I think my next step is going to be to write a function to step through pretty much the entire damned memory on this camera looking for the AF LED... Having the blue LED will help tremendously in this process because we can have it blink patterns to indicate where it is in the scanning process to help avoid going crazy...
Here's what I was thinking would make sense to write a function to do:
For (i = 0 to big number, incrementing by 1) {
Try to blink an LED at address at that step value (i.e. increment counter by one, but increment LED address by appropriate number of bytes)
Wait a few seconds
Blink the blue LED with a pattern indicating which one was just blinked
}
The function to blink the blue LED could be done in binary... with a 1-second window for each blip.
Before blinking out a number - have it blink rapidly 4 times in one second- and then pause for a second...
Blink once in the one second window - logical 1
Blink twice in the one second window - logical 0
we could put a pause of 0.125 seconds in between each bit just to help prevent a string of a lot of zeroes becoming hard to count...
Or... we could just do it morse-code style and do dots and dashes with some kind of flicker at the beginning to indicate it is starting a new number.
The benefit of reading it out in binary is that it would be easy to be human readable with a relatively constant time to read out each bit... If we had something where it blinked it out in base-10, that would be a pain in the butt to sit there and count... With the binary output you could just jot down things on paper and then if you missed it... just jot down the next one
Maybe this is overkill, but I think that leveraging the existing blue LED to find the AF LED would be a good use of our time. Unless, of course, someone can find my a blue-sensitive phototransistor... I know they must exist, I just can't find them. (Blue LEDs are in the 450 to 470 nm range for light). I'm willing to buy one and ship it out here, but even then if we could find the AF LED we could have the firmware dumped before the phototransistor would get here.
On another note - can the LED backlight be flickered at a fast enough speed to dump the firmware? I know it would be a big battery drain, but if any of us have phototransistors that could be sensitive to that moreso than the blue - that might be a good approach. That one is in a known range at least...
Anyway... those are my thoughts for now. What do the rest of you think?