I'm wanting to current sense a 10A load with high precision. I'm planning to use a current shunt resistor and then amplify/filter this signal before sending it to an ADC. Which design flow is typically more precise?

  1. Using a high precision differential op-amp (amplify + level shift) -> buffer -> RC filter -> ADC
  2. Using an RC filter -> internal PGA of an ADC to amplify the signal, for example, the ADS131M02IPWT, datasheet

The intuitive answer is #1, but then I'm wondering why most ADCs have a PGA stage, because it's good enough for the typical application?

  • 3
    \$\begingroup\$ What does "high precision" mean to you?? \$\endgroup\$ Commented May 30, 2023 at 19:22
  • 1
    \$\begingroup\$ I'm trying to achieve <1% error for an energy measurement. \$\endgroup\$
    – ken
    Commented May 30, 2023 at 19:28
  • \$\begingroup\$ Is the current sense resistor referred to ground, the same ground as the ADC? If not, what is its voltage with respect ground? \$\endgroup\$ Commented May 30, 2023 at 19:28
  • \$\begingroup\$ My application is AC and I'm putting the sense resistor on Neutral. \$\endgroup\$
    – ken
    Commented May 30, 2023 at 19:38
  • 2
    \$\begingroup\$ You can easily get <1% with either approach. #1 requires more parts, #2 gives you less flexibility on the input signal. Whichever you choose be careful if you are planning on building these for any length of time; specialized, single-source parts like the one you mentioned can have a tendency to become hard to get. I would use a popular ADC with a big user base as a basis for your decision. \$\endgroup\$ Commented May 30, 2023 at 19:50

1 Answer 1


I'm trying to achieve <1% error

Then you should use a Hall Effect sensor instead:

  • isolated
  • 1 % accurate
  • Outputs 0 to 5 V, same as the input range of your ADC.
  • \$\begingroup\$ I'm considering cost so I have to go with a shunt resistor - I'd rather spend more on a better op-amp/ADC than going with other sensing techniques. \$\endgroup\$
    – ken
    Commented May 30, 2023 at 19:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.