Timothy J. Weber replyed:

Wild guess: When it's waiting for a button press at power-on, some
timers are running.  Once you start the first step, maybe timers and
interrupts get synchronized for eternity, and if you're lucky it's in a
good way that masks race conditions, and if you're not, you... lose the
race.

Can you look at a call tree and determine whether there could be a stack
overflow during interrupt handling?  Or are there other possible race
conditions between main code and interrupts?

Can you hack it to remove the button presses at power on, so that timing
is standardized, and if so, does that change anything?
-- 
Timothy J. Weber

Russ then says:

There are no intentional interrupts, interrupt routine is just a 
return.  I have not done
anything to intentionally enable or disable interrupts.  I have stripped 
the code down
to straight line thru it with not button presses, just some blinks then 
infinite loop of stepping:
still erratic.

Russ
-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist