CCD linearity measurement by incremental binning (Conference Presentation)

The traditional method for measuring CCD non-linearity using constant flux and variable exposure time is compared to a faster and potentially more accurate method that requires just two exposures at low intensity. Signal is varied by parallel-binning a successively greater number of lines. The binned image is compared to a conventional image with identical illumination, co-added digitally in the same pattern. This method requires no shutter calibration and is insensitive to illumination drift or charge transfer inefficiency. With some additional software effort, it can be arranged to tolerate illumination non-uniformity. We present data obtained by both methods for the 64 CD231-C6 outputs and discuss automated performance optimization. We find that the widely held assumption that non-linearity only affects high signals is incorrect when optimizing for minimum gain change over the full signal range, in which case slope increases with signal initially, then peaks in the mid-range before dropping again.