Like I said the fluctuation was in my circuit. So, I fixed that problem, but now I'm not getting an accurate ADC value. It is off by 4% even with the calibration bytes used.

Ok, so what do your calibration bytes read? Have you worked through the calibration algorithm manually and checked that it matches up?

Yeah, stepped through the calibration and its applying the bytes per the algorithm I duplicated from the App Note. So, I kind of question the algorithm itself. It should be increasing the ADC count to get to the right value, but in one part of the algorithm it is actually subtracting/reducing the value.

I'm using a z8F2480 and had good results with my ADC, but I was using an external vref. We have some voltage dividers to detect board rev and current role. I found my calculations work out to within 1/2 counts of the measurement. I'll reread my specs, I didn't notice an ADC calibration step.

However, when I tried to apply the calibration bytes algorithm for the internal temperature sensor I couldn't get reasonable similarity between any of the 3 devices I tried.

I'm using the internal 2.0V reference. Maybe I should try an external Vref. I might try my 3.0V supply voltage.

I see that the Z8F042A has a little different description for the ADC in the Specifications than the Z8F2480. I'm going to read that and the app note and see if I get any new inspiration.

My desire to use the temperature feature was a minor feature so I abandoned it. I'd still like to conquer it.

If your test code is short, try asking support to explain the discrepancy.