Author: Tom Droege Date: 960806 Revision: #1 960806 Key Words: CCD, instrumentation
I took a few measurements on triplet #3. Full system noise level was 26, 28, and 30 e- for the three cameras. By full system noise, I mean that I cool the camera (in this case to -15 C) and run it long enough in drift scan mode so the dark current is stable. This is then a "practical" noise number and includes electronic noise, dark current contribution, and the noise from the ccd amplifier in the chip(Read Noise). To test the electronic noise, I unplugged the camera head lead. This gives a noise of 5 e-. At -15 C, the KAF-0400 noise should be about .1 e-/sec or 47 e- for the 470 sec exposure from the drift scan operation.
So:
(Total noise)^2 = Dark Electrons + (Electronic Noise)^2 +
(Read Noise)^2
28^2=47+25+(Read Noise)^2
(Read Noise0^2 = 712
Read Noise = 27 e-
Kodak lists 13 e- typical and 20 e- max for the Read Noise.
One possible reason for the higher number could be light leaks from running the camera on the bench. I have had some problem with this in the past.
In any case, it does not amount to a hill of beans.
In drift scan operation, the dominant noise contribution is the sky background. For me this is typically 20,000 e-. This just swamps any of the noise contributions above. I estimate that 20,000 e- represents mag 18.5 sky. To start being a significant contribution the sky electrons would have to be similar to the Read Noise of 30 e- or 900 electrons. This would require a sky of mag 21.8 per square arc second.
I don't think any of us will have to worry that anything but the sky background is important.
This also shows that cooling is not so important. To get 20,000 e- from dark current one would have to operate at about 35 C. So why cool at all? Well, we could probably run uncooled. The problem is, that there are "hot" pixels. These are pixels that do not obey the double every 5 C rule. At some temperature they turn on and leak a lot of current. This shows up as the room temperature streaks that one sees when doing the test sequence in the note. I find that most of these turn off by about 0 C. I think there is not much reason to push the temperature below this value. There is good reason, however to run at constant temperature since the output offset changes with temperature, and this looks just like a leakage current to the electronics.