And what fascinates me is that the issue keeps arising, regardless of the steps people take, though often so seldom they're convinced it's "cured."

Do you have a logic analyzer with enough channels to capture all the pins on the MCU? If so, a "picture" of what happens when you touch the crystal, which action, BTW, may not always cause the flash corruption that you suspect, might shed light on what's really going on.

I've long suspected that there's something in the common architecture that permits the core to run on internally during active RESET when power is out of spec. Now, that may not be the case with your experience, but it would be good to know what RESET, among other signals, are doing during this "blessed event."

My own experience with 805x "RESET issues" began with an MCU using external program store consisting of a BBRAM, and I was able to recreate the event from time to time, but not consistently. There was a "supervisor" chip in the circuit and when it triggered its RESET output probably due to decaying Vcc, it became noticeable that addresses and nRD, nWR, and nPS were occasionally active despite the active presence of RESET. I found that this occurred with old Intel parts, old AMD parts, and newer Maxim/Dallas and Philips parts, as those were what I had on hand, actually, on the work surface.

Not everyone believes that placing a "supervisor? IC in the circuit is a solid fix for the resulting program-store corruption, though most of those you'd be likely to hear from are persuaded that that is a solution to the problem. The consensus is, however, that all that's needed when you have flash corruption is a supervisor that watches the level of Vcc. I am not among their number, however, though I've had little recent opportunity to pursue this matter.

If you can capture, even with your oscilloscope, a picture of the behavior of RESET and Vcc when this sort of thing occurs, it might shed light on the matter. I don't know how much your finger might have loaded the Vcc line. Maybe the flash corruption is something as simple as forcing an accidental brownout.

It's always good to know what sorts of things will cause your system to "break".

RE