I'm using a unity gain buffer to drive a power amplifier voltage reference. Here's a schematic for the buffer configuration.

enter image description here

The op-amp in question is a LMV321.

I've set the voltage divider at the amplifier's non-inverting input to output 2.85V. Unexpectedly, I'm measuring the voltage at the inverting input and output as between 2.70V and 2.75V (it changes somewhat which I'm also surprised by). The measured non-inverting input voltage is stable at 2.854V. As you can see from the schematic, the supply voltage is 3.3V (3.28V measured). When the device is off, I measure the DC resistance between the output and ground as \$2.328k\Omega\$.

The max input offset voltage specified in the datasheet is 9mV, not the more than 100mV I'm getting. I've kept the output voltage well below the 3.3V supply (especially considering this is a rail-to-rail op-amp).

The load is the a reference input voltage of a SE5004L power amplifier, which is supposed to be supplied with a voltage between 2.8 and 2.9V. I'm having a little trouble determining what the input impedance of that pin should be although I guess it has to be about 2.328k (op-amp output impedance in parallel with it, which is so high that my multimeter reads OL when attempting to measure the resistance). I unfortunately don't have an extra power amp to measure the impedance, and I'd rather not have to take it off and resolder it. Please let me know if I've omitted relevant information.

Any thoughts on what might be going wrong here?

  • 1
    \$\begingroup\$ Not your main problem but when seeking to minimize offset voltage it is recommended that the DCR looking outward from the inverting terminal should be the same as that looking outward from the non-inverting terminal. In this case that means adding a roughly 1k resistor in the feedback path. This is done to minimize the effect of bias current. This op-amp has such low bias current it may be negligible. Then again, if you change it to a different one it could matter. \$\endgroup\$
    – user57037
    Mar 17, 2020 at 20:49

1 Answer 1


I believe you have exceeded the input common-mode range that the opamp supports.

The data-sheet does not give an explicit range for 3.3V operation but when operating from 2.7V or 5V the maximum is about 0.8V from the positive rail - this would be 2.5V in your situation.

Although the device has a rail-rail output it does not support operation close to the positive rail on the input.

You have a couple of options: substitute another pin-compatible device that does support rail-to-rail input or reconfigure the circuit with some gain, for example 1.5 times, such that the input is within the allowed range. Section of datasheet with common mode range


  • \$\begingroup\$ Ah nice catch, that would explain it. I'm looking for a pin-compatible device with rail-to-rail input per your recommendation and it seems another major consideration I need to make is the max current the opamp can supply. I'm having a bit of trouble finding current draw info on the PA datasheet, but it looks like the bias control current (10mA) on page 3 may be what I'm looking for. Have I interpreted this correctly? Are there any other obvious pitfalls to avoid? \$\endgroup\$
    – MattHusz
    Mar 17, 2020 at 22:54
  • 1
    \$\begingroup\$ @MattHusz - The datasheet is not the greatest. I agree that the IEN BIAS current is the nearest thing I can see to being the current into the VREF pin. RF device companies are renowned for being bad at providing DC data. The only other thing I can think of is making sure the tolerance on your 3.3V supply is good enough, it probably is. \$\endgroup\$ Mar 18, 2020 at 0:25
  • \$\begingroup\$ In case anyone else comes across this and is curious, IEN BIAS current is indeed the current into the VREF pin. I contacted skyworks and they confirmed it. \$\endgroup\$
    – MattHusz
    Mar 18, 2020 at 20:24
  • \$\begingroup\$ I replaced the opamp with the lvm321 from stmicroelectronics which supports rail-to-rail input and the output is now 2.854V as expected. \$\endgroup\$
    – MattHusz
    Mar 19, 2020 at 20:45

Your Answer

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

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