virtins.com

Thanks for your answer !

Of course !
I should have thought that the accuracy of the substraction method would be problematic.
Thanks for sharing the bandwidth used for HD computing: I'll adapt and use the same.

A lot of (other) work this week, but I hope to work on this during next week-end.
More to come then.

Thanks again !

Thanks once again !

It works, of course.

I'm now consistently getting the exact same figure than REW with "AES17-2015 standard notch" distortion setting.
I also get exactly the same THD by computing from Band Energy than the "official" THD DDP.
Of course, there is anoise benefit to this Notch
It is quite constant with level

Thanks for the report.

RJA4000 wrote: I also get exactly the same THD by computing from Band Energy than the "official" THD DDP.

The "official" THD cannot be seen from your screenshot, but it should be nearly the same as the one with AES notch filter, that is, the increase of fundamental energy due to the much wider bandwidth of AES notch filter should be negligibly small when the fundamental is well above the noise floor.

It can be seen in the your screenshots that the fundamental has a quite wide skirt, probably because the ADC is RME ADI-2 Pro but the DAC is not?

MI 3.9.14 has added an "f1 Notch" option in [Spectrum Analyzer Processing] to define the bandwidth used for calculating the fundamental power, in accordance with the notch filtering method specified in the AES17-2015 standard for THD and THD+N measurements. If the fundamental exhibits a wide spectral skirt, selecting the "f1 Notch" option will result in a lower measured THD+N.