How an Audiophile Network Switch Can(not) Improve the Sound (Ep:74) 

Here's the thing -- you might not've been totally lying, and this is something I'm going to make a video about. More often than not, if you play something once, then again, with no changes, people will perceive a difference. Why? Familiarity as far as I can tell. They are simply getting accustomed to it. So they actually ARE hearing a change.

@@dougschneider8243 I agree. The thing with music is there’s so much going on, you’re never focusing on the exact same part/instruments twice. And you can’t pause to scrutinize like with video .

@@HansenMath100% true, which makes these definitive judgements that so many make so suspect.

The whole "system is not revealing enough" comment is almost always a farce. I can easily shut it down the moment I list off the equipment I have within a few feet of me -- state-of-the-art electronics, speakers, etc. Plus, we always have something coming and going -- and I travel the world to some of the best hi-fi companies. Furthermore, we have a great electronics lab and we measure most of the speakers in the NRC's anechoic chamber. But even apart from that, put on a good set of headphones and play a music file through any decent headphone amp. You're going to hear EVERYTHING that's possible to hear. It's nonsense and, yeah, I'm sick of tired of hearing that excuse.

Great post!

Thanks for your detailed response. I'm really happy to read it. Whenever I make something a little bit technical, I always know I run the risk of someone smashing back. Your justification for building the power supplies is perfect -- because you "can." That's great. About dedicated lines -- these ARE a thing. In fact, in our lab, all amp measurements are done on a 20A dedicated circuit. The power-output measurements are greatly improved over a 15A shared circuit. This is actually the #1 upgrade, as opposed to, say, a power conditioner.

Excellent way of putting it in in true perspective. Thanks 👍

I guess nobody says a smpsu+cheap switch makes the Stones sound like the Beatles.. we are talking not about function but a relatively small difference. And btw the thing about smpsus is not their problematic DC output (although there are differences) but their hash into the power net if (if!!) picked up by a sensitive audio device. In an all SMPSU powered hifi system with D-class amps I would assume that hash is no issue..

Let us know!

I'll be interested.

"hi-end LPS from well known manufacturers" - why buy, you can make your own, basic skills needed.

I think the point you bring up about "issues" is a good one. Why put in a power conditioner if there are no power issues? I don't believe I have any power issues in my house, but I do know I had them when I lived on the 22nd floor of an apartment building. Sometimes I'd have almost brownout conditions. That's why dedicated powerlines are a very good thing. But internet switches, I think a lot more work needs to be done with those.

Great points

Glad it was helpful!

Thanks! Happy that you found it sane.

For the most part, I agree. I'd change my mind if someone can demonstration the effect(s), mind you. That said, right now I feel people are buying based on dubious reviewer stories and misinformation.

It's not easy to explain logic and facts, is it? That said, they always fall back on the "noise" issue. Well, "burning in" or whatever you want to call it doesn't affect noise.

just one hell of a good video. this is why i subscribed

We are happy you found it useful!

@@WeeWeeJumbo Very glad to hear!

Packets get sent and resent all the time. An audiophile switch is not going to make a single bit of difference about that. Furthermore, all this talk about "processors" is what fueled the FLAC vs. WAV debates. Processing happens in computers -- and sometimes processors work more or less. That's just part of the life of a computer. As for the ridiculous notion that this can affect the sound quality of a FLAC file has been pointed out by me and by others -- the most time-intensive tasks on a computer are I/O tasks, such as reading from a drive. With a FLAC file, there's a lot less to read. But the who processor thing is, nevertheless, a red herring.

@@dougschneider8243 I meant the DAC as a processor. If data with the wrong checksum is thrown out and data with the correct checksum is buffered in the stream there must be some effect.

@@kenwong9796Hi, I think this is rarely if ever an issue. Honestly, the internet wouldn't work the way it does if we had these errors getting through. You watched my video without a problem. Passing music data around is NOT a problem. And that's why we have buffers. Furthermore, jitter reclocking ONLY matters right at the point of conversion. Right there. I made a video about that here: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-xwAspcxqYmQ.htmlsi=vva-Y3lp-8U1ML3N

A bit late but... My FLAC files give less than 1GB for an album. That can be transferred on 1Gigabit network in less than 20secs, say. Resending packets is rare, but even so, not an issue. Overhead is nothing compared to all the other stuff going on. Do you watch streamed movies? How often is there a glitch? And video is more demanding than audio.

Compared with the number of packet resends that happen before streamed audio gets to your router, how important do you think that last step will be in the whole chain ? Especially since the distance is almost zero, so the chance of getting a access conflict or a transfer error is also close to zero. The very funny thing is, some of these high-end audiophile switches advertise that their switch uses especially slower chips hence transfer rates (i.e. 100mbit/s in stead of ≥ 1Gbps) , to avoid creating noise ... If it wasn't so sad, you'd be rolling on the floor ... A faster processor is capable of handling any resend/error with even more ease then a slower one. And regarding noise : putting your audio on a different power circuit might be useful (I think this is one of the few things that really make a difference, besides working balanced wherever possible), and you might consider optical connections, although - if digital noise is your concern - one shouldn't underestimate the potential for noise-induction of an sfp module (besides the digital circuit everyone contains LC lowpass filter ...)

But you'll be a happy person if you see this before.

What kind of digital source(s) is plugged into it?

@@dougschneider8243 Right now I am using Bryston BDA-3.

@@dougschneider8243 I am using a Bryston BDA-3 DAC fed via USB by a SOtM Trifecta. The SOtM gear is being fed by the UpTone EtherREGEN. I sounds very nice. It is a huge improvement over feeding the BDA-3 DAC directly via USB from my windows laptop. Night and Day.

Thank you for the feedback!

Providing the CD and hi-res are created from the same master file, yes. Many times, the differences people are hearing are different masterings of recordings.

All ears -- go ahead and demonstrate the measurements that show this. But also, know the measurements have to show the impact on the converter.

​​@@dougschneider8243My claim is about reducing network latency and jitter with modification. If you understand about sampling jitter and how reducing network jitter can reduce sampling jitter in da conversion, that may work I guess?

Unless you're streaming UDP or have so much packet loss you're running out of buffer network latency and jitter will have absolutely zero impact on audio quality

@@dougschneider8243 I'm intrigued but I'm not sure if I can take wave of comments from unreasonable people.

@@v0ldy54 Yamaha concluded that cable and network also affect jitter.

All cars have to provide safety for their passengers in case of an accident, which is enforced by the law. Reality is that in some cars you can get easily killed, while other hi-end cars protect your life much better.

@@andru2625 In case you haven't noticed we are not talking about cars. Not all metaphors are valid.. There are not many ways to provide galvanic isolation on ethernet ports. You do it either through transformers or through optical conversion. If a device does not provide optical conversion then no matter how audiophile it purports to be and how much money it asks from you, it is doing no more than your run-of-the-mill network switch.

This is all very easy to explain. But let me first say that I never said that a transport, say, can't improve the sound quality. But a lot of that depends upon the quality of the DAC design. Some DACs have very poor jitter rejection, for example, so they can be influenced by the connecting device. Same with noise -- you have a poorly designed digital receiver susceptible to noise, well, that could then pass through. This can all show up in measurements. But I used an EMM Labs DAC, which has outstanding jitter rejection (it completely regenerates the signal) and equally outstanding noise rejection, so it's basically impervious to these things. But with regards to network switches, again, "jitter" isn't an issue when it's passing around the data stream that isn't even time correlated at that point in the signal chain. As someone pointed out elsewhere, inherent in switch design at any price is galvanic isolation to deal with things like noise. Regarding Hans Beekhuyzen's videos, I see that's good that he's backtracking on what he previously claimed because he came onto my Facebook wall several months ago spouting BS about network switches and jitter. I mentioned to him there that he should look into packeted data transmission and what that means. That said, these things are all measurable, but I think what should be taken just as seriously is the listening tests -- in particular, are people really hearing differences or if they are imagining them. I think that you'll see that in not too long of a time, many people will realize that their expensive network switches are a complete waste of money and they won't be able to sell them -- and they would probably be better off buying a better-designed DAC instead.

@@dougschneider8243 You are very logical when you talk about the streamer (digital transport) and how it affects the DAC. You don´t seem to need measurements to believe in them? I don´t understand why you don´t apply the exact same logic when you talk about another digital transport (switch)? It is the exact same thing. Just because one of them is not a classical HiFi component doesn´t mean it still does the exact same thing, from a noise reduction perspective. You need to forget about networking and data, it has absolutely nothing to do with that. The noise travel on the digital circuits, on its ground plane, and eventually turns into jitter since it affects the clocking. (If i understood Hans B´s explanation correctly)

@@tobiaskervefors8496 You seem to have lost me on these comments. I *do* believe in measurements strongly -- that's why we have invested so much in our network lab and also doing measurements in a true anechoic chamber. But as far as forgetting "about networking and data," that would be a BIG mistake. Networking and the way data travels along these networks is what streaming is relying upon to work properly. There's not magic there -- they're using the existing internet infrastructure.

@@dougschneider8243 ok, but which part in Hans two videos is it that is unclear? He has put a lot of effort in explaining the exact logic behind why network switches impacts the DAC to sound better. I had to watch it several times before i understood, but it really is all in there! I get feeling that you haven´t actually watched closely what he is saying?

The guy at Alpha Audio (Jaap) is a cisco certified network architect that didn´t believe in this, until he clearly heard the impact himself. You can even hear it yourself on their live streams when they test different switches. He has worked really hard to try to understand why switches impacts the sound and they have invested in really good measurement equipment and you can watch them measure the switches, on the DAC side, in "live" streams. They are really trying to answer your questions and give you objective data. Of course they get annoyed if getting questioned when they are doing everything they can to help explain for us all, also with measurements.

Ones marketing specifically to audiophiles.

A shiny box...LOL !

Something with a price tag of at least 15 times the actual worth of the product

I'm going to push back just a little bit. I've been using Wi-Fi off and on since about 2007 (when I got a Slim Devices Transporter). It can work exceedingly well. The only caveat is the implementation. If it's poorly implemented in the receiving circuit, it can generate the problems you'd expect, like noise. But if it's well implemented, it cuts the electrical connection between components and can work extremely well.

@@dougschneider8243 are there any receiving units you'd recommend?

@@dougschneider8243 How does one know if the receiving unit has bad implementation? I notice a few high end streamers now are gutting the wifi altogether to let the consumer figure it out - is there a quality extender you know of?

@@mat.b. Measurements can help tell the story. Unfortunately, it's not possible to measure every component out there.

"As for all the talk about A/B double blind test … many want that but they don’t do that when they audition or review equipment." Most people don't, but if you really want to suss out if you are truly hearing a difference, it could be good to do it.

@@dougschneider8243… that I do not disagree. I can’t.

Yes, it changes it -- and therefore breaks the electrical connection -- but on the receiving end, you have all kinds of electronics again to receive and decode it. So is that better or worse?

It would be weird for that to happen, but it's also possible that the "audiophile designer" who created the switch degraded the performance of it such that it wasn't passing the data properly anymore. As I said, without doubt, all these high-priced network switches are simply modifications of existing mass-market ones.

Wow! Some very lofty claims in there! I particularly like how the solder will affect the bass -- on a box that transmits digital data.

Or not. That's just not how data works.

The data is buffered inside the DAC or DSP and has no knowledge of how it got there so you must be imagining an improvement. Also all ethernet ports are isolated through transformers etc so the claims are bogus ;)

You should try audiophile grade oxygen free air now. The effect is drop dead magnificent.

Perhaps you can inform me. When I created this video, I saw no measurements showing the output from any DACs. If that's changed, can you tell me where they are? As for the "perfect" measurements that you're talking about, I've never seen a product with perfect measurements, nor have I seen a product where the measurements are identical to another, meaning they couldn't sound difference. There are always differences.

Hi Doug, you might want to check out the measurements here: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-dstO0650oKk.html As to my comments on "perfect" measuring DACs using the standard set of audio measurements, go the Audio Science Review website and check their measurements of a Topping D-90 vs a Mola Mola Tambaqui DAC. I put perfect in quotes because, of course, no audio components actually measures perfectly, there is always going to be some element of distortion and noise. While amplifiers and speakers are never going to measure "perfectly", we now have DACs which do. By "perfect" I mean a standard measurement set where all the distortion and noise products are at levels which are non-audible in an audio system playing in a a room. For example, no one is going to hear distortion products at -120 dB in any audio system playing in a room. We have DACs now which measure at these levels, and hence are essentially "perfect" as to any audible noise or distortion on their outputs, and the Tambaqui and D-90 are among these "perfect" DACs. Now there are those who claim these two DACs could not sound any different from one another because the measurements show they are "perfect"! I am not among those persons, and have heard both DACs, and clearly the Tambaqui is another class when it comes to sonics from the D-90, despite that fact that they both measure "perfectly". The point which I was trying to make, is that the standard set of audio measurements is not entirely adequate for determining the performance of an audio component. I do not believe it is impossible to measure everything, but I do believe we do not measure everything, or at least the standard set of audio measurements do not measure everything which matters for sound quality. Hence: it is possible for Network switches to make for a sonic difference which does not show up in the analog output of a connected DAC via the standard set of measurements. I work with a company which develops, manufactures, and distributes Network connected Hi End audio products, (we do not make a switch). And the things we have found which matter to sound quality continually surprise us (we do own an AP and other sophisticated test gear). Some of these things are currently measurable by us, and some are not. For example, most exasperating for me is the effect a better Ethernet clock has on sound quality-it makes an astonishing difference, certainly of a level which is meaningful to audiophiles. Why, I cannot say for sure, one of our digital engineers has a theory (currently unproven) of why, which he is currently trying to find a measurement for, to allow us to quantify these differences-right now, all I know for sure, is that a clock with lower close in phase for the Ethernet clock, does improve sound quality in my testing, and expectation bias before this test was to assume that it would make no difference! As a professional reviewer, I generally respect your work, but in this area it appears that you may have made some assumptions without actually testing things to hear differences for yourself. I would suggest a more open mind might be in order before claiming it is "impossible" for things like Ethernet clock performance to matter to sound quality.

@@barrowsworm1226 Audio Science Review is hardly a voice of authority in hi-fi design -- and they certainly don't speak to me. That said, in that video, can you please tell me where he's showing how these switches affect the output of the DAC?

I am no big fan of ASR either, they are totally closed minded and kind of ridiculous (and they are apparently deaf, haha). But they have published relevant measurements for quite a few products, and those measurements are accurate now that Amir has learned how to actually use his AP (he made some mistakes in the beginning). I am only referring you to them as a source of the measurements I was talking about. Did you actually watch the video I linked??? They demonstrate, specifically, that higher noise on the Ethernet input to streamer (the streamer under test has a DAC as well) results in higher jitter on the analog output of the DAC. That is what they tested.@@dougschneider8243

@@barrowsworm1226So just for you, I watched it all again. Now, unless I'm missing something completely, there are no analog measurements. Instead, he's attached probes somewhere near the clock and he appears to be analyzing its output. That's not the analog output of the DAC. Correct me if I'm wrong and he's doing that somewhere I haven't noticed. More important is that he raises more questions than answers. One question I have is if his test is even valid. He seems to have concocted something and is producing measured outputs, but is this actually showing anything? I'm not going to pretend to answer that, but I did send this video to a couple of digital designers to see if, in fact, it is showing something relevant. And another question I have is why he isolated this test to just one DAC. Surely, a review publication has multiple DACs on hand -- I have five within my line of sight. Why, if he's proving that switches make a difference, didn't he repeat this across multiple DACs to see if they all behave the same? All told, I'm not sure what exactly he is showing -- and if it means anything.

I hope this doesn't sound too harsh, but this bit of nonsense about a "revealing enough system" comes up often. In my own listening rooms I have speakers from Revel, Estelon, Vivid Audio, and more. I have electronics from EMM Labs, Moon by Simaudio, Anthem, Purifi Audio, and more. This is not only some of the best equipment in the world, the products have been measured in our labs and we know it's all working correctly. What's more, that's just me -- we have numerous writers with equipment that's just as good and "revealing."

what type of "revealing"? single 8 inch widerange driven by tubes? or something like TOTL Genelecs, PSI, Neumann?

@@AudriusN HiFiMan HE1000se or better. HoloAudio Bliss. Mainly HQPlayer to DSD.

@@b00m3rh4nd_sol such a scary words, but you still don't know how networks work. Run a wireshark on that switch and where and which packets/frames are labeled as audio. good luck, dare you.

And that's the gist of my video!

No one claims that the data is altered or lost in any way, the damage in Ethernet connected audio is due to noise coupling into the DAC, and then affecting the DAC clock, and DAC analog output stage.

@@barrowsworm1226 bollocks. Google doesn't use any of that nonsense in their data centers yet video you watch is uncorrupted.

I believe if you revisit Hans Beekhuyzen's channel, his thinking has "switched" on switches. In years past, he proclaimed about reclocking. Shortly after my video came out explaining why, when data is transferred around the internet and through switches, there's no embedded clock in the data stream that matters, he put out a video saying clocking through switches doesn't matter. I could be wrong, but check it out.

@@soundstagenetwork I'll check out his channel now, cheers

Ears can be extremely valuable to gauge sound quality -- that is, of course, what matters most. It's what we hear. But listening tests themselves are tremendously unreliable, so to use your ears, you have to set up listening tests properly to make sure you're not fooling yourself.

@@dougschneider8243 So why are you testing audio streamers (transports) and give opinion on how they sound, without measurements to back that claim up ? I don´t get why you are cherry picking and review audio streamer transports (that is still only digital data, one´s and zero´s) but then suddenly refuse to believe in your own ears when reviewing switches. It makes no sense what so ever.

Some switches work (also) at OSI level 3, handling packets (in addition frames). I don't think we need to go into that level of "correctness " to make the point he's making . It's already hard enough to convince audiophiles the last meter of power (cable) cannot make a difference if the house power wiring is just using ordinary 2,5 mm cable ... (Let's not even start about the influence a starting washing machine would have when on the same circuit 🙄)

come on, flaming the argument just because of a very tiny terminology difference, considering the arguments are for non-technical non computer engineering people is too much

I often think that when we publish a video that we should put a timer on to see how long it takes someone to say that a system isn't revealing enough. Did you know that we have more than 20 writers and, on almost a daily basis, equipment from all over the world comes through our doors? I personally have access to the least and most expensive audio equipment on the planet. Furthermore, we measure most of it in state-of-the-art facilities. Revealing, huh? Try to find anything more revealing at a given time than the equipment that's around here. Sorry, try another tactic.

It is not about tactic, but just about experience in serious listening. And if I can hear things changing, when I insert different switches in my simple system, and you say there is no difference for you , I just can assume your gear does not reveal a difference. And since I an not the only one HEARING a difference; I cant say it is true what you write here. Thats it. Simple as that. No tactics, just listening carefully. @@dougschneider8243