Debayering a DSLR's Bayer matrix.

[freiform]

So far, so good. Heat gun at 320 °C and a small nozzle did the trick. The glass cracked starting at the places I applied physical force, hopefully the bondings are still intact. 

 

Sven

[nikivan]

There is a company offering this mod. In order to move the sensor so far away from the processing board, a long connecting cable is required. Do you know where one can buy such cable? 

[Stub Mandrel]

6 hours ago, nikivan said:

There is a company offering this mod. In order to move the sensor so far away from the processing board, a long connecting cable is required. Do you know where one can buy such cable? 

Probably custom made, do you have a picture of the original connections?

Seems a bit odd - is it cooled? If so the big lump of aluminium heatsink is going to be counter-productive.

Edited September 25, 2019 by Stub Mandrel

[Ahousefavor]

I know it's been a month since anyone posted but I just came across this and gave it a try on a sensor that was just sitting around. My question is that everyone says that theres 2 layers to remove (lens and cfa) but when I was doing it there was a 3rd one... did i just go to far? You can still see some microlens layer around the edges. Then the green cfa layer. Then theres kind of a clearish layer then what I guess is just the sensor

 

..please excuse the glass... it made me mad lol

Edited October 7, 2019 by Ahousefavor

[Noobulosity]

Just finished up adding a TEC system to cool my sensor.  This is from my Canon 7D, mono-modded.  I'm starting small with a 30W TEC.  I'm curious to see how the noise drops as the sensor cools.  I ran into significant noise shooting in mono, as compared to full-color images, but that could also be partly due to the increased sensitivity of the sensor.  I had banding in horizontal and vertical directions.  I'm hoping the TEC will help to reduce that noise significantly.

Has anyone else noticed banding/noise as an issue with these scraped mono sensors?

[Noobulosity]

On 06/10/2019 at 21:20, Ahousefavor said:

I know it's been a month since anyone posted but I just came across this and gave it a try on a sensor that was just sitting around. My question is that everyone says that theres 2 layers to remove (lens and cfa) but when I was doing it there was a 3rd one... did i just go to far? You can still see some microlens layer around the edges. Then the green cfa layer. Then theres kind of a clearish layer then what I guess is just the sensor

 

..please excuse the glass... it made me mad lol

 

That's probably too deep.  It looks like you scraped through the protective undercoating.  But, you can pop the sensor back in and test it to see if it still functions.

Broken glass gets into pretty sticky territory.  It doesn't take much to scratch the sensor too deeply and hit a photosite or surrounding circuitry.  And broken glass is really sharp.  It can easily scratch through the polymer coatings.

I'll be surprised if that sensor still works, but it's worth trying it to see what happens.

[Noobulosity]

Any ideas out there on plugging up light leaks through that left side of the camera, where I have my cold finger protruding?  It's coming out right by the data ports (HDMI, USB, etc.), and I have a feeling there's some light making it through the actual ports themselves.  I currently have the area covered with electrical tape, which does the job.  But, I'm looking for something that won't peel up.  The tape seems to have a tough time sticking onto the rubber grip and port cover areas.  Should I plug these ports somehow?  Fill them with liquid electrical tape?

I was shining a high-powered flashlight around those areas (probably 500 lumens or more, as it's a 1k lumen flashlight at about 3/4 power).  This really helped me find all of my light leaks.  Most of them were around the seam around the back panel of the camera where it attaches to the rest of the body.  I did notice some light made it in through the top small LCD screen that displays the current settings, too.  Some black electrical tape slapped over them helped plug up the light leaks pretty well.  I'm just looking for something that's less...  "Red/Green"-looking.  I'm leaning toward the liquid electrical tape, but always up for suggestions.

[Noobulosity]

Also...  Has anyone else noticed significantly-increased banding noise after removing the Bayer matrix?  I'm seeing a lot of banding noise in stretched dark frames.  And I noticed a lot of it came out when I took an image of the Eastern Veil nebula in H-alpha.  Now...  I don't think I dithered in that image at all, so that will likely help.  But, can dithering even out ALL of that banding?  I'm not entirely sure.

The stretched dark frame below was done in Photoshop by just doing a Levels adjustment and dropping the white point down to 3.  (Yep...  3.)

Unprocessed stack from DSS in H-alpha:

Slightly-cooled, highly-stretched 120s dark frame, EXIF temp 27°C (but, probably a few degrees cooler than that):

Edited October 20, 2019 by Noobulosity

[Stub Mandrel]

On 16/10/2019 at 23:36, Noobulosity said:

Has anyone else noticed banding/noise as an issue with these scraped mono sensors?

Very slight.

I can't tell from the picture, but make sure your cold finger is insulated everywhere!

[Noobulosity]

17 hours ago, Stub Mandrel said:

Very slight.

I can't tell from the picture, but make sure your cold finger is insulated everywhere!

Aside from the exposed areas touching the sensor itself and the TEC side outside the camera, it's been sprayed with a rubber material. I'm sure it's not perfectly sealed off, but it should be pretty well sealed inside the camera.

I plan to get some liquid electrical tape to touch some areas up.

Are you referring to electrically insulated, or thermally insulated? Or both?

[Stub Mandrel]

6 hours ago, Noobulosity said:

thermally insulated?

Thermally.  Any insulation you can add makes a difference, a bare spot of copper can lead to failure, it seems.

I saw the black colour, looks a good idea.

[Noobulosity]

I added a larger TEC (50W, this time) and heatsink.  It dropped the EXIF temperature very quickly, and condensation began forming within a few minutes of having it turned on.  I needed to use a desktop PC power supply to get this one started, as the laptop PSU had some type of short-circuit protection preventing it from running.

I noticed something interesting while I was messing around with my camera again.  I went back and looked at a photo with light leaks, and saw vertical lines.  But, it's not "banding noise", per se...  it almost looks like it's showing the gap between pixels, now that I'm using DCRAW to ignore the Bayer interpretation for RGB.  I think that's what I'm seeing as vertical banding in my images.  I hope that dithering will reduce that significantly.  Any other ideas as to how to eliminate this?  Can I "re-debayer" the image as mono, but in a way that's more like when it's interpreted as RGB to get rid of those pixel columns?

Edited October 23, 2019 by Noobulosity

[Noobulosity]

Ended up testing a bit more, and this interpretation seems to be a bit more blended.  Feels like fuzzier details, though.  I didn't get this with daytime photos, so this just feels very confusing.

dcraw64 -r 1 1 1 1 -d -4 -T -b 16 *.cr2

(FYI, I did stretch the image a touch, which is why it appears brighter.)

Edited October 23, 2019 by Noobulosity

[Gina]

Odd!  I never had that when I was playing with debayering DSLRs.  I used DCRAW too.  I gave up when I was spending more on semi-duff DSLRs than a proper astro camera cost.  It was fun though.

[Noobulosity]

21 hours ago, Gina said:

Odd!  I never had that when I was playing with debayering DSLRs.  I used DCRAW too.  I gave up when I was spending more on semi-duff DSLRs than a proper astro camera cost.  It was fun though.

I have to agree. This has been educational and quite enjoyable. I know it'll never be a "real" astro camera. But, I've learned a lot in the process.

[Stub Mandrel]

On 23/10/2019 at 04:12, Noobulosity said:

Ended up testing a bit more, and this interpretation seems to be a bit more blended.  Feels like fuzzier details, though.  I didn't get this with daytime photos, so this just feels very confusing.

dcraw64 -r 1 1 1 1 -d -4 -T -b 16 *.cr2

(FYI, I did stretch the image a touch, which is why it appears brighter.)

As the stripes are every eight column I suspect you have killed a data line OR got  a dodgy connection on one of the ribbon connectors.

[Noobulosity]

I'll try some more daytime shots to see what I get. It wasn't doing this before, and I tried to stay away from the sensor surface when I added the cooler. I can also try reseating the ribbon cables.

[Noobulosity]

Unfortunately, disassembly/reassembly didn't fix the issue.  I still have the regular intervals of vertical dead pixel lines.  I wonder if I touched something adding the copper cold finger and messed something up on the sensor.  It's tough to see what's actually between the sensor and its PCB, so I wouldn't be surprised if I scratched something with the copper plate as I slipped it between the two.  I recall there being a burr on the copper plate when I tried putting it in initially.  I did sand that off and it went in fine.  But, maybe I scratched through something in that process.

As much as I really, REALLY want to keep at this, I think my attempts at this project are done.  It was fun and rewarding.  I just can't keep sinking money into it, at this point.

Any idea if dithering might help to minimize the impact of these dead lines?  Or is that just an exercise in futility?

Anyone want to buy my Canon 7D and pick up the project?  Shoot me a PM.

Edited October 29, 2019 by Noobulosity

[alterelektroniker]

On 19/07/2019 at 03:52, YellowPacman said:

Also don't cover the golden wires with an epoxy, they will be teared as temperature rise.

You are right about that! Don't do that. 

This said, loooong time ago I work in an electronics factory and we did wire bonding Chip on Board modules, or COB as we call them. Bond wires where sealed with an special epoxy that had no thermal movement. Brand was Hysol then, now owned by Henkel, here is the link to what they have for that application now:

https://www.henkel-adhesives.com/us/en/products/potting-encapsulating-injection-molding-compounds/glob-top-material.html

And be careful, this not removable.

Alternately, google for Glob Top Adhesive-Sealants.

Edited November 1, 2019 by alterelektroniker

[Stub Mandrel]

4 hours ago, alterelektroniker said:

Alternately, google for Glob Top Adhesive-Sealants.

I want a tube of ANYTHING called Glob Top. I don't care what it does.

[zfedoran]

I managed to debayer a Sony ICX453AQ sensor (Cam86, QHY 8 pro, Nikon D40/D50/D70/D70s). I've heard that the glass on this sensor is quite hard to get off in one peace, so went super slow with it. I did eventually get it off without breaking it (used a blade to work around the edges, then heat gun to free the glass).
 

 

I was hoping that the sensor would be old enough to work with the scotch tape method (I've done a Canon 20D, which was super easy thanks to the CFA coming off with tape), however, this sensor does not release the CFA without scraping. I'm not sure if there are solvents out there that can do this sensor, but acetone didn't really work (except for help with cleaning).

Anyways, I've sealed it up, plugged it in, still works

 

Edited November 4, 2019 by zfedoran

[Gina]

Do you have a mono image?

[Stub Mandrel]

8 hours ago, zfedoran said:

I was hoping that the sensor would be old enough to work with the scotch tape method (I've done a Canon 20D, which was super easy thanks to the CFA coming off with tape),

Cool - I have an old 10D do you think that would work on it?

[zfedoran]

3 hours ago, Stub Mandrel said:

Cool - I have an old 10D do you think that would work on it?

I imagine it would work since it is older. If you do try it out, the trick is to get started with a small scrape, just enough to lift a few CFA pixels. Then apply the tape, pressing firmly, then peal back slowly. You'll need to take a lot of care to ensure the static cling of the tape doesn't make it jump somewhere you didn't intend (like the gold wires).

Take a look here: https://www.cloudynights.com/topic/486810-thinking-about-removing-the-cfa-from-my-20d-thoughts/

 

Just a side note though, I've done this on two canon 20d's and honestly, it isn't worth the effort. The camera is quite dated and really annoying to work with software wise. Other than that, it also suffers from an outdated ADC process that introduces noise.

[zfedoran]

8 hours ago, Gina said:

Do you have a mono image?

I did do a few tests, all looks good but I don't have an enclosure to put this in yet. I'll get back to you once I can take a proper flat frame.

Also, I tried something new for closing the sensor back up. I went with black hot glue, just in case I ever want to open it up again. It seems to have had a desirable secondary effect, I think it formed a slight vacuum.

I carefully put down shavings of the glue around the glass (not on top or under). Then I carefully hit the glass with heat from a heat gun. This obviously melted the glue in place, but then also sucked it under the glass, indicating that there is a pressure differential.

It will be interesting to see if it lasts and if moisture will be an issue.

 

Edited November 4, 2019 by zfedoran