Cruncher Posted May 5, 2020 Posted May 5, 2020 @Assisi "A standards compliant switch may be fine for your purposes." - why would you say this ? you know nothing about my set up or my 'purposes'. Not that it should matter.....but ... (1) Yes, I have done a lots of listening tests and tested a lots of gear. (2) I use is what has been listed in this site as a 'Audiophile Ethernet switch'.
gwurb Posted May 6, 2020 Posted May 6, 2020 (edited) 14 hours ago, Stereophilus said: I can understand a perceived frustration with a generic piece of equipment subtly altered and rebadged with inflated pricing and astronomical claims. I personally have not purchased or experienced this. But in my experience, what you have generalised here is not true. Take the EtherRegen as an example... it’s an “audiophile” switch, but there is nothing that has any similarity to its design in generic switches. It is a unique design. You would need to listen to it to decide if it is worth your coin. Another example: the Aurender W20. There own PCB, fanless design, unique battery power supply, ultra low noise USB and dual spdif outputs designed specifically for audio. And software similarly designed uniquely for the hardware. You cannot buy the equivalent from Dell, or Apple, or HP. Once again, the only way to know if it is worth your coin is to try it. That is Oranges to Apples. In terms of board layout and design this: https://uptoneaudio.com/products/etherregen is completely different ballpark to this: https://cruxaudio.com/collections/thunder-data/products/bonn-n8-high-end-audio-network-switch This is a dedicated audio product: https://aurender.com/w20se/ This is a modified PC: https://antipodes.audio/product/antipodes-cx/ I haven't heard EtherRegen but from the board it looks like it is doing a few things differently to a standard switch. Will it improve anything; I would doubt it but I haven't heard it so I'll hold my final opinion on EtherRegen until I hear it. Edited May 6, 2020 by gwurb
Assisi Posted May 6, 2020 Posted May 6, 2020 13 hours ago, Cruncher said: @Assisi "A standards compliant switch may be fine for your purposes." - why would you say this ? you know nothing about my set up or my 'purposes'. Not that it should matter.....but ... (1) Yes, I have done a lots of listening tests and tested a lots of gear. (2) I use is what has been listed in this site as a 'Audiophile Ethernet switch'. I agree. I do not know actually know anything about your setup. This thread is about regular switches. Reading the tenor of your post I took a pserspective that you were not supportive of audio grade switches and other devices. Others do have a similar position on the limited or no value of various audio grade devices without having tried anything. My position is that I have a switch that has for me beneficial outcome. My switch is IEC power cable in. No switch mode or Linear power device connections. So I did not have an issue in that respect. Out of curiosity what switch do you use? John
Stereophilus Posted May 6, 2020 Posted May 6, 2020 1 hour ago, gwurb said: That is Oranges to Apples. In terms of board layout and design this: https://uptoneaudio.com/products/etherregen is completely different ballpark to this: https://cruxaudio.com/collections/thunder-data/products/bonn-n8-high-end-audio-network-switch This is a dedicated audio product: https://aurender.com/w20se/ This is a modified PC: https://antipodes.audio/product/antipodes-cx/ I haven't heard EtherRegen but from the board it looks like it is doing a few things differently to a standard switch. Will it improve anything; I would doubt it but I haven't heard it so I'll hold my final opinion on EtherRegen until I hear it. No, it’s not Oranges to Apples, I’m specifically rebutting you, when you said “my point is: current audiophile servers and switches are almost identical to regular pc and switching/routing devices”. The ER and the Aurender W20 are examples of a current audiophile server and switch respectively. And, whilst I agree that the Antipodes CX is a heavily modified PC, it’s modifications to the motherboard, USB outputs, power supply, audio dedicated operating system, and inbuilt modified ethernet switch still make it a unique audio component, with a distinct sound quality which is quite different to a modified Mac mini or HP laptop. Yes, I own all three and have compared them. 1
davewantsmoore Posted May 6, 2020 Posted May 6, 2020 On 05/05/2020 at 4:33 AM, Cruncher said: Music runs across IP, and for me the Elephant in the room is the end device quality. Latency and Jitter in a IP stack is significantly higher than at the Ethernet layer and this is not even including the O/S. Whther or not this "computer jitter" matters for audio, depends on the implementation of the DAC. On 05/05/2020 at 4:33 AM, Cruncher said: End device latency and jitter is where there is a lot of research and money is being focused from the High Performance computing and High Frequency trading communities and has resulted in technologies such as High Performance MPI, RDMA, custom silicon, usNIC and other user space and stack by pass methods. But this jitter is enormous to the point of being wholely irrelevant for audio. 1ns of jitter is bad (big). It is the bare minimum standard. 160ns may as well be three weeks. On 05/05/2020 at 4:33 AM, Cruncher said: The things I struggle with : (1) I see a lot of talk with absolute conviction about hearing ~ +/- 200 nSec of Ethernet Jitter (especially from people who sell audiophile Ethernet, but fail to supply measurements or compliance and test data) but I see no talk about hearing end device jitter ( there are always two end devices ) and they contribute ~ +/- 100,000 nSec of Jitter each. What people must understand is that what is audible is what the DAC does. Nothing else. Jitter from upstream devices could modulate the signal (leak through, so to speak) reaching the DAC ..... but this is not something which can be generalised. It depends on the DAC. So, we can't ask "what does 200ns of network jitter sound like?" in a vacum. It depends on how much, if any, of it, leaks into the (analog) audio. On 05/05/2020 at 4:33 AM, Cruncher said: emotional and expensive Indeed. I don't think either of these are very useful On 05/05/2020 at 4:33 AM, Cruncher said: (3) Upstream Effects on a DAC (tight coupling). I have read upstream devices such as Sources and Ethernet switches effects the quality of music from a DAC due to delay and jitter. If "upstream" jitter "leaks" into the DAC (it should not). On 05/05/2020 at 4:33 AM, Cruncher said: I also have read DAC manufactures using very accurate clocks (33 femto-seconds jitter). Tightly timed clock (across the whole jitter frequency spectrum, including low Hz) are needed for good sounding audio. On 05/05/2020 at 4:33 AM, Cruncher said: If upstream jitter could effect the DAC then it would have to be completely coupled to the input You'll need to be more specific with you "have to be". .... but in short, it should not affect the DAC..... if it does, redesign the DAC/implementation. Lowering the jitter in the network beyond what "not broken" networks have, is IMVHO, silly. On 05/05/2020 at 4:33 AM, Cruncher said: This buffer would provide some level of decoupling. Yes. On 05/05/2020 at 4:33 AM, Cruncher said: Do I think that someone swapped out a Ethernet switch and heard a difference ? Yes. But I don't think it was because the new switch had 2 uSec less latency ( the switch is likely to be 0.001% of the system jitter). I think the new switch was within standards, or less out of standards. To me if you replace something on a Ethernet device and it improves it was not built properly in the first place. e.g by changing the switch they got rid of the nasty cheap wall wart. Yes, it may also be mains or cable radiated noise entering the DAC (from the switch).... although again, DAC manufacturers should ideally address this issue. On 05/05/2020 at 4:33 AM, Cruncher said: A regular properly made, standards compliant switch will work fine. Yes , unless the system is succeptible to jitter leaking through the ethernet phy.... or to noise from the swithes power supply. In which case, get system components which aren't. 1
Cruncher Posted May 6, 2020 Posted May 6, 2020 3 hours ago, davewantsmoore said: Yes , unless the system is succeptible to jitter leaking through the ethernet phy.... What do you mean by the Jitter leaking through the Ethernet phy ?
davewantsmoore Posted May 7, 2020 Posted May 7, 2020 17 hours ago, Cruncher said: What do you mean by the Jitter leaking through the Ethernet phy ? In the DAC (for example, although there are lots of version of this problem) you will have incoming ethernet chip ("phy"sical layer) with a clock, etc..... then there will be "something" .... then there will be digital audio (typically I2S) with a new clock. If (for example) you carefully watch the I2S audio signals..... while modulating the arriving ethernet signals, eg. their timing, or their noise components, etc..... and things that you do to the ethernet, are visible in the I2S ..... then that is what I mean by "leaking through". As you have alluded earlier...... this should not happen. But it is possible that it does. None of needs to be expesive or fancy it just need to be designed/implemented well.
frednork Posted May 7, 2020 Posted May 7, 2020 2 hours ago, davewantsmoore said: In the DAC (for example, although there are lots of version of this problem) you will have incoming ethernet chip ("phy"sical layer) with a clock, etc..... then there will be "something" .... then there will be digital audio (typically I2S) with a new clock. If (for example) you carefully watch the I2S audio signals..... while modulating the arriving ethernet signals, eg. their timing, or their noise components, etc..... and things that you do to the ethernet, are visible in the I2S ..... then that is what I mean by "leaking through". As you have alluded earlier...... this should not happen. But it is possible that it does. None of needs to be expesive or fancy it just need to be designed/implemented well. So care to share some dacs that have this sorted already so we dont have to try and fix it before it happens?
Cruncher Posted May 8, 2020 Posted May 8, 2020 (edited) 22 hours ago, davewantsmoore said: ("phy"sical layer) with a clock, etc..... then there will be "something" .... then there will be digital audio (typically I2S) with a new clock. @davewantsmoore Have you personally observed / measured the Phy clock interfering with the I2S clock or you quoting someone else ? Edited May 8, 2020 by Cruncher
dbastin Posted May 10, 2020 Author Posted May 10, 2020 (edited) On 30/04/2020 at 10:27 PM, dbastin said: The Ubiquiti EdgeSwitch, with my audiophile ethernet and power cables that cost 20 times the EdgeSwitch, actually sounds very satisfying to me, so it can certainly suffice. This is interesting ... I'm starting to conclude there's a number of ways to solve the problem of Noise over Ethetnet (NoE - to coin a new acronym ). Fibre, grounding ... Firstly, @Stereophilus has noted elsewhere a simple $50 Netgear Switch + JCAT grounding conditioner equals the EtherRegen (ER) without the JCAT grounding conditioner, and that conditioner did not improve ER. The JCAT grounding conditioner is a cable from device to AC plug with a soecial quartz mixture in the sheild to help reduce noise. https://jcat.eu/product/usb-lan-ground-conditioner/ I recently delved inside my Antipodes Edge and placed a single Synergistic Research ECT on the chip just a few mm behind the ethernet port - all the traces on the PCB from the ethernet port lead to this chip, so I dubbed it the data doorway. That made a considerable improvement. https://www.synergisticresearch.com/isolation/accessories/ect During that session I swapped out the Nordost Sort Kones with Synergistic Research MIG 2.0 under my Devialet Pro. That was OMG level of extra resolution. But at the end of that session I felt the midrange, while ultra revealing, was 'hard' like metal. Next session, following the theme of treating data doorways, I openned up the EdgeSwitch and placed a single ECT on the isolation transformer that servers the 3 ports I am using, which are closest to the crystal clock and main chip (see photo). And a further OMG level improvement. Its So good. Now, the midrange is now back in balance and while not as revealing (presumably the hardness was exaggerating), its seems very right to me. There's still a touch of smearing in Siblances. While tweaking around more I ended up disconnecting the EdgeSwitch from the rest of the network. The music kept playing Roon radio, I just couldn't control Roon. Now this is total isolation, perhaps better than fibre. But I must say, with this total isolation, it may be only slightly better SQ - it will take some study to understand the differences, if I can be bothered.. And the switch is not grounded yet (other than via power)! At AU$90 ish each new, I used 2 at AU$180, and will use another on the data doorway of Devialet to address the final leg, bringing it to AU$270 ... very effective value for money ... an absolute no brainer at the used price I paid. Bear in mind, the EdgeSwitch is supported by: 1. power via Shunyata Triton V2, Shunyata ztron Alpha Digital power cord and Core Audio Technologies KAIA LPS and quality DC cable 2. connected to the Edge and Devialet with 2 x US$1000 rrp ethernet cables, and a Wireworld Starlight Cat 8 to wall outlet. This 'support' has proven very effective at also suppressing noise, and RRP is about 25 x the switch RRP. And my system total RRP just tips 6 figures. This is not to brag, but for context. Even in a lesser system, $200 for switch + $200-300 for ECTs is still ridiculous value, considering what it does for SQ of a system that transmitts data over ethernet. I would compare the benefit to a good power conditioner, at a fraction of the cost. Of course, this approach could be applied with a lower cost power, ethernet cables, switch and some 'quantum' type dots or chips if they address high frequency noise. Even an Ubiquiti 5 port EdgeRouter at $100 could most likely be set up as a switch - my EdgeRouter X enables any port to be configuired as a switch port. Imagine what an ECT would do on a low cost audiophile switch, like Bonn 8? I hope this is of interest. Edited May 10, 2020 by dbastin
Guest rmpfyf Posted May 10, 2020 Posted May 10, 2020 2 hours ago, dbastin said: This is interesting ... I'm starting to conclude there's a number of ways to solve the problem of Noise over Ethetnet (NoE - to coin a new acronym ). Fibre, grounding ... Firstly, @Stereophilus has noted elsewhere a simple $50 Netgear Switch + JCAT grounding conditioner equals the EtherRegen (ER) without the JCAT grounding conditioner, and that conditioner did not improve ER. The JCAT grounding conditioner is a cable from device to AC plug with a soecial quartz mixture in the sheild to help reduce noise. https://jcat.eu/product/usb-lan-ground-conditioner/ I recently delved inside my Antipodes Edge and placed a single Synergistic Research ECT on the chip just a few mm behind the ethernet port - all the traces on the PCB from the ethernet port lead to this chip, so I dubbed it the data doorway. That made a considerable improvement. https://www.synergisticresearch.com/isolation/accessories/ect During that session I swapped out the Nordost Sort Kones with Synergistic Research MIG 2.0 under my Devialet Pro. That was OMG level of extra resolution. But at the end of that session I felt the midrange, while ultra revealing, was 'hard' like metal. Next session, following the theme of treating data doorways, I openned up the EdgeSwitch and placed a single ECT on the isolation transformer that servers the 3 ports I am using, which are closest to the crystal clock and main chip (see photo). And a further OMG level improvement. Its So good. Now, the midrange is now back in balance and while not as revealing (presumably the hardness was exaggerating), its seems very right to me. There's still a touch of smearing in Siblances. While tweaking around more I ended up disconnecting the EdgeSwitch from the rest of the network. The music kept playing Roon radio, I just couldn't control Roon. Now this is total isolation, perhaps better than fibre. But I must say, with this total isolation, it may be only slightly better SQ - it will take some study to understand the differences, if I can be bothered.. And the switch is not grounded yet (other than via power)! At AU$90 ish each new, I used 2 at AU$180, and will use another on the data doorway of Devialet to address the final leg, bringing it to AU$270 ... very effective value for money ... an absolute no brainer at the used price I paid. Bear in mind, the EdgeSwitch is supported by: 1. power via Shunyata Triton V2, Shunyata ztron Alpha Digital power cord and Core Audio Technologies KAIA LPS and quality DC cable 2. connected to the Edge and Devialet with 2 x US$1000 rrp ethernet cables, and a Wireworld Starlight Cat 8 to wall outlet. This 'support' has proven very effective at also suppressing noise, and RRP is about 25 x the switch RRP. And my system total RRP just tips 6 figures. This is not to brag, but for context. Even in a lesser system, $200 for switch + $200-300 for ECTs is still ridiculous value, considering what it does for SQ of a system that transmitts data over ethernet. I would compare the benefit to a good power conditioner, at a fraction of the cost. Of course, this approach could be applied with a lower cost power, ethernet cables, switch and some 'quantum' type dots or chips if they address high frequency noise. Even an Ubiquiti 5 port EdgeRouter at $100 could most likely be set up as a switch - my EdgeRouter X enables any port to be configuired as a switch port. Imagine what an ECT would do on a low cost audiophile switch, like Bonn 8? I hope this is of interest. No surprises here - though it sounds like you're a user crying out for fibre. If your device can't take it, fibre as far as you can go off a decent switch to a decent SFP>Eth switch and a very, very short last leg into your device.
dbastin Posted May 10, 2020 Author Posted May 10, 2020 (edited) 3 hours ago, rmpfyf said: No surprises here As you quoted my entire post, what exactly are you not surprised about? I intend to try fibre as follows: - Edge Router X SFP > fibre > EdgeSwitch - ER > fibre > EdgeSwitch - EdgeSwitch > fibre > ER And share my observations on Ethernet Cables for Audio Part C. Edited May 10, 2020 by dbastin
Guest rmpfyf Posted May 10, 2020 Posted May 10, 2020 15 minutes ago, dbastin said: As you quoted my entire post, what exactly are you not surprised about? I intend to try fibre as follows: - Edge Router X SFP > fibre > EdgeSwitch - ER > fibre > EdgeSwitch - EdgeSwitch > fibre > ER And share my observations on Ethernet Cables for Audio Part C. You're diminishing noise sources. You've significantly left the jitter characteristics in the clock and switching chains. @recur has posted extensively on what would be better than an Edgeswitch in these capacities. You'd then go with some decent to get back to Ethernet - EtherRegen or similar - on as short a cable as possible.
dbastin Posted May 11, 2020 Author Posted May 11, 2020 20 hours ago, rmpfyf said: @recur has posted extensively on what would be better than an Edgeswitch in these capacities. He also recommended Ubiquiti as a low cost good performance starting point (new) - as well as outlining a few enterprise level switches for buyers of used gear.. Even JCAT, which sells the Uber M12 switch, highly recommend Ubiquiti EdgeSwitch.
Guest rmpfyf Posted May 11, 2020 Posted May 11, 2020 (edited) 10 minutes ago, dbastin said: He also recommended Ubiquiti as a low cost good performance starting point (new) - as well as outlining a few enterprise level switches for buyers of used gear.. Even JCAT, which sells the Uber M12 switch, highly recommend Ubiquiti EdgeSwitch. Not surprising about JCAT, it's trendy It's a good switch. There are other makes similarly good. But if you want something that's likely to clock out with better regularity, the Enterprise level stuff is a better bet. Edit: You could theoretically make a cheapie work within some severe usage constraints, but a nice enterprise switch on fibre is going to be hard to beat. Let alone you could mod it if you like... Edited May 11, 2020 by rmpfyf
dbastin Posted May 15, 2020 Author Posted May 15, 2020 This has been restored from the Great Debate sub forum and its close monitoring by moderators. It was put there due to an incident some time ago, but we've been well behaved since then. Please be civil - so this doesn't go back to the Great Debate, or shut down. And so we dont discourage helpful contributions. 1 1
dbastin Posted August 29, 2020 Author Posted August 29, 2020 (edited) Some info of interest here - multiple Cisco switches in series. Page 8 ... Edited August 29, 2020 by dbastin
Hydrology Posted September 2, 2020 Posted September 2, 2020 English Electric just arrived. Very solid unit. Some pics..
PCOWandre Posted September 3, 2020 Posted September 3, 2020 What's on the inside to attempt to justify four hundred and fifty pounds? 3
anhton82 Posted November 29, 2020 Posted November 29, 2020 Many thanks for useful infor from this topic. I have bought a Sotm sms 200 and I will follow step by step to upgrade my system for better sound quality. Thank you !
BugPowderDust Posted February 13, 2021 Posted February 13, 2021 I stumbled across this youtube video this morning. While it's designed for Ravenna / AES67, the building blocks of functionality that it suggests are precisely what any suitably qualified networking engineer would tell you (and I did earlier in this thread). Network segmentation, traffic marking and prioritisation and then policing if you need to send traffic over a network boundary. In short, a cheap Cisco switch like the SG350 (a 10 port managed gigabit desktop switch for $380 at Scorptech) has all the functionality you need. The video here is for a distributed switch setup, which I'd wager most of you won't have in your home, so you can lose the need for multicast and PTP in your home streaming environment. It's got some decent explanations on the various concepts too so if you're new to networking, it's not a bad primer. If anyone wants to understand any of this more clearly, please reach out. I'd be happy to explain and give guidance as to how far you may need to go in a home environment to optimise it. 3
Dropbear67 Posted April 1, 2021 Posted April 1, 2021 unless you are streaming in UDP OR your TCP stream can't keep up with your bitrate AND your receiving software has zero buffering then the type of switch you have , or your ethernet path length (how many switches it passes through) makes no difference. 4
dbastin Posted April 28, 2021 Author Posted April 28, 2021 Interesting comments from Paul McGowan, seems to support 'low noise' ethernet cables, but switches ???
BugPowderDust Posted April 28, 2021 Posted April 28, 2021 @dbastin if Paul's past behaviour is anything to go by, once he starts selling PS Audio ethernet switches, there will now be some magical reason why you need one.
dbastin Posted April 29, 2021 Author Posted April 29, 2021 2 hours ago, BugPowderDust said: some magical reason why you need one Given he is in the network bowels of PS Audio, I suspect his network people have convinced him of the position he has taken and also discouraged him from experimenting on the basis it would be a waste of time. If this is the case, his company and products might benefit from employing some more open minded engineers. He mentions he couldn't tell the difference between wired and wifi ethernet on his best system. If that is the extent of his experience on this subject, I am surprised he has not been more inquisitive and not surprised of the conclusion he has reached. I'll add that in view of the above it is unlikely his DS DAC Bridge has given any special attention to mitigating ethernet noise. I am now on a mission to find those endpoints that do. I also note he essentially dissed the SOtM switch which has been a benchmark for some time. Well, he did say he was being controversial .. perhaps intentionally.
Recommended Posts