Best audiophile switch

[italia]

Let me just throw in my two cents – I think optical isolation is overrated. I’ve tried it in my setup (at various points in the network), and it was always a trade-off: slightly better precision at the expense of tonal richness and fluidity.

I feel it's difficult to generalize about optical (galvanic) isolation as there are many potential variables.

1) What is it being compared to? Pure ethernet connection from router through to streamer? What grade of ethernet cable? Length of run? Signal degradation of long runs of ethernet can be an issue.
Or Wifi connection? I believe Wifi invokes 'noise' that make it the worst of the connection to streamer options.

2) What specific equipment and cabling is being utilized for the 'ethernet to fiber to ethernet' isolation path? There's a significant difference in low end fiber media converters versus audiophile switches with SFP ports. Likewise there is a noticeable difference in SFP transceivers, fiber optic cable, and fiber optic cable connectors.

I replaced a wifi based streaming setup where the streamer was connected with a 1 meter ethernet cable to a mesh satellite unit, with an 'ethernet to fiber to ethernet' set up (void of wifi) utilizing a 30 meters long fiber optic cable. The reduction in noise and improvement in sound quality in my system was noticeable. Upgrading FMC's, switches, SFP transceivers, fiber cable/ connectors along the way elevated my sound quality.

This said, if my router were 1 meter from my streamer/ audio system, I suspect a pure ethernet connection void of galvanic isolation would be a solid option.

 

[kennyb123]

Let me just throw in my two cents – I think optical isolation is overrated. I’ve tried it in my setup (at various points in the network), and it was always a trade-off: slightly better precision at the expense of tonal richness and fluidity.

This can certainly be true about optical. But it can also NOT be true about optical. Implementation matters and performance can fall anywhere between dreadful and wonderful depending on the care taken. This is true about everything in this hobby.

Where optical may differ though is that it may have a ceiling. My own experience tells me that well-implemented optical may have fewer tradeoffs compared to copper up to a certain price point, but copper can then scale well beyond this - but that can be prohibitively expensive at longer lengths.

I considered my fiber span to be well-implemented once it got it to the point where I no longer heard it doing harm relative to what I heard with it disconnected. And yet, it got seriously trounced by what I believe was a $3500 run of QSA Lanedri Gamma Infinity. That was significantly more than I spent on my fiber span so it should not be surprising that this copper run was significantly better. The key difference was transparency. The QSA-treated cable just did a better job of getting out of the way of the music. It didn’t seem as if there were any promising options that would have taken my fiber to a level where it would be competitive with this QSA Ethernet cable, but I can’t say for sure. Regardless, I have some of QSA Lanedri’s Discovery Veridian incoming and that will be evaluated against my fiber span once it arrives. This will have a total cost at around $1150 as I need it to span 8m, so it’ll be interesting to hear how it compares.

 

[NigelB]

As has been said, although optical is theoretically the way to go because the "cable" itself can carry no noise, it's all about implementation. The downside, perhaps the only downside, is that in some FMC implementations the RFI noise generated by the back conversion from optical to electrical undoes a good measure of the benefit of the optical connection by introducing noise of its own. This is why practice doesn't always live up to the theoretical promise and we hear mixed reports.

It doesn't, have to be either/or, of course. In fact, I know a good few people for whom their sonic nirvana combines the two technnologies highly effectively:

> long optical from router > high quality FMC with high quality PSU > high quality short network cable > high quality switch with high quality PSU > high quality short network cable > streamer (hopefully a high quality streamer!).

In this scenario, the optical link has done the heavy lifting and all the switch needs to clean up is the noise which the FMC's back conversion generates; as this is a high quality FMC, there isn't much of this.

 

[Mr. Heptamer]

I made several trials with switches. I started with Melco S10, Reiki, Tempus. All of them gave a certain improvement. But the very last one, the Taiko extreme switch showed the best performance in my setup. I have only to feed the server/streamer and nothing else. Nevertheless a personal test in the own audio setup is always recommended.

 

[Di-fi]

I agree that optical looks perfect on paper, but in practice the implementation (FMCs, PSUs, cabling/length) sets the ceiling. Either a well-implemented fibre span eventually hits its limits compared to high-end copper, or the RFI from back conversion undermines the theoretical advantage unless it’s dealt with carefully.

That’s why Wackerd’s result with the Taiko switch is interesting. Since it has an SFP port and can take different transceivers (on both ends), it may well be reducing that back-conversion noise to a much lower level. @wackerd , if you don’t mind sharing a bit more about how you’re using the Taiko’s SFP input in your switch, I think it would really help the rest of us connect the dots between theory and practice.

 

[Lee]

Have you tried grounding the UEF?

I use SR’s Active Ground Block on mine and it helps performance.

 

[MarkusBarkus]

sharing a bit more about how you’re using the Taiko’s SFP input in your switch, I think it would really help the rest of us connect the dots between theory and practice.

...there is a good bit of info on the Taiko web:

https://taikoaudio.com/taiko-2020/wp-content/uploads/2025/02/Extreme-Network-Card-Switch-Summary-FAQ_v53.x80569.pdf

Briefly, they recommend the DAC cable due to power consumption and noise related to power consumption. They say a single SFP consumes more power than their entire switch design. There is also info regarding SFP vs. RJ45 processing overhead.

I was using optical before adding the Taiko switch, and later their router. I continued long after adding both components, contrary to recommendation, because sometimes I'm just like that.

Eventually, I tried the recommended set-up, using the DAC cables, and they were/are better here.

 

[NigelB]

...there is a good bit of info on the Taiko web:

https://taikoaudio.com/taiko-2020/wp-content/uploads/2025/02/Extreme-Network-Card-Switch-Summary-FAQ_v53.x80569.pdf

Briefly, they recommend the DAC cable due to power consumption and noise related to power consumption. They say a single SFP consumes more power than their entire switch design. There is also info regarding SFP vs. RJ45 processing overhead.

I was using optical before adding the Taiko switch, and later their router. I continued long after adding both components, contrary to recommendation, because sometimes I'm just like that.

Eventually, I tried the recommended set-up, using the DAC cables, and they were/are better here.

Interesting. So for clarity, even with presumably a very good implementation, you find that you prefer non-optical to optical. Of course you are still using the SFP ports on both devices as the Direct Attach Copper (DAC - an unfortunately familiar acronym!) cable requires this.

Do both devices also have RJ45 ports? If so, have you compared SFP port > DAC cable > SFP port with RJ45 port > copper ethernet > RJ45 port?

Thanks

 

[Di-fi]

...there is a good bit of info on the Taiko web:

https://taikoaudio.com/taiko-2020/wp-content/uploads/2025/02/Extreme-Network-Card-Switch-Summary-FAQ_v53.x80569.pdf

Briefly, they recommend the DAC cable due to power consumption and noise related to power consumption. They say a single SFP consumes more power than their entire switch design. There is also info regarding SFP vs. RJ45 processing overhead.

I was using optical before adding the Taiko switch, and later their router. I continued long after adding both components, contrary to recommendation, because sometimes I'm just like that.

Eventually, I tried the recommended set-up, using the DAC cables, and they were/are better here.

Thanks for sharing. So you were running fibre using the Taiko switch’s only two SFP ports before switching to DAC cable, right? I wonder which transceivers you used, since Taiko notes that power consumption and module quality can affect noise.
This fits with what others observed: RFI from back conversion can reduce fibre’s theoretical benefit, and even good SFP transceiver setups may vary. I think SFP quality explains much of the mixed feedback on fibre: power draw, back-conversion electronics, and how the PSU, switch/router, and cable lengths interact all matter.
With the Taiko switch’s ultra-low-noise design* (and likely other switches with a sfp cage) , SFP transceivers can actually add more noise than they remove, so here DAC cables often give the cleanest results. Note that the streamer/dac needs a sfp port to make this work.

*A single fiber SFP module consumes more power than our entire switch design, and the additional noise this generates is multiple times higher than that of the switch itself.

 

[Mr. Heptamer]

I agree that optical looks perfect on paper, but in practice the implementation (FMCs, PSUs, cabling/length) sets the ceiling. Either a well-implemented fibre span eventually hits its limits compared to high-end copper, or the RFI from back conversion undermines the theoretical advantage unless it’s dealt with carefully.

That’s why Wackerd’s result with the Taiko switch is interesting. Since it has an SFP port and can take different transceivers (on both ends), it may well be reducing that back-conversion noise to a much lower level. @wackerd , if you don’t mind sharing a bit more about how you’re using the Taiko’s SFP input in your switch, I think it would really help the rest of us connect the dots between theory and practice.

Well, I use the switch SFP to connect a RJ45 Ethernet cable coming from the FB router (Fritzbox without SFP). And then again a RJ45 from output to the server/streamer which has no SFP, unfortunately. My former SFP trials by use of RJ45 converter to bridge the 2m distance from the FB router to the switch ended up without audible advantages. But one must know that I switched off the WiFi in FB router to avoid disturbances by high power consum. The Wifi is generated by an access point far away from my audio system.

 

[NigelB]

Thanks for sharing. So you were running fibre using the Taiko switch’s only two SFP ports before switching to DAC cable, right? I wonder which transceivers you used, since Taiko notes that power consumption and module quality can affect noise.
This fits with what others observed: RFI from back conversion can reduce fibre’s theoretical benefit, and even good SFP transceiver setups may vary. I think SFP quality explains much of the mixed feedback on fibre: power draw, back-conversion electronics, and how the PSU, switch/router, and cable lengths interact all matter.
With the Taiko switch’s ultra-low-noise design* (and likely other switches with a sfp cage) , SFP transceivers can actually add more noise than they remove, so here DAC cables often give the cleanest results. Note that the streamer/dac needs a sfp port to make this work.

*A single fiber SFP module consumes more power than our entire switch design, and the additional noise this generates is multiple times higher than that of the switch itself.

Much harder to argue with an assertion containg the word "can" than one which contains "will"!

The choice of SFP transceiver is important, yes, but I doubt there is a generality here that because SFP transceivers can be noisy, DAC is typically better than optical. Most people find optical to bring significant benefits. I'd suggest anyone thnking about trying a Direct Attach Copper cable instead of 2 SFP's and fibre/fiber compares DAC connection not only with optical but with optical-followed-by-a-switch. For many, this really is the best of both worlds, combining two noise-mitigating technologies.

 

[Di-fi]

For many, this really is the best of both worlds, combining two noise-mitigating technologies.

Agreed. That’s exactly why it would be interesting to hear from more Taiko switch users. The Taiko is unusual in having only two SFP ports, and yet some users report preferring DAC cables over fibre. Do they still keep optical further upstream? If not, of course, using a DAC cable to a streamer means giving up the optical isolation (and the transceivers).
So results will depend on how the different noise floors play out in each setup. What it does show, though, is that SFP transceivers can introduce noise, which also ties into Kenny’s point that optical/transceivers, while very effective up to a certain point, can hit a ceiling compared to copper.

 

[NigelB]

Agreed. That’s exactly why it would be interesting to hear from more Taiko switch users. The Taiko is unusual in having only two SFP ports, and yet some users report preferring DAC cables over fibre. Do they still keep optical further upstream? If not, of course, using a DAC cable to a streamer means giving up the optical isolation (and the transceivers).
So results will depend on how the different noise floors play out in each setup. What it does show, though, is that SFP transceivers can introduce noise, which also ties into Kenny’s point that optical/transceivers, while very effective up to a certain point, can hit a ceiling compared to copper.

The whole point of having a separate switch after the FMC is that the noise from the optical-electrical back conversion is generated in the upstream device and the downstream switch then cleans up whatever conducted RFI comes out of the FMC on the copper ethernet cable. If these are in the same device, as with the Taiko, which is connected to the streamer, the noise from the back conversion is also directly connected to the streamer. This won't perform as cleanly as keeping the FMC one step removed from the streamer.

I know a few people who have compared a switch upstream of an FMC with an FMC upstream of a switch. The latter has always won the day, and I can't say I'm surprised.

All the best.

 

[Di-fi]

The whole point of having a separate switch after the FMC is that the noise from the optical-electrical back conversion is generated in the upstream device and the downstream switch then cleans up whatever conducted RFI comes out of the FMC on the copper ethernet cable. If these are in the same device, as with the Taiko, which is connected to the streamer, the noise from the back conversion is also directly connected to the streamer. This won't perform as cleanly as keeping the FMC one step removed from the streamer.

I know a few people who have compared a switch upstream of an FMC with an FMC upstream of a switch. The latter has always won the day, and I can't say I'm surprised.

All the best.

I see your point. Keeping the FMC one step removed from the streamer makes a lot of sense for many, and examples support that. At the same time, it does make me wonder more generally: why do companies besides Taiko, like Network Acoustics (Tempus) and XACT (N1), include dedicated SFP ports on their switches? They must see a path to making it work well in practice. Personally, I like the idea of integration if SQ holds up, since it keeps the system simpler. Maybe others here who’ve used those designs can share how they made best use of the SFP input.
Cheers.

 

[agisthos]

I see your point. Keeping the FMC one step removed from the streamer makes a lot of sense for many, and examples support that. At the same time, it does make me wonder more generally: why do companies besides Taiko, like Network Acoustics (Tempus) and XACT (N1), include dedicated SFP ports on their switches?

They offer the SFP port mainly to be used as the incoming data line from the router, not so much as the outgoing to a streamer with SFP capability.
Note both Network Acoustics and XACT recommend using the LAN ports for connection to a streamer.

 

[darkstar_xiii]

After swapping the positions of my switch and optical isolator, I settled on this config as sounding best (to my ears):

Reiki SuperSwitch Master Pro → LHY Optical Isolation → Antipodes K50

I found this to provide vocal clarity and tone that aligned best with my personal preferences. The differences were small, however.

 

[MarkusBarkus]

...I have used a few different SFPs in my network over the past five to seven years, or so, as well as a few different optical cables and FMCs.

For 2-3 years I used a Sonore OM, which was excellent, and much better than the cheaper Amazon off-shore options. I also used the high-end Sonore power supply with the OM.

The network experiments began with daisy-chaining the "expired" Cisco 2960 switches. I settled on two, but had as many as four during my investigations. I also used the UpTone eRG switch, which I still have/use in a different capacity. I had ethernet filters, USB cleaners, various cables, expensive or not.

I swapped over to fiber because I had a 35' run from the network closet, through the walls/ceiling, to the basement audio room.

The original copper network ethernet cable passed through spaces that had Romex wiring, can-light "pig-tails" and other lines not run by me. I thought it might be a bit of an electro-magnetic soup.

I am also about a quarter-mile away from a company that develops sat comm products for military and civilian use. Oh, the antennas I have seen! And Humvees testing products, etc.

I assumed I was in a great location for RF/EM traffic.

However, not having the tools or knowledge, I couldn't practically test this impact in my home. So, I thought it might be a best-practice to make the 35' run with optical, to reduce the exposure of my network to all this junk. That was the genesis of my investigations with optical solutions.

Eventually, fiber infrastructure came to my neighborhood, so I added this to the home, in addition to the copper ISP. Audio is on the fiber feed; everything else is on the copper. Sounds great.

I had previously tried to create a virtual version of this concept using the EdgeX device, but that was a huge PITA. I never did get it working, even with help. IMO this device is not user-friendly, unless you have more-than-casual networking skills.

Eventually, the Taiko network products arrived, and those guys seemed to have a different take on network noise and solutions for audio.

I added the Taiko Switch to my network, and switched back to copper. BTW, this is easy for me, because I ran CAT6, CAT8, SM and MM fiber at the same time. Materials are cheap. Labor/time is expensive. I ran it myself, so easy to manage (for me).

For about a year, or so, I had a modded Buffalo switch upstream from the Taiko Switch, contrary to general guidance to use only the Taiko product.

Eventually, I removed the Buffalo, and just used the Taiko Switch. After about a year, I added the Taiko Router, and went with the DAC cabling suggested by Taiko. I retain that set-up currently, using OEM adapters/transceivers from Taiko, although, one of the cables is an Amphenol from AfterDark.

Everything sounds amazing.

I have omitted some specific details re: product models, etc. as I am waiting for a friend to come out from surgery, and don't have access to the products to double-check for you folks.

Net-net:
--Good power, even to cheap products is an improvement.
--Better products are better, incrementally.
--Wi-fi transmissions may not be the killer you think, but managing the band-width and radio power is good.
--Copper can be as good or better than optical.
--You need to test this stuff in your place to determine the outcome that matters to you.

 

[NigelB]

After swapping the positions of my switch and optical isolator, I settled on this config as sounding best (to my ears):

Reiki SuperSwitch Master Pro → LHY Optical Isolation → Antipodes K50

I found this to provide vocal clarity and tone that aligned best with my personal preferences. The differences were small, however.

Now that's a first! Always good to know, thanks for confirming.

 

[NigelB]

...I have used a few different SFPs in my network over the past five to seven years, or so, as well as a few different optical cables and FMCs.

For 2-3 years I used a Sonore OM, which was excellent, and much better than the cheaper Amazon off-shore options. I also used the high-end Sonore power supply with the OM.

The network experiments began with daisy-chaining the "expired" Cisco 2960 switches. I settled on two, but had as many as four during my investigations. I also used the UpTone eRG switch, which I still have/use in a different capacity. I had ethernet filters, USB cleaners, various cables, expensive or not.

I swapped over to fiber because I had a 35' run from the network closet, through the walls/ceiling, to the basement audio room.

The original copper network ethernet cable passed through spaces that had Romex wiring, can-light "pig-tails" and other lines not run by me. I thought it might be a bit of an electro-magnetic soup.

I am also about a quarter-mile away from a company that develops sat comm products for military and civilian use. Oh, the antennas I have seen! And Humvees testing products, etc.

I assumed I was in a great location for RF/EM traffic.

However, not having the tools or knowledge, I couldn't practically test this impact in my home. So, I thought it might be a best-practice to make the 35' run with optical, to reduce the exposure of my network to all this junk. That was the genesis of my investigations with optical solutions.

Eventually, fiber infrastructure came to my neighborhood, so I added this to the home, in addition to the copper ISP. Audio is on the fiber feed; everything else is on the copper. Sounds great.

I had previously tried to create a virtual version of this concept using the EdgeX device, but that was a huge PITA. I never did get it working, even with help. IMO this device is not user-friendly, unless you have more-than-casual networking skills.

Eventually, the Taiko network products arrived, and those guys seemed to have a different take on network noise and solutions for audio.

I added the Taiko Switch to my network, and switched back to copper. BTW, this is easy for me, because I ran CAT6, CAT8, SM and MM fiber at the same time. Materials are cheap. Labor/time is expensive. I ran it myself, so easy to manage (for me).

For about a year, or so, I had a modded Buffalo switch upstream from the Taiko Switch, contrary to general guidance to use only the Taiko product.

Eventually, I removed the Buffalo, and just used the Taiko Switch. After about a year, I added the Taiko Router, and went with the DAC cabling suggested by Taiko. I retain that set-up currently, using OEM adapters/transceivers from Taiko, although, one of the cables is an Amphenol from AfterDark.

Everything sounds amazing.

I have omitted some specific details re: product models, etc. as I am waiting for a friend to come out from surgery, and don't have access to the products to double-check for you folks.

Net-net:
--Good power, even to cheap products is an improvement.
--Better products are better, incrementally.
--Wi-fi transmissions may not be the killer you think, but managing the band-width and radio power is good.
--Copper can be as good or better than optical.
--You need to test this stuff in your place to determine the outcome that matters to you.

Excellent, informative post telling a personal listening story and your own happy ending. Nice one.

 

[MarkusBarkus]

...thanks Nigel. I am enjoying the thread even though I am not actively engaged in experimenting at the moment. Never know where knowledge rears its furry head!

I think there are multiple solutions to the network enhancement model, and the best one perhaps becomes the best one given budget, physical access, etc. for a given case.

I am with you and others here that the final scrub just before entering your server/streamer, is probably the sensible approach. I am doing that with the Taiko stuff too, although aesthetically, I would prefer all network gear upstairs and out of sight/site! Cheers...