T.H.E Show 2014: Day 3 Morning

Stephen Mejias, formerly of Stereophile, now VP of Communications at AudioQuest, performed a most convincing cable comparison using the company's top-of-the-line AudioQuest Diamond Ethernet cable ($695 for 0.75m, $1195 for 1.5m). In a system whose JRiver-equipped MacBook Pro fed CD-quality files to an Audio Research Reference DAC, Audio Research Reference 75 amplifier, and Vandersteen Treos outfitted with a double bi-wire pair of AQ WEL Signature cabling, Roy Orbison's "Crying" sounded one-dimensionally flat and a little bright on top via a generic Ethernet cable. When Stephen switched to the AQ Diamond Ethernet cable, voice and instruments suddenly and dramatically acquired depth, air, dimensionality, and subtlety.

"The sound is more rounded and full," I scribbled gleefully. "Sounds like a different track altogether." At demo's end, Stephen noted that AudioQuest's Ethernet line begins with Pearl ($29/1.5m).

In my second listen to the pairing of Vandersteen Model 7 loudspeakers ($52,000/pair including speaker wire) and Vandersteen M7-HPA monoblock amplifiers (also $52,000/pair), I noted the lovely midrange and wonderful depiction of cymbals and trumpet and Lee Morgan's Tom Cat.

"Thoroughly enjoyable, flows really nicely," I wrote of a listening session overseen by Richard Vandersteen (pictured) and DJ'd by Shane Buettner. Even on CD, the system nailed the air and depth of the orchestra. While the sound was very non-fatiguing, the loudspeakers' powered bass didn't fully define the bottom line of double basses in my recording of Mahler's Symphony 2. I've also heard richer sounding cellos. But overall, the presentation was quite excellent."

The liquid-cooled M7-HPAs are high-pass amps designed specifically to handle 100Hz and up on the Vandersteen 7s and Quattro 5A Carbons. Sharing the honors were Audio Research's Reference 10 preamplifier ($30,000), Reference Phono 10 phono stage ($30,000), and Reference CD9 transport/DAC ($13,000); a Basis Inspiration table with Lyra Atlas cartridge, Harmonic Resolution Systems stands and bases, and AudioQuest cabling.

As I soon discovered in the Vapor Audio room, don't try to evaluate music sourced from the Antipodes DV music server ($3990) while it's in the midst of ripping files, because sound is compromised. Once the task was completed, a 24/192 reduction of Ivan Fischer's DSD recording of Mahler Symphony 4 exhibited fabulous depth and air, and Shelby Lynne's voice was again worth lovin'.

A listen to Reference Recordings' Rachmaninoff Symphonic Dances also showcased the excellent, full-range sound and engaging depth produced by Vapor Audio Joule White loudspeakers ($12,995/pair). Also heard: Clayton Audio M300 monoblocks ($16,500/pair), DAC conversion via Empirical Audio's Overdrive ($6399) and Short Block USB filter ($199), a Final Drive Transformer Buffer ($2999), VH Audio Plasmatron power conditioning ($4000), cabling from Antipodes and Verastarr, and Hi-Fi Racks racks.

The pairing of YG Acoustics Hailey loudspeakers ($42,800/pair) with Audionet's new MAX monoblocks ($30,500/pair), PRE G2 preamp ($23,350), and DNC streaming DAC ($10,100) was an absolute winner. Through a Kronos Limited Edition turntable and Black Beauty tonearm ($40,000 total) with unspecified cartridge, and Kubala•Sosna's Xpander power distribution box ($4800 and Elation! Cabling, the vinyl version of Count Basie's 88 Basin Street sounded exceptionally warm and full, with realistic piano timbres. Sound was also extended in all directions, and left me feeling as though I was hearing all Basie's recording engineer(s) had hoped music lovers would hear.

A 45 rpm pressing of de Falla's Three Cornered Hat, conducted by Ansermet, delivered fabulous depth and a warmer sound that I had previously heard from YG loudspeakers. This was definitely one of the finer systems at T.H.E. Show Newport Beach.

Yet another winner came via Dynamic Design AV Anniversary Edition 15 Series cabling, which together with Stillpoints Aperture Acoustic Room treatment ($600/2x2 panel), produced the most liquid sound I've ever heard from ModWright electronics. Johnny Hartmann's voice sounded simply fabulous, and playback of Reference Recordings' Fiesta LP was distinguished by outstanding depth.

The system also included ModWright Instruments' flagship electronics, new modified Sony HAP-Z1ES music server, and modified tubed Oppo BDP-105; Annalyric Systems AC-8 AC conditioner ($1800), and Marten Getz loudspeakers ($20,000/pair), and a Helius Alexia turntable/tonearm with Scantech Helikon cartridge. Note that the Stillpoints panels are claimed to act as an all-in-one resonator/absorber/diffusor, and are said to self-regulate depending upon volume.

In the Nordost room, the company demmed its new Heimdall 2 headphone cable ($799/2m with adapters) on both Sennheiser HD800 and Audeze LCD-3 headphones connected to an Auralic Gemini headphone amp. As eager as I was to hear the difference—I'll try listening at home on Audeze LCD-2 headphones—time necessitated that I leave the comparison to Tyll Hertsens of InnerFidelity.com, and instead investigate Nordost's new Blue Heaven ($214.99/1m) and Heimdall 2 ($459/1m) iKables. Available in various configurations, the iKables are designed for portable players, iPhones, and anything with a mini-output.

Compared to a generic cable transmitting a 24/192 conversion of a DSD file of the LA 4 through an Astell&Kern 120 player, Valhalla 2 cabling, and Nola KO loudspeakers equipped with Nordost Sort Füt loudspeaker footers, the Blue Heaven iKable delivered a lot more harmonics on guitar, and sounded rounder, fuller, and smoother. The Heimdall 2's sound was even cleaner, clearer, and more impactful, to the point where I felt I could finally hear the leading pluck of the guitar's strings before the full body of sound commanded my attention. A return to the generic cable confirmed how shallow its presentation was. A second, equally convincing comparison on a track by Leonard Cohen confirmed that the Heimdall 2 iKable delivered the warmest, smoothest, and fullest sound of the lot, with lots more delicious height, depth, and width to the presentation.

Jim Thompson of EgglestonWorks was on hand to show off his new Emma loudspeaker ($4000/pair) and Nico bookshelfs ($3000/pair). Both are entry-level loudspeakers that use proprietary drivers, and whose front ports are designed to work well in smaller rooms. The Emmas delivered very neutral and smooth sound via a Rogue Pharoah amplifier, Hegel DAC, and Hemingway cabling. Percussion on a track of dubious musical value from an Usher CD sampler was also quite impressive, albeit a mite toned down on top.

Albert von Schweikert has of late been towing his imposing VR 100 XS loudspeakers ($140,000/pair including two subwoofers) to shows. When paired with Constellation Audio's Centaur 500 monoblocks and what may have been a Virgo II preamp, a custom Esoteric CD player, and Master Built cabling, I noted how lovely, crystalline, and pure the highs of Myung Whun Chung's piano sounded. Bass, however, seemed a bit disconnected, perhaps because of my proximity to the speakers in such a small room, and the midrange lacked the last iota of liquidity. Regardless, subtle dynamic contrasts that are often glossed over by lesser speakers and electronics were delivered with notable excellence, making for an exceedingly musical presentation.

In the first of two rooms from Audio Skies of California, GamuT RS5 loudspeakers, electronics and cabling may have sounded a little gray in the midrange and bass in the Hilton, but their bass was exceedingly profound, and they came to life higher up. On an orchestral recording whose title I didn't note because the exhibitor was scribbling a component list in my notebook while I listened, the triangle sounded quite fine, and the spatial presentation was excellent. Doing the honors were GamuT's RS5 loudspeakers ($36,900/pair), M250i monoblocks ($12,300/pair), D3i preamp ($9990), and CD3 CD player ($8830), as well as Wormhole cabling. I'm not sure I heard the Kid Thomas turntable/Pear Audio Blue tonearm/Ortofon cartridge/Lehman phono stage combo.

COMMENTS
otaku's picture

"When Stephen switched to the AQ Diamond Ethernet cable..."

I don't see an Ethernet connection on the Reference DAC. Are you saying the sound changed when changing the Ethernet cable between the Mac and the LAN???

eugovector's picture

Yep, and it sounded "like a different track altogether" and "voice and instruments suddenly and dramatically acquired depth, air, dimensionality, and subtlety"...from an ethernet cable.

It's a good thing that science is hard and common sense is in short supply, otherwise some manufacturers would have a hard time emptying their snake oil warehouses.

And before anyone tells me to "trust my ears", I've heard this Demo more than once. Wireless to generic wires and up the product range to diamond, before going back down to wireless. Not only did the demonstrator use every cheap trick in the book (expectation bias, raising the volume level between cables, condescension of those who couldn't discern a difference, pseudo-science, irrelevant anecdotes, and on and on...), but when asked to do a simple blind (connect one Ethernet or wireless, don't tell the listeners which it is) they refused which was funny seeing as how moments before they described the differences just like stereophile has: night and day.

yonk's picture

NFW. The claim that a LAN connection with protocols from the Internet (TCP/IP) can improve sound quality is as outrageous as claiming that your HiFi can clone Humans.

TCP/IP was designed for and accomplishes TOTALLY RELIABLE ASYNCHRONOUS transfer of data on serial links using modems on crappy phone lines at 9.6 characters per sec.

The ENGINEERING and SCIENCE just does not support such a claim. I'd believe "aliens were helping" first.

Give me a TRAFFIC TRACE that shows this cable changes the delivery characteristics of the datagrams that the audio stream is broken down into.

You would need to have absolutely horrific datagram corruption and loss before an audio stream on Gigabit Ethernet is compromised.

QED

-Yonk IT and Network professional for 35 years

heavystarch's picture

A-FUCKING-MEN. Brilliant points.

jeffro's picture

The Reference DAC does indeed have an Ethernet port and is capable of accessing music files from NAS.

However, any reasonable person having a basic understanding of Ethernet and TCP/IP will have a problem with statements like "When Stephen switched to the AQ Diamond Ethernet cable, voice and instruments suddenly and dramatically acquired depth, air, dimensionality, and subtlety." Especially in a NAS environment, where the buffers on the client side make any "jitter" on the LAN irrelevant.

Personally I can't get past Audioquest's statement that Ethernet cables are directional and should be oriented in the direction of music flow.

remlab's picture

I hope you know what you got yourself into..

eugovector's picture

I'm sure it's similar to ex-congressional employees turning lobbyist. As long as you are willing to deny reason or sell your ethics, you can make a very nice living.

heavystarch's picture

Your comparison to a former government employee turned lobbyist is quite apropos. I wonder how much his previous employment will help in gaining more reviews for Audioquest cables...time will tell.

On a related note I too have been to one of the AudioQuest cable swap demos and never heard the difference.

Cable should definitely be made of quality materials, follow proper connection methods, sound geometry and engineering methods. However now that we have cables that cost as much as luxury cars we have created a market segment that makes our whole industry appear kooky to the world outside audiophiledom.

This is plain bad for our industry as it will be a big turn off for most folks who might otherwise step into the wonderful world of audiophiledom.

John Atkinson - will you provide measurements for cables like you do for other components? I see speakers and electronics get tested at your hand but I don't recall cables getting the same analytical treatment in stereophile...am I wrong? I better go home and look over my old S'philes...

eugovector's picture

This is the real issue, as you put it so well:

"However now that we have cables that cost as much as luxury cars we have created a market segment that makes our whole industry appear kooky to the world outside audiophiledom.

This is plain bad for our industry as it will be a big turn off for most folks who might otherwise step into the wonderful world of audiophiledom."

If there truly is a difference, prove it with a simple test, objective or subjective, and put the naysayers to rest. Otherwise you are killing the golden goose; fleecing the current crop of rich old fools while turning off the next generation. This might make for a good retirement plan, but you're killing the industry and love of good sound for the next generation who haven't dismissed science (yet).

v1m's picture

From writing unreadable columns to selling undetectable cables: it's such a fine line.

zimmer74's picture

Here come the "scientists"! Cueing Thomas Dolby, "She Blinded Me With..." The issue of ethernet cabling and sound quality has been extensively discussed on many audio forums; the most intelligent treatment, in my opinion, is on the Naim forum. Anyway, as with such topics as digital coaxial cables, component isolation, jitter, etc., the scientists usually lose, because they live in a constrained reality. "Bits is bits" is a joke in some circles, but not everybody gets the joke. The ultimate irony is that audio is a hobby, and the goal of the hobby is to enjoy music. Not everybody acknowledges that.

John Atkinson's picture
zimmer74 wrote:
with such topics as digital coaxial cables, component isolation, jitter, etc., the scientists usually lose, because they live in a constrained reality.

I wrote about a similar matter in the April issue, after experiencing a difference in sound quality with audio data sourced via a WiFi link and via a wired Ethernet connection:

There should be no audible difference between a wired and a wireless datalink. Both are packet-switched systems with received data fed into a buffer. But, to my horror, I did hear an improvement between a WiFi link and a hardwired Ethernet connection.

I have no explanation for this audible difference, unless, to paraphrase the mathematician Kurt Gödel, in any formal system that is sufficiently complex for everything to be explained, there will intrinsically be things that cannot be explained.

John Atkinson
Editor, Stereophile

abderrahim's picture

I can assure you that Gödel's incompleteness theorems are irrelevent in this case!
greetings from morocco

John Atkinson's picture
abderrahim wrote:
I can assure you that I can assure you that Gödel's incompleteness theorems are irrelevant in this case!are irrelevant in this case!

Though you haven't offered any support for your statement, I believe you are incorrect. All the skeptical comments made in this thread are made from looking at the network as an information system: that if the actual information is not altered, then there can be no difference in sound quality. You have to go outside of that myopic viewpoint, to examine a system where digital data and its transmission are actually represented by an analog electrical system.

In that Gödelian light, it is possible that there could be changes in the manner in which those digital data are used to reconstruct an original analog audio signal, even when the actual bits are handled without error. This is something that is irrelevant from the information viewpoint, in the transmission of digital data representing something other than audio.

John Atkinson
Editor, Stereophile

abderrahim's picture

OK, let's assume that this is "sufficiently complex" that there are some properties of the cable that influence the sound in an unexplainable (by nature) way, then the design of the cable didn't rely on science. (my guess: they used magic too!)

John Atkinson's picture
abderrahim wrote:

OK, let's assume that this is "sufficiently complex" that there are some properties of the cable that influence the sound in an unexplainable (by nature) way, then the design of the cable didn't rely on science.

Again, you are not thinking outside of the box. The system's electrical properties will change when you change the cable, or replace a hard-wired Ethernet connection with a WiFi link. In your ideal, IT-based view, it should not make a difference when the "bits" are fed to a D/A converter. But from an audio engineering viewpoint, it might do so. Without further investigation, you should remain agnostic, as do I.

abderrahim wrote:

my guess: they used magic too!

And why would you say that? As I said before, you have not provided any support for your arguments, other than to belittle others. You are arguing from faith, not actual knowledge.

John Atkinson
Editor, Stereophile

Larry Ho's picture

Some simple tests could show the latency is 20% higher, jitter is much more with Wi-Fi standard.
One example from PC Magazine, http://www.pcmag.com/article2/0,2817,2381754,00.asp

Under the basic 7-layer TCP/IP transmission model, the underlying layer could make sure the DATA will be sent and received correctly. But it is NEVER designed for real time transmission.

With a good design buffer, the difference between latency and transmission induced jitter will be smaller. But it definitely will be there... One extreme example will be, think about the case we are playing WiFi music on its 'one bar left' situation. I believe John's subjective assessment here.

What really puzzles me: Why the material of ethernet cable will make the difference? I tested expensive audiophile ethernet cable and data center grade Cat-6 cable, Cat-6 is better!

Larry

jeffro's picture

Curious, what system or component does this observation pertain to? I saw your comments in Letters to the Editor, but without any context.

I would love to see Stereophile cover this topic in more detail. Especially if measurements and blind testing were involved (realizing, of course, that what we hear cannot always be measured). At the very least, I think it would benefit Stereophile readers to better understand the difference between different types of digital interconnects.

Good cables do make a difference that we can hear, especially in the analog domain, and perhaps to a lesser extent, with baseband digital signals (i.e., clocked serial transmission). And while I have no doubt that AQ makes some very fine Ethernet cables, I can't help being skeptical that they impact sound quality in any meaningful way. Certainly not in a "night and day" kind of way. And while it's still early days for computer audio, I have yet to see a manufacturer offer any science- or engineering-based rationale for why their Ethernet cables improve sound quality.

eugovector's picture

Can you tell the difference between an Audioquest Pearl and Diamond ethernet cable?

eugovector's picture

Maybe Jason can chime in on this one? Surely, when the assertion is made that a cable can make a track sound completely different, hyperbole or not, something as simple as discerning between a $20 and $500 Ethernet cable wouldn't be hard to do. Did you mean what you wrote Jason? Could you tell the difference?

audioengr's picture

This is not backed-up by an empirical testing, but it is possible that Ethernet like USB is affected by common-mode noise, so that indirectly the playback device can experience higher jitter. This is why CM filters and galvanic isolation helps with USB. The cable still matters there if earth ground is connected between the two systems.

Another thing to consider is that particularly with Mac, the audio stack may be involved with WiFi and not with wired Ethernet. Different software, different sound.

Steve N.
Empirical Audio

Bongo Fury's picture

If the ethernet cable impacts the sound, what about the rest of the signal chain?
Does the ethernet from a cable modem to the computer similarly impact the sound?
The cabling the ISP uses?
Certainly the modem itself must?
And routers if used, surely that adds a veil?
Brand of hard drive, SSD vs platter, type of controller?

So many variables that must make a difference.

mrhyfy's picture

Cisco may want to know about these superior cables!

yonk2's picture

I don't disagree that the final test is always in the listening! However claims DO have some basis in science and engineering - especially digital which unlike analog CAN be quantified.

Cisco doesn't need to "know" about these cables (I know it was sarcasm) but ask me how I know - and known for 19 years.

The difference between analog and digital using Internet protocols is that the bits DO tell the story. You say you laugh about "bits is bits" but when you transfer a file from your NAS to your DAC or DAC via Computer->USB you are making a BIT PERFECT copy of the file. No Magic. No Mystery. No BS. It can be proved. Provide a traffic trace of the data copy via both cables. No Mystery. The protocol and the datagrams can be captured and viewed. No Doubt. Tell me how the Internet could function if the proper delivery of data was questionable. No BS. A whole lot more would "sound worse" than someone's HiFi.

No amount of "interference" can affect the data. It either gets to it's destination or it doesn't On or Off. 0s or 1s.

Blinded me by science vs. baffling me with BS.

I have totally lost all respect for Stereophile and it's Editors. You really have no clue on digital.

- Yonk IT and Network professional for 35 years

Oh, and it was really funny to have had my original account disabled after the posting above. That really made it hard to post again .. that just adds to my loss of respect.

Show me the traffic trace from the network. Do an SHA or MD5 hash of the file after copy from one place to another. It really is that simple.

John Atkinson's picture
yonk2 wrote:

I have totally lost all respect for Stereophile and it's [sic] Editors. You really have no clue on digital.

Again, you are looking at the system purely from an information point of view. Even if the analog signal representing the digital information is received correctly so that the bits are correctly represented, there are other mechanisms that can affect the sound when those bits are used to reconstruct an analog signal. Other posters have mentioned grounding problems and jitter. Only if those bits are never used to reconstruct an analog signal will you be correct.

yonk2 wrote:

IT and Network professional for 35 years

And I studied audio engineering and electronics 45 years ago. So what? What has your chest beating to do with the results of the simple listening test described in this article. Have you ever performed any such listening tests?

yonk2 wrote:

Oh, and it was really funny to have had my original account disabled after the posting above. That really made it hard to post again .. that just adds to my loss of respect.

The only account from a legitimate poster to this site that I have recently disabled was from "Jazz Purist," who was flaming other readers. Unless you are admitting that you are "Jazz Purist," your account was not disabled.

John Atkinson
Editor, Stereophile

yonk2's picture

And digital networks are not audio engineering.

It is IT (purely informational) and it can be diagnosed and quantified. If it sounds different then something else is the cause. If it sounds different then the network is broken all else being the same. Nothing other than the bits that make up the datagrams matter. There are no "mechanisms" in a TCP/IP stream of data that makes up a disk file that can possibly effect the sound. It's either there, or it's not.

Again. Show me the traffic trace. I don't need to, nor care to listen. The issues you are citing are like arguing if the sun will come up tomorrow. I don't need to open my window tomorrow to make sure - unless my name is Heisenberg. :-)

Oh, and I've spent PLENTY of hours in listening rooms. I'm also a season ticket holder for the NC Symphony and NC Master Chorale so I know what music is SUPPOSED to sound like. I have a huge 24bit digital library including the newly remastered Led Zep by Jimmy Page - awesome.
Chesky absolutely rocks for high quality digital and the recent Lorde - Pure Heroin is incredible as well.

John Atkinson's picture
yonk2 wrote:

Nothing other than the bits that make up the datagrams matter.

No-one is arguing that the bits are changed or not transmitted perfectly. However, when audio data is presented to a D/A converter, even if the bits are all correct, the timing of exactly when each data word is converted to analog matters. The right bit at the wrong time results in analog distortion. This is called "jitter" and has been examined at length in the audio engineering literature. This is why I keep explaining to you that looking at the transmissions of bits from an IT perspective is only relevant when those bits are never used to recreate an original analog signal.

John Atkinson
Editor, Stereophile

Bongo Fury's picture

If the difference in performance is that dramatic, doesn't it suggest the digital input section of the DAC is rather significantly lacking? Especially for a component that aspires to a high level of performance?

I’m an analog guy who only recently has started to pay attention to DAC reviews as my musician son has gotten more into headphones. But it hasn't taken long to notice that even some of the reasonably priced headphone oriented DACs discuss addressing this issue, reclocking and such.

Given that the issue is well known, do the SP DAC reviews include evaluations with a variety of cables? Seems knowing which DACs can or can’t gracefully deal with some variation in cables, let alone what I can only speculate may be larger differences in output side hardware timing, would be valuable consumer information.

John Atkinson's picture
Bongo Fury wrote:

If the difference in performance is that dramatic, doesn't it suggest the digital input section of the DAC is rather significantly lacking? Especially for a component that aspires to a high level of performance?

Yes it does. But to be fair to the DAC AudioQuest used in their dems, I have only rarely encountered a digital product that is totally immune to issues of grounding etc. When I test products, I experiment with all the ways of grounding the device under test to the Audio Precision analyzer and use the set-up that gives the lowest level of noise. When testing USB inputs, I use my laptop on battery power because I have found with some devices that the computer mains power supply can inject noise into the product's output.

Bongo Fury wrote:

Given that the issue is well known, do the SP DAC reviews include evaluations with a variety of cables?

Testing every product with every possible cable would take way too much time to be practicable. I have therefore used the same cables for some years in my lab testing. And Stereophile's reviewers always note what cables they use in their auditioning.

John Atkinson
Editor, Stereophile

Yeti-hunter's picture

While the description of jitter (the variability in time delay between received bits) is accurate, there are two significant lapses in logic when applying that concept to the impact that data cables have in an audio system: time scale and basic material properties.

First, I believe that it has been acknowledged that all the information that will ever be in a digital signal is contained in discrete 1’s and 0’s. If all the bits that were part of the original recording are delivered exactly the same in the exact same order, there is no difference in the files. There is no more and no less information to be had or missed.

The issue is the assertion that the timing of the arrival of those bits, or more accurately, the variation in the timing of the arrival (jitter) of those bits at a DAC, induced by a physical data cable will influence the sound. While this is a rational assertion on its face, there are two significant problems with this line of reasoning.

Time scale. The time that it takes for an electronic signal to move through 1 meter of Cat5 Ethernet cable is something on the order of 5 nanoseconds, plus or minus some fractions depending on the type of insulation (dielectric) used. A nanosecond is one billionth of a second and this scale of time is so far beyond human perception that it is completely irrelevant to human experience. At this scale, a jitter condition supposedly induced by the properties of a cable at a length that is commonly used for interconnects is so far beyond any human ability to perceive it, that it functionally doesn’t exist.

The second problem with this line of reasoning is the implication that the cable itself has a humanly detectable influence over jitter (variations in the time that it takes to receive individual packets or collections of data bits), in a cable length that is commonly associated with audio interconnects. Attenuation in a cable can account for machine measurable jitter over very long runs of tens of meters. But again, at the cable lengths used in home audio systems, any jitter induced by the physical properties of a properly constructed Cat5e cable are going to be so infinitesimally small that they are irrelevant both to machines handling the bits and to any human perception. In order for the jitter theory to be true, the cable itself would have to be an active medium where it was acting on specific bits of an electronic signal: delaying some 1’s and 0’s in a differentiated way from others on a time scale that was several orders of magnitude higher than the measured transmission rate of the cable itself – making the cable a storage device. How a passive medium like a copper cable specifically designed to accurately transmit data could delay some pieces of electronic data differently from others in the same signal stream passing through it, to the extent that it was humanly detectable, in cable lengths that are commonly used in audio, is beyond rational explanation.

If one is looking for jitter (or other artifacts of electronic processing), the active electronics in the end systems are the place to go. These components are where the bits of data are pulled from storage, assembled and packaged for delivery, transmitted across the cable, received and unpacked and converted to what we eventually hear as music. These are active bit handling processes and if detectable audio impacts due to jitter (or other errors) were to occur, it would be in these components. The point is that swapping one well-constructed passive digital data transport cable for another will not in itself introduce or eliminate relevant variations in the time that individual bits within a common data stream are received by a DAC. To the degree that cable induced variations occur, they exist at time scale so small that no human could perceive it.

However, the assertion (and noted experience) that swapping good-condition Ethernet data cables, had a dramatic audible impact on a bit perfect data stream delivered through the same system, is indicative of another very real and powerful phenomena: the placebo effect. This is why all drug studies are conducted as “double-blind”, because real sick people can feel better taking a sugar pill, if they think that it is supposed to help them. While those experiencing the placebo effect actually perceive to be feeling different with some even showing demonstrable short-term benefits, there was no actual compound introduced to cause any change at all. Those results are solely the work of the immensely powerful human mind.

John Atkinson's picture
Yeti-hunter wrote:
The issue is the assertion that the timing of the arrival of those bits, or more accurately, the variation in the timing of the arrival (jitter) of those bits at a DAC, induced by a physical data cable will influence the sound...Time scale. The time that it takes for an electronic signal to move through 1 meter of Cat5 Ethernet cable is something on the order of 5 nanoseconds, plus or minus some fractions depending on the type of insulation (dielectric) used. A nanosecond is one billionth of a second and this scale of time is so far beyond human perception that it is completely irrelevant to human experience. At this scale, a jitter condition supposedly induced by the properties of a cable at a length that is commonly used for interconnects is so far beyond any human ability to perceive it, that it functionally doesn’t exist.
The time taken by the datastream through the Ethernet cable is irrelevant. It is variations in the timing of the transitions between samples and data words that are being discussed here, which is referred to as "jitter." With 44.kHz data, the interval between samples is 22.67 microseconds, 22,000 times larger than a nanosecond, which, as you say, is a very small amount of time. But you can show mathematically that, with 16-bit data, if there is not to be measurable distortion above the 16-bit noise floor, the uncertainty in the timing of each conversion of the 16-bit word to analog needs to be five orders of magnitude lower than that interval, ie, IIRC, around 0.2ns or 200 picoseconds. See my 1990 article at www.stereophile.com/reference/1290jitter/index.html for a discussion of this phenomenon.

With 24-bit data, the math shows that the requirement is another two orders of magnitude more stringent. Changes in the local electrical conditions of the DAC chip in the converter that are introduced by, say, a change in Ethernet cable, from an unshielded generic cable to the fully shielded AudioQuest will change the inherent timing uncertainty and hence, possibly, the sound quality. As I keep saying, you have move outside the simplistic "bits are bits"model and examine the system's properties as a whole. hence my reference to Godel.

John Atkinson
Editor, Stereophile

John Atkinson's picture
John Atkinson wrote:
yonk2 wrote:

Oh, and it was really funny to have had my original account disabled after the posting above. That really made it hard to post again .. that just adds to my loss of respect.

The only account from a legitimate poster to this site that I have recently disabled was from "Jazz Purist," who was flaming other readers. Unless you are admitting that you are "Jazz Purist," your account was not disabled.

I just checked the user administration page and your "yonk" account is marked as "active," ie, not blocked.

John Atkinson
Editor, Stereophile

yonk2's picture

I completely understand jitter. I have a lot of experience with IP telephony. Unlike TCP/IP the protocols which telephony use ARE sensitive to network congestion and performance.

In moving a file with TCP/IP jitter is not a factor. Jitter comes into play when the computer pushes the bits down the USB to the DAC, but by that time the track I'm listening to (Mac-Pro and Audirvana+) has been read into memory and the network is no longer a factor. If your computer was so old and so wimpy (in which case no modern DAC playing software would work anyway) then you *could* have an issue where memory was constrained and the network transfer on-going such that interrupts could effect the music play. However I'd argue that no reasonably new laptop or Mac-mini or mac-pro would experience this. In any case I'd take the money and instead of buying snake-oil cables improve the computer to completely eliminate any possibility of this side-effect.

Biffstar's picture

I too am a networked systems engineer and I categorically agree with yonk2. You can't improve, per se, the transfer of digital data.

Unless a double-blind test is performed between an "audiophile" network patch cable versus a generic cable and chosen correctly every time, or a complete bitstream dump between the two cables is displayed with noticable differences between the two, saying that an audible improvement is evident during a brief tradeshow demonstration using an overpriced cat5 is the biggest case of placebo effect since the dawn of civilization.

We have yet to see any blind test results or bitstream comparisons. Is it any surprise that these manufacturers or journalists/reviewers haven't provided any?

Plus, if you were to expand on the rationale provided by the defenders of these cables (jitter, interference, etc), the entire digital audio tract would need to be brought up to snuff as well to match. Do the circuit boards in the source computer use gold traces and gold-plated USB ports? Are the pins on the CPU, memory slots and DIMMS themselves gold plated and of extremely high quality with shielding? Is the SATA connector to the source drive of as high a quality as the network cable too? No, of course not.

Plain and simply, anybody that believes that these cables actually make any sort of audible or measureable improvement is an ignorant fool with their head in the clouds. The entire argument can be summed up with two words: PROVE IT.

adamdea's picture

I have a question for John- granted that you insist that it cannot be ruled out that a dramatic sonic improvement can be generated by a change in the Ethernet cable feeding the Mac mini serving the DAC by asynch USB ...
A) Is there any sense in which this is more likely than that the same sort of change could be generated by placing a single grain of sand on the cable
B) is there any reason to suppose that the change in cable would be likely consistently to produce, by the mechanism you have identified, better rather than worse results either in the same system over time or across different systems?

mrhyfy's picture

If this magical cable makes such difference can you just imagine the difference an Audioquest switch / router would make?? I suspect the esoteric cable manufacturers regret the lost sales opportunities that internetworking standards have caused.

eugovector's picture

You're right. John, what do you say to "audiophiles" on the topic of what router they should use? Surely this is more important than an ethernet cable.

Antipodes_Audio's picture

I will no doubt be seen as biased because we make music servers, and worse still, audio cables. But I must leap to John's defense. As John has tried to say, when we study engineering we are not studying the real world. We are studying relatively simple abstractions from the real world that work sufficiently to be termed scientific knowledge. The moment you treat such knowledge as irrefutable fact, particularly outside the context of the experiments used to support the knowledge, we are not being scientific. To be more precise, inducing your knowledge about the theory of digital transmission to decide what can and cannot sound different is unscientific without experimentation to prove it.

yonk2's picture

I said earlier: "I don't disagree that the final test is always in the listening! However claims DO have some basis in science and engineering - especially digital which unlike analog CAN be quantified."

Do you disagree that listening is subjective, or in scientific terms "qualitative vs quantitative?"

I keep saying "SHOW ME THE TRAFFIC TRACE." The performance characteristics of digital packet networks is well understood and quantified. "we are not being scientific" is obfuscation and total BS.

It's like arguing the "theory of evolution" vs "creation science." Show me the data - if you dare!

YOU are ignoring the science and engineering and relying on mysticism.

HiFi is not the "real world." It the closest approximation (and actually a rather poor one) compared to the real world of live music sitting/standing in front of the musicians. If you can spend enough money and have a really good room you can approximate the real world to the best that the recording itself can resolve.

Antipodes_Audio's picture

The subject of blind testing has been brought up. We routinely use blind testing and even conduct public tests. A frequent one we do is showing that DACs are not resilient to noise interference with the incoming signal, and this applies to USB, Ethernet etc. It is not about the bits information being obscured, but about the way DACs work. If you assume a buffer/reclock step can eliminate noise and jitter then I can agree with the nay-sayers here. But we have regularly demonstrated that buffering/reclocking/resampling in the playback chain changes things but does not eliminate jitter and noise.

The only buffering/reclocking that works as claimed is to transport the bits, write them to storage, finish writing it to storage and then read from storage. The storage cannot record jitter and noise, just the bits. But when you write to and read from the buffer at the same time and at different rates, it is audibly clear that it is far from perfect.

The issue/question with streaming DACs is that they rely on buffering and reclocking in the DAC to overcome the asynchronous nature of Ethernet.

yonk2's picture

All I can say to that is that you clearly do not understand modern computer networking and operating systems. I don't disagree about the jitter re-clocking, blah, blah, blah.

Are you telling me a Disc player (blu-ray, SACD, CD, etc) does not buffer the bits someplace? Nonsense. The player oversamples the spinning media - the more times the better. Where is the oversample data stored? In some memory buffer. That's the COPY which then gets fed to it's DtoA converter.

Yes, the NAS->Computer->USB->DAC path has more intermediate copies, but the only critical one is the buffer which feeds the USB to the DAC where the clock needs to be as perfect as possible.

As I've said before, everything that's being played and fed to the DAC is already a Bit-Perfect copy in computer memory (i.e. Audirvana+ Application). There are no noises or artifacts in the buffer. The protocols take care of that as sure as the Sun will shine. The quality of the USB and DAC is the critical point - nothing else. With very little equipment I can PROVE this - unlike analog, where even measuring the signal can cause undesired/unknown changes.

The "asynchronous nature of Ethernet" has no effect on the bit stream. It's simply a bit delivery medium. The Internet would simply fall apart if what you believe it true. Do the math of the capability / use:

GigE is 1,000,000,000 bits/sec - take about 20% for "overhead" (protocols and framing) and that leaves 800 million bits/sec. A 24/192 audio stream is (stereo rounded UP) is 10 million bits/sec which is 1.25% (worst case) capacity of the link.

Antipodes_Audio's picture

For a start, John made an observation. Observations can be erroneous. Take it for what it is. He did not present it as a scientifically conducted rigorous experiment. I appreciate that people describing observations that do not gel with another person's strongly held belief can be upsetting to that other person. But that is not an excuse to launch personal attacks. Call it out for what it is - a random observation. Your fellow man is smart enough to know the difference. By the way, I am not claiming to hear any difference between Ethernet cables as I have not tried yet. John's statement merely suggests to me that I should try them out, no more than that.

As to science, as stated my firm completes blind tests on the result (what people experience), and over the years we have found this to be more useful that limited measurements for starting the process. The existence of measured differences does not prove there is an audible difference. And equally the absence of measured differences does not mean there isn't an audible difference.

We have got reliable blind test results between cables that show nothing much above the noise floor in a null test. You can put your faith in the null test and believe the experiments were faulty. Or you can conduct the experiments in a variety of ways to remove bias and see what happens then. We prefer to use the latter as we don't accept that a limited model null test is conclusive in the light of the observations. I expect you would take a different view and I can respect that choice, but caution against drawing a wide set of conclusions from a limited measurement. It is one thing to have a mental model of how a stereo system works and within that decide a null test is conclusive. A real audio system may involve influences and relationships that are not covered in your model and we won't find them if we close our minds to them.

The complexity of the processes that the brain uses to get from receiving sound pressure waves to experiencing musical enjoyment is not well enough understood that anyone can tell you what distortions in what quantity upset that process and what don't. When we conduct blind tests and get consistent results, we decide that the difference is meaningful. The benefit of measurements is different. Having identified a difference in blind tests we theorise as to what might be causing it. Then we try to measure the cause. Then we try to manipulate the cause and measure that we have been successful (in what we thought was important). But then we do blind tests to check whether our theory has produced something useful.

Naimdude's picture

Stevey Mejias... So much passion, so much promise... A man who wrote passionately about budget audio and claimed in an article, that the Furutech 250$ wall outlet did not improve the sound from his gear.

Now, SM tells people that 4000$ USB and HDMI cables make a difference... Disappointment is too weak a word...

I hope Mrs. Mejias is happy spending your new and bigger income...

eugovector's picture

With this trajectory, he'll be at synergistic in a year. Jason will be so jealous.

X