Design of Audio Interconnects

An insight into RFC Audio Cable design philosopy

Design of Audio Interconnects

So what’s in an audio interconnect cable and what does it do?  The choice of audiophile interconnects and cables can seem bewildering so to help you choose an audio interconnect, RFC has produced this guide.

Interconnects come in various types, the main ones being:

Balanced: Designed for balanced audio circuits and consisting of a “hot” (positive), “cold” (negative polarity) and “Ground”. The two signal conductors have a signal which is symmetrical in respect of ground with the “hot” wire being the in phase signal (zero degrees) and the “cold” wire being the out of phase (-180 degrees). Since these two conductors have equal impedance relative to ground, and the circuit only measures differences between them, then any noise induced into the wires is cancelled out at the receiving end.

Unlike unbalanced (or single ended) connections, the ground conductor between the pieces of kit being connected carries no signal, so any measurable voltage differences between their chassis cannot be added to the signal as ground noise current. To further explain: think of the hot and cold conductors as carrying the same amplitude signal but as a mirror image of one another with the hot swinging above the zero volt reference and the cold swinging below the zero volt reference, so the signal is carried differentially, with their levels being identical but their polarities being opposite. The polarity alters with the frequency of the signal and the amplitude, (or voltage) of the frequency is the sum of the differences between the voltages in each conductor.

So if the hot conductor carries 1volt, the cold will carry minus 1 volt and the signal level will be 1V-(-1V) = 2V. Noise which induced by RFI and EMI (Radio Frequency Interference from strong RF transmissions and Electromagnetic Interference from power transformers and high voltage power lines) is cancelled out via common mode noise rejection. This works on the principle that the noise induced into one conductor will be cancelled out by the noise in the other. For example. If we take our example of 2V signal made up from 1-(- 1V) = 2V, inducing the same noise in each of say o.5V we get (1+0.5)-(-1+0.5) = 2V, which is the same as the signal value without induced noise.

To ensure that as near as possible, the same value of the interference “noise” or voltage is induced in each conductor, the conductors are closely twisted together, the closer the twist, the more effective the common mode noise rejection becomes. Whilst most balanced cable types tend to be twin core plus shield, there is another configuration that is even more effective at EMI rejection and this is known as the Star Quad configuration consisting of four conductors (two twisted pairs twisted together) . This configuration further reduces the antenna loop area (in fact reducing the effects of EMI by a factor of 10). Whilst this configuration also increases capacitance, for line level applications that have sources with low output impedance and relatively short cable lengths, it results in very good low noise cable. This is the configuration that RFC’s Fidelity Hyperion balanced interconnect uses. Balanced audio leads are terminated in three pin XLR plugs and can also terminated for studio use in 1/4 inch stereo Jack plugs.

Unbalanced: Most hifi circuits are of this configuration where the signal is what’s known as “Single Ended”, ie it has one signal conductor (and one ground return conductor which for coaxial cable is the shield) and is referenced to a zero V signal return, or ground. The shield or ground couples the kit being connected and any voltage differences between the kit chassis are carried in the ground as noise current. Since the shield, or ground wire has finite resistance, the result is a voltage of magnitude “x” which exists along the ground. Voltage “x” is in series (unlike in balanced connections where it is not in series) with the signal voltage and it’s equal to the common mode current multiplied by shield resistance. Since it is in series with the signal, it will add directly to it at the receiver so has no mechanism for common mode noise rejection. By reducing the shield or ground resistance, the Signal to Noise ratio can be increased thus mitigating the effects of that noise on the signal. Typical arrangements include twisted pairs, Litz Braid and coaxial cable terminated in RCA phono plugs. It can bee seen that as there is no common mode noise rejection in unbalanced circuits, the low noise performance of an unbalanced interconnect becomes largely dependant upon two things: The value of the ground conductor resistance and effective shielding of the signal from RFI.

Effective Unbalanced Cables:

The most effective form of unbalanced cable by far is the shielded coaxial (coax) cable. This uses asymmetrical conductors, typically a central signal conductor wire, surrounded by a braided or spiral shield. In order to be effective at reducing external noise, the ground or shield must have very good coverage and for reducing the impacts of external noise, must have lower impedance than the signal wire (ie be more conductive).

There is a common misconception that a simple un-shielded twisted equal pair (sometimes referred to as the “perfect” twisted pair) somehow cancels noise in an unbalanced circuit, but since there’s no common mode noise rejection the signal will pick up noise as well as raising capacitance which is a factor of each conductor’s diameter and distance apart (plus the insulation between them). By twisting an unbalanced pair, the magnitude of some EMI noise can be mitigated (but not avoided) by reducing antenna loop area and this is likely where the misconceptions are about rejecting noise (ie it helps but it doesn’t prevent it from happening). Generally speaking, unshielded twisted pair signal cables used in unbalanced circuits and the term “high fidelity” are not to be confused!

Multi braided cables (often mistakenly referred to as “Litz braid”) usually employ a greater number of strands for the ground return than the signal so the theory is that the ground return is less resistive than the signal which mitigates against noise and that the greater number of wires used forms a better “shield effect” than the simple twisted pair. This configuration is arguably more effective at reducing common mode noise than a twisted pair by virtue of the better conductivity of the ground, but it also has several drawbacks. Firstly, by increasing the number of individual wires, capacitance is raised. Secondly, using more wire strands than a twisted pair offers no real increased shielding effect against RFI and EMI. The gaps between the ground conductors and the signal are simply too large, as with the twisted pair configuration.

Generally, for low noise environments, and in line level applications, braid can be satisfactory if the number of conductors is limited to 3 (one signal, two return) and the wires individually insulated in a something like cotton (air gaps in the material provide a low dielectric constant and cotton is a good damping material). Satisfactory interconnects are possible using this method.  However, they are no match for an overall shielded  cable.

Wires loosely threaded through PTFE tubing as seems the fashion amongst some sellers is not recommended and is largely provided as it’s easy to construct and convenient. The distance between the conductors is variable as the air gap is inconsistent, there are triboelectric effects, particularly where silver conductors are concerned, this type of construction is not recommended.  Silver is often described as somehow “brighter” sounding but this misconception may be linked to the use of silver in PTFE dielectrics which can result in HF “ringing”.  Silver, like copper, has no sound. It is a conductor. the main advantage of using it, is to enable thinner core conductors which in any given geometry can reduce capacitance but often at the increased risk of core fracture at the connector joint.

 

Types of Shielded Unbalanced Cables:

There are two types commonly employed for audio signal leads: single core coax and twin core coax. The former is a simple single core arrangement but suffers from one problem, in that the ground return needed to complete the signal also carries any additional noise from RFI and EMI which is additive to the noise induced into the signal. One way of mitigating this is to use twin core shielded unbalanced cable. The shield is connected at one end only (usually the source end) and acts effectively to pick up and shunt unwanted RFI and EMI noise to the source component ground without being induced into the signal or ground return.

That’s the theory; in practice there are limitations of this design as some noise inevitably will be picked up through gaps in the shielding. It also raises capacitance, but providing cable length isn’t excessive (and for almost all conceivable home audio applications, it won’t be), the capacitance will be within perfectly acceptable limits for most line level sources which are generally have low source impedance (ie the circuit is not as prone to HF signal loss through higher capacitive loading). Often, to combat the capacitance increase, the overall coaxial cable thickness is increased. Both single and twin core coaxial cables make very good high fidelity signal interconnects.

One thing is often missed though, especially where phono and CDP/DAC interconnects (to the amplifier) are concerned:  the signal has a ground return, which is part of the circuit so total signal transmission distance includes this return portion and many CDPs for example do benefit audibly by keeping interconnect lengths as short as possible to avoid things like internal signal reflections.

Types of Shielding:

There are two basic groups of shield employed for signal cables: the outer screen and quite often an inner screen located usually just underneath the main outer shield. Each has a different function. The outer shield, usually copper, is used in unbalanced single core interconnects to act as the signal return and to act as a barrier against unwanted noise getting through to the signal conductor (s). In balanced circuits it does not form part of the signal return. To be effective, it must be tied to a ground at both ends (single core coaxial cable in unbalanced circuits and twin core used in balanced circuits) or at one end for twin core coaxial used in unbalanced circuits.

There is a misconception that twin core coaxial cable in unbalanced circuits must have the shield grounded at both ends to be effective, but this is not true. People often quote the issue of a shield being connected at one end acting like an antenna and inducting noise in the signal. Providing it’s grounded to a component chassis which itself has a ground connection to the AC power source ground, any stray noise will be shunted to that ground. At frequencies below 100KHz (ie all audio frequencies) the transfer impedance of the shield will equal its DC resistance. At frequencies above 100KHz skin effect will increase transfer impedance but we do not have to consider that for audio frequencies.

It’s not sufficient just to have a low impedance outer copper screen. That screen must also have good optical coverage, ie have few gaps for interference to get though. The highest quality braided screens have around 97% optical coverage and the weave means that the coverage remains constant even if the cable is flexed or bent, so it is a very effective shield. Its is however more expensive to make than the spiral shielding found on less expensive interconnects. The other benefit of a lapped or woven shield is that is has very low inductance, so provides excellent low transfer impedance at higher frequencies.

Spiral shielding is also effective and is more flexible than braided shielding but repeated bending and flexing can open small gaps in the shield barrier reducing its long term effectiveness. Spiral shields by their very construction act like an inductor coil so have very high inductance and this means that their transfer impedance increases with frequency thus they become less effective as the frequency rises, but again this is generally not a problem for audio frequencies, just those above 100 KHz.

The inner shield is generally made of Mylar Foil. Its function is principally to block any RFI getting through the outer shield as it has a 100% coverage. It is not as conductive as copper and with flexing and over time, the conductive plating can wear and allow gaps for RFI to migrate through.

There is also another type of inner shield which is sometimes referred to as the electrostatic shield. Whenever the cable is moved or flexed, the contact of the copper outer shield with the insulation creates electrostatic charge which in certain circumstances can be heard as small crackles through the loudspeakers. In effect the cable is acting as a capacitor with the outer shield being one plate, the insulation being the dielectric and the inner core being the other plate. When one or both of these “plates” are moved or deflected, a voltage is created which is coupled as electrostatic noise in the signal. The electrostatic shield is there to bleed off the small charges that are formed by cable movement. This type of electric charge created in a cable is often referred to as triboelectric charge and is the reason why interconnect cables DO suffer from microphonic noise as vibrations (as well flexing of the cable) also produces small triboelectric charges. However, the value of the microphonic noise in home audio set ups is generally quite low as to be of little consequence, but the theory is still valid.

 

So what matters?

1. LCR Parameters

The term “LCR” is commonly used to describe Inductance, Capacitance and Resistance (or in the case of AC signals, Impedance). This is principally what matters for all audio signal cables. Interconnects differ from speaker cables in that they are principally in a low current, high impedance circuit and speaker cables are in a high current circuit driving low impedance ‘speaker loads.

Whats the difference? Well in low current systems of milliamps but with driving voltages of between a few millivolts and say 5 volts, the signal is less prone to inductance effects on the signal due to the low value of current being passed. Inductance is the ability of a conductor to store energy in a magnetic field, with the property that in doing so it creates an opposing voltage proportional to the rate of change of current in a circuit. In effect, the current flowing down a signal conductor creates a tiny electromagnetic field but in signal applications the value is so low as to be negligible in terms of inductive reactance (not so in speaker cables).

So with tiny phono or line-level currents at audio frequencies, inductance isn’t that important, nor is resistance (or impedance) with the exception of shield or ground resistance, since the output impedance of a source component is much lower than the input impedance of the receiving amplifier (therefore the cable resistance is not going to affect the signal unless it is inconceivably high).

This just leaves Capacitance. Many of you may have read that capacitance at audio frequencies is not important, but it is, very important, especially where phono signals are concerned. High cable capacitance combined with the source impedance can form a low pass filter. The higher the source impedance, the greater the impacts of capacitance and the more the HF will be rolled off. A good example is with phono signals. Consider a high inductance source such as a MM cartridge. As frequency increases, so does source impedance due to the high inductive nature of the MM cartridge design. Therefore as frequency rises and output impedance rises, combining that with a high capacitive load (either high capacitance cable or excessive cable lengths) will make HF response suffer.

The other deleterious effect of capacitance in low level audio signal cables is that the storage and release of energy caused has an effect on the audio signal, whether subjective “detractors” believe this or not, it is an audio fact, NOT an audio myth.  An audio signal, unlike a single form RF signal, is made up of multiple complex waveforms varying in amplitude and frequency.  These include harmonic waveforms which make up things like specific tones and timbres.  They are related by phase (time difference between signals) amongst other things and capacitance, even low level capacitance, affects these phase relationships.  However subtle those impacts are, they nevertheless change in some fundamental way the nature of the signal such that the waveform loses part of the true timbrel accuracy of the original signal, so in general, the lower the cable capacitance the better, hence the shorter the cable the better.  How audible subtle shifts in phase accuracy are depend on may factors not least the rest of the audio chain, but seeking perfection or at least good practice is a worthwhile goal if one wishes to do everything possible to preserve the audio signal.

Many MM cartridges like to see a load generally of lower than 400 or 500pF. Taking the phono stage input into consideration (typically 150 to 250 pF) you can see the effects of adding capacitance from a cable of say 3m long with 100pF/m capacitance. It may have a marked effect on signal. For this reason, single core coax is best suited to longer runs for phono leads since it’s capacitance is lower. This effect is not as pronounced for MC cartridges as their outputs, hence inductance values tend to be lower.

2. Phase Shift and Skin Effect

A factor in signal propagation is how that propagation (and the propagation velocity) is affected by high capacitance. Use of dielectric insulators with high dielectric constants (PVC being one such example) can lead to unacceptably high capacitance in cables which can affect signal propagation at extremely low levels of audibility. Sometimes this is referred to as signal smearing but the term is sometimes mis-used since “smearing” is also related to phase shift and it’s effects on harmonic series signal wave forms.

So what is “phase shift?” It simply describes the displacement of two signals with time (relative to one another). To better understand the effects of phase shift, it is useful to first understand how music (and more to the point) musical notes are made up. In audio applications, fundamental frequency and “tone” are often and mistakenly considered to be the same thing and music mistakenly considered to be formed when a series of single waveform tones (or fundamental notes) all come together at varying amplitudes to create that music . It this were true, then there would be NO audible difference between a note of say “C” played on a trumpet and the same “C” note played on a violin, so why do they sound different? Well, because each instrument produces a frequency waveform made up of different parts, and the fundamental note is just one part of the “whole” note.

What are these “parts”? Well a musical note from an instrument is where the whole frequency waveform vibrates at a fundamental frequency but is actually made up of the fundamentals and overtones, collectively called the “partials”. Overtones vibrate above the level of the fundamentals. Together, along with other waveform sinusoidal components they are called the “harmonic series”. The harmonic series for the waveforms that every instrument produce act together to form the characteristic sound of that instrument and each has a different “tone” by virtue of the harmonics in play, and the nature of the rise and decay (signal amplitude differences) of those waveforms relative to time. A pure Sine wave of a specific frequency has no harmonic content but only one simple fundamental frequency forming a fundamental note or tone.

When a signal is fed into one end of an interconnect, the ideal interconnect should be able to pass that signal whilst maintaining the harmonic series waveforms and their associated phase differences and amplitudes. In reality, other factors come into play whereby distortion or loss of part of the harmonic series happens from recording through to replay via  source components, amplification and output from the loudspeakers. Room reflections also impact heavily on phase distortions so does it really matter if a signal cable doesn’t preserve what manages to make it onto the recording and on to to loudspeakers? Being perfectionist about conservation of signal integrity means, yes it should matter. However, whether any signal losses as “phase shift differences” as already mentioned, are audible depends on so many variables.  The losses via signal cables are likely to be minimal in terms of the overall audio chain from recording through to the loudspeakers, so it’s not really worth losing sleep over. You get more phase shift differences from in-room reflections than you do with most cables for example.  However, good practice says if it’s worth doing, it’s worth doing right.

This leads onto skin effect. So much has been written about the impacts of skin effects at audio frequencies often very poorly with very little explanation of how it affects sound at audio frequencies. Skin effect has little real impact on fundamental notes at audio frequencies. However, it does have an affect on harmonics, and therefore (it is sometimes argued) the degree of “realism” on the timbre of some musical notes.

Skin effect causes phase lag, and phase lag affects harmonics, so the components of harmonic series waveforms that extend well beyond audio frequency thresholds can be affected, and in turn (it is argued) can affect how that sound is perceived. How does this relate to interconnect design? Well minimising phase shift is simply a matter of designing signal conductors to avoid excessive phase lag caused by skin effect. Take a typical signal conductor of say up to 1mm diameter. The value for phase shift caused by skin effect equals one Radian per skin depth. At 20KHz, skin depth is around 0.51mm, so up to 20KHz phase lag is never greater than one Radian. At 10Khz, skin depth is greater at about 0.7mm and by 60Hz, skin depth is much greater at about 8.5mm. At all audio frequencies for conductors of 1mm or less, the whole conductor is therefore used and phase differences are limited to one Radian or less (phase shift drops with frequency).  This may not apply to the overtones in the harmonic series though so some phase lag of part of the signal may still have an impact (technically speaking) in the integrity of the the original signal.

There are other advantages however in making signal conductors smaller than 1mm in most instances by lowering capacitance between signal conductor and shield for any given diameter of cable.

The practical consideration of conductor size is more one of flexibility and strength. Too small and it could break easily, the smallest recommended size for mechanical integrity being about 0.4mm diameter which is about 26AWG. Signal conductors are commonly constructed of multi-strands of copper for good reason. A solid core of say 0.4mm is very prone to breakage and doesn’t withstand being flexed or put under repeated strain as well as a multi-stranded cable, so by making the cable of many smaller strands twisted together, the conductor is made more durable.

The very best theoretical signal core would be made up of lots of very fine individually shielded cores, commonly referred to as a Litz construction. These get over the phase lag (skin effect ) issues with harmonics and can be made into very small bundles (helping to keep capacitance low) whilst maintaining reasonable joint strength. They are, by their construction, more expensive to produce and therefore more expensive to purchase.  They are far from easy to terminate for the home cable builder and are best left to the professional to terminate properly.

 

Cable choice – Summary

Summing up, a well designed interconnect cable might then be said to have the following very basic characteristics:

  • low capacitance
  • Multi strand signal conductor of 0.4-1.0mm (Litz is best but multistrand copper for all intents and purposes is fine)
  • effective and low impedance shielding

There are lots of manufacturers who claim other serious audible affects in low current signal cables but for the most part much of this is misinformation where no evidence is forthcoming to back up those claims and that includes skin effect and phase shift. There are however a small number of very reputable cable manufacturers who pride themselves in their attention to detail, and go the extra mile to seek perfection for total signal integrity and this may be reflected in their cable designs.

Whether you’ll ever notice the difference between such expensive cables (as large sums of money are usually associated with their design and specialist manufacture) and simpler designs I’ll leave for you to decide, suffice to say that those cables are generally very expensive and may be worth consideration if your system is ultimately revealing and your pockets very deep! Avoid pseudo scientific claims and anything subjective in terms of claimed design benefits. If there’s no evidence to back it up, it becomes a matter of opinion and marketing.

 

Construction Quality – unbalanced cables

This matters. If poorly shielded cable with poor dielectric insulation is used and the connectors in particular are of a poor quality, then noise will enter the signal, the signal will be degraded (noise/higher capacitance) and where the transfer of that signal matters most (ie at the connector’s junction with the receiving phono socket) then further losses will result. The standard of termination is important too. Where the conductors are joined to the plug, that joint must be secure and not increase unduly the impedance or the capacitance. The use of shielded non-magnetic plugs is always preferable to non shielded or magnetic plug types.

Magnetism in plug bodies can be caused by the layering techniques used in nickel plating and magnetic fields cause distortion in low current signals. Shielding is provided by the plug body being conductive (low impedance), as with the shield, so that noise is effectively shunted to ground. There is little point in having a very effective shield and then connecting it to a high impedance RCA plug!

The best type of solder to use for soldered joints is solder alloy composed with similar metals to the one’s being joined to with a low eutectic melting point (so that excess heat doesn’t damage insulators or alter the physical properties of conductors, plus the melting temperature is lower than the individual melting points of component metals employed so that the solder slurry sets as a true joint rather than a connection of individually setting metals which create lossy grain boundaries) and high conductivity.

Keeping the joints neat and free of cold joints is vital too, so its best not to attempt soldering your own connections unless you are confident in your soldering skills! Mechanical joints are better still, since a gas tight cold weld can be formed via crimping, but there are very few plugs on the market suitable for this type of joint, although some of the better factory made cables are terminated in this way.

 

Construction Quality – Balanced Cables

This matters for the same reasons as for unbalanced cables.   Whilst most if not all noise may be rejected by common mode noise rejection, it is still preferable to utilise non magnetic plugs with high build and contact quality. Worth considering the pin quality too, as repeated removal and insertion can lead to wear on the conductors and a loss of signal integrity. High quality plating, as with RCA phono plugs is important for longevity and reliability.

What Cable Do You Need?

The answer depends on your intended use.

1. Unbalanced Cables:

For Phono Applications it is recommended to use a quality single core coaxial cable to minimise capacitance (or a well designed twin core coax cable) and ensure sufficient shield coverage of 95% or greater of the cable together with low capacitance non-magnetic plugs for best results. Keeping the cables to a minimum length matters, so aim for 2m or less if possible. If this is not possible, you may need to find out what your MM phono preamp recommended capacitive loading is and then work out the cable capacitance (total) needs to be to keep things under around the ballpark of 500pF for the circuit (this is less important for MC cartridge loading but it’s always wise to check the specifications for individual cartridges).

For Line Level Applications either a single or twin core cable which satisfies similar criteria as for phono leads and (up to a point) you can get away with much longer cable runs without serious degradation of the signal although the shorter the better is the best mantra to practice.  If budget’s an issue, go for a high quality single core coax. This will yield decent results. If you think you have a problem with ground loops or are in a particularly “noisy” environment then a high quality shielded twin core may fit the bill and may sometimes be more effective at preventing RFI and EMI from being induced into the signal.

2. Balanced Cables

Pretty much as for unbalanced cables in respect of capacitance issues on phono leads, but otherwise you can make these as longer (generally speaking) without serious effects on the signal from external noise. 5m or more for line level applications. Consider using good quality connectors.

Summing up

When considering a cable purchase, is it better to just use the freebee cables that came with your system or to invest in decent quality cables?

Hopefully some of the information above demonstrates very clearly why a well designed audio interconnect cable of good quality is by far superior to the cheaply made and poorly designed freebee cables that may have come with your kit.

You have to decide that, but bearing in mind the above, a good quality cable will always have benefits over a low quality cable of similar design and many enthusiasts are prepared to pay a small premium for quality, which is a sensible thing as with many other purchase decisions in life. The final arbiter of the decision is yourself, so learn to trust what your ears tell you!!!

Finally: be wary of “snake oil” salesmen offering the next magic interconnect using “new technology” or magic fairy dust.  Quality costs up to a certain price point and there’s more to the design of interconnects than meets the eye but good, sensibly design durable audio interconnect cables needn’t cost the earth nor do they involve rocket science.