INA226: Need to monitor max current 1A, what's the best choice of shunt resistance value

Hello Zhengquan,

Well strictly speaking the best choice for a resistor will be one with 0.1% tolerance and is properly assembled onto board. The maximum input offset (Vos_input) is ±10µV so any shunt voltage sensed (Vshunt) could be ±10uV off for the absolute worst case with no calibration. So you want the shunt resistor to be large enough so that at 1A, the Vshunt >>10µV. When this is the case the gain error (0.02% typical) will dominate over Vos_input.

The maximum full-scale input (Vshunt that can be sensed) is ±81.92mV. If we set Vshunt to be 81.92mV when load current is 1A, the Rshunt = 81.92mΩ. The closest 0.1% resistance value is 81.6mΩ and the nearest 1% resistance value is 80.6mΩ (although this depends on the supplier). Anyway by choosing a resistor lower than 81.92mV, you will account for any current noise at 1A. So if 81.6mΩ is shunt resistance, then it will take 1.003921A to saturate the input of INA226. Customer may want to decrease Rshunt further depending on what temperature range they are considering. A temperature delta will increase Vos and gain according to datasheet specs. It will also induce a change in Rshunt resistance depending on Rshunt datasheet. Customer needs to take into account what value Vshunt will be at maximum temperature and then set Rshunt so that Vshunt is not saturating at currents < 1A. The common-mode voltage (VCM) and power supply (VS) also contribute to Vos. Please view our training courses or read through the circuit article below on understanding how to calculate amplifier error.
training.ti.com/getting-started-current-sense-amplifiers
www.ti.com/.../sboa336.pdf
www.ti.com/.../sboa179.pdf

If Rshunt is 81.6mΩ, VCM= 12V, VS = 3.3V, and ambient temperature is 25°C, then Vos = 10µV according to datasheet. Thus without calibration, customer cannot sense currents lower than 10µV/81.6mΩ = 122.5µA worst-case. If customer perfectly calibrates out the offset for every part, then they could theoretically sense current down to INA226 LSB of 2.5µV. Although load current noise may make this current level impossible to resolve.

Hope this helps.
Peter Iliya
Current Sensing Applications