>      I need a very accurate linear position sensor, that can measure verticle
>      object travel of approx 1.5 inches. Cost is an issue. It will be an input
>      for a PIC16F84, so a digital signal would be less work.

Depending upon your accuracy requirements, a slide pot might work reasonably
well, combined with a resistor, cap, and a comparator (uses two port pins).
I have found this circuit seems to give results repeatable results of 1 part
in 4096, though readings take awhile (the faster the readings, the worse the
repeatability.  Also, the approach requires the use of an interrupt to poll
the comparator at 10KHz or so (the faster the better).  Unlike the normal
RC approaches, the comparator approach uses the pot as a ratiometric device,
so contact resistance should not be a problem.

Alternatively, you might try something a little different and use a casette
tape head along with a length of casette tape.  Using a manchester-style
coding, you could track absolute position to within 0.0025" or so provided
the tape was not moved too slowly or jerkily.  With a two-track tape and
somewhat better coding, the resolution could probably be improved to 0.001"
or less.