Optical Network Configurations

[Superdad]

20 hours ago, lmitche said:

My understanding is that the MMs use leds, not true lasers, so that may cause the difference in SQ.

 

Hi Larry:

I'll demur to the eggspurts here, but AFAIK even the $13 multi-mode SFP transceivers use VCSEL lasers, not LEDs.
A lot of single-mode transceivers us DFB (distributed feedback) lasers. 

[lmitche]

1 hour ago, Superdad said:

 

Hi Larry:

I'll demur to the eggspurts here, but AFAIK even the $13 multi-mode SFP transceivers use VCSEL lasers, not LEDs.
A lot of single-mode transceivers us DFB (distributed feedback) lasers. 

Hi Alex,

 

I am certainly no expert (or eggspurt) on lasers or leds or the differences between them. My understanding that MM SFPs use leds, not true lasers could easily be wrong. Thanks for the correction.

 

I do observe a significant increase in SQ listening to SM vs MM SFP+s. There is a clarity and focus that doesn't happen with the MM SFPs.And now, finisar SFP+s sound better then fs.com SFPs of the same spec.

 

That is the extent of my very boring and tedious SFP rolling journey.

 

Nevertheless I am happy with the outcome.

[rickca]

8 hours ago, Superdad said:

I'll demur to the eggspurts here, but AFAIK even the $13 multi-mode SFP transceivers use VCSEL lasers, not LEDs.

This article clears things up.

https://www.ofsoptics.com/faq-guide-to-laser-optimized-fiber/

[lmitche]

4 hours ago, rickca said:

This article clears things up.

https://www.ofsoptics.com/faq-guide-to-laser-optimized-fiber/

Thanks Rick,

 

Good link, This paragraph explains the behavior of VCSEL lasers used in Multi-Mode Fibers (MMF):

 

"To complicate matters, the power profile of a VCSEL is nonuniform and fluctuates constantly. It changes sharply across its face, varies from VCSEL to VCSEL and changes with temperature and vibrational fluctuations. Consequently, individual VCSELs will excite different modes in a certain fiber at any given time. And because different modes carry varying amounts of power, the fiber’s bandwidth can vary in an unpredictable manner."

 

Anything that is nonuniform in power demand is going to diminish our SQ.

 

Here is another resource:  https://www.optcore.net/transceiver-laser-types/

 

Here it is explained that for single mode fiber (SMF) SFP+s the DFB (distributed feedback).lasers have a "prominent feature of their excellent monochromaticity (ie, spectral purity". Could that mean that DFB laser current demand is more constant then VCSEL? If so, does that explain the SQ difference between MMF and SMF SFP+s?

 

Neither of these articles explains why the finisar SFP+s sound better then the fs.com SFP+s. My current hypothesis is that the EMI/RFI emissions vary between manufacturers and that the finisars emit much less noise than the FS.coms.

 

 

[jabbr]

From 2015 so doesn't have latest 100,200,400 gbe products but lists the laser type across pluggable modules ...

https://www.finisar.com/sites/default/files/resources/finisar_optical_transceiver_product_guide_3_2015_web.pdf

 

@lmitche : according to https://en.wikipedia.org/wiki/Distributed_feedback_laser there are two ways to modulate the DFB laser, the most direct is to vary current but better to keep the laser on and modulate its output (EML) ... the area of "silicon photonics" is white hot (sic) right now https://en.wikipedia.org/wiki/Silicon_photonics ... you can use a single laser always on and split into 4 channels each of which is modulated to get a total of 100 Gbe, and this is very cost effective. (cable with 4 fibers each direction). Alternatively you can use 4 lasers each of a different frequency, modulate each individually and then combine into a single singlemode fiber for duplex 100Gbe communications.

 

I'm gonna say that a company which is capable of producing a functional 100Gbe module will be able to do it in a very low noise way (otherwise it just doesn't function) and ideally that technology will flow down to its 10 and 25 Gbe modules, and these when run at 1Gbe would be expected to have the vey lowest noise. At least that is my own approach. 

[barrows]

5 hours ago, lmitche said:

Anything that is nonuniform in power demand is going to diminish our SQ.

 

Hey man, I find this discussion very worthwhile and illuminating in many ways.  But,, while I agree with the basic idea of the above, I think the statement is a bit too absolute to be (always) true, especially without objective verification.  I would suggest that we would need to establish where these power supply perturbations actually may have audible consequences, lest we drive folks down paths which may be a bit crazy, and perhaps even a waste of time and effort.

For example, i would submit that such power supply perturbations are only going to have a possibility of influencing the actual sound quality (in the analog domain, loudspeakers, or headphones) IF they make their way to the USB output.  I can accept that in the case of these power supply perturbations getting to the USB output, that they may affect sound quality.  This could be related to the USB ground, the data lines, or between the ground and power line.  I would also suggest, that even if these power supply supply perturbations are present on the power feed for the SFP module (and the same would apply to the processor chip, memory, etc, etc), some Renderers may be well enough engineered that these power supply perturbations do not make it the USB output.  I would also point out that some well equipped engineers can (and do) measure the USB output for the presence of any such problems. 

 

@jabbr, with your preference for Finisar product, do you suppose that Finisar SFP transceivers may have benefits for 1Gbe SFP (non +) as well as SFP+?

[plissken]

20 minutes ago, barrows said:

i would submit that such power supply perturbations are only going to have a possibility of influencing the actual sound quality (in the analog domain, loudspeakers, or headphones) IF they make their way to the USB output.  I can accept that in the case of these power supply perturbations getting to the USB output, that they may affect sound quality.  

 

Also it can't be ignored that the CPU itself is always going to be modulating power like crazy compared to a network connection. Same for all the I/O demand/load. Buffers, Interrupts, Applications.

 

My Solar Flare SFP+ NIC support two adapters. Could totally setup LACP LAG to a switch LAG and quite easily put in an FS and Finistar SFP+ module and test out in real time.

 

Knowing what I know about how non-realtime audio actually works I'm doubtful but I have asked, but have not seen, the setup that this is effecting. Still curious about that.

[Superdad]

41 minutes ago, barrows said:

I would also suggest, that even if these power supply supply perturbations are present on the power feed for the SFP module (and the same would apply to the processor chip, memory, etc, etc), some Renderers may be well enough engineered that these power supply perturbations do not make it the USB output.

 

Hi Barrows:

Well you can answer the question for yourself by changing power supply to an opticalModule FMC feeding an opticalRendu. Lot's of Sonore clients doing just that and reporting differences...

[barrows]

56 minutes ago, Superdad said:

 

Hi Barrows:

Well you can answer the question for yourself by changing power supply to an opticalModule FMC feeding an opticalRendu. Lot's of Sonore clients doing just that and reporting differences...

Hi Alex, I am trying to keep the thread generic as to "brands" and such...  Of course i use an oM (with a Sonore Power Supply of my design with single digit µVs of noise), and run SMF to my Signature Rendu SEoptical.

 

But, the power supply quality on the oM in this instance may produce improvements for other reasons: such as lowering the phase noise on the oscillator in the oM...  So I cannot for sure say that power supply perturbances caused by the SFP transceiver is the reason an improved power supply might improve audible performance.  Anyway, I was thinking more about the SFP transceiver in the Renderer side, where any power supply perturbance might have a path to the USB output (or not!).  But these are all unproven allegations as far as I am concerned, and I consider it highly speculative to suggest that they do effect sound quality, given that the only evidence of such is entirely subjective (and neither is it agreed upon) at this point.  This does not mean that these things do not effect sound quality, just that while it might be OK to speak in absolutes about such esoteric, and unproven things in the "massive" thread, at other places on AS I would prefer to keep things a little bit more down to earth.  Such that, when extreme claims are made (from a technical perspective), they are actually accompanied by some level of at least technical explanation for such, if not actual proof through measures.

This thread, for example, as started by @jabbr, has kept to a fairly reasonable level of discussion in terms of avoiding esoteric claims of "massive" improvements, and sticking to technical matters which actually have at least some degree of technical merit behind them.  There already is a place on these forums reserved for outrageous claims of "massive" improvements which have no solid basis in their technical merit, which is fine with me.    

[lmitche]

2 hours ago, barrows said:

Hi Alex, I am trying to keep the thread generic as to "brands" and such...  Of course i use an oM (with a Sonore Power Supply of my design with single digit µVs of noise), and run SMF to my Signature Rendu SEoptical.

 

But, the power supply quality on the oM in this instance may produce improvements for other reasons: such as lowering the phase noise on the oscillator in the oM...  So I cannot for sure say that power supply perturbances caused by the SFP transceiver is the reason an improved power supply might improve audible performance.  Anyway, I was thinking more about the SFP transceiver in the Renderer side, where any power supply perturbance might have a path to the USB output (or not!).  But these are all unproven allegations as far as I am concerned, and I consider it highly speculative to suggest that they do effect sound quality, given that the only evidence of such is entirely subjective (and neither is it agreed upon) at this point.  This does not mean that these things do not effect sound quality, just that while it might be OK to speak in absolutes about such esoteric, and unproven things in the "massive" thread, at other places on AS I would prefer to keep things a little bit more down to earth.  Such that, when extreme claims are made (from a technical perspective), they are actually accompanied by some level of at least technical explanation for such, if not actual proof through measures.

This thread, for example, as started by @jabbr, has kept to a fairly reasonable level of discussion in terms of avoiding esoteric claims of "massive" improvements, and sticking to technical matters which actually have at least some degree of technical merit behind them.  There already is a place on these forums reserved for outrageous claims of "massive" improvements which have no solid basis in their technical merit, which is fine with me.    

Barrows, why do you use SMF vs MMF from the Sonore OM to Signature Rendu SE Optical?

[barrows]

1 minute ago, lmitche said:

Barrows, why do you use SMF vs MMF from the Sonore OM to Signature Rendu SE Optical?

yes, I am using SMF with Cisco LX SFP transceivers.

[lmitche]

3 minutes ago, barrows said:

yes, I am using SMF with Cisco LX SFP transceivers.

Can you please share why you chose SMF SFPs and fiber over MMF SFPs and fiber?

[barrows]

6 minutes ago, lmitche said:

Can you please share why you chose SMF SFPs and fiber over MMF SFPs and fiber?

Because some have suggested that there are advantages to using SMF, and it is easy to try.  It did not sound worse, but there was not any notable improvement either.  I did not do strict, level matched, careful A/B comparisons either, so there may be a small difference, but nothing really very significant here.

[lmitche]

36 minutes ago, barrows said:

Because some have suggested that there are advantages to using SMF, and it is easy to try.  It did not sound worse, but there was not any notable improvement either.  I did not do strict, level matched, careful A/B comparisons either, so there may be a small difference, but nothing really very significant here.

That's interesting, many thanks. I was curious to know if your rationale for the SMF choice was based on something measured or observed through listening.

 

I wouldn't say that the change from MMF to SMF here is massive, but it's significant enough to justify the expense of purchasing two used SMF SFP+s, $24 for two plus $10 for 5m of fiber.. In my case these are used on a 10gbps link between a PCIE card in my server and an SPF connected USB 3 controller with two ports. The difference in clarity was immediate.

 

The SQ difference between the fs.com and finisar SFP+s  is another story and is dependent on which DAC is in use. With one, there was little difference. With the other, it had a major impact on the sound adding audible distortions to the image. That led me to hypothesize that the EMI/RFI emissions impacted the SQ on the second DAC. These distortions disappear with the Finisar SPF+s.

 

I will stop posting here now.

[barrows]

@lmitche, no need to stop posting here, I would like to continue to hear about your experiences with optical network configurations.  And I have nothing against subjective evaluations either, just that when we get into areas which are really esoteric, i would think a bit of restraint is in order.  As one who works with a manufacturer of Network based audio products, it important to understand where I am coming from.  Threads, such as the "massive" thread, while they may be very interesting for the devoted participants there, are very intimidating to potential customers looking to switch to Ethernet distributed audio.  A lot of potential customers looking to set up an Ethernet based system take one brief look at that thread, and turn their backs on the idea of implementing an Ethernet based system.  I mean, and they start asking questions like: do I need a Mutec clock master to slave all my Ethernet gear to!  Imagine how intimidating that can be to someone new to Ethernet distributed audio.

 

All I am  saying is that absolute statements, such as power supply perturbations are always going to result in sound quality issues, could perhaps be prefaced with: "IME", or "in my system".  When I hear these remarks about gear upstream of the optical fiber isolation, my own eyebrows raise a bit.  This does not mean I feel such is "impossible", just that it is unlikely, and needs to be verified by a bit more than only subjective means before suggesting that it is "fact".

 

My question when one raises this possibility is: then what are the implications for audio streams from the internet, like Tidal or Qobuz ?  I am sure they are not using silver gold alloy DC cables, for example, for all of their DC supplies, probably not using Mutec level clock generators either, and there is likely not a single linear power supply in the path from their servers to one's home.  Not to mention the accumulating clock phase noise of the hundreds of clocks the data encounters along the way...  

Now I play only local files here anyway, myself, so i do not know if Tidal and Qobuz hi res. are just as unlistenable as I find 128 Kb/s MP3, perhaps they are...

[plissken]

37 minutes ago, lmitche said:

That's interesting, many thanks. I was curious to know if your rationale for the SMF choice was based on something measured or observed through listening.

 

I wouldn't say that the change from MMF to SMF here is massive, but it's significant enough to justify the expense of purchasing two used SMF SFP+s, $24 for two plus $10 for 5m of fiber.. In my case these are used on a 10gbps link between a PCIE card in my server and an SPF connected USB 3 controller with two ports. The difference in clarity was immediate.

 

The SQ difference between the fs.com and finisar SFP+s  is another story and is dependent on which DAC is in use. With one, there was little difference. With the other, it had a major impact on the sound adding audible distortions to the image. That led me to hypothesize that the EMI/RFI emissions impacted the SQ on the second DAC. These distortions disappear with the Finisar SPF+s.

 

I will stop posting here now.

 

What end points? What server? What DAC? How can I attempt to recreate what you have?

 

Anyways I have some Finisar 10G LC on order.

[jabbr]

On 6/14/2020 at 2:01 PM, barrows said:

 

@jabbr, with your preference for Finisar product, do you suppose that Finisar SFP transceivers may have benefits for 1Gbe SFP (non +) as well as SFP+?

 

I am happy for others to give their own sonic impressions.  That is not a simple question. Finisar is a fine company and folks should be confident in the quality -- indeed "branded" SFP from major network vendors might use Finisar optics.

 

What I *fully* trust is that they meet specifications. That means low noise. The 10G specs are simply much tighter and more detailed than 1g, so I have more confidence that the 10G parts will be very low noise.

 

NICs also, that means when plopped down into a bog PCIe slot supplied by a bog standard ATX supply, the 10G NICs will still be very low noise, and the 100g NICs even lower, which means of course that they implement their own ultralow noise power supplies. Not just low noise but verified ultra low noise.

 

The new SFP28 parts are backward compatible with SFP(+) and I am seeing a trend where they supply 25Gbe, 10Gbe, 5Gbe, 2.5Gbe and 1Gbe so I assume that when run at 1Gbe, these NICs, SFP28 modules get the same low noise board layout, low noise onboard power, low noise onboard clocks etc etc etc 

[barrows]

18 minutes ago, jabbr said:

The new SFP28 parts are backward compatible with SFP(+) and I am seeing a trend where they supply 25Gbe, 10Gbe, 5Gbe, 2.5Gbe and 1Gbe so I assume that when run at 1Gbe, these NICs, SFP28 modules get the same low noise board layout, low noise onboard power, low noise onboard clocks etc etc etc 

Oh, please excuse my pestering and ignorance, but i want to be perfectly clear on this before i make a purchase error:

 

You are saying that SFP+ modules (I have seen that they appear to be able to work at both 10 Gbe and 1 Gbe) will work in regular (non +) SFP cages, right?  So then one can have the extra precision and perhaps low noise benefit of the 10 Gbe transceivers, ruling at lower speed in a  1 Gbe environment, for, perhaps an even greater noise benefit.  I also need to be able to advise Sonore customers, so i "cannot" afford to be wrong in this (Sonore optical products use standard Gigabit SFP cages).

 

Thanks! 

[tgb]

Thanks barrows for this huge question...  & to all for this nice discussion.

dear jabbr, the entire world is waiting for your answer ! :-)

[Superdad]

2 hours ago, barrows said:

You are saying that SFP+ modules (I have seen that they appear to be able to work at both 10 Gbe and 1 Gbe) will work in regular (non +) SFP cages, right? 

 

Generally no.  While one can order dual-rate SFP+ modules that will work at 1 Gigabit in a standard SFP port, most are 10Gbe (or higher) and will not negotiate down when plugged into a 1Gbe SFP port.

However, one can use 1Gbe transceivers in SFP+ ports.

 

Hope that helps.

--Alex C.

[jabbr]

2 hours ago, barrows said:

Oh, please excuse my pestering and ignorance, but i want to be perfectly clear on this before i make a purchase error:

 

You are saying that SFP+ modules (I have seen that they appear to be able to work at both 10 Gbe and 1 Gbe) will work in regular (non +) SFP cages, right?  So then one can have the extra precision and perhaps low noise benefit of the 10 Gbe transceivers, ruling at lower speed in a  1 Gbe environment, for, perhaps an even greater noise benefit.  I also need to be able to advise Sonore customers, so i "cannot" afford to be wrong in this (Sonore optical products use standard Gigabit SFP cages).

 

Thanks! 

Great point. It’s the entire unit that has to be built to 10G in order to need to meet 10G standards. So a 10G switch can have a port run at 1G.

 

There can be power requirement issues, so while an SFP module should work in a SFP(+) port in general you cannot expect an SFP(+) module to work in an SFP port.

 

Short answer: no, longer answer: you could test and certify specific SFP modules for use in their equipment. Intel, for example, publishes a list of SFP(+) modules certified for the x520 NIC

 

The reason I started this thread was to allow people to post specific SFP/SFP(+) modules that work in specific equipment. 

[kerisabe]

So what are the best sfp module to be used in:

 

1. d-link fiber optic converters

2. cisco 2960 48-TC L

 

thank you

[Markus87]

On 4/24/2020 at 12:43 AM, jabbr said:

So these are the Intel branded modules: https://www.intel.com/content/www/us/en/support/articles/000005528/network-and-i-o/ethernet-products.html

 

and you can see that FTLX1471D3BCV is the base Finisar LR module

 

check that your switch supports dual mode -- most should

 

Quote

There are dual speed Finisars 10/1gbs and what I use.

 

Let's see this one is a dual single mode: https://www.finisar.com/sites/default/files/downloads/finisar_ftlx1475d3bcv_10g-1g_dual_rate_10gbase-lr-lw_1000base-lx_10km_1310nm_single_mode_datacom_sfp_optical_transceiver_productspecrevb1.pdf

 

Hi @jabbr,

 

In the posts above you mentioned two different Finisar model numbers FTLX1471D3BCV & FTLX1475D3BCV

I was trying to figure the difference but the specs look almost identical. Both should be 1G/10G dual mode compliant.

 

Is there a relevant difference between both?

Which do you use?

Would you recommend one over the other?

 

Cheers,

Markus

[jabbr]

4 hours ago, Markus87 said:

 

In the posts above you mentioned two different Finisar model numbers FTLX1471D3BCV & FTLX1475D3BCV

I was trying to figure the difference but the specs look almost identical. Both should be 1G/10G dual mode compliant.

 

 

Yes I see, never thought about that. I think Intel OEMs the first so I prob got those, but honestly haven't bought 10G modules (nor 1G) for several years.

[The Computer Audiophile]

Hi @jabbr, will this Intel card work with the Ubiquiti UF-SM-10G-S BiDi SFP+ module?

 

Intel 82599EN Single-Port 10G SFP+ PCIe 2.0 x8, Ethernet Network Interface Card - https://www.fs.com/products/75851.html

 

Ubiquiti UF-SM-10G-S - https://www.ui.com/accessories/fiber-modules-cable/