• State TI Thinks Resolved
  • Locked Locked
  • Replies 1 reply
  • Answers 1 answer
  • Subscribers 47 subscribers
  • Views 675 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

INA226: calibration method

Kelly Bai
Intellectual 1330 points
Part Number: INA226


Hi team,

 About INA226,  not very sure how INA226 adjust the result due to actual measurement,  I made a understanding of it as follows. Could you kindly explain if it is working as this? Thanks!

 

 

For example, Calibration Register content is 0x1900, and finally I calculate the current is 33A, but external ampmeter result is 33.3A.

Then Calibration Register content should be adjusted to =0x1900*(33.3A/33A)=0x193A ?

thanks.

  • 0
    TI__Mastermind 32395 points

    Hello Kelly,

    Your example calculation is correct. This procedure does not need any other coefficients, it is only a system gain calibration that corrects INA226 gain error, shunt tolerance, shunt sense layout resistance gain error mostly. As with any calibration, it is only as good as the external calibrated measurement (e.g., 33.3A in you example) and it should be better than the total error which includes INA226 inherent gain error, shunt tolerance and layout.

    This procedure does not technically correct for device offset (10µV max), but this is not usually a problem since offset error only contributes significantly to total error at lower currents (or when Vsense < 1mV). If you must actually maintain very tight accuracy for low currents (low Vsense), then you could perform a 2-point linear calibration, but this will introduces system and manufacturing complexity as you would need two load currents (e.g., 33.30A and 1.00A for example) which are both measured accurately. You would have to solve for system gain and offset with standard y=mx+b (or load_adc = gain*load_real + offset) and then load this into firmware of host processor.

    Sincerely,

    Peter