Implement ADC self-calibration

[jcard0na]

Implement the ACD self-calibration procedure, which is defined in the RM377 reference manual as follows:

The ADC has a calibration feature. During the procedure, the ADC calculates a calibration factor which is internally applied to the ADC until the next ADC power-off. The application must not use the ADC during calibration and must wait until it is complete. Calibration should be performed before starting A/D conversion.

[jcard0na]

Thanks for looking at this, @jamwaffles

Looks good but I'm a bit hesitant about the 1000 loop iterations. Would it be better to use a while loop and make the assumption that the hardware will always clear ADCAL?

Yes, I'm not too proud about that loop, but also I'm kind paranoid about introducing potential infinite loops so deep down. If you are happy with it, I'll change it to a while loop. At least on my hardware ADCAL seems to always be cleared.

Also, the manual says:

Calibration can only be initiated when the ADC is disabled (when ADEN = 0)

Is that ensured in the context of calibrate()? I'm not familiar with the ADC code in this crate.

Yep, I'm checking for that here: https://github.com/stm32-rs/stm32l0xx-hal/pull/233/files#diff-09d1f58a7f7c271d63f1a9a00c0c48c886e094bc98c207a8d7688c9f874259bbR132

[jcard0na]

By the way, on my STM32L072, the ADC errors dropped from 6% to 0.05% after calibration(), so it is a big deal :straight_ruler:

[jamwaffles]

Let's go for a while loop... we can change it if someone complains 😁

Yep, I'm checking for that here

Ah thanks, I missed that... 💤

[jamwaffles]

By the way, on my STM32L072, the ADC errors dropped from 6% to 0.05% after calibration(), so it is a big deal 📏

Nice! It's great to validate these changes, and that's a huge improvement.

[jamwaffles]

Thanks!