I'm on my way out right now, but I'll take a closer look at this later this afternoon. However, for the time being:



The errors are not a problem. I almost always get a few. Those are errors that were detected and subsequently rewritten successfully. The error count is just for informational purposes.



Do you observe this behavior only by touching pin 2? Not by touching any of the other pins? I'll have to think about this at greater length when I get back.



As has been seen on this forum in quite a few threads, getting SPI ISP to work often is very dependent on cable lengths, types, constructions, etc. All my work with the SBC and VisISP-52 has been with a purchased, shielded DB25 Female->DB25 Male connector. I'm suspecting that hacking off the Centronics connector and installing the female DB25 is letting enough noise get into the lines that SCK is no longer a solid, well-defined signal. That's just an initial guess, though. What I can't explain is why just touching pin 2 would make a difference.

If it's possible, try it with a shielded DB25 Female->DB25 Male cable; obviously the shorter the better, but the one I've worked with is 1.5m (about 5'3") and I haven't had any problems.

If you're trying with the hacked cable, make sure all pins 18-25 on the DB25 are connected to ground. This reduces interference between the signal lines.

Regards,
Craig Steiner