Ethernet Cable Shootout: Comparing Network Cables from $5 to more than $1,000 USD

Why on Earth do I do this?

I rolled my eyes, sat back, listened and shook my head. I sighed.

The midrange had tightened, with the close-mic’d acoustic guitar gaining tonal and timbral definition as well as more perceived weight to the wooden body the softly-plucked notes were emanating from. This and the brush on drum skins were slightly more forward and higher vertically in their spatial presentation between my speakers and the entire sound stage had opened up.

What precipitated these sonic changes in my hi-fi?

Swapping the Ethernet cable between the source component and the network switch feeding it.

This type of behaviour begs for harassment online or from friends and family. Spending more than a dozen hours over three days swapping cables back-and-forth and listening for inherent sonic changes probably qualifies me for for some type of therapy. But, as an audiophile this type of experimentation is something that becomes important because the best ethernet cable, like the best amplifier, does not add or subtract anything from the source signal being passed along to the component – if it is doing the job it was designed for – you should only be hearing more of the information embedded in that carrier. Many argue that digital cables cannot make a difference, they are welcome to this view, I can only pass along what I have heard, and I have heard the differences that ethernet cables – like AC cables – can make to a sound system.

Besides, I know there are many music lovers out there putting together network-enabled, digital-based sound systems who wonder silently or aloud in the darkness of their listening rooms, “Do ethernet cables make a difference?”

I didn’t plan to do this comparison, it came about while I was preparing to move, self-isolating from coronavirus, and realized I had amassed a small collection of ethernet cables ranging in price from $5 to well beyond $1,000 USD. Some grabbed in a pinch at Best Buy, others sent to me from boutique, high-end companies that specialize in bespoke cables. There are no measurements, double-blind tests or specifications to refer to – just me spending a good chunk of my week listening to the same album over and over, taking notes. This is not meant to be a definitive comparison, it is merely one man, five cables, and a window onto what you might expect if you were to try the same thing at home.

The Linn Klimax DSM.

I conducted this comparison using one of the highest-resolving network music-player/DAC combos I’ve ever had on hand: the Linn Klimax DSM ($23,380 USD). I used it because I’d been listening to it for three weeks straight with a Pass Labs XP-12 and X150.8 pre/power amp combo feeding my Harbeth M40.1 loudspeakers, and I had this setup’s sound locked down in my head.

Pass Labs pre and power amplification.

My thinking was simple; if you’re listening for changes in a cable, you want a source capable of truly imposing resolution and clarity. The Linn fit the bill. I had it on a dedicated, audio-only D-Link 100/1000 five-port switch shared with a Roon Nucleus+ which was feeding the Klimax a mix of dematerialized albums (TIDAL, Qobuz) and locally-stored Redbook, high-res PCM and DSD files. Cables were only compared between the Linn and the switch, with a ‘control’ three-foot Blue Jeans CAT6 cable running from the Nucleus+ to the switch and a 25-foot CAT6 network patch cable connecting the switch to my router.

Linn Klimax DSM and Pass Labs XP-12 preamplifier.

The cables in the shootout were: StarTech yellow CAT5e three-foot network patch cable bought from Best Buy for about $5 when I grabbed it a couple years ago. Blue Jeans three-foot Bonded Pair CAT6 in black ordered direct online for $10 a few weeks ago. A 2m totaldac ethernet cable currently going for $425 USD with Euro exchange. A 1m Final Touch Audio Matis ethernet cable with a current MSRP of $805 USD. And lastly a 2m TelluriumQ Black Diamond ethernet cable that retails for £780/m – $1,760 USD for 2m with current exchange.

All the cables had at least 80-100 hours on them, some much more, so they were all what I would personally consider ‘burned-in’ and settled down (some people think burn-in is bullshit, I feel bad for anyone who can’t hear the difference between a a component or cable used in hi-fi that is box fresh vs having a 100 hours on it. YMMV). The album of choice for this completely subjective experiment was a Bandcamp 24-bit/96kHz FLAC of Seu Jorge & Rogê – Night Dreamer Direct-to-Disc Sessions. The single-take recordings are cut live to acetate on Artone Studios’ vintage Neumann VMS70 lathe, and are then taken downstairs to the Record Industry pressing plant to be galvanized and pressed. Needless to say, this album sounds spectacular. I also ordered the LP.

StarTech yellow CAT5e.

1. StarTech yellow CAT5e: Yellow, plain, flexible, standard terminations. I started with the StarTech cable and the first cut “Saravá” which opens with Jorge’s gentle and playful guitar emanating from the centre-left of the sound stage. It spread out life-sized for about three-feet horizontally/vertically starting about a foot to the right of the lefthand loudspeaker. Jorge and Rogê’s vocals didn’t have as much ‘chest’ to them as I would’ve liked when they float in about 20-seconds later in front of and on top of the guitar, but Jorge’s voice maintained the expected spatial relationship to the guitar body he played in the X&Y planes. Peu Meurray and Pretinho da Serrinha’s drum work figured prominently almost dead-centre between the speakers and were stacked behind the guitar and vocals. Tonally and timbrally the body of the guitar had real woody body volume to its impact. On “A Forca,” the two trade delicate, but powerful licks on their six-strings – easy to tell apart because of timbral accuracy – with the various chimes, dog barks and bird calls adding a noticeable amount of air and space – along with their associated spatial cues – to the recording. The sound stage was large, splayed out between the speakers, but oddly, like there was an invisible line, stayed within the outer boundaries of the speaker cabinets. The players staging in the Z-plane was rather shallow overall. Percussive hits had, tangible snap to them and bass/kick-drum depth was commendable. Everything seemed to have an expressed neutrality, as opposed to any warming effect, that cables may sometimes introduce.

2. totaldac ethernet cable: Black, heavy, covered in mesh wrapping around what seems to be a hundred or so one-inch metal rings fitted like armour around the inner core of the cable itself. Terminations looked to be standard. Surprisingly flexible (to a point) because of the small width of each ring. “Saravá” opens and immediately there is an increase in not only resolution, but color and bounce too, with the size and alacrity of the guitar swelling compared to the StarTech. Gains are noticeable to visceral projection of presence to guitar, vocals and percussion. Frequency extremes have opened up, leapt past their previous boundaries. Timbral harmonics are far more fleshed out and there is a substantial increase in depth and breadth to the sound stage. Voices and instruments shift slightly further to the right compared to the StarTech, percussion digs deeper, seemingly dropping down to pulse from between a lower octave. More textured squeak from fingers on strings, hyper-realism to the sound of hand slaps, and brushes on drum skins seemed countable. The X,Y & Z plane layering of instruments comes across as a natural, organic and human mixture of the players as opposed to what now seemed like a ‘model stage’ populated with two-dimensional players/instruments on it through the StarTech. A constraint to the overall recorded space that wasn’t noticeable until the StarTech cable was swapped out for the totaldac, with an effortless flow and ebb to the emotional tenor of “Carminhao” now in place.

TelluriumQ Black Diamond.

3. TelluriumQ Black Diamond: Black braid over super-stiff cable that pushed everything around and sometimes made connections flex because the cable wouldn’t. Longer, large-sprung, high-grade terminations which in tight-fitting spaces for connection were difficult to disengage. Marked as directional. Within 30 seconds of the TelluriumQ cable being in the mix you have a noticeable refinement to the overall tonal balance of the music’s presentation that was not there with either the totaldac or StarTech. There’s a slight tightening to the sound compared to the totaldac. Bass depth is substantial and comparable to the totaldac, with similar spectral-like decay to bells, whistles, zingers, voices and metallic percussive stikes on top. The upper registers came off as ever-so-slightly smoothed out, losing some of their spark and jump that came through with the totaldac. Spatial presentation is deep and wide, with the players once again laid out in a very natural and relaxed manner that came across as more ‘bro mix’ than ‘studio piece’ – if that makes sense. I guess I’d say it managed to capture the easygoing emotional tenor of the recording session as Rogê and Jorge are musical soul mates of a kind. On the cut “Meu Brasil” there’s an underlying kick drum that comes in occasionally, which through the totaldac/TQ has concrete solidity to it, whereas via the StarTech it lacked the same definition. Vocal inflection was crisp, detailed and not noticeably sibilant (neither was it with either the other cables so far). Less warmth overall – tonally – than the totoaldac, but also less perceived grunt and grit – less quick and dirty I guess I’d venture in comparison – cleaner, tighter and slightly brighter on vocals. The totaldac darkened things up just slightly – the TQ is the most tonally coherent cable evenly across the frequecny spectrum to this point.

Final Touch Audio Matis.

4. Final Touch Audio Matis: Multicolored black and gold braid weave that is a flattened oval in cross-section. Very light and flexible with medium-length, large-sprung, high-grade terminations. The Matis presented big, weighty body to the lower midrange and bottom end with the most spread out and deep sound stage of any of the cables so far – instruments were placed behind the front wall and outside speaker boundaries by a good couple of feet. Treble was sweet, extended and very open – lots of air and space around those high-hat, cymbal crashes, bells, whistles, and upper-register vocals – with in-the-room, holographic believability/presence to the overall spatial sound stage. Tonality and timbre had richness with voices and instruments, guitars and percussion coming through with gritty, organic, human textures. Chesty vocal emanations, fret work, string pressure, and wooden-body slaps… floor tom and hanging tom skin whacks/brushes could all be ‘felt’ viscerally against my body. Kick drum and bass on “Onda Carioca” had concussive chest impact which reminded me of being next to small stages at punk shows in my youth, with the drum kit punching the hardest and tightest on “Meu Brasil” so far. While “Saravá” opened up again to the lower/mid left in the sound stage, it migrated immediately to the centre when Jorge and Rogê’s vocals harmonized in. Overall spatial presence was deeper, wider and had a far more relaxed co-mingling of instruments and voices that all the more expensive cables seem to share. Unravelling each instrument’s piece woven between one another, song-after-song, was the easiest through the Matis of the cables. You could hang on one musical thread if you wanted, but there was no stopping pulling back to take in the whole emotional fabric of each track. In fact, I found it difficult to stay focused on one thread, innately preferring the overview and associated mental connection it inferred.

Blue Jeans three-foot Bonded Pair CAT6: Plain black cables with standard terminations, lightweight with good flexibility. Rogê’s plaintive vocal intro on “Meu Brasil” was nicely forward and high in the mix, with solid layering to his guitar and the accompanying percussive slam. Surprisingly wide and deep sound stage compared to the StarTech, closer to the higher-priced cables in both this regard and believability to players arranged on the stage. Bass slam, midrange punch and treble sparkle/air were a definite step-up from the StarTech too, but unsurprisingly, not in the same class as the totaldac, TQ or FTA cables. Excellent spatial cues in reproduction of the recorded space on vocals too. The Blue Jeans also did a great job of siloing individual players; Rogê and Jorge’s vocal harmonizing in particular – the two were presented as if sitting on chairs beside one another about five-feet apart throughout the album. There was no spatial drift; artists and instruments were locked in place track-to-track. Pitch stability was unwavering too. Timbral accuracy let one easily discern between the two six-string guitar bodies, with tonal color shifted more towards warm than neutral – which I prefer on intimate, close-mic’d acoustic recordings like this album. Transient speed and PRaT was fast and driven, with excellent resolution and detail retrieval on “Vem Me Salvar.” The only cable that skewed towards offering a more analytical view of the music was the StarTech. Through the rest of the cables – even the Blue Jeans – the emotional tenor of songs tended to supercede the other noticeable traits.

The input matters.

Conclusion

If you’re hooking up a network player, or network player/DAC – as I did here – then you will need an ethernet cable to pass along the cloud-based binary libraries of TIDAL, Qobuz, Spotify, Amazon, etc. So, hopefully this was helpful in showing that this often overlooked ink to access music holds the ability to critically influence what your source is passing along to your ears. Each cable used in this listening session allowed more or less – with differing presentations – of what the Linn Klimax DSM was capable of translating from its binary diet. None of the cables changed the fundamental sound signature, tenor, timbre, tactility or musicality of the Linn, what they did do was allow a window into how the different construction and materials involved with each cable affected the flavoring of the spectrum in the audible frequency band. I will say, that the sound became more natural and lifelike when using cables constructed specifically for audio performance – this also translated to costlier. You can take my findings for what they’re worth: the listening opinion of one man. Digital audio is a strange, complex and oftentimes obsessive pursuit of seeking sonic perfection where there really is none; there is only the mountain that each individual climbs to attain their own summit of what they qualify as audio perfection. In listening to these cables I came to prefer and enjoy certain attributes of what each brought from the Linn Klimax. Choosing one to keep in this system is a non-issue for me because this is not my reference system, it’s merely a transient system like many that come and go, such is the nature of the hi-fi reviewer. Since I’m often changing and swapping gear, and sometimes whole systems, I often must compromise on how each new iteration is built and fleshed out based on what I have ‘in stock.’ Similarly, you will do the same when you build your own systems, based on things like available space, source preference (digital, vinyl – both), and most importantly; budget. I can’t tell you which cable would be best for you, because the experiment was specific to a place, time and certain components. I can say that it was time well spent because it helped show me again, that like all things in hi-fi, curating the best-sounding system you can is a balancing act where one is always trying to emphasize strengths and minimize weaknesses.

COMMENTS
AllanMarcus's picture

The expectation bias is real. It is not possible for there to be a sound difference with ethernet cables, with the exception of a really poorly designed system that ,might possible pick up RFI.

Someone do a double blind, or even single blind test and let us know.

Ug.

Ejlif's picture

Rafe clearly heard some big differences. I guess you can never take his word for any differences he hears on anything from now on in fact any reviewer who says they hear something from a product you don't believe can show a difference in sound cannot ever be trusted again right?

jeffhenning's picture

There are two things to examine Ethernet cables for:

• Data throughput integrity
• RFI/EMI shielding and how easily the cable can accept & cause RFI/EMI

Ethernet cables have no "sound" unless they are causing data packets to get lost or corrupted. Even then, what you are dealing with is how the DAC will deal with the lost data and whether there is any form of error correction in the DAC. The cable either passes the data properly enough to be decoded or doesn't. It's that simple.

I hope no one gives this "shoot out" any credence. Without any empirical data, this review is of no value.

The best Ethernet cabling to use is single twisted pair (STP) CAT 6 that uses solid copper conductors rather than copper-plated tin. Having gold-plated connectors helps as well. This type of cable also offers shielding for each pair of wires (positive & negative) as well as the entire group of 8 wires.

Monoprice offers a really nice, 25 foot STP Ethernet cable for about $13. that matches all these criteria.

There is no reason for a $500 Ethernet cable to exist.

Rafe Arnott's picture
... is welcome here. But it is nothing more than that; simply your opinion.

This shootout was never represented as anything more than listening notes within the context of a single system. I made it clear that there were no 'measurements,' 'data,' etc.

You, like others, are free to draw their own conclusions, and as with all things in this hobby, everything is subjective.

Dismissing someone's impressions because they don't jive with yours doesn't prove anything.

Also, telling people what the 'the best cable' is without any evidence seems disingenuous on your part. I made no such claims in this comparison, I wrote to the sonic attributes of the cables and had personal preferences for certain attributes with certain cables – but I declared no winner or 'best cable.'

Dick James's picture

There is no reason for a digital cable, Ethernet or USB, to have a "sound." The bits are either error-free or they aren't, and with Ethernet, the packets are resent until correct if there are errors. A buffer stores the bits before the DAC gets them. Most DAC chips don't do any error-correcting as that's done up the digital chain before the DAC.

This debate is similar to the power-cable debate: Most people's power generator is hundreds or thousands of miles from their home, but yet they think the last few feet from their wall to their electronics makes a huge difference.

Richard D. George's picture

Rafe:
Thank you for the comparison.
I am "crazy" enough that I had custom installers use (relatively) high quality Audioquest CAT6 in-wall throughout two houses.

All:
Read "Grinding Your Own Gears" at Darko.Audio. The subject is USB cables but John Darko's witty exchange makes excellent points that apply here.

stevebythebay's picture

So, what was the switch you used? Apart from the bits getting to the DAC (which are rarely if ever going to fail to get to the DAC from the source) was there any perceived lower noise using one cable over another? There's so much more to what takes place in just this network chain than getting all the bits from point A to B.

Rafe Arnott's picture
The DGS-105 D-Link: https://us.dlink.com/en/products/dgs-105-5-port-unmanaged-gigabit-desktop-switch

Perceived noise floor differences tended to favor the three more expensive cables as there was more low-level information available to discern. The Blue Jeans tended this way more so then the StarTech as well.

I can only attribute this to the 'audiophile' cables being designed for better RF/EMI shielding in their construction/design, which is a common practice in all high-quality cables on the market, be they AC, RCA, XLR, USB, etc.

Thanks for chiming in!

Hope you enjoyed the piece, it was actually fun to do. Something most people in this hobby seem to forget.

stevebythebay's picture

That lowly switch (had one, upgraded to used Cisco 2960 and most recently an Uptone EtherREGEN) will be a quite limiting factor. The noise level in the network is huge with conventional switches, even good ones like the Cisco. And a true purpose built Ethernet cable like recent ones from Shunyata Research do an even better job at reducing electromagnetic interference among its other features. So, when you get a chance down the road take another crack at this topic.

PeterMusic's picture

Thanks, Rafe. This is a very important topic. Even the critics such as AllenMarcus acknowledge as much when they note the RFI issue. Our systems are screaming bundles of electrical and other noise that must be tamed. I did not appreciate this until similar "shootouts" at home. If you're dropping 5 figures on a stereo, and you're not dropping a few thousand on cables and noise reduction, you owe it to yourself to borrow a few samples from your local dealer. Cheers

AllanMarcus's picture

I have borrowed expensive cables, and made some too. I can hear no difference. I'm lucky. Simple, well made cables (Like for Bluejeans) works as well as really expensive ones from AQ (for me, at least).

I use monoprice power cords. I splurged and got 16AWG. I use Monoprice speaker cable for my 5' run for the mains. I also splurged and got 12AWG cable. I compared with some named brand stuff for both cables and could hear no difference. I compared headphone cables (home made vs $2000); no difference to me. I compared USB cables and RCA cables. No difference. If ANYONE has any published data on blind tests or any way of measuring the differences, please post.

I sat in a room at ListenUp with a bunch of other people as AQ demoed their top of the line optical cable vs a target $20 cable. Neither I nor most of the people in the room heard a difference (It was not a blind test). One or two people claimed they could hear a difference. 10 others could not.

As for Ethernet, generally there's a NAS or server, the data is moved from there to a client, and then fed to DAC via USB, Optical, or Coax. I imagine the file is completely transferred into memory on the client, then fed to the DAC, so the "quality" of error correct transmission from the NAS to the client is irrelevant to thing like "timing". We would need to understand how the specific client deals with network based files. Even if were truly streaming, I would imagine the client buffers the error corrected file information before sending it to the DAC, which would also eliminate any ethernet related issues.

The issue of EMI noise? Someone prove it, please. Should not be hard to reproduce in a Lab.

Mr Underhill's picture

Thanks Rafe,

Interesting. Over recent years I have extensively used: NUCs; microRendu; ultraRendu; and, SOtM Trifecta. With these boxes I found upstream changes to be readily heard; whether they were improvements or not was another matter!

More recently I have been using a Linn Klimax DS/1 Renew. This I found REMARKABLY indifferent to upstream widgets, cables and other changes.

On a side note:
I have recently started using the Meridian 210 streamer with my old Audio Note 2.1 DAC - it is BLOODY BRILLIANT. Best digital I have every had at home.

Thx for the hard work,

M

Cout's picture

The only thing I would want to know about an ethernet cable is how it might improve or degrade the measurements of my DAC. Show me a cable that improves the SINAD of my DAC by a couple of points and you've got my attention.

gilde's picture

Your audio over Ethernet is traveling on top of robust protocols that automatically recover from errors. It's like copying files on a computer - they don't get degraded. So data loss is a non-issue.

The extent that an Ethernet cable allows electromagnetic noise to be introduced to your system is hard to say - but the sonic differences the author mentioned are hard to attribute to noise.

romaz's picture

It's interesting that Allan Marcus is accusing you of expectation bias based on his own expectation bias that an Ethernet cable cannot possibly impact sound. He has probably never even attempted to conduct listening tests even though they are easy enough to do. Having performed group double blind testing on several occasions with various cables including Ethernet cables, sound differences among various Ethernet cables are not hard to discern. With some cables, the differences are subtle and with others, quite stark and golden ear definitely not required. My guess is that Allan won't take the time to do this testing for himself because he'll claim he doesn't have time to be bothered with it or that it's beneath him and yet he has the time to read your review knowing what it's about and to comment.

bcredman's picture

Many skeptics of subjective listening reports of audible differences in digital connectors use the old "bits are bits" argument that implies that bits are some immutable inviolable quantity.

As someone with advanced degrees in engineering, physics, and mathematics who has performed research and testing on digital electronic systems (for sensor systems rather than for audio) for a few decades, I would like to state that the "bits are bits" argument is an oversimplification of reality.

In all digital systems the transmission of the idealized bits used in the elementary fundamental theory of digital processing and communications occurs over analog channels of finite bandwidth and with all sorts of distorting effects on the shape of the binary waveform in the real world.

The finite bandwidth has the effect of causing binary waveforms to have finite transition times instead the of the zero transition times of the idealized bits. Poor impedance matching at terminations causes not only power transfer loss (resulting in signal-to-noise ratio loss), but also reflections back and forth along the cable resulting in interference effects that can cause amplitude fluctuations.

These phenomena can have measurable effects on signal-to-noise ratio (SNR), bit error rate (BER), timing jitter, and the performance of error correction codes. These deleterious effects are in addition to those caused by poor shielding from EMI/RFI.

In my experience, usually, but not always, the effects of poor EMI/RFI shielding are far worse than these other effects, so that they obscure these other effects until the EMI/RFI is reduced to a level that these other smaller effects become apparent. Therefore, I agree that poor EMI/RFI shielding should be the primary suspect in causing audible differences between digital cables. However, there can be other causes even after any RFI/EMI shielding issues have been resolved.

One major difference between ethernet cables of different categories is the analog bandwidth of the cable, ranging from 100 MHz for Cat 5/5e to 1000 MHz for Cat 7a, for example. Although even the lowest of these is far above the bandwidth of audio signals or audio code rates, that is not relevant for this discussion. What is relevant is that these bandwidths effect the transition times for the edges of the bits. This is important in that digital filters are most often designed assuming idealized zero transition time bits so that the finite transition time of the real received binary waveform causes a mismatch between the actual waveform and the digital filter. This mismatch causes a lower SNR, higher BER, higher timing errors, and higher code cross talk.

These issues are at least partially addressed in ethernet cable specifications such as Power Sum Equal-Level Far-End Crosstalk (PS-ELFEXT), Near-End Crosstalk (NEXT), Power-Sum Alien Crosstalk, Attenuation, and Return Loss (RL) with generally improving specifications with increasing Cat number. In addition, Cat 7 cables uses four individually shielded pairs inside an overall shield and have far more stringent specifications for crosstalk and system noise. The fully shielded cable virtually eliminates crosstalk between the pairs and is highly resistant to EMI/RFI.

How well the cables from different manufacturers meet or exceed the specifications may differ among manufacturers.

Therefore, "bits are bits" is an oversimplification that is at best a first-order approximation to the much more complicated real world situation. It is sufficient for keeping the formulae simple enough for undergraduate electronics engineers and hobbyists to work with, but is misleading when it is taken as the last word on digital performance in the face of real world complications.

Because of the differences in specifications between the different ethernet cable categories, the article would have been better if it had given the category designations for the "boutique" cables used instead of just for the more generic cables used.

Otherwise, I found the reviewer's subjective impressions of the sound using different cables interesting and well expressed in the writing.

As with all subjective impressions, the reviewers impressions may be very different from any other individual's impressions and value judgments, but this does not make them useless anymore than subjective reviews of restaurants, wines, movies, or music by others would be considered useless. It's just that subjective reviews should not be taken as "gospel" anymore than slogans like "bits are bits" or "perfect sound forever" should be taken as "gospel."

Absolute Zero's picture

"Many skeptics of subjective listening reports of audible differences in digital connectors use the old "bits are bits" argument that implies that bits are some immutable inviolable quantity."

There's not a singular argument however. You're the one now trying to over simplify.

1. Bit's are Bit's. It's why we have CRC32, MD5, etc...
2. We are talking about non-realtime, asynch data transfer. It's hurry up and wait:

https://youtu.be/O7Ht58vdA8w?t=564

"I would like to state that the "bits are bits" argument is an oversimplification of reality."

Agreed and it's not the only argument being made however. So please, no red herrings.

"In all digital systems the transmission of the idealized bits used in the elementary fundamental theory of digital processing and communications occurs over analog channels of finite bandwidth and with all sorts of distorting effects on the shape of the binary waveform in the real world."

Yes, all these different encoding mechanisms ride on an analog carrier, NRT, PAM, PWM, PCM, etc...

"The finite bandwidth has the effect of causing binary waveforms to have finite transition times instead the of the zero transition times of the idealized bits. Poor impedance matching at terminations causes not only power transfer loss (resulting in signal-to-noise ratio loss), but also reflections back and forth along the cable resulting in interference effects that can cause amplitude fluctuations."

Yes they can, they are overseen by IEEE / ISO / BiSCI etc engineering bodies. Buy in spec Ethernet cables (Blue Jeans, Panduit, Tripplite).

"These phenomena can have measurable effects on signal-to-noise ratio (SNR), bit error rate (BER), timing jitter, and the performance of error correction codes. These deleterious effects are in addition to those caused by poor shielding from EMI/RFI."

UTP CAT5 is pretty much noise immune to 30MHz. Also I find it disingenuous when we are talking about cable runs that are averaging 2-3 meters for most people, out of an allowable 100 meters. Also most modern PHY can transmit out to 380 feet before BER starts to rear it's ugly head at 1000Mbps PAM 16. Remember the supported spec is 328 feet / 100 meters.

"In my experience, usually, but not always, the effects of poor EMI/RFI shielding are far worse than these other effects, so that they obscure these other effects until the EMI/RFI is reduced to a level that these other smaller effects become apparent. Therefore, I agree that poor EMI/RFI shielding should be the primary suspect in causing audible differences between digital cables. However, there can be other causes even after any RFI/EMI shielding issues have been resolved."

Read T.I's paper on 10/100 LAN PHY and radiated emissions. Read Seimons paper "The Antenna Myth". Bottom line: you need in spec cabling. Nothing more.

"One major difference between Ethernet cables of different categories is the analog bandwidth of the cable, ranging from 100 MHz for Cat 5/5e to 1000 MHz for Cat 7a, for example. Although even the lowest of these is far above the bandwidth of audio signals or audio code rates, that is not relevant for this discussion. What is relevant is that these bandwidths effect the transition times for the edges of the bits. This is important in that digital filters are most often designed assuming idealized zero transition time bits so that the finite transition time of the real received binary waveform causes a mismatch between the actual waveform and the digital filter. This mismatch causes a lower SNR, higher BER, higher timing errors, and higher code cross talk."

I think you are purposefully misleading here, IMO. 'Although even the lowest of these is far above the bandwidth of audio signals or audio code rates, that is not relevant for this discussion' It's about this EXACTLY. I run 10GBe fiber. My computer streamer is capable of 332MB/s. I can cache a flac track in 300ms and my network connection drops to 0Kbps.

What is only relevant is what happens during that 300ms window. You can unplug the cable for pete's sake.

"These issues are at least partially addressed in ethernet cable specifications such as Power Sum Equal-Level Far-End Crosstalk (PS-ELFEXT), Near-End Crosstalk (NEXT), Power-Sum Alien Crosstalk, Attenuation, and Return Loss (RL) with generally improving specifications with increasing Cat number. In addition, Cat 7 cables uses four individually shielded pairs inside an overall shield and have far more stringent specifications for crosstalk and system noise. The fully shielded cable virtually eliminates crosstalk between the pairs and is highly resistant to EMI/RFI"

Specs get tighter as speeds increase. That's why after 10GBe you are looking at optical connectivity.

"How well the cables from different manufacturers meet or exceed the specifications may differ among manufacturers."

Not seeing your above point. Get a cable that is certified and move on.

"Therefore, "bits are bits" is an oversimplification that is at best a first-order approximation to the much more complicated real world situation. It is sufficient for keeping the formulae simple enough for undergraduate electronics engineers and hobbyists to work with, but is misleading when it is taken as the last word on digital performance in the face of real world complications."

And none of this has anything to do with the output of a DAC, UNLESS, you can find and post empirical DAC analog outputs being measured and displaying such effects.

"Because of the differences in specifications between the different ethernet cable categories, the article would have been better if it had given the category designations for the "boutique" cables used instead of just for the more generic cables used."

Why? even CAT5e can to 10GBe up to 147 feet. In a fully optimized system we are talking 1250MB/s. That's a PCM 16/44.1 album in 500ms on average.

"Otherwise, I found the reviewer's subjective impressions of the sound using different cables interesting and well expressed in the writing.
As with all subjective impressions, the reviewers impressions may be very different from any other individual's impressions and value judgments, but this does not make them useless anymore than subjective reviews of restaurants, wines, movies, or music by others would be considered useless. It's just that subjective reviews should not be taken as "gospel" anymore than slogans like "bits are bits" or "perfect sound forever" should be taken as "gospel." "

The reviewer is listening to audio that is being read back out of local buffer. It's data that was most likely read into buffer anywhere from a few seconds to a minute beforehand.

Now if we could have de-biased evaluation....

whell's picture

Your system probably costs as much as my house. If I ever get to the point that I want to spend that much on my gear, then I'll care. Until then, not so much.

X