Thanks to Bob, Olin, Lawrence, Mike and Vasile for the replies & thoughts.

Several mentioned latchup as the likely cause of the device failures/damage.
I tend to agree, in fact this was the opinion of this customers' lead
engineer.

As to the possibility of using the Power-OK signal from the power supply,
unfortunately this will not help if the problem is latchup (unless all power
and I/O signals were switched on when the OK signal was active - requiring a
magor redesign)... it would help if the cause was bus contention, but that
would require a significant design change to a board that is past it's prime
and only has this issue with one customer. With our current budget and
manpower issues, a redisign is very unlikely.

A summarry of the more common triggers for latchup follow, feel free to add
to this if you are aware of others.

1.) Supply voltages exceeding thier absolute maximum ratings.
2.) Input or output pin voltage exceeding either supply by more than one
diode voltage drop.
3.) Improper power supply sequencing.
4.) Transient irradiation; gamma rays, x- rays and other ionizing radiation
(such radiation can induce photocurrents which can active parasitic SCRs
within the device).
5.) Electro Static Discharge to an I/O pin or a trace/device connected to an
I/O pin (see numbers 2, 1).
6.) Inadequate power supply filtering or transient protection (see number
1).
7.) Coincedences of PCB track layout coupling voltages or inducing currents
that result in overvoltage to an I/O pin or supply pin (see numbers 2, 1).

It may be possible to add overvoltage protection and power filtering to the
system as some suggest, I'll look into this. (I was surprised to read the
degree of off voltage that Vasile mentioned!).

Thanks again.
Bruce.

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.