[Comm2011] SA107

[Konrad Kuijken]

Hi all,

I have been playing with the cross-talk, trying to understand it and  
see whether we can get rid of it in software.
Results are very encouraging! I attach a few plots, based on a random  
exposure (OMEGA.2011-04-06T08:12:05.192.fits) .

-------------- next part --------------
A non-text attachment was scrubbed...
Name: xtalkpanels.gif
Type: image/gif
Size: 22210 bytes
Desc: not available
Url : http://listman.astro-wise.org/pipermail/comm2011/attachments/20110526/b9f8b6a5/attachment-0004.gif 
-------------- next part --------------


In the panels plot I have plotted the flux on each CCD vs the flux of  
the brighter pixels on another. Horizontal scale from 0-65535 ADU,  
vertical scale is centered on the median ADU count +/- 300. From left  
to right, CCD 93 vs 94,95,96; CCD94 vs 93,95,96; CCD95 vs 93,94,96;  
and CCD96 vs 93,94,95. (the double sequence on ccd93 is due to the  
shadow of the filter in the B-filter exposure). Particularly the last  
three panels are interesting. The slope shows that on CCD96 also non- 
saturated pixels on other CCDs have an effect. You can see those as  
'fake' positive and negative sources on CCD96.
The linear relationship is encouraging, it is easy to fit, and to  
correct away. Worst amplitude I've seen is -200ADU response on CCD96  
for a 60000 signal on #95.
A complication is that crosstalk due to saturated pixels does not  
follow the same relation. eg the bottom left panel (effect of CCD96 on  
CCD 95) shows negligible slope, but saturated pixels cause a ghost of  
about 100ADU amplitude. So the effect of saturated pixels has to be  
fitted separately, but it is again a 1-parameter fit (an ADU offset).

An example of correcting this way for the saturated and unsaturated  
crosstalk is in the three panels below, each time the same pixels on  
ccd 94,5,6. Blinks between before and after. Particularly no.96 is  
peppered with ghosts, and they disappear very nicely!
There is one issue remaining, (can be seen well in the three ghosts on  
ccd95): the saturated ghosts leave a single-pixel wide 'ring'. How to  
deal with this I am still playing with. But this seems very  
encouraging already. It would be a real pain to have to deal with the  
unsaturated ghosts in particular: they look just like faint stars, and  
are not easily dithered away.
It will be interesting to see how variable these crosstalk levels are  
when measured in this way.
More later!
K

-------------- next part --------------
A non-text attachment was scrubbed...
Name: x94.gif
Type: image/gif
Size: 647724 bytes
Desc: not available
Url : http://listman.astro-wise.org/pipermail/comm2011/attachments/20110526/b9f8b6a5/attachment-0005.gif 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: x95.gif
Type: image/gif
Size: 587785 bytes
Desc: not available
Url : http://listman.astro-wise.org/pipermail/comm2011/attachments/20110526/b9f8b6a5/attachment-0006.gif 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: x96.gif
Type: image/gif
Size: 587593 bytes
Desc: not available
Url : http://listman.astro-wise.org/pipermail/comm2011/attachments/20110526/b9f8b6a5/attachment-0007.gif 
-------------- next part --------------