Dear Forum:

I have built a power subsystem using the ISL28022 digital power monitor. I can read the bus voltage very accurately but I am seeing a scale error in the current sense value of about +7%. My sense resistor tolerance is 1%, so I expected no worse than 1.5% performance.

In my application, Vshuntfs is 40mV, Rshunt is 4 miliohms so the computed full sale current is 10A [Eq.2.]and the computed CurrentLSB is 0.000305 A [Eq. 3]. This leads me to compute an integer calibration value of 33574 or 0x8326 [Eq. 4]. I realize that the value written into the calibration register is only 15 bits because D0 is set to zero.

My question is what could be causing my scale error of about 7%? If I scale back the computed calibration factor by 7%, I get very good current measurements. A couple of thoughts come to mind. Could a a non-optimum Kelvin connection between the sense resistor and the VINP/VINN amplifier connection cause this level of error? Or could my routing of the main current path in and out of the sense resistors explain this kind of scale error? Lastly, as shown in the partial schematic below, could the filter R4/R5 series input resistors that are part of the filter affect the scale factor computation?

Thanks for any guidance!

Craig