Forums https://www.decware.com/cgi-bin/yabb22/YaBB.pl AUDIO FORUMS >> Digital >> Can Ethernet Cables Make a Difference? https://www.decware.com/cgi-bin/yabb22/YaBB.pl?num=1544551190 Message started by Lonely Raven on 12/11/18 at 17:59:50

Title: Can Ethernet Cables Make a Difference? Post by Lonely Raven on 12/11/18 at 17:59:50 An older article, but still good info and food for thought on digital, and network cabling. https://darko.audio/2014/12/global-feedback-can-ethernet-cables-make-a-difference/

Title: Re: Can Ethernet Cables Make a Difference? Post by Colin Booth on 12/16/18 at 07:51:51 (I hate doing this but: 15 years professional linux system administration and five years hobbiest of the same before that) I'm strongly in the "can, shouldn't" camp. Theoretically if your cable is so bad that the two computers have to re-transmit the data so often that you get a buffer underrun then yes, it'll have an impact. But assuming you are using a cable that can reliably handle the demands of basic home networking there should be literally zero impact on the end product. So the short of it is: don't use super garbage cables, don't have insanely long runs of unshielded cable (20+ meters), and don't do something stupid like wrap them around house power. If you're not interested in technical commentary about TCP/IP and why high-cost ethernet doesn't make any sense, stop here. Still here? Cool, lets rock! I'm probably going to end up being terminology heavy and definition light, so feel free to ask for clarification. While the author is correct that the transport medium (the cable) caries a pulse-modulated analog signal, the timing and error correction data built into each packet at the protocol level means that a busted packet will be dropped and resent. In essence, there are only two possibilities: the packet arrives unmolested, the packet is thrown out and retransmitted. Both of these events are entirely transparent to an end-user application which sees an unbroken stream of data. Again, and this is super important, IP is a datagram-based protocol which sends data in discrete chunks, TCP is a connection protocol sitting on top of IP doing quality control and managing successful delivery, and your operating system handles pulling the data out of the packets and reassembling them into a correctly ordered stream. Because of this layering, your application sees a continuous stream of data that by definition must look identical to the equivalent bytesteam generated and consumed locally. Before we go any further I have a brief aside: all of this only applies to TCP/IP based transmissions. If you use a protocol like UDP (also over IP) or IGMP (another internet layer protocol like IP) you do not get any retransmission guarantees, ordered reassembly guarantees, or any of that. Generally speaking though, point-to-point transmissions (like streaming music from your media server to your playback computer) don't need the speed benefits of UDP, and don't need the multicast functionality of IGMP. As an aside to the aside, the reason internet TV occasionally has graphics glitches is because internet TV uses IGMP as a transport and if a busted packet arrives there's nothing the protocol can do. So now that we've established that TCP provides delivery guarantees (regardless of the quality of the physical link that it is going over) it should be obvious that a packet switched TCP network (a category that includes every home network) is incapable of having data breakage in a way that degrades playback quality. This isn't to say that it can't have data breakage that makes things impossible to listen to, but that would be because the underlying network was hostile to all forms of data transfer, most likely making the connection unusable. So, bringing things back around to the beginning, as long as your cable runs are good enough to push the bandwidth needed to copy the data stream faster than your playback (roughly 1 MB/s for 192 khz 24 bit uncompressed audio, basically the peak of what anyone should be shoving across the wire) plus whatever other networking demands you have there will be zero difference between $1/foot and $100/foot cables as the protocol literally makes that impossible. Questions? Comments? EDIT: one note here, I'm talking about home networking situations where your media lives on one computer, your playback system is on another, and you're streaming either the file or the decoded PCM data across the wire (or DSD I guess, for the six of you with DSD media collections). I am not talking about anything using a multicast or UDP based broadcast system like internet radio does. Mostly because anything without delivery guarantees going across the internet is guaranteed to suck, and no amount of good-vs-amazing cable will make up for that (remember, TCP/IP was designed around delivery guarantees in unreliable environments). EDIT THE SECOND: Doing an objective test is easy: copy a file across your network from your file server to your playback system and then run a file hasher on them both (for mac/linux/bsd systems I suggest the command line sha1/sha1sum, for windows I'm sure you can find something) - the hashes should be identical. The subjective test is basically the same: play back the file over the network, then play back the file from the local copy - they should sound identical. If they hash to different values your network sucks, if they sound different but hash the same you're most likely imagining things as the files are essentially identical. (Yes I'm somewhat taking the piss here, but the whole thing is silly to folks with a compsci background)

Title: Re: Can Ethernet Cables Make a Difference? Post by Archie on 12/16/18 at 17:45:42 This is a debate that I don't get involved with since I don't listen digitally.  However, a lot of effort has been spent on both sides of the argument when "simple" double blind tests can put the whole thing to rest.   And just because results may not meet expectations (theory) they still ought to be the gold standard. But I also understand that the debate is sometimes 99% of the fun.   :)

Title: Re: Can Ethernet Cables Make a Difference? Post by Colin Booth on 12/17/18 at 04:41:56 The reason it bothers me so much is that you don't even need double blind tests to do it. You can literally do a byte-for-byte comparison of the data coming off the wire against the same data locally and know if you're getting the same results. Plus, the failure modes for any kind of digital transport are real obvious (dropouts, blips, etc) since it's a serialized bytestream as opposed to the whole spectrum of degradation that analog signals can have. It is true that some folks won't trust it unless you do a blind test, but transmitting across the wire into a digital frequency analyzer really should be all it takes. Generally speaking, the gold standard should be anything that can reliably transfer data at 100 Mbit or better, which means most short runs of cat5, and any run of cat6 or better where the wires aren't actively damaged. If it's good enough for home networking, it's good enough for any variety of data transfer: encoded audio, video, or whatever.

Title: Re: Can Ethernet Cables Make a Difference? Post by Archie on 12/17/18 at 15:34:36 The problem is that despite "byte-for-byte" perfection, some still hear a difference.  If that difference shows up in a statistically significant way through double blind tests, then there is a problem with the theory.  Otherwise it's a placebo.  Although even at that, some would argue that if we think it sounds better then it does.  It all eventually comes down to a personal experience -- even the being driven crazy by the "non-science" types.   ::)

Title: Re: Can Ethernet Cables Make a Difference? Post by ScottNC on 12/17/18 at 16:31:11 This is a great post. I have to say I couldn’t claim to be able to explain this to anyone, but it makes sense to me. It is an interesting discussion and there are some good sides brought to the table here and for me personally, I’m thinking my (modified somehow by Andrew at LGC) cable is doing its job because when I upgraded the next one, my USB upgrade it made a significant difference.  Thanks for posting, hoping to see more on this subject. Best, Scott

Title: Re: Can Ethernet Cables Make a Difference? Post by Colin Booth on 12/17/18 at 16:34:17 I'm not sure if anyone has done double-blind (or ideally ABX) testing of ethernet as a data transport in any sort of meaningful way. The link in the OP explicitly said that they were sighted tests. My feeling (backed up by 40 years of reliable data transfer in computer networking) says that spending money on well built data interconnects is a peace of mind thing - good cable made by a company that you trust means you aren't worried about equipment failure, and if $30 for a five-foot run is what that costs, then so be it. I agree that at the end of the day we experience the world in a subjective way, it's in fact one of the major reasons why I like vacuum tube amplification and Steve's approach to it - generally simple circuits driving tubes that change the sound in interesting and enjoyable ways. It's why I said above that people most likely perceive differences in sighted tests when comparing the free cable you get with a $50 router and a cable that costs as much as the router. It's also why music sounds better after a nice beer or a glass of scotch. For people who don't know how the digital sausage is made, a fancier cable subjectively might sound better, even when the playback computer sees the same data at the same rate across each cable (those people also live in a more magical world, something I occasionally envy, but alas my career has been built around grinding digital sausage). Scott: I can't reliably speak for USB audio in terms of exactly how interconnect quality can affect a signal (my own opinion is that if you have a cable that can reliably transmit a clean signal that the cable isn't the issue, but I also know that the USB audio spec is more complicated than ethernet). With ethernet (wired or not), there is no "TCP/IP audio spec", so audio data gets transferred identically to everything else, and the transport protocol has to be able to deal with out of order delivery, dropped and resent frames, and so on. As for the modification my guess is that it was removing the RJ/45 connectors and replacing them with ones that do better strain relief, not anything that will make your cable transmit better but should increase their life expectancy if you're jamming them up behind stuff. As for "doing their job" or not, I've never argued that more expensive cables aren't doing their job, I've been arguing that every cable that can pass a ratings test can do the job just as well.

Title: Re: Can Ethernet Cables Make a Difference? Post by Lonely Raven on 12/19/18 at 04:40:19 It's after 2am and I need to be up in a few hours, so I'll keep this brief. I agree with you Colin, when we're talking TCP/IP the data is going to get there, and get there accurately...otherwise the stockmarket would be a mess, right? That said, doing a hash comparison only proves the data got from sending computer to receiving computer - I'm not listening to packets, nor am I listening to my computer, I'm listening to my DAC. So what we need to look at and evaluate is what happens once those packets are hitting the 99 cent made in China network chip, what happens when it hands that data off to the DAC, and if a network cable can have any effect on this what so ever. Back in the day everyone was saying  USB cables can't possibly make a difference! My printer doesn't care if it's 99 cent cable or $400 audiophool cable...You're all delusional!! But then once people started looking at how the streaming audio formats work, how the bits are transmitted, how the bits have *timing* attached and how a cable can skew all of that and actually indeed change the sound...well then, we will just move the goal posts and it's all placebo because you can't possibly hear that!! What I'm getting at is that if people are hearing a difference in Network cables, and we *know* that it can't be lost data due to how TCP/IP works, then we are possibly missing something, or too focused on the wrong something(s); just like we were with USB before that. Let's stop hashing the same hash about checking file hashes (sorry, I couldn't resist LOL), because that hyper focusing on the fact that the network gets the data from point A to point B perfectly is just a distraction from looking at what *could* affect sound. Just to be clear, I agree 100% that the network gets the data there 100% reliably. If there is a sound difference, it's something else. Also, audio streaming via USB, TOSLINK, Coax, is completely different and has been objectively and subjectively proven that cables can make a difference. Man I hope that made sense, it's 2:40am and I'm not confident I'm coherent. LOL

Title: Re: Can Ethernet Cables Make a Difference? Post by StanleyG on 01/18/19 at 05:52:14 I don't like it but I switched out my basic (TP Link) ethernet switch for a TrendNet with apparently an aftermarket linear power supply.  What I don't like is there is an audible improvement... Okay, appears to be an improvement. I bought it as an 'Optical Starter Kit' from Small Green Computer in anticipation of purchasing a soon-to-be released Sonore opticalRendu. I have been mesmerized by the discussions on Computer Audiophile (now Audiophile Style) about it, Sonore's other upcoming optical products, Uptone Audio's upcoming EtherRegen, and the whole idea of femto clocks and phase noise affecting Sound Quality. Not optical SPDIF but optical ethernet which if bits are bits over copper ethernet what will they be over optical?  It all sounds better on the new switch which I don't even have the optical section going yet (4 copper, 1 optical connection on it) and that switch is just a commercial one except the upgraded power supply. It defies reason. https://audiophilestyle.com/forums/topic/55217-sonore-opticalrendu/?page=1 https://audiophilestyle.com/forums/topic/55216-sonore-systemoptique/?page=1 https://audiophilestyle.com/forums/topic/38968-etherregen-early-general-details-please-dont-ask-too-many-questions-yet/?page=1

Title: Re: Can Ethernet Cables Make a Difference? Post by Geno on 01/18/19 at 20:27:52 Thanks for the info Stan.  I was planning on purchasing the UltraRendu in the near future, but will wait to see what the reviews are on the new optical player first.

Title: Re: Can Ethernet Cables Make a Difference? Post by Lonely Raven on 01/18/19 at 20:30:49 Wait Computer Audiophile changed their name? Optical Ethernet Switches to improve sound quality! Crap, I feel like I'm behind in things. Especially if Uptone is doing something Ethernet related - that guy seems to have his head on straight and everything I've heard from him has surprised me. I guess it's time to go further down the rabbit hole and see what kind of stuff we'd been missing on the Ethernet side.

Title: Re: Can Ethernet Cables Make a Difference? Post by StanleyG on 01/18/19 at 21:59:01 Sounds like you are diving into the digital ocean, Geno.  Cool!  The micro- and ultraRendu both deliver excellent sound quality and I expect the optical- to sound a little better even.   Contrary to my earlier comments about all the players/streamers that I had used reading from attached drives, the -Rendu devices only receive files via ethernet (optical or copper).  Just wanted to clarify. The USB port on them are their outputs and there is no option for internal storage so I don't think of them as players exactly though I guess they essentially are.  They require third-party software to function, though several options are built-in or pre-installed. Plus they require a server of some kind to stream files to them, either your own files or over the web. Not difficult if you have a wired network in place, especially if you are planning to use Roon.  A little trickier to go wireless and open-source (free), but I did it (now I am running all wired and Roon as that is arguably better sound-wise). I wish I had this forum as a resource when I was spending hours of trial and error to figure some of this stuff out.  Still learning, call it a fun adventure!

Title: Re: Can Ethernet Cables Make a Difference? Post by Geno on 01/18/19 at 23:49:47 Yep. I’ve spent a lot of time on the Small Green Computer website. I think that I have an understanding of the server-player-dac connections. I also realize that I’ll have to use a tablet to navigate and control the music. I can live with that. It gets me away from playing files from my laptop. And it looks like with the new technology, there is a much cleaner sound. By the way, I see you have Heresy’s.  I had a pair of Forte’s for over 30 years😊

Title: Re: Can Ethernet Cables Make a Difference? Post by StanleyG on 01/19/19 at 03:21:08 I don't think I have thirty years left in me!

Title: Re: Can Ethernet Cables Make a Difference? Post by Melvin on 01/20/19 at 12:47:44 I've been following the recent Ethernet developments over on Audiophile Style as well .. bleeding-edge stuff. Over the past few months I felt the time was right for me to explore Ethernet/UPnP/NAS options. I'm not very comfortable with networking so I started slowly and, after replacing my Apple router with a mesh system, I tried out some UPnP and control point apps. Once I was more comfortable with this direction and liked what I was hearing I purchased a microRendu at a nicely discounted price. The refinement in sound quality and ease of operation has put a big smile on my face. My system has never sounded so good. Not sure if I'll go down the optical Ethernet path in the future but I will continue to keep an eye on it. Exciting times.

Title: Re: Can Ethernet Cables Make a Difference? Post by StanleyG on 01/26/19 at 05:10:32 So my second "Optical Starter Kit" from Small Green Computer arrived and so did 15m of optical cable from SF Cable. I put the two switches next to each other, left most of the cable coiled, and simulated my intended interim setup without moving anything else or fishing any cable behind the couch or to the other side of the room. Plug and play! My first impression was it's not quite as stunning a difference as was swapping in the better switch with linear power supply (see my earlier post), but 50 feet of optical ethernet is maybe a skosh quieter in the quiet places than the original 30 feet of copper alone.  Blacker blacks as they say in the hifi magazines. Could be placebo effect. The 30 feet of copper is still there between the server and the first switch (will change to 3 feet when all done), then the experimental 50 feet of optical, a second identical switch, and finally about .5 meters of AudioQuest Cinnamon ethernet before the ultraRendu. As I listen a coupla hours now, my impression (placebo or not) is expanding to, 'it just sounds better'.

Title: Re: Can Ethernet Cables Make a Difference? Post by GeriTheBiker on 04/09/24 at 04:27:17 I know, this is a hundred years old topic, probably no one is reading it anymore, but there is one thing seems to me was missed here. First of all, thanks Colin B for the Networking Lecture, pretty good explanation. I am also playing in Team Skeptical. The article on Darko says: "Why does an Antipodes DXe sound different to a MacMini when feeding, say, an AURALiC Vega?" Okay, so, how is this ensures that the two player transmits the same data??? How can we know that those things do not do something with the data internally, and changing the stream? Did he measure it? If yes, how? I think we can agree, that it is just physically impossible to have difference in sound, if the same bit-flow reaches the DAC. So before we go to the listening room, I really would like to see a measurement of the bit flow on both sides of the wire or devices that are compared. Or a measurement what reaches the DAC. You know, those stupid hashes... Also, there is one thing was not touched. What about streaming? We cannot even know what kind of cables the data goes through (probably after reading it from the storage it is converted to a fiber transmit, then the fabric of the storage, then some kind of light-to-electric conversion, then all kind of routers, then whatever-whatever... And the data is packaged multiple times for different physical transmitting medium. Way much more complex than what Colin described.) before reaching that 3feet of $200 Ethernet wire... I just cannot think how that short piece of wire would have impact on the sound quality, while transmitting digital data? Or when we stream from the Net, we should replace and save the expensive wire for later use? (yes, i am cynical, sorry...). I guess we should not even talk about WiFi... One more thing: How is that possible, that every segment of the of the Internet can connect perfectly using the $3 wires, and only the 3 feet audiophil playing part must have a $300 wire? So, unless I see some kind of measurement that shows difference in the data streams, I stick to the $3 wire. On the other hand, if you hear a difference between the wires, pls do not be offended by my opinion.

Title: Re: Can Ethernet Cables Make a Difference? Post by CAJames on 04/09/24 at 08:34:18 Here is a discussion from a couple years ago about streaming that might be interesting: https://www.decware.com/cgi-bin/yabb22/YaBB.pl?num=1633788897 I have a different take on how networks impact digital music than Colin B.

Title: Re: Can Ethernet Cables Make a Difference? Post by GeriTheBiker on 04/10/24 at 05:48:02 That fiber thread made me think if there could be a possible way the CatX cable adds some kind of noise? While I am still sure that the network data does not change, and you get the correct bit-stream on the receiving side as Colin B described, there still could be a way. So first I did a little home test, from my WD NAS that is connecting to a NetGear Switch, from there a Cat6 wire comes down to my room, then wall mount, then patch cable to a Thunderbolt HUB, then USB C interface to my notebook, and in a different USB connects to the DAC/AMP. I did not hear any difference between the song I played from the NAS and from locally. I even removed the connection to the dock when played the song locally. Is it conclusive? I don't think so... Let's forget the possibility that the difference is small, and my ears are not sensitive enough to hear the difference. But. My setup is certainly different than yours... Hence I cannot say that there could not be a difference in your case, and vica versa. That thread you sent made me thinking what could be the reason (if any) for the difference? The only explanation I could think of, and it does not have an impact on networking, is if there is some kind of electrical noise on the CatX cable, and it reaches the analog side of the DAC. I have no idea if this is possible? Does anyone have knowledge about it? If this, or something similar is the case, then it could differ even between DACs. But if the isolation with the fiber works, or makes a difference, then WiFi should have the same effect, as it also does not have electrical connection between the "server" and the "client/player". (Of course the connection has to be reliable enough to deliver the data on time...). Also, DAC makers could build their product to do the separation internally, and then this possible (?) issue would be avoided.

Forums » Powered by YaBB 2.2.2! YaBB © 2000-2008. All Rights Reserved.