Calibrate SAADC to improve gain error

[Ronald Reichmuth]

Hello,

according the datasheet the nRF52832 has an gain error of 0.02%/◦C and E_G1 of -3 ... +4%.
This is too high for our requirements. Is there a way to improve this error in production? eg. calibration?

BR

[Priyanka]

Hi Ronald,

Did you take a look at this anomaly.

Regards.

Priyanka

[hmolesworth]

Both offset and gain error of the SAADC can be calibrated at the present temperature by using a reference resistor of known value (eg 20k 0.1%) and connecting it to differential SAADC measurements first in one direction then in the other direction. This requires two pins for excitation voltage at each end of the reference resistor (or 1 pin if one end of the reference resistor is Gnd) and 2 pins from the physical ends of the resistor for the reversible SAADC connection, 3 or 4 pins in all, though with significant capacitance across the resistor and some loss of accuracy 2 pins only can be used (if short on spare pins) by first charging the parallel CR then measuring quickly; in the 2-pin case a higher value resistor is required to maximise the CR time constant. In both the 3-pin and 2-pin cases 2 analogue input pins are still required for true reversible differential measurement. The reversible differential SAADC measurement gives the SAADC offset, and the difference between measured resistance and known resistance the SAADC gain error using VDD reference; for fixed voltage reference 2 resistors may be required.

This post shows how; it's actually tailored to RTD temperature measurement but the RTD is simply a ratiometric resistance measurement using a  reference resistor. Using 14-bit mode it is possible to get an ENOB of close to 12-bits on the SAADC.

seeking-more-information-on-nrf52833-adc