> I'm curious - what brand/manufacturer of crystal?
>
> Have you had an opportunity to 'characterize' the crystal
> (measure it's parameters) on an xtal test set?
>
> (Perhaps a lower-than-average-Q 'rock' in conjuction
>  with an oscillator whose gain is in the low end of the
>  acceptable spectrum is the overall problem ...)

A change in crystal characteristics was one of the things high on my list of
suspect candidates.
Unfortunately, it was as I said
    > > an otherwise intractable problem.
ie I had tried changing everything else in sight (and a few things not in
sight) including the crystal. The change occurred with processor date code
change. ALL devices prior to this date code work 100%. About 50% with new
date code fail. Manufacturers (in Taiwan) assure me that the crystal type,
manufacturer etc has not been changed. I was sent 11 sample boards with the
fault. Changing the crystal to ones available locally (in New Zealand) does
not alter result.

The problem APPEARS to be related to the processors sensitivity to latch up
during power on if certain voltages are present on pins when Vdd is applied
with certain (or uncertain :-) ) rise times. Suffice it to say that
everything I am doing is inside datasheet spec but processor does not always
start (whic is ouitside datasheet spec :-) ).

Even clamping all (sensible) pins to Vdd with Schottky diodes does not
prevent maloperation.
Once in the errant mode the oscillator does not start and the processor will
not return to normal until Vdd drops to below about 0.1v. This can take some
10's of seconfds in a normal circuit as once silicon junctions stop being
biased on current draw is due almost only to purely passive loads.

The 22uH made all but one of the samples work OK - the last may have other
problems or be an especially bad example.



        Russell McMahon

--
http://www.piclist.com hint: To leave the PICList
mailto:piclist-unsubscribe-request@mitvma.mit.edu