Was it in here and was it svart who cautioned about using PICs sometimes for critical control functions in realtime apps, because of some crazy initial condition glitches?
I have steered clear of micro programming, preferring to leave it to experts (and I have a friend who is very very good indeed), and wanting to not be the pacing item for every bit of hardware on projects. But now and again I hear the siren song and am tempted.
When my friend and I hacked out a very intricate design with an elaborate (for a powered speaker system anyway) control surface on the right satellite, we used two PICs---iirc 16C621 parts, with extensive command, power and speaker/headphone signals all travelling on a total of four wires (some of the details of that system are in patent purgatory right now in fact).
Despite flagrant violation of one of management's prime directives, namely "No Invention on the Critical Path", everything worked like a charm, almost miraculously, but flux contamination of some units' teeny TSSOP package parts in the sats caused the clock oscillator to get biased into a region of low gain, until the machine was on for a few minutes. Not good.
While we were all trying to figure out what was going on, the Chinese subcontractor making them did some of their own research, and discovered that the crystal required a negative resistance to oscillate, and wondered where they could purchase negative resistors :razz: