Why testing at 5W is probably the worst possible choice

[pma]

I have read about the 5w more than 40 years ago, maybe in J. Hiraga publications.
It was measured at mid priced seat. I estimated at that time to get the same measured level with 5w amp, the speaker used was 90db.
Here for music in theater is 95db max.
Put your system at maximum comfortable level and with scope measure the peak level and convert it what it would be with 90db speaker.
In 60's to 70's, we had 2.5w by ECL82,amps, and hardly we could push to max in radios, tvs or turn tables without disturbing the neighborhood.

I hope you take into account speaker sensitivity. If 5W is fine with 92dB/W/m sensitivity, then 20W is needed with 86dB/W/m and 40W with 83dB/W/m. Look at average sensitivity of nowadays speakers, it is much lower than in the times of old tube radios. Such simplified comparisons with the old days equipment make no sense. And, Hiraga wanted to defend his low power designs.

 

[Hayk]

The subject is, why 5w. And this number rings me the bell of Olsen. Indeed, nowadays the efficiency of speakers have become very low to get low bass in small volume but 90 dB is not uncommon.
I am sure you have a scope, adjust the volume to maximum comfortable level and tell us what is peak voltage you get for a given speaker.

 

[pma]

The subject is, why 5w. And this number rings me the bell of Olsen.

That is exactly the subject. And I am afraid that to most ASR readers, including the chief reviewer, the mechanism of distortion creation in amplifiers remains unknown. Thus this thread. I have prepared another example, with the same amplifier, different PSU voltage, and higher measuring sensitivity. Once it is measured at near to optimal class AB bias and then with minimum bias at class B. I am posting both THD and THD+N vs. power plots.

1. THD vs. power

The legend shows THD at 5W. Though at 5W there is not much difference, at 0.1W the difference is huge. And 0.1W is still quite normal listening level whenwe do not listen loud.

2. THD+N vs. power

The plots below include noise, so they are an equivalent to SINAD measurements at ASR.

Again, at 5W there is almost no difference, but at 0.1W the difference is huge, almost 10x. At low volume, the crossover distortion smears the signal in class B, due to crossover distortion. And please note that in class B the THD and THD+N plots would overlap, because distortion dominates over noise in the whole range of signal amplitudes.

Finally, below you will find a 1kHz record at 0.28W output power for both bias settings (B vs. AB). Please feel free to analyze it or listen to it.
My final question is, what does the 5W test tell us about crossover distortion impact at lower power level or about similar dead time class D distortion? I think that my opinion is clear.

 

[Hayabusa]

Testing amplifiers at 5W/1kHz and using such parameter as a comparative parameter is IMO probably the worst choice. Why? Below please check the example of a conventional small class AB amplifier and its THD vs. power plots where idle current of the output stage is the parameter. Covering several mA to about 300 mA. For low values of idle current, crossover distortion rises for smaller signal amplitudes. But at 5W, it is almost independent of the idle current. At least 3 points should be used for amp comparisons. Take it into account, @amirm !

View attachment 522035

Amir also puts these plots.. What are you complaining about?

 

[pma]

Another good signal to reveal nonlinearity at low level is a 3rd octave noise.

BTW I am not speaking about THD+N vs. frequency plots, as indicated in post #24. The DUT amp in post #1 has had distortion almost independent of frequency. The plots vs. frequency are irrelevant for the described linearity issue. The issue is even static transfer function nonlinearity that is not disclosed in the 5W benchmark test.

 

[Hayk]

Just for knowledge.
The German DIN 45500-6 to qualify as HIFI, the power in stereo should be minimum 2x6w.