|Know, that this is a very cheap single ramp converter.
|It needs to be calibrated often against the internal refs, and mchip have
|already programmed the calibration data into *EPROM*.

One problem with the ramp converter on the PIC14000's from what
I've heard is that like all latched-count single-ramp convertors
it is extremely sensitive to input noise.  A brief glitch on the
input may cause the A/D to latch its count prematurely.  An alt-
ernative A/D technique which I've used with good success is app-
licable to any PIC with the addition of a simple comparator chip,
a resistor, and a cap.  [Details of technique discussed in other
emails before].  While my technique is a bit slow to respond to
rapidly changing signals, it has excellent noise immunity and its
accuracy is dependent only upon the following factors:

[1] Offset errors or (to a much smaller extent) delayed response
    in the comparator.

[2] The stability of the PIC's "high" and "low" output levels.  If
    these levels are unstable (e.g. because the PIC is driving high-
    current I/O devices) an external chip with good 5.0 and 0.0 volt
    outputs may be used.

Note that the device does not put any significant load on the PIC
port output that it uses, nor does it depend upon the accuracy of
R and C (a smaller RC will result in faster readings but add some
"random" error to the readings)  The one drawback is the necessity
of a uniform periodic service routine.

Has anyone else used my A/D method?  Any results good/bad?