I was thinking about this problem and I think I've come up with a
relatively simple way to synchronize two PICs (which could be extended
to as many as you would like):

Drive the PICs with an external oscillator, and also drive the external
oscillator into the clock input of a D flip-flop (for each PIC).  Both
PICs need to programmed with the same code (at least for the first few
lines), and preferably no start-up timer.  Dedicate one output for
synchronization, and the output is pulled low from an external resistor
so it reads low on power-up.  The first thing the program is to do is to
make it an output and set the sync pin high.  These sync pins are put
into the D input of the flip-flops, and the outputs (Q) are put into an
XNOR gate.  The output of the XNOR gate goes into the /MCLRs of the
PICs.   As long as the sync pins change to the same state at the same
time, the reset is not asserted.  If there is ever a point where the two
sync pins are not the same (on a clock rising edge), it is reset.

From this technique you would either synchronize really quickly, or the
chips would continually be reset.  If there is any randomness in what
phase of the clock cycle the PIC comes up in, it would oscillate briefly
until it found the right phase and then lock in once it was correct.  If
you ever wanted to make sure they were all synchronized again- change
the sync pin on one, which would force a reset on them all.

Comments?

Matt

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
"[PIC]:" PIC only "[EE]:" engineering "[OT]:" off topic "[AD]:" ad's