>I use the method you described, charging the cap for several RC periods
>till it's pretty much flat.
>Then I enter the discharge timing loop, and all I can do is exit when the
>pic tells me the pin is low.
>The loop is two instructions at 1 MHz, and I'm getting timing values in the
>range of 0AXh for 10k, so I've got plenty of resolution, but I see it in
>the data, and on the scope, that the discharge termination point (and
>presumably the low threshold) is bouncing around like crazy.

Only difference is I used it the opposite way around, measuring the time it
takes to charge. Your method should be more accurate.

Actually I counted T0 interrupts, so my loops were very long and I used a
bigger cap, but that should make my results even worse than yours.

Don't know why it's not working for you, but you're smart so I imagine
you'll squash the gremlin.

Cheers,

Bob

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.