I am looking for more insight into how the AD7746 performs its offset calibration. I am using the EVAL-AD7746 board and I notice that when I perform 2 sequential offset calibration procedures without moving/touching any part of my test setup I get two different values. For instance, after my first offset calibration the registers return 10000111 and 10001010 respectively and after the second offset calibration, the registers return 10000111 and 10001110 respectively. It's not a huge difference in terms of a difference between measurements made with the first calibration vs the second calibration, however it got me thinking. The spec sheet says to get 4 aF resolution, 1000 data points need to be averaged at the longest conversion times (~227 ms). It only takes at most a few seconds to complete an offset calibration. So:
Is the AD7746 not making the offset calibration at the highest resolution that it can possibly measure?
How many conversions does it make when performing an offset calibration?
To get a more repeatable offset calibration, should I make 1000 uncalibrated measurements, average them, and then convert that parasitic capacitance to binary to feed back into the offset calibration registers?
Thanks in advance!
