Hum Rejection in Unbalanced Audio Cables
A test of cable designs and shield types
Audio cables have a simple mission when it comes to noise from electromagnetic interference sources: keep it out! In this article, we address just how a good audio cable carries out that mission, and compare a few cable configurations and shield types in an effort to see just what is the best strategy for excluding EMI from the audio signal path.
Shielding, Shielding and More Shielding
Most consumer audio gear runs audio signals in an "unbalanced" form, with the signal carried on one wire, accompanied in the cable by another wire or shield which connects the signal grounds of the two pieces of equipment together. If the jacks on the back of your equipment are for "RCA" type connections, your audio circuits are this most common, unbalanced type.
As we've discussed elsewhere in our articles on cable design, in these unbalanced circuits, there's really only one way to keep noise out of the signal path: shielding. Loose electromagnetic energy that hits the cable needs to be intercepted by a shield and shunted to ground; if it gets past the shield, and reaches the signal conductor, there's no way to separate it from the original, intended signal. For this reason, coaxial cable, with its all-encompassing shield, is the best cable design for unbalanced audio circuits.
There is a common misconception that twisting the two conductors of an interconnect cable together can somehow contribute to noise rejection, and even that it can perform as good a job of noise rejection as a good shield can. But the phenomenon by which twisted-pair wires reject noise--called common-mode noise rejection--works only in balanced audio circuits, not in the unbalanced circuits that are nearly universal in consumer audio gear.
So, shielding it is, and the better the shield, the better the noise rejection. But what kind of shield? If you browse through the specs of a large number of coaxial cables, you'll find a number of different shield types. Some cables have a spiral shield wound around the dielectric; some have a braided shield woven from tiny wires; some have nothing but foil; some have a braid and foil; and some have multiple braid and/or foil layers.
One of the reasons for this variety of shield types is that different types of shields are effective at removing different types of noise. For example, noise can come from a power cord laid alongside a cable for a distance; the electromagnetic field set up around the power cord will induce a small current flow in the neighboring cable. This sort of low-frequency noise is usually referred to as EMI--"Electro-Magnetic Interference." Another type of noise source, much higher in frequency, is radio waves--these might come from broadcast stations, cell phone relay antennas, amateur radio stations--anywhere radio signals come from. This higher-frequency noise is usually referred to as RFI--"Radio Frequency Interference."
Depending upon what sort of signal a cable is carrying, we may be worried about EMI, RFI, or both. For video signals, both EMI and RFI are problems; low-frequency noise can distort a picture--for example, inserting "hum bars" into it--while high-frequency noise will be mistaken for part of the video signal itself and may appear, for example, as "snow" in the image. For analog audio, which operates at much lower frequencies, EMI is the principal problem. RFI can enter your system through audio cables, but unless it is particularly strong, ordinarily does not present a noise control issue.
The two principal shield types, braid and foil, operate differently with respect to EMI and RFI for a few reasons, principal among which are (1) coverage and (2) conductivity. Braid shields, because they consist of wires woven together, always have holes in them, and are limited to about 95% coverage. These holes are not particularly significant at low frequencies, because the wavelength of low-frequency interference is so long--an easy, if somewhat inexact, way to think about this is that these long wavelengths are just too big to fit through the holes in the braid. At higher frequencies, however, these holes start to become significant, and the limited coverage of the braid reduces its shielding effectiveness. Foil shields, by contrast, provide very high coverage--if they're well-made, 100% coverage--so that there just aren't any holes for the noise to get through.
Simple, then--use foil shielding for everything, and you've got 100% shield coverage, ergo 100% shield effectiveness--right? Wrong. Another principal difference between foil and braid shielding is that foil is not nearly as conductive as a dense copper braid. When EMI or RFI hit the shield, the shield doesn't just absorb or reflect all of the noise--it needs to provide a path to ground for the noise to travel down, and it needs to steer as much of the noise down that path as possible; otherwise, a significant portion of the energy will bypass the shield and still work its mischief. Think of the shield as a lightning rod; a lightning rod made out of thin aluminum foil might attract lightning, but it wouldn't provide significant protection because of its limited ability to dissipate energy.
The competing demands of coverage and conductivity lead to the shielding configuration we see on the best video cables--foil and braid, working in tandem. The foil shield is effective over a wide range of frequencies, but is too thin to provide an optimum path to ground; the braid shield has reduced effectiveness at higher frequencies, but is highly conductive and mechanically solid. The combination intercepts noise over a wide range of frequencies, shunts it effectively to ground, and provides a cable which will continue to perform these jobs well even after it's been heavily handled.
We have always found foil/braid shielding to provide excellent noise rejection results in audio cable applications as well--but we've also wondered just how much the foil was contributing to the equation. EMI often is very high-energy noise, and the importance of a conductive path to ground is therefore usually greater with EMI than with RFI. Meanwhile, the 100% coverage offered by the foil shield is of less importance, because the frequencies with which we're principally concerned are so low.
Some coaxial cables, particularly those designed for applications requiring high flexibility, use not one, but two high-coverage braid shields. Because these shields simply have a lot more copper in them than there is in a single-braid shield, they're very highly conductive. We wondered whether these double-braid shielded cables might outperform braid-and-foil cables in rejecting high-energy, low frequency noise, and devised a test to try it out.
We chose six coaxial cables to test; of these, two had double-braid shields (Canare LV-77S and Belden 1505F) (note: our LC-1 audio cable hadn't yet been built, so wasn't included in this test--but see below), two had braid-and-foil shields (Belden 1694A and Canare L-5CFB), and two had single-braid shields (Belden 89259 and Canare LV-61S). In addition, we decided to test out a balanced audio cable, wired in an unbalanced configuration to RCA plugs, because some people feel that this type of construction is better than coaxial cable for noise rejection. For this last cable we chose Belden 1800F, an excellent balanced audio cable with a "french braid" shield--two spiral shields, wound in opposite directions and interwoven. We wired it in what seems to be the most common configuration used in audiophile circles--one wire of the twisted pair to the center of the RCA plug, one to the outer ring, and the shield attached to the outer ring but only at the "source" end of the cable.
We put together twenty-foot interconnects in each of these seven cables. Now, the task was to expose them all to a high-energy, low-frequency source of interference. To do this, we strapped them all tightly to a heavy-duty extension cord. We plugged in the extension cord, and plugged a space heater into its outlet to get a significant current flow running through the cord; this was important because the strength of the magnetic field around the cord, and hence of the EMI exposure of the cables, is directly related to the amount of current flowing through it.
We listened to the induced 60-cycle hum in each of these cables through an amplifier, and did A/B comparisons to determine which cables were loudest (that is, admitted the most hum). At all times the listener was blind, not knowing which two cables were being compared.
After a series of comparisons, we were able to generate a best-to-worst hum rejection ranking for the seven cables. These results are reflected in the table below, together with some related cable characteristics, listed in order from best to worst:
|Cable||Shield Type||Shield Resistance|
|Canare LV-77S||Double Braid||1.8 ohms/1000 ft.|
|Belden 1505F||Double Braid||2.4 ohms/1000 ft.|
|Canare L-5CFB||Braid/Foil||2.1 ohms/1000 ft.|
|Belden 1694A||Braid/Foil||2.8 ohms/1000 ft.|
|Belden 89259||Single Braid||2.6 ohms/1000 ft.|
|Canare LV-61S||Single Braid||4.0 ohms/1000 ft.|
|Belden 1800F||French Braid||5.0 ohms/1000 ft.|
One note: although the differences between cables several entries apart in this list were quite dramatic, some of the comparisons between adjacent cables on the list were quite close.
The two points that jumped out at us when we examined these data were (1) the consistency of the shield type hierarchy--double braids outperforming the braid/foil cables, braid/foil outperforming the single braid, and the single braid outperforming the "quasi-balanced" unbalanced cable, and (2) the very close correlation between the shield resistance (inverse of conductivity) and the cable's performance.
The 1800F performed miserably, which was no particular surprise. First, it isn't a coaxial cable and therefore isn't well-suited for use in an unbalanced circuit. Second, it was wired in a manner which, though popular with some users, makes no apparent electrical sense. None of this should be held against 1800F, which, when attached properly to balanced connectors and used in a balanced circuit, is a truly superb cable.
Our conclusion is that for rejection of low-frequency, high-energy noise, just as electrical theory would suggest, a coaxial cable with a highly-conductive double braid shield will outperform similar braid/foil and single braid cables. When we recently had our own unbalanced audio cable, BJC LC-1, custom-built by Belden, we went with the same shield configuration found on the best performer of these: Canare LV-77S, but with a softer texture for better flexibility and lower capacitance for less high-frequency rolloff. This new cable replaces our prior recommendation of LV-77S for subwoofer use and Belden 1505F for general audio use.
Does this mean, then, that a double-braid cable will always be the best choice from a noise rejection standpoint? Not necessarily. First, for video and RF applications, where RFI is an important consideration, the 100% coverage offered by a good foil/braid combination (e.g., Belden 1694A) will outperform a braid-only shield. Second, it's possible, in an RF-noisy environment (for example, a home close to radio and TV broadcast antennas), for RF to play a role in audio quality as well. Strong RF signals can enter audio circuits and be "rectified," stripping their high-frequency content and causing audible interference. In such a situation, a braid/foil shield may well provide the more effective shield; but for most users, in most environments, EMI is a far more likely cause of audio mischief than RFI, and the highly conductive double braid will reduce this low-frequency noise more effectively.