Re: [avr-gcc-list] ADC anomaly

[Mattias Svensson]

>>The 3 LSB difference in your case could be
>>temperature drift due to the
>>device heating up during your several hours test.
>>

> I thought of that, but the temperature is pretty
> constant during the test. I'll check that out though.
> The resistors in the circuit are 1% so I wouldn;t 
> expect that much of an error. Seems that the system
> settles to 508 counts, but is 511 or 512 after a
> reset. Only a second so I don't think that temperature
> is a factor in such a short time with uninteruppted
> power.

Try using the ADC noise canceling techniques mentioned in the datasheet
if you need better accuracy.

Maybe you are doing something with the device peripherals that causes 
interferance on chip level...

/Mattias