[Alan R. complains about static causing resets in a PIC]

I had the identical fault happen in a product I did a few
years ago. It too was reset-intolerant, since it was a timing
application that relied in part on a timecode burned into
the PIC's ROM. Once 'born', a device couldn't be reset without
returning to its 'date of birth'......which took it out of
sync with 'real time'. When I talked to Microchip about it, they
spent some time looking at the problem in detail.

The conclusion was that when a static zap hits the CPU, it
causes currents to flow across the die. The differential
voltage across the die is enough to trip the reset circuit
and cause the usual reset sequence. It only takes a couple of
nanoseconds....but that's what you get from direct static
discharge.

I solved the problem by shipping the product in an anti-static
bag, packed in individual cardboard boxes. It was primarily
a shipping-related problem (styro-peanuts, etc.) so this was
enough to get me out of the woods. Besides, I had 11,000 of them
already built, so a redesign was out of the question.

In the course of studying this, I determined that you could
minimize the problem with the usual ESD control measures. Find
a way to suppress that static zap, or route it somewhere else.
Copper foil tape works well; you can get it at your local
stained glass supply company, where it's used for joining the
glass panes together. Check with Horowitz & Hill (Art of Electronics)
for ESD control ideas, as well as Compliance Engineering
magazine. "Grounding and Shielding Techniques in Instrumentation"
is another good book to look at.

You're probably in the unenviable position of having a bunch
of these widgets already built....in which case you have my
deepest sympathy. In that case about all you can do is experiment
to find a way to shield the PIC. Don't bother trying to tie the
reset pin to different potentials; I tied it to VCC with a
jumper wire (A short one!) and it still hiccupped on an ESD
hit. Unless you can find a way to keep the static out of the
part, you're stuck.

In defense of the PIC, I applied some *really horrible* zaps to
the pins.....they'd reset, but the parts always worked afterwards.
When Microchip says that their parts are ESD-hardened, they
mean that they'll *survive* ESD hits.....not that they'll
keep running flawlessly *through* them. Contrast that with a
486 chip.....where if you look at it funny it'll blow up.

Mike Jedrzejewski (Jed-er-eff-skee) at the Microchip San Jose
sales office was a lot of help in tracking this down. I think
he's lurking here somewhere. (Hi Mike!)

Good Luck! (Yer gonna need it!)

forbesm@csos.orst.edu
Mark G. Forbes