I'm having some issues with 10-bit A/D conversions that I've never seen before, and wanted to know if anyone on the list had ever seen, or would have any insight into...

I'm seeing a variable gain in the A/D result, apparently dependent on the frequency of the sampled signal.  I'm primarily interested in reading a 1 to 6 hz ac signal.  The ADC works perfectly at DC, and works perfectly above 20hz.

My test setup right now is the frequency generator and the scope are both connected directly to the AN0 input, on the PIC pin itself.  The only other thing on the board that is on this pin is a voltage divider, made up of 2 1k resistors to set my bias at 2.5v, and an open connector where the normal analog signal would be applied.
I'm sampling at 192hz, which is plenty slow, and time samples as follows:
turn on the A/D module, delay 24uS (calculated min. Taq of 11.5uS), do the conversion, delay 100uS, turn the A/D module off, store ADRESH and ADRESL, delay remainder of 5.2mS necessary for 192hz sample rate, and repeat.  I've also tried this with a full millisecond aquisition time, and a full millisecond to do the conversion, with the same results.  Also tried witl all possible ADC clock sources, with identical results.  Once I've read all 192 samples, I spit out all the sampled values via the serial port, and can capture and plot the data.

Waveforms look perfect, just the amplitude is off.  The gain that I see is anywhere from 2 to 5 times, depending on the frequency ranging between .5hz and 10hz.

I've tried this on multiple boards, with different PICs: 16F877A, 16F777, 18F452, even an 18F4431 which has the different A/D module from the dsPIC.

Anybody ever seen anything like this, or have any suggestions?

Any pointers would be greatly appreciated!

Thanks,
Mark

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads