Optical Network Configurations

[SQFIRST]

1 minute ago, Iving said:

a little birdie told me to use SFP on the eR side.

It is a good idea to go single mode SFP on the ER side. 

[Iving]

5 minutes ago, SQFIRST said:

It is a good idea to go single mode SFP on the ER side. 

 

Understood

[SQFIRST]

1 hour ago, audiobomber said:

How about 10G SFP+ and fiber between the computer and qnap switch, then copper to an opticalModule Deluxe, with 1G Finisar SFP and fiber to the ER. The oMD v2 is phenomenal.

Agreed that leveraging the higher quality FMC will yield benefit. But introducing copper will require including the 10G switch into clean power domain. By leaving the switch as optical only, it can even stay on exposed circuit without affecting the network.

 

47 minutes ago, Quokka_61 said:

'PC > fiber > copper > fiber > copper > device' seems a bit superfluous, especially since it's often advised to use as few as possible steps in an audio chain.

Agreed. The idea suggested was stemming from wanting to leverage the higher quality electronics in the OM.

 

[Jud]

2 hours ago, Iving said:

 

SFP transceiver both sides of switch? OK ty

 

edit: but that's a 10G switch with SFP transceivers both sides (from PC and to eR respectively). In that case wouldn't a 1Gb switch with SFP transceivers be just as suitable?

 

Perhaps I haven't followed ...

 

Another switch would be just as suitable, but would not be built to the 10G spec (lower jitter). The switch I mentioned is inexpensive, built to 10G spec, and leaves room for upgrading in the future.

[audiobomber]

7 hours ago, Quokka_61 said:

'PC > fiber > copper > fiber > copper > device' seems a bit superfluous, especially since it's often advised to use as few as possible steps in an audio chain.

Fewest steps makes sense in an analogue audio chain, but network audio benefits from extra steps, if implemented properly. If less were better, we wouldn't use a switch and just use the router as a switch. Anyone who knows about network audio will tell you to use a switch, and most of us use and recommend more than one.

 

I am totally skeptical that a consumer grade 10g switch can compare with a product designed for audio, like the ER or opticalModule Deluxe. Putting an OMD between the 10G switch and ER makes sense to me, because the signal gains whatever it can from the 10G conversion, then benefits from the refinement of the Swenson-designed oMD. 

[dbastin]

5 hours ago, audiobomber said:

I am totally skeptical that a consumer grade 10g switch can compare with a product designed for audio, like the ER

Just for interest, I recently swapped the order from

ER > fibre > CRS 305 to

CRS 305 > fibre > ER

 

Nothing else was changed.  CRS 305 is my router.  No wired ethernet is connected to it except WAN via Synergistic Research Atmosphere X Reference which has the sheil disconnected at tge downstream end, and connected to ground.  Upstream is a mandatory national broadband network termination device (eg like a modem). Downstream is Antipodes CX Oladra.

 

The difference in sound is not so obvious a novice would immediately notice.  It took me a while get a fix on the difference, some of that was probably 'clock' type things getting back towards optimal.  Refinement is a great descriptor - in that sense ER Is 'better'.

 

Both the ER and CRS 305 sit on Synergistic Research Tranquility POD (original) which in my view have a greater impact than the difference between ER and CRS 305 -the difference may have been greater without the PODs.  I'd guess the CRS 305 with a POD is really quite competitive with ER without a POD.

 

I consider PODs are really worthwhile addition to the network and CX which has a new Carbon version which 'conditions' my battery powered WAP wifi connection to Devialet.  I focus the PODs on network devices that are not available in 'audiophile' products where I moved them to after they were replaced by Tranquility Bases used with other components.

[audiobomber]

2 hours ago, dbastin said:

Just for interest, I recently swapped the order from

ER > fibre > CRS 305 to

CRS 305 > fibre > ER

 

Nothing else was changed.  CRS 305 is my router.  No wired ethernet is connected to it except WAN via Synergistic Research Atmosphere X Reference which has the sheil disconnected at tge downstream end, and connected to ground.  Upstream is a mandatory national broadband network termination device (eg like a modem). Downstream is Antipodes CX Oladra.

 

The difference in sound is not so obvious a novice would immediately notice.  It took me a while get a fix on the difference, some of that was probably 'clock' type things getting back towards optimal.  Refinement is a great descriptor - in that sense ER Is 'better'.

 

Both the ER and CRS 305 sit on Synergistic Research Tranquility POD (original) which in my view have a greater impact than the difference between ER and CRS 305 -the difference may have been greater without the PODs.  I'd guess the CRS 305 with a POD is really quite competitive with ER without a POD.

 

I consider PODs are really worthwhile addition to the network and CX which has a new Carbon version which 'conditions' my battery powered WAP wifi connection to Devialet.  I focus the PODs on network devices that are not available in 'audiophile' products where I moved them to after they were replaced by Tranquility Bases used with other components.

What power supplies do you use on the switches? Which SFP's are you using?

[Quokka_61]

9 hours ago, audiobomber said:

Fewest steps makes sense in an analogue audio chain, but network audio benefits from extra steps, if implemented properly. If less were better, we wouldn't use a switch and just use the router as a switch. Anyone who knows about network audio will tell you to use a switch, and most of us use and recommend more than one.

 

@audiobomber thank you for pointing out an ambiguity in where I wrote that 'it's often advised to use as few as possible steps in an audio chain.'. In hindsight I should have used the words 'as few as necessary'. What's necessary is for every audiophile to discover and decide, and this is why getting at a solution for a given problem is often such a challenge. However, nowhere did I write that one shouldn't use a switch. In my two posts prior to the one you quoted, I even referred to the two switches that are part of my audio chain, a QNAP QSW-M408-MC and a MikroTik CRS305-1G-4S+ IN. What I am still not sure of, however, is how much sense it makes to get rid of electromagnetic interference by introducing a fiber tract in the audio chain, only to introduce the same interference, at whatever level, by again adding copper downstream, necessitating a second tract of fiber to again get rid of electromagnetic interference introduced in the step before last. If I have understood anything of the many hundreds of pages I have read at this forum and recommendations by specialists like Jussi Laako, a single tract of fiber at a stage as late as possible in the audio chain, may well lead to the same outcome. But just to be sure, I am far from being an authority on the subject, and if the chain you described and the opticalModule Deluxe you referred to gives you the sound quality you strive to achieve, fair enough and good for you, there's surely more than one road to Rome.  

[SQFIRST]

17 hours ago, Jud said:

built to the 10G spec (lower jitter)

That's a key point, thank you.

It is essential that device is IEEE 802.3ae compliant for 10G benefit.

 

[dbastin]

21 hours ago, audiobomber said:

What power supplies do you use on the switches? Which SFP's are you using?

The SFPs are Cisco 1G from Afterdark which have some sort of tape on them.  I intend to change those when I have time to re-read what works between 10G CRS 305 and 1G ER - as I recall we either use 1G both ends (and tse the 'best' 1G sounding, eg. Cisco AOC) or shell out heaps for SFP28 to autonegotiate the ER end to 1G.

 

The LPSs are W4S PS1 at downstream end (which is powered via dedicated line, Synergistic Research (SR) Powercell SE and Atmosphere and Galileo SX power cords).  The upstream end is Antipodes ODAPS with fairly basic Furutech power cord into a wall outlet on a general household circuit.  The downstream device also has a SR DC SX cable.

 

Some of the benefits of this arrangement is that the devices either end of fibre are on totally seperate power lines.

[jabbr]

On 2/12/2024 at 6:25 AM, dbastin said:

The SFPs are Cisco 1G from Afterdark which have some sort of tape on them.  I intend to change those when I have time to re-read what works between 10G CRS 305 and 1G ER - as I recall we either use 1G both ends (and tse the 'best' 1G sounding, eg. Cisco AOC) or shell out heaps for SFP28 to autonegotiate the ER end to 1G.

 

Ha ha I'm not sure there is a need for folks to "shell out heaps for SFP28" modules because you'll also need a 25G switch.

 

To be 100% clear, and to reduce the need for folks to read through every post in this thread: I am *not* claiming that 10G "sounds better" than 1g. Both with fiberoptic will eliminate transmission of common mode noise through the cable. When folks theorize about phase noise being an issue, or need for fancy clocks etc, I have pointed out that as of the 10G+ standards, phase noise must be controlled in the end to end connection including rejection of upstream noise. This is by design. It doesn't mean that a 1g device can't be designed to reject upstream noise but a 10G device **must** reject upstream noise. That's all.

 

I've also pointred out that the Finisar devices are state of the art (as are Avago) but that doesn't mean that corresponds to improved SQ --- people seem to like them though.

 

On 2/12/2024 at 6:25 AM, dbastin said:

 

The LPSs are W4S PS1 at downstream end (which is powered via dedicated line, Synergistic Research (SR) Powercell SE and Atmosphere and Galileo SX power cords).  The upstream end is Antipodes ODAPS with fairly basic Furutech power cord into a wall outlet on a general household circuit.  The downstream device also has a SR DC SX cable.

 

Some of the benefits of this arrangement is that the devices either end of fibre are on totally seperate power lines.

 

Exactly! I don't worry about special PSUs in my server room, indeed you can't swap PSU in my 100G switch and my 100G cards have on board linear power chips etc. A big advantage of fiber is that you can use noisy servers in noisy server rooms (my basement) on one power line and transmit clean signal to clean power in your listening area (mine is behind a balanced isolating transformer)

[dbastin]

15 minutes ago, jabbr said:

A big advantage of fiber is that you can use noisy servers in noisy server rooms (my basement) on one power line and transmit clean signal to clean power in your listening area

Unfortunately my recent experience demonstrated that to be an incorrect assumption in my case.  I made several changes to the upstream fibre device (which was initially ER) and to the device upstream of that and I couid notice a difference with every change.  Actually, the changes I made have together gained so much improvements it is the most beneficial expenditure I have made on any part I my hifi system in quite a long time when comparing with other similar expenditures.

 

More interesting than that surprising discovery is that I also use wifi to connect to my endpoint (Devialet Pro), so the recent changes have translated to being audible despite fibre and wifi isolation.

 

In short the changes were installing Synergistic Research Tranquility PODs and MIG 2, Ground wires and ground block, a better shelf and Furutech AC.

 

So the general tip from this is it might pay to experiment to test assumptions.

 

37 minutes ago, jabbr said:

I'm not sure there is a need for folks to "shell out heaps for SFP28" modules because you'll also need a 25G switch.

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.

[audiophilac]

I think, please use the dual rate SFP transceiver between the Mikrotik CRS305 and etherRegen.

[audiobomber]

1 hour ago, jabbr said:

 

Ha ha I'm not sure there is a need for folks to "shell out heaps for SFP28" modules because you'll also need a 25G switch.

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.

 

1 hour ago, jabbr said:

I don't worry about special PSUs in my server room

I have to agree with @dbastin, changes upstream of the fiber optic conversion are clearly audible, including switches, power supplies, ethernet cables and SFP's. 

[audiom3]

6 minutes ago, 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. 

I think you might be over-simplifying your 10G comment and proof just a bit.  Three years ago when I switched my network to fiber, very little was known vs today...  Many of us didn't take anything as fact in our experimentation.  We spent loads of time, effort and expense in finding what worked best for us and in our setups.  Heck, I have no way or even the desire to measure to see if 10G has lower jitter than 1G.  My proof that 10G is better in my setup (audio and video in my case) lies in my experimentation.  Seeing/hearing is believing.

As this entire thread provides real-world feedback while performing heaps of trial and error, everything should be taken with a grain of salt.  It's up to any/everyone to experiment for themselves, in their setup to hear what sounds best for them...  Sure, we all read posts and even try product A or B which comes recommended.  But in the end, we have to determine what is/was better for us.  I've had an eR and sold it.  It's beloved on so many forums but it ended up not being 'all that' in my setup.  Not only in terms of sound quality but in terms of system/network hiccups.  Roon did not care for the eR and would skip songs long before the end.  Since removing, I have never, ever had Roon FF to the next track until the current track ended. 

I currently have and utilize a Sonore oM.  And as obvious as some transceivers make in audibility, or SM vs MM fiber types, linear vs SMPSUs, I could not tell you if the oM sounds any better than any of my 10Gtek FMCs (I have other 'cheapies' too) with the same linear power supply being used.  One day, I will try to do a thorough A/B between the two but I have no desire to do so right now.  I already own the Sonore, so it's easier to just keep it in the chain because it works.

[Jud]

1 hour ago, 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.

 

I have to agree with @dbastin, changes upstream of the fiber optic conversion are clearly audible, including switches, power supplies, ethernet cables and SFP's. 

 

This is a bit too much generalization from a small and not always relevant universe. At least some of the use of 100M came about because at the time of design (a few years ago), higher speed parts were electrically noisier. Currently that is very probably no longer the case, and with optical connections it doesn't matter anyway.

[Iving]

10G experiment unambiguously disappointing! Comments welcome!

 

System is PC > etherRegen > Device (dedicated PC plays flac to RedNet Device - offline Dante network)

 

Default/benchmark was: PC > Startech PEX1000SFP2 > fiber SM cable > eR > Device

I had gotten this sounding truly superlative. After years of personal investment (including W10 Pro tuning 'n' all that). But just as SQ peaking, got latency blips/audible dropouts. After painful investigation turned out the culprit was the Card. Happened to be Realtek.

 

Last stages of the investigation involved swapping SM fiber for AOC (which successfully ruled out Startech SFP1000LXST Transceivers). Curious (let's say provisional) jump in SQ. I say curious because SM supposed to be better, no?

 

Anyway - el cheapo 10GTek i120 Card (which ruled in Startech Card as problem) was superlative regards latency - ultra stable - but SQ was coarse.

I have a shortlist - surprisingly short actually - of quality 1Gbe Cards that could replace the PEX1000SFP2. I tried a Startech P021GI-NETWORK-CARD (i350). The two ports didn't behave identically. Neither particularly stable wrt latency - but certainly not faulty either. I was too busy diagnosing to notice SQ with this Card - but I think probably it was OK.

 

Then a 10G experiment after some chat here ...

 

Installed 10G Startech PEX20000SFPI no problem. To my surprise, couldn't get it to work direct to eR with Startech SFP1000LXST Transceivers/SM fiber. I expected 1Gbe Transceivers to be "backwards compatible" (against 10G PEX20000SFPI) and play nice with eR being 1Gbps. No dice. If I insert this Transceiver into PEX20000SFPI, all I get is an exclamation mark in DevMan ("Windows stopped device because it has problem"). I tried setting Speed & Duplex to 1Gb, but no joy. Could I have missed something?

 

Tried AOC from PEX20000SFPI to QNAP QSW-2104-2S – and other side of QNAP - SM fiber with Startech SFP1000LXST Transceivers to eR. @Jud was right! It "just worked"! Sat down in eager anticipation.

 

Now I know the Card and Switch are not broken in.

And the QNAP is on its basic adapter - no LPS.

But the SQ is garbage. Sounds like my system did years go. Just unrefined. I can almost hear real distortion listening to Nana Mouskouri sing (yes I do that - provided she is singing this version of "I Gave My Love A Cherry").

 

[btw - tried RJ45 copper from QNAP straight to Device (i.e. no eR). Also worked. SQ similarly sub par.]

 

In my own estimation it's exceedingly unlikely that hardware "break in" - or QNAP switch being on stock PSU - accounts for this. Everything "works". Dante Latency Graph solid. Just sounds soulless. No body. As if all the LPSs and mains noise mitigators were in my system were taken out - or something.

Hoping you guys can give me radical insight, without which I shall have to abandon the 10G choo choo train and go back to 1Gbe with much relief. Many thanks

[SQFIRST]

Nice to see your update @Iving

 

That is the type of difference between managed vs unmanaged and manufacturer compatibility specifications. In this case the switch allowed backward compatibility but the pc card strangely did not. 

 

Sounds like you can chuck the 10G distraction and get on with your search for a new card. Your hard work on your PC is showing as it gives you a great signal. 

 

Thank you for the enjoyable song from Nana!

[Jud]

1 hour ago, Iving said:

10G experiment unambiguously disappointing! Comments welcome!

 

System is PC > etherRegen > Device (dedicated PC plays flac to RedNet Device - offline Dante network)

 

Default/benchmark was: PC > Startech PEX1000SFP2 > fiber SM cable > eR > Device

I had gotten this sounding truly superlative. After years of personal investment (including W10 Pro tuning 'n' all that). But just as SQ peaking, got latency blips/audible dropouts. After painful investigation turned out the culprit was the Card. Happened to be Realtek.

 

Last stages of the investigation involved swapping SM fiber for AOC (which successfully ruled out Startech SFP1000LXST Transceivers). Curious (let's say provisional) jump in SQ. I say curious because SM supposed to be better, no?

 

Anyway - el cheapo 10GTek i120 Card (which ruled in Startech Card as problem) was superlative regards latency - ultra stable - but SQ was coarse.

I have a shortlist - surprisingly short actually - of quality 1Gbe Cards that could replace the PEX1000SFP2. I tried a Startech P021GI-NETWORK-CARD (i350). The two ports didn't behave identically. Neither particularly stable wrt latency - but certainly not faulty either. I was too busy diagnosing to notice SQ with this Card - but I think probably it was OK.

 

Then a 10G experiment after some chat here ...

 

Installed 10G Startech PEX20000SFPI no problem. To my surprise, couldn't get it to work direct to eR with Startech SFP1000LXST Transceivers/SM fiber. I expected 1Gbe Transceivers to be "backwards compatible" (against 10G PEX20000SFPI) and play nice with eR being 1Gbps. No dice. If I insert this Transceiver into PEX20000SFPI, all I get is an exclamation mark in DevMan ("Windows stopped device because it has problem"). I tried setting Speed & Duplex to 1Gb, but no joy. Could I have missed something?

 

Tried AOC from PEX20000SFPI to QNAP QSW-2104-2S – and other side of QNAP - SM fiber with Startech SFP1000LXST Transceivers to eR. @Jud was right! It "just worked"! Sat down in eager anticipation.

 

Now I know the Card and Switch are not broken in.

And the QNAP is on its basic adapter - no LPS.

But the SQ is garbage. Sounds like my system did years go. Just unrefined. I can almost hear real distortion listening to Nana Mouskouri sing (yes I do that - provided she is singing this version of "I Gave My Love A Cherry").

 

[btw - tried RJ45 copper from QNAP straight to Device (i.e. no eR). Also worked. SQ similarly sub par.]

 

In my own estimation it's exceedingly unlikely that hardware "break in" - or QNAP switch being on stock PSU - accounts for this. Everything "works". Dante Latency Graph solid. Just sounds soulless. No body. As if all the LPSs and mains noise mitigators were in my system were taken out - or something.

Hoping you guys can give me radical insight, without which I shall have to abandon the 10G choo choo train and go back to 1Gbe with much relief. Many thanks

 

I wish I had "radical insight," but nothing occurs at the moment. 🙁 Very sorry this didn't work out in your system. I'm very happy with the SQ in mine, so either there's some other difference in our systems, or between our ears! 🙂

 

I hope at the end of all this you find both reliability and the sound quality you desire.

[Iving]

29 minutes ago, Jud said:

I wish I had "radical insight," but nothing occurs at the moment. 🙁 Very sorry this didn't work out in your system. I'm very happy with the SQ in mine, so either there's some other difference in our systems, or between our ears! 🙂

 

I hope at the end of all this you find both reliability and the sound quality you desire.

 

No doubt our systems are constructed very differently, and the respective components play different parts of different Gestalts.

 

My experiment is still interesting tho'. Explanations?

- something to do with feeding an eR

- something to do with Dante/RedNet

- something to do with QNAP switch "noise" being on a cheap stock ps adapter

- something to do with PC > AOC > switch > SM fiber - i.e. two different cable types and/or the "conversion" in between.

But still mine is just a PC > Device hop via an eR and prima facie result says 10G stringent specs not necessarily an audio advantage.

 

btw - been swapping about -

(i) put the spiking (but otherwise Ace) Card back in just to check my ears/sanity - I'm not in doubt -

and

(ii) whilst AOC presents "something" in the 1Gb hop PC > eR, it sounds like it's waiting to settle in and won't. The SM fiber is more natural and that'll be part of how I try to get back to where I was.

[Quokka_61]

2 hours ago, Iving said:

If I insert this Transceiver into PEX20000SFPI, all I get is an exclamation mark in DevMan ("Windows stopped device because it has problem"). I tried setting Speed & Duplex to 1Gb, but no joy. Could I have missed something?

 

Does the driver software for the PEX20000SFPI card offer the option 'Auto-negotiate'? If so, did you check it, in addition to setting Speed & Duplex to 1Gb? I don't know whether this applies to your hardware, but I do remember that not setting this option in the OS of my MikroTik CRS305-1G-4S+, caused me quite some hassle before I discovered checking this option made all the difference between dead silence from my loudspeakers and a working fiber connection...

[Iving]

7 minutes ago, Quokka_61 said:

 

Does the driver software for the PEX20000SFPI card offer the option 'Auto-negotiate'? If so, did you check it, in addition to setting Speed & Duplex to 1Gb? I don't know whether this applies to your hardware, but I do remember that not setting this option in the OS of my MikroTik CRS305-1G-4S+, caused me quite some hassle before I discovered checking this option made all the difference between dead silence from my loudspeakers and a working fiber connection...

 

Yes. Auto neg. is default.

Tried 1Gb setting as fix.

[SQFIRST]

On 2/13/2024 at 5:26 PM, Iving said:

My experiment is still interesting tho'. Explanations?

 

Sharing some thoughts and not any explanation as such. I learned from reading the forums here and experimenting and trying things out has been helpful for me.

 

There was a chance that 10G would not have an impact in your setup as it is a compact network with one source pc. If there were no network jitter issues to begin with then the extra switch will be redundant and as you are finding, can affect negatively with its overhead.

The ER is effective as the last network device and I do not see it being as a factor in the 10G experiment. Rather, in my setup it serves as the 1G refinement device after the 10G cleaning. The AOC vs SM consideration is similar to what I have observed which is that MM fiber (850nm) sounds slightly muted compared to SM Fiber (1310nm) which seems to allow full dynamic range. Which one sounds better probably depends on which sound profile the audio gear leans towards. You can swap out the AOC for a SM 10G pair of transceivers to further test that out if you feel inclined.

 

In my more 'open' network, with Internet streaming, multiple Lan servers and Wifi; the sound with and without 10G is very different and in favor of 10G.

 

Appreciate the fact that you took the time to check it out and share your findings. Hope you get your setup back up and running soon.

[Iving]

7 hours ago, SQFIRST said:

 

Sharing some thoughts and not any explanation as such. I learned from reading the forums here and experimenting and trying things out has been helpful for me.

 

There was a chance that 10G would not have an impact in your setup as it is a compact network with one source pc. If there were no network jitter issues to begin with then the extra switch will be redundant and as you are finding, can affect negatively with its overhead.

The ER is effective as the last network device and I do not see it being as a factor in the 10G experiment. Rather, in my setup it serves as the 1G refinement device after the 10G cleaning. The AOC vs SM consideration is similar to what I have observed which is that MM fiber (850nm) sounds slightly muted compared to SM Fiber (1310nm) which seems to allow full dynamic range. Which one sounds better probably depends on which sound profile the audio gear leans towards. You can swap out the AOC for a SM 10G pair of transceivers to further test that out if you feel inclined.

 

In my more 'open' network, with Internet streaming, multiple Lan servers and Wifi; the sound with and without 10G is very different and in favor of 10G.

 

Appreciate the fact that you took the time to check it out and share your findings. Hope you get your setup back up and running soon.

 

Interesting reflections and can't help but agree all of it.

 

Yes it's not an eR thing. As reported I also went QNAP straight to Device (no eR) and the SQ was similarly compromised.

 

If I may say, I think my first instincts were right. All I did was add hops and noise.

 

Whatever theoretical (or actual) advantage afforded by 10G was outweighed by disruption of whatever kind ...

 

... in my system!

 

Thanks for your re-iterated good wishes and these are returned in abundance!

[SQFIRST]

 

On 2/13/2024 at 9:31 AM, jabbr said:

A big advantage of fiber is that you can use noisy servers in noisy server rooms (my basement) on one power line and transmit clean signal to clean power in your listening area

This has been one of the most beneficial advantage with 10G introduction in my setup. Prior to that when I would read such statements it was a bit of a head scratcher. It just made sense to actually try a 10G fiber end to end path. It isolates fully. No amount of speculation can provide clarity and one has to experience for themselves.

 

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.