The detector saturates at different count levels depending on its bias voltage. You should try to keep your data values below the 1% linearity deviation for your selected bias. Using a single fixed bias (as done with the old NICMOS array) is inappropriate because of non- linearities and bad pixel artifacts in the SBRC 256 x 256 InSb arrays. Larger biases allow more electrons to be collected before saturation, but they also cause increased numbers of high dark current pixels and low-level non-linearities. Fortunately, these effects mostly disappear when object and sky frames are subtracted. The table below gives ADU/sec that keep nonlinear deviations below 1%. For more details see the CSHELL Manual Appendix F.
VDET Bias (mV) ADU/sec -3.4 -300 2196 -3.3 -400 2323 -3.2 -500 2405 -3.1 -600 2448 -2.9 -800 9911 -2.7 -1000 10861
LAST UPDATE: February 1, 2010