We have been characterizing the first 25 boards of a small pilot run. We us the CC2541's ADC to measure the supply voltage (using the internal AVDD / 3 channel) and current across a 2 ohm resistor. The ADC We use the internal 1.24V reference and operate in 12-bit mode (single-ended so really 2047 counts) - our code is very similar to HalAdcRead().
We have a test setup that can sweep the supply voltage and record the ADC measurements of the supply at each dwell point. From this, we construct a least squares linear fit to the data. With these first 25 boards, we have observed what appears to be an ADC input offset voltage adding 15mV - 50mV to the measurement (these values are with respect to 1.24V, not 3*1.24V). This seems very large, variable, and unexpected. Can someone explain this?
Additionally, the current measurement channel has a voltage offset that is about half as great as the supply voltage channel, however it is sometimes additive and sometimes subtractive.
Are these results to be expected from this part? Am I misunderstanding something about the ADC specifications?
Thanks,
Mark