A circuit I'm working seems to have data corrupted by noise
caused by the PIC's PortD output occassionally (intentionally)
changing from 11111111 to 00000000. Traces and wires in
the vicinity of the port show a simultaneous 0 -> 1 blip at that
instant, which might be interfering with another data source.
This document shows why it happens

http://www.altera.com/literature/wp/wp_grndbnce.pdf

Perhaps something to be aware of when bug-hunting

(thanks to PIC-lurker Brian for suggesting it)

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