# Thermal EMF on shunts

I tested the accuracy of a shunt resistor of 20 μΩ by applying short pulses of 100 A. At lower current levels near 0, I noticed the sensed current had a variation of ±300 mA. I did some research and learned about the thermovoltage effect which I assume is causing this inaccurate readings.

To compensate this effect, I got the Seebeck coefficient of the shunt (1.5 μV/K) and the internal heat resistance (1 K/W). I registered the data for the sensed currents in a table. From what I found online, this is the calculation steps to make:

• Power dissipation of the shunt: P = I2·R.
• Max. temperature between two junctions: ΔT = P·1 K/W.
• Thermal voltage: ΔV = 1.5 μV/K·ΔT.
• ΔI = ΔV/20 μΩ.
• Icompensated = I - ΔI.

I would like to know if these calculations are correct? Should I always substract the thermal voltage or does it depend on whether the sensed current is positive or negative?

• There is a lot more to shunt compensation then thermal EMF. d-nb.info/1142816079/34 Commented Feb 15 at 10:49