Article: SOtM sNH-10G Network Switch Review

[plissken]

That's not a parallel setup. It's serial.  I've done ad-hoc runs were we need to get to something 800 feet away and using two switches to do it where fiber wasn't an option.

 

The frames get rebuilt 100% from the ground up as the go from port to port.

 

So with the play back machine being Windows 10 but I don't see the playback software other than Tidal or Qobuz I'll mention this:

 

When I went out to Denver years ago to a roughly $50,000 setup, the participant, using Tidal, couldn't tell when the customer $1500 12 cable or the 100 foot $14 cable was in use.

 

Another tidbit: They didn't even know they were listening to entire tracks from Tidal where I removed the cable 30 seconds in and let entire 5 to 11 minute tracks play.

 

I call shenanigans on this.

 

 

[plissken]

1 hour ago, The Computer Audiophile said:

Their native language and country of origin don't matter in the least.

 

Hey Chris, what does matter then?

 

My personal opinion is this review lacks any integrity knowing what I know about non-realtime audio reproduction, i.e. you can pull the plug and the music still plays.

 

I would love to see the article writer or SoTM sit blind and using Tidal on Windows 10 tell me during playback when the cable is in the computer or not. Do you think we could arrange this some time?

[plissken]

42 minutes ago, STC said:

 

I asked the same and told it could. 

 

 

I've implemented both AVB and AES 67 and I read through the 7 replies.

 

Yamamoto explained what AVB does for you. Sound quality was NEVER mentioned.

 

Protocols like AVB and AES 67 guarantee a quality of service for audio applications that have realtime needs that are Ethernet connected. These protocols have no bearing on reproduction of but only the production of audio.

 

[plissken]

1 hour ago, The Computer Audiophile said:

That said, once in a while we publish articles that are controversial. 

 

Let's talk about an article that talks about why it can't affect audio playback.

1 hour ago, The Computer Audiophile said:

 

One of the huge benefits of an uncensored comment section is reading the opinions of everyone who is respectful while providing opinions and / or factual information. 

 

I think a larger benefit would be a piece that presents the flip side with some actual data.

[plissken]

1 hour ago, incus said:

All subjective experience is bunk?

 

Nope, just opinion and uncorroborated.

[plissken]

7 minutes ago, incus said:

Actually pretty corroborated if you read up on switches.

 

And as for opinion, well it's all "just" opinion, isn't it? One opinion says only currently-existing measurement devices tell the truth.

 

I'm willing to bet you money you fall apart in a single blind evaluation.

 

 

 

[plissken]

46 minutes ago, incus said:

You are an external master-clocker, I see. So you know what this can bring. Obviously lower jitter, lower phase noise, shunting of high frequency incursions, etc. all have their impact.

 

I dare you to answer this one simple question:

 

What happens to all of what you posted when the Ethernet cable is pulled and the music plays on?

[plissken]

10 hours ago, LowMidHigh said:

 

When I stream using Airplay, the music stops. 

 

In general, the internet orange light flashes repeatedly, so network traffic is ongoing. 

 

Does that help you understand? 

 

The 'review' mentioned Tidal explicitly.  Tidal will cache the entire track locally.

 

So AGAIN, when using this SoTM switch, using Tidal (which was what was used, try and keep up here, reading skills are important), you pull the plug and the music still plays what do you think happens to the sound quality?

[plissken]

10 hours ago, thyname said:

 

‘You keep repeating the same thing over and over again, for years, in hundreds of forums and threads. Eureka!

What's even funnier is that over the years the subjectivist won't answer it because the know that their position on Ethernet cable and switches won't hold up.

 

 

[plissken]

1 minute ago, thyname said:

 

What's even funnier is that you got pretty elaborate replies, with numbers, data, rationale, and everything many times in the past, and yet, you choose to disregard and ignore any of them.

 

It goes like this:

 

Other people: X cable, Y switch, makes a difference for me.

 

You: BS! you must be hallucinating. Pull the Ethernet cable out! Does the sound continue? Yeah!

 

Links to your unicorn post's of 'What's even funnier is that you got pretty elaborate replies, with numbers, data, rationale, and everything'

 

Cause what you just said certainly hasn't happened.

[plissken]

16 minutes ago, The Computer Audiophile said:

My guess is the data has already traversed the noisy switch and unplugging just to listen to one track isn’t practical. Thus, keeping a source of noise connected all the time is a requirement. 

 

You should be able to tell WHEN the cable is removed because the sound should improve. A cable that is unplugged during playback can just as easily be plugged back in during playback.

 

If you can't tell which is which you have a quandary as a subjectivist:

 

Let's walk through the guess you presented:

 

'data has already traversed the noisy switch'

 

Using that we can say that the ultimate fidelity at that point and time is when you unplug the Ethernet cable. That that portion of the playback chain can not be improved upon.

 

So if you have a $90 switch feeding your playback stack, and you have someone randomly unplug and plug the Ethernet cable and you can't tell when that was, then logic dictates that a $1500 switch, or two of them, can't improve your outcome.

 

I've offered before to setup a multi-homed playback computer. I can even do multiple switches and do a HSRP configuration up stream. This will allow full on comparison of a $1800 setup ($1500 switch and $300 cable) and and $100 setup ($90 switch and $10 cable) in real time.

 

[plissken]

17 minutes ago, charlesphoto said:

 

This question never made sense: if the ethernet cable has transmitted ‘noise’ or jitter, etc to the streamer/DAC,

 

Ethernet cables transfer data. They don't transfer jitter. They can carry noise.

 

 

17 minutes ago, charlesphoto said:

and then the dac/streamer plays that in it’s buffer, won’t it have the same ‘sound’ as the just pulled cable if the buffer continues to play? Alwasy seemed like a straw argument to me...

 

It only seems like a straw argument until it's realized that buffers only store data. They can't store jitter, they can't store noise.

 

17 minutes ago, charlesphoto said:

 

Anyway, just go get a used Cisco 2960 Catalyst (fanless 8, 12.or 16 port models) for anywhere from $25-$200, and have lots of $ left over for gear that will make a bigger difference like amp, speakers, etc. The switch ‘sounds’ better than any consumer ones I used, and with a decent built in SMPS no worrying over power supplies. 

 

I have at the moment: 8 2960, 4 3750V2, 2 4000 Nexus, 2 3725, some Fortinet, Kemp, and Palo Alto gear sitting here.

[plissken]

3 minutes ago, thyname said:

 

LOL!! two people already just told you:

 

 

 

The question remains unanswered:

 

What happens to the SQ when the cable is pulled. It's a simple question that yet remains unanswered.

 

Those responses are conjecture. I'm asking what actually happens WHEN they do it.

 

It's why I offer $2000 to someones $500 for this to be done in a controlled environment.

[plissken]

18 minutes ago, The Computer Audiophile said:

This is something we should look at more. 

 

Ethernet cables transfer whatever is in the packets. For IP telephony jitter is a huge issue because the tolerances are so much worse than audio. It seems like there is plenty of jitter involved with Ethernet.

 

I'm not following you about Ethernet not transferring jitter. Can you provide more details?

 

Jitter is simply this: Take a signal with a certain specified clock rate like 125Mhz and and allow that rate to +/- a bit (skew). That is a form of jitter.

 

For one thing a certain amount is allowed. For a second we have input and output buffers and most sit between clock domain boundaries. Buffers eliminate jitter 100% guaranteed.

 

The packets remain unaffected by jitter once they've hit a buffer. The final one being in the DAC where the audio clock rate is applied.

 

Pulling the Ethernet cable, technically, introduces the worst kind of jitter imaginable. Many high end streamers have enough buffer to hold 30-240 seconds of audio (format dependent).

 

Jitter is never stored, It's never part of the data.

[plissken]

26 minutes ago, Jud said:

 

Yes, exactly.  What happens if/when the noise gets into the DAC chip and clock?

 

Then a dipshit designed the DAC.

[plissken]

29 minutes ago, Jud said:

 

Yes, exactly.  What happens if/when the noise gets into the DAC chip and clock?  (I recall at least one ESS white paper with info on this.)  It can cause jitter.  And/or it can get into the analog side of the system through ground.  Now I'm just speaking conceptually, and not about whether the levels of this stuff would be audible.

 

You will see 60hz mains noise creep it's way over Ethernet cables, even making it past the magnetic isolation transformers. This is why I'm a proponent of WiFi over either copper or optical Ethernet.

 

If you must go wired and you are wanting quality components: Get a used Cisco Catalyst 2960 8T for $30 and an Intel Server NIC for $20 if using a computer.

 

WiFi is the defacto, highest fidelity, connection available to the audiophile.

[plissken]

30 minutes ago, The Computer Audiophile said:

That's a very good point. If one cause of sonic change is noise and that noise is coming from upstream of the DAC Ethernet interface, one should be able to notice when the cable is unplugged.

 

That's all I'm proposing here. If you start Tidal and playback your favorite track and have someone randomly disconnect/connect the cable and you can't tell any difference there is literally no amount of money you can spend on the portion of your playback chain to improve things.

 

$1500 for a 'audiophile' switch is just insanity.

[plissken]

35 minutes ago, The Computer Audiophile said:

I've also heard about measurements with and without Ethernet cables plugged into a DAC and being able to see differences on the analog output. I need to dig further into this to see if these differences matter and what they actually show.

 

That is certainly possible. You could even take a switch, desolder the the isolation transformer, zip up the traces together and hear all manner of shit. But the bottom line is that the isolation devices available already do a killer job:

 

Their job is to block noise and they do a good job of it already in equipment measured in the $10's of dollars. Not the $1000's when it comes to networking equipment.

[plissken]

12 minutes ago, Jud said:

 

Noise is always going to be present, the only question is how much (how effectively the design protects against that noise).  Lots of methods of isolation from transmitted noise are noisy themselves, so complete isolation isn't the complete answer either.

 

More is better than less Jud. WiFi certainly eliminates AC hum/ground loop.

[plissken]

7 minutes ago, The Computer Audiophile said:

I like optical because one gets the reliability of wired and the isolation of wireless. 

 

 

 

Takes the same effort to run an optical cable as it does Ethernet 🙂

 

Also the $30 2960 8t has an SFP port on it.

[plissken]

2 minutes ago, Jud said:

 

Optical is a lovely idea.  But just like any other isolation method, the question is what level of noise the isolation method itself creates (in the case of optical, the receiver circuitry to convert the optical signal back to an electrical one).

I had a white paper at one point comparing the power supply rail characteristics and optical was something like 10^3 better in consumption than WiFi. So that's a win.

[plissken]

11 minutes ago, incus said:

1) I still don't understand the experiment you are asking us to do. Where in the chain am I meant to pull the ethernet cable?

 

In the context of this review using a Windows 10 Computer and Tidal with the switch upstream of the computer, where would you think?

11 minutes ago, incus said:

 

 It is my understanding that no current device can eliminate everything from upstream.

 

WiFi and Optical most certainly can.

11 minutes ago, incus said:

 

So you want each juncture from modem to DAC to have the most accurate and quietly powered clock possible as well as the best possible noise isolation and/or filtering. So if you pull an ethernet cable from some random place in the chain -- assuming there is some random component buffering the signal -- and there is no change - doesn't that simply mean that the component doing the buffering is no worse than anything that came before it?

 

You'll do well to ponder some more on what I said: If you start playback of music, in this case Win10 and Tidal, and you pull the Ethernet cable and SQ doesn't change....

[plissken]

16 minutes ago, The Computer Audiophile said:

I believe this is false.

Actually it's true.

 

In order to read off of a buffer some form of clocking needs to be present to read it out in an ordered fashion.

 

Keep in mind that one thing buffers do is enable linkage between two clock domain boundaries.

[plissken]

9 minutes ago, The Computer Audiophile said:

Not true with all digital signals. Ethernet sure, but a USB to AES converter doesn't reclock the data.

But AES isn't mentioned in the article and we are talking about a specific implementation.

If there is no buffering involved in your example I could see the point.

 

A quick glance at the Wiki for AES-EBU does show a clock and it's not the same clock on the USB bus (8Khz?).

 

So something is going on unless you mean something else.

[plissken]

5 hours ago, PeterSt said:

2. The continuous processing involved when connected to the Internet (or normally live Ethernet) is not present in the audio playing PC (it is in the PC which holds the music data). 

 

 

Like that matters. I posted years ago a screenshot all the process caching that goes one with even optimized PC has going on. A few more aren't going to matter.

 

If you couldn't hear the 'noise' the 120+ processes are generating you aren't going to hear a few more.

 

This ignorance is beyond the pale.