What external capacitor? Is one required for this to work reliably?
Many recommend installing a .1ufd cap between ground and the Aref pin. It can help remove a little noise from the reference voltage. Wouldn't account for the error size you are seeing.
Can anyone who was worked with this cast some light on it please? Thanks very much.
You didn't post your code so I don't know what to say about the error size you are seeing, other then it seems too high to me. My first question is what constant value are you using for the 1.1vdc reference in your sketch? 1.1vdc is a nominal datasheet value ( can vary from 1.0 to 1.2) and every chip will have a certain variation around this value, but still pretty constant. Coding Badly had a method where you could measure the actual band gap value for your specific chip by measurement of the Aref pin with a decent quality DVM, after first selecting it as the reference in a small 'test' sketch. I choose to use a back 'door method' where I powered the chip with a known and accuratly measured Vcc and keep 'tweeking' the bandgap constant value in the sketch until the calculated and displayed value did indeed match my measured Vcc. After that I checked that the 'dynamic calibration' value used to correct analogRead() values were accurate as the Vcc was adjusted downward. I was pleased with the results but sorry I don't recall the actual accuracy value, but I'm pretty sure they were in the 1-2% max error range.
The Bandgap value I ended up with for my specific Seeeduino mega (1280) board was Aref = 1115 millivolts and my Arduino 328p board was Aref = 1056 millivolts. Once I arrived at those specific values I was quite pleased with the accuracy results. I didn't write down the accuracy result at least that I can fine, but possibly I might have mentioned them in the long thread that Coding Badly and I were work with?
So what bandgap value did you use?
Lefty