Re: [sdiy] Request brain dump on balanced lines

[jhaible at debitel.net]

Hi Don,

>By the way, my comments shouldn't be interpreted to mean that I hear
>no difference between a typical tube amp design and a typical
>transistor amp design.  I just don't think the difference lies in some
>weird abstraction involving the second harmonic.

Yes, but an abstraction that prefers lower order harmonics against
higher order harmonics covers a remarkable amount of 
phenomena that are considered relevant for a pleasant sound.

Last week I was sitting in a High Frequency seminary from Agilent,
and they explained the error sources of their spectrum analyzers.
They had an animated graphic which showed how 2nd and 3rd
harmonics were created from nonlinear effects in their mixers,
and how 10dB of increased input level will bring up 2nd harmonics
by 20dB, and 3rd harmonics by 30dB. 
I know this is nothing esoteric, but for me, seeing these animated 
spectral lines jumping up, was like an eye-opener to understand
the pleasantness or unpleasantness of different types of distortion.

Asuming there is a certain threshold level for perception of each 
harmonic (highly depending on th econtext, of course), the lower
order harmonics will "fade in", whereas the higher harmonics
will "suddenly jump into our perception".  3rd will be worse
than 2nd in this respect, but nothing compared to 6th, 7th or
whatever. I think this "jumping" fits very well to your thoughts
from August (as I remember them).

>In a post in August I mentioned that I believe our sense of hearing
>has been optimized over tens of millions of years of evolution to
>recognize mechanical processes by their sonic fingerprints. 


>(To that end, I'll even claim that they typical opamp-style transistor
>amp is not an amplifier at all, it's a servo.)

And the servo works nicely if it is fast enough, and as long as it
does not hit its limits of operation. 
There's a lot to be said about the speed, but even with
an ideal response time there's the other issue:

We tend to think of an opamp with high negative feedback as "linear",
while in fact it's a transfer courve that starts horizontal (negative rail),
goes on as a quite linear ramp, and finally becomes horizontal
again. (positive rail). 

If we develop this into a Taylor series, there will be a lot of higher order
terms - depending on the smoothness of the clipping.

Of course this ist just a different way of saying "hard clipping is
not so good", but my point is that this fits perfectly to the theory of
higher order harmonics being the culprit. 

Servo - yes, there's more to it. ButI leave the response time issues for
you - or others - to explain. 

In a way, even a slight feedback from the output winding to 
the driver stage of a tube amp is a "servo". 
Not sure about the local feedback of a emitter rersistor
in a common emitter stage - for very high frequencies
this might be some delayed effect, too. Gladly not for
audio, though. (Just thinking in a random way here.)

JH.