The ATmega328P data sheet in table 29-11 on page 324 notes that the internal 1.1V reference (VBG) may vary between 1.0V and 1.2V. In this answer this is called "not stable or accurate".
Question: Is this "unstable" or "inaccurate" or even both if I define it as
- "unstable" if it varies over the day, say with temperature or VCC or or some such.
- "inaccurate" if it is a rather stable voltage for a specific Nano-board I have in hand, just not the typical 1.1V.
In the latter case I could measure the value accurately and scale my analog measurements with the measured value instead of the 1.1V.
1 Answer 1
I would say it was "stable but not accurate". The internal reference uses a bandgap voltage reference, which is pretty stable over the operating temperature range of the IC.
The output of the bandgap reference is fed through an amplifier to buffer it, which may introduce some instability with temperature, but this is not documented so cannot even be estimated.
The inaccuracy comes from minor differences in the manufacturing process. These don't change over time, so once calibrated for a precise value of voltage reference you should not see much in the way of variance in that value for different temperatures or power supply fluctuations.