OPA1611 strange distortion behavior, any explanation?

[Kokoriantz]

Normally, an opamp will increase its distortion by increasing the closed loop gain with the same ratio.
Why with +10 2k load it has lower thd at high frequencies and even not the double than with+1 2k

[moffy]

It is possible that the THD is coming from the input stage, for low loads, higher gain means a lower voltage excursion on the input stage, which would explain the -1 and +10 correlation at 2k load.

[Kokoriantz]

If you inspect the -1 2k cyan curve, it is nearly the same as +10 2k. But your idea makes me think about input common mode distortion.

[moffy]

If you inspect the -1 2k cyan curve, it is nearly the same as +10 2k. But your idea makes me think about input common mode distortion.

That's the input stage I mentioned, consider the difference between the +1 and -1 at 2k load, the only real difference, apart from the -1 having a noise gain of 2, is the signal level that the input stage is seeing.

[David Hess]

It could be common mode input distortion in the non-inverting configuration.  The output signal levels are all the same, so the 10x case has 1/10th the common mode input signal.

However the -1 cases show a large effect of output loading on distortion, as expected.  They are not including the loading of the feedback network, and the non-inverting 10x case will have about 1/10th of the load from the feedback network, lowering distortion.

We need more information about the test configuration.

[Kokoriantz]

If the -1 is increased to -10, I suppose the distortion will be 10 times higher than +10 which has more common mode voltage.
It looks like this opamp is optimized for +10.

[David Hess]

If the -1 is increased to -10, I suppose the distortion will be 10 times higher than +10 which has more common mode voltage.
It looks like this opamp is optimized for +10.

There is no common mode distortion in inverting mode.

-10x may have lower distortion because the load from the feedback capacitance resistance is likely 1/10th as much.  We cannot tell without details about the test configuration.

[floobydust]

It's a special test circuit for the OPA1611 distortion test.

[David Hess]

That tells us what we need to know.

Feedback resistance R2 should be included as part of the load resistance calculation.  In the +1 gain test, the load resistance is much lower than stated.

It is not uncommon for low distortion or precision operational amplifiers to have difficulty driving their own feedback network when the feedback network must be low impedance to limit noise.  Some circuit designs will include a buffer between the output and feedback network to help drive a low impedance feedback network and load.

I think raising the noise gain with R3 will lower the amount of common mode distortion from unequal source impedance, at least as shown.

[Kokoriantz]

I think it is the noise which is higher with -1 as the input impedance is 10 times higher than with +10. The feedback resistor is just 10% higher with +10 to be 5.5k vs 5k paralleled with the 2k load.

[David Hess]

The problem is that the +1 gain case has a much greater output load, increasing distortion, than the +10 gain case.