ADC Oversample with Scan - sanity check question

[snoopy20]

I just want to double check something before I optimise an implementation.

If burst is enabled on multiple channels, let's say x256 for two channels, internally is the process doing:

Read first channels x256 times then read second channels x256 times

or

Read first channel, Read second channel, repeat 256 times

The documentation suggests it is the prior method however the implementation under-the-hood would give the same outcome just with the reads being done in or out of order.

docs.nordicsemi.com/.../saadc.html

(Obviously I can just use the scan feature and am doing so but it's a memory hog.)

[AHaug]

Hi,

Could you specify which SDK you're working with? 

Kind regards,
Andreas

[snoopy20]

I'm not, direct manipulation.

[AHaug]

So nRF5/bare metal, then?

Kind regards,
Andreas

[AHaug]

Anyways, Read first channels x256 times then read second channels x256 times should be correct. See  RE: How does the SAADC "Scan" mode use the "BURST=1" for more details. Carstens answer matches my understanding of it atleast

Kind regards,
Andreas 

[snoopy20]

It's my understanding as well, but under-the-hood it might be doing each in sequence when multiple channels are enabled. It would need an engineer familiar with the underlying hardware implementation to verify.

[AHaug]

Hi,

Both quotes below, i.e "Read first channels x256 times then read second channels x256 times should be correct" and the comments from Carstein are coming from a discussion with engineers who are familiar with the adc hardware implementation on the nrf52810

Let me know if you have any follow up questions or if that answer keeps both of us on the sane side of the sanity check!

Kind regards,
Andreas