Background: This is from my Mono DSLR camera that I have been working on for some time now. I finally have a sensor that is producing reasonable results but there is a problem which may not be related to the sensor surface itself.
Test: Last night, I took a series of four test lights of Eta C, all shot with the Canon 450D with mono sensor, ISO 1600, all same duration (300"), through my RC08 (no filter other than a front mounted UV/IR Cut). The attached light is one of the four light subs and they are all identical.
Also attached is one of four Darks shot just after the Lights, all are identical, shot on the same scope and camera with same settings and duration.
I have also attached one Offset Bias Frame, one of four also shot through the same setup at 1/4000". Note that the camera cooler was turned off during this test session, so these were shot on a warm night (about 22C), so noise will feature in the Light and Dark. Also the sensor still has some cleanup to be done on the surface and on the border, there is evidence of small bits of the Bayer Matrix coating still on the surface - ignore these. There was some thin high cloud around and this may have impacted the Light slightly.
Questions:
1. The thin line through the Light is puzzling, the sensor is obviously detecting data as there are some stars right on the line showing up. What could cause the line? Note I did have the sensor 'shield' off at the time (which is an earthed aluminium cover that encloses the back of the sensor board assembly). Is this simply internal signal from the camera electronics that the shield would have cancelled out?
2. In both the Light and the Dark there is a narrow vertical light beam on the left centre that rises from a sunburst sort of light source at the bottom. I suspect this is a light leak. I had my Teleskop-Express EOS filter drawer on the front of the camera but there was no filter in it at the time. It is possible that light could have leaked through the drawer sides or front as there were a few small light sources in the observatory (Red leds on heater controller, computer screen, etc).
3. The Bias Offset Frame shows none of the horizontal line or vertical light problems but this is consistent with the very short duration of the frame (1/4000').
My plan is to reshoot, when I can, without the filter drawer (just an EOS bayonet and T- Adaptor) to try and confirm the filter drawer as the light leak source. I also plan on reinstalling the sensor board shield for a retest to see if it impacts that dark line across the Light.
Further to this test image, I have stacked the four Light test subs with Darks, Bias/offsets, and Flats to see if I could clean up the artifacts via DSS. Attached is the final image, and I have also included a Flat example so that you can see the lines clearly. I am at the point of assuming that the lines are sensor related.
The question now is can the lines be "processed out" in some way?
The lines (black one) looks like a bad column, makes individual subs not look great but via dithering it can be processed out via stacking (same as getting rid of planes and satellites).
Re the Flat, I simply stuck a piece of white photocopy paper in the filter drawer and set the camera to ISO 100 and auto exposure (I believe it shot at 1/500th). The purpose was to determine the sensor performance not for normal image processing. I have disassembled the camera again today and cleaned the sensor again. Inspected the sensor for possible foreign matter, pin shorts, etc and retested. The Flat has improved but the two lines are still there. I will accept it is damaged and replace it. In the mean time I will try to do some dithered subs when the sky permits. Thanks.
Suggest that the light source may be a super hot pixel (or group), possibly generating light as well as dark current - yours is nowhere near as bad, but looks somewhat like http://www.iceinspace.com.au/forum/s...d.php?t=127858
The horizontal column? defect is minor and can be processed out as others have suggested, either using column substitution or dithering and statistical rejection. Many popular chips have column defects from new, so don't sweat on it.
Overall though, provided that the defects remain stable and can be handled in processing, looks like a winner to me - nice work.
Ray, et all, how do I do column substitution? Is there a particular software tool i need? ill retest without the filter drawer when I can, to see if it is leaking light.
Thanks for you help.
Ray, et all, how do I do column substitution? Is there a particular software tool i need? ill retest without the filter drawer when I can, to see if it is leaking light.
Thanks for you help.
hi Glen. Had a look around and most packages seem to have a function for cleaning up bad columns - I use PixInsight, which has that ability in "cosmetic correction", although I haven't ever used it. Also use Nebulosity, which doesn't have anything. I read that DSS has a column repair function, but I don't know anything about it. If you are keen enough, maybe try the software linked to in Sander Pool's post in this thread.http://pixinsight.com/forum/index.ph...c=4003.10;wap2
If you don't have suitable software, dither when gathering images and use outlier rejection to tidy up the bad data when stacking - as others have suggested. That would mean that you need do nothing to the individual subs.
Dunk, even though it would be a very similiar sensor I would not want to risk it. The one I have working is my fourth attempt, I think you might have seen an image from the first one that I had paid CentralDS to do. In hindsight it was close to being ok, but I tried to clean up the stuff they left on the surface and that ruined it - clearly their chemical stripping process did not work very well, at least on the 450D sensor. They did refund my payment in the end.
I found a low cost Chinese source for the sensors, and they were OEM, so I was paying about $38USD for each one I ordered, so the cost didn't bother me too much. I gave up on JTWs recommended 'glass polishing' method as it was messy and I think it resulted in shorting out of a sensor. The method that did work was the old scraping off of the CFA and microlenses, and I used a couple of different modified paddle pop stick tools for that. I broke a chip wire (the little gold ones) on one sensor that was looking promising, and I don't have the tools to try and repair that one.
Of course, never one to leave this sort of thing alone, I might try just one more in months to come, to get one without a bad column, but i could well end up with another failure. I should probably just cut my losses and settle with imaging with this one.
I have re-run DSS with changes to the setting parametres to try and clean up the image in stacking. Here is the resulting image, which is a definite improvement. I can still see the horizontal line but the vertical line appears to be gone now. Now to learn dithering. Thanks
Backyard EOS should have a dithering function, at least APT does. It will talk to PHD to do the nudging. It also works wonders with the Canon pattern noise, but I'm not sure whether you'd see that without the Bayer array
dithering would clean that up nicely.
