The performance of the array can be checked at the telescope by recording calibration frames. The read noise and dark current are determined by setting all wheels to their blank position. For the read noise measurement first record an average dark frame by typing:
CASPIR/DARK/TIME=0.3/CYCLES=100/METHOD=2
then record another dark frame with only one cycle:
CASPIR/DARK/TIME=0.3/CYCLES=1/METHOD=2,
subtract the average dark frame from this frame and determine the
standard deviation of the pixel values. This is the read noise for
two reads of the array (a difference of the end of the integration
ramp and the reset voltage). The single-read read noise is the
standard deviation of this frame divided by 
2 and multipled by
9 e 
/ADU. Values of 
40 e are expected.
The dark current measurement is complicated by the long settling time (a few seconds) of the array after resetting. This means that it is not meaningful to subtract a short exposure dark frame from a long exposure dark frame to determine the dark current. To measure the dark current, record two long exposure dark frames of different duration with one cycle each, e.g.:
CASPIR/DARK/TIME=100/CYCLES=1/METHOD=2
CASPIR/DARK/TIME=50/CYCLES=1/METHOD=2,
subtract the two frames and determine the mean pixel value. The dark
current is the mean pixel value divided by the integration time
difference between the frames and multipled by 9 e /ADU.
Mean values of 13 e /s/pixel are expected from this
measurement, but significant numbers of pixels have dark currents in
excess of 50 e /s/pixel, as shown in Figure 3.
For the 5 s integration time used in imaging observations, the average
dark current is 30 e /s/pixel. For the longer integration
times typical of spectroscopic observations, this value corresponds to
the 90th percentile of the cummulative dark current distribution.
Figure 3: Dark current as a function of time since pixel reset. The
plotted points are average dark current measurements formed by
differencing dark exposures with durations indicated by the extent
of the horizontal error bars. The vertical error bars indicate the
standard deviation of dark current values across the array. The
solid line is the 90th percentile of the cummulative dark current
distribution for each measurement (i.e., 90% of pixels have a dark
current lower than this value).
The signal strength can be checked by recording images of the standard
stars listed in Appendix G. Measure the total
sky-subtracted signal in the stellar image and convert to instrumental
magnitudes using 
. Correct for
extinction using the mean extinction corrections listed in Appendix
C and form the zenith zero point offsets for each
filter, defined to be 
. These can be compared
with the typical values listed in Appendix C.
Typical sky brightness figures are also listed in Appendix C. These can be checked using a dark-subtracted sky frame and the zero point offsets determined above, or the zero point offsets listed in Appendix C.