Old detector

The zero level pixels in the engineering grade array, first believed to be pixels dying because of repeated thermal cycling, increased in number gradually from Jan-2003 through 2004 from 1% to around 5%, but suddenly in 2005 jumped to $\sim$30%. However, it was found that the overal signal level had drifted $\sim$4000 ADUs towards zero, causing a large increase of apparent zero level pixels. Only slight modifications to the bias voltages of the detector were needed to restore the original reset level. Since May-June 2005 the number of zero pixels on the engineering grade array is again around 1%. The reason for the varying reset level is believed to be some drift in the array itself.