Speaker Wires and Audio Cables: Separating Sense from Nonsense (Vol.2, No.2: 1979)

"There's the full moon, Dr. Van Helsing. It's time to connect the pure silver cable. " And other such widely debated matters.

There was a time, not so very long ago, when speakers were connected to the amplifier with lamp cord stripped at the ends and wrapped around screw terminals, and the other components in the system were plugged in by means of plain, off-the-shelf phono cable. Those innocent, hype-free days are gone forever; specialized, high-technology wiring has become de rigueur for the serious audiophile, in much the same way as a jogging suit for the jogger.

Is this just another trendy affectation or does it have some basis in electrical science? Are the differences real and audible, or are they wishfully mythicized by cultists? Some time ago we set out to find the answers by measuring the relevant electrical characteristics of a large number of wires and cables of different configurations and listening to each in a known reference system. Initially we even considered the possibility of a brand-by-brand test report, such as we might publish on speakers or preamplifiers, but we soon realized that our results would then be subject to simplistic mis interpretation by those who can't live without numerical scores and rankings. A systems approach that takes into consideration the basic nature of electrical interfaces quickly dispels any notion that one particular brand of wire or cable can end up as the ''winner'' even if it happens to have certain advantages over others under given circumstances. It simply isn't the same kind of problem as finding out which preamp is best. We shall therefore concentrate here on the overriding issue of reality vs. fantasy, with only incidental brand recommendations.

Copper, silver, platinum or kryptonite?

One superstition should be disposed of right up front, before we get involved in more complicated matters. Please note once and for all that electrons retain no memory of the metal they have flowed through, be it copper, silver, gold, platinum or whatever. This, of course, has nothing to do with the use of silver or gold to reduce small-area contact resistance and oxidation, which is a totally different subject. But you may rest assured that an electrical signal that has traveled through a length of silver cable is absolutely indistinguishable from what it would have been if it had traveled through a length of copper cable of equal resistance and reactance.

Anyone who tells you the contrary is either an outright charlatan or a duped victim of vampire tales about the nature of metals. We have checked this out with some very advanced students of the periodic table, metallurgy, solid-state physics and electromagnetism, and they just turn their eyes heavenward with a God-give-me-patience expression when confronted with the silver cable fad in audio. We have also performed some fairly conclusive experiments of our own.

To wit:

We set up a simple, unambiguous but very high-quality single-amped stereo signal path by completely turning off the tweeters of a pair of Vandersteen Model II speakers. substituting a pair of Pyramid Model T-1 ribbon tweeters (properly phased and pulse-aligned with the midrange). connecting this speaker setup to a perfect sample of the Hafler DH-200 power amp (which happens to be very happy with this load), and driving the whole thing out of our Reference A 'front end’ (mostly Cotter). Our assumption was that any changes made in the line-level cable between the front end and back end of this system, if indeed such changes altered the signal quality, would be easily audible and unequivocally attributable to the cable differences. To magnify the possible differences, we used 10 meters (32.8 feet) of cable in each case.

The big shoot-out was between 10 meters of MLAS Ltd pure silver coaxial cable and 10 meters of M.A. Cotter Co. triaxial cable (similar to the former Verion triaxial), with identical top-quality phono plugs soldered to each.

Although pure copper is only 5 percent less conductive than pure silver, there was a greater difference between the two cables because of the gauges and constructions involved; the Cotter inserted 7.2 ohms in series with each channel, the MLAS only 0.73 ohms. The shunt capacitance per channel added by the full length of each cable was 900 pF with the Cotter and 800 pF with the MLAS. None of these four numbers had the slightest significance at the interfacing impedances of the hookup. More significant was the fact that each cable had excellent RF shielding; the triaxial could be expected to be somewhat superior in this respect under extreme conditions of RF interference, but the important point is that the comparison was not between a well-shielded and a poorly shielded cable, as we suspect had been the case in some of the listening tests cited by the silver cultists. RFI can definitely cause audible degradations of the signal, and some audio cables are very marginally RF shielded. But that has nothing to do with silver vs. copper.

Need we spell out the results at this point? Blind testing by a number of exceptionally keen-eared auditioners revealed, in the words of the French chef in that notorious TV commercial, "no differawnce!" Both cables sounded exactly the same in repeated A-B substitutions. If there was the slightest preference for one of the cables, it was perhaps in favor of the copper triaxial, but ever so rarely, vaguely and inconsistently. (Do we have more RFI in our lab than we think?) We call it an obvious draw. And we call silver cable, at $24 per meter per channel (without plugs!) an exploitation of the moneyed audio neurotic, precisely the sort of thing that makes high-end audio appear fatuously snobbish and repellently doctrinaire in the eyes of so many intelligent but non-techno-freak music lovers.

It shouldn't really be necessary to add that the '100% pure copper' fad, which is the equivalent syndrome in speaker wire selection (heavy-gauge silver speaker wire being too costly even for faddists), must be classified in the same subdivision of vampire lore. Because another thing that electrons can't remember, children, is whether or not the length of metal they have flowed through contained a few little impurities. The only possible electrical effect of the latter would be a minute change in resistance. We haven't performed any experiments in this area, since we don't have a metallurgical laboratory; nor do those who profess to hear an improvement when some manufacturer tells them that this here is 100% pure copper wire, yes sir.

Then why does it sound better?

This brings us to the broader issue of why perfectly levelheaded and open-minded audiophiles hear an improvement when they replace their old audio cable or speaker wire with a new super-fidelity design. We're inclined to believe that most of the time the perceived difference in sound is really there-what they report is true but not necessarily because of the superior quality or technological advantages of the new wiring. Here are some of the possible alternative explanations:

Breaking an old metal-to-metal connection that has been undisturbed for many months and may be oxidized to some degree could in some cases significantly reduce contact resistance, diode effects (i.e. rectification) and the resultant nonlinearities, some of which may have been marginally audible. This is somewhat more likely at the higher impedances and lower signal levels of front-end electronics connected with audio cables than at either end of the speaker wire. In other words, just the self-cleaning action of breaking the connections and plugging back the old cables may also make everything sound better.

Or consider this. In an RF-infested environment such as most of us inhabit, where every cubic foot of space is teeming with CB, police radio, TV and innumerable other signals, the entire cable and wire harness of a home stereo system acts as a huge ''antenna farm.'' Adequate grounding and shielding, combined with proper circuit design, will minimize RFI in the front-end electronics, but many power amplifiers are also vulnerable to, and largely unprotected against, RFI backing into their output terminals via the speaker wire. Since RF antennas are tunable and directional, just moving the speaker wire around can change RFI sensitivity for the better or worse, let alone changing the construction of the wire, which can also have an effect but with little or no predictability. (A possible exception: Mogami speaker wire, which is of self-shielding coaxial construction and may conceivably provide a consistent minimum level of RFI protection, although that's not what Mogami believers confess as the prime article of their faith.)

Thus the improvement or deterioration heard in the sound after changing the speaker wire may be simply an RF antenna tuning and orientation effect.

Another cause of audible differences, almost invariably but quite incorrectly attributed to the inherent virtues or deficiencies of speaker wires, is the effect of small series inductances and of small shunt capacitances on the stability of certain amplifiers. Any typical length of speaker wire represents a series L, ranging from a fraction of one microhenry to 8 or 10 or more microhenries, and a shunt C, ranging from a few dozen to a good many thousand picofarads. (This in addition to the series R everyone talks about in connection with power loss and damping factor.) Those values are right in the ball park for either stabilizing or destabilizing a feedback amplifier that can be upset by complex load impedances. A perfect example is the Bedini Mod el 25/25 power amplifier reviewed elsewhere in this issue.

With a 2-microfarad capacitor plugged directly into its output terminals while passing a low-amplitude square wave into an 8-ohm resistive load, this amplifier goes into uncontrolled oscillation and blows its power supply fuses. Nevertheless, when driving electrostatic loudspeakers that are fairly closely modeled by such an RC load, the little Bedini is happy as a lark and sounds gorgeous. Why? Because the couple of microhenries of series inductance introduced by the speaker wire provides a stabilizing trim that isolates the speaker capacitance at the higher frequencies. (The amplifier appears to have no such protection internally.) Now what if we switched to some ultralow-inductance speaker of the braided variety (Polk, etc.)? The isolation would be greatly reduced, a marginal oscillatory condition might reappear, and innocent golden ears would conclude that the new speaker leads are of poor design because the sound is now worse. Whereas if the amplifier had been, say, the Rappaport AMP-1 (now defunct, alas), which is almost totally insensitive to load capacitance, there would have been 'no differawnce.' Conversely, the lower series inductance and higher shunt capacitance of the braided cable might have been a highly beneficial trim for some other slightly peculiar amplifier circuit loaded by some other slightly peculiar speaker. There are more things in amplifiers and speakers, Horatio, than are dreamt of in your cable philosophy ...

Other facts and other fictions.

The idea that a length of cable conducting an audio signal can be modeled as a transmission line, with a characteristic impedance (like 300-ohm TV antenna lead-in), is another fallacy that needs to be disposed of here. Transmission-line effects can begin to come into play when the length of the transmission path is at least a quarter wavelength of the frequency being transmitted. For argument's sake, we'll call 40 kHz to be the highest frequency of interest in the accurate reproduction of music. (See also Part II of the seminar transcript, elsewhere in this issue.) The electrical quarter wavelength of 40 kHz is well over one mile. The electrical quarter wavelength of 18 Hz, at the bottom end of the useful audio spectrum, is over 2500 miles. Therefore, to talk about transmission lines and characteristic impedances in audio wiring is the rankest nonsense. There are no signal losses at audio frequencies due to mismatched cable impedances. If your Polk Audio Soundcable, for example, sounds better than the conventional speaker wire you were using before, the reason is not that its characteristic impedance (the square root of its L/C ratio) has been carefully manipulated to come out at 8 ohms, to match the nominal impedance of your speaker. You already know what some of the real reasons might be.

Then there is the much-discussed skin effect, the tendency of alternating current to flow near the surface of a conductor rather than uniformly through its entire thickness.

This is claimed to create too much resistance to signals at the higher frequencies unless very finely stranded wire is used, with each strand individually insulated (i. e. litz wire).

Again, an examination of the basic physics of this phenomenon shows that it becomes significant only as we approach the megahertz region. We have measured the 10 kHz AC resistance (that is to say the R component of the total Z at 10 kHz) of many different types of speaker wire and can report that it exceeds the 1 kHz AC resistance by no more than 5 or 6 percent in worst-case examples. Since the entire speaker wire represents only 1 to 4 percent of the total load in a typical speaker installation, the power loss due to skin effect alone is most likely of the order of one or two millibels (1 mB = 0.01 dB) at the higher audio frequencies, totally swamped by the more significant though still small losses due to series inductance. Once again, academic bugaboos without a quantitative perspective lead only to equipment hypochondria instead of audible realities.

As for those inductive losses, let's look at a worst-case possibility. In the latest modification of the Beveridge 'System 3' speaker, the impedance of the highly capacitive electrostatic 'line source' is approximately 1 ohm at 20 kHz. Pretty hairy. Now let's assume that in a biamped setup the line source is driven through 15 meters (just about 50 feet) of ordinary No. 14 speaker wire. (Would any audiophile use thinner wire for such a long run?) We measured 0.66 micro henry per meter in this type of wire, so that the total series inductance comes to 10 microhenries. That represents an impedance of about 1/4 ohms at 20 kHz. Combined with the resistance of approximately 0.3 ohm of the 10-meter wire, the total rms impedance at 20 kHz comes to just a little over 1 1/4 ohms, let us say 1.3 ohms. In other words, there will be a larger voltage drop across the wire at this frequency than across the speaker itself, reducing the 20 kHz voltage drive to the speaker by something like 5 dB compared to a noninductive connection of the same resistance. Now that's not academic; it will cause a significant roll-off that should be avoided, especially since the speaker happens to be already rolled off to some degree in the top octave. The solution would be to place the amplifier near the speaker and use as little wire as possible, or alternately to switch to a very low-inductance speaker cable like the Polk and hope it doesn't make the amplifier go unstable with the unisolated capacitive load. This is obviously an extreme example, chosen to prove that speaker wire inductance can be an issue, although more often than not you can safely forget about it. What you, shouldn't forget is that the 'best' choice can turn into the worst under exceptional circumstances. Understanding the total system is the only insurance.

The realistic criteria.

What, then, are the genuinely desirable characteristics of audio cable and speaker wire as we step out of the Transylvanian night vapors into the broad daylight of scientific inquiry? In the case of shielded cables going into and out of the front end of an audio system, we believe the most important criteria are good, clean contacts and effective shielding against both hum pickup and RFI. The ultimate solution, covering all bases, would be a triaxial cable with Camac connectors. Unfortunately, there exists no audio equipment today ready to accept such a cable; in fact the Verion/Cotter type of triaxial cable with conventional phono plugs is still somewhat difficult to interface with equipment having coax ial jacks grounded on the shield side. As for Camac plugs, only Mark Levinson equipment accepts them and only the two-wired kind. In any event, avoid audio cables with cheap, flimsy plugs and light open-mesh or single-spiral shielding.

And never trust a connection until you have tugged at it and found it unshakable and totally noise-free.

A good dielectric, such as Teflon, is also an important requirement in a quality audio cable; dielectric materials chosen with cheap and easy fabrication in mind often exhibit capacitance changes with varying signal frequency and voltage, which may in extreme cases be the cause of spurious modulations of the signal. Since the dielectric is seldom specified in ready-made audio cables, price is generally the best indication of quality, although there may be unfortunate exceptions.

Very low capacitance per unit length matters only in audio cables driven from a high-impedance source. For ex ample, the excellent little Precision Fidelity C7 tube pre amplifier, which has no flat-gain line amplifier stage, may present an output impedance as high as 7000 ohms on account of the level potentiometer used in its passive output section. This preamp definitely needs low-capacitance out put cables, especially for longer runs, otherwise the high frequencies will be rolled off. On the other hand, a preamp like the Hegeman (Hapi), with its 15-ohm output impedance, couldn't care less about the output cable capacitance. You could use 50 meters of the highest capacitance cable you can find and it wouldn't make a bit of difference. Or take tone arm cables. With low-output moving-coil cartridges the cable capacitance is immaterial; with moving-magnet and moving-iron cartridges it enters very much into the correct load calculation. Again, it's the difference in impedance.

Don't let anyone tell you that you must have low-capacitance cables for good sound; ask him to explain why that would help you at the impedances that exist in your particular system.

In speaker wires, the situation is also far from black and-white. Low DC resistance is generally mentioned as the most important criterion, to minimize the influence on damping factor and to waste as little of the available amplifier power as possible. Granted-but, again, watch out for quantitative reality as against qualitative theorizing. The DC resistance of Monster Cable, advertised as ''the high definition speaker wire" and almost as thick as a pair of pencils, is a very impressive 0.01 ohm per meter (0.003 ohm per foot). Conventional No. 14 speaker wire is only half as good in this respect: 0.02 ohm per meter. But a 5-meter length of No. 14 and a 10-meter length of Monster Cable would each add exactly 0.1 ohm to the source impedance seen by the speaker and/or the load impedance seen by amplifier, so that you can't talk about a ''better'" wire with out specifying how much you're using. If you have the amplifier close to the speakers and need only 2 meters (62 feet) of speaker lead per channel, we solemnly guarantee that you won't hear a difference even if you use No. 18 wire. On the other hand, for wiring a 15-meter long recreation room with the speaker leads routed around the baseboard, Monster Cable would be highly desirable. You've got to think numbers, not labels.

Series inductance can also be critical, as we've seen, but it seldom is-and when it is, you must know whether you need higher or lower inductance for your particular amplifier/speaker interface. Most wires you would normally consider for a quality installation are in the range of 0.5 to 1 microhenry per meter; only the Polk type of braided speaker cable is of a totally different order, measuring as low as 0.035 microhenry per meter. With typical amplifiers and typical dynamic speakers exhibiting a rising impedance at the higher frequencies, you need not worry about this criterion.

The capacitance of speaker wire should be of no con sequence to an amplifier that can drive electrostatic speakers with their incomparably higher capacitance, but a few amplifiers can be made unstable specifically with medium capacitance loads in the double-oh to single-oh microfarad region and should therefore not be connected to the speakers with the Polk type of cable, which is 10 to 40 times more capacitive than others. The speaker wire with the lowest capacitance measured in our tests was the imported ILV Lucas, closely followed by ordinary No. 14. Anything in the 40 to 70 picofarads-per-meter range can be considered very low-capacitance speaker wire.

The most farfetched idea about speaker wire performance comes to us from France. It calls attention to the possibility that the distance between the plus and minus leads will be minutely varied by the magnetic field force between the two wires as well as by the acoustical energy in the listening room. This would cause a fluctuation of the energy storage in the speaker leads and thereby modulate the audio signal.

Wild, isn't it-but not completely without plausibility, especially at current levels of several amperes, which are quite common in loud playback through large amplifiers. According to this theory, very rigid speaker cable with solid (un stranded) wires will minimize the effect. We have absolutely no opinion on the subject but are willing to concede that this kind of undesirable modulation might be marginally audible under worst-case conditions (such as the third round of Pernod without water). We haven't been able to verify it.

In fact, we've been able to hear very few and only very small differences among speaker wires and audio cables in our tests so far-and none that couldn't be easily explained by one or more of the considerations discussed above. But then the moon may not have been in the right phase, and unfortunately we were fresh out of wolfsbane . . .

Recommendations

Since there are obviously no unqualified ''bests'" in this product category, we just want to mention informally some brands that have given us good results.

For the most effective hum and RFI shielding, nothing we know of equals the now extinct Verion Triaxial audio cable. The Cotter company (Verion's successor) is using almost exactly the same cable at the output of their transformers and electronic modules but has just barely begun to make it available as a separately purchasable product. Their version, incidentally, has a greatly improved phono plug with a springy ground contact that always grips tight but slips on and off with ease. Really nice.

In a very low-capacitance shielded audio cable, our favorite so far is Denon Audio Cord (of the order of 50 pF per meter). It's thick, rugged, very limp, and comes with high-quality phono plugs.

Our favorite deluxe speaker wire at the moment is Monster Cable. At 0.01 ohm, 70 pF and 0.7 microhenry per meter, it seems highly suitable for just about any application except long, long runs to crazily capacitive speakers such as the Beveridge System 3. Also, it's very limp and flexible, with a transparent vinyl jacket that allows you to inspect the condition of the finely stranded wire at any point. What's more, Monster Cable dealers are equipped to prepare the ends of any length of Monster Cable exactly the way you want it, with spade lugs, Pomona-type double banana plugs, color coding, etc. A very classy, well-thought-out product.

Our second choice would probably be ILV Lucas cable, which is somewhat higher in DC resistance and series inductance but quite a bit lower in capacitance. It may be more readily available to our overseas readers.

As for ordinary speaker wire, No. 14 or even No. 16, there's not a thing wrong with it. Unless there's a special problem as discussed above, you're unlikely to gain anything by switching from it to one of the super cables. No. 18 should be used only for very short runs.

In all cases, remember-electrons obey only the laws of nature, not the dictates of fashion in the audio salons nor the incantations of the high-end shamans and warlocks.

---------

[adapted from TAC]

---------

Also see:

A Spats of Speaker Systems, Large and pro, Good and Bad: Axiom TLT-1 Beveridge System 3 B&W DM7 DCM 'Time Window' (further improved) Fried Model C (improved) Magneplanar Model MG-1; Onkyo Model F-5000; Perspective MK2 QLN I; Sound Lab R-1; Swallow CM70; Vandersteen Model IIA

Why We're So Mean, Vindictive, Arrogant, Negative--and Truthful

Various audio and high-fidelity magazines