[sdiy] Re: Twisted

[Magnus Danielson]

From: "JH." <jhaible at debitel.net>
Subject: Twisted (was: Re: [sdiy] "Time Winding" in Audio Cables ???)
Date: Tue, 12 Jul 2005 01:19:45 +0200
Message-ID: <005f01c5866f$087b3030$0200a8c0 at jhsilent>

> > Everybody always obsesses about the shield too.  We should just
> > remember what the shield is for.  It is a shield against stray
> > electric fields, not magnetic fields.
> 
> Are you sure? I thought a coax structure is good for electric *and* magnetic
> shielding. But it's been a long time since I looked at that.

Actually, it is a bit more complex than that. In a real coax (one solid wire,
with symmetric isolation and then solid metalic tube) it is correct to the
level that assymetried between the center-wire and shield-tube will reduce
the magnetic canceling, but the electrostatic shielding is unaffected from such
asymmetries. Those coaxes are good, really good, but darn hell to work with,
hook up behind synths, running around with on stage and all that. So, we use a
braided shield instead. The shielding effectiveness lowers as we go up in
frequency and eventually will the holes in the shield become large relative to
a cycle and then the magnetic shielding goes away. This usually happends around
100 MHz. Also, the magnetic isolation is only effective above the cut-off,
which is around 1 kHz.

Now, that was the single-wire coax. We don't use that all the time. Another
end of the spectrum is the twisted pair. Twisted pair gives magnetic shielding
up to some frequency, but not that good and also reduces with frequency. Again,
asymmetries will reduce the canceling. The electorstatic shielding is basically
non-existent, but their is in the far-field about the same coupling to both
wires, so some canceling can be expected.

OK, we can now combine these and we end up with a balanced signal system with
shield. For ultra-low frequencies (audio) the shield basically handles all the
electrostatic shielding and the twisting all the magnetically shielding.
However, remember that the shielding for magnetics becomes lower at higher
frequencies! This can bite us since RF can come into our audio-path.

That was only the short story.

> > That's why we like to
> > use balanced lines in studios.  Using a twisted pair that results in
> > equal magnetic and electric coupling in both wires of the pair, and
> > permits the receiving amp's common mode rejection (either by using a
> > transformer or active balanced input) to cancel the coupled signals.
> > Actually, the twist is far more important than the shield.
> 
> Twisted pair inside a shield is certainly the best solution.
> And balanced connection, of course.
> 
> I think the problem with magnetic fields and unbalanced coax cables
> is not that the coax won't shield the voltage between inner and outer
> conductor (differential mode) against the influence of the magnetic field
> (I think it does ...), But that the magnetic field will cause common mode
> errors along the shield, from one end to the other, the dreaded
> ground loops in non-balanced connections.
> 
> Does this make sense?

It does, but then add the refined knowledge that magnetic shielding is only
effective above some cutoff frequency, only to drop above another cutoff
frequency (that of the braid). A cable with braided shield and an underlying
metalfilm is even better in electrostatic shielding.

The trouble is that we usually deal with a little too much oversimplifications
in our ideas of cables. I had to re-learn some myself. EMC is a bitch, there is
soo much non-ideal properties so you go nuts. Good EMC probes, good spec and
some peace and quite in the lab usually gives away what the main culprits are.

As for ultra-low frequencies, you can find components specified like this:

Frequency range: DC-8 GHz (1)

(1) DC = 300 kHz

What is this??? I guess a look on a standard network analyser (say Agilent
E5071A just to pick one from the top of my head) frequency range and you have
a good clue. :O)

So, frequencies below DC must be ultra-low! ;O)

Cheers,
Magnus