Ansel Sermersheim wrote:


>>>>>> "Peter" == Peter L Peres <plp@ACTCOM.CO.IL> writes:
>
>> imho, you can obtain 20 usec loops relatively easily with a PIC at 4
>> MHz for transmission ( == 50 kHz), but for a decent receiver at that
>> frequency you'd need to use a faster clock. What are your trying to
>> modulate ? 4-bit voice can be passed through such a channel but
>> don't tell me about the quality (you'd use crude ADPCM at 8
>
>I'm using FSK to pass digital data.  I'm only going to be using about
>2400 bps, with two filters to receive.  I can easily poll at 415 usec
>to read that.
>
>However, it would be a lot easier if I don't have to rig something
>similar to that for xmit.
>
>> Peter
>-Ansel


The piezo and film devices normal require a high voltage, but very little
current. The current requirements from the raw supply normal requires you to
use an off chip transistor (NPN or N-Channel MOSFET) to switch the current.
Use this to drive a transformer single ended. Tune the transformer leakage
inductance to be complementary to your capacitive piezo device at your
center frequency, this can provide a large voltage boost at high
effieciency. It is not hard to produce 600Vpp this way.

If I was trying to receive FSK I would use a simple bandpass filter wide
enough for both frequencies and then feed that into TMR0 or TMR1. Every
1/(3*2400) seconds read the change in the counter and just compare against
the center frequency to determine if the input is a 0 or 1. To synchronize
with the start pulse you would expect to see the mean frequency on the pulse
which contains the edge, if it closer to the 1 level then shift your phase
back in time, and if it is closer to the 0 level then shift it forward in
time.

Chip Weller