Noise in TASS images

[Tom Droege]

Posted to CCD-world:
-+-+-+-
We have been looking at noise in tass images.  We have been looking mostly
at dark frames, but the problem is also seen in images of the sky.  All
data is from CCD442A devices which are TEC cooled.  

What I expected to see in a histogram of the pixel values was a gaussian
distribution with a tail to the high side - the leakier pixels.  What is
seen is a structure.  The structure has the form of regularly spaced peaks
decreasing in size at the larger values.  The peaks are farther apart at
high temperatures, and slowly merge together as the temperature is
decreased so that at -30 C the peaks merge and the expected distribution is
seen.

A short note on this by Michael Richmond can be seen at:

        http://a188-L009.rit.edu/tass/markiv/noise/peaks.html

A longer note for background is tass TN-51 which can be seen on the tass
home page:
http://www.tass-survey.org   Go to the technical area, go to technical
notes, pick TN-51. 

At first I thought this was a bit problem.  It is not.  It seems only
related to temperature.  Some but not all things I have eliminated:

1) Bit Problems
2) Cable cross talk
3) Interactions from various drive motors.
4) Interactions with TEC drive servo.  There are no temperature dependant
oscillations.   
5) ADC problems
6) Temperature dependant output amplifier oscillations. 

Note that there are dual cameras in the Mark IV system, with separate ADCs
for each channel.  The conversion data follows a common path to the
computer.  Data is acquired simultaneously on the V and I channel.  When
operated with the V channel at one temperature and the I channel at
another, the cooler camera has the closest spaced peaks.  Except for the
front end ADC, the data paths and the analysis paths are common.  One can
reverse the temperatures and the peak spacing reverses, thought I have only
done this indirectly.  

Why should the noise have preferred values?  It is beginning to look like
physics to me.  Is there any physics going on that could produce these
peaks?  This is a very solid problem which has persisted for several weeks
since first notices.  It occurs on 3 different 2k cameras.

Any comments or research direction would be appreciated.

Tom Droege


-+-+-+-
For information about CCD-world, send email to owner-CCD-world at cfht.hawaii.edu.