A set of 10 full frame bias images was collected (images 349889.fits to 349909.fits, see Data Collected). The LED light source was disconnected while these images were made since there is some fear that the LED may be emitting a small amount of light even while "turned off" and there is no shutter in the test system.
The bias images were median stacked and the resulting image is shown below:
As in Noise Test, note the overscan is split into two at the boundary between the 20 on-chip, overscan pixels and 20 "true" overscan pixels (obtained by reading beyond the physical end of the CCD), the last pixel in the on-chip overscan pixels reads anomolously high and should not be used in data reduction.
The same hot column and major trap are observed as in Low Level Flats (indicating that the traps also contains hot pixels). Two additional hot pixels are noted at (635,1975) and (637,1975) - the fact that these were not found in the low level flats suggests that they are stable enough to subtract out with a bias image. The hot pixel at lower right does not subtract out.
There is a slight slope in the bias image (not present in the overscan), for which we have no certain explanation, although it may arise from the release of deferred charge during readout. We recommend that bias images be used for bias subtraction in astronomical observations, and not simple overscan calculations, except in calculating a scale factor for a bias image.
The mean bias level is 1516.5 ADU. No coherent external interference
signals are present in a single bias frames.