INA231: Current Resolution

Hello Lior,

Thanks for considering to use Texas Instruments.  If 30 uA is your max expected current and you are considering the INA231, you will be sinking more current than you intended because of the input bias of the current monitor.  According to the Electrical characteristics table, the INA231 typically pulls 10uA at a 12V common mode.  But as you can see from the device used to generate the figure below, It could be higher than that.  If you do a low side measurement, then the input bias current might be more reasonable for your application.  Since there are multiple error sources involved in determining accuracy or error, I would need to know your operating conditions (supply, common mode, temperature, and shunt value) to have a closer approximation of the error you might see.  But from input bias alone, I would say your error is: (input bias)/(ideal load current)*100.  This error will get significantly worse with the lower end of your current range.  I recommend you consider the INA190, an analog output device with an input bias of 3nA. 

Hello Lior,

That thread you referenced I think would be a good compromise for measuring the low current you want while allowing for a digital communication. The INA190 has in input bias of 3na. So if we only consider input bias error with % error of 1%, the minimum sense current would be (input bias)/(1%/100)=(3E-9A)/ 0.01=300nA. However, we should also consider Vos which is related to input bias current. Vos % error of 1% would result from a Vsense voltage of (Vos)/(1%/100)=(15E-6V)/0.01=1.5mV. From that value, your minimum Rsense would need to be >(1.5mV/300nA)=5kohm. If your INA190 Vsense voltage is 1.5mV, then the lowest output voltage (Vout-Ref) you could get with the lowest gain variant would be 37.5mV. Worst case for INA219, this adds 200uV/37.5mV*100 = 0.5% voltage offset error. There are other error sources but voltage offset will dominate at low sense currents.

As you want to measure 20-30uA all of these errors should improve (decrease) assuming you choose a shunt such that yields a voltage across your INA190 shunt greater than or equal to 1.5mV when your load is 20uA.

As one would argue that the error sources of both INA190 and INA219 are uncorrelated, you could sum squares of all the error terms and then take the square root of that value to get your typical expected error. For a worse case scenario you may consider calculating the RSS error for each and then summing.  If you need more information on the various error sources and how to calculate the RSS error, I would recommend looking at our training videos here posted on the right side. On a given video page, there should be the option to download the slides if you just want to quickly click through the content.