We probably should note that flat untwisted wires are, just like
twisted pair, a *balanced* transmission line. In fact they are
virtually identical! The flat line is twisted pair with zero
twist, that's all.
(Of course in the case of speaker wire, that is not significant
at all. The impedance is very low and the power levels are very
high, hence the SNR is always going to be *very* high, even in
a very hostile environment.)
Zip cord! (Oh, that's what I use for speaker wire anyway...
That is certainly an interesting URL. Unfortunately it is very
poorly stated, though almost 100% correct (and only one
interesting nit pick about something that is not quite
"correct"). I can see where you are being confused by what it
says, for example, because many of the things you've said that
are absolutely incorrect appear to be supported by the way he
presents his information. But what he actually means is not
what you are taking it for. (I won't go into it here, because
it is all in the other two articles I've posted.)
However, as a brain teaser, I'll point one technical error on
that web page. This is not a serious criticism, because this is
*way* over the heads of his intended audience, and I would not
have mentioned this either if I were writing that article. (I
would word it a little differently though... as you will see.)
But when I talked to the computer experts that were
familiar with this noise cancellation phenomenon, I
realized not many except the real gurus realized the
cancellation occurred in the balanced line receiver
and NOT the cable itself!
Again, its NOT the twisted pair cable that gets rid of
noise, its the balanced recievers! IN FACT, the
balanced line twisted pair wires are NOT shielded so
that BOTH wires get equal noise, so the noise (if
equal) will be cancelled out!
In fact, for most but not for all twisted pair cables that is
true. The exception is interesting though. For long telephone
feeder cables, those that are more than a few hundreds of feet
long, a typical cable is laid in sections, for example 3000'
feet long. At each splice box the shield that surrounds the
entire cable (which might have anything from 25 to maybe 600 or
even more twisted cable pairs) is grounded very carefully, and
is bonded to the next section.
That has the effect of canceling noise in each and every pair.
(Not by shielding it either, but by cancellation.) But before I
describe how that works, let me point out that the exact same
effect can be had by grounding all unused pairs at both ends!
(Clearly _that_ is not an effect of shielding!) Of course that
works well only if there are many spare pairs...
So, the question then is just how does that cancel noise in the
twisted pair conductors that make up a multi-pair cable?
Any noise source will generate an electrical field that causes a
voltage and a current to appear on a conductor. If there is
good continuity, the voltage will be low and the current will be
high. With poor continuity the opposite is true. Also the
potential is directly related to the area (i.e., the size) of
the conductor. Hence many pairs tied together, or the outer
shield, will have more potential than a single smaller
conductor. And by grounding the shield (and maybe all spare
pairs too) at both ends we cause the *greatest* amount of
possible current to flow from any given noise source that is
causing induction. Typically each pair that is in use will be
terminated in an impedance somewhere between 100 and 1200 Ohms,
so compared to the directly grounded shield and conductors there
will be much less noise current in the used pairs.
Okay? Now comes the fun part. The large noise current in the
grounded conductors causes an equal and opposite potential to be
induced into each of the other pairs! It therefore cancels some
of the induction directly from the noise source.
Hence, while it is commonly the case that most of the noise is
canceled by common mode rejection at the balanced receiver,
that is not the only noise reduction mechanism for twisted pair
cables.