Jump to content


short exposures on DSI III have a severe response non-uniformity

  • Please log in to reply
3 replies to this topic

#1 rbelikov



  • Members
  • PipPip
  • 10 posts

Posted 29 January 2012 - 03:19 PM

I have a monochrome DSI III and whenever I take a uniformly illuminated flat
field with a short exposure (e.g. 0.0001s) , I see a huge gradient from top to
bottom. The top of the image is ~20K counts and the bottom is ~50K counts, with
a smooth transition from top to bottom. I know it's the CCD that's non-uniform
rather than the field itself because when I rotate or shift the CCD, the orientation of
the gradient doesn't change on the image. This effect does go away for longer
exposures -- it's almost imperceptible for exposues 0.002s and longer (where the
field is quite flat), but I'd like to be able to use 0.0001s exposures.

Is this normal? Is there any way to fix this? (I didn't see an option to do flat
field compensation in Envisage, but even if there was, a 20-50K gradient is just too cumbersome to
work with even with flat fielding!)

This came up when I was trying to image the sun. The bottom of the image at
0.0001s was consistently 2x brighter than the top no matter how I turned the DSI
(which then led me to do an actual flat field test). I realize the work-around
can be to use a stronger solar filter and longer exposures, but it would be nice to be able to use 0.0001s exposures!

#2 rbelikov



  • Members
  • PipPip
  • 10 posts

Posted 31 January 2012 - 04:30 PM

If anybody's interested, I believe I figured this out. Essentially all CCDs suffer from this effect, and basically it means that exposures less than about 1ms are not very useful on the DSI III (and I assume most other CCDs).

In a CCD, typically pixels are always collecting light, even during the readout process. During the readout process, the charges in all the pixels are shifted vertically up until all charges have been transferred off the active area of the CCD. This charge transfer process apparently takes ~0.1ms for the DSI III. If there is enough light to generate photoelectrons during these 0.1ms, then they will contaminate the image that's being read out. So, effectively, every time you take an image, what you get is the true image, plus a vertically-blurred ghost image, if you will, corresponding to what would happen if you quickly vertically shifted the image across the CCD in 0.1ms.

This effect would look similar to blooming when imaging bright stars (a faint vertical ray on one side of the star), but it is different from blooming and anti-blooming will not help it. For example, if we assume that the CCD has 1000 rows and takes 0.1ms to fully transfer, and that you have a star that's 65K counts across a 10-pixel wide region during a 1ms exposure (and 0 elsewhere), you will have a streak below it with amplitude of 65K*0.1/1000*10 = 65 counts. This can actually be calibrated out very easily and I'm surprised I've never seen mention of this before.

#3 MistrBadgr


    Advanced Member

  • Administrators
  • 2957 posts
  • LocationBroken Arrow, Oklahoma

Posted 31 January 2012 - 06:11 PM

Thanks for the research!  I do not normally look at imaging posts, since I am more involved with beginner and intermediate scopes.  Your post caught my eyejust now as the most recent.  This is good information.

Best Regards,

Bill Steen
Bill Steen, Sky Hunters' Haven Observatory, Broken Arrow, Oklahoma

#4 JohnGraham


    Advanced Member

  • Members
  • PipPipPip
  • 85 posts
  • LocationKettering, Ohio

Posted 17 May 2012 - 12:52 PM

Hmmm, have you tried this with a bright target like the moon? I've noticed some bleed from top-to-bottom at the edge of the moon, but not on the moon itself. I've used exposures in the 0.0005ish range with good results. I've since switched to an DMK21/DBK21 for my lunar and planetary imaging so I haven't done this is a while.


0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users