I'm using the OPA4990 Op Amp from Ti for a project and I noticed today that the offset voltage seems to be larger than the datasheet specifies.

I'm using one of the op-amps as a voltage follower in the unity gain buffer configuration. I measured the voltage present at the input to the positive terminal of the Op Amp and saw 3.300 volts. When I measure the voltage present on the negative terminal and the output of the Op Amp I see 3.320 volts. This seems like an effective offset voltage of 20 mV. The data sheet lists a maximum offset voltage of 1.5 mV and the bias currents for this part mean that the input resistance shouldn't be affecting anything. The output of this Op Amp is fed directly into a TLV1704 comparator so the load current the Op Amp has to provide is extremely low.

For now, the other Op Amps in the package are properly terminated/tied off so they aren't saturated and I don't think they should be affecting anything. The voltage rail I'm trying to measure is the main power rail which the op amp is also powered off of. Any ideas what could be causing this? I tried changing the input voltage across the input range and saw this effective offset voltage change with the magnitude of the input voltage.


simulate this circuit – Schematic created using CircuitLab

  • \$\begingroup\$ Depending on the materials used and the thermal environment, a few millivolts might be a thermocouple effect. \$\endgroup\$
    – Whit3rd
    Jan 6 at 1:48
  • 1
    \$\begingroup\$ What is the input resistance of your measuring device? \$\endgroup\$
    – John Doty
    Jan 6 at 2:01
  • \$\begingroup\$ I also think it's the voltmeter input resistor that has drawn current from the op amp's negative terminal. \$\endgroup\$
    – Vincent
    Jan 6 at 9:38

2 Answers 2


I suspect meter loading measuring the voltage divider.

Not counting component tolerance.

If you calculate \$V_+\$ without loading for a 10V input the result is 3.33...V.

If you calculate using a 10M\$\Omega\$ meter in parallel with the 100K resistor a 22mV decrease is observed. The meter will not load the output measurement.

So what you are seeing is correct.

  • \$\begingroup\$ I suspected this was the answer when you first posted it but I was able to verify this on the actual circuit. Additionally, measuring the voltage difference between the inverting and non-inverting terminals yields a 0.001 volt reading on the multimeter. \$\endgroup\$
    – cEEa
    Jan 11 at 17:46
  • \$\begingroup\$ The multimeter will load that measurement as well.@cEEa \$\endgroup\$
    – RussellH
    Jan 12 at 18:10

Try measuring the voltage from the non-inverting input of the op-amp to the output of the op-amp. You might also want to parallel R1 with something like 100nF of capacitance to avoid excessive RF and other noise pickup.

You can use a narrower voltage range on the meter, and loading effects will be minimized.


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.