I tried but I don't think the result is the most appropriate .
For example , I did this:
1 ) First, I used the PicoScope signal generator (AWG output) with no load on the output and the THD is ~0.05 % at 1 kHz. In fact, varies between 0.05 % and 100 % (quite frustrating);
2 ) I connected the generator output audio input signal (it is a transistor amplifier having input like in the picture attached below):
After I increased voltage almost 0.5Vrms, distortion THD reached about 0.3 % on the input of the amplifier (amplifier isn't powered).
3 ) in this step I powered audio amplifier and gave its input signal in order to reach half of the maximum power that can be delivered (half the maximum peak to peak voltage of the output). In this case THD increases from ~0.75 % to 21 % ? ? I tried to decrease the frequency band, as you said, at 6kHz, but the distortion is a few percent to half of the nominal power. I think a lot.
The phenomenon described above is happening on three types of audio amplifiers with different schematics, including TDA7294?? Why? I mean, if tested three different audio amplifiers, all three have the same results (very close) in terms of input and output THD amplifier ?? How can this happen even more as one of the three amplifiers was fed from another voltage source?
Another weird aspect is the measurement of THD + N (dBc). In a first test, when the spectrum analyzer was set to 1 MHz, at maximum power amplifier was obtained THD+N = 48.68dBc and THD = 0.368%. Then, I reduced the bandwidth of at spectrum analyzer, in this case 6kHz, and THD+N reached ~45dBc and THD=~1%. Now, if you use this website:http://www.sengpielaudio.com/calculator-thd.htm
you'll see 48.86dBc means 0.368% and 45dBc means 0.562%. Now, in other words, you want to tell me that: the THD is higher than THD + N? or equal ?? Is impossible.
One thing rather frustrating is refresh rate measured values (very high). By the way, how I can reduce this refresh rate ?
The fact that, under certain conditions, PicoScope software can not calculate exactly a THD and take steps like the one from 0.7% to 21% shows that the software is not optimized, does not know the difference between useful signal and noise. You should mix the noise filtering method using the software - I don't know if you did this or you don't did that. So, I don't think you've managed to make it work very well. Surely something must be improved.