Hi,


There has always been snake oil associated with the high-end audio market. One of the hypes even includes high performance speaker cables. Which aside from I-R drops source impedance of the amplifier (damping factor) and complex speaker loads seems like snake oil.

Just for grins some time ago, I characterized a lot of different interconnects for audio systems.

I used my HP-4192A to characterize RLC parameters over frequencies from 5 Hz to 150kHz. I also used my HP8903A combined with my HP3585A Spectrum Analyzer terminated in multiple loads (the analyzer includes 1 Meg, 75 Ohm and 50 Ohm inputs - which allowed me to externally terminate what I wanted on the Spectrum Analyzer). My HP8903A has 600 Ohm output capable of balanced and unbalanced drive. I also used my HP3336B which has a 75 Ohm output which I varied the output drive impedance with series resistors. I matched the before and after insertion loss plots of course. I used a wideband audio power amplifier, the tracking generator in the HP3585A, to drive an EMI current injection probe to couple noise to the cable shield and observed the resulting noise susceptibility on the Spectrum Analyzer. For pulse noise, I used an HP8165A programmable pulse generator input to the Audio Amplifier (within 200 kHz).

I found ALOT of capacitance variance from coaxial cabling due to construction. Most of that was centered around the type of shield that was used and the spacing and dielectric type used for each sample.

Each sample was at least 5 feet long and no longer than 8 feet long. The variance in capacitance was no surprise as these are known quantities for signal integrity impedance calculations (see IPC-2141A or IPC-2251 which I participated on the revision committees). By the way, the overall impedance was measured at 1 MHz and was all over the map too. This was measured using 2 different methods (IEEE-4143, Beldon and HP also have good application notes for cable impedance measurement of Low and High Frequency Cabling, and characterizing individual L and C with applicable formula).

The cables were a mix of commonly available cables and included balanced and un-balanced types and shield types included the spiral wrapped copper, 60% and 85% optical coverage overbraids, foil, and foil/braid combinations. I also tested some Twisted pairs and Twisted Shielded Pairs (like M27500 - MIL-Spec. Uncontrolled Twisted Shielded Pair, and commercial differential cables).

RLC at 1 MHz results were mixed with: I don't think anyone has heard 1 MHz from their system before. This is bogus, but the only real effective way to measure the parameters. Although I did have a cassette deck dirty record playback switch a long time ago, that routed the 65 kHz bias oscillator through my pre-amp, power amp and cooked one of my AR tweeters once. All I saw was was a pinned power meter and saw a puff of smoke come from my speaker grill. Easy to replace though, although the cassette deck received shall we say a little "rough handling"...

Capacitance of Coaxial Cables ranging from 20 pF/ft to 180 pF/ft.
Capacitance of Twisted Pair cabling from 13 pF/ft to 25 pF/ft.


Bottom line was from 20 Hz to 20 kHz there was very little affect.

1) The impedance mattered some depending on drive and destination impedances.
a) High Impedance driving High impedance (10k Ohm typical) mattered very little at audio frequencies except with very small signalling.
b) High Impedance driving Higher impedance (47k Ohm typical) rolled off the highs to some degree at very small signalling.
c) Low impedance drive (< 150 to 600 Ohms) and high impedance destination didn't matter with audio frequencies. Of course you wouldn't want to do this with digital signalling unless you like ringing due to higher edge rates. It's best to match the impedances of the source, cable, and destination.
d) Low impedance drive (< 150 to 600 Ohms) and low impedance destination was ideal and I-R drops mattered over very long runs. Differential was always better.

2) The Capacitance mattered very little at Audio Frequencies except at matching Phono impedances due to low drive from small signal sources and higher input impedances (like 47k Ohms). The reason it didn't matter was that many pre-amps have a low drive impedance capable of driving some capacitance and the inputs are usually high (10k Ohms).

3) The inductance didn't matter that much because it provided some capacitance isolation from the driver.

4) Resistance mattered when the source and load impedances were low, and then it was only I-R drop , not impedance.

5) Shield construction mattered as far as noise susceptibility.

a)The foil was the worst at audio sweep and impulse noise susceptibility.
b) The spiral foil was better
c) The 60% and 85% and foil/braid overbraid was still better
d) The Twisted Pair cable was the best (provided the shield was terminated properly)

Foil does not really help at low frequencies (below 100kHz).

With higher frequencies the above became and adverse part of the transmission path and degradation occurred.

These were measurement only tests and not subjective listening tests. That came later after I built my own cables.


What did I do? I built short custom cables out of measured low capacitance coax and differential TSP cabling for all of the cabling in my system. I characterized and optimized them per the above method, then closed the magazine article to hide all of the non-sense.

Ross