Erik Reikes wrote:

> It seems to me with such big load caps and the additional load of the
> resistor that maybe its having trouble starting out.

  10 ohms as specified is negligible.  Recall the DC impedance of the
OSC2 line is already well over 50 ohms.

> I seem to remember somewhere reading that the load caps should be 2x
> the capacitance of your crystal, so mine should probably each be
> 20pF's.

  Twice the specified load capacity of the crystal (because they are
effectively in series) minus the capacity of the PIC pins (which are in
parallel with the load caps).  *This parameter isn't specified*!  The
latest revision appears to be DS30430C on the website, and unless
they've altered it without changing the revision code, this is stamped
"Preliminary" on the specs sections.  Will this persist while the 16F84
becomes a "legacy" device like the 16C84 whose datasheet *has* been
removed from the website?

  Assume about 10 pF (may be excessive) for the pin capacitance.  If
crystal loading was 16 pF, twice this would be 32 pF, drop 10 and you
get 22 pF.

Richard Prosser wrote:

> Try adding a high value resistor (1-10MEG) across the xtal. This tends
> to linearise the amplifier slightly, allowing a noise build up that
> starts oscillation. (or something).

  But this is supposed to be provided in the chip.  Indeed, it is
purported to be the *only* difference between LP, XT and HS modes, so
the same effect would/ should result from selecting HS mode.  Some have
actually suggested this in the past!

> We do it anyway and the accepted reason is that it speeds the start-up
> of the xtal.

  It reduces the "Q".  This is the same thing as speeding start-up.

> Also - 47pF caps is a bit strong - try 22 or 10pf instead.

  Sounds good.
--
  Cheers,
        Paul B.