I thought this was interesting. I'd like Serge or someone similar to let me know if the science bits are sound. He openly declares where he departs from measurements and launches into assertion, but I'm trying to get a feel for the general soundness of his approach, and if the science elements are sound then that provides a context for the more obviously subjective elements.
[align=center]Tip #25: IN PURSUIT OF THE PERFECT CONDUCTOR
by Dick Olsher (March 2001)[/align]
[align=left]Note: The is an abridged and updated version of an article originally published in the January 1996 issue of Fi Magazine. Most of the updates are minor in nature, except for the addition of MAC wireand Kimber Kable Select KS 1030 interconnect to the recommended list.[/align]
[align=left]The notion of cable as a high-end component, so contentious in the 80s, is today rather well accepted by most serious listeners. However, the underlying technical foundation for the sonic character of cable remains nearly as controversial and fuzzy today as it was back then. Well...NOT if you're an electrical engineer. Most of these folks look at the corpus of a cable through entirely materialistic eyes; mainly because they are taught to analyze a circuit in terms of a few fundamental electromagnetic quantities. They idealize and model a cable's behavior as a function of its lumped electrical parameters: resistance ®, inductance (L), and capacitance ©. These parameters are ordained by the cable's dielectric materials and "geometry," by which I mean conductor wire gauge, the number of strands, the spacing between strands in each leg, and the spacing between the "go" and "return" legs. On this basis it is possible to calculate a cable's impedance. And because the calculations agree with the meter readings, this model must be judged as physically valid. [/align]
[align=left]Taking this one step further, it isn't difficult to find instances where a cable's basic electrical parameters do make a sonic difference. Adding the loudspeaker impedance to the model, it is easy to show that since the cable impedance is in series with that of the loudspeaker, a speaker cable can actually modify the loudspeaker's frequency response. It soaks up a greater percentage of the amp's output power as its own impedance rises and that of the load decreases. Imagine a worst case scenario where a 2-ohm dip in the speaker's impedance curve meets up with a 1-ohm cable impedance; the result being about a 3 dB drop in output at that frequency. In general, it's fair to say that cable impedance contributes subtle EQ effects to the overall system sound. In addition, a high-impedance speaker cable increases the apparent output impedance of the power amp. The obvious consequence of that is that the amp's damping factor is reduced. Another not so obvious consequence involves the "de-tuning" of the loudspeaker's crossover network. Crossovers are designed to look into specific source and load impedances, so an increase in the amp's output impedance works to shift the actual crossover point from its intended frequency. All of these effects clearly impact the sound, but because their precise nature is unpredictable and usually deleterious, I would therefore be the last guy on this planet to recommend high-impedance speaker cable. On the other hand, it's important to note that a high-impedance interconnect is normally not a sonic detriment because it operates in a high impedance circuit; a power amp's input impedance typically being on the order of 50,000 ohms. But interconnect capacitance can be an issue if the preamp's output impedance is unusually high. The cable capacitance and output impedance form a low pass network that rolls off the treble. Just how critical is this effect? Lets take a look at a typical scenario. The capacitance of most interconnects measures between 100 and 150 pico Farad (pF) per meter. No serious preamp should have an output impedance greater than 500 ohms, but lets stretch that a bit to 2,500 ohms for the benefit of some tube preamps. Using figures of 150 pF per meter, a 20-foot run of interconnect, and a realistic upper limit of 2,500 ohms for the preamp's output impedance, results in a calculated -3 dB frequency of 73 kHz - and that folks is good enough for me. Undoubtedly, it is a cable's electrical "personality" that gives rise to specific system interactions and much confusion about the merits of any given cable. For example, a high-inductance speaker cable, which would be expected to sound dull in a tonally balanced system, may in fact be hailed as "sonic nirvana" in the context of a bright-sounding system.[/align]
[align=left]A Question of Perception[/align]
[align=left]The decisive mistake engineers make is to ascribe ALL sonic differences to a cable's electrical parameters. It's about as absurd as trying to divine a person's character from his weight, height, and rectal temperature. Card-carrying members of engineering societies by and large tenaciously uphold scientific dogma pounded into their heads during many years of schooling (as a reformed engineer, I should know). Put a gun to their head, and they will maintain till their dying breath that the RLC paradigm is the truth, the whole truth, and nothing but the truth. In the past ten years, a handful of investigators have shown that the RLC paradigm just doesn't go far enough, and that there are other factors that do indeed affect signal transmission. This is also true for other passive parts such as caps and resistors where a simplistic test-bench measurement oriented paradigm has failed to fully account for sonic differences. Turn a Meter Head loose with an Audio Precision System and have him try to differentiate between a mass-market receiver and a Mark Levinson, or for that matter, between a run of #16 awg zip cord and an equal length of high-end cable. It's like trying to judge fine wines on the basis of a chemical analysis. Such measurements are in general not predictive of human perceptions. [/align]
[align=left]To paraphrase Rene Descartes, I hear, therefore I am. Meaning, that the whole is greater than the sum of the parts. While it is possible to dissect a sound field with a variety of frequency and time domain measurements, these meter readings or waterfall plots in no way add up to reflect the emotional reaction I might experience. No wonder science has had such a hard time defining perceptual attributes. Take timbre, for example. The American National Standards Institute defines timbre as "that attribute of a tone by which a listener can judge that two sounds of the same loudness and pitch are dissimilar." Pretty vague if you ask me. The following layman's definition is no better: the perceivable difference between a clarinet middle C and a violin middle C is timbre. To quote Handel (Listening: An Introduction to the Perception of Auditory Events: MIT Press), "timbre is not reducible to an acoustical property that automatically yields a clarinet note or a violin note." Timbre [has] to be judged subjectively. Human vision presents us with a similar perceptual dilemma. The wavelength of visible light can be measured very precisely with a spectrometer. At a wavelength of 520 nm, light is perceived as green; at 470 nm it is blue. But at what wavelength does it change from green to blue? That's a question that measurements cannot settle. There's an infinite number of blue-green shades between these wavelengths, so in the end the answer depends totally on the observer. [/align]
[align=left]It is precisely because sound perception is nearly impossible to predict on the basis of objective measurements that subjective audio reviewing was invented some 40 years ago by J. Gordon Holt. And that's why the ear must be the final arbiter in all things musical. [/align]
[align=left]Beyond the RLC Paradigm[/align]
[align=left]My main thesis is that cable sound is profoundly influenced by its constituent materials. In hindsight, this assertion strikes me as not only sensible but also as self evident. A long-standing audiophile tenet holds that a component's sound is determined to a large extent by its parts quality. Hence, in the case of an active component such as a power amp, we can safely state that identical circuits executed with different ingredients will sound dissimilar. Ditto for passive components such as caps where, for example, the choice of dielectric makes a world of difference. By extension, it is logical to expect that the following factors would impact cable sound: conductor type, its purity, crystal granularity, and choice of cable insulation or dielectric. As with capacitors, the quality of the dielectric is critical in determining the harmonic lucidity of a cable. A fraction of the signal soaks into the dielectric, to be released later in time with a rather slow decay. This has the effect of smearing musical transients and causing textures to turn grainy or gritty. Certainly, dielectrics with small "memory" effects are best for audio applications. The main issue then revolves around the choice of conductor material, which brings to mind questions such as: "is silver superior to copper?" or "is 6N copper audibly better than OFC?" [/align]
[align=left]The Inner View[/align]
[align=left]To understand why factors such as conductor purity are important in signal transmission, it's imperative to examine the conductor's inner world - a microscopic realm inhabited by the ubiquitous electron. Metals are by definition good conductors of electricity. The noblest of these, copper, silver, and gold, are distinguished from insulators by virtue of possessing many more free electrons. These electrons, being detached from atoms in the crystal lattice, are the mobile charge carriers that perform the actual conduction. The poplar view of how a conductor works, has electrons buzzing down Electric Avenue at the speed of light. If that were the case, the cost of a relativistic electron accelerator would be less than a single dollar: simply connect a piece of cable to the terminals of a battery. The truth of the matter is that electrons meander down the length of a conductor at a snail's pace. It is the electromagnetic field associated with the flowing electrons that conducts the signal at the speed of a photon torpedo. This field surrounds the cable, cutting through the space around the conductor. The audio signal travels through the conductor at a fraction of the speed of light in vacuum. The transmission speed increases slightly with signal frequency, so that the bass is slightly delayed relative to the treble. Without the application of an external voltage, free-electron motion is thermal in nature and totally random. In other words, there is no net current flow along the wire. Recall that electric current is defined as the net charge passing a cross section of the wire per second. Thus, in the absence of an applied voltage, there are as many electrons coming and going at any point in the wire, so that the net current is zero. The application of an external voltage to the wire provides a velocity component to the free electron that is parallel to the field. Electrons drift down the crystal lattice, pushed along by the voltage gradient, until they bump into atoms in their path. These atomic collisions stop the forward drift and again deflect the electrons in a random direction. After each collision, the electrons are accelerated again by the potential gradient. The result is random electronic motion characterized by an average drift velocity down the conductor. In a sense it's like a microscopic pin ball game with the electrons (the pin balls) crashing into massive bumpers (the atoms). In both cases, the situation is chaotic; it's impossible to predict the path of either the pin ball or the electron. However, it is possible to calculate the average electronic drift velocity. For a 20-gauge copper conductor and a current of 1 ampere, the actual velocity is on the order of 1 foot per hour. It would in this instance take an electron on average about a day to complete a journey down a 24-foot cable! [/align]
[align=left]The continual scattering of electrons by atoms is a form of friction and constitutes the main source of electrical resistance. The amount of scattering is a function of temperature. As temperature is increased, the atoms vibrate more strongly about their mean position in the crystal lattice and scatter electrons more often. It's a well known fact that the resistivity of a conductor increases with temperature. And as copper, for example, is cooled toward absolute zero its resistance decreases. Ideally, at absolute zero where all thermal motion comes to a screeching halt, its resistance would be zero. I said ideally, because for this to happen requires a perfectly ordered crystal lattice. Quantum mechanics predicts that electrons will not be scattered by a perfect crystal. [/align]
[align=left]Unfortunately a conductor wire is far form a perfect crystal. Numerous lattice defects are introduced during the casting of the metal and drawing of the wire, and these create resistance by interfering with electron motion. A wire, though it may look like a homogeneous mass to the naked eye, is in reality made of a multitude of small crystal grains. Because each grain boundary acts to scatter electrons, metallurgists over the years have invented numerous heat treatments and even a continuous casting process all designed to grow larger grains and thus minimize the number of boundaries.[/align]
[align=left]Just as important an impediment to electron conduction are impurity atoms (mostly oxygen) which lodge in the crystal lattice or precipitate along grain boundaries. The level of impurity is a function of the grade of copper and is often given in parts per million (ppm). Copper as a raw material is available in a number of purity grades. Ordinary Tough Pitch Copper (TPC) is only 99.5% pure. Electrolytically refined copper is 99.9% pure and may be designated as three-nines or 3N in purity. This is the stuff of which common electrical conductors are made of. The next level up is 4N or 99.99% pure. The most significant impurity in 3N copper consists of oxygen atoms at a level of over 200 ppm. Being electronegative, oxygen is adept at latching on to free electrons.[/align]
[align=left]The problems of ordinary zip cord are multiplied by its multi-strand construction. First, there's the problem of strand interaction. Second, by greatly increasing the surface area of the conductor, oxidation of the copper over time is significantly increased over a solid-core design of equivalent gauge. Because copper oxide is a semiconductor material, it behaves as a microscopic diode to rectify low-level audio signals. In terms of purity, Oxygen Free Copper (OFC) represents, in my opinion, the minimum starting purity for audio use. This is a 4N pure material, whose oxygen contamination level is only about a fifth (40 - 60 ppm) of that of TPC. [/align]
[align=left]As an audiophile, the absolutely first question that should come to mind when considering a prospective cable purchase is: what grade is the conductor material?[/align]
[align=left]In 1987 the Nippon Mining company, Ltd., succeeded in implementing copper purification technology suitable for commercial scale production of 6N high-purity copper. The 6N designation, of course, refers to the six nines in its level of purity: 99.99997%. It is a factor of 100 more pure than TPC, with less than 20 ppm oxygen content at crystal grain boundaries. Nippon Mining has also developed the technology for drawing various gauges of 6N copper and of heat-treating or annealing the wire to make it suitable for audio applications. Sold under the trade name of Stressfree 6N, and available for the first time in large quantities and in a variety of forms, the 6N grade has found its way into a diverse cross section of audio products: speaker cable, interconnects, phono cartridge wire, internal component wiring, voice coils, crossover coils, and even power cords. Acrotec's high-purity speaker cable and interconnects remain, to this day, in my stable of reference products.[/align]
[align=left]A New Paradigm[/align]
[align=left]I've already mentioned the use of various heat treatments to increase grain size. However, the efficacy of the various treatments depends on the level of impurities present. Grain growing conditions are optimum when the impurity level is lowest. Hence, OFC wires have much finer grains compared with Stressfree 6N copper. The crystal grains in 6N copper are considerably larger, so that the total number of boundaries is about 80 to 100 times smaller than in 4N copper for similar gauge wire.[/align]
[align=left]It is known that impurity atoms precipitate preferentially at grain boundary sites. Ono and Kato, of Nippon Mining Company, discuss in a 1989 paper (87th AES Convention, October 1989, Preprint 2865) several ideas as to how impurity atoms at such sites may compromise audio signals. First of all, the precipitate impurities act as a nonconductive "wall" impeding current flow by scattering electrons. They also postulate the formation of microscopic "capacitors" at grain boundaries which cause signal phase shifts. Along the same lines, since these impurities attract and trap free electrons, it is possible that trapped electrons are released fractionally later in time and thus in a manner similar to the memory effect of a dielectric act to smear transient detail. At least such theories correlate with my own observations of the sonic impact of decreasing crystalline granularity and increasing conductor purity.[/align]
[align=left]I can still recall Acrotec's demo many years ago at a Winter CES, where they compared 6N copper to 3N or plain vanilla copper. The cables were otherwise identical, as far as geometry and dielectrics. The richer, smoother, and more detailed presentation of the 6N cable still resonates in my memory banks. Over the years, I've had the chance to get to know Acrotec's 6N cable quite intimately, and then the 8N cables arrived at my door step. With the only variable being copper purity and grain size, I was surprised to find that as good as the 6N stuff was, the 8N cable was even more delicate and refined sounding. After all, we're contrasting 99.9999% pure copper with copper that's 99.999999% pure. While those last two nines represent a mere parts per billion improvement in purity, they were sonically audible.[/align]
[align=left]The Silver Bullet[/align]
[align=left]This is about as good a time as any to broach the subject of silver as a conductor material. As a conductor of both heat and electricity silver has no equal. The resistivity of 4N copper is 4.4% higher than that of 4N silver. It is only second to gold in malleability and ductility. One ounce of silver can be drawn into a fine wire about 30 miles long! It is commonly available in a number of grades. A common alloy form is sterling which contains 92.5% silver and 7.5% of usually copper. Jewelry silver is an alloy containing 80% silver and 20% copper. Lab-grade silver is 4N, and 5N as well as 6N grades are available at a significant premium. Outside of the high-end cable industry, silver is used for making printed circuit boards and as a plating for other conductors. High-purity silver wire is clearly a phenomenon of the high-end audio scene, where Siltech, Kimber, Audio Note, and AudioQuest have been the most noteworthy proponents of silver wire.[/align]
[align=left]Available evidence indicates that silver is sonically less affected by oxygen impurity levels than is copper. Silver like copper oxidizes, but unlike copper oxide, silver oxide is a good conductor. That's probably the reason that 4N silver sounds much better than 4N copper. Another practical advantage is silver's greater resistance to mechanical damage. It is more readily shaped and drawn into wire with less crystal damage, so that the granularity of the wire and the number of crystal lattice dislocations is reduced. From this perspective it's easy to argue that silver is superior to copper. My listening tests suggest that when properly annealed and at sufficiently great purity (eg, 8N) copper is sonically the equal of 4N (but not 6N) silver and that other factors beyond materials come into play (see discussion on skin effect). However, OFC and even 6N copper are easily outdistanced by 4N silver in terms of liquidity of harmonic textures, preservation of low-level detail, and microdynamics. [/align]
[align=left]Siltech, the Dutch cable specialist, has adopted the technical position that certain residues in refined silver are instrumental in improving signal transmission. Because Siltech's silver supply is a secondary product of gold mining, Siltech came to realize that gold inclusions in refined silver have a salutary effect on sound quality. That naturally suggests a gold-silver marriage for enhanced signal conduction, which Siltech has implemented in the FTM line. [/align]
[align=left]I do not find silver to be inherently bright sounding, as others have. In my experience, some of the brightest sounding cables have turned out to be stranded copper designs. The worst of the lot could fry a gnat at 10 yards. In contrast, some of the sweetest and most liquid cables I've heard to date have been pure-silver Litz construction. For example, the Audio Note AN-SPX facilitates the upper octaves with remarkable finesse and grace - without even a hint of brightness. Of course, any cable can sound bright under certain conditions, and there were in fact instances where silver cable did sound bright in my system. Take the Kimber 4AG. Its airy and extended top highlights any inherent speaker brightness. Substituted for a rolled off cable and matched with a metal-dome tweeter, I can understand why the finger might be pointed at silver as the culprit. When cable is used as an EQ band aid to correct a tonal problem, substitution of a neutral cable into the chain is bound to cause problems. In the case of the Kimber KCAG interconnect, the situation turned out to be a bit more complicated. Being unshielded, it is prone to RF pickup. Its slightly bright sound in my system was totally mitigated by the application of AudioQuest's RF Stoppers.[/align]
[align=left]Motherhood, Apple Pie, and the Skin Effect[/align]
[align=left]Conductor and dielectric materials impact cable sound in several critical areas. First, the degree to which harmonic textures and colors are retrieved with a semblance of the purity and liquidity of live music is contingent on the level of purity and lack of granularity of the conductor material itself. Second, the "pedigree" of the conductor and dielectric materials influence resolution of low-level detail - including elucidation of the recording venue's reverberant signature. Finally, without the intervention of pure conductors, much of reproduced music's microdynamics would be squashed. [/align]
[align=left]But that's not the end of the story. There's space, the final frontier, and the proper domain of the skin effect. Unless a cable design copes with the skin effect, it will be unable to paint a spatially believable soundstage complete with tightly focused image outlines. Let me go on record publicly to affirm the skin effect's essential role in audio cable design. It has been known for about 100 years that a conductor's self inductance forces high-frequency currents outward, toward the conductor's skin layer. Conductor cross section shrinks with increasing frequency; the electrical consequence being, of course, an increase in cable impedance with frequency. In the time domain, the skin effect may be said to cause phase shifts which delay bass and mid frequencies relative to treble frequencies. Because the skin effect is small over the audio bandwidth, it has been discounted by engineers as a serious factor in cable sound. Yet, listening tests tell another story. [/align]
[align=left]The concept of skin depth is crucial for understanding the implications of the skin effect for cable design. Skin depth is defined for a given frequency and conductor material as the distance into the wire at which the signal decreases by a factor of 2.718. Skin depth obviously decreases with increasing [/align]
[align=left]frequency. The following Table shows calculated values of the skin depth for circular conductors and various materials at a frequency of 20 kHz.[/align]
[align=left] Material Skin Depth in mm at 20 kHz
------------- -----------------------------------------
Aluminum 0.59[/align]
[align=left] Copper 0.47[/align]
[align=left] Carbon 1.34[/align]
[align=left] Gold 0.55[/align]
[align=left] Silver 0.45[/align]
[align=left] Lead 1.64[/align]
[align=left] Titanium 2.34
---------------------------------------------------------------[/align]
[align=left]To minimize the impact of the skin effect, conductor radius should be small relative to the skin depth at the highest frequency of interest. For copper at 20 kHz, the skin depth is 0.47 mm - or about the radius of a 19-gauge (awg) wire. For impedance to remain as [uniform] as possible over the audio bandwidth, 19-gauge or finer wire is indicated. The finer the gauge, the more uniform the impedance magnitude becomes but at the cost of a higher DC resistance, which you should realize by now is almost irrelevant for interconnect design. [/align]
[align=left]Mapleshade/Insound have taken this approach to its practical limit by using a nearly invisible micro-conductor for their interconnect - the thinnest that can be handled by human hands. The conductor is so fine that a good sneeze during assembly, when the wire is still outside the shield, can spell the end. [/align]
[align=left]An important observation to be gleaned from the skin depth Table is that the "worse" the conductor material, the greater the skin depth. Comparing copper to carbon, the skin depth for carbon is a factor of 2.85 larger. That means that a 19-gauge carbon fiber is already small relative to its skin depth. Therefore, at a given frequency, the more resistive the material, the more uniform is the current distribution in the wire. Another way of saying it is that the worse the conductivity, the smaller the skin effect. No wonder then that some Dutch audio pros have embraced the use of carbon in audio cable. In particular, van den Hul has introduced an interconnect based on carbon-fiber technology. As you can see, there is a method to his "madness," but my feeling is that such designs emphasize the skin effect portion of the design formula to the detriment of conductor materials. [/align]
[align=left]High-resistivity interconnects may sport loop resistances in excess of 10 ohms per meter. That's about 100 to 200 times the typical loop resistance of a "normal" interconnect. The world record in this regard (at least for a commercial product) is surely held by Dave Magnan, whose Ultra High-Resolution Signature interconnect uses a conductive polymer for one of its legs. The resultant loop resistance is on the order of 35,000 ohms per meter! Such a level of resistance is bound to cause trouble with some preamps and power amps.[/align]
[align=left]The traditional approach to taming the skin effect while at the same time maintaining low-resistance values involves the use of Litz construction. Litz refers to cables in which many individually insulated strands are used in parallel and the strand radius is smaller than the skin depth at the highest frequency of interest. There is also the requirement for a braiding that allows each strand to occupy as much of a position along the conductor surface as every other strand. In this way, each strand will see the same flux, on average, giving the conductor as a whole better current distribution. Of course, Litz cable opens up a can of worms in terms of interaction or intermodulation between strands, and has provoked some innovative fixes such as the Cardas Golden Section stranding. Cables that have been optimized in regard to impedance uniformity, sound more coherent spatially. Image outlines are brought into sharp focus, massed voices are easier to resolve, and depth perspective is fully resolved. My first exposure to a simple solid-core design was via the TARA Labs Space & Time speaker cable. I was amazed at the time by just how cohesive the harmonic envelope became. The cable's impact was akin to that of a focus knob on a slide projector, as it integrated harmonics into a tight space within the soundstage. [/align]
[align=center]Tip #25: IN PURSUIT OF THE PERFECT CONDUCTOR
by Dick Olsher (March 2001)[/align]
[align=left]Note: The is an abridged and updated version of an article originally published in the January 1996 issue of Fi Magazine. Most of the updates are minor in nature, except for the addition of MAC wireand Kimber Kable Select KS 1030 interconnect to the recommended list.[/align]
[align=left]The notion of cable as a high-end component, so contentious in the 80s, is today rather well accepted by most serious listeners. However, the underlying technical foundation for the sonic character of cable remains nearly as controversial and fuzzy today as it was back then. Well...NOT if you're an electrical engineer. Most of these folks look at the corpus of a cable through entirely materialistic eyes; mainly because they are taught to analyze a circuit in terms of a few fundamental electromagnetic quantities. They idealize and model a cable's behavior as a function of its lumped electrical parameters: resistance ®, inductance (L), and capacitance ©. These parameters are ordained by the cable's dielectric materials and "geometry," by which I mean conductor wire gauge, the number of strands, the spacing between strands in each leg, and the spacing between the "go" and "return" legs. On this basis it is possible to calculate a cable's impedance. And because the calculations agree with the meter readings, this model must be judged as physically valid. [/align]
[align=left]Taking this one step further, it isn't difficult to find instances where a cable's basic electrical parameters do make a sonic difference. Adding the loudspeaker impedance to the model, it is easy to show that since the cable impedance is in series with that of the loudspeaker, a speaker cable can actually modify the loudspeaker's frequency response. It soaks up a greater percentage of the amp's output power as its own impedance rises and that of the load decreases. Imagine a worst case scenario where a 2-ohm dip in the speaker's impedance curve meets up with a 1-ohm cable impedance; the result being about a 3 dB drop in output at that frequency. In general, it's fair to say that cable impedance contributes subtle EQ effects to the overall system sound. In addition, a high-impedance speaker cable increases the apparent output impedance of the power amp. The obvious consequence of that is that the amp's damping factor is reduced. Another not so obvious consequence involves the "de-tuning" of the loudspeaker's crossover network. Crossovers are designed to look into specific source and load impedances, so an increase in the amp's output impedance works to shift the actual crossover point from its intended frequency. All of these effects clearly impact the sound, but because their precise nature is unpredictable and usually deleterious, I would therefore be the last guy on this planet to recommend high-impedance speaker cable. On the other hand, it's important to note that a high-impedance interconnect is normally not a sonic detriment because it operates in a high impedance circuit; a power amp's input impedance typically being on the order of 50,000 ohms. But interconnect capacitance can be an issue if the preamp's output impedance is unusually high. The cable capacitance and output impedance form a low pass network that rolls off the treble. Just how critical is this effect? Lets take a look at a typical scenario. The capacitance of most interconnects measures between 100 and 150 pico Farad (pF) per meter. No serious preamp should have an output impedance greater than 500 ohms, but lets stretch that a bit to 2,500 ohms for the benefit of some tube preamps. Using figures of 150 pF per meter, a 20-foot run of interconnect, and a realistic upper limit of 2,500 ohms for the preamp's output impedance, results in a calculated -3 dB frequency of 73 kHz - and that folks is good enough for me. Undoubtedly, it is a cable's electrical "personality" that gives rise to specific system interactions and much confusion about the merits of any given cable. For example, a high-inductance speaker cable, which would be expected to sound dull in a tonally balanced system, may in fact be hailed as "sonic nirvana" in the context of a bright-sounding system.[/align]
[align=left]A Question of Perception[/align]
[align=left]The decisive mistake engineers make is to ascribe ALL sonic differences to a cable's electrical parameters. It's about as absurd as trying to divine a person's character from his weight, height, and rectal temperature. Card-carrying members of engineering societies by and large tenaciously uphold scientific dogma pounded into their heads during many years of schooling (as a reformed engineer, I should know). Put a gun to their head, and they will maintain till their dying breath that the RLC paradigm is the truth, the whole truth, and nothing but the truth. In the past ten years, a handful of investigators have shown that the RLC paradigm just doesn't go far enough, and that there are other factors that do indeed affect signal transmission. This is also true for other passive parts such as caps and resistors where a simplistic test-bench measurement oriented paradigm has failed to fully account for sonic differences. Turn a Meter Head loose with an Audio Precision System and have him try to differentiate between a mass-market receiver and a Mark Levinson, or for that matter, between a run of #16 awg zip cord and an equal length of high-end cable. It's like trying to judge fine wines on the basis of a chemical analysis. Such measurements are in general not predictive of human perceptions. [/align]
[align=left]To paraphrase Rene Descartes, I hear, therefore I am. Meaning, that the whole is greater than the sum of the parts. While it is possible to dissect a sound field with a variety of frequency and time domain measurements, these meter readings or waterfall plots in no way add up to reflect the emotional reaction I might experience. No wonder science has had such a hard time defining perceptual attributes. Take timbre, for example. The American National Standards Institute defines timbre as "that attribute of a tone by which a listener can judge that two sounds of the same loudness and pitch are dissimilar." Pretty vague if you ask me. The following layman's definition is no better: the perceivable difference between a clarinet middle C and a violin middle C is timbre. To quote Handel (Listening: An Introduction to the Perception of Auditory Events: MIT Press), "timbre is not reducible to an acoustical property that automatically yields a clarinet note or a violin note." Timbre [has] to be judged subjectively. Human vision presents us with a similar perceptual dilemma. The wavelength of visible light can be measured very precisely with a spectrometer. At a wavelength of 520 nm, light is perceived as green; at 470 nm it is blue. But at what wavelength does it change from green to blue? That's a question that measurements cannot settle. There's an infinite number of blue-green shades between these wavelengths, so in the end the answer depends totally on the observer. [/align]
[align=left]It is precisely because sound perception is nearly impossible to predict on the basis of objective measurements that subjective audio reviewing was invented some 40 years ago by J. Gordon Holt. And that's why the ear must be the final arbiter in all things musical. [/align]
[align=left]Beyond the RLC Paradigm[/align]
[align=left]My main thesis is that cable sound is profoundly influenced by its constituent materials. In hindsight, this assertion strikes me as not only sensible but also as self evident. A long-standing audiophile tenet holds that a component's sound is determined to a large extent by its parts quality. Hence, in the case of an active component such as a power amp, we can safely state that identical circuits executed with different ingredients will sound dissimilar. Ditto for passive components such as caps where, for example, the choice of dielectric makes a world of difference. By extension, it is logical to expect that the following factors would impact cable sound: conductor type, its purity, crystal granularity, and choice of cable insulation or dielectric. As with capacitors, the quality of the dielectric is critical in determining the harmonic lucidity of a cable. A fraction of the signal soaks into the dielectric, to be released later in time with a rather slow decay. This has the effect of smearing musical transients and causing textures to turn grainy or gritty. Certainly, dielectrics with small "memory" effects are best for audio applications. The main issue then revolves around the choice of conductor material, which brings to mind questions such as: "is silver superior to copper?" or "is 6N copper audibly better than OFC?" [/align]
[align=left]The Inner View[/align]
[align=left]To understand why factors such as conductor purity are important in signal transmission, it's imperative to examine the conductor's inner world - a microscopic realm inhabited by the ubiquitous electron. Metals are by definition good conductors of electricity. The noblest of these, copper, silver, and gold, are distinguished from insulators by virtue of possessing many more free electrons. These electrons, being detached from atoms in the crystal lattice, are the mobile charge carriers that perform the actual conduction. The poplar view of how a conductor works, has electrons buzzing down Electric Avenue at the speed of light. If that were the case, the cost of a relativistic electron accelerator would be less than a single dollar: simply connect a piece of cable to the terminals of a battery. The truth of the matter is that electrons meander down the length of a conductor at a snail's pace. It is the electromagnetic field associated with the flowing electrons that conducts the signal at the speed of a photon torpedo. This field surrounds the cable, cutting through the space around the conductor. The audio signal travels through the conductor at a fraction of the speed of light in vacuum. The transmission speed increases slightly with signal frequency, so that the bass is slightly delayed relative to the treble. Without the application of an external voltage, free-electron motion is thermal in nature and totally random. In other words, there is no net current flow along the wire. Recall that electric current is defined as the net charge passing a cross section of the wire per second. Thus, in the absence of an applied voltage, there are as many electrons coming and going at any point in the wire, so that the net current is zero. The application of an external voltage to the wire provides a velocity component to the free electron that is parallel to the field. Electrons drift down the crystal lattice, pushed along by the voltage gradient, until they bump into atoms in their path. These atomic collisions stop the forward drift and again deflect the electrons in a random direction. After each collision, the electrons are accelerated again by the potential gradient. The result is random electronic motion characterized by an average drift velocity down the conductor. In a sense it's like a microscopic pin ball game with the electrons (the pin balls) crashing into massive bumpers (the atoms). In both cases, the situation is chaotic; it's impossible to predict the path of either the pin ball or the electron. However, it is possible to calculate the average electronic drift velocity. For a 20-gauge copper conductor and a current of 1 ampere, the actual velocity is on the order of 1 foot per hour. It would in this instance take an electron on average about a day to complete a journey down a 24-foot cable! [/align]
[align=left]The continual scattering of electrons by atoms is a form of friction and constitutes the main source of electrical resistance. The amount of scattering is a function of temperature. As temperature is increased, the atoms vibrate more strongly about their mean position in the crystal lattice and scatter electrons more often. It's a well known fact that the resistivity of a conductor increases with temperature. And as copper, for example, is cooled toward absolute zero its resistance decreases. Ideally, at absolute zero where all thermal motion comes to a screeching halt, its resistance would be zero. I said ideally, because for this to happen requires a perfectly ordered crystal lattice. Quantum mechanics predicts that electrons will not be scattered by a perfect crystal. [/align]
[align=left]Unfortunately a conductor wire is far form a perfect crystal. Numerous lattice defects are introduced during the casting of the metal and drawing of the wire, and these create resistance by interfering with electron motion. A wire, though it may look like a homogeneous mass to the naked eye, is in reality made of a multitude of small crystal grains. Because each grain boundary acts to scatter electrons, metallurgists over the years have invented numerous heat treatments and even a continuous casting process all designed to grow larger grains and thus minimize the number of boundaries.[/align]
[align=left]Just as important an impediment to electron conduction are impurity atoms (mostly oxygen) which lodge in the crystal lattice or precipitate along grain boundaries. The level of impurity is a function of the grade of copper and is often given in parts per million (ppm). Copper as a raw material is available in a number of purity grades. Ordinary Tough Pitch Copper (TPC) is only 99.5% pure. Electrolytically refined copper is 99.9% pure and may be designated as three-nines or 3N in purity. This is the stuff of which common electrical conductors are made of. The next level up is 4N or 99.99% pure. The most significant impurity in 3N copper consists of oxygen atoms at a level of over 200 ppm. Being electronegative, oxygen is adept at latching on to free electrons.[/align]
[align=left]The problems of ordinary zip cord are multiplied by its multi-strand construction. First, there's the problem of strand interaction. Second, by greatly increasing the surface area of the conductor, oxidation of the copper over time is significantly increased over a solid-core design of equivalent gauge. Because copper oxide is a semiconductor material, it behaves as a microscopic diode to rectify low-level audio signals. In terms of purity, Oxygen Free Copper (OFC) represents, in my opinion, the minimum starting purity for audio use. This is a 4N pure material, whose oxygen contamination level is only about a fifth (40 - 60 ppm) of that of TPC. [/align]
[align=left]As an audiophile, the absolutely first question that should come to mind when considering a prospective cable purchase is: what grade is the conductor material?[/align]
[align=left]In 1987 the Nippon Mining company, Ltd., succeeded in implementing copper purification technology suitable for commercial scale production of 6N high-purity copper. The 6N designation, of course, refers to the six nines in its level of purity: 99.99997%. It is a factor of 100 more pure than TPC, with less than 20 ppm oxygen content at crystal grain boundaries. Nippon Mining has also developed the technology for drawing various gauges of 6N copper and of heat-treating or annealing the wire to make it suitable for audio applications. Sold under the trade name of Stressfree 6N, and available for the first time in large quantities and in a variety of forms, the 6N grade has found its way into a diverse cross section of audio products: speaker cable, interconnects, phono cartridge wire, internal component wiring, voice coils, crossover coils, and even power cords. Acrotec's high-purity speaker cable and interconnects remain, to this day, in my stable of reference products.[/align]
[align=left]A New Paradigm[/align]
[align=left]I've already mentioned the use of various heat treatments to increase grain size. However, the efficacy of the various treatments depends on the level of impurities present. Grain growing conditions are optimum when the impurity level is lowest. Hence, OFC wires have much finer grains compared with Stressfree 6N copper. The crystal grains in 6N copper are considerably larger, so that the total number of boundaries is about 80 to 100 times smaller than in 4N copper for similar gauge wire.[/align]
[align=left]It is known that impurity atoms precipitate preferentially at grain boundary sites. Ono and Kato, of Nippon Mining Company, discuss in a 1989 paper (87th AES Convention, October 1989, Preprint 2865) several ideas as to how impurity atoms at such sites may compromise audio signals. First of all, the precipitate impurities act as a nonconductive "wall" impeding current flow by scattering electrons. They also postulate the formation of microscopic "capacitors" at grain boundaries which cause signal phase shifts. Along the same lines, since these impurities attract and trap free electrons, it is possible that trapped electrons are released fractionally later in time and thus in a manner similar to the memory effect of a dielectric act to smear transient detail. At least such theories correlate with my own observations of the sonic impact of decreasing crystalline granularity and increasing conductor purity.[/align]
[align=left]I can still recall Acrotec's demo many years ago at a Winter CES, where they compared 6N copper to 3N or plain vanilla copper. The cables were otherwise identical, as far as geometry and dielectrics. The richer, smoother, and more detailed presentation of the 6N cable still resonates in my memory banks. Over the years, I've had the chance to get to know Acrotec's 6N cable quite intimately, and then the 8N cables arrived at my door step. With the only variable being copper purity and grain size, I was surprised to find that as good as the 6N stuff was, the 8N cable was even more delicate and refined sounding. After all, we're contrasting 99.9999% pure copper with copper that's 99.999999% pure. While those last two nines represent a mere parts per billion improvement in purity, they were sonically audible.[/align]
[align=left]The Silver Bullet[/align]
[align=left]This is about as good a time as any to broach the subject of silver as a conductor material. As a conductor of both heat and electricity silver has no equal. The resistivity of 4N copper is 4.4% higher than that of 4N silver. It is only second to gold in malleability and ductility. One ounce of silver can be drawn into a fine wire about 30 miles long! It is commonly available in a number of grades. A common alloy form is sterling which contains 92.5% silver and 7.5% of usually copper. Jewelry silver is an alloy containing 80% silver and 20% copper. Lab-grade silver is 4N, and 5N as well as 6N grades are available at a significant premium. Outside of the high-end cable industry, silver is used for making printed circuit boards and as a plating for other conductors. High-purity silver wire is clearly a phenomenon of the high-end audio scene, where Siltech, Kimber, Audio Note, and AudioQuest have been the most noteworthy proponents of silver wire.[/align]
[align=left]Available evidence indicates that silver is sonically less affected by oxygen impurity levels than is copper. Silver like copper oxidizes, but unlike copper oxide, silver oxide is a good conductor. That's probably the reason that 4N silver sounds much better than 4N copper. Another practical advantage is silver's greater resistance to mechanical damage. It is more readily shaped and drawn into wire with less crystal damage, so that the granularity of the wire and the number of crystal lattice dislocations is reduced. From this perspective it's easy to argue that silver is superior to copper. My listening tests suggest that when properly annealed and at sufficiently great purity (eg, 8N) copper is sonically the equal of 4N (but not 6N) silver and that other factors beyond materials come into play (see discussion on skin effect). However, OFC and even 6N copper are easily outdistanced by 4N silver in terms of liquidity of harmonic textures, preservation of low-level detail, and microdynamics. [/align]
[align=left]Siltech, the Dutch cable specialist, has adopted the technical position that certain residues in refined silver are instrumental in improving signal transmission. Because Siltech's silver supply is a secondary product of gold mining, Siltech came to realize that gold inclusions in refined silver have a salutary effect on sound quality. That naturally suggests a gold-silver marriage for enhanced signal conduction, which Siltech has implemented in the FTM line. [/align]
[align=left]I do not find silver to be inherently bright sounding, as others have. In my experience, some of the brightest sounding cables have turned out to be stranded copper designs. The worst of the lot could fry a gnat at 10 yards. In contrast, some of the sweetest and most liquid cables I've heard to date have been pure-silver Litz construction. For example, the Audio Note AN-SPX facilitates the upper octaves with remarkable finesse and grace - without even a hint of brightness. Of course, any cable can sound bright under certain conditions, and there were in fact instances where silver cable did sound bright in my system. Take the Kimber 4AG. Its airy and extended top highlights any inherent speaker brightness. Substituted for a rolled off cable and matched with a metal-dome tweeter, I can understand why the finger might be pointed at silver as the culprit. When cable is used as an EQ band aid to correct a tonal problem, substitution of a neutral cable into the chain is bound to cause problems. In the case of the Kimber KCAG interconnect, the situation turned out to be a bit more complicated. Being unshielded, it is prone to RF pickup. Its slightly bright sound in my system was totally mitigated by the application of AudioQuest's RF Stoppers.[/align]
[align=left]Motherhood, Apple Pie, and the Skin Effect[/align]
[align=left]Conductor and dielectric materials impact cable sound in several critical areas. First, the degree to which harmonic textures and colors are retrieved with a semblance of the purity and liquidity of live music is contingent on the level of purity and lack of granularity of the conductor material itself. Second, the "pedigree" of the conductor and dielectric materials influence resolution of low-level detail - including elucidation of the recording venue's reverberant signature. Finally, without the intervention of pure conductors, much of reproduced music's microdynamics would be squashed. [/align]
[align=left]But that's not the end of the story. There's space, the final frontier, and the proper domain of the skin effect. Unless a cable design copes with the skin effect, it will be unable to paint a spatially believable soundstage complete with tightly focused image outlines. Let me go on record publicly to affirm the skin effect's essential role in audio cable design. It has been known for about 100 years that a conductor's self inductance forces high-frequency currents outward, toward the conductor's skin layer. Conductor cross section shrinks with increasing frequency; the electrical consequence being, of course, an increase in cable impedance with frequency. In the time domain, the skin effect may be said to cause phase shifts which delay bass and mid frequencies relative to treble frequencies. Because the skin effect is small over the audio bandwidth, it has been discounted by engineers as a serious factor in cable sound. Yet, listening tests tell another story. [/align]
[align=left]The concept of skin depth is crucial for understanding the implications of the skin effect for cable design. Skin depth is defined for a given frequency and conductor material as the distance into the wire at which the signal decreases by a factor of 2.718. Skin depth obviously decreases with increasing [/align]
[align=left]frequency. The following Table shows calculated values of the skin depth for circular conductors and various materials at a frequency of 20 kHz.[/align]
[align=left] Material Skin Depth in mm at 20 kHz
------------- -----------------------------------------
Aluminum 0.59[/align]
[align=left] Copper 0.47[/align]
[align=left] Carbon 1.34[/align]
[align=left] Gold 0.55[/align]
[align=left] Silver 0.45[/align]
[align=left] Lead 1.64[/align]
[align=left] Titanium 2.34
---------------------------------------------------------------[/align]
[align=left]To minimize the impact of the skin effect, conductor radius should be small relative to the skin depth at the highest frequency of interest. For copper at 20 kHz, the skin depth is 0.47 mm - or about the radius of a 19-gauge (awg) wire. For impedance to remain as [uniform] as possible over the audio bandwidth, 19-gauge or finer wire is indicated. The finer the gauge, the more uniform the impedance magnitude becomes but at the cost of a higher DC resistance, which you should realize by now is almost irrelevant for interconnect design. [/align]
[align=left]Mapleshade/Insound have taken this approach to its practical limit by using a nearly invisible micro-conductor for their interconnect - the thinnest that can be handled by human hands. The conductor is so fine that a good sneeze during assembly, when the wire is still outside the shield, can spell the end. [/align]
[align=left]An important observation to be gleaned from the skin depth Table is that the "worse" the conductor material, the greater the skin depth. Comparing copper to carbon, the skin depth for carbon is a factor of 2.85 larger. That means that a 19-gauge carbon fiber is already small relative to its skin depth. Therefore, at a given frequency, the more resistive the material, the more uniform is the current distribution in the wire. Another way of saying it is that the worse the conductivity, the smaller the skin effect. No wonder then that some Dutch audio pros have embraced the use of carbon in audio cable. In particular, van den Hul has introduced an interconnect based on carbon-fiber technology. As you can see, there is a method to his "madness," but my feeling is that such designs emphasize the skin effect portion of the design formula to the detriment of conductor materials. [/align]
[align=left]High-resistivity interconnects may sport loop resistances in excess of 10 ohms per meter. That's about 100 to 200 times the typical loop resistance of a "normal" interconnect. The world record in this regard (at least for a commercial product) is surely held by Dave Magnan, whose Ultra High-Resolution Signature interconnect uses a conductive polymer for one of its legs. The resultant loop resistance is on the order of 35,000 ohms per meter! Such a level of resistance is bound to cause trouble with some preamps and power amps.[/align]
[align=left]The traditional approach to taming the skin effect while at the same time maintaining low-resistance values involves the use of Litz construction. Litz refers to cables in which many individually insulated strands are used in parallel and the strand radius is smaller than the skin depth at the highest frequency of interest. There is also the requirement for a braiding that allows each strand to occupy as much of a position along the conductor surface as every other strand. In this way, each strand will see the same flux, on average, giving the conductor as a whole better current distribution. Of course, Litz cable opens up a can of worms in terms of interaction or intermodulation between strands, and has provoked some innovative fixes such as the Cardas Golden Section stranding. Cables that have been optimized in regard to impedance uniformity, sound more coherent spatially. Image outlines are brought into sharp focus, massed voices are easier to resolve, and depth perspective is fully resolved. My first exposure to a simple solid-core design was via the TARA Labs Space & Time speaker cable. I was amazed at the time by just how cohesive the harmonic envelope became. The cable's impact was akin to that of a focus knob on a slide projector, as it integrated harmonics into a tight space within the soundstage. [/align]
