Hi,
we implemented the INA226 power monitor into a measurement application. The shunt input has been
been equiped with the low-pass R=10 Ohm, C= 100 nF (as recommended at page 12, in fig. 22 at the data sheet).
Regarding to the datasheet, selecting there resistors has minimal effect to the accuracy.
The datasheet provides, that the max. shunt voltage gain error is 0.1% of the fullscale.
Our measurements exposed errors > 0.2% FS.
A failure analysis has revealed, that there is a not negligible voltage drop at the resistors.
Based on the assumption, that the input impedance of the shunt input is the same as the Vbus input impedance 830 kOhm there should'n be any significant voltage drops at the LP-filter resistors.
The shunt input impedance seems to be around 12k but there is no specification relating the shunt input impedance.
This leads to the conclusion, that the Vbus and the shunt input impedance are identically.
Our problem is, that even long wires from the shunt resistor are affecting the measurement accuracy of the shunt input.
We performed the reference measurements with a Fluke 715 voltage calibrator and a Keithley 2001 DMM connected directly to the
LP-filtered input of the INA226 shunt input.
Originally we decided to use the INA226 power monitor for our measurement applications due to the fantastic A/D specs and so a
calibration of seemed to be unnecessary, because we wanted to avoid a calibration in any case.
Could someone on part of TI please confirm, that the input impedances of Vbus and Shunt are different and why this particular information has not been published?
Thank you.
Dimitri