1. Miss Pernelle Bernardi
COROT WEEK Liège 4-7/12/2002.
(working on Saturday!!!)
The CCD flight models
Miss Pernelle Bernardi
Mr Vincent Lapeyrere
Mr Tristan Buey
and the CCD team at work
From Meudon : Régis Schmidt, Bertrand le Ruyet, Jêrome Parisot, Didier Tiphène
From CNES : Olivier Gilard, Guy Rolland (for irradiation test)
Funding by CNES.
CorotWeek 3, Liège 4-7/12/2002

2. What’s up?? PAST. Test of electrical model on 4 chips are finished:
---> Constraints for flight electronic readout and processing.
Irradiation test are finished:
---> Performances for the end of life.
Present and future.
Test and calibration of the 10 CCDs flight models:
---> Selection for the flight camera.
---> Delivery of the CCDs in March 2003.
Test of the CCD mechanical models:
---> Delivery for the MIQ Camera in December 2003.
Bench dedicated to scientist for photometric experiment:
---> From May 2003.
CorotWeek 3, Liège 4-7/12/2002

3. Results on the flight models
2 CCDs flight models have already been tested on the bench.
A third one is under test at that time.
We test one CCD every 3 weeks.
High homogenety between the CCDs (7 EM and 2 FM).
We have about 20Go of raw images per CCD and 1Go of reduced images.
2 steps to reach the performances of the CCDs:
* Measurements of the characteristics on the bench.
Working point, Dark current, PRNU, Gain, Coeff Temp, Saturation…
* Software models (expected system performances and CCD characteristics).
CCD information
????
Happiest flying mascot!!!
CorotWeek 3, Liège 4-7/12/2002

4. Working point (1) Optimize 3 bias voltages VOD, VRD, VOG
Measurement of the video signal for different values of the bias voltages:
Same working range for the 3 CCDs FM:
CCD N°
VOD
VRD
VOG
056
26.5  0.2 V
13.2  0.2 V
-1.9  0.2 V
071
076
CorotWeek 3, Liège 4-7/12/2002

5. Sensibility is given in e-/mv.
Working point (2)
Sensitivity of the video signal to the bias voltages : static measurement, in dynamic it will depend on the frequency and some compensations exist.
About same order at worst frequency (100kHz, readout frequency).
Higher sensitivities = 4e-/mV
Specification for the electronics: 1mV peak to peak ---> equivalent noise of few e-.
Sensibility is given in e-/mv.
CCD N°
VOD
VRD
VOG
056
2.6
1.0
1.9
071
3.9
1.8
2.0
076
3.7
1.2
2.5
CorotWeek 3, Liège 4-7/12/2002

6. Histogram of the mean value of 32*32 pixels windows over the CCD
Dark at –40°C
Dark mean value: Spec: < 0.5e-/px/s at –40°C
No cosmetic (all pixels < 100e-/px/s)
CCD N°
Mean value (e-/px/s)
Local max (e-/px/s)
80% of windows<
056
0.11
0.23
0.13
076
0.42
Histogram of the mean value of 32*32 pixels windows over the CCD
Mean value
80% of windows
CorotWeek 3, Liège 4-7/12/2002

7. PRNU (Pixel response Non Uniformity) (1)
Obtained with flat illumination
l= 420nm, Dl = 10nm
l= 700nm, Dl = 10nm
l= 950nm, Dl = 10nm
Give the uniformity of the
thickness
---> Fringing Pattern
Give the uniformity
of the state surface and AR coating ---> Surface Pattern
Give the uniformity of the
physical characteristics
---> High homogeneity
CorotWeek 3, Liège 4-7/12/2002

8. PRNU (Pixel response Non Uniformity) (2)
64*64 windows of 32*32 pixels (no side effects by removing pixels).
In each window (i,j), we calculate
- the mean value mi,j
- the standard deviation si,j
Global PRNU = dispersion of the mean values of the windows, standard deviation of the mi,j.
The Global PRNU is not a relevant parameter.
Local PRNU = dispersion of the pixels values inside each window, mean value of the si,j.
CorotWeek 3, Liège 4-7/12/2002

9. CCD Gain (µV/e-) Results at –40°C:
The CCD is illuminated with white LEDs at several time flashes
Calculation of the mean value: m
the variance of the difference of 2 images with the same flash: s
GT is the gain of the complete video chain (e-/ADU)
1 ADU of the ADC = 76.3mV
Results at –40°C:
CCD N°
Left side
Right side
056
4.33
4.28
076
3.97
4.15
CorotWeek 3, Liège 4-7/12/2002

10. CCD Gain versus temperature
Global gain (e-/ADU) measured at different temperatures from –45°C to 30°C
The electronics gain is constant  the variation of the global gain is the same than the variation of CCD gain.
CCD N°
Temperature coefficient of the CCD Gain (ppm/K)
056
-1100
076
-900
CorotWeek 3, Liège 4-7/12/2002

11. Full Well Capacity in Flat Field
Image pixel Saturation
CCD N°
FWC (ke-)
056 87
076
110
E2V data: 2 groups of CCDs
- 3 CCDs with a FWC < 90ke-
- 7 CCDs with a FWC > 100ke-
CorotWeek 3, Liège 4-7/12/2002

12. Full Well Capacity with a PSF
2 sizes of PSF
3 positions on the CCD
Right output
Left output
F = 20 pixels
First results:
* Saturation occurs at lower values when the PSF is far from the output.
* Depends on the PSF size.
Traps? Results soon… in CorotWeek4
F = 50 pixels
CorotWeek 3, Liège 4-7/12/2002

13. Temperature Coefficient of Quantum Efficiency
Measure of the flux at:
- different wavelengths from 400nm to 950nm
- different temperatures from –45°C to –30°C
Correction by the temperature coefficient of the CCD gain (aG ~ -1000ppm/K)
aR = temperature coefficient of the CCD response
aG = temperature coefficient of the CCD gain
aQ = temperature coefficient of the quantum efficiency
But the temperature coefficient of the CCD response is about 1500ppm/K
CorotWeek 3, Liège 4-7/12/2002

14. Results of the irradiation tests (1).
3 chips tested (4210) with protons at 4 different energies (30 to 60Mev).
Tested characteristics :
Working point.
Gain and Full Well Capacity.
Dark Current (mean value, defect).
Pixel Response Non Uniformity.
No transient were visible due to the too high proton flux (more than one impact on a single pixel on each image).
CorotWeek 3, Liège 4-7/12/2002

15. Results of the irradiation tests (2).
Unless the dark current all the other characteristics of the CCD will not be degraded enough to impact on the system performances.
Irradiation will induce in the dark current :
Local defect (spatial and temporal) Global evolution
<0.5e-/s ---> 3 to 10e-/s
Rapid evolution with the temperature gives strong constraint on the CCD operational temperature
---> Under -40°C.
No possible recovering on board… Aïe!!
Defect are less than 1/10000 pixels.
CorotWeek 3, Liège 4-7/12/2002

16. Selection of the flight models
The aim of this work is to develop methods to :
Determine if all CCDs are able to fly
Find criteria for each program
By software models using :
The expected performances of the system (PSF, pointing, thermal stability…).
The characteristics of the CCDs measured on the bench and also at E2V.
4 CCDs have to be selected, the parameters of choice have to be optimized for the asteroseismology et exoplanets.
Could avoid random choice if more than 4 CCDs are still alive after the test campaign!!!
CorotWeek 3, Liège 4-7/12/2002

17. CCD parameters Shouldn’t be used : Should be used for sorting :
Dark current
Noise
White pixels
Pixel response non uniformity
Cosmetic defects
Function of the wavelength
Gain
Function of the temperature
Quantum efficiency
Number of electron collected
Function of temperature
Pixels capacity
Saturation
Shouldn’t be used :
Non linearity
Dominated by the electronic
Transfer efficiency
good
Read out noise
Irradiation sensitivity
Not predictable
CorotWeek 3, Liège 4-7/12/2002

18. Value of the dark current
Map of the dark noise
Dark noise is due to the poissonian statistic of the dark current :
Value of the dark current
Psf
Mask for photometry
Here each pixel indicates the max value in the 32*32 pixels windows
We obtain an image, each pixel indicates the dark noise at that position
CorotWeek 3, Liège 4-7/12/2002

19. Map of the jitter noise
3 different flat fields for 3 different parts of the bandwidth
Each flat field is convolved with the corresponding PSF
The jitter noise is calculated with the sum of the 3 images
Flat fields
Convolution with the PSF (exo or seismo) nm
Computation of the jitter noise
sum nm
Each pixel indicates the max value of the jitter noise inside the 32*32 pixels windows nm
CorotWeek 3, Liège 4-7/12/2002

20. Jitter and dark noise On the curve we plot :
Photon noise versus the star magnitude.
Noise specification.
Average Value of the jitter and dark noises.
Jitter noise
Average ~ 7.5 ppm
Noise maps computed with seismo psf
Unit : ppm
Dark noise
Average ~ 11 e-
Average value calculated over all the CCD surface
Photon noise
Specification (1/10 photon noise)
Dark noise
Jitter noise
CorotWeek 3, Liège 4-7/12/2002
Unit : e-

21. Jitter and dark noise
Jitter noise is greater than the specification on a jitter defect
Jitter noise
Average ~ 7.5 ppm
Noise maps computed with seismo psf
Unit : ppm
Dark noise
Average ~ 11 e-
Jitter defect
Photon noise
Specification (1/10 photon noise)
Dark noise
Jitter noise
CorotWeek 3, Liège 4-7/12/2002
Unit : e-

22. Other parameters
Variation of global CCD response with temperature (Gain and quantum efficiency)
Coefficient ~ 1500 ppm/K
=> Variation of about 10 ppm (with T=5.10-3K)
Quantum efficiency
Values (E2V data):
Pixel capacity
Values (E2V data)
CorotWeek 3, Liège 4-7/12/2002

23. Parameters priority For exoplanets : For seismology :
Dark noise (strong defects, will evolve with radiations!)
Jitter noise map (strong defects)
Quantum efficiency
Pixel capacity
Temperature coefficient
For seismology :
Pixel capacity
Quantum efficiency
Temperature coefficient
Jitter noise map
CorotWeek 3, Liège 4-7/12/2002

24. CCD quality All important parameters on the same diagram
Compare CCDs between them
=> Example of 2 CCDs (seismology PSF) :
Best CCD
CorotWeek 3, Liège 4-7/12/2002

25. Conclusion Methods are developed to sort the CCDs Now :
Applied these methods on the bench data
Choose 2 CCDs for each scientific program
CorotWeek 3, Liège 4-7/12/2002