I built the calibration tool mentioned and prototyped on the previous log, and it is open source and documented on another Hackaday project here. This calibration tool can provide a precision current supply to MetaShunt, which we can use for calibration and performance testing.

I recently analyzed the accuracy of the precision current supply here, and showed that the accuracy is quite good, typically better than +/-0.3%. A calibration script using a fixed reference device was developed and used on MetaShunt V2. This script is open source and listed here.

This calibration script was used on a MetaShunt V2, and then several days later a calibration test script was developed here, and tested with that MetaShunt and an adjustable reference voltage configured current supply. This was a physically different current supply. For this test, 75 different current commands were tested, from 50nA to 250mA, distributed linearly on a log scale. For each current command, I measure for 1 second and compute a mean and standard deviation of the roughly 6,000 measurements that are received back from MetaShunt V2. I then plotted the error of the mean measurement and a 2-Sigma error for each. The plot below shows the accuracy across the current supply scale. Across the entire scale, the MetaShunt V2 provides highly accurate mean measurements, within +/-0.6% of the commanded current. Combined with the +/-0.3% error of the current supply, this indicates better than +/-1% absolute accuracy for 1s mean measurements across 50nA to 250mA. At very low current, the standard deviation varies significantly, likely due to the high impedance of the measurements at this scale.