I have a product that currently uses the 16F84. I am getting field failures
where the data EEPROM has changed (it contain calibration values written at
test). The product previously used the 16C84. I have seen no instances of
the 16C84 spontaneously changing the data EEPROM value.

I can duplicate the failure in the bench by inducing a brown-out condition
in the 'F84. I do this by starting the circuit normally, then letting the
voltage decay (but not to zero), then raising the voltage again. It does
not suprise me that the chip might malfunction under these conditions, but
two things are evident:
1. There is a high probability that the data EEPROM will be written.
2. The watchdog does not reset the chip, so it stays in the malfuncting
state until power is removed.

Of course, the mechanism of failure in the field might be a different one...

Has anyone else seen performance like this? It looks like I'll have to add
a power supply supervisor (the Dallas DS1233D looks like a good candidate).
It is interesting that this problem only started with the "F" series. Does
anyone have any insight on why it might happen with 'F84s but not 'C84s?

Regards, Jim
Jim Ham, Porcine Associates
(415)326-2669 fax(415)326-1071
"http://www.porcine.com"