Wow. After Ian brought the discussion back to the crystal I was feeling a bit foolish for pursuing the MOSFET line of troubleshooting for so long, because 1) we've seen a similar issue caused by crystal & caps before, and 2) we all know "MOSFETs don't fail that way" right? Lol. I guess they do fail that way after all. I take no credit for the solution of the problem, I'm merely confused and amused!
Look into it later when the dust is clearing off the crater.
Just reviewing the "rmcrae" thread, and it appears that the MOSFET was the problem there as well. I recently changed to IRF640 from IRF840, and it appears to me that the IRF640 might be a bit more fragile than the bullet-proof 840. It has the advantage of a lower Rds, which means that it does not generate heat, but on the other hand, it seems to be more of a problem for failures.
I wonder if it is an ESD problem? I don't ask people to take any special precautions, but that might be a problem.