The findings here were very interesting. Also, you don’t need a trained ear to hear the differences. Once the switches were detached from the server – the sound immediately changes. So noise is always present – let’s find out by how much.
The Innuos server and Roon does get a little weird when you unplug the cable midway. Sometimes I’ll have to wait a few minutes for it to “refresh.” At times, it was very quick – and those were the times I did my listening.
When I unplugged the SOtM sNH-10G, the sound had less of an edge, a more solid midrange, better coherence, better focus, and just sound fuller. The tone itself didn’t change much – although slightly warmer. It’s not as “excited” and sounds smoother and more dynamic. When I plugged the SOtM back in, there’s a little bit more grain and the sound was a little thinner. Although it wasn’t night and day, the main giveaway is a more diffused midrange.
The differences were far larger with the ZyXEL. When unplugged, the sound was pretty much identical to what I heard when the SOtM was unplugged. Plugging the ZyXEL back in and you start to get an idea of just how much an attached switch could alter the sound of your music server. The ZyXEL is relatively dark sounding and has a much more veiled sound. It’s not nearly as transparent as the SOtM. Not even close, really.
From my experience, a local file almost always sounds better than a streamed file. So far, without exception. It’s always quieter, more detailed, and imparts better cohesiveness.
In short, the performance gap is widened even further when playing local files (with switches detached). It sounds crisper, quieter, more liquid, and more transparent. The soundstage gets deeper and the music just breathes better. Plugging either switch back in – and the sound gets a little fuzzier and diffused. This is not so much the case for the SOtM, but is extremely dramatic with the ZyXEL.
If the sound quality with the switch disconnected is our reference point, it’s clear that the SOtM sNH-10 Ethernet switch does a really good job at isolating noise. The bottom line is – the SOtM gets you WAY closer to what it sounds like without having a switch attached. With the ZyXEL attached, sonic performance takes an enormous hit.
To summarize, less noise meant:
1 April 2022, Wiltshire, England: Chord Company has applied its 38-year cable expertise to produce truly unique… Read More
From November 25 through December 25, 2021, Danacable will hold a Giving2021 sale on standard headphone products… Read More
ELEVATE YOUR COMPUTER AUDIO TO THE NEXT LEVEL. JCAT INTRODUCES OPTIMO NANO - POWER SUPPLY UPGRADE… Read More
So...I spent the last four months listening to the world's best linear power supplies. If… Read More
AudioWise Inc. announces the launch of PGGB.IO - PGGB PCM Remastering (https://pggb.io), cloud version… Read More
Undoubtedly, your amplifier could be major decider of the sound of your entire system.… Read More
This website uses cookies.
View Comments
Chris Conneker of Computer Audiophile concurs with your POV, as do many others.
There is an alternative to the wildly expensive SOtM switch.
Paul Pang offers two Audio Grade Switches. One @ $219.00 and the other at $1190.00
https://ppaproduct.blogspot.com/2015/07/audio-grade-switcher.html?m=1
Pang suggests that two of the $1190.00 switches takes fidelity to an even higher level.
A follow up review comparing the Pang modded switches to the SOtM and Cisco switches would be most welcome.
"MEASUREMENT MORONS"
Thank you!!!
You stated my my view to the letter, with 1 exception. I call them :"Flat Landers" rather than Measurement Morons.
As we know "Flat Landers" KNEW that the world was flat, in much the same way that audio naysayers KNOW that the things that they either can't afford or don't want to pay for cannot possibly change the sound.
We as humans are still the equivalent to infants in our knowledge.
Just because WE DON'T YET KNOW how to perform the appropriate measurements to satisfy the Measurement Morons, does not mean changes do not exist.
But here's what I do know any piece of equipment that doesn't have a signal to noise ratio of CD quality or better is a no go for me. Then start listening with Pet Sounds. More likely than not I'm done, the piece of equipment can't do it. By the time I get to my reference banjo and harmonica recordings I have eliminated nine out ten pieces of equipment.
And I can safely say a lot is known about what measurements are important.
A continuity tester is all I use for audio cables. The rest is in the listening.
I was wondering can this switch be deployed as a noise filter? due to circumstances I have to use a wireless bride to connect to my innuos so if I add the SOTM switch between my wireless bridge and the innuos, will it improve the sound quality?
Most definitely. I'm only saying that because I've tried using an ASUS bridge directly before. Sounds better with the SOtM.
What about going direct from router to Streamer/DAC without a switch?
You don’t buy such a switch because you need more ports, you do it because of its re-clocking and lower noise capacitors etc. Is it worth the price tag? -> you have to decide for yourself.
You, silly, silly human being. Thanks for the laughs!
Agreed. I have been in professional audio for 40 years, and this is easily the silliest article I’ve ever made the mistake of reading.
How can this be a audio switch, it doesn't even support IEEE802.1BA standard. Go lookup a Catalyst 9200, 3850, 9300 and get a real switch instead.
A rebranded cheap switch and call it audiophile doesn't make the product better, even if it costs 10 times more than normal.
Sorry, I really don't believe this review. And the speakers in the test are more like midrange too.
OK, you're probably a bachelor in engineering, but not in networks. :)
You obviously have a faulty music player. The fact that plugging a cable in can alter the sound of the player means that the players designers did a bad job of designing the EMI / RFI suppression system in the music player.
There is so much buffering going on in file playback that in a well designed system, only two factors are relevant, file integrity and sample clock performance. In a badly designed system, where shielding and interference suppression were disregarded, all sorts of things start to mess with the system.
My question is why you are barking up the wrong tree and kludging a "fix" that means paying snake-oil money to address a problem that should not exist in a well designed system.
You said it better than I could have, a well designed DAC should be immune to what is plugged into it. I suspect a lot of audiophile digital gear is indeed poorly engineered.
You guys obviously aren't audiophiles. Tell me...which DAC or music server/player is completely immune to noise? You wouldn't be able to point out a single damn product - because it doesn't exist. Even "well-designed" gear is susceptible. Also, the level of suppression matters and could adjust how one enjoys the sound. Which goes back to "If it measures well, it doesn't mean it'll sound good to everyone."
They all are susceptible, to a varying degree. Agreed. But the thing that matters is whether they will degrade their output as a result of it. "Audiophile" gear has a long history of not being designed to be robust and tolerant, interface-wise, probably because there are observable problems with constraining bandwidth in the analog domain. But this is digital data. If the bits that come in are identical to what is being sent from disk (and the CRC32 on the Ethernet frames pretty much guarantees that, together with the TCP checksumming and retransmission going on, for which we have buffering), the resulting waveform will not deviate as a function of the transfer.
In my experience, as a broadcast network engineer doing several orders of magnitude more complicated things than essentially copying a file over a short cable, (like transfering 100+ channels of real-time live sound 600km over dedicated networks and playing some of them back in surround, which of course requires path delay alignment in the microseconds range to be believable) the cable and the switch are negligable. As long as they perform to standard and lose no bits, and the switch supports the timing protocols. (This is of no concern to your test case, because you can buffer and will use your local sample clock oscillator steered to the values derived from metadata in the file. I am listening to my GPS-controlled OXCO putting IEEE1588-2008 packets on the wire, and using that as sample clock steering. Much harder. And measurable.)
Why am I going on about this? Because _everything_ is done to a budget. It simply does not make sense to start wandering off in these directions before the problems with the DAC are solved. Because, I'm 100% certain that that is where you'll find the root cause.
Paying large amounts of money for Ethernet gear simply does not make sense from a systems engineering point of view, not here, where no optics are involved, no distances are critical, and no bandwidths are even remotely constrained.
I'd start with getting a good oscillator for the DAC, one that is stable and has negligable jitter. Probably means that it needs to be an OXCO, i.e ovenized and separate. Or rubidium, steered by GPS, if you want long-term stability.
Oh, and don't forget, make sure you do not have ground loops running in the Ethernet cable screen. Potential equalization with short and low-impedance cables is very important; as is insensitivity to Pin 1 problems (Muncy, JAES, 1994).
This is /engineering/ -- it is not voodoo. It is treated as voodoo by people who earn large amounts of money from people who want to believe in it. But in the end, there are simple reasons why things work and not, especially so in digital transmission.
Put your money in record players and loudspeakers instead. That is where money makes a direct difference. The rest is simple. At least in comparison.
And compare double-blind. We have 100+ years of solid psychology research pointing to it being completely necessary to get anything remotely repeatably true out of subjective evaluation.
Thanks.
fantastic reply, Måns !!
Jay's response of "You guys obviously aren’t audiophiles." is the perfect red-flag for cult'ish, mob thinking over a quest for objective truth. In a time where lots of anti-science rhetoric is flying about, that and other monikers like "measurement morons" cast the audiophile community as a sort of voodoo religion.
For years, I've called myself an audiophile... I love music & I simply strive to re-create it to the best of my ability. That ability is a combination of budget, understanding & limits of my own perception. First, we need to acknowledge that this industry is FULL of snake-oil. Some is malicious, some is not... but where there's a buck to be made, you can guarantee somebody will be there waiting for a sucker. Nobody wants to be that sucker and most would never want to admit to it either.
Anyway, in the realm of analog hi-fi, I think there's still plenty of wiggle-room for some crazy-complex electrical interactions masquerading as some kind of witchcraft. From the DAC-forward, there's a complex system where each component up to & including the electromagnetic thing moving the air has it's own butterfly-effect on the net result.
I see lots of folks applying that same type of thinking to the even MORE complex digital system on the other side of the DAC... so many that really want to believe that the same principals apply to the packets, frames & serial-streams of binary data. And while it is a modulated signal that can have errors, those have been mitigated by the checksums, error-correcting protocols & buffers. Yes, if you're streaming audio via. UDP, it can contain glitches... but that's not how most of this media is streamed.... it's treated like files over a application-level protocol. Think about it... if you send a file from one computer to another over ethernet and it gets corrupted... something is very very wrong!!
Audiophile switches... *SMH* it's like somebody found a band of zealots out in the desert, desperately looking for a new god...
TCP checksumming? Make your homework!
Music- and Videostreaming does not use TCP/IP but UDP/IP and Google has even an own Protocol -> less overhead, but no checksumming or resending of lost packets. Therefore quality of components and cables matter even more or error correction has to fill the gaps, which will be audible when reaching a certain threshold. Digital is not fool proof, it can go wrong as well.
You are 100 % certain that his DAC is "bad". Have you measured his DAC or the same model ? Have you checked the manufacturer's website for specs ?
Maybe you should offer to replace the clock for him with a money back guarantee?
Otherwise why should he follow your guesses at his expense ?
I've worked in networking for 20 years, TCP-IP is a lossless protocol, if a packet gets corrupted, then it's re-sent... What switch you use has zero impact on the quality of your sound.
Following your theory,all content on the internet would suffer degradation every time is duplicated, and 90% of it would be corrupt - as as a "software developer" you should know that is crap, every file has a checksum, any corruption would be detected.
I'm not saying the packets have errors - they don't. Otherwise, it would be clearly audible. I'm saying the cable itself indeed changes the sound at the output due to how it modulates noise and interacts with the connected components.
That DAC does not exist.
Seth, You suspect without scientific proof ? WTH ?
Have to agree with you here. Although, with a tiny budget like his it is probable he might get bad results even if there were a decent component here and there in the signal chain. In my humble opinion, no point reviewing a switch before the rest is in order.
Do you have scientific proof that Mr. Bacon's music player is incorrectly designed ?
Thanks for the laughs! I applaud your ability to write absolute utter nonsense for 5 pages straight and still making it sound like an article.
Oh boy.... wait until someone flags the Master on this review --- the one and only Dr. Ethan Winer! Then the hell will break lose.
How does this compare to the TLS you reviewed in Sept 18?