Compensating multi-stage feedback (was: RE: all tranny vca+ )

[Magnus Danielson]

From: Don Tillman <don at till.com>
Subject: Re: Compensating multi-stage feedback (was: RE: all tranny vca+ )
Date: Sat, 1 Jul 2000 19:18:50 -0700 (PDT)

>    Date: Sun, 02 Jul 2000 02:49:51 +0200
>    From: Magnus Danielson <cfmd at swipnet.se>
> 
>    Would you _please_ elaborate on how 0.001% is more awful than 5%.
>    I'd really like to know. Where's the evidence?
> 
> Distortion can take many forms; heck, any form.  Some of it might not
> show up in standardized measurements, some might only show up as
> intermodulation distortion on more complex waves, some of it might
> show up at certain levels or on certain types of waveforms.  Harry
> pointed out crossover distortion; that's a *great* example.
> 
> Product examples, perhaps?  My personal experience is with the old
> Peavy mixers and the mixer section of the Yamaha MT2X 4-track cassette
> deck.  These units sound incredibly bad.  If your signal goes through
> one you can tell immediately, really, everything sounds very thin.
> There are only a few opamps in the audio path, I don't remember what
> kind, and the mixers are spec'd at something like 0.01% distortion,
> yet they suck all the life out of the music.  There are other
> opamp-based mixers that sound great of course.
> 
> On the other hand there are tube amps rated at over 1% distortion that
> sound beautiful.

Certainly, but now comes my key question:

Are you sure that what you perceive is due to types of distorsion?
Are you _really_ sure?

My point is, I do not object to people hearing something sounding good or bad,
but what I have learned is that very few people are able to pinpoint the
correct process of destruction or enhancement (as below). There are many
possible processes to obscure sounds as it passes through systems, some are
linear and some are non-linear. Then we have the placebo type of effects.

As I have previously stated I am sure that a lot of diffrent knowledge allready
exists on this list.

Now, one of the problems with "standardized measurements" is that each such
measurement is aimed at some specific process or to preform some approximate
overall measurement. If we are to be able to characterize the pros and cons of
diffrent solutions we need to understand how these processes works in order to
make a suitable measurement and also evaluate the result.

> And then there's the Aphex Aural Exciter and the BBE Sonic Maximizer
> that intentionally add distortion to bring out certain aspects of the
> music.  Okay, that's a reach.  

Certainly, and there the Aphex Aural Exciter is a rather complex scheme to
acheive this sweetening. I find these devices interesting. Just as I find
tubes interesting. I have recently got myself a number of books relating to
tubes and tube amps.

> Other classic examples are the Pioneer and Technics hifi receivers
> from the mid-70's distortion spec race.  These companies had marketing
> departments that understood that it generally wasn't practical for
> their customers perform reasonable listening tests and purchased based
> on distortion specs instead.  So suddenly there were receivers on the
> market with huge amounts of feedback for a 0.001% distortion spec, and
> they sold well but sounded terrible.

Well, was the measurement applicable to the process of destruction?

>    "You know that's not true"
>    Howha! That's strong!
> 
> :-)  
> 
> I was actually trying to point out that our language has a tendancy
> to obscure what's really happening.  When we start talking percentages
> it not only implies that we're measuring in the same units, but that
> our measurements are appropriate to begin with.  

Certainly, but your formulation may not have been the best.

Now, to summerize what I have heard and what I know of which may destroy or
enhance sound:

Linear filtering
Impulse responce (_should_ be the result of linear filtering)
Cross-over distorsion (small signal distorsion)
Linearity (medium signal distorsion)
Clipp and clipp behaviour (large signal distorsion)
Feedback loop behaviour on distorsion
Amplifier load impedance behaviour
RF intermodulation (any non-linear section will act as a mixer)
Noise and noise modulation
Cross-talk

I migth have forgot some in the process since this was just from the top of my 
head. We could and possibly should elaborate on each of these. If someone sees
something I missed, please let me know. I've tried to summarice the generic
process but it may occur at many places. For instance, non-linear properties
may be found in any passive component such as capacitors, transformers and
speakers.

Cheers,
Magnus