Taiko Audio SGM Extreme : the Crème de la Crème
- Thread starter CKKeung
- Start date
Hi RGMD11,
I was surprised by how the Sotm switches performed. It is quite obviously a good product, I do strongly suspect the "silver wire" upgrade to be the cause for it's less then pleasant midrange performance. This all by itself is quite surprising as it is only a ~2 1/2" piece of silver wire used as a DC lead to power the optional clock board. @romaz gave me a green light on replacing it with copper:
The Ether Regen is most certainly good value for money.
Hi Matt,
The copper B-side is 100Mbit, I have the Telegartner M12 MAGIC switch here which also runs at a 100Mbit speed. Both of these, though "pleasant" sounding, come with reduced levels of detail, ambient retrieval, dynamics and envelopment. Anecdotally I have spend some time testing 10Gbit networking which do take these type of traits up another notch, but unfortunately have not managed to find a way around a certain associated mechanical quality. Years ago, before our servers even existed, I used to reduce my networking speeds to 100Mbit (fiber) to get a less "digital" sound, most certainly a valid way to get better sound from say a mac mini, or an Intel Nuc. To answer your question directly, both fiber and copper connected to the A-side running at 1Gb.
Hi Emile,
It is a great post for your in depth reporting on the network switches. I would also like to share my current setup of Uptone EtherRegen (eR).
Signal path:
DIY Roon Server (Jcat Femto Net Card) -> 13m Supra Cat8 -> eR Port A -> eR Port B ->1.5m Pachanko "Aphelion" -> dCS Vivaldi Upsampler
eR:
External Clock: Mutec Ref10 SE-120
Clock Cable: Wireworld Silver Starlight 7 - 1.5m
Power Supply: Uptone JS-2 (12v)
DC Cable: Pachanko Server LPSU cable - 1m
Ground Box: Tripoint Troy Signature BLK
Ground Cable: Tripoint Troy Signature Silver - 2m
Sound qualities in terms of clarity, density, resolution, micro dynamics and soundstage are improved quite a lot by switching to an audio-graded DC cable and by connecting eR to an external clock and a ground box.
I am planning to replace the above-mentioned 13m Supra Cat8 with a fiber optic cable to see any further improvement over my current setup.
Thank you again for your valuable write-up.
Attachments
Last edited:
Hi Emile, thanks for your advice.
Yes. I am using Double Crown power cables for Spectral Preamp and dCS Vivaldi DAC.
Hi Emaile. Thanks for a great report.
In the other 4 cases, you connect the switch to Extreme using fiber, but with the M12, you had a somewhat convoluted setup (a) with copper - fiber - copper to M12 and (b) use copper to Extreme instead of fiber.
Can you throw some light on why pls?
Thanks so much for taking the deep-dive into this stuff, Emile!!
I'm looking forward to hearing where we can find this alternative SFP...
Well, it certainly indicates once again that your Extreme high-power CPU server works differently. In my Melco with low power CPU, the pass from A to B does not produce this effect. We are really just beginning to understand all the complexities of audio networks, which is surprising because as everyone knows: bit are bit
What an excellent report, Emile. Thank you. Much appreciated. I had to finally register here, so I can post.
Interesting that you start your discussion talking about glare. I have noticed that certain type of network injected noise could be perceived as glare. This is typically a digital curtain between you and the music. In some cases it is something that people may like, even prefer. It’s also interesting what happens when this digital curtain / glare disappears. All of a sudden you realize what the instruments really sound like, but one may be too attached to the glare-y sound and may even prefer it. It is indeed a way to tweak your system.
One of my takeaways from Emile’s report is how dependent the network tweaking is on the local environment. It’s very important to keep in mind the context of his tests. And people should be careful how they read his report. He did not say that the etherREGEN is better than the SOTM. He said that the etherREGEN “was initially disappointing” but in a very specific scenario it sounded really good (and better than the SOTM combination he has tried). He also said that he has “done a brief check with a low power server prototype with no SFP port present running an audiophile Linux variant, and the results are very different”. You have to keep those things in mind. I hope the person who commented earlier that he changed his purchasing decision based on that report has an Extreme server or is planning to buy one. There are other reports that can be found comparing switches with different results in a different context.
I have been playing a lot with my home network recently. As some others have reported, replacing the all-in-one cable modem/router/firewall/etc. box the cable companies in the US provide with a dedicated cable modem and a dedicated router powered by good linear power supplies helps a lot. It makes a bigger difference than my etherREGEN switch.
I know Uptone was engineering the etherREGEN switch to be insusceptible to upstream devices and were thinking that the power supply would probably not matter. That’s what all DAC’s manufacturers would tell you too - it’s an asynchronous signal with galvanic isolation and reclocking, so the upstream device does not matter. Well, probably everyone reading this thread knows better. It is a similar story with the etherREGEN - although they created an excellent product that I like a lot, I found my system even more susceptible to upstream devices than before. The added resolution makes my system more sensitive on upstream network devices so they need good attention too. And powering the etherREGEN with a better than the stock power supply is an obvious improvement.
Back to my home network experiments. Modding the cable modem and routers (i.e. replacing the crappy caps in the power section) also helps. Better DC cables also help. Vibration control helps too.
I was doing an A/B comparison the other day by changing the DC power cable on my cable modem. I had two cables of the same length, materials, geometry and connectors, both broken in. One was shielded and the other one was not. I streamed a Qobuz track, switched the DC cable to my cable modem, and listened to the same track again. You could easily hear the effects of a shield on the DC cable to my cable modem! This was mind blowing. And to be honest, if someone else told me that I would not believe it! I would be like the guy with the white hat from another forum talking about measurements and expectation bias. BUT... I was able to clearly hear the difference from a shield on a DC cable connecting my cable modem.
I played with different routers and WiFi access points. And yes, they sound differently. Using a router without build-in WiFi between the cable modem and the etherREGEN removed a lot of that digital glare. From that experience I feel like the noise generated by the WiFi routers have big contribution to the digital glare. I am using Ubiquiti EdgeRouter X (and the SFP version) routers currently. I like those devices a lot. They have 12V DC input (unlike some of the bigger Ubiquiti devices), so they are easy to power up with LPS. They have a lot of functionality in a really small box with a really simple and nice PCB layout. And they are cheap and easy to configure (if you know what you are doing). For example, it took seconds to create a dedicated VLAN for my server on one of the ports, so my local network broadcast messages would never reach my server. It enables DHCP and routing automatically for you.
My network looks like this at the moment: Cable modem —> Ubiquiti EdgeRouter X SFP —> etherREGEN —> JCAT NIC on my server, all powered by good LPS, cables, etc. Let's ignore the ethernet cables to keep this post more simplified, but I have been playing with a lot of those too. That network set up sounds really good, but I need to add WiFi back. Luckily that router has a SFP port, so I am now playing with various ways of connecting a WiFi access point that is isolated by fiber. I have a second Ubiquiti router, a bunch of media convertors, so there are plenty of options, and I need to find which one works best for my network. A lot of tests ahead of me, but you get the idea of what I am trying to do.
Having said all that, I am curious what the entire upstream network looks like at Emile’s lab. I am also curious if Emile has done any testings with the JCAT Net Femto card in the Extreme. If the Telegartner GOLD M12 switch via copper to the Extreme sounds better than anything else, a logical next step would be to try connecting it to a JCAT NIC. But as we all know Emile is always many steps ahead of us. I was laughing at his comment about the etherREGEN grounding - of course he has tried that and probably every other combination with every switch we can think about. It's so good to have people like him in the industry.
Interesting that you start your discussion talking about glare. I have noticed that certain type of network injected noise could be perceived as glare. This is typically a digital curtain between you and the music. In some cases it is something that people may like, even prefer. It’s also interesting what happens when this digital curtain / glare disappears. All of a sudden you realize what the instruments really sound like, but one may be too attached to the glare-y sound and may even prefer it. It is indeed a way to tweak your system.
One of my takeaways from Emile’s report is how dependent the network tweaking is on the local environment. It’s very important to keep in mind the context of his tests. And people should be careful how they read his report. He did not say that the etherREGEN is better than the SOTM. He said that the etherREGEN “was initially disappointing” but in a very specific scenario it sounded really good (and better than the SOTM combination he has tried). He also said that he has “done a brief check with a low power server prototype with no SFP port present running an audiophile Linux variant, and the results are very different”. You have to keep those things in mind. I hope the person who commented earlier that he changed his purchasing decision based on that report has an Extreme server or is planning to buy one. There are other reports that can be found comparing switches with different results in a different context.
I have been playing a lot with my home network recently. As some others have reported, replacing the all-in-one cable modem/router/firewall/etc. box the cable companies in the US provide with a dedicated cable modem and a dedicated router powered by good linear power supplies helps a lot. It makes a bigger difference than my etherREGEN switch.
I know Uptone was engineering the etherREGEN switch to be insusceptible to upstream devices and were thinking that the power supply would probably not matter. That’s what all DAC’s manufacturers would tell you too - it’s an asynchronous signal with galvanic isolation and reclocking, so the upstream device does not matter. Well, probably everyone reading this thread knows better. It is a similar story with the etherREGEN - although they created an excellent product that I like a lot, I found my system even more susceptible to upstream devices than before. The added resolution makes my system more sensitive on upstream network devices so they need good attention too. And powering the etherREGEN with a better than the stock power supply is an obvious improvement.
Back to my home network experiments. Modding the cable modem and routers (i.e. replacing the crappy caps in the power section) also helps. Better DC cables also help. Vibration control helps too.
I was doing an A/B comparison the other day by changing the DC power cable on my cable modem. I had two cables of the same length, materials, geometry and connectors, both broken in. One was shielded and the other one was not. I streamed a Qobuz track, switched the DC cable to my cable modem, and listened to the same track again. You could easily hear the effects of a shield on the DC cable to my cable modem! This was mind blowing. And to be honest, if someone else told me that I would not believe it! I would be like the guy with the white hat from another forum talking about measurements and expectation bias. BUT... I was able to clearly hear the difference from a shield on a DC cable connecting my cable modem.
I played with different routers and WiFi access points. And yes, they sound differently. Using a router without build-in WiFi between the cable modem and the etherREGEN removed a lot of that digital glare. From that experience I feel like the noise generated by the WiFi routers have big contribution to the digital glare. I am using Ubiquiti EdgeRouter X (and the SFP version) routers currently. I like those devices a lot. They have 12V DC input (unlike some of the bigger Ubiquiti devices), so they are easy to power up with LPS. They have a lot of functionality in a really small box with a really simple and nice PCB layout. And they are cheap and easy to configure (if you know what you are doing). For example, it took seconds to create a dedicated VLAN for my server on one of the ports, so my local network broadcast messages would never reach my server. It enables DHCP and routing automatically for you.
My network looks like this at the moment: Cable modem —> Ubiquiti EdgeRouter X SFP —> etherREGEN —> JCAT NIC on my server, all powered by good LPS, cables, etc. Let's ignore the ethernet cables to keep this post more simplified, but I have been playing with a lot of those too. That network set up sounds really good, but I need to add WiFi back. Luckily that router has a SFP port, so I am now playing with various ways of connecting a WiFi access point that is isolated by fiber. I have a second Ubiquiti router, a bunch of media convertors, so there are plenty of options, and I need to find which one works best for my network. A lot of tests ahead of me, but you get the idea of what I am trying to do.
Having said all that, I am curious what the entire upstream network looks like at Emile’s lab. I am also curious if Emile has done any testings with the JCAT Net Femto card in the Extreme. If the Telegartner GOLD M12 switch via copper to the Extreme sounds better than anything else, a logical next step would be to try connecting it to a JCAT NIC. But as we all know Emile is always many steps ahead of us. I was laughing at his comment about the etherREGEN grounding - of course he has tried that and probably every other combination with every switch we can think about
. It's so good to have people like him in the industry.
Last edited:
Nenon, fantastic post. Tbh, I'm torn in response btwn patting you on the back for yr efforts and offering you some tranqs Lol.
Is this really the parallel realm in digital of analog finagling?
Ie in the 80s when cdp afficionados were simply plugging a cdp in, popping on a cd and saying "perfect sound forever", analog mavens were continually adjusting arm/cart setup, phono loading, tt isolation, lp cleaning etc.
And three to four decades later, life has done a 180, the digital guys are fussing modems, routers, clocks, switches, usb v i2s etc, and attendant LPSs, keeping noise away any which way possible.
History loves to repeat itself. Now perfect sound forever in digital means as much or more attention to detail than analog Lol.
Is this really the parallel realm in digital of analog finagling?
Ie in the 80s when cdp afficionados were simply plugging a cdp in, popping on a cd and saying "perfect sound forever", analog mavens were continually adjusting arm/cart setup, phono loading, tt isolation, lp cleaning etc.
And three to four decades later, life has done a 180, the digital guys are fussing modems, routers, clocks, switches, usb v i2s etc, and attendant LPSs, keeping noise away any which way possible.
History loves to repeat itself. Now perfect sound forever in digital means as much or more attention to detail than analog Lol.
These Audiophile Switch comparisons are an excellent work by Emile.
But we have to take into account that the impact of these switches is inferior to that of other components in systems with the Extreme.
Matt
One easy and worthwhile tweak people can make to their network is a lan shorting plug. You can either buy the Acoustic Revive model or make one following the Kanaimaru link on the url below. Best opened in Chrome for auto translate.
They are quite easy to make with the Telegartner rj45 plug and I used Takman metal film resistors. Plugging it into my modem gave an appreciable drop in noise / hash. Fitting two however dulled the presentation, making the sound over damped.
https://acousticrevive.jp/portfolio-item/rlt-1/
They are quite easy to make with the Telegartner rj45 plug and I used Takman metal film resistors. Plugging it into my modem gave an appreciable drop in noise / hash. Fitting two however dulled the presentation, making the sound over damped.
https://acousticrevive.jp/portfolio-item/rlt-1/
steve williams
Site Founder, Site Owner, Administrator
Absolutely. Emile has said countlessly that a good USB cable, a good power cord and footers trumps everything else
What an excellent report, Emile. Thank you. Much appreciated. I had to finally register here, so I can post.
Interesting that you start your discussion talking about glare. I have noticed that certain type of network injected noise could be perceived as glare. This is typically a digital curtain between you and the music. In some cases it is something that people may like, even prefer. It’s also interesting what happens when this digital curtain / glare disappears. All of a sudden you realize what the instruments really sound like, but one may be too attached to the glare-y sound and may even prefer it. It is indeed a way to tweak your system.
One of my takeaways from Emile’s report is how dependent the network tweaking is on the local environment. It’s very important to keep in mind the context of his tests. And people should be careful how they read his report. He did not say that the etherREGEN is better than the SOTM. He said that the etherREGEN “was initially disappointing” but in a very specific scenario it sounded really good (and better than the SOTM combination he has tried). He also said that he has “done a brief check with a low power server prototype with no SFP port present running an audiophile Linux variant, and the results are very different”. You have to keep those things in mind. I hope the person who commented earlier that he changed his purchasing decision based on that report has an Extreme server or is planning to buy one. There are other reports that can be found comparing switches with different results in a different context.
I have been playing a lot with my home network recently. As some others have reported, replacing the all-in-one cable modem/router/firewall/etc. box the cable companies in the US provide with a dedicated cable modem and a dedicated router powered by good linear power supplies helps a lot. It makes a bigger difference than my etherREGEN switch.
I know Uptone was engineering the etherREGEN switch to be insusceptible to upstream devices and were thinking that the power supply would probably not matter. That’s what all DAC’s manufacturers would tell you too - it’s an asynchronous signal with galvanic isolation and reclocking, so the upstream device does not matter. Well, probably everyone reading this thread knows better. It is a similar story with the etherREGEN - although they created an excellent product that I like a lot, I found my system even more susceptible to upstream devices than before. The added resolution makes my system more sensitive on upstream network devices so they need good attention too. And powering the etherREGEN with a better than the stock power supply is an obvious improvement.
Back to my home network experiments. Modding the cable modem and routers (i.e. replacing the crappy caps in the power section) also helps. Better DC cables also help. Vibration control helps too.
I was doing an A/B comparison the other day by changing the DC power cable on my cable modem. I had two cables of the same length, materials, geometry and connectors, both broken in. One was shielded and the other one was not. I streamed a Qobuz track, switched the DC cable to my cable modem, and listened to the same track again. You could easily hear the effects of a shield on the DC cable to my cable modem! This was mind blowing. And to be honest, if someone else told me that I would not believe it! I would be like the guy with the white hat from another forum talking about measurements and expectation bias. BUT... I was able to clearly hear the difference from a shield on a DC cable connecting my cable modem.
I played with different routers and WiFi access points. And yes, they sound differently. Using a router without build-in WiFi between the cable modem and the etherREGEN removed a lot of that digital glare. From that experience I feel like the noise generated by the WiFi routers have big contribution to the digital glare. I am using Ubiquiti EdgeRouter X (and the SFP version) routers currently. I like those devices a lot. They have 12V DC input (unlike some of the bigger Ubiquiti devices), so they are easy to power up with LPS. They have a lot of functionality in a really small box with a really simple and nice PCB layout. And they are cheap and easy to configure (if you know what you are doing). For example, it took seconds to create a dedicated VLAN for my server on one of the ports, so my local network broadcast messages would never reach my server. It enables DHCP and routing automatically for you.
My network looks like this at the moment: Cable modem —> Ubiquiti EdgeRouter X SFP —> etherREGEN —> JCAT NIC on my server, all powered by good LPS, cables, etc. Let's ignore the ethernet cables to keep this post more simplified, but I have been playing with a lot of those too. That network set up sounds really good, but I need to add WiFi back. Luckily that router has a SFP port, so I am now playing with various ways of connecting a WiFi access point that is isolated by fiber. I have a second Ubiquiti router, a bunch of media convertors, so there are plenty of options, and I need to find which one works best for my network. A lot of tests ahead of me, but you get the idea of what I am trying to do.
Having said all that, I am curious what the entire upstream network looks like at Emile’s lab. I am also curious if Emile has done any testings with the JCAT Net Femto card in the Extreme. If the Telegartner GOLD M12 switch via copper to the Extreme sounds better than anything else, a logical next step would be to try connecting it to a JCAT NIC. But as we all know Emile is always many steps ahead of us. I was laughing at his comment about the etherREGEN grounding - of course he has tried that and probably every other combination with every switch we can think about
Another great write-up, Nenon !
So that turns out to be a bit of an issue, I can order them and they will be branded then, but the delivery time is around 2 months, I'll take a look around for readily available alternatives. They are being sold somewhere by someone, more likely multiple brands, but which ones is not going to be disclosed. It's a 1310nm 20km modules which is interesting as if I remember correctly those are being supplied by MSB with their USB Pro ISL module. Photos in case someone has something similar looking in their stash, btw note the receiving/transmitting ports are actually marked with arrows for that earlier discussion here.
