On Mon, 30 Apr 2001, Russell McMahon wrote:

>  What sort of caps are you using?
> Presumably Mylar or similar rather than eg ceramic.
>
> Does transferring this from the ICE to a "real" PIC produce an identical
> result?
>
> Presumably when you say "non-linearity" you mean non linear compared to an
> exponential curve.
>
> Might it help to eg go low, then high, and time for say 5 time constants
> after the low to high transition and then go low again and count the time
> taken? That way, even though the 5 time constants "should" pretty much have
> the cap at top rail, driving it there for a consistent period after it
> passes through low-high threshold may reduce variability.

Good points.

I'm curious how this "2-instruction loop" is implemented. How can this be done
AND count the number of cycles? I know one way, but from the context I'm sure
that's not being done.

To achieve the +/- 2% repeatibility to which Bob alludes you'll have to be
extremely careful in keeping track of ALL cycles and ALL voltages.

First the voltages: you've got ground, Vcc, and the two switching thresholds of
the I/O pin (the low-to-high and the high-to-low).

Others here have more experience than me with the PIC I/O thresholds. But one
thing I'd comment on is that the thresholds are sensitive to variations in
Vcc. If the sensitivity is linear then this approach is ratiometric. But, I
don't know how the input thresholds vary. It'd be instructive to create a simple
program to measure the input voltage thresholds:

l1:
    bcf   ioport, bitx    ; The ioport we'll monitor
l2: btfss ioport, bity    ; the ioport/bit we'll test
     goto l1
    bsf   ioport, bitx
    goto  l2

Attach a variable power supply (or a the wiper of a pot) to the ioport you wish
test. Run this program and monitor the other I/O pin.

This will give you a good idea of the thresholds at a given vcc. In your lab
notebook, you may wish to record the results for a few different Vcc's
(e.g. 4.5, 4.75, 5.0, 5.25 etc)

-----------------

Now, the algorithm I'd take is this:

1) Charge the cap. You can either do this through your unknown resistor or
tie the cap to another I/O pin that can quickly charge the cap. The latter
makes more sense since you can use the "discharge" I/O as your input to monitor
the capacitor voltage, so I'll assume that in what follows.

Eg:

        |
PIC I/O +----+------------+
        |    |            |
        |    |            /
        |   ===  C Ref    \ R unknown
        |    |            /
        |    |            |
            ---          ---
           ///          ///


2) Begin charging the cap. Make sure it's fully charged! The pic I/O pin has
fairly low output impedance, so should charge the cap in a much shorter time
than your unknown resistor.

3) At the same time you change the output pin that was charging the cap to an
input, start a timer. Note that the I/O pin that was charging the cap is now
monitoring the cap's voltage.

4) When the PIC I/O goes low, stop the timer.

------

Sounds easy enough.

But to get repeatable results you'll need an accurate counter. I don't know
which PIC you're using. If you're using one with the Timer 1 peripheral (like
the F877) then the timer is trivial. If you're confined to the '84 then you'll
probably want to grab my 3-instruction cycle resolution software pulse width
measurer. It's part of the gpsim distribution and can be found in CVS:

http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/gpsim/examples/14bit/

the file is pulse_measure.asm


Scott

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email listserv@mitvma.mit.edu with SET PICList DIGEST in the body