) : the procedure is exactly the same as for SW,
only, since 
= 0, the
curves merge together when TO 
FROM, and they are
horizontal
when 
. If FROM = 30 ADUs and TO = 10 ADUs :
= 36. If FROM = 30 ADUs and TO = 100 ADUs : = 28.
The experimental numbers are within a factor of 2 of this fit.
On Figure , it can be seen that after readouts the image in some cases is still somewhat
distorted.
Stabilization after changes of flux or integration time is the main problem encountered with this array. The worst situations occur in two cases, a) illuminating the array previously at dark, or b) after a saturating flux.
The first problem can be overcome by keeping always light on the array and sorting the sequences of observations by decreasing fluxes. Dark calibration must be done rarely, and must be preferentially placed at the end of the observing sequence.
After saturation, the array keeps memory of the illuminating scene for a period as long as half an hour. The only way we found to avoid polluting the following observations with the memory image is to saturate the whole array with a uniform illumination. With an ICD flash of a couple of readouts, the array looses memory of the previous scene, and is ready to image a new field. The drawback of this method is that after the flash, the overall responsivity of the array will be below the average value for about 10 minutes. This effect will decrease the absolute photometric accuracy, but will keep a reasonable relative photometric accuracy inside an image.