Debayering a DSLR's Bayer matrix.

[Gina]

I think if someone devises a reliable method they could make a few Bob as many people would pay to have it done but be too scared to try themselves.

Sent from my Nexus 7 using Tapatalk

I agree.  "Reliable" is the crux of the matter

[Gina]

I've been thinking about option 1 myself. I was considering placing the sensor in a sealed metal tin (like a biscuit tin), filled with some desiccant and leave it over night to dry. Then put it in the oven to re-set the epoxy.

Yes, I was thinking of a sealed container with desiccant but the problem is to lower the cover glass onto the frame.  It has to be clear of the sensor and its frame to allow dry air to get to the sensor.  I'm thinking in terms of a transparent plastic box with the cover glass attached to a lever on an axle going through grommets in the sides of the box to seal.  Not sure what to do about the glue though.

[Gina]

Back to the debayering...  I have had a very very quick go at debayering the 1100D duff sensor with Dremel look-alike and felt buffing tool plus paint scratch repair paste.  Here is a photo of the result.

When I get the epoxy resin I can practice adding resin to the area around the sensor - see if I can cover the gold wires but keep below of the frame so that a cover glass could go back on when I work on the real job (a working sensor).

[russellhq]

Maybe it would help to orientate the sensor in the vertical position and do 1 row at a time. That way, gravity will tend the epoxy to flow away from the sensor.

[Gina]

Maybe it would help to orientate the sensor in the vertical position and do 1 row at a time. That way, gravity will tend the epoxy to flow away from the sensor.

Yes, but then it might run over the frame where the cover glass is going to go.  Anyway, time will tell

Edited April 22, 2014 by Gina

[Gina]

Still looking out the bits and pieces to make up the test rig.  Can't find the Peltier TEC I was going to use - a 12v 3A one from Farnel and more efficient than the Chinese ones.  However, I do have a 3.3v 2A one (TEC1-4902) about 7W I/P power and a 50mm square CPU cooler.  I have that setup running now and the cold side has already collected a thick layer of ice

[pixueto]

Seems JTW use epoxy resin to protect the gold wires and that definitely seems to be the way to go.  But it has to be the right sort of epoxy as I found out for myself!

Do you use the felt tips dry?  I ask because I tried that with a dry tip on the dead 1100D but even after some time it just polished the surface - I think it will need the car paint polish I used with the manual method.

It's very easy to go too deep and destroy the sensor that way.  Some parts cause dead pixels, some dead rows and/or columns and some complete failure - I've seen all of these!

I've yet to try a blowtorch on a complete cover glass but I have hopes of success.

Yes Gina, do use the felt tips dry; just be patient and keep going and eventually the CFA will start to come off. I wouldn't use any wet polish as it will be almost impossible to maintain an even surface and your flat frames will be a mess. I've found that, if some epoxy ends up on the sensor and you remove it with a sensor swab and isopropanol/eclipse fluid, you will also be removing the CFA in some places  making the dry removal easier when you apply the rotary tool.

Here you have a photo of my sensor. You can see that it is even with a perfect golden colour (except for a tiny bit on the left botton corner which needs the dry felt tip to be applied further). My suggestion is that if the coloured side edges hadn't been polished when trying to remove the CFA, it is possible that the sensor would still be working as the microscope shows no damage. In that case, the flat frames would have been perfect. I think it's worth having a go entirely and uniformely debayering the centre of the sensor without touching the edges and see if the sensor still works perfectly as I suspect.

[pixueto]

Yes, but then it might run over the frame where the cover glass is going to go.  Anyway, time will tell

Guys, don't worry that much about the epoxy going into the sensor. Just make sure you have a sensor swab to clean it with isopropanol. You can use the swab too to hold the resin in the right area. Some epoxy on the sensor which is removed with isopropanol and a swab is actually good as it will take away some of the CFA and that will help later on with the rotary tool.

[Gina]

Have you tried using a dry tool on an 1100D sensor?  These have a much tougher CFA than earlier models.  I tried dry on a 350D sensor (I think) and the CFA came off eventually.  I tried dry on the 1100D and even after about 5 mins continuous buffing the sensor was polished and very hot but still a strong green colour - no CFA had come off at all and the buffing tool was still white.  Maybe the tools you are using are different, harder perhaps.  I have some of the pointed tools mentioned earlier, on order from ebay and I'll see how they work.  I can say I was surprised how even the CFA removal was with the paint scratch polishing paste - quite different from what I achieved manually.

[pixueto]

Have you tried using a dry tool on an 1100D sensor?  These have a much tougher CFA than earlier models.  I tried dry on a 350D sensor (I think) and the CFA came off eventually.  I tried dry on the 1100D and even after about 5 mins continuous buffing the sensor was polished and very hot but still a strong green colour - no CFA had come off at all and the buffing tool was still white.  Maybe the tools you are using are different, harder perhaps.  I have some of the pointed tools mentioned earlier, on order from ebay and I'll see how they work.  I can say I was surprised how even the CFA removal was with the paint scratch polishing paste - quite different from what I achieved manually.

Unfortunately, I haven't tried with a 1100D yet. At page 74 in this thread you can see the tools that I have used, though. I think, Gina, a good idea would be to smear the sensor with the mapplin epoxy then remove it with isopropanol/ecipse fluid before it cures. This will help you break into the CFA, making the work with the dremel much, much easier. I wsed to use a needle and break some bayer matrix along the coloured edges to help starting the process but, obviously, I will stay well clear of those edges from now!

Edited April 22, 2014 by pixueto

[Gina]

Unfortunately, I haven't tried with a 1100D yet. At page 74 in this thread you can see the tools that I have used, though. I think, Gina, a good idea would be to smear the sensor with the mapplin epoxy then remove it with isopropanol/ecipse fluid before it cures. This will help you break into the CFA, making the work with the dremel much, much easier. I wsed to use a needle and break some bayer matrix along the coloured edges to help starting the process but, obviously, I will stay well clear of those edges from now!

I'll try the the epoxy/IPA approach.

I saw your post on page 74 and I've ordered those conical tools but the earlier link shows several options and I just wondered which of those you got.

<snip>

These are the tools that I've used:

http://www.ebay.co.uk/itm/261211368960?var=560197392052&ssPageName=STRK:MEWNX:IT&_trksid=p3984.m1439.l2649

[pixueto]

I'll try the the epoxy/IPA approach.

I saw your post on page 74 and I've ordered those conical tools but the earlier link shows several options and I just wondered which of those you got.

In the link:

http://www.ebay.co.uk/itm/261211368960?var=560197392052&ssPageName=STRK:MEWNX:IT&_trksid=p3984.m1439.l2649

It is the 'Solid hard felt wheel' the one that worked best for me. The conical ones that I also used are too soft so based on what you are saying it may be better not to use them with the 1100D. To get to the edges I used the same tool but it's too big and it removes most of the CFA in the coloured edges (which we know it isn't a good idea). I am now waiting for the tiny tips made of silicone that Rinaldo used only for the edges but I had to order them from China and it's taking ages.

Edited April 22, 2014 by pixueto

[Gina]

I've sorted out a clear plastic box that I can use to dry the air round the sensor and apply the cover glass.  This will also contain colour changing silica gel bags - of which I have plenty   I plan to have a DHT22 digital temperature and humidity sensor also in the box with the cable going through a gland and connected to an Arduino to check the dew point of the air.  That way I can be sure the air is dry enough.

[Gina]

I think I may have the answer to the glue problem - double sided adhesive tape.  Placed around the frame of the sensor.  Then the cover glass could be moved into position and gently pushed home onto the sticky tape.  Likewise the cover glass can be attached to the lever with double sided adhesive sponge.  With the adhesive sponge on the outside of the cover glass any residue can be easily cleaned off.

[russellhq]

Double sided tape sounds like a good idea. And I'm interested to see the lever arrangement you come up with.

[Gina]

In the link:

http://www.ebay.co.uk/itm/261211368960?var=560197392052&ssPageName=STRK:MEWNX:IT&_trksid=p3984.m1439.l2649

It is the 'Solid hard felt wheel' the one that worked best for me. The conical ones that I also used are too soft so based on what you are saying it may be better not to use them with the 1100D. To get to the edges I used the same tool but it's too big and it removes most of the CFA in the coloured edges (which we know it isn't a good idea). I am now waiting for the tiny tips made of silicone that Rinaldo used only for the edges but I had to order them from China and it's taking ages.

Thanks   They look similar to the ones in my kit but I've ordered one to see.  Shold be here Friday or Saturday.

[Gina]

Double sided tape sounds like a good idea. And I'm interested to see the lever arrangement you come up with.

I'm still thinking about the lever - I have some ideas but not decided yet.  I may do a SketchUp model.  Trouble is SketchUp doesn't cater for a moving model AFAIK.  Also, I don't know when I'm going to get the replacement control board for my 3D printer so I may have to build it the hard way.

The sort of thing I have in mind is to hold the sensor assembly with the sensor facing slightly downwards so that any dust doesn't settle on it.  And the cover glass with the inside face also facing slightly downwards.  Then the cover glass swung round and onto the sensor frame once the humidity has dropped sufficiently.  We haven't had a fire for a few days now so I'm hoping the dust level indoors is suitably low

[russellhq]

I wonder if you could make some kind of bladder or bellow that would allow you to control a lever from outside the box while keeping it sealed at the same time?

[Gina]

I wonder if you could make some kind of bladder or bellow that would allow you to control a lever from outside the box while keeping it sealed at the same time?

Yes, I should think so   I was thinking of mounting the lever on a smooth shaft going through grommets.  With a nice close fit of course to make a seal.

Edited April 22, 2014 by Gina

[Gina]

Here are a couple of screenshots from SketchUp showing the idea.  Front side of box removed for clarity.  Also, I haven't shown the grommets or the box details.  This is just a rough sketch but it is to scale.  The arcs show the path of the cover glass as its fitted onto the sensor frame.  The shaft shown is a 6mm aluminium rod but I may choose a different size depending on the size of grommets - I have quite a variety. I might use silicone grease on the grommets to aid sealing.

I still have to work out how the sensor assembly will be held in the right place and orientation and also the lever on the shaft and attachment to the cover glass.  It will be next week (or later) before I'm ready to replace the cover glass on the working sensor, as there's plenty to do before then so I might wait until I get the 3D printer working again.  Then I can print parts with just the right dimensions, angles etc.

Edited April 23, 2014 by Gina

[Gina]

I've set up a drying test with the box, DHT22 T & RH sensor and Arduino.  Here's a screenshot of the first few minutes of drying and a photo of the setup.

[Gina]

Another 20m of drying.

[russellhq]

Looking good so far, might be an over night job though?

[Gina]

Looking good so far, might be an over night job though?

Oh yes, I think so.  However, this is with just two silica gel bags and I shall put as many in as I can.  If I can print the parts I can arrange to keep the bags away from the mechanism - barriers with holes in.

Here's a screenshot of the serial output with another couple of bags added.

[Gina]

Thought I'd try an experiment so I unplugged the USB lead and put the box and Arduino in the freezer for a while.  Then brought it back into the living room and reconnected to the PC.  Here are the results for the first 15 minutes.  This shows how the dew point depends on temperature as well as humidity. 

Edited April 23, 2014 by Gina