SAADC Cross-Channel Sensitivity Issue

RE: SAADC Cross-Channel Sensitivity Issue

Kenneth — Wed, 03 Feb 2021 09:41:06 GMT

[quote user="Atoo508"]As an aside, the sensors these inputs are to be connected to have an impedance of 1MOhm. For the nRF52810, is this to be considered a *very* high impedance?[/quote]

That is higher than supported yes, you can find a table here that show maximum source resistance for different sample time:
https://infocenter.nordicsemi.com/topic/ps_nrf52810/saadc.html#concept_qh3_spp_qr

For higher impedance I recommend to add a >1nF capacitance on the input, which will be able to feed the internal (~2.5pF) capacitor that is charged during sampling.

RE: SAADC Cross-Channel Sensitivity Issue

Atoo508 — Tue, 02 Feb 2021 19:32:10 GMT

Thank you for your reply Kenneth.

Yes, I did mistakenly strip out the while loop where the code resides and the power management is called. I can confirm the chip does go to sleep after each set of 400 reads (4 channels, 100 samples).

I took a look at the code from the ticket you attached above and applied it to my program, particularly where the m_buffer_pool[0] is used in the conversion function, instead of converting the result buffer. I also set the SAADC into single buffering mode, as with a 10ms sampling interval and 400 samples I don't think single buffering would make an impact.

I also checked the original configuration without calibration and the same behavior was still occurring.

It would appear that the buffer conversion being done with only one channel connected to something and the rest floating was causing strange behavior, both for sampling and calibration. I tested your solution, and the original config with other all other inputs connected to ground and the values were all as expected. In short, don't leave other channels floating... simple really.

As an aside, the sensors these inputs are to be connected to have an impedance of 1MOhm. For the nRF52810, is this to be considered a *very* high impedance?

RE: SAADC Cross-Channel Sensitivity Issue

Kenneth — Tue, 02 Feb 2021 15:07:17 GMT

[quote user=""]Could the issue be that the other inputs are floating when one is being sampled, or is there something wrong with my config and the way the data is sampled?[/quote]

Unless the sources you are measuring are very high impedance I don't think it's likely they will impact each other, you can for instance try to add a small 1nF on the individual input for test also. But if you want reasonable measurements you should connect the pins to something. I think your issue is a configuration or usage issue, for instance does it occur if you are not calibrating the input? You main() doesn't have an infinite while() loop or go to sleep, so I guess you have removed part the code here.

I did help a customer the other day with the saadc and calibration, maybe that can be of use here:
https://devzone.nordicsemi.com/f/nordic-q-a/70921/how-to-get-nrf_drv_saadc_evt-_calibratedone/291601#291601