It is found that the consistency of each power-on is not good NRF52832

[Francis Feng]

We used the ADC nrf52832 to collect the reference source voltage, 1.240V. It is found that the consistency of each power-on is not good, for example, sometimes it is 2806 after power-on and sometimes 2847 after power-on (stable for a period of time, after collecting the average value of 500 code values after filtering). However, the consistency of continuous power is acceptable, such as 2806 after a certain power on, continuous power, repeated measurement, are close to 2806.

What is the reason for this?

The gain used is 1/3. 2806 corresponds to 1233mV, 2847 corresponds to 1251mV, a difference of 18mV

[mathiaso]

Hi, Francis.

Have you tried running the calibration at start-up?

"The ADC has a temperature-dependent offset. If the ADC is to operate over a large temperature range, we recommend running CALIBRATEOFFSET at regular intervals, a CALIBRATEDONE event will be fired when the calibration is complete."

Best regards,
Mathias

[Francis Feng]

There was no change in temperature and it was calibrated. We did another experiment and collected an additional 2.48V benchmark. It is found that the power-on consistency of the two benchmarks is not good. Please help to analyze the reason.

[mathiaso]

Ok. Do you see this on several devices or only one device?

When performing the offset calibration, it might be that you end up on the border between two different steps if you see two different levels after power-on.

It could be an idea to test oversampling. If enabling oversampling on channel 0 before offset calibration, the calibration will run with oversampling. This takes a bit more time but will average out noise.