Letters responding to this article were published in February, June & August 1996 (Vol.19 Nos.2, 6 & 8):
Cable standards
Editor: Thank you for publishing a scientific study of cable performance using in-circuit measurements ("What a Difference a Wire Makes," December '95, p.95). Author Ben Duncan suggests that low inductance correlates with good measured results. The obvious deficiency is that he did not provide inductance (L), capacitance (C), and resistance (R) measurements for the eight cables he studied in the article. Would it be possible to ask him to supply such measurements? I also wish Stereophile would start measuring speaker wire and interconnects. Martin Colloms published measured results for a number of cables in the July 1990 issue of Hi-Fi News & Record Review. Note that another important effect of cables is their frequency response. High L values tend to roll-off the high frequencies, while the long "settling times" Ben Duncan measured may audibly show as loss of transparency and low-level detail. Given at least the two measurable effects (frequency response and settling times), I'm not surprised that some people report hearing differences among cables. Of course it remains, for another article, to correlate the long settling times with sonic attributes, in a well-controlled blind listening experiment. From the very little data I was able to obtain on only four speaker cables from two manufacturers, I found inductance variation by a factor of 25 and capacitance variation by a factor of 86! How much variation would I have found had I looked at 50 cables?
The FDA requires extensive testing and evaluation of drugs before approval. Makers of bottled and canned foods and drinks are required to report calorie and nutrient contents in a specific way. Tobacco companies are required to publish in their adverts nicotine and tar contents in their cigarettes. Amplifier makers are required by the FTC to report power rating in a very specifc way. Car makers have to report tens, possibly hundreds, of measurements on their cars. Is it too much to require cable and interconnect manufacturers to publish inductance, capacitance, and resistance specifications in their adverts? Maybe it's time for Stereophile to lead in this direction.
Finally, Mr. Duncan's "Further Reading" list (p.105) was very short. Max Kelson has recently posted a list of 36 references on the topic on the rec.audio.high.end (R.A.H.E.) internet newsgroup.—Bahjat F. Qaqish, Chapel Hill, NC
Ben Duncan's list of measured R,L, and C was published in May 1996. I agree that for cable manufacturers to specify these parameters is a good idea. Certainly we should start publishing this information in our cable reviews. Mr. Duncan's further reading list was only intended to be a starting point. If the interest exists, we'll publish a complete list of references. Incidentally, an important earlier article which revealed measured speaker cable differences was written by Ken Cowans, the co-founder of Kinergetics. It appeared in Stereophile in January 1990 (Vol.13 No.1, p.103).—John Atkinson
Cable sonics
Editor: While I generally concur with Stereophile's position regarding cable sonics, recent experiments of mine suggest that you may be doing your readers a disservice. Specifically, I've been playing around with Teflon-insulated, silver-plated, reasonably priced cables from D.H. Labs (stranded) and Apature (solid-core). If my ears and my MartinLogan speakers are not deceiving me, I have found these cables to compare quite favorably with excessively priced esoterica. I may expand my quest to include products from Alpha, et al. Your disservice is in not pursuing similar research. Why have your reviewers never attempted a comparison of exotic cables with similarly constructed off-the-shelf products? Heck, many of the exotic cable manufacturers are really marketeers who merely put their logos on products manufactured under more pedestrian auspices. Surely this would have been suitable grist for the old "Cheapskate" or "Getting Real" mills. Could it be that companies like Alpha and National Wire don't advertise in your periodical? Or is that spending less than $10/ft. induces Audiophilia nervosa in otherwise responsible journalists?—Steve Podvoll, Gardena, CA
As far as I'm aware, the brands of cables we recommend feature proprietary constructions and materials. This is not to say that there are not also good inexpensive cables to be had. We have been remiss in not auditioning a wider selection of such cables. We'll try harder in the future.—John Atkinson
Cable standards
Editor: The obfuscation of information appears to be alive and well in the audio cable industry. Many a so-called "white paper" has been published in attempt to explain away various cable design techniques and their audibility. These same publications have yet to address the real cause of cable differences: the lack of impedance standards for sources and loads! In support of the above idea, it should be said that many years ago (ca 1976), the powers-that-be realized that there were problems in the hi-fi industry with the measurement and specification system regarding FM broadcast tuners. Everyone was using testing methods that attempted to make their products look better than the competition. They (the Institute of High Fidelity or IHF) set about finding and publishing a group of methods for measuring and specifying FM broadcast tuners. This action turned out to be a great success; soon thereafter, most (but not all) manufacturers of FM tuners started using the IHF methods to specify their products. Not being an insider, I can only assume that the manufacturers realized that consumers would catch on to what was happening and the "numbers" farce would become obvious to everyone.
Zoom forward to 1996...witness the proliferation of companies that specialize in the arcane field of audio cable design. The real problem is this: the fault is not the physics of cables, but poor product design.
The lack of standards for audio component engineers has lead to a seriously audible problem which is being "corrected" by the cable industry. Very few companies (Conrad-Johnson comes to mind) have attempted to make their products follow any kind of "standard" regarding source capabilities and load impedances. The "integration" method of designing, wherein a component is designed for use with only a specific product (usually from the same company), is often frowned upon by the audio community for its lack of adaptability. The engineer has designed a product which will perform as advertised when presented with a load of known electrical characteristics.
Designing a product that will predictably work well into all known loads is an impossible task. The best an engineer can do is hope that the consumer will use reasonable loads for signals. The result of this is easy to predict: any cable that calms the errant signals present with improperly terminated loads will be deemed a blessing, and will subsequently be referred to as a superior cable. The fact that no cable works well in all installations outlines this idea very well.
I wonder why equipment designers have not created a de facto standard in attempt to eliminate this whole thing...must be a money thing.—Gary Mattie, Kolbermoor, Germany
Cable physics
Editor: I am writing regarding Ben Duncan's "What a Difference a Wire Makes" (December '95, p.95). The time-domain results shown are hardly surprising, especially given that his test signal "representing a music transient" bears little similarity to a musical signal at all. Musical signals have an unmistakable amplitude envelope, a fact well-known to synthesizer designers. Percussive sounds may start with a rapid onset, but they all end with a controlled decay. The end of the [1kHz] toneburst used in his tests bears more similarity to a squarewave than to a musical source. The maximum slew rate of a sinewave occurs at the zero crossing, and this is where the tone burst suddenly ends. For example, look at his fig.7 (Vol.18 No.12, p.100). Cover up the waveform to the left of the "4.00m" mark. Note that the end of the toneburst not only looks like the rising edge of a squarewave, [this is a visual illusion due to the large vertical amplification used in BD's graphs—Ed.], but also the circuit behaves as if it were (and for good reason), with harmonics similar to those shown in my fig.1.
Cables with greater inductance and amplifiers with lower damping factor will have a higher impedance at higher frequencies, which permits greater ringing with transient signals. The end of a 15kHz toneburst could certainly be considered "transient" since its slew rate is 15 times greater than at 1kHz. Whether or not the >100kHz ringing observed is "common" is unproven without a larger sample of amplifiers. Whether it makes an audible difference is yet another question that perhaps bats could answer.
The amplifier I tested does not exhibit this "imperfection" even at frequencies greater than 15kHz, in spite of the amplifier's relatively large negative feedback. On the other hand, this amplifier's damping factor remains constant and high across the audio spectrum. So which will make the greater difference—the amplifier, cable, or speaker? If Mr. Duncan had used a different amplifier, the cable effects at 15kHz would have been masked less by the amplifier's artifacts.
Even if we were to accept a toneburst with no amplitude envelope control as a possible musical source from an electronic instrument, no evidence is shown that the measured effects are actually audible. If they were in any way audible, then I'm certain that data would have been prominently displayed. Then the phrase "meaningful differences" would be undeniable. If no one can hear it, how important is it?
I can find nothing in Mr. Duncan's results that suggests "logic" concerning special cables for mains. I have never seen current waveforms into power supplies that even remotely resemble an isolated double-sine toneburst. The fundamental frequencies involved are so low (50Hz/60Hz) that even the harmonics will not be profoundly affected by mains-cable inductance (even without considering the tens of thousands of microfarads of filter capacitance that follow in a typical power supply). What degree of real improvement could be expected by changing the last meter or two of cable at the end of several kilometers of mains wiring?
Mr. Duncan's data concur with mine: The effects of a speaker cable, though measurable, are small compared to amplifier artifacts. Once again, we are left with clearly measurable differences between speaker cables, but no proof that these differences are audible.—Fred E. Davis, Connecticut
Cables redux
Editor: I read Fred Davis's letter in the June Stereophile (p.28) with interest. He appears to admit the validity of my cable measurements! But to maintain his belief system, he falls back on simplistic and flawed logic: Small numbers aren't (at all, ever) audible behind (different) big numbers. I find that to counter real-world experience.
Sure, the burst I used isn't like a lot of music waveforms, but it's far, far more like them than the test signals used in 99% of audio electronic path measurements. Such waveforms also will occur for real, when zero voltage switching (zvs) is used to obviate clicks. Such waveforms are therefore embedded in a great deal of recorded music produced on modern mixers that employ zvs analog switching, as well as electronically produced house music. Like many electronic engineers, Mr. Davis might best refrain from defining music unless he is in the thick of it.
Mr. Davis's comment about the quality of the AC mains supply is unimaginative. I have some 11kV (British) power-line frameworks acting as Babylonian sculptures in my garden, so I know a little about the wide spacing, hence high inductance of AC-power transmission lines. Using high-capacitance/low-inductance cable in the final few yards forms a useful circuit known as a low-pass filter. Unlike a single capacitor, the transmission-line capacitance of a speaker cable made from a cable of high mutual inductance will attenuate stuff far above audio, without the high-Q resonances of ordinary capacitors, over a very wide range of radio frequencies, quite possibly up to GHz frequencies. I am sure I don't need to further spell out the benefit of using low-inductance cables for AC power connection in high-resolution audio systems.—Ben Duncan, Tattershall, Lincoln, England
I believe that this letter should bring the Great Cable Debate to a (temporary) close. Malcolm Omar Hawksford's article last October ("The Essex Echo," Vol.18 No.10) showed that, working from first principles, cables have effects on signals propagated through them not predicted by simplistic considerations of the electrical characteristics (L, C, and R). Ben Duncan's article last December ("What a Difference a Wire Makes," Vol.18 No.12, with addenda in May and June '96) revealed measurable differences similar to those predicted by Professor Hawksford's theoretical work. The Audio Establishment, in the form of wire expert Fred Davis, mutters that while such effects can be measured, they don't matter. The rest of us, having found that cable choice can sometimes have a surprisingly large effect on the sounds of our systems, gets on with listening to our music.—John Atkinson
Editor: Thank you for publishing a scientific study of cable performance using in-circuit measurements ("What a Difference a Wire Makes," December '95, p.95). Author Ben Duncan suggests that low inductance correlates with good measured results. The obvious deficiency is that he did not provide inductance (L), capacitance (C), and resistance (R) measurements for the eight cables he studied in the article. Would it be possible to ask him to supply such measurements? I also wish Stereophile would start measuring speaker wire and interconnects. Martin Colloms published measured results for a number of cables in the July 1990 issue of Hi-Fi News & Record Review. Note that another important effect of cables is their frequency response. High L values tend to roll-off the high frequencies, while the long "settling times" Ben Duncan measured may audibly show as loss of transparency and low-level detail. Given at least the two measurable effects (frequency response and settling times), I'm not surprised that some people report hearing differences among cables. Of course it remains, for another article, to correlate the long settling times with sonic attributes, in a well-controlled blind listening experiment. From the very little data I was able to obtain on only four speaker cables from two manufacturers, I found inductance variation by a factor of 25 and capacitance variation by a factor of 86! How much variation would I have found had I looked at 50 cables?
Editor: While I generally concur with Stereophile's position regarding cable sonics, recent experiments of mine suggest that you may be doing your readers a disservice. Specifically, I've been playing around with Teflon-insulated, silver-plated, reasonably priced cables from D.H. Labs (stranded) and Apature (solid-core). If my ears and my MartinLogan speakers are not deceiving me, I have found these cables to compare quite favorably with excessively priced esoterica. I may expand my quest to include products from Alpha, et al. Your disservice is in not pursuing similar research. Why have your reviewers never attempted a comparison of exotic cables with similarly constructed off-the-shelf products? Heck, many of the exotic cable manufacturers are really marketeers who merely put their logos on products manufactured under more pedestrian auspices. Surely this would have been suitable grist for the old "Cheapskate" or "Getting Real" mills. Could it be that companies like Alpha and National Wire don't advertise in your periodical? Or is that spending less than $10/ft. induces Audiophilia nervosa in otherwise responsible journalists?—Steve Podvoll, Gardena, CA
Editor: The obfuscation of information appears to be alive and well in the audio cable industry. Many a so-called "white paper" has been published in attempt to explain away various cable design techniques and their audibility. These same publications have yet to address the real cause of cable differences: the lack of impedance standards for sources and loads! In support of the above idea, it should be said that many years ago (ca 1976), the powers-that-be realized that there were problems in the hi-fi industry with the measurement and specification system regarding FM broadcast tuners. Everyone was using testing methods that attempted to make their products look better than the competition. They (the Institute of High Fidelity or IHF) set about finding and publishing a group of methods for measuring and specifying FM broadcast tuners. This action turned out to be a great success; soon thereafter, most (but not all) manufacturers of FM tuners started using the IHF methods to specify their products. Not being an insider, I can only assume that the manufacturers realized that consumers would catch on to what was happening and the "numbers" farce would become obvious to everyone.
Editor: I am writing regarding Ben Duncan's "What a Difference a Wire Makes" (December '95, p.95). The time-domain results shown are hardly surprising, especially given that his test signal "representing a music transient" bears little similarity to a musical signal at all. Musical signals have an unmistakable amplitude envelope, a fact well-known to synthesizer designers. Percussive sounds may start with a rapid onset, but they all end with a controlled decay. The end of the [1kHz] toneburst used in his tests bears more similarity to a squarewave than to a musical source. The maximum slew rate of a sinewave occurs at the zero crossing, and this is where the tone burst suddenly ends. For example, look at his fig.7 (Vol.18 No.12, p.100). Cover up the waveform to the left of the "4.00m" mark. Note that the end of the toneburst not only looks like the rising edge of a squarewave, [this is a visual illusion due to the large vertical amplification used in BD's graphs—Ed.], but also the circuit behaves as if it were (and for good reason), with harmonics similar to those shown in my fig.1.
Editor: I read Fred Davis's letter in the June Stereophile (p.28) with interest. He appears to admit the validity of my cable measurements! But to maintain his belief system, he falls back on simplistic and flawed logic: Small numbers aren't (at all, ever) audible behind (different) big numbers. I find that to counter real-world experience.
