Sean Breheny wrote:
<<cut>>
> What I AM surprised at (or at least was surprised) is that the
> oscillator SEEMS to continue to run (although with a different DC bias
> level) during sleep (which only lasts 50uS,which seems about right for the
> conversion length).

The data book says that the PIC stops driving the oscilator pin during
sleep.   For a CMOS technology this means that there will be very little
loading on the crystal from this time onwards, if you have a good
osciliscope then as you say, 50 uS may not be sufficient time to
significantly damp the oscilations.   The power up timer bit in config
gives 72 mS before reset finishes, 144 x more than your sample period !

regards,
Graham Daniel.