### Re: How do speaker wires affect audio frequencies?

Date: Mon Oct 23 10:22:24 2000
Posted By: Bruno Putzeys, Staff, Electrpacoustics and Analog Electronics, Philips ITCL
Area of science: Engineering
ID: 971893383.Eg
Message:
```
Eew! Tuf one...

For one thing, it's barely about wires, it's about "cables", i.e. a bundle
containing both conductors. The electrical behaviour of separate wires is
quite nondescript as you'll understand from the following.

The simplified electrical model of a cable has 4 parameters:
*Inductance/length:
measure at one end with the other end shorted. Increases as the distance
between the wires increases, i.e. the loop area encircled.
*Resistance/length:
measure with both conductors in series.
*Conductivity/length:
conductivity of the insulating material. Measure with the cable open on the
other end. Likely to be hundreds of megaohms.
*Capacitance/length
Measure with the cable open at the other end. Increases with increasing
outside area of the conductors and with decreasing distance.

The resistance of the cable can indeed be reduced ad infinitum by using
sufficiently thick conductor gauge. The insulator conductivity can equally
be zeroed out by using an insulator material of decent quality.

The combination of the capacitance and inductance yields a well-known cable
parameter called characteristic impedance, calculated as Zc=SQRT(L/C). In
high-frequency applications this is a cable's primary parameter as it
determines the required resistance that should be placed at the ends of the
cable to exactly absorb an incoming signal without sending part of it
ricocheting back to the source.
This effect is only noticeable for cables that are long with respect to the
signal's wavelength, so any contribution of a cable's characteristic Z to
sonic character is highly doubtful.
Nevertheless, Zc is often used by cable salespeople to advertise their
product. The reasoning goes that loudspeaker loads are "typically" 8 ohms
and "normal speaker cable" has a Zc much higher than that.
It's easily debunked if one keeps in mind that the output impedance of an
amplifier is very low compared to the loudspeaker's input impedance (and
rightly so), which itself is in turn highly variable in function of
frequency. Otherwise put, the power end of the audio chain by design does
not use impedance matching, nor should it.

The capacitance of a cable alone is always so low that a normal amplifier
wouldn't even notice it. Even a cable with 1nF of capacitance does not
detract appreciably from the system bandwidth when the amplifier has a
moderately low output impedance.

That leaves us with the inductance. A normal cable (2 parallel conductors)
can have an inductance of several microHenrys. At 20kHz a 2 microHenry
inductance has an impedance of 0.25 Ohms, which is high compared to both
the output impedance of the amplifier and to the resistance of the
(doubtlessly extremely thick) cable. As such it is the only parameter that
has any influence comparable to that of the resistance. Indeed it is quite
simple to demonstrate using a square wave source, a good amplifier, 5 yds
of cable and a loudspeaker (-like load) that different cables show a
different picture on the oscilloscope when measuring on the loudspeaker
end.

You will find that most "audiophile" cables have some kind of twisted-pair
construction. Some of these constructions demonstrably reduce inductance,
but always at the expense of increased capacitance.

To which extent these effects do anything to the sound is difficult to
fathom, keeping into account the loudspeaker's own series resistance which
also places itself in the current path and which bluntly swamps out
anything as subtle as amplifier output impedance, cable resistances etc.

Other effects called into court in the cable argument are cable losses
(certain dielectrics show excess resistive losses when subjected to a high
frequency voltage) and its dual, skin-effect (excess resistive losses in a
conductor when it is subjected to a high-frequency current). Both effects
are linear and can be inserted into the simple model by simple addition of
extra reactive and resistive elements, which in turn are even smaller than
the main elements.

You now probably expect me to say that there can be no difference in sound
as far as cables are concerned. I'm not going to say that. I've performed
enough listening tests, blind and double-blind, to be certain that there
are differences (though fairly subtle ones) to be heard. So what I'm saying
is that normal "cable science" as outlined above just doesn't cut it. One
physicist once told me that science always relies on gross
oversimplification of matters and that building a truly accurate behavioral
model of something as simple as a cable is almost beyond human capability,
so he thought it extremely arrogant of some people to claim that the sonic
difference cannot exist, simply based on the common cable model.

The hunt for a decent explanation for what I and others are hearing is
still open. It is also very clear that no-one (and especially the people
who make money selling cables) has a clue as to what is going on. It would
only be better if they were just honest and said "I don't know", rather
than flooding the place with quack theories that won't help anyone.

So, here's one practical advice: just experiment with some other cable in
your system and listen if you hear any difference. If you don't, it won't
make you happier so don't buy it. If you do, ask yourself if it's worth the
money.

Hmm. Simple after all...

Cheers,

Bruno

```

Current Queue | Current Queue for Engineering | Engineering archives