I'm also about to tinker with the touch-switch circuitry in the 16F1936, which AFAIK is the same module in the 16F726/7. I attended a Microchip embedded dev seminar a couple weeks ago and played with some of the demo units they had at the seminar and it was very reliable, and I tried it with some sheets of paper in-between my finger and the touch pads, to simulate an enclosure/cover. No issues there either.
Apparently it will require "tuning", but not really for each individual unit if the units are built consistently, but more to tune for the "environment" (surrounding circuitry, enclosure, touch-pad specifics, etc). One great suggestion is to have one extra similar but unused/unuseable dummy switch, that will serve as a baseline for the other switches thresholds -- essentially a "control" in science-experiment terminology.
However, I do intend to test my app across several units to see how the thresholds vary due to variations in component tolerances, user positioning, etc.
Cheers,
-Neil.
