Interconnects: why they affect the sound so much

[Al asked this question. I thus started this thread. Elk]

Al wrote:

I would like someone who understands better than I do to start a thread on interconnects and why they effect our sound so much and our devices with input and output imp as to best specs of our interconnects to be used.

Usb cables do effect things , but the worse offender is coax and AES. The change is dramatic and is either good or bad .

This is something that effects us all and for me I feel little is understood as to the why . The cable makers do not post real specs either .

The wire and the connectors usaelly do not match in imp. Are there other factors . Companies like Belden will post real specs but we all buy products for hundreds of dollars with little to no specs published . They usually have patent pending for something but no real specs.

any thoughts on this ??

So, I have learned a lot recently about cabling. My premise was always that audio, sorry, analog audio signals, are simple in respect to propagation down a wire. We sent them down Radio Shack RCAs for years, why bother with anything else? Then that “anything else” arrived in the form of high-end cables. BS! Just trying to take my money. Then I heard a set of better interconnects and got my ass handed to me. Turns out, I was wrong. I’ll bet we’ve all had the same or similar experience at some point. Back to the premise of “simple” analog audio signals: wrong again!

They are a hot mess of differing frequency and phase. Screw with any of them too much and you audibly degrade the signal. A “cable” is not a “wire” any more than an amp is not a straight wire with gain. A nice lofty goal, but many gremlins inside. Even a solitary wire can screw up audio when the skin effect comes into play (higher frequencies tend to travel faster down the exterior of the wire). Add in multiple wires and “insulation” (dielectric) and you now have inductance and capacitance to deal with. Maybe not a big deal if the cable is 2mm long, but 1m, 2m or more and you have a rather impressive, unintentional filter going. Oh yeah, I forgot DCR.

If you measure cables from end to end you get numbers for R,L and C. These tell you virtually nothing about how the cable will sound by themselves, assuming that their values are not crazy high. What DOES matter is that the electrical environment that each individual wire in the cable sees is identical AND that the overall values for R,L and C are as low as possible at the same time. The designs for RCA, balanced and speaker cable are very very different. A little bit is shared, but by and large that are different. The RCA uses a solitary 25ga copper wire and is shielded very differently than the balanced interconnect, which has a double pair of conductors in a unique arrangement. These cables are all copper, btw. It turned out that the design of the cables trumped the contribution of other materials. I have heard this myself and found that these same cables made with OF single crystal copper sounded marginally different (better?). Are they worth whatever the additional cost will be? That is for you to decide.

Yes, I’ve left lots of questions unanswered. Details that I have left out on purpose because I tend to screw things up when I explain them. I am a geek but not an engineer and those details are better left to them. In this case they will be from the engineer who invented the cables and when Belden says it’s OK, I will put them here first. I’ve never been given a clear explanation of just how a cable “works.” I knew this or that but there is a whole series of basic electrical principals to painstakingly get right before the cable “sounds good,” or as I have come to think of it, damages the signal as little as possible. I will post more on the sound of the cables that I have but I have promised Belden to write up my impressions first. Comments from others have put their new cables in the upper end of high-end and I am in agreement with this as far as the balanced interconnect goes. Pretty neat.

I found this article interesting and informative:

As good a place to start as any…?


scotte1 said I found this article interesting and informative:

More here (from the article series):

He’s on course there. There is a good deal more to the story beyond the basics. The commentary is typical of the hyper polarized camps with the usual absurd name calling to boot. Even Galen had engaged in some debate on A-Gon about cable design and the “experts” chimed in and told him that he was wrong. Pretty big claim for someone to call a cable engineer with 30yrs of experience a know-nothing. Another reason to like our forum!1_gif

When we start looking at interconnects in the usual unbalanced mode, we are dealing with a low impedance connection to a high impedance connection. If this were a balanced connection, (i.e. 3 pin configuration mic cable with 600 ohm transformers on both ends) we would deal with cables in a different manner. Having said that, since we are speaking to unbalanced connections, impedance match really doesn’t come into play, however, some simple rules about cables still do matter. One of those is shielding. A second item to consider is capacitance. Personally for my interconnects, I utilize high quality quad shielded RG6, with RCA crimp style terminations.

Impedance at audio is mixed with the very old and very new. We’ve all heard of 600-ohm cables, yes? These are impossible to make at RF frequencies without excessive cable size. At audio, the value is only true for a very narrow frequency range where the velocity of the cable is very low, raising the impedance.

Lets look at what IMPEDANCE calculations are, and quick see what the measured values tell us about audio cable.

At audio, the “impedance” can’t use the two reactive variables like RF, SQRT of L over C. We need to use an equation that is valid in the audio frequency range that takes FREQUENCY into account. Here is why frequency needs to be considered;

These are a few examples of cable velocity and frequency relationships. ALL cables will do this through the audio band. Look at the values at the far left of the trace. Do we see 90% velocity at low frequencies? No, we see values as low as 10% at 1 KHz or even less at 20 Hz!

To get an idea of the impedance of a low frequency cable, we need to use the equation

101670 / (Cap X Vp) = Impedance

Capacitance is stable through the measurement range as the chart below illustrates. The traces are for 1 KHz and 10 KHz values. ALL cables will exhibit stable capacitance with frequency, this is just a single example of a speaker cable.

                     1 KHz 	10 KHz 

AVG (pF) 14.893 14.441
STD DEV 00.166 00.202

Capacitance is stable so the attribute that influences cable impedance is the Velocity of the signal in the cables. The actual measured CHART for ICONOCLAST RCA and XLR show two things, first, each cable is IDENTICAL to the other cable reactively, and second the IMPEDANCE follows expected deviation across the frequency range.

What does the impedance do at the very low end? Yes, it goes UP as the Vp drops to 10% or less at 100 Hz. We only see a FLATTENING of the trace in the RF region above 100KHz.

The concept of a 600-ohm cable was derived through much study on LONG length telephone audio signals, and to excellent effect. Notice that the 600-ohm value in the example above is at just above 1 KHz, dead nuts in the frequency human hearing is centered around.

The “impedance” is NOT consistent at all, so the reference to 600-ohms is where that “spectral density” of the signal is greatest, and needs to be managed the closest to ideal with the same resistive load at the end of the cable to minimize reflections. But this can only happen for a cable that is a “transmission line” and for audio cables, about the only cable that meets that is a analog telephone cable as it is LONG enough to actually contain several wavelength is a cable length. Generally you need at least 10 wavelength in a cable length to act like a transmission line. From a PURE physics standpoint, ONE would work with ultimate physical perfection.

We have a problem, though, as the WAVELENGTH is different at every frequency from our frequency and Vp graph. What to do? Well, use the worst case. Lets us the 1 KHz value as an example that has a 20% value of the speed of light in a vacuum.

Wavelength = Wave speed / Frequency = 0.2(983,571,056 feet/second)/ 1000 = 196,714 feet

What do we see? An answer that is at a BARE MIMUMUM under the best of circumstances 196,714 feet long! This, for ONE wavelength at 1,000 Hz.

We don’t use cables this long in our systems. Our interconnects are terminated into INFINITY as an ideal number (actually 47K-ohms or the where abouts). This looks like a VERY large RESISTOR across the ends of the RCA or XLR cable.

A high impedance interconnect (sees a high impedance load, not 600-ohms!) has a very small current draw, but there is current as the cable looks like a capacitor that is being charged at the send end, and the signal amplitude observed at the receive end.

Capacitors don’t charge instantly, even though the current jumps to “infinity” at the start of a charge cycle and drops to ZERO after the capacitor (our cable) is fully charged. Circuits have capacitive reactance that defines how fast the capacitr can charge. Every circuit is different and the value of capacitive reactance is frequency dependent.

Capacitive Reactance, Xc = 1 / (2 F C)

- pie
F = frequency
C = capacitance

The use of high input ”impedance” (really resistance as we don’t have true impedance at audio) alleviates the cable from transmission line properties and falls back on a more lump sum attribute of capacitance and inductance. Inductance counts, too, as this value determines the instantaneous peak current applied to the cable at the transmit end that charges the capacitor…it DOES take CURRENT to CHANGE the voltage signal.

XL = 2 * *f * L

Notice this is the reciprocal of capacitive reactance. It takes TIME to change the voltage across a capacitor (what the cable looks like) and applied current is what changes the voltage level. The inductive reactance determines the initial maximum value of that current.

SUMMARY – All audio interconnect cables are high impedance (really resistive) terminated leads, and their capacitive and inductive reactance, that is based on their DESIGN, determine the voltage waveform distortions at the receiving end. With an ideal infinitely high load, the cables “impedance” is mitigated since no matter how high the cable impedance is, it is still “zero” relative to the infinitely high load value.

The current loop value is THROUGH the capacitance of the cable, and that value goes to ZERO once the cable is charged. Technically the LOAD can’t be seen by the cable as it isn’t there at infinity (not fully true, of course).

Each frequency is a different time based reactive set of values and alters the signal from an ideal instantaneous voltage value. When the voltage is zero the current is infinity, and when the voltage is at the maximum current is at zero. It takes TIME to move the voltage and current top opposite values. Add to this the SPEED that a voltage signal can travel down the wire (design’s dielectric value) which is also time dependent, and we can see wires aren’t perfect.

Using high impedance leads is a very good way to mitigate cable non linearities over speaker cables that see a low impedance cable terminated into a low impedance reactive load. We should be fortunate we don’t have 600-ohm transmission-line audio interconnect cables.