This post describes a "fix" for a poorly starting oscillator circuit. It MAY
be of help to others in similar situations (PIC or otherwise). I'd be
interested in knowing if others have had this sort of problem and/or used
this fix.

________________

I have had problems with oscillator starting on a Z8 processor. This is made
worse when the supply comes up really slowly, as it sometimes does in this
product for technical reasons.

The problem is limited to some processors from one date coded batch. I
needed a "fix" for this - replacement was not an option for reasons outside
my control.

As far as I can tell (from 1/3 of a world away) no other components have
changed.

The "normal" tricks of changing the size of the capacitors on either
oscillator pin, and adding parallel resistance and / or series resistance
all had some effect but were not 100% successful.

A fix that APPEARS to have worked is to add a 20 pF capacitor ACROSS the
crystal. While some ways of looking at the typical oscillator circuit
consider that the two normal caps are effectively in series and therefore
are equivalent to C1xC2/(C1+C2), in this case increasing each normal cap by
twice the parallel cap (40 pF each) did NOT produce the same result.

Using only a 10 pF parallel cap worked OK while 40 pF was too much. As the
parallel cap is increased oscillator amplitude decreases. 20 pF seems a good
compromise in this case.

The Z8 has an internal oscillator feedback/bias resistor and no external one
should be needed. Interestingly, the bad batch of CPUs measures about 500 k
between oscillator pins while normal ones measure about 900k to 1000k.
Despite this, adding a further parallel R actually helps.



        Russell McMahon

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics