During our testing, the PIC18F258 ADC is very odd. The big problem is that
there is alway a big offset. For example, the expected value should be
0x200, but the data read is 0x210. The ADC is a 10-bit converter and we use
20 Mhz oscillator. We just connect the ADC with a power supply in a constant
voltage for the test purpose

We are using the PIC with external mux's and internal mux's.  There are four
A/D pins on this chip, one designated for high res RTDs, one for low res
RDS, one for TCs, and one for analog inputs.  The software is writen such
that all 20 input channels are scanned in sequence.

I took a voltmeter and oscilloscope and verified that the signals to the A/D
pins were clean.  However, during this multiscan mode, the A/Ds read
significantly higher than they should have based on the coresponding input
voltage.  This can be anywhere from 5-10 bits off to as high as 50 bits off!
Obviously completely unacceptable for accuracy reasons.

We have alot of experience with A/Ds and immediately looked at channel
settling times.  We were allowing 3ms for the signal to settle, at least 2ms
longer than required.  The signals in were very clean, and we run an
averaging routing in software.  Even with all of this, still the terrible
results.

If we disable the multi-channel scan, and run the chip slowly, accuracy
improves greatly.  We have seen accuracy to 1 or 2 bits using a calibrated
voltmeter to verify.  However, with no input changes, the minute the multi
channel scan is repeated, we immediately get wild, yet predictable,
readings.

We basically hardwired a voltage source to the A/D pin, eliminating any
question of external settling times for that channel.  Still, when multi
channel scanning is enabled, we get horrible results.

We see the problem on the emulator and in actual hardware.

Please advise where the problem is?

Thanks in advance

Luke

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