[sdiy] Request brain dump on balanced lines

[Magnus Danielson]

From: Don Tillman <don at till.com>
Subject: Re: [sdiy] Request brain dump on balanced lines
Date: Sun, 13 Oct 2002 13:29:26 -0700

>    > Date: Sun, 13 Oct 2002 10:47:05 -0700 (PDT)
>    > From: Tim Ressel <madhun2001 at yahoo.com>
>    > 
>    > The best explanation I have heard is this: all semiconductor
>    > amplifiers add 3rd harmonic distortion, which is unplesant to the
>    > ear. Transformers tend to add 2nd harmonic distortion (as do tube
>    > amps), and 2nd harmonic distortion is pleasant to the ear.
>    > 
>    > There are as many opinions on this as there are people, but this
>    > explanation makes sense to me.
> 
> Really?  
> 
> If a tube stage was exactly a square-law device it would only have 2nd
> harmonic distortion.  But tubes aren't exactly square-law devices,
> more like 1.5-law, so a single tube stage does contribute third
> harmonic distortion.  And if there's feedback involved, that will also
> change the distortion products.  Multiple stages changes the
> distortion products more.
> 
> And a single transistor stage has significantly more 2nd harmonic
> distortion than third.
> 
> But more practically, most amps work in a balanced configuration that
> cancels out all odd order distortion products, depending on how well
> balanced things are.

If you read old tube literature, this is the _benefit_ of balanced designs that
they stress. There they are out to reduce distorsion as such. That they in the
process struc on what some may call "audiophile oil" is a sideeffect.

> Further, this explanation assumes that our sense of hearing cares a lot
> about harmonic distortion, which is not necessarily true.

Indeed.

> So while this explanation is the popular one -- it comes from that
> engineering math class where you get to derrive the Fourier spectra of
> exp(sin(wt)) and (sin(wt))^2 -- I think it completely lacks
> credibility.

There is allways a problem to translate subjective judgements into physical
properties. Especially when you have non-linear systems you have a total
expolsion of odd properties all over the place, and pin-pointing the right one
for a certain (beleived common) subject property is a hard thing to do.
Just look back on the various THD and IMD wars for instance.

In a similar sense we tend to aclaim a certain type of element (say tube or
transistor) to uniquely have a certain subjective property. First of all may it
be that a certain usuage of a device do display this property, but other usages
do not. In a similar sense may other alternative devices in their typical usage
not show this property. If you actually look on how things are used, you learn
that there isn't many basic variations in actual use. There are many different
ways to use a transistor for instance. Few people use MOSFETs like tubes etc,
they are mostly used like "diffrent transistors". Thus, the basic concepts
behind the designs also vary. In the end, the whole discussion ends up with a
mudslinging between different groups of people and the signal to noise ratio
in the discussion is so bad that you really have to use up all your energy to
listen to the sound arguments from either of the groups.

If we recall the argumentations there is these issues as I recall them:

- basic distorsion properties of tubes vs. those of bipolar junction
  transistors.
- high negative feedback vs. that of low negative of positive feedback
- crossover distorsion in class A, AB, B and C type of stages
- use of transformers in signalpath (output, input etc) and their contribution
  to distorsion and (often forgot) other propeties in the curcuit
- use of other non-linear components (like capacitors)

and the list of differences could go on...

In the end, there is not only one but a whole bunch of different design
decissions which makes up the real difference. Being able to tell exactly which
design decission which gives exactly which property which gives exactly which
subjective judgement is hairy to say the least. IMHO there is no 100% agreement
in the subjective judgement to start with.

My conclusion is not that it is worthless to attempt to find these things out,
rather the opposite. If you can quantify which technical issues control a
certain subjective judgement you have got yourself a powerfull tool. However,
all to often people jump over the basics and want to go for the conclusion and
then they miss all the gory and hary details. It is not selldom you find all
these forms of judgements and tests flawed in one way or another.

Cheers,
Magnus