Why (Ethernet) switches matter for high-end audio

Puma Cat

Well-known member
Joined
Apr 5, 2013
Messages
2,723
Location
East Bay, CA
Very informative and accurate video by Hans Beekhuyzen

Hans does an excellent job diagramming and explaining why switches at various price points and designs matter and have such a significant influence for music-streaming-based audio applications. In particular, he explains why crap el-cheapo "consumer-grade" switches sound so bad, and why really good audio-grade Ethernet switches provide such an improvement, e.g. their ability to mitigate noise factors, ground-plane noise, high-source impedance leakage current, phase noise and threshold jitter, which has a notable impact on timing of the digital conversion.

Why switches influence the sound quality - YouTube
 
I really was a non believer in audio grade switches until a moment I placed Edison Creation Silent OCXO Extreme into the chain. After 10 seconds i was converted fully and now I'm into this camp so much so that I ordered also his Fiber Box 2 which based on several buddies who I trust brings also significant uplift in sonics.
 
I really was a non believer in audio grade switches until a moment I placed Edison Creation Silent OCXO Extreme into the chain. After 10 seconds i was converted fully and now I'm into this camp so much so that I ordered also his Fiber Box 2 which based on several buddies who I trust brings also significant uplift in sonics.

Yeah, I've heard really good things about the Ediscreation switches.

The Fiber Box 2 should also be an upgrade as it will eliminate the impact of high-source impedance leakage current. Keep us updated...
 
Regarding Edison switch - really good buddy of mine as a first switch got one of the top Anszus models. It made profound diff in his rig. After that for his second system he got cheaper version of Edison switch which proved to make even bigger diff vs Anszus - go figure...

Will keep you updated once I pick up the Fiber Box - ah BTW that guy from above then took Fiber Box from Edison...
 
Regarding Edison switch - really good buddy of mine as a first switch got one of the top Anszus models. It made profound diff in his rig. After that for his second system he got cheaper version of Edison switch which proved to make even bigger diff vs Anszus - go figure...

Will keep you updated once I pick up the Fiber Box - ah BTW that guy from above then took Fiber Box from Edison...

The Edison switches are definitely nice. And as you mention, more expensive does not mean "better".

We've got some high end switches on the way and we're going to be doing a video on the Paul Pang QUAD switch in the next 2 weeks.
 
I have the Ediscreation basic switch and two Fibre box II’s .

Great products

My next upgrade is to move out one of the standard level FibreBoxs and order an extreme level Fibre box’s powered by Ediscreations top of the line Firebird separate LPS. This will go between my Innuos and my Auralic Aries G2.1

I like the incremental improvement the Ediscreation switch made but it’s my experience that if you put the Fibre Box in the right spot the improvements are some of the best dollar spent to sonic gains ratio available in your streaming network .

For the record, I have tried many other products before with varying degrees of success.
 
I also like a good switch and love my Bonn. A few days ago I watched this video and understand the part how digital can go wrong.

The question that is still open for me: Why do all those small timing errors happen only in the last switch and not in all the servers, my home Fritzbox router, and the long internet cables, which bring the internet music to my house?
 
I also like a good switch and love my Bonn. A few days ago I watched this video and understand the part how digital can go wrong.

The question that is still open for me: Why do all those small timing errors happen only in the last switch and not in all the servers, my home Fritzbox router, and the long internet cables, which bring the internet music to my house?

They don't only occur at the last device, they occur all along the way, including the router, the "upstream" server, in any "switch" connections (e.g. RJ45 interfaces) it may have, etc. Also, switch-mode power supplies that generally power these devices, adds leakage current and thus, threshold jitter and the cables can contribute common-mode noise. Also, remember it is only the "data", i.e., the music file, that is encoded in digital form. The actual signals are analog voltages. If they were "perfect", they would take the form of a perfect square wave, e.g. the signal transitioning from 0V to 2V "instantaneously" but nothing is perfect, and there is a time component that causes electrical voltage signal to be not perfectly square and on top of that, the noise components are overlaid on top of the square wave.

So it, actually looks like this:
Phase%20Noise%20pics.png


Have you read the white paper on this by John Swenson? I've linked to it at least half a dozen times on this forum.
 
They don't only occur at the last device, they occur all along the way, including the router, the "upstream" server, in any "switch" connections (e.g. RJ45 interfaces) it may have, etc. Also, switch-mode power supplies that generally power these devices, adds leakage current and thus, threshold jitter and the cables can contribute common-mode noise. Also, remember it is only the "data", i.e., the music file, that is encoded in digital form. The actual signals are analog voltages. If they were "perfect", they would take the form of a perfect square wave, e.g. the signal transitioning from 0V to 2V "instantaneously" but nothing is perfect, and there is a time component that causes electrical voltage signal to be not perfectly square and on top of that, the noise components are overlaid on top of the square wave.

So it, actually looks like this:
Phase%20Noise%20pics.png


Have you read the white paper on this by John Swenson? I've linked to it at least half a dozen times on this forum.

...and/or...

The timing issue could well be placed at the hands of the very most END device (the audiophile gear itself). Since there is no real "standard" that audio equipment manufacturers follow (unlike makers of Ethernet Network gear such as Cisco/Nexus, Juniper, et.al) rather they seem to choose their own Ethernet recipe for what is best for their design even down to how to decipher link-light-status at the RJ-45 or fiber ports. Not to mention, very little if any Audio gear to my knowledge allows a user to adjust duplex/speed settings or allow timing/clocking adjustments within a managed GUI. Just look at the various manufacturers of Ethernet connected "audio" gear or digital-doo-dads/filters/converters that have no standard on link-light-status colors for 10/100/1000 speeds or blink/no blink for "activity" or error rates.

One simple thing to do if you're having timing or physical errors in your audio network (or suspect such within your audio setup) is to simply unplug the END-MOST device (streamer, server, etc.) and plug your laptop computer into that cable. Go out and download something huge off the internet or start a stream of some sort and watch your Laptop NIC diags for any errors incurred. IF you see no errors counting up, then the issue is likely with communications to that Audiophile piece of kit/NIC. IF you see errors stacking up, then most certainly, look upstream one step at a time doing the same test in order to diagnose the problem and find the culprit. Linear troubleshooting in this manner can ferret out most issues.
 
Does the quality of a switch matter if one downloads music from Tidal / Qobuz for offline listening vs real time streaming?
 
Does the quality of a switch matter if one downloads music from Tidal / Qobuz for offline listening vs real time streaming?

If you send data via a network connection from a music file resident on a hard drive or flash drive that's connected to an "upstream" networked music server (e.g., a Roon Core, NAS, SonicTransporter, Pink Faun, etc.) via an Ethernet switch connecting the music server to a "downstream" network bridge, streamer, or streamer/DAC, then, yes, it matters. Even the winding of the cores of the isolation transformers in the RJ45 jacks matters, just as an example. These devices, e.g. an EtherREGEN or SOtM sNH-10G, Melco switch, etc. also "sound best" is they are "grounded" to a ground-plane noise reduction hub, e.g. an Altaira, or even the GP-NR ground post of a Shunyata power distributor.

Back before I had the (frickin' amazing) Lumin P1 and was using my Lampi Baltic 3, here's a shot of my ER connected via an Alpha CGC ground cable to the Altaira below. Connecting just the ER to the Altaira makes for BIG improvement on audio quality; it is not subtle. Many thanks to John Swenson, who's a helluva good Ethernet engineer, for putting a proper ground terminal on ER.

ER-Ground-Terminal.jpg
 
Sure glad my music files are stored on the music server itself to avoid all these issues with streaming.

This is probably also why locally saved standard 16/44.1 FLAC rips typically sound better than even "Hi Rez" files streamed from online.

Just think of all of the servers, hubs, repeaters, questionable connections, etc, etc a streamed file from Qobuz and Tidal have to go through before it enters your home. And that's not even taking into account your own internet/network gear it has to go through still.
 
Very informative and accurate video by Hans Beekhuyzen

Hans does an excellent job diagramming and explaining why switches at various price points and designs matter and have such a significant influence for music-streaming-based audio applications. In particular, he explains why crap el-cheapo "consumer-grade" switches sound so bad, and why really good audio-grade Ethernet switches provide such an improvement, e.g. their ability to mitigate noise factors, ground-plane noise, high-source impedance leakage current, phase noise and threshold jitter, which has a notable impact on timing of the digital conversion.

Why switches influence the sound quality - YouTube
The one thing I noticed about the video is that the author never tested anything. He goes on and on attempting to explain the theory behind his believes but does not actually test anything. I assume that he owns testing equipment (they show in the background) but does not actually use it (at least for the video). I was hoping that he would actually test several network switches and show why one was better than the other.
 
Trying to wrap my head around audiophile switches. Does it improve sound quality if streaming via wifi? e.g., Auralic recommends wifi vs ethernet.
 
Trying to wrap my head around audiophile switches. Does it improve sound quality if streaming via wifi? e.g., Auralic recommends wifi vs ethernet.

Via wifi? No. You need to be connected directly to the audiophile switch.


Sent from my iPhone using Tapatalk Pro
 
Via wifi? No. You need to be connected directly to the audiophile switch.


Sent from my iPhone using Tapatalk Pro

My room is quite a distance away from the router. How would I overcome this? Should I get a powerline adapter, then plug in an audiophile switch? Would the adapter introduce noise into the AC line?
 
My room is quite a distance away from the router. How would I overcome this? Should I get a powerline adapter, then plug in an audiophile switch? Would the adapter introduce noise into the AC line?

In order to answer your question, Tommy, we need to know your system components and how it is currently configured. For example, what are the components in the main rack where you listen and how do you get content from the source (e.g. a Roon Core, music server, etc) to the DAC and amplification componentry? How does your DAC connect to the "content" server? Via USB from a network bridge (a network bridge is just that, a "bridge" between your network and the DAC)? Also, when you say "quite a distance away", what is that distance, exactly?

As as example, my first streaming setup was comprised of a Sonore microRendu, which is a "network bridge". Content from my "music server" in the spare bedroom's Mac Mini was sent via WiFi to an Apple Airport Express near the main rack. From this Airport Express, I connected to the microRendu via an Ethernet cable at one end, and a USB cable from the other end of the microRendu to the USB input on my Gungnir DAC. The Sonore microRendu made for a significant improvement in audio quality compared to connecting a laptop or mini-computer directly to the DAC via a USB cable. Nowadays, there are lots of ways to do this, and I can safely say many folks have found that the best audio quality is obtained when connected directly from the "music server" computer or NAS (wherever it is located) to the network bridge or streamer via Ethernet cable or optical fiber (NOT TOSLINK), as Mike has alluded to, and not via Wi-Fi.

If you don't have Ethernet switches that support fiber (which has a number of advantages over copper Ethernet) to connect from the "upstream" switch to the "downstream" switch, then you're going to need a long run of Ethernet cable from the "remote server" (that connects to the Roon Core or NAS) to the main rack, as shown.
Ethernet-base-setup.jpg


This will work, but be very mindful that copper Ethernet can pass low-source and high-source impedance leakage current, which has a notable and audible effect on audio quality. This is why guys are finding that a run of optical fiber is really helpful, because the optical fiber can't pass leakage current (the signal is encoded and transmitted as light pulses, thus no leakage current) between the upstream music server and the downstream network bridge or streamer, and DAC.

Shown here is one way to use a run of fiber with "generic" fiber media convertors (FMCs). Bear in mind that these generic FMCs all use cheap-*ss SMPS (switch-mode power supplies), which should be replaced with Jameco Reliapro wall-wart style linear power supplies (which cost about $12/each) because the cheap-*ss SMPS put all sorts of nasty leakage current into the switches, and the problem that leakage current creates is threshold jitter, the impact of which is quite audible. These generic FMCs also use crap clocks, which contribute phase noise, the impact of which is also quite audible.
Generic-FMCs.jpg


Alternatively, if you have an EtherREGEN or SOtM s-NH-10G, each of which have SFP cages for use with optical transceivers, you can put the switch with fiber input 'downstream" near the main rack, as shown here. In this configuration, an audio grade FMC, the Sonore OpticalModule is used upstream, and the SFP cage (for the optical connection) of EtherREGEN is used "downstream", near the main audio rack. The Etheregen then connects to a network bridge, which serves to convert an Ethernet input to a USB output. EtherREGEN and the SOtM switches also utilize much better clocks than the sh*t clocks used in el-cheapo generic switches.
Optical-Setup.jpg


I've found this last set-up to provide both the best audio qualiity AND reliability of consistently being able to make a connection from the music server to the DAC. Also, bear in mind that the EtherREGEN has been discontinued, so if you want an audio-quality Ethernet switch that also supports connection via optical fiber, you'll need to get a SOtM s-NH-10G. Moroever, the original Sonore OpticalModule has also been discontinued, and replaced with the higher-specification OpticalModule Deluxe. Both the SOtM switch and the OM Deluxe will also require clean, quiet, linear power supplies.

Lastly, different optical transceivers sound different, as well. Lots of guys like the Finisar optical transceivers for making these optical connections.
 
Back
Top