Martin Peach wrote:

>A CMOS 555 works well, temperature drift may be an issue. The input voltage
>has to be kept in the 1/3 to 2/3 Vcc range. You can get very high
resolution
>on slowly changing signals such as temperature by counting for longer.
>Accuracy depends on the linearity of the VCO transfer function, which can
be
>corrected for if necessary. For higher sampling rates measure the width of
>single pulses.

I have seen 555 but I wasn't very convinced on his precision about
voltage/frequency conversion on this chip (applying a voltage to the CONTROL
VOLTAGE input, pin 5, ok?).
However, i will do some tests on it: i have only a pin free (RA4) and this
solution is compatible.

Marcelo wrote:
>>I would use a LM331, Voltage-to-Frequency converter.

It could be very good for my use, but the price is significantly over the 3$
(in Italy) and isn't very easy to find: however, it seems a bit more
accurate than 555 solution.

To Vasile: many thanks for your hint!  When I'm started in pics I have read
the Microchip App.Notes on this argument, but I supposed that  4 pin were
needed for this kind of measure. Also the code related was too "big" for my
program space.  My actual problem is that the '84 program memory is VERY
near to full and olnly RAa is free, because I use it also for other types of
frequency sensor (example: a 4093-based circuit for water conductivity
measurement).
I think that the code that you have suggested (very interesting) isn't
possible to merge in my current logger code: it is probably the right moment
to start in 16F87xx.

Many thanks to All for suggestions

Marco

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics