Silent Angel Bonn N8 Audio Grade Ethernet Switch

[radix]

hi sq225917,
well...
please have a try : change a clock (or quartz) and then you'll talk about.
no test... well... I'd be like you : WTF changing a clock on a switch ? Theoricaly : useless (of course vs tcp/ip protocol)

but streaming is is not based on tcp/ip, it's real-time stuff !
thus... open a switch / take off the basic quartz / stick a brand new tcxo or oxco : and then => give your conclusions !

no brainer guys, I know that I know nothing

@Jinjuku also addressed this, but I'll add my 2 cents.

Digital audio -- that is AES/spdif/toslink (they are basically the same protocols) -- are real time. They transfer digital audio sound samples in time slots. Here clock accuracy is important, up to some degree. There's other threads about digital audio clocks.

Streaming protocols are most definitely not real time. Digital audio uses the time slot of a sample to determine when to play it. Streaming protocols put a timestamp in the packet to say when to play the clock sample in the packet. This allows the player to buffer the samples from the packets and even reorganize them in time (e.g. packet 2 arrived before packet 1 and they get reordered so they play correctly). Depending on your network, there may be significant buffering to mask out delay and jitter variations.

A lot of streaming protocols use TCP, which means "packet" is not really that important as it's transferred as a byte stream and TCP will take care of re-ordering packets and re-transmit dropped or delayed packets. TCP is most definitely not real-time.

Marc

 

[Jinjuku]

Depending on your network, there may be significant buffering to mask out delay and jitter variations.

Part of my network is running 10GBE. I just did a file copy and hit 498MB/s. That never gets old.

To put it into perspective ~45 times faster than a well known, hot running, audiophile switch that cost more than 300% of my Cisco/SolarFlare/Broadcom Fiber setup.

Can literally queue up an entire FLAC album in 1 second.

 

[Jinjuku]

I love it. ~4GB in about 5 seconds.

 

[sq225917]

No no no, you guys have it all wrong. What you're actually saying is that the switch can spend 99% of its working day concentrating on sending noise as only 1% of its time is spent sending data. Now it all makes sense.....

 

[Jinjuku]

No no no, you guys have it all wrong. What you're actually saying is that the switch can spend 99% of its working day concentrating on sending noise as only 1% of its time is spent sending data. Now it all makes sense.....

Also faster switches can send more noise in the same interval as a slower switch. So we need dial up modems over old POTS networks. We'll have Uptone and audio quest get right on that.

 

[sarumbear]

but streaming is is not based on tcp/ip, it's real-time stuff !

Shall we see for instance what “stuff” Spotify use? (Spoiler: it’s TCP)

https://www.hugoribeiro.com.br/biblioteca-digital/Zhung-Understanding_User_Behavior_in_Spotify.pdf

 

[Gremlins]

Hello Amir,
I was a bit too straight in my previous blabla, I'll be more cool, sorry about that ;-)

What I say is : your paragidm to do some measurements over the audio bandwidth, on device that work at much high bandwidth, is not relevant.

Let's take a farily basic tweak :
=> change the quartz or the basic oscillator on a switch (good TXCO, or OCXO with external PS for instance ; TCXO can use the PS from the PCB, OXCO requires more amps thus complete external PS is required)
=> you will get a improvement on SQ : more space between instruments / more depth etc...
(don't tell this is my own idea & feeback, numerous tweakers did it...)
Clocking in these device is standard => 25MHz
25MHz is completly out of the audio bandwidth of course, BUT => nice effect on the SQ with the proper tweak, clock change for instance.

I apologize since I don't know your background etc...
but did you tweak such devices ? to see effects on SQ ?
fairly basic work though...

@ radix :
you said a bloody true stuff => "Ethernet is designed to work in noise"
of course you're right (there are SMPS all around network devices etc.. for instance)
but let's see the Ethernet that way : like a "professional juggler", thus if you leave him doing his job : no fail. No fail unless you grab his arm during a few second & so on... but it's unfair vs this nice juggler & never happen if you refer to the analogy with data transfer in a switch.

well...
we could discuss about it even longer....
the basic question to people that state that "move along, no issue there" : did you ever tweak a switch with a better clock ? basic work though...

I made various tweaks on my hifi/network gear :
1. part of them I expected "an improvement" led to : equal... or worse SQ
2.-part of them I expected "an improvement" led to : improvements ! cool !
psycho induced effect (placebo) between 1 & 2 : of course not ! otherwise I would get 100% SQ improvements all over my tests LOL

blabla bla... i stop there (useless exchange ? ) : do experiment clock swap & then you'll change your mind
Rgds

Man, some guys understand nothing in audio gear, they create their own knowledge and illusions ..

When you go to dentist, do tell him what to do? No, you trust him because he knows more than you, period

 

[sarumbear]

Man, some guys understand nothing in audio gear, they create their own knowledge and illusions ..

When you go to dentist, do tell him what to do? No, you trust him because he knows more than you, period

Oh no, many in this forum will question the dentist and make a video that dentistry is not what it seems.

We don’t need no education…

 

[radix]

@ radix :
you said a bloody true stuff => "Ethernet is designed to work in noise"
of course you're right (there are SMPS all around network devices etc.. for instance)
but let's see the Ethernet that way : like a "professional juggler", thus if you leave him doing his job : no fail. No fail unless you grab his arm during a few second & so on... but it's unfair vs this nice juggler & never happen if you refer to the analogy with data transfer in a switch.

well...
we could discuss about it even longer....
the basic question to people that state that "move along, no issue there" : did you ever tweak a switch with a better clock ? basic work though...

I made various tweaks on my hifi/network gear :
1. part of them I expected "an improvement" led to : equal... or worse SQ
2.-part of them I expected "an improvement" led to : improvements ! cool !
psycho induced effect (placebo) between 1 & 2 : of course not ! otherwise I would get 100% SQ improvements all over my tests LOL

blabla bla... i stop there (useless exchange ? ) : do experiment clock swap & then you'll change your mind
Rgds

@phile2,

What I am looking for is a mechanism by which upgrading an oscillator on a switch can change the sound quality of a typical streaming protocol. A simile to a Juggler is not convincing. A switch is not like a juggler.

I have done experiments with audio and video streaming over different network conditions (delay, packet loss) and my background suggests there is no mechanism for a properly spec'd oscillator to cause changes in audio dynamics as you describe. Either packets are decoded correctly or they are not. There's a really astronomically small probability of a false packet acceptance (a corrupted Ethernet packet making its way past all the error correction and checksums).

Rather than talk abstractly, maybe you could offer one specific switch and one specific upgrade and one specific listening experience and how you think that changed the sound quality. And do that with enough details (or links to other sites) that someone could perform the same change. That is something someone could replicate and evaluate.

Marc

 

[Gremlins]

@phile2,

What I am looking for is a mechanism by which upgrading an oscillator on a switch can change the sound quality of a typical streaming protocol. A simile to a Juggler is not convincing. A switch is not like a juggler.

I have done experiments with audio and video streaming over different network conditions (delay, packet loss) and my background suggests there is no mechanism for a properly spec'd oscillator to cause changes in audio dynamics as you describe. Either packets are decoded correctly or they are not. There's a really astronomically small probability of a false packet acceptance (a corrupted Ethernet packet making its way past all the error correction and checksums).

Rather than talk abstractly, maybe you could offer one specific switch and one specific upgrade and one specific listening experience and how you think that changed the sound quality. And do that with enough details (or links to other sites) that someone could perform the same change. That is something someone could replicate and evaluate.

Marc

Yeah, but you know that some audiophiles can here that missing packet, even if its one over a million .. he's an audiophile after all .. he hear things that comon people or mesurements do not

 

[sq225917]

the same guys who can hear ssd vs hd, and boggo idc cable vs audiophile grade sata bollocks

 

[Gremlins]

Yeah, this one
In our hobby , we have guys that should attempt guiness wolrd records for human hearing


Btw, @Amir : it would be great to make an article with measurements about treshold hearing related to age. Many guys than can aford expensive gear are olders, but some of these guys can hear things in internet switch and cables .. how come? They can hear differences between amps with similar specs and mesurements etc etc
How can you hear a cable when you are over a certain age if your bandwith is naturally limited?

 

[Killingbeans]

Btw, @Amir : it would be great to make an article with measurements about treshold hearing related to age.

There's this: https://www.audiosciencereview.com/...ds/most-useful-diagram-of-hearing-loss.26779/

How can you hear a cable when you are over a certain age if your bandwith is naturally limited?

Same reason you can if you have the hearing of a newborn baby. Preconceptions fuel the ego like a bucket of gasoline fuels a bonfire.

 

[radix]

Yeah, this one
In our hobby , we have guys that should attempt guiness wolrd records for human hearing


Btw, @Amir : it would be great to make an article with measurements about treshold hearing related to age. Many guys than can aford expensive gear are olders, but some of these guys can hear things in internet switch and cables .. how come? They can hear differences between amps with similar specs and mesurements etc etc
How can you hear a cable when you are over a certain age if your bandwith is naturally limited?

If the perceived difference is sound stage and definition, I'm not sure how one objectively measures it. It seems like it would be specific to the room, the position in the room, the speakers, the overall crosstalk, and of course the mix. Is there a way to measure those? I've not seen one, but I'm by no means an expert on this.

Maybe one could ask, how much could the transport of audio affect soundstage? I can see how the transmission of FM could lead to soundstage differences, as it's sending L+R and L-R and depending on how one receives those it would have a big effect. But for separate and likely near lossless transport of L and R, how could the transport affect it? I'm sure one could instrument an open source stream player and measure packet statistics or do experiments with different latency/reliability.

Anyway, I'm mostly interested in how one would setup an experiment to support or refute such an argument, beyond the THD+N and SINAD and FR measurements. I personally don't see a mechanism by which the Ethernet switch makes an audible difference. Perhaps one could instrument an open source stream client and show there's no difference in the playout buffers between one switch and another?

 

[Murmon]

Hi guys! I want to share my project.
NeoZuxel project!
Compiled on the basis of Zyxel GS-108B v3.
Fitted details: clock Neutrino 2 Compact Audio Reference Clocks (http://www.newclassd.com/index.php?page=168&hv=1)
and
Ultra Wide Bandwidth Mk 2 Voltage regulator (http://www.newclassd.com/index.php?page=250) NewClassD.
Power to the switch comes from a 12-volt LPS. It has a branch inside, so 12 volts go to the clock,
and 5 volts are supplied to the board through the regulator.
Everything fit into the switch case.
From additional materials for fastening and insulating the clock and the regulator, electrical tape was needed,
a strip of double-sided tape, a bolt and a sleeve. Tools: cutter pliers, soldering iron, soldering accessories and a pair of screwdrivers.

 

[Killingbeans]

@Murmon I love dicking around with my soldering iron, but I have to ask... was there any practical reason for this project?

 

[sq225917]

As stated.

 

[Murmon]

@Murmon I love dicking around with my soldering iron, but I have to ask... was there any practical reason for this project?

Certainly. Analog sound, detail and silence. All this is the best!

 

[sq225917]

Analogue sound, from a digital device?

 

[Murmon]

Analogue sound, from a digital device?

Yes, right. Try it, all this is possible and not as expensive as silent angel, bonn n8 ....