questions around SAADC offset calibration

RE: questions around SAADC offset calibration

Jørgen Holmefjord — Thu, 20 Jul 2017 13:25:46 GMT

In this test you will be dominated by gain error, not offset. The gain error will still remain. If you want to check how the offset changes with offset calibration then you need to apply 0 input. Do you get more stable values as you increase the oversampling?

It's interesting if the gain error changes offset calibration, because it should not. To be able to separate gain and offset error you need two points, 0 and 1.8 V input.

RE: questions around SAADC offset calibration

peter — Wed, 12 Jul 2017 15:57:35 GMT

Hi! thanks for the quick reply. Oversampling does not seem to help. I ran test series on some of the devices with 4x, 32x, 128x and 256x oversampling. The difference from 4x to 128x was at maximum around 2 points on the adc (4096 scale), whereas the offset calibration itself, still went around 50 points further from the non-calibrated average to a total of around 90 points off of the actual value (the worst case precision of the input voltage is 15 points on the adc scale).

Actual values: 1.8V reference input should be 3072, non-calibrated value e.g. 3030, calibrated value 2980.

I also tried doing calibration once and never again, and also after every sample (time between samples was 3 seconds) which leads to the same results.

RE: questions around SAADC offset calibration

Jørgen Holmefjord — Wed, 12 Jul 2017 12:21:19 GMT

This could be caused by noise. Can you try calibration with OVERSAMPLE = 5 (32x)?