I don't disagree with the concern, but I'd be
interested in hearing
estimates on the probability of this taking place. Has anyone had it
happen to them?
I've seen it twice. Once on a rarely used machine on power-up, Once
on a machine in continuous operation. I doubt its more likely to
happen in one case (continuous use vs rarely used) than the other, it
just doesn't go unnoticed in the continuous use case. Then again I've
seen the mains side of a transformer short to +5 on (non-computer)
machine, also in continuous use.
For me, it always depends upon how easily I can replace the machine.
On a 286 PC clone no unique hardware, just flip the switch. On a
TRS-80 Model 3 you might want to turn it off if you smell smoke. On a
Sol-20, pull out the Variac and monitor voltage and current. On a
Compucolor II consider adding current and voltage limiting hardware.
On an original Apple I, microscopic inspection and every very low
voltage test you can think, take the old home-brew power supply and
put it in a display case, buy a brand new current and voltage limited
bench supply, and then recheck all the voltages and limit settings
before connecting. The closer it gets to one-of-a-kind, the more
careful you probably want to be.