Full detailed sound at 30 - 40 - 50 dB

I want to come back to the preference for tubes with high output impedance playing at low level. Why could this be? I have two explanations. The first is that a particular tube amplifier may have a non flat frequency response emphasizing low end and top. This is not infrequenctly the case (see the Stereophile response graphs for many of them), and would act as a traditional loudness control. The second explanation could be that underpowered amplifers actually compress the signal. They depress the peaks, and this has the effect of raising the relative level of the quieter passages, for more 'detail'. This is then a crude version of the digital compression software as used by FM radio stations to keep the signal within the limited dynamic range of the FM format. Yes it works, but is it realistic? No. But then, no low level playing is a realistic representation of the real thing. Enjoy the baby - my kids have now reached the age when they like to play the audio at it's limits....

There is always more to it...

@atmasphere 

Hi Ralph. I have a follow up question if you don't mind, as I mentioned I would be re-reading/re-digesting the rather informative information you've already provided.

Regarding the "weighting' where distortion is more problematic depending on which harmonic the distortion occurs, I note that manufacturers - including your own product specifications - provide "total harmonic distortion" and in some cases "intermodulation distortion". Other than subjective listening, how might a customer objectively derive from specifications alone at which harmonics the distortion is occurring? I guess what I'm asking is, is there a way to understand the individual numbers that comprise "total" harmonic distortion? Thanks.

Other than subjective listening, how might a customer objectively derive from specifications alone at which harmonics the distortion is occurring?

That is a good question. You can get some of that from the spectral analysis. this can be tricky- I've seen some of this done on our amps, but the methodology is was incorrect (one speaker terminal got accidentally grounded by the test equipment) so the readings were off. If the measurement technique doesn't introduce problems, this can be one way of sorting things out. 

But if the total is low, the  components must be even lower (or theoretically at most equal).