|The CCDs used in COROT mission are large format chips (4280 from Marconi Applied Technologies, former EEV) with the main characteristics (typical number) :
For mode details : DataSheet.
Different CCDs will be buy for different uses :
Each CCDs will have a serial number given by the manufacture and we will give them a name (a data base make the link between the name and the serial number). This name will be choose in the great list of the Egyptian gods and divinities, that will be good for our culture !!
This is the list of the main characteristics measured on the test bench at Meudon .
It's the bias voltage and clock level situated at the center of the useful range (values that allow a well readout).
The whole bias are determinated with the following criteria's :
This is the measure of the number of e- per ADU converter, knowing the electronic gain we can then calculated the CCD gain (µV/e-).
The gain is measured on the two output as a function of the temperature, about 3 to 4 measurements from 20°C to -40°C.
We plot the curve varaince=f(mean), the PENTE given by the linear part of this curve is the inverse of the gain multiplied by two. For low light flux the readout noise is dominated and for high flux the pixel is saturated (definition of the Full Well Capacity when the curve became non linear), between the two the curve is linear.
It's the measured signal in absence of any light on the CCD, it is a function of the temperature et linear with the exposure time (contribution of the image zone). The readout process will also contribute to the dark current with a different law in function of the temperature and the readout type (windowed, binning...). This different contributions have to be measured and compared.
The curves given the dark current in function of the temperature are obtained by removing the different cosmetic defects. With an AIMO CCD the dark current is very low (less than 0.1e-/s at -40°C) and so the exposure time is very long (tens of minutes) to achieve accuracy measurement. A very high efficiency protection against stray light is needed. Some over measurements will be done to study the defects.
The average dark current is given by the law :
dark(T) = dark(293K) 1.14E6 T3 e-9080/T
where T is the temperature. Valid for the "normal" pixel during the exposure within the pixel (all line phases at low level).
When the charge transfer the CCD is no more operating in AIMO mode and then the law for the dark current is given by :
dark(T) = dark(293K) 1.14E6 T3 e-9080/T
where T is the temperature.
For the dark defects (with pixels) they will follow different laws in temperature with greater variation than the "normal" ones (with the same law as for the non AIMO but with different coefficients).
To know the CCD quality it also needs the dispersion of the dark current over the surface : the Dark Signal Non Uniformity (DSNU). It's measured over 32*32 pixels windows, the mean value and standard-deviation are calculated.
It is the correspondence between the number of photoelectrons and the CCD output signal amplitude. For our application this measure is not of great importance, but we need to know the differential linearity between the pixels.
We used a defocalized spot to measure the linearity over a large number of pixel or with flash illumination over the whole CCD during the exposure.
This is the measurement of the quantum efficiency variation as a function of the temperature. We calculated the PRNU relative variation over the whole CCD (quantum efficiency) and the PRNU local variation (modification of the non-uniformity). It is a function of the wavelength.
To obtain it the CCD is illuminated in flat-field over the whole optical bandwidth for different temperatures.
It is the measure of the quantum efficiency dispersion over the pixels. It is a function of the wavelength and the temperature.
In our application we need to know the PRNU over small region on the CCD, and calculated it over 32*32 pixels windows (4096 windows on the CCD). For each windows we calculate the mean value of the pixel and the standard deviation.
The local PRNU is defined as the percent of the windows with a given value of standard deviation. The value given in the different documents corresponds to 80% of the windows having a lower value.
The global PRNU is defined as the dispersion of all the mean value of the 4096 windows.
Remark : with the illumination non uniformity the value of the global PRNU is over estimated in comparison with the real value. For the local PRNU we take care that this non uniformity is enough small.
It is the ration between the number of photons arriving on the CCD and the collected photoelectrons, it's function of the wavelength (and also of the temperature).
On our bench we do not measure this parameter, but we will compare the CCDs together for 6-8 different wavelengths and in white light.
The precision of the measurement can only be around 30% (du to the bench configuration).
It is the measurement of the charge transfer efficiency between the line (vertical CTE) and the pixel of the register (horizontal CTE). It's expressed as the percent of the charge transferred between 2 lines or 2 pixels : typically the value is 0.99999 ....
The method used to measure it is the analysis of the pixel and line overscan within the non transferred charge will be present.
We study them with the overscan lines (charges can be relapsed within) and with a special transfer in lines (the charges go forward and backward).
For each CCD the noise of the whole chain : CCD + Electronics is measured. This noise expressed in ADU or µV can be calculated with the gain.
The noise of the readout electronics is measured independently with specific simulator.