 205 4 FLIMframelength(multiplesof10ns) 209 2 FLIMbinwidth(fs)211 89 Unused300 1 PDEmeasurement301 2 Startwavelength(nm) 303 2 Stopwavelength(nm) 305 2 Step(nm)307 717 Unused



Inordertoeasilyvisualizeandelaboratedatastoredintheproprietary.spc3,.spcfand.spcefileformats,a

pluginforImageJsoftware(http://rsb.info.nih.gov/ij/)isalsoprovided.Thepluginiscompatiblebothwith

ImageJ (tested on v. 1.49q) and with Fiji (http://fiji.sc/Fiji), which is essentially a distribution of ImageJ

togetherwith several useful plugins.Note that file header format and file signature have been changed

fromSDKversion1.0.1toSDK1.1.1.ImageJplugin1.1.1willnotreadolderfiles,howeveracommandline

tooltoconvertSPC3filefrom1.0.1formatto1.1.1formatisprovidedintheSDK.Thisisalsousefultoread

.spcefilesprovidedwith1.0.1cameraswhenusingthenewplugin.

Inordertoinstallthepluginfollowthisprocedure:

• forImageJ(testedonv.1.49q),launchImageJandselectthemenuitemPlugins→Install…,thenfollow

the instructions and select the file SPC3_Reader.java provided in SPC3 USB key. The "Save Plugin,

MacroorScript..."dialogappearswithalistofyourcurrentlyinstalledImageJplugins.Selectafolder,

suchas the Input-Output folder, andsavetheSPC3_Reader.java file there.Donotchangethename

SPC3_Reader.java,as it followsImageJnamingconventionsandit isnotseenbyImageJ if itdoesnot

containanunderscore.

• forFiji,launchFijiandselectthemenuitemPlugins→InstallPlugin…,thenfollowtheinstructionsand

selectthefileSPC3_Reader.javaprovidedinSPC3USBkey.RestartFiji.

In order to use the plugin select SPC3_Reader under thePluginsmenu. Itwill be in the location you’ve

previouslychosenforImageJ,orattheendofthemenuforFiji.An"OpenSPC3file..."dialogappears.Select

the.spc3,.spcfor.spcefileyouwishtoload.ImageJwillnowaskyouhowtodealwithsigneddata.

Ifdatais8bit,therearethreeoptions:

• Convertdatato32bitfloat

• Importunsigneddataasisandconvertsigneddatato32bitfloat

• Import all data as is (thismeans ImageJ will interpret signed data as unsigned, e.g. the number -1,

whichin8bitbinaryis11111111,willbeshownas255)

Ifdatais16bit,therearefouroptions:

• Convertdatato32bitfloat

• Importunsigneddataasisandconvertsigneddatato32bitfloat

• Import all data as is (thismeans ImageJ will interpret signed data as unsigned, e.g. the number -1,

whichinbinaryis11111111,willbeshownas255)

• ImportalldataasisanduseanImageJcalibrationfunctiontomapsigneddata(thismeansImageJwill

remapdatatoshowsignedvalues,e.g.thenumber-1,whichin16bitbinaryis1111111111111111and

willreadas65535withoutremapping,willbeproperlyshownas-1).



If theselected file isproperly formattedandnoerroroccurred,an imagewindowforeachSPC3counter

enabledopens.Framescanbeskimmedusingthecontrolleratthebottomofeachwindow.Selectingfrom

themenuthecommandImage→ShowInfo…willshowallcameraparametersfortheacquisition.



Systemrequirements§ High-speedUSB3.0interfaceforfullspeedacquisition,compatiblewithUSB2.0

§ Hostcomputer(minimumrequirements)

o 2GHzprocessorand1GBofRAM

o SSDrecommendedforfullspeedcontinuousacquisition.

§ Supportedoperatingsystems

o VisualSPC3

§ MicrosoftWindows7,8,32or64bitversions

o SDKandImageJPlugin

§ MicrosoftWindows,7,8,32or64bitversions

§ LinuxUbuntu12.04 LTS, CentOS6.5,6.6,6.7orcompatibledistributions,32or64

bitversions.Differentdistributionsshouldwork,butwerenottested.

§ MacOSX10.8andabove

§ TestedImageJversions(differentversionsshouldwork,butwerenottested)

o StandaloneImageJ:version1.49q

o Fijidistribution:version2.0.0-rc-28/1.50b

CopyrightanddisclaimerNo part of this manual, including the products and software described in it, may be reproduced,

transmitted,transcribed,storedinaretrievalsystem,ortranslatedintoanylanguageinanyformorbyany

means, except for the documentation kept by the purchaser for backup purposes, without the express

writtenpermissionofMicroPhotonDevices S.r.l.All trademarksmentionedherein areproperty of their

respectivecompanies.MicroPhotonDevicesS.r.l.reservestherighttomodifyorchangethedesignandthe

specificationstheproductsdescribedinthisdocumentwithoutnotice.

Single Photon Counting Camera - Micro Photon Devices · SPC3 is a single photon counting camera...

Documents

Transcript of Single Photon Counting Camera - Micro Photon Devices · SPC3 is a single photon counting camera...