CCD Fundamentals（CCD基础）.ppt

1、CCD Fundamentals,Shanghai, June 11, 2007,Imaging CCD,Spectroscopic CCD,Fill the bucket with light!,Fill the bucket with light!,Convert Photons into Electrons MOS Photodetector (Metal-Oxide-Semiconductor),Electrical Charge,What to do with the captured Electrons?,Electrical Charge is shifted,Imaging A

2、rea,Serial Register,Shift them,Electrical Charge is shifted,Output Node,How the Electrons are moved within CCD?Each Pixel has Phases,Charge Transfer,Pixel 1,Pixel 2,Direction of Charge transfer,Basic CCD Array,Channel Stop,Control Electrodes,Drive Pulse Connections,Image Section,Output,Readout Secti

3、on (Serial register),Drive Pulse Connections,Preamplifier Structure,Typical CCD Structure,Preamplifier,Output Node,Active Array (Imaging Area),Full Frame CCD,Serial Register,Serial Register,Preamplifier,Output Node,Active Array,Serial Register,Preamplifier,Output Node,Active Array,Front-Illuminated

4、CCD Polysilicon Gates Absorb Light,Incoming Light,Potential Well,Silicon,Silicon Dioxide,Polysilicon Gate,Electrical Connection,QE for Front-illuminated CCD,Incoming Light,Potential Well,Silicon,Silicon Dioxide,Polysilicon Gate,Electrical Connection,Back-illumination offers higher QE,QE for Back-ill

5、uminated CCD,Enhancing UV Sensitivity with Proprietary CCD Coatings Unichrome (Phosphor composition),What is the Full Well?,Single Pixel of Serial register,Output Node,Single Pixel of Imaging Area,Full Well defines how many electrons Single pixel can store,How much is the Full Well?,Pixel of Imaging

6、 Area Typically 100-250 ke- Pixel of Serial Register Typically 250-500 ke- Output Mode Typically 600-1000 ke- Why are they different? Because of Binning,Higher Signal to Noise Ratio Faster Readout,All at the expense of Spatial and Spectral Resolution!,BINNING,Serial Register,Preamplifier,Output Node

7、,Active Array,Binning,How Binning works (2 x 2),Serial Register,Preamplifier,Output Node,Active Array,How Binning works,Serial Register,Preamplifier,Output Node,Active Array,How Binning works,Serial Register,Preamplifier,Output Node,Active Array,How Binning works,Serial Register,Preamplifier,Output

8、Node,Active Array,How Binning works,Where the electrons go from the Output Node?,Electrons are converted into V signal,It is difficult to count Electrons,V is analog signal it has to be converted into digital signal,Therefore it is sent to the Analog-to-Digital Converter (ADC),Serial Register,Preamp

9、lifier,Output Node,Active Array,From Output Node,ADC,Serial Register,Preamplifier,Output Node,Active Array,To ADC,Dynamic Range of ADC,Lets say we collected 250,000 e- (Full Well),Ideally we would like to have 250,000 gradations of digital signal,New CCDs have 16-bit Dynamic range 16 bit = 216 = 65,

10、536 counts (ADU),But digital electronics capability is limited,Older CCDs used to have Dynamic range of 12-bit 12 bit = 212 = 4,048 counts (ADU),What is the electron Gain?,Ok, we collected 250,000 e-,With 12-bit ADC we have 4,048 gradations of signal,It means wed use 250,000/4,048 = 62e- per each di

11、gital count (electron Gain = 62e-/count) It also means we cant distinguish between 1.62 e-,With 16-bit ADC we have 65,536 gradations of signal,It means wed use 250,000/65,536 = 4e- per digital count (electron Gain = 4e-/count) It also means we cant distinguish between 1.4 e-,PI-Acton CCDs have 3 ele

12、ctron Gains,High (4 e-/count) Medium (2 e-/count) Low (1 e-/count),Which Gain to use?,What if?,ADC over-filling,ADC under-filling,Serial Register,Preamplifier for high signals,Larger Output Node,Active Array,We can handle low signals. What about high signals?,Output Node,Preamplifier for low signals

13、,Dual output CCD,What is the advantage of Double output?,Smaller Output Node capacity = 250,000 e- Can handle Gains = 1, 2, 4 e-/count Larger Output Node capacity = 1,000,000 e- Can handle Gains = 4, 8, 16 e-/count Together they can handle very wide range of signals,Why dont use larger Output Node w

14、ithall different Gains?,In principal, its possible to use 1,000,000 e- Node with all 1, 2, 4, 8, 16 e-/count Gains,Problem with large Node it generates higher Read Noise,Lower signals can not tolerate high Read Noise,The Fundamental problem of NOISE!,In ideal world,Electrical Charge is shifted,Imagi

15、ng Area,Serial Register,In reality,Noise,Noise = ( Ns2 + Nr2 + Nd2 ) 1/2 Ns2 = Photon Noise e- Nr = Read Noise RMS e- Nd = Dark Charge e-,Photon Noise,Statistical variation in incident photon number Governed by Laws of physics Can not be reduced,Photon Noise = Signal,Signal = 1 e- Photon Noise = 1 e

16、-,Signal = 10,000 e- Photon Noise = 100 e-,Read Noise(Preamplifier Noise),Produced while converting electron charge into ADU Read Noise = consts (depends on particular electronic circuits) Can be reduced by optimizing read-out electronics,Typical Read Noise for our detectors = 2.5-20 e-/pixel Read N

17、oise occurs only once for all Binned pixels,Dark Noise,Charge Thermally generated within silicon layer of CCD Can be reduced by cooling CCD,Cooling alternatives: - Thermo-Electric (TE) -80.-100 C Liquid Nitrogen (LN) -120 C,Anatomy of TE-Cooled CCD Camera,Mount to Spectrograph,Shutter,Vacuum Window,

18、CCD Device,Peltier Stack,Preamp Board,Noise,Noise = ( Ns2 + Nr2 + Nd2 ) 1/2 Ns2 = Photon Noise e- Nr = Read Noise RMS e- Nd = Dark Charge e-,Signal to Noise Ratio =,S,N,To be honest with you,Noise level by itself does not mean much,Noise level only makes sense when its compared with the level of useful Signal,S/N Ratio is the fundamental measure of Spectral Quality,Signal,Signal = I * QE * T,Signal electrons produced on CCD by conversion F