Once the reset anomaly fix has been implemented, another issue with subtracted CDS frames arises. This is low frequency banding . This banding is shown below in Figure 5.
In our system the bands are +/-20e p-v. As reported in previous papers this banding can be corrected using the average of a few columns for one quadrant to flatten the image. In Figure 5 we see that the peaks and valleys occur at exactly the same places in the four quadrants of the detector so the correction from one quadrant can be used to correct the other three quadrants also. In our system we synchronise the readout for four detectors and see low frequency banding on all four detectors. However the peaks and valleys occur at different places on the different detectors. Thus each detector must be corrected separately.
On an imager such as WFCAM, in a crowded star field, there may be many stars near the edge of the detector which would not then allow this edge to be used for the correction described. It may therefore be sensible to design an under scan region onto each detector. This could take the form of a simple mask which sits above the detector as part of a radiation shield and blanks off a few columns of one quadrant of each detector on a mosaic.
Was this article helpful?