Digital Cables compared – AES/EBU, Coax,Toslink and ST Glass
Do digital cables really make a difference in sound? And if so, what are the differences?
Skeptics and die-hard technicians claim that differences in sound are impossible with digital cables, as long as the receiving device re-clocks the signal. Well, I have experimented with many different DA converters of all sorts, entry-level consumer, pro audio, and super high-end, that have reclocking, resampling, extra RAM or a combination of techniques and I haven’t come across a single DAC that does not sound different depending on the source or cable.
I haven’t based this on only a few digital components but have tested many: Mark Levinson no.36, no.360S and 390S, Wadia 12, 25 ,27ix GNS, 861, dCS Delius/Purcell, Krell KPS20i, Weiss DAC2, Ayre QB9, Arcam rDAC, DAD AX24, Meridian, Audiomeca, Audio Aero and many more. All these DACs and CD players showed comparable differences between digital cables used. I have also tested many digital cables, too many even to mention here.
Firstly, I have to mention that normal people (not obsessive audiophiles:-) would probably categorize the differences between cables I am talking about as nuances. I do feel however that these nuances can mean a lot in a well-balanced high-end setup. Even if you’re not looking to squeeze the last drop of quality from your system it may still be interesting to note that differences can be heard even though simplified conventional wisdom may dictate that it should not be possible. Finally, I should note that not all components are sensitive to cable influences in the same measure. Some are so impervious to make the differences pretty much meaningless.
DIGITAL SIGNAL TRANSMISSION
My personal concern is not so much the technical background but rather the audible differences themselves. Nevertheless, below I have copied a section from the CH Precision website to help illustrate why cables can indeed influence the quality of digital transmission.
“Digital signal’s quality is too often overlooked by simplistic statements such as “It is digital therefore it is perfect”. This is true as long as we stay in the digital domain. But there are two main cases where it is no more true: when a digital-to-analog (D/A) or an analog-to-digital (A/D) conversion takes place and when several digital devices need to be synchronized. In these cases, extreme care must be taken to digital (especially clock) signal’s quality, more precisely to their perfect timing (edge shape and time-placement accuracy). There are several ways to synchronize a digital source and a DAC. The most common ones include analog Phase-Locked Loop (PLL), asynchronous Sample Rate Converters (SRC) and First-In-First-Out (FIFO) buffers. Each of these techniques has its own advantages, but also its own drawbacks.”
WHAT ARE THE AUDIBLE DIFFERENCES?
The most prominent differences are known as PRAT. PRAT stands for pace, rhythm, and timing. In layman’s terms, this means temporal changes rather than tonal. I have also noted tonal differences between digital cables but for these differences, you need to audition two very different cables. For example an Apogee Wyde Eye and Belden RG59. In that case, you’re comparing solid core in a tight sleeve to litze in a very flexible sleeve. If you compare cables that are more alike, as is usually the case, the cables more or less resemble each other in tonality. And then it is easy to say that the cables sound alike. But listen more closely and you’ll find that one cable can sound slower and more rounded than another that is sounding faster and more dynamic.
The more you listen for differences in cables the better you learn to listen beyond the generic “sound” of a cable. There will appear to be more subtle changes such as fluidity in the treble, fullness of bass or lack thereof, overall warmth, decay of subtle sounds and tightness of attack in transients. The more careful you listen, the more you’ll hear.
three of these cables are of the RG-59 type, and so should adhere to strict specs, yet they all sound different. The fourth cable (with the tinned shielding) is a special double-shielded variant for higher-spec video use over long distances or in signal unfriendly environments.
This may sound incredible, and I know that many technicians would declare me crazy but I’ve listened to every RG59 coaxial cable I could lay hands on and I can tell you that there are large differences between the various brands. Even going from one Belden type to the next, there are big differences. The differences in sound are not usually timbral in nature and hardly concern detailing but can best be described as changes in transient sharpness which is related to the perceived speed and the sense of warmth. One cable can be very fast and dynamic sounding while another can sound slow and overly smooth.
WHAT ABOUT AES/EBU?
AES/EBU is a professional standard and although it can run over a coax cable, and the AES flag can be either on or off, the native format of the signal is balanced and thus it is meant to run over a balanced. The format is intended not to sound better (as many audiophiles assume) but to be more solid over great lengths and under harsh conditions. For instance, in recording studios and post-production facilities where there can be enormous lengths of cable between any two components and 100 meters is not uncommon. This is where the balanced signal comes into play. Just like with balanced analog signals, in the digital form, there is a plus and minus side which are in opposite phase to each other. Any disturbance picked up by the cable is likely to be present in both positive and negative conductors. A phase comparator in the component at the receiving end subtracts both signals and consequently gets rid of the disturbance. This is possible because the disturbance will be in phase on both conductors while the music signals are in counter phase.
For this to work, a source needs to present the digital data to the output in balanced form. Some components use true differential circuitry but most convert the single-ended signal to a balanced form by means of active circuitry or transformers. One could imagine that any extra processing going on can have a negative effect on the sound. But whether this is the real culprit or it is another factor, there’s something to AES/EBU that makes it sound different from SPDIF. And it’s not only the cable, either. I’ve tested many, and have also used AES inputs and outputs with coax cable in balanced and unbalanced form. In all cases, the AES input sounded different from the SPDIF coax input.
WHAT ARE THE DIFFERENCES IN SOUND BETWEEN S/PDIF AND AES/EBU?
Under most circumstances where I tested this, I find AES/EBU to sound overly controlled or restrained and typically a little too clean. And perhaps because of its very tidy sound, AES/EBU often fails to excite me. The bass is clean and articulated and technically probably precise but via SPDIF, the bass is more solid, bigger and more physical. Of course, opinions will always vary and you may well actually prefer AES/EBU over coaxial. That’s all fine. This article is merely meant to raise awareness of the fact that there are differences and AES/EBU need not automatically achieve the best balance.
The good thing is that the differences in cables are a little less clear with AES/EBU than they are with SPDIF and even very standard cables can do very well.
Update 2017: with the Aurender N10 connected to a Wadia 521 and many other DACs I now find that coax and AES/EBU are much closer to each other in sound than I have heard before.
ST GLASS VERSUS TOSLINK
This is (in the audiophile world) an old standard by now. It comes from the telecom world and is still used there extensively. Unfortunately, not many manufacturers of audio equipment today use this standard. An exception is M2Tech with its HiFace EVO USB interface. This little unit actually allowed me to compare a 25 meter long Belden RG59 coaxial cable to the same length ST glass cable directly into the Levinson no.360S DAC. ST Glass transmission is optical but don’t confuse it with Toslink. First, there is a laser, not a LED in the transmitter, second, the conductor itself is made from very thin glass, not plastic. Lastly, the wavelength is very much higher and is wholly appropriate, even oversized, for consumer digital audio whereas Toslink is actually underspecified. Toslink connections tend to soften transients and lessen dynamic behavior compared to an electrical connection. This isn’t the case with ST glass. Instead, the musical portrayal is very much as dynamic and speedy as with RG59 but devoid of the added sharpness and “zing” of the latter cable. At the same time, ST Glass is more detailed and refined than RG59. Do note that in certain system setups, RG59 might sound subjectively better just because of its more aggressive character. ST Glass is so much so devoid of artificial sharpness that it may appear to “rock” less. In my setup (Jeff Rowland and Magnepan speakers) I prefer ST glass, especially since I need to use a very long 25-meter cable. Whether or not ST glass is better for you too, depends on your setup and your taste. Toslink, however, is always the lesser choice, unless you have a setup that is perhaps somewhat edgy and overaggressive, in which case Toslink might soften things up a bit. But this will be at the expense of some dynamics and speed. The other good thing about ST glass is that it is designed for long distances. And by that, I mean hundreds of meters. I have yet to do the actual comparison (and will do of course) but in theory, a length of 25 meters sounds the same a length of 1 meter. From what I gather in my 25-meter cable comparison, I have no reason to doubt this. The ST Glass cable I tested was a standard no-name cable that cost less than 40 us dollars for 25 meters. That’s even cheaper than Belden RG59!
BNC to cinch adapters are cheap and handy and seem to do their job just fine. However, when you start comparing, it becomes obvious that these adapters do add some edge to the sound and depending on your system’s balance and your own sensitivity to this, are best avoided. For example, when using equipment having both BNC and cinch inputs and outputs, Belden RG59 with crimped on BNC connectors has a nicely neutral, yet dynamic and lively character with excellent focus, drive, and attack. When comparing to the same cable with cinch connectors and BNC adapters, you notice that the resultant sound is more “impressive”, with more presence in the treble and seemingly more attack. But you quickly realize that this also makes the cable sound a bit aggressive and unnatural. In overly smooth setups, however, this can give some nice extra “bite” to the sound.
BNC VERSUS COAX
Some components offer both Cinch and BNC connections for SPDIF. In the cases where I have compared these outputs, usually, BNC sounds cleaner with sharper focus. This makes sense because it is technically also supposed to be the best connection. However, depending on circumstances you may also prefer the cinch outputs because these tend to sound beefier in the bass and arguably a little more colorful.
Update September 2020
Having reviewed the Jorma AES/EBU cable has turned some of my long-standing views on the subject upside down. It’s not just the cable, though, but also my preferences that have changed over time, along with system changes. Overall, my current view is that any connection can sound good or not so good, depending on the quality of the cables and the implementation of the interfaces. Please read the review for more info.