I've been thinking about a capacitive sensor.  You might use a couple of
conductive parallel plates, with the mass changing the spacing.  I'd probably
use the capacitor in an oscillator, and then count pulses with a PIC.

Then you'd have to correct the non-linearities, either in software or in the
oscillator.  (Maybe a 1/n transfer-function V-F converter?)  It appears the
resolution would be a lot better at one end than the other.

I imagine it would be pretty drifty with temperature.  Mechanically, it could
be simple and rugged.


Then again, all the good scales I've ever seen were based on either load
cells or quartz crystals.


later,
newell