CMOS ACTIVE PIXEL SENSOR TECHNOLOGY - AN …

1 PhotobitTECHNOLOGYTECHNOLOGYCMOS ACTIVE PIXEL SENSOR TECHNOLOGY -AN INTRODUCTION-AN INTRODUCTION Dr. Eric R. Fossum ER FOSSUMP hotobitTECHNOLOGYIiStTECHNOLOGYI maging SystemsTiming &OpticsTiming &DriverCircuitsDetectorPixelPixelArrayAS PA nalog Signal ProcessorAnalog to Digital ConverterDigital Signal ProcessorADC ER FOSSUMD igital Signal ProcessorDSPP hotobitTECHNOLOGYPh tTECHNOLOGYP hotons f(f) A photon is a bullet of light (quantum view of light waves) Energy is related to wavelength: E = hc/ High energyLow energyUltra-iltInfra-Short wavelengthLong wavelengthvioletredYiiifildkdYour eye is sensitive from violet to dark red450 nm - 700 nmSilicon is sensitive from violet to the near infrared (not thermal)400 nm1050 nm400 nm -1050 nmSilicon is also sensitive to higher photon energies ( x-rays)

2 ER FOSSUMP hotobitTECHNOLOGYS ources of PhotonsTECHNOLOGYS ources of PhotonsThi Three main sources SolarIncandescent lamp Incandescent lamp Fluorescent lamp Typically characterized by color temperature Typically characterized by color temperature Temperature of blackbody with roughly same color distribution Higher temp means more blue, less red -> looks white ER FOSSUMP hotobitTECHNOLOGYSPECTRAL (nm)NIRUV ER FOSSUMP hotobitTECHNOLOGYOtiTECHNOLOGYO ptics Optics (lenses mirrors) COLLECT and FOCUS lightOptics (lenses, mirrors) COLLECT and FOCUS light Optics are measured by focal length (zoom-effect) and by speed or F-number (ratio of focal length to diameter)Li hti 1/4F2li ht Light on SENSOR is 1/4F2x light on scene Higher F-number means less light on image SENSOR Lower F-number mean more light Iris or diaphragm adjusts F-number by changing effective lens diameter Typical video cameras are F/2 (so faceplate illuminance is 1/16 scene)

3 Standard lens sizes are measured in inches (sort of lens diameter)Standard lens sizes are measured in inches (sort of lens diameter) 2/3 , 1/2 , 1/3 , 1/4 etc. Smaller lens sizes weigh less and cost less - important for portablesSll iillid ll ili (f Smaller sizes require smaller image sensors and smaller PIXEL sizes (for same # of pixels) and are less sensitive to light Actual image SENSOR diagonal is much less than lens size ER FOSSUM For example, 1/3 format ( mm) needs mm diagonal !PhotobitTECHNOLOGYP ixel SizesTECHNOLOGYP ixel Sizes A PIXEL is a picture element A PIXEL is a picture element More pixels means more RESOLUTION More pixels means more DATA (could be a bad thing) Modern PIXEL sizes are between 4 and 10 microns Pixels are square if designed for a computer display Pixels have a rectangular shape for TV applicationsFORMATC olumns RowsTotal #PixelsCIF352288100 ER lTECHNOLOGYP ixelsVpdpCpdCpdIpdVssvoltagevoltagetime ER FOSSUM timePhotobitTECHNOLOGYC lFiltA(CFA))

4 TECHNOLOGYC olor Filter Array (CFA) Each PIXEL gets covered by a colored filterEach PIXEL gets covered by a colored filter We use red, green, blue (RGB) CFA - best match for RGB displays in Bayer patternpyy p Could use complementary colors (cyan, yellow, magenta) (nm)Re ER FOSSUMW avelength (nm)PhotobitTECHNOLOGYFill F tTECHNOLOGYFill Factor A PIXEL is divided into a sensing portion and a A PIXEL is divided into a sensing portion and a readout portion Fill factor is the ratio of sensing area to total areaFill factor is the ratio of sensing area to total area and is typically about 20% for CCDs and CMOS APST otal PIXEL areaSitiSensitive area ER FOSSUMP hotobitTECHNOLOGYMilTECHNOLOGYM icrolenses Microlenses funnel light away from non-sensitive portions of pixelMicrolens layerColor filter layeryMetal opaque

5 LayerPhotodiodeSilicon substrateSilicon substrate ER FOSSUMP hotobitTECHNOLOGYP hotons to ElectronsTECHNOLOGYP hotons to Electrons Photon is absorbed by silicon and converted to an electron-hole pair. ypTypically the electrons (e -) are collected and holes discarded. Quantum efficiency (QE) is ratio of collected electrons to incident photons over whole PIXEL * and is always less than unity for visible light, ppyygand is wavelength dependent. Depends on: fill factorfill factor microlenses designti l h ttt vertical photosensor structure fundamental physics of silicon* some people just use the defined sensitive area to get a higher value but we consider this cheating.

6 One always needs to check how it is defined. ER FOSSUMP hotobitTECHNOLOGYQtEffi iTECHNOLOGYQ uantum EfficiencyAPS htffi ibl t CCDAPS has quantum efficiency comparable to Virtual PhaseCCDCMOS APSP hotogate-TypePhilips FT8FT 370 ILT CCD040050060070080090010001100 Wavelength (nm)KAI-370 ILT CCD ER FOSSUMW avelength (nm)PhotobitTECHNOLOGYCMOS A tiPilTECHNOLOGYCMOS ACTIVE PIXEL Each PIXEL has its own output amplifierPi lXYPhotodetector Pixels are X-Y addressed Key is low noise dti itActive amplifierreadout circuit Best of CCD detection/readout dCMOSA ctive amplifierRowSelectand CMOS integrationColumnOt tOutput ER FOSSUMP hotobitTECHNOLOGYE lectrons to VoltsTECHNOLOGYE lectrons to Volts Th CONVERSION GAIN dihfl The CONVERSION GAIN determines the amount of volts per electron for the pixels.

7 Typical value range is 1-10 V/e-Typical value range is 110 V/e Higher value is good but limits total signal that can be handled at 10 V/e -, 1 volt max = 100,000 electrons maxg ,, PIXEL amplifier should introduce minimum noise Noise is measured in equivalent number of electrons 200 V rms @ 10 V/e - = 20 e - rms noise ER FOSSUMP hotobitTECHNOLOGYV olts to BitsTECHNOLOGYV olts to Bits The ANALOG-TO-DIGITALBITSThe ANALOGTODIGITAL CONVERTER (ADC) converts the voltage from the pixels to a digital ADChANALOG SIGNAL2551 VADCREFBITS The ADC may have an ANALOG SIGNAL PROCESSOR (ASP) to reduce noise and provide additional gain before Vpg ADC has a reference voltage, 1 Volt, so that digital word is scaled against the fltAlfreference voltage.

8 A lower reference voltage is similar to adding gain. On-chip ADCs trade power for resolution Vpp(number of bits) 8 bits is enough for teleconferencing. 10-12 bit i d i d fdi it l till0000 V ER FOSSUM12 bits is desired for digital still tt BitTECHNOLOGYP hotons to BitsOPTICSMICRO-LENSQUANTUMEFFICIENCY(QE )CONVERSIONGAIN(uV/e-)ASPADCCOLORFILTERA RRAY(CFA)1/4F2 SFlIldEl tAlifi tiVlt(CFA)Sl tScenetoSensorFunnelsLightIncludesFill-F actor(no cheating)ElectronstoVoltsAmplificationan dNoiseSuppressionVoltstoBitsSelectsColor Suppression ER FOSSUMP hotobitTECHNOLOGYPtCt PiTECHNOLOGYPost-Capture ProcessingSd8b()fhil SENSOR produces 8b (or more)

9 For each PIXEL System eventually wants 24b (RGB) for each pixelColor interpolation required Color interpolation required Color balancing ( white balance ) required Additional processing used for aperture correction, processing used for aperture correction, etc. Communication is usually bandwidth limited Compression often required Interfaces need to have data format and output according to interface specificationI t ft ll /d t ftti dUSB NTSC Interface controller/data formatter required , USB, NTSC ER FOSSUMP hotobitTECHNOLOGYC olor InterpolationTECHNOLOGYC olor Interpolation Goal is to get best approximation for RGB at each PIXEL site Many possible approaches.

10 Have blue, need green & red G = average of 4 neighboring greens R = average of 4 neighboring reds Have green, need blue & red B = average of 2 neighboring bluesRaverage of 2 neighboring reds R = average of 2 neighboring reds Have red, need green & blue G=average of 4 neighboring greensG average of 4 neighboring greens B = average of 4 neighboring blues ER FOSSUMP hotobitTECHNOLOGYCl CtiTECHNOLOGYC olor Correction Need to correct colorsNeed to correct colors relative intensities (white balance) remove mixtures (off-diagonal elements) RrrrR'bgrNo perfect solution due to filter = BGbbbgggB'G'bbgrNo perfect solution due to filter overlaps and eye responseBest-fit by weights to standard co