Why we do not show both THD and THD+N (SINAD) vs. frequency plots at various frequencies?

[pma]

We can see plots of THD+N (SINAD) vs. power at almost all power amplifier reviews. My question is, why we do not see THD (without noise component) vs. power plots as well?

My point is as follows - for many amplifiers, noise component may mask the distortion behaviour. Distortion profile is thus masked and it is difficult to say, from cumulative THD+N, what is happening in the amplifier. As an example, please see the measurements of my A250W class AB amplifier compared to my NC252MP. Both THD vs. power and THD+N vs. power plots are shown. Measuring BW = 22kHz. Load 4 ohm. Both amps have similar power, 200W+.

Please note how the "N" component in the THD+N plots masks and hides distortion profile and behaviour at various signal frequencies. Please also note that the 1kHz measurement gives the best results.

 

[DonH56]

Good question! Old analog analyzers made it difficult to isolate harmonics; you needed a complex (expensive) switched or variable filter bank with high-order filters. Now, DSP and FFTs make it easy so all you should get is noise in the harmonic bins, hopefully negligible. It would also be nice to see just a noise plot with the fundamental ignored. Follow the IEEE or other test standards to choose appropriate frequencies and you don't even need to window the FFT.

 

[YesChickenNuggets]

pma, big fan of your posts

How are we to interpret the thd+n and thd outside of the data itself, what are the implications on what we hear?

 

[pma]

pma, big fan of your posts

How are we to interpret the thd+n and thd outside of the data itself, what are the implications on what we hear?

Unfortunately, I do not have any simple answer. Very very high THD+N means audible noise or supply hum/buzz. THD worse than 1% (-40dBr) may or may not become audible. It depends on spectrum components. Personally, I prefer numerous spectrum measurements (HD, IMD, multitones) at many levels with various loads to those cumulative THD+N vs. level or frequency plots. THD+N plots have to be deciphered and then you need individual measurements of spectra. Or to plot the harmonics, not only the cumulative value. An example below:

 

[lashto]

Unfortunately, I do not have any simple answer. Very very high THD+N means audible noise or supply hum/buzz. THD worse than 1% (-40dBr) may or may not become audible. It depends on spectrum components. Personally, I prefer numerous spectrum measurements (HD, IMD, multitones) at many levels with various loads to those cumulative THD+N vs. level or frequency plots. THD+N plots have to be deciphered and then you need individual measurements of spectra. Or to plot the harmonics, not only the cumulative value. An example below:

View attachment 252181

Nice graph, much more useful than the 'simple' THD vs. power. I also like the way archimago shows the HD. I would vote for any of those variants to be added as a standard measurement. Ideally both

And btw, does it take a lot more time/effort than ASR's standard THD vs power graph?

 

[Head_Unit]

My question is, why we do not see THD (without noise component)

ASR's standard THD vs power graph?

As noted in the old school days it was difficult/expensive to separate out the noise. I presume @amirm choose to follow the "old" method for some degree of comparability to others measurements (Stereophile etc who keep the same thing they have been doing a long time for consistency).

 

[Hapo]

 

[SSS]

Needless discussion because today a lot of other valid measurements possible. SINAD is from the past but still used for radios and radio telecommunation as far as I know.

 

[SIY]

And btw, does it take a lot more time/effort than ASR's standard THD vs power graph?

No.