Optical Network Configurations

[muski]

The impact of upstream RF noise is an interesting question. I recently changed ISPs from Xfinity cable to AT&T Fiber. I switched for reasons of price and  upstream speed and wasn't even thinking about any impact on my audio chain. But during my first post-installation listening session, I immediately noticed the soundprint of reduced RF in the chain, wondered where it came from, and realized I'd changed ISPs.

It could be that the AT&T Fiber modem is quieter than the Arris SB8200 on Xfinity, or because of a new network topology my audio subnet was now optically isolated from the AT&T Fiber modem. Either way, this is all upstream of my final optical network run, and yet I could hear a very significant difference.

 

All of this brings to mind my favorite AS topic of all time, which asks if we are all chasing improvements to reduce
RFI/EMI?  I'd say yes...sure feels like RF/EMI is the Achille's heel of all digital music reproduction.

 

P.S. I've tested about half a dozen SFPs. To my ears the Finisar FTLX1475D3BCV sounded best. It's 10G and single mode. One is in my Sonore Optical Module Deluxe (fed by an Innous Phoenix Net), and the other is in my Sonore opticalRendu.

[audiobomber]

6 hours ago, muski said:

I've tested about half a dozen SFPs. To my ears the Finisar FTLX1475D3BCV sounded best.

I am curious as to which SFPs?

[muski]

These are the other ones I've tried...

TP-Link TL-SM311LM

Finisar FTLF1321P1BTL

Finisar FTLX8574D3BCV

 

[audiom3]

2 minutes ago, muski said:

These are the other ones I've tried...

TP-Link TL-SM311LM

Finisar FTLF1321P1BTL

Finisar FTLX8574D3BCV

 

@audiobomberif you want to know of other transceivers, I may have listed some ealier in this thread. I have over 100 SFP/SFP+ transceivers. I have used the three above and countless ones from GiTi, 10Gtek, Wiimark (I believe that's the name), Intel, Finisar among others. 

[jabbr]

On 2/13/2024 at 10:20 AM, dbastin said:

I thought I read in this thread (or the EtherRegen thread) the SFP28 can auto-negotiate speed to work in 1G and 10G devices. It was a way to work around the constraint that ER needs a 1G SFP module even when using a 10G switch like CRS 305 with 10G modules at tgat end..  Maybe my recollection is wrong or what I read is.

 

The SFP28 autonegotiation is for an SFP28 capable switch with an SFP28 module ... so yes it autonegotiates but at a cost

[jabbr]

On 2/13/2024 at 11:29 AM, audiobomber said:

Thanks for clarifying, because it seems that some people have taken the "10G has lower jitter" statement as proof that 10G is better than 1G for audio. Actually, I run my audio systems at 100Mbps where possible, as suggested by some audiophile companies. 

 

The SOtM renderer in my desktop system offers 100BASE-T as an option, and it sounds better than 1000BASE-T according to most users.

https://audiophilestyle.com/forums/topic/69285-about-cpu-frequency-function-on-the-sms-200-and-sms-200ultra /page/2/#elControls_1270660_menu

 

My main system also runs at 100Mbps, because I use an Audio Sensibility Signature two-pair ethernet cable ahead of the opticalModule FMC. I use a four-pair Audio Sensibility Signature I2S ethernet cable downstream, between the ER A-side and my PlayPoint renderer. I initially had the cables reversed so that the system was gigabit up to the ER, but the current arrangement, with lower speed feeding the oMD, SFPs and ER sounds significantly better.

 

Do what sounds best to you. There is no "proof" that the lower jitter of 10G has better SQ.

[jabbr]

On 2/15/2024 at 11:43 AM, SQFIRST said:

It also helps to read the above quote keeping in mind that @jabbr has probably not used 1000-BASEX (1G) fiber since soon after starting this thread. Most need not go as high as 100G as the benefits start showing 10G onwards.

Many of my endpoints are still 1G so yeah I still have and use 1G SFP modules... my switches are all 10-100G

[ericuco]

This is my system that I have used for years now, at least the network portion. Mix of 1Gbe & 10Gbe as well as multi-mode & single-mode. Both Mikrotik switches are managed so you can see what the settings are.

 

[dbastin]

9 hours ago, jabbr said:

 

The SFP28 autonegotiation is for an SFP28 capable switch with an SFP28 module ... so yes it autonegotiates but at a cost

Thanks, I thought I had read SFP28 can autonegotiate in the ER 1G SFP cage.  That is where it would be udeal to put the best SQ SFP module but ER might accept only 1G, apparently it won't work with 10G modules but I can't recall if dual rate 1/10G modules work.

[SQFIRST]

@jabbr It's understandable that a mix of 1G is likely as most endpoints, especially for audio, are designed with 1G. I was implying that for proper network isolation where even with 1G fiber the upstream components are intrusive, then just 1G fiber may not be sufficient. That was the case for my setup. 

 

@ericuco thanks for sharing your network diagram. It's a nice example of 10G audio isolation from servers. Just curios whether the upper 10G run is to isolate the HQP from NUC and router or something else?

[SQFIRST]

20 hours ago, jabbr said:

There is no "proof" that the lower jitter of 10G has better SQ.

Something more sciency than ears?  :)

 

[Jud]

38 minutes ago, SQFIRST said:

Something more sciency than ears?  :)

 

 

I'm personally not aware of peer-reviewed academic research into how or whether that level of difference in jitter might affect our perceptions.

 

But when a 10G fiber output from my PC and a 10G fiber switch each cost less than a week's worth of my grocery shopping, well, why not?

[jabbr]

2 hours ago, SQFIRST said:

Something more sciency than ears?  :)

 

I'm unable to tell what someone else hears with their setup. People should listen to what they like.

 

What I can do is suggest a system with low jitter and CMN rejection, 10G fiber does that by design. One segment of 10G will essentially eliminate upstream jitter so I don't see a final 1G hop to the endpoint as an issue from an engineering point of view. 

 

I have been using 10G-100G switches for awhile and have tested a bunch of NICs some designed for ultralow jitter ie better than plain 10G. I haven't been able to tell a difference swapping high speed fiber NICs at the server. I think I've documented my own trials on this thread but I've used a variety of NICs as well as switches from very well established vendors for example this: NVIDIA MSN2100-CB2F Spectrum 100GbE 1U Open Ethernet Switch ... I'm not suggesting that anyone need this for home audio just that personally I can't hear a SQ difference once jitter goes below some threshold.

 

[jabbr]

2 hours ago, SQFIRST said:

@jabbr It's understandable that a mix of 1G is likely as most endpoints, especially for audio, are designed with 1G. I was implying that for proper network isolation where even with 1G fiber the upstream components are intrusive, then just 1G fiber may not be sufficient. That was the case for my setup. 

 

Sure! and I'm saying that 1G was not designed/specified to block upstream jitter whereas 10G was explicitly designed that way...

 

[jabbr]

1 hour ago, Jud said:

 

I'm personally not aware of peer-reviewed academic research into how or whether that level of difference in jitter might affect our perceptions.

also we are talking about network switch jitter not DAC clock jitter which might actually be important!

1 hour ago, Jud said:

 

But when a 10G fiber output from my PC and a 10G fiber switch each cost less than a week's worth of my grocery shopping, well, why not?

 

exactlty the Mikrotik 10g and 100g switches -- among others! -- are so cost effective why not? (I certaintly didn't pay $32k for my NVidia/Mellanox switch ... got it on eBay and a very nice Mellanox engineer helped my get it working!)

[ericuco]

7 hours ago, SQFIRST said:

 

@ericuco thanks for sharing your network diagram. It's a nice example of 10G audio isolation from servers. Just curios whether the upper 10G run is to isolate the HQP from NUC and router or something else?

No special reason, the Intel NIC runs at 10Gbe so why not.  I would run everything at 10Gbe if possible.

[muski]

11 hours ago, jabbr said:

for example this: NVIDIA MSN2100-CB2F Spectrum 100GbE 1U Open Ethernet Switch

 

 

“with 16 ports running at 100GbE, the SN2100 carries a whopping throughput of 3.2Tb/s”

 

Nice. I want one!

 

[SQFIRST]

21 hours ago, jabbr said:

I'm unable to tell what someone else hears with their setup. People should listen to what they like.

 

22 hours ago, Jud said:

I'm personally not aware of peer-reviewed academic research into how or whether that level of difference in jitter might affect our perceptions.

 

That was just an attempt at humor guys :)

Sufficient feedback, coverage and information about 10G exists on this thread that a listening test is good enough as proof for hobbyists like myself. Of course, any in depth technical paper can be helpful too and there are the protocol information from IEEE.org for those wanting technical material.

Just trying to give back as I gained from this site and appreciate all the good information.

[SQFIRST]

21 hours ago, jabbr said:

also we are talking about network switch jitter not DAC clock jitter which might actually be important!

 

This is a good distinction for the component factor. For setups using USB-Audio into DAC the experience will be different from those that have Network DACs.

I use the Lumin X1 streamer-dac which has only Ethernet as input so the network jitter has high impact. 

[jabbr]

5 minutes ago, SQFIRST said:

 

This is a good distinction for the component factor. For setups using USB-Audio into DAC the experience will be different from those that have Network DACs.

I use the Lumin X1 streamer-dac which has only Ethernet as input so the network jitter has high impact. 

 

Yeah **honestly** I couldn't predict how much input ethernet "jitter" effects SQ on the Lumin. It might not at all because buffers etc. There's been alot of hand wrangling to suggest that it might but in the decade that this has been debated and despite promises of tests etc I haven't seen an actual shred of evidence to say that the Lumin, to give a specific example, SQ is at all affected by input ethernet jitter.

 

Now I've chosen to remove the possibility that it might be important by using 10g switches with a 1g output but really its theoretical and subjective reports which don't use modern highspeed fiberoptic ethernet (eg 10G and faster)

 

It would be interesting to get something definitive from Lumin if they have any data...

[SQFIRST]

As someone that uses the X1 daily for last couple of years I can attest to the fact that network quality does have high impact, regardless of buffer or any inbuilt isolation. I wish that it were not the case but it took some effort to get to the sound that I like.

 

Examples:
Roon sounds totally different running through a 10G segment. Without 10G the 1G sound is frankly unlistenable to my ears.
Control point and background traffic still matters. There is some background network activity taking place even with audio being buffered that has impact. This is similar to removing the Lan connection and listening. To mitigate this 10G works effectively as well.

 

Whether these can be categorized as 'jitter' I don't know but there is certainly some networking aspect that still makes a difference to the X1 and the network signal. Reading feedback on network issues from the others on here they seem very similar.

[Jud]

22 hours ago, jabbr said:

On 2/20/2024 at 10:07 AM, Jud said:

 

also we are talking about network switch jitter not DAC clock jitter which might actually be important!

 

Just idly wondering whether, since we're dealing with the timing of zero crossing points of an analog wave, this can be additive. Likely not, or I'm sure all sorts of computer and telecommunications equipment wouldn't work otherwise.

 

And I'm also idly curious about what happens when what comes over the network and runs through a USB connection hits the DAC and the DAC clock. I know the DAC clock governs the timing of the async connection, just wondering what jitter from the incoming signal causes electrically on the DAC receiver end.

 

Those idle thoughts aside, as I said above I'm not aware of any peer reviewed research showing any audible effects.

[jabbr]

4 minutes ago, Jud said:

 

Just idly wondering whether, since we're dealing with the timing of zero crossing points of an analog wave, this can be additive. Likely not, or I'm sure all sorts of computer and telecommunications equipment wouldn't work otherwise.

 

And I'm also idly curious about what happens when what comes over the network and runs through a USB connection hits the DAC and the DAC clock. I know the DAC clock governs the timing of the async connection, just wondering what jitter from the incoming signal causes electrically on the DAC receiver end.

 

Those idle thoughts aside, as I said above I'm not aware of any peer reviewed research showing any audible effects.

 

At this point all discussion of such noise has been theoretical and despite promises that "any day now" we are going to see data, after 10 years there hasn't been any -- for audio systems.

 

Correct it can't be additive for high speed modern computers and networks. For 10G+ networks there has been masses of data and there are devices which measure this because its a big industry. Differential noise, whether ground plane, supply plane, or phase, isn't allowed to propagate through a 10G+ switch ie it isn't allowed to be additive and there need to be circuits which are designed explicitly so the noise isn't additive. At 100G, for example, even a tiny bit of noise can throw the system completely off. Its not just peer reviewed research, its by specification and design and compliance testing.

 

 

[jabbr]

Hey @Jud I see you have a fiberoptic USB cable, what do you think?

[audiom3]

4 minutes ago, jabbr said:

Hey @Jud I see you have a fiberoptic USB cable, what do you think?

I don't have a fiberoptic USB but I do own a pair of Wireworld Stellar Optical HDMI cables.  I use them on my Zidoo UHD3000.  One into my OLED and the other into my Lyngdorf SSP.  The Zidoo is a Roon Endpoint for 5.1 SACD and DVD-A files.  So I do get to listen to hi-rez music through it.  They're certainly not true optical cables but a hybrid of copper+optical but I am very pleased with them.  Never once experienced intermittent blackouts or audio drop outs that every other cable has had at some point.