This short report describes the results of my analysis of a few short dark frames taken after the Stargrasp team made repairs to the HDI dewar in November, 2013.
On UT 2013 November 22, at about 18:30, I took five 1-second dark images; the data are c6618, indices 2-6. At that time, the CCD temperature was about -116 Celsius, and the camera pressure about 50 microtorr. All the data were taken in 4-amp readout mode.
I re-formatted the raw HDI images into 4 individual quadrant files per image, but otherwise did not modify the pixel values in any way. I defined 3 boxes, each 100x100 pixels, centered at (200, 200), (700, 700) and (1200, 1200). I then chose all possible pairs of images in each quadrant. For each pair, I
The result is a set of 27 measurements (9 pairs times 3 boxes per pair) of the statistics of differences between two dark frames. We can convert the sigma of the gaussian to a measure of the readout noise, using a value for the gain measured a few weeks ago:
1 1.3 e-
readout noise (e-) = sigma (ADU) * ------- * --------
sqrt(2) ADU
We find
quadrant avg sigma readout noise readout noise
(ADU) (ADU) (e-)
---------------------------------------------------------------
1 8.35 5.9 7.7
2 7.76 5.5 7.1
3 8.17 5.8 7.5
4 7.37 5.2 6.8
---------------------------------------------------------------
These values are similar to those measured on UT Oct 19, as described in Tech Note 3.
Peter Onaka's notes titled November 2013 trip log (vacuum repair) state
15. Took test bias, dark and object exposures. The noise is back down to 4 to 5 adus.
My values are slightly higher, but that might be due to differences in the manner of calculation.