Debayering Nikon D70 sensor

[luka]

I thought this is worth a separate post instead of burying it inside our camera build thread.

Basically there are 2 steps, one is removal of the cover glass and the other is the actual debayering. I will split the info in several posts to avoid a single huge post.

I used a toothpick to scrape the sensor to remove the CFA (colour filter array). There are several other methods that were suggested on the net.
1. Dichoromethane (DCM) apparently dissolves/softens the CFA. DCM is often found in paint strippers. I did not use this method as I could not find a suitable paint stripper. The ones I found had gel inside which was impossible to get rid of.
2. Ethyl acetate is supposed to soften the CFA. I did not have any to try but I tried isopropanol and could not notice any difference compared to just using the toothpick.

Tools used: scalpel blade, heat gun, toothpicks, stereoscope, extremely steady hands, patience (lots of it).

For this method (and probably any other method) a stereoscope with 10x to 20x magnification is necessary. You could get away with smaller magnifiers to debayer the middle of the sensor but for the corners the accuracy is critical. See photo which shows the tiny gold wires which will break if you look at them the wrong way
Some people also use epoxy to protect the gold wires but with a stereoscope this is not necessary.

To be continued...

[luka]

Removal of the cover glass:
- Use a sharp blade and scrape off all the resin around the sensor. Scrape both ways, using sharp and blunt edge of the blade. You will need to get every little bit off. It comes off as yellowish powder (see photo 1). If you have any left you will run into problems later. Even when you think it is off scrape some more. Change angles of the blade and scrape more.
- When you are 100% sure that the resin is gone, you can use the heat gun to heat up the glass and get it separated from the resin under it. Glass will expand when heated and break the bond.
- Start heating in one corner and wait until the colour changes. See photo 2. Keep pushing the "discolouration" until it is all around the sensor and the glass is separated.

Make sure you let the sensor cool down. I held the metal frame by my hands and when that got too hot to hold I stopped for 5min and let the sensor cool down. Filip used a wet cloth to keep the sensor cool.
I did this over the course of several hours and it took me about 10 heating sessions.
DO NOT OVERCOOK THE SENSOR.

For my broken test sensor I used max temperature of the heat gun (450 degrees C from memory). The glass got cloudy but could be wiped clean. I managed to get the glass without cracking.
For my other sensor I used 340 degrees C and the glass cracked at the very end.
I am not sure which temperature would work the best but I would go with the max if I do it again. Higher temperature will heat the glass faster and hence create a larger temperature differential which will break the bond to the resin faster. However, the sensor will heat up faster as well and faster breaking of the bond may lead to cracking of the sensor. Not sure

Once the glass is off you have to be extremely careful not to damage the gold wires.

To be continued...

[luka]

Debayering of the sensor:
Again, the stereoscope is critical for this step. This step took me over an hour and I took several breaks to rest my hand.

I used a toothpick but people found that sharpened artist's brushes work well. I tried using sharpened piece of plastic but that scraped the test sensor. Also different brand toothpicks worked differently, probably because they are made from different wood.

- Start in the middle of the sensor and scrape. Keep scraping... You will notice that the microlens array will come off easily (slight change in the shade of green) but the CFA will take few minutes to scrape through. You could use something sharp to get a head start but you run a great risk of damaging the sensor.
- When CFA comes off it will be obvious. The lightly greenish colour will change to completely shiny colour.
- I believe that the CFA is abrasive and once it starts coming off, the toothpick will suddenly start working really well. If you break it get some CFA on the tip of the new toothpick before resuming.
- Once CFA starts coming off, just keep scraping and expanding the debayered section. Every now and then you could pop the sensor into a DSLR and check your progress.
- You can use compressed gas to clean the sensor. It will not break the gold wires.
- When you finish the middle you will need to be extra careful with the edges:
1. Golden wires are no go zone. But stereoscope will help you stay clear.
2. Edges... from golden wires towards the inside there is a blue line (with electronic circuitry inside), then there is a golden wire and then there seem to be another line that is covered by CFA. The latter has a slightly different colour than the rest of the CFA. Photo is above.
DO NOT TOUCH. STAY AWAY.

I did touch the golden line once or twice and took the top layer off but I got lucky and the sensor survived. I did not go all the way to the sensor edges. Too risky but I may go back to it.

Hopefully someone will find these notes useful. As I said there are few ways to tackle this problem and I liked the simple scraping. This method worked OK for Nikon D70 and D40 sensor but different sensors use different technologies and this method may not work well for them.

Note: This is an extremely delicate procedure, especially debayering the edges. Chances are that you will kill your sensor. There is a 99 pages long thread here that has lots of information. Some people killed over 5 sensors before giving up. Again, you are likely to kill your sensor if you try this. Do it at your own risk.

And a good source of new sensors are broken DSLRs on ebay

[luka]

And one more time, if you do this you are likely to kill your sensor

[gehelem (Gilles)]

Ouch ! need to hang hard to do that !
I don't know if i'll ever do it...
Thank you
Gilles.

[billdan (Bill)]

You're a brave man Luka, I wish I had the guts to remove the CFA from my QHY12. If I was brave enough to go down that path I would buy some used sensors to practice on.

In that SGL thread, if I remember correctly , R.A.C (from NZ) used a UV lamp to soften the CFA array and then applied sticky tape to peel of the array.
I have never read of anyone else using that technique, maybe RAC just got lucky.

Cheers
Bill

[glend (Glen)]

Ha ha, as a sensor killer myself, who may hold the record, good to see your giving it a go. Keep im mind that most debayered sensors will exhibit some artifacts of the process and dithering will be required to calibrate them out. That said, even new high end ccds can have dead columns so don't feel too bad if you have a few spots or flaws. Learn to do flats and dithering to correct.

CentralDS does, or did, offer a chemical debayering process, but it was not cheap, and they offer no guarantees nor even test prior to return.

The best mono sensor is one from the factory all others are a compromise.

[luka]

I have attached the flat. It is still work in progress as I need to clean up the leftovers in the middle but I am tempted to go back to the edges... just need to build up some courage first

Bill, I read about the UV lamp. I think it worked on one sensor but had no effect on a sensor from a different camera. Generally speaking the older sensors are easier to work with. In 300D or few other older cameras the sensor glass can be popped off with a blade and the bayer matrix scraped very easily. But the newer ones are generally more difficult to debayer. Glen did 450D and from what I read it was much, much more difficult than 300D.