AutoStarSuite.Net

Your Source for Meade AutoStar, AutoStarSuite, LPI, DSI and Telescope information
Welcome to AutoStarSuite.Net Sign in | Join | Help
Home Blogs Forums Photos Files Astronomy News

Yet more dark frame problems with DSI III

Last post 01-10-2011, 5:05 AM by JonW. 10 replies.
Sort Posts:
  •  03-15-2010, 4:29 PM

    Yet more dark frame problems with DSI III

    Hi All,

    I'm still getting problems with dark subtraction in Envisage. I have a recurring problem where the dark subtract is taking the low end of the histogram to practically 0, in fact, the saved frames bear no resemblence at all to live image, and are noisy and have nowhere near the dynamic range of the image captured. When this has happened previously, removing the old darks and shooting new ones has cleared it up, but not this time.

    Normally with my set-up the gain is set to 100 and the offset to 45, this gives a noise floor of about 4500 ADUs. When this problem strikes, noise floor is around 10 ADUs. I increased the offset to 55 which gives a raw noise floor of 10,000 ADUs, but again this sinks to around 10 ADUs with dark subtract selected.

    Any ideas?

    Thanks, JonW
    10" Meade LX200 Classic
    8" Mead LX-90
    TMB 80/480
    DSIc, DSI IIIc, QSI 583ws
    Isle of Wight
  •  03-15-2010, 9:22 PM

    Re: Yet more dark frame problems with DSI III

    Hi, Jon

    Is this with the DSI-IIIC?

    How much did the temp drop during the session?

    Just guessing out loud here, but I think I would try taking darks with offset at 45, and lights with offset at 50 or 55. The theory being that if the camera cools down and the thermal noise component falls off, you will be less likely to over-subtract the lights.

    Another thing you can do (if you saved the original files before dark subtraction) would be to use something like AIP4WIN to manually add a pedestal of 2,000 or so to the lights before applying the darks.

    Hopefully someone else with more recent experience will jump in here & help out.

    Chuck
    Chuck Reese
    CDIP Certified: Nebula, Galaxy, Solar System, Stars & Clusters Imager
    CDIP Mentor: Nebulae & Solar System Categories
    For information on the Certified DSI Imager Program, see:
    http://autostarsuite.net/forums/1427/ShowForum.aspx
  •  03-16-2010, 12:52 PM

    Re: Yet more dark frame problems with DSI III

    Hi Chuck,

    Yes the problem is with the IIIc. There was only a 0.5C tempertaure drop between taking the darks and the the flats at the end of the session. Even the flats (also dark subtracted, but for a shorter exposure, look bad... So it's not just a long exposure problem).

    Unfortunately, I've only been saving the dark subtracted frames, so have lost 4 really good sessions now... Though I should have investigated further this time as the histogram minimum was so low.

    I'm getting a bit worried that the stability of the DSI IIIc amplifier may be questionable, and the errors are due to jumps in gain, however at the moment I need to look into this more systematically. I'm hoping this is not the case and it's some obscure software "feature" or set-up problem. I might try and capture some frames in Nebulosity as a cross check next time this feature shows up.

    Will do some more investigation tonight and post the outcome.

    Thanks,

    JonW


    10" Meade LX200 Classic
    8" Mead LX-90
    TMB 80/480
    DSIc, DSI IIIc, QSI 583ws
    Isle of Wight
  •  03-16-2010, 4:43 PM

    Re: Yet more dark frame problems with DSI III

    Attachment: Dark Noise.jpg
    Further update:

    1) Ran the DSI IIIc at 19.0C, gain at 100 and offset at 49. Noise is around 8000ADU level for 1S frames. Left it to settle for about 1/2hr
    2) Shot some 10-frame darks for 1S and 15S.
    3) enabled dark subtract, took some frames with lens cap still on and the histogram disappears of the bottom end...
    4) reset the offset to 53, and this brings the noise back to about 4000 ADUs.

    I'm guessing that in this case it's like subtracting two sets of noise about the same mean, and in order to see the noise in the dark subtracted frames, the bias point has to be shifted up a little. But I'm a bit surprised at the size of the shift needed, and why this has not been a regular problem. I've never previously adjusted the offset after enabling the dark subtract?

    The only thing I can think of that has recently changed is that background light level has been significantly reduced by using a light pollution filter. Previously, the background level would have been skyglow dominated and well above the camera thermal noise. I'm thinking that maybe the positive benefit of reducing the skyglow impact, means I need to take more care in setting the CCD offset?

    Whilst looking into this I've also noticed a strange behaviour where the first frame after changing the exposure time, or starting a stack of frames, seems to be different from the steady state. I've attached an AIP4WIN screen-shot that shows the dark line at the top, and brighter background in the first frame (0001) than the second (0002) and subsequent frames. Small changes, but noticeable...

    JonW

    10" Meade LX200 Classic
    8" Mead LX-90
    TMB 80/480
    DSIc, DSI IIIc, QSI 583ws
    Isle of Wight
  •  03-16-2010, 11:43 PM

    Re: Yet more dark frame problems with DSI III

    OK, last item first

    The "first dark at a new time" problem was with the DSI's from a long time back, at least to the II series and maybe even the original. I thought the developer did something in one of the last few releases to fix that, but I can't remember now exactly when that was or how it worked. What version of the software are you running? If 7.05 or later you should be good. What I used to do was let it take the first dark then start the dark frames over again. Since I used a cooler to keep the temp steady, it wasn't a big deal for me.

    As for the other, the offset problem, I agree it shouldn't be that way. I don't trust the temp indicator for two reasons. One, the temp sensor is place at somewhat of a remove from the chip. I'm not sure what it is measuring exactly, but it is not really the chip temperature, strictly speaking. Two, I had at least one (maybe it was both?) DSI-III where the cold finger didn't reach the chip by quite a bit (like, a 1/16th inch or more). You can fill the gap with thermal paste and it should help, but it really isn't intended to bridge a gap like that. It's more intended to fill in the little pores in the surface of the metals to eliminate any tiny air spaces from acting as an insulator. Again, I installed an after-market cooler which corrected this problem.

    So I guess I would have to recommend you pop the back off the camera if you have not done so already & see what's what. I'm betting you will find you have been monitoring the case temperature and not the true temp temperature, and once you solve that problem things will behave as expected.

    HTH,

    Chuck
    Chuck Reese
    CDIP Certified: Nebula, Galaxy, Solar System, Stars & Clusters Imager
    CDIP Mentor: Nebulae & Solar System Categories
    For information on the Certified DSI Imager Program, see:
    http://autostarsuite.net/forums/1427/ShowForum.aspx
  •  03-17-2010, 4:30 AM

    Re: Yet more dark frame problems with DSI III

    Hi Chuck, I use 7.05... I didn't get as far to check what dark frames actually got stacked, so maybe they ignore the first one?

    On the main point, I think I'll do as you suggest and check out the thermal path from CCD to heatsink as a wise precaution. It may explain things if the thermal path is inconsistant and the chip temp is moving around semi-independantly from the heatsink (and temperature monitor point) and the wrong darks are being selected... Hadn't thought of that one. Particularly as the problem seems to be more apparant when there are more darks to choose from (I had a library with 1C spacing for my main exposure times). This would also stack up with the odd bright frame, or the frames with unexplained dark current breakthrough as reported in the previous post.

    Will report back once I've looked into it further...

    Thanks, JonW


    10" Meade LX200 Classic
    8" Mead LX-90
    TMB 80/480
    DSIc, DSI IIIc, QSI 583ws
    Isle of Wight
  •  03-17-2010, 6:19 PM

    Re: Yet more dark frame problems with DSI III

    Hi Chuck,

    I opened the DSI IIIc up this evening, and the CCD chip was not very well connected thermally as you suspected. There was about 3mm of heatsink paste between the chip and heatsink, and it looked like expansion/contraction had partially pulled the two apart (there was no adhesive force holdingthe circuit board to the heatsink, it literally fell away under its own weight). I replaced the paste with a set of copper shims and "silver fox" and was fairly optimistic of an improvement but now not so convinced. There now seem to be far fewer hot pixels towards the mddle and top of the histogram, but the dark spread at the bottom is twice what it was before (6300-14300 ADUs vs 5400-9300 ADUs at the same indicated temperature yesterday). I'll fit another shim tomorrow and see if that improves things.

    This problem has also made me much more aware of the impact of dark current variation temperature, it's incredible what dfference even 0.5C makes.

    JonW
    10" Meade LX200 Classic
    8" Mead LX-90
    TMB 80/480
    DSIc, DSI IIIc, QSI 583ws
    Isle of Wight
  •  03-18-2010, 5:44 PM

    Re: Yet more dark frame problems with DSI III

    The final solution...

    I stripped the DSI down again and measured the locations of the internal mounting points and the heat sink pillar. It became clear that the circuit board (and hence the chip) is not parallel to the heat sink pillar, so this is what I did... First cut out a 12x14mm piece of 1.6mm copper. Then lapped this onto the heatsink pillar to get a good interface between copper plate and heatsink. I then lapped the other side of the copper until the circuit board stood off the mounting pillars by 1mm. The heatsink and chip interfaces were then coated with a thin coat of silver fox heatsink compound and the DSI re-assembled. Now the careful bit, the allen screws have to be tightened up in a diagonal sequence, and the gap between the two shell halves reduced to 1.5mm. At this point if you overtighten, you'll probably break the chip, you need to leave enough flex in the o-rings under the allen head screws and the seal between the two halves to absorb any expansion. If the screws are tightened unevenly, the thermal contact with the copper shim will be compromised (this was my problem with the first attempt) as the chip and shim will not be parallel. I've now got significantly better dark current performance, particularly a massive reduction in the amp glow. This seems to have cured the unexplained dark breakthrough and black spots... but need more than a sample of one. I'll post some pixel statistics for the before and after to close this post down.

    The other problem, occasional horizontal light and dark barring, seems to have been cured by re-installing the DSI drivers.

    My advice is to keep an eye your dark frame variation over time, because it's a really sensitive test of the DSI state of health.

    JonW


    10" Meade LX200 Classic
    8" Mead LX-90
    TMB 80/480
    DSIc, DSI IIIc, QSI 583ws
    Isle of Wight
  •  03-19-2010, 7:10 PM

    Re: Yet more dark frame problems with DSI III

    Thanks for the update, Jim. Glad you figured it out.

    Did you by chance take any pictures of your project?

    Chuck
    Chuck Reese
    CDIP Certified: Nebula, Galaxy, Solar System, Stars & Clusters Imager
    CDIP Mentor: Nebulae & Solar System Categories
    For information on the Certified DSI Imager Program, see:
    http://autostarsuite.net/forums/1427/ShowForum.aspx
  •  03-20-2010, 8:16 AM

    Re: Yet more dark frame problems with DSI III

    Hi Chuck,

    unfortunately I didn't take any pictures. I would also not advise anyone to split the DSI III and do anything unless they have a real problem... The approach I've described is relatively high risk as the pressure contact between chip and heat sink is maintained partly through the circuit board and seal/o-rings being kept under compression. The standard build process Mead has used (an adhesive heat sink paste) minimises the risk of thermal changes putting increased pressure on the chip, which could either distort or break the chip.

    The gains of this modification are slight, measurements made with the DSI III in the fridge, indicate that the dark current is reduced, and before and after darks indicate that the DSI III can run 1.5C hotter indicated for the same dark current. This is pretty small, so not worth it unless you run into the same problems with the adhesive/heat sink compound separating from the chip. There's also a very slight improvement in the dark noise.

    At least it's back up and running reliably, and I've learned a lot more about the sensitivity of the DSI III dark current variation with temperature along the way.

    JonW
    10" Meade LX200 Classic
    8" Mead LX-90
    TMB 80/480
    DSIc, DSI IIIc, QSI 583ws
    Isle of Wight
  •  01-10-2011, 5:05 AM

    Re: Yet more dark frame problems with DSI III

    Hi All,

    Just a word of warning if you find yourself in the situation of having to re-make the heatsink to chip joint following a strip down of the DSI. My re-worked DSI worked fine until last month, then I started getting intermittent frames and finally no video output at all, just gaussian noise. I stripped the DSI down again and discovered that a minute (about a hair's thickness...) amount of the heatsink compound had dropped onto one of the imaging chip legs and shorted it to ground. Despite being careful with the heatsink compound, I must have left a small residue on the PCB by accident, which over time fell onto the chip leg. Fortunately, having cleaned up the PCB, the DSI now seems to be up and running again.

    So, as well as using the absolute minimum of heatsink compound, it's a good idea to double check the PCB and chip with a bright light before reassembly. In fact, it may well be better to use a non electrically conductive compound or heatsink pad and put up with the lower heat transfer. I was lucky on this occasion it could have had far worse consequences.

    JonW


    10" Meade LX200 Classic
    8" Mead LX-90
    TMB 80/480
    DSIc, DSI IIIc, QSI 583ws
    Isle of Wight
View as RSS news feed in XML

AutoStarSuite.Net

Powered by Community Server, by Telligent Systems