Debayering a DSLR's Bayer matrix.

[Gina]

So it looks like some other part of the camera that may be at fault?

James

Yes, I think so.

[Gina]

If you can get it in a freezer you'll find it cools down far quicker! Well that's what I found anyways

Yes, I can put it in the freezer - there's a socket nearby to keep the netbook charged if necessary. I'm a bit worried about condensation though.

[russellhq]

Yes, I can put it in the freezer - there's a socket nearby to keep the netbook charged if necessary. I'm a bit worried about condensation though.

I think you should be OK, the freezer is much colder so any moisture should already be frozen. I ran my 1100D in the freezer and it got down to -6C from memory.

[Gina]

It's moisture in the air inside the camera I'm worried about.

[Gina]

Been trying to remove the cover glass on the second working 1100D sensor without total success. I think it's alright though. As before I gradually and smoothly worked round the edge of the glass with the hot air and on the second go it started lifting, showing diffraction fringes on a short length of the bond. Let it cool again and started heating from one end of that as I went round there was a "ping" and a piece of the glass cracked and shot off to the side several inches. So I let it cool again and applied less heat but more glass cracked and more lifted. With the glass now in several pieces I tipped the fragments off the sensor and then continued round the still bonded part. After several more goes most of the bond was broken leaving a small area in one corner that just wouldn't loosen at all with a section of glass next to it that had lifted. The remaining glass is just clear of the imaging area. I tried lifting the glass at the broken bond with the end of a craft knife but I couldn't shift it using moderate force - I didn't want to risk breaking the glass and risking the fine gold wires.

Here are some photos. The fine gold wires seem intact as far as I can see. I'm considering replacing the sensor in the camera and testing it.

[Gina]

Put sensor back in it's camera (taking out the other sensor) and reassembled. Now getting Err 70 in EOS Utility

[Astroscot2]

I admire you guys and gals pushing this forward, Gina, you must be running out of cameras now !

Edited August 20, 2013 by Astroscot2

[Rottweiler]

The 30s exposure completes (can hear the shutter) and then I get err70. Now cooled down and I'm getting err70 immediately on starting time exposure.

I now have one of my ambient cooled 1100Ds (Rebel T3) stripped down to sensor cover glass ready to apply heat. Before that I tested it in both EOS Utility and APT latest version and all was working fine - took 10 x 10s subs in APT. Sensor was running at 25C.

The reason the error appears at the end of the exposure is because it is detected on readout. I saw the image you posted with the horizontal and verticle traps. This is from going a little too deep with the scraper, maybe try a wider tip. It does seem the wooden scraper is safer than the metal one as luis said, I think using a metal scraper is something that comes with practice. Although I have to say we will be sticking with our polish only method, since we also have occasional troubles from scraping methods.. The polish choices, I would stick to anything used normally for optical purposes. It is being that it is confirmed again and again that newer sensors are just plain harder to debayer than older ones.

[Gina]

I admire you guys and gals pushing this forward, Gina, you must be running out of cameras now !

Yes indeed! I have one modified 1100D with red filter removed and cold finger for ambient cooling and a duff 1100d arriving sometime in the future which hopefully has a working sensor. The other ambient cooled camera I've put away somewhere and can't find ATM. So I can do no more until I get the parts or find my missing 1100D. Edited August 20, 2013 by Gina

[Gina]

I have Peltier cooling working on that partly working debayered sensor. Running APT with 1s exposures at ISO 100 with 20s pauses. Started with EXIF T of 25C and all well but by the time it cooled to 22C it stopped working! That ties in with it working in the other camera indoors at 25-27C but not outside where it was cooler. So this sensor wont work below 23C in either camera.

[Gina]

The reason the error appears at the end of the exposure is because it is detected on readout. I saw the image you posted with the horizontal and verticle traps. This is from going a little too deep with the scraper, maybe try a wider tip. It does seem the wooden scraper is safer than the metal one as luis said, I think using a metal scraper is something that comes with practice. Although I have to say we will be sticking with our polish only method, since we also have occasional troubles from scraping methods.. The polish choices, I would stick to anything used normally for optical purposes. It is being that it is confirmed again and again that newer sensors are just plain harder to debayer than older ones.

Thank you I won't try a metal scraper again - maybe a hardwood one, possibly. The car paint cutting back polish I have works with the 1100D sensors.

The problem with the cover glass on that last 1100D sensor is a worry - getting it off killed the sensor it seems

[Gina]

My progress may be summed up as four dud 1100D sensors and one dud 350D sensor Maybe I should take up knitting like other women of my age...

[Gina]

I tried knitting some time ago got it out of my system Found it too boring... Just feeling rather frustrated ATM!

[JamesF]

I tried knitting some time ago got it out of my system Found it too boring... Just feeling rather frustrated ATM!

Trying making a machine that knits instead

I understand your frustration though. Does the first 1100D sensor you modded still work in the other body, or is that the one that fails if it gets too cold?

James

[Gina]

Trying making a machine that knits instead

I understand your frustration though. Does the first 1100D sensor you modded still work in the other body, or is that the one that fails if it gets too cold?

James

Thanks James It fails when too cold. To spell it out :-

Body no. 1 - Stripped down to the absolute minimum for astro with cold finger and Peltier cooling.

Body no. 2 - Mostly complete except for back section - colour correction filter removed for astro use. Cold finger fitted to sensor and brought out at the back.

Sensor no.1 - Out of body no. 1, debayered with 3/4 of sensor working but with some dead rows and columns.

Sensor no. 2 - Out of body no. 2. cover glass removed, all gold wires visually intact but sensor totally dud - Err 70 and black image

Sensor no.1 appeared to be working (just dead rows and columns) before I did the final cleaning up. I didn't try it cold until after the final cleaning up so don't know at what stage it stopped working when cold.

Sensor no.2 didn't get as far as debayering - I killed it in removing the cover glass. I think I must have over cooked it

Sensor no.1 works in both bodies in warm conditions - fails when cool giveing Err 70 but shows an image.

Edited August 20, 2013 by Gina

[Gina]

It's a shame that Sensor no.1 doesn't work when cold otherwise I think it would be quite useable.

[russellhq]

What do you think the white spots are in the lower left of the frame?

[Gina]

I think they must be places where I went too far and destroyed some pixels. Don't kow why they're white rather than black though.

[JohnC64]

Looking on ebay, i have found a few places that are selling the sensors for 350D for about £30 - £40 (searched for Canon 350D sensor)

http://www.ebay.co.uk/itm/CANON-DSLR-REBEL-XT-350D-CCD-SENSOR-ORIGINAL-REPAIR-PART-/271256723891?pt=Digital_Camera_Accessories&hash=item3f282921b3

http://www.ebay.co.uk/itm/GENUINE-CANON-EOS-REBEL-XT-350D-CCD-SENSOR-REPAIR-PARTS-/281017340502?pt=Digital_Camera_Accessories&hash=item416df05256

Although it looks like they come from the states

John

[Chris]

Oh dear

My progress may be summed up as four dud 1100D sensors and one dud 350D sensor Maybe I should take up knitting like other women of my age...

I feel for you having such bad luck after all the effort you've put in its a lot of dead sensors and expense so far on this thread, its definately given me second thoughts about giving this a go with my next ebay camera, I might just good old fashioned mod it!

[pixueto]

Oh dear

I feel for you having such bad luck after all the effort you've put in its a lot of dead sensors and expense so far on this thread, its definately given me second thoughts about giving this a go with my next ebay camera, I might just good old fashioned mod it!

Well, it'll be me the next one to have a go. I'm now going to polish the 1000D sensor that is already partially debayered with fine glass polish diluted with 99.999% pure isopropanol as suggested by Rotweiller. I'll apply it with a sensor swab cut to half size. Then, I''ll move on to the 350D and I'll try the Dichloroethane approach with that one. I've got all the ingredients ready. Trouble is removing the glass over the 350D sensor using heat. That is a big worry after the recent senor failures with that procedure. As I said earlier, I'll avoid using the epoxy (with the heat, I'm afraid I have no choice but I'll be using a torch pen). I won't remove the CFA completely from the fringes until I see if I can remove all the bayer array leftovers without damaging the sensor. And if I succeed with that (remains to be seen), I'll inspect the kind of flat frames the camera is producing.

Edited August 20, 2013 by pixueto

[Chris]

There must be another way to get the cover glass off other than heating the adhesive to 300-400C! I noted earlier that Gina used a sticky pad to stop shards of glass breaking the gold wires, I don't know if the following would work: cut a sticky pad to fit inside the glued area of the cover glass, then score the glass around the outside of the sticky pad repeatedly unitl you could lift out the glass using the sticky pad. I bit like using a router and a guide fence. The glass doesn't look that thick so maybe a simple blade might work to score it?

Chris

[pixueto]

There must be another way to get the cover glass off other than heating the adhesive to 300-400C! I noted earlier that Gina used a sticky pad to stop shards of glass breaking the gold wires, I don't know if the following would work: cut a sticky pad to fit inside the glued area of the cover glass, then score the glass around the outside of the sticky pad repeatedly unitl you could lift out the glass using the sticky pad. I bit like using a router and a guide fence. The glass doesn't look that thick so maybe a simple blade might work to score it?

Chris

Both the canon 1000D and 450D have cover glasses that can be removed intact without applying any heat. It's the 1100D and 350D the ones that seem to give trouble. A pity because the 1100D is a very powerful camera and, as for the 350D, there is a chance that the solvent may remove the BCA without scraping (I'll let you know if that's the case when I try it).

However, the 450D is an impressive camera for astrophotography. If the CFA leftovers can be polished, and the scratches removed while the flat frames are good in my 1000D, there is virtually no difference between that model's sensor and the 450D (I know that because I have been practicing with a dead 450D).

Edited August 20, 2013 by pixueto

[Oddsocks]

For cover glass removal this technique might be worth a try.

Glue the end of a piece of flat metal bar around 200mm long and of a cross section shape that is the same size (for cold removal), or slightly smaller than the cover glass window (for hot removal) using super glue.

The glued metal surface needs to be as flat as possible so that it bonds to the glass window with the same surface tension across the entire area.

Now when you try to prise off the glass window either cold or hot you can apply a little sideways force to the bar at the same time lifting the edge of the window.

This should spread the stresses across the entire window and help prevent it from shattering.

If all goes well and the window comes off in one piece then place the metal bar and glass window in a close-able glass container and add a solution of acetone (in the open air...not indoors!!!!) for several hours to dissolve the super glue, the two pieces can then be separated with a thin scalpel or hobby blade.

I used this technique a few times for removing radiation protection lead-glass (Pb) port windows from obsolete medical X-ray sources so that the X-ray source housings could be reused.

(Pb glass in X-ray sources is very thick ~1cm but extremely soft and fragile, you couldn't handle it with normal tools, it would crumble away but the glued-to-a-bar method worked a treat)

Gina, have you tried finishing and polishing the faulty sensors and then washing with a mild solvent and rinse with distilled water, sounds crazy but you only need a microscopic piece of dislodged conductive silicone to be embedded in a sensitive area of the wafer and this may be causing the temperature related errors, because the gold wires are not very springy if they were damaged you would expect them to be permanently broken and not make/break with temperature change but the wafer surface is much more easily disrupted and a tiny piece of silicone bridging two conductors could easily make/break with temperature.

Whatever the outcome this has been a great thread to follow, much better than anything on TV, has anyone approached you yet for the film and book rights?

William.

[alistairsam]

Hi William,

Thanks for the tip about the Bar. I've thought of that and will be doing something similar with my 600d. darn it, its taking ages to arrive.

I think I'll do a 450D instead.

Cheers