So I finally got around to making my first attempt at the sensor modification. It's great to have 92 pages of R&D to go on! Just wanted to add some more bits on this chain to give back a little bit. My first attempt was not a 100% success, but the sensor still works. I've ordered two more sensors (just $100 for the pair) and feel much more confident now. I plan to use the current sensor as a tester for any new ideas.
Cover glass and CFA removal for the 550D definitely seem to fall on the harder end of the difficulty spectrum. Cover glass was on tight, but in the end defeated by a butane torch and a razor blade.
For CFA removal I first tried the scraping method using a wooden tool but did not have a lot of success here. For the sensor on this model, the CFA did not break away as it seems it does on some other models and I felt like I was doing more harm than good. See Image 1 for a representative flat taken after about 30 min of scraping.
You can see from that image that this also resulted in 2 dead rows of pixels. Once I saw this I gave up any hope that I could limit this endeavor to just one sensor and became much more carefree about the whole process. So, I got out the dremel tool....
Using some glass polish and the felt polishing bit, I polished the CFA away using the dremel tool. I also experimented with some glue to protect the gold wires. I used Loctite Go2 glue and it seems to have worked fine. I had been worried about the glue breaking the wires as it cured, but seems it's OK.
The other two images are a flat and dark frame after dremel polishing and glue application. You can see I over polished in a few spots, and there's a hot pixel and row, but I'm confident I can mitigate those issues on a second attempt. It could be better, but I'm hoping the lessons learned from the 'Mark-1' will translate into better results on the next go-around. I plan to try some dry polishing using the dremel to clean up the edges. If successful I may go without the glass polish on the next attempt.
Thanks again guys!