A set of 9 full-frame bias images were obtained. Basic data are shown below - the first column is the image name, the second is the image label (as given by the FITS card OBJECT), the third is the exposure time in seconds (FITS card INTTIME) and the last two are the image dimensions in pixels (NAXIS1 and NAXIS2). The value in FITS card INTTIME is used as the exposure integration time in all following discussion.
314227b.fits Loral3 test 1.000 2088 2048
314228b.fits Loral3 test 1.000 2088 2048
314229b.fits Loral3 test 1.000 2088 2048
314230b.fits Loral3 test 1.000 2088 2048
314231b.fits Loral3 test 1.000 2088 2048
314232b.fits Loral3 test 1.000 2088 2048
314233b.fits Loral3 test 1.000 2088 2048
314234b.fits Loral3 test 1.000 2088 2048
314235b.fits Loral3 test 1.000 2088 2048
A median stack of the 9 bias images was obtained. The mean row of this image (Figure 1, 2) is characterised by an initially normal column, followed by a high then a low column and then a 100 column wide burst of coherent interference signal with a period of 2 columns. This is superimposed on a second interference pattern with a spatial period of 4 columns which extends across the entire image. This latter interference pattern is present in each of the raw bias frames with an amplitude as given in Table 2. The pattern can switch in phase by 180 degrees between frames, apparently randomly.
In the median stacked bias image, column number 2049 is at the normal bias level but it is followed by two columns that are above the normal level and then the rest of the overscan which falls below the bias level and appears to rise exponentially to that level, although there is not sufficient overscan to be certain of this. Note that this structure is not consistent with a true bias overscan (where the readout is clocked identically in the overscan as in the imaging area of the CCD), or a device with no prescan pixels.
The mean of the mean row away from the anomolous ends of the row is
ADU. The amplitude of the interference signal of
wavelength 4 pixels is 0.35 ADU (median stacked image).
The mean column (Figure 2 of the median stack of bias
images is characterised by an initial high row (by a few ADU) followed
by another row which is another 1 ADU higher and then an exponential
decline over 
columns to the nominal bias level. In the rest
of the column there is a low-spatial-frequency wave (wavelength 
rows) of amplitude 
ADU. This wave appears to reflect
similar variations (of generally larger amplitude) in the raw bias
frames as listed in Table 2.
There is no evidence for a light leak during readout.
Figure 3 is a 2-d amplitide spectrum of the median bias stack, note the complicated structure in the horizontal direction but almost complete uniformity in the vertical direction. Figure 4 is the mean row FFT of the median bias stack. Note the complex, but low-amplitude structure.
Figure 5 is the histogram of the single bias image 314227b.fits, columns (100:2000). The distribution appears to be reasonably Gaussian.