For of us who believes (and understand) why switches matter to sound, what would be a good "normal" switch to use as base for DIY? I am thinking about switching crystal to something good (Crystek 575 perhaps) and bypass the switched voltage regulator with my own linear voltage regulator board and run it from batteries.
I have listened to UpTone etherRegen and it definitely sound better than my regular switch, and in my views its nothing magical going on. Its simple less electronic noise going into the streamer, and by generating a good signal (good clocks) it will be easier for the streamer to reconstruct the data which leads to less electronic noise generated internally.
Don't confuse the issue with bit-perfect, error correction etc, those issues has very little to do with why a switch sounds worse or better.
I have listened to UpTone etherRegen and it definitely sound better than my regular switch, and in my views its nothing magical going on. Its simple less electronic noise going into the streamer, and by generating a good signal (good clocks) it will be easier for the streamer to reconstruct the data which leads to less electronic noise generated internally.
Don't confuse the issue with bit-perfect, error correction etc, those issues has very little to do with why a switch sounds worse or better.
... but are unable to explain even a fraction of.
What is claimed without proof can be dismissed without proof.
I need more popcorn
Jan
From Archimago. He says it very well
Archimago's Musings: MUSINGS/MEASUREMENTS: Netgear Nighthawk S8000 (and audiophile ethernet switches)
Here is a review of UpTone etherRegen (which is not biased and verified with good measurements.
UpTone Audio EtherREGEN Switch Review | Audio Science Review (ASR) Forum
Lat but not least
Archimago's Musings: MUSINGS/MEASUREMENTS: Netgear Nighthawk S8000 (and audiophile ethernet switches)
Here is a review of UpTone etherRegen (which is not biased and verified with good measurements.
UpTone Audio EtherREGEN Switch Review | Audio Science Review (ASR) Forum
Conclusions
Computer, networking and streaming architecture instructs us that an external switch cannot have any effects on an Audio DAC. But it is always good to put some hard data behind this. And that is what the measurements show. That no matter how deep we dig down in the waveforms coming out of the DAC, no difference exists between a cheap, generic switch and UpTone EtherRegen. Even when going down to incredible -160 dB which is equivalent to a 27 bit audio word (3 bit more than any 24 bit content), there is still no difference.
Measurements conclusively demonstrate that EtherRegen did not change jitter, noise or distortion of the DAC. It further had no impact on its clock speed, or output voltage.
What is that? You hear otherwise? Well, I tested it and there is no audible difference either. I suggest you repeat my test. If you do hear a difference, have someone switch cables behind your back. You can do this quickly, or wait days or weeks to switch. Just don't look and keep a log of 10 trials. If you can tell better than 8 out of 10 times that there is an audible difference with EtherRegen, then you have conquered the impossible!
Now, I am not beyond paying more for something if it looks better, feels better, etc. Unless you have an upset stomach and like to use the EtherRegen as a hot water bottle replacement (don't laugh, I have done that with my laptop!), there is no such benefit. Even if there were, I would not pay $640 for it. $100 maybe. But $640?
Lat but not least
(...a disturbingly large group)
(an equally disturbing, but tiny group)
Two answers:
1. Whatever switch it takes to transmit data without corruption or dropped packets
2. Whatever it takes to convince yourself you're hearing magic.
Pick either, or both. The one with the thick milled and gold anodized front panel would be a good start for 2.
But the problem is: you don't know what the noise is, it's amplitude, spectrum, mode of transmission, and most importantly, its effect on the transmitted data. Ignoring, of course, that Ethernet is fundamentally designed for noise immunity, or it just plain wouldn't work.
No confusion at all. But if it's bit-perfect, there can be no sonic effect by definition, because no changes have occurred. Hence the term, "Bit Perfect".
There is no error correction other than resending dropped or corrupted packets. The data stream over Ethernet is packetized, and must be reassembled into a "stream" at each end, again, by definition and design. There simply is no method of organized data corruption, or organized data modification, than can improve the sound of the resulting reassembled bit stream beyond "bit perfect", which must occur in any data transmission system, or it's single and primary function is a complete failure.
All noise in the transmitted data manifests as potential data corruption above some threshold. Either a packet makes it uncorrupted, or it's resent until it either does or the data connection fails. Unrecoverable packet corruption results in catastrophic failure. Recovered packets are reassembled into a bit-perfect stream. There's really no middle ground.
Sorry, that's how it works. No component can become a passive DSP.
I find it hard to believe that people still believe in the "bits-are-bits" argument. 10 years ago yes, but today?!
Why do you think USB tweaks like ISO Regen works? Why do you think toslink can sound better than coax (toslink will always have more jitter)? Why do you think LPS sound better than switched mode power?
You don't need a very high end system to hear difference for the things I mentioned, and yet none of them "change the bits" in any way. In fact, the answer to those questions I mentioned is the same: they reduce the electronic noise going into the DAC and hence into the analogue stage.
For some reason, maybe because of galvanic isolation, people seem to think ethernet is somehow immune to noise.
But if you can't hear differences in your system thats fine I guess, but in my system ethernet switch does matter, as well as tweaks like LAN isolator, using battery for switch and more. In other words, just like USB and other digital input. It would actually be strange if ethernet switch didn't matter.
If you want to test the effect of electronic noise into the DAC, try this on a transparent system:
1. Connect the DAC to a gaming computer with fast CPU and graphics card
2. Play and listen when computer is idle
3. Play and listen while playing a resource intense computer game.
Electronic noise affects a lot in a DAC, but most audible is the effect it has on the clock, and the direct effect it has on the analogue stage. Decrease in clock accuracy will produce clock-jitter, and effect on the analogue state will reduce details and clarity.
1. Connect the DAC to a gaming computer with fast CPU and graphics card
2. Play and listen when computer is idle
3. Play and listen while playing a resource intense computer game.
Electronic noise affects a lot in a DAC, but most audible is the effect it has on the clock, and the direct effect it has on the analogue stage. Decrease in clock accuracy will produce clock-jitter, and effect on the analogue state will reduce details and clarity.
Noise incursion has nothing to do with bit perfection.
As with other things I don't have is a single game on any device except for solitaire or mines left there by the installer.
I didn't come here to debate with "bits-are-bits" people, I wanted a suggestion about a good ethernet switch to use as base for DIY. Then I kinda got jumped and had to defend myself
Anyway, what I want to do it take a regular switch, replace switched power regulators with linear ones and replace its crystal with a more accurate one. Maybe replace some capacitors as well. But I want the rest of the switch to be of good quality, and of course the layout has to be such that its possible to do.
Maybe Faboz who started this thread knows by now? Maybe the D-Link DES-108 is a good starting point:
I have been playing with switches in my home as I bought some arm-based SBCs. In general, if you have switches, they will stablize network latency and speed during large file transfer as wifi routers/carrier gateways typically suck at it (not built for that task). If network transfer carries issues, you will see speed slow down and latency (probably) getting worse, as data gets re-transferred.
But you barely need any speical purpose ones and don't mess with QoC. Just keep your lan cables away from mains cables.
On DIY, most of these switches have large ground plane, cold joints easy to form. I probably won't bother touch it...
But you barely need any speical purpose ones and don't mess with QoC. Just keep your lan cables away from mains cables.
On DIY, most of these switches have large ground plane, cold joints easy to form. I probably won't bother touch it...
Last edited:
What you came here for is irrelevant, this board offers mostly trolling, so that's what you get
A simple experiment helped me made up my mind with respect to switches. That came after some frustrating mods on the Dlink, none of which made it sound better, on the contrary.
So, here is the idea behind the experiment. Most streamers have an option to play files or even albums from memory, either decoded or not. This makes it possible to unplug both the switch and the ethernet cable. Unplugging the cable should be done at the streamer side as well, quite audible.
So, comparing a cheap switch with a commercial CAT6 cable vs no switch and no cable showed that in my system there is a clearly audible but modest difference in favour of the perfect switch and perfect cable. Would i make a serious effort in order to approximate the "no switch, no cable sound"? Absolutely not. Even minor changes to amplifier topology and parts are more audible and musically important.
Perhaps my dac and system lack the resolution. Perhaps multiple levels of reclocking make such phenomena less audible. Dunno, but now i can sleep well. Highly recommended to try.
What you came here for is irrelevant, this board offers mostly trolling, so that's what you get
From both sides.
Don't change the issue because you run out of facts (well, you didn't have any to start with, so ...).
Nobody said Ethernet is immune to noise, but it was explained to you (I assume you did understand) that the audio packets are re-send until they assemble bit-perfect or the link fails. Nobody said jitter doesn't matter. Nobody said all DAC outputs are perfect (hint: they aren't). What everybody is bending over backwards is to show you, with facts and figures, that the Ethernet switch cannot make a difference. But you seem to prefer to hang on to fairy tales rather than do what most intelligent humans are good at, namely reason towards the facts.
Jan
Last edited:
If you change a bit in a word, you change the word. If you don't change a single bit, you make no changes to any word. That's why.
Off topic. We're talking Ethernet switches, and audio over IP. Very different discussion. The fact that you don't recognize the difference is the source of your misconception.
It is largely because it must be, and has been designed that way. It wouldn't work if it wasn't immune to most common noise sources.
So, in your scientific testing, have you done any measurements? Controlled subjective testing? Or just listened for what you (think) you hear?
USB and Ethernet are completely different animals. Please do some research.
This is clearly a specific case related to a specific system. You suggest a result that may or may not be present on any other system.
Sure it does. If there's enough noise, bits cannot be properly recovered, and become corrupted.
Why would anyone do this? Get a Cisco W-C2960CG-8TC. The PSU is already correctly done, regulated, filtered, etc., etc., EIA power cord. Yes, you'll want to configure it with the management GUI, but all other efforts are futile.I didn't come here to debate with "bits-are-bits" people, I wanted a suggestion about a good ethernet switch to use as base for DIY. Then I kinda got jumped and had to defend myself
Anyway, what I want to do it take a regular switch, replace switched power regulators with linear ones and replace its crystal with a more accurate one. Maybe replace some capacitors as well. But I want the rest of the switch to be of good quality, and of course the layout has to be such that its possible to do.
Maybe Faboz who started this thread knows by now? Maybe the D-Link DES-108 is a good starting point:
If data gets through any switch with no corruption, the job is done. Whatever noise there may have been has already been mitigated, or the data would be corrupted. I don't know why this is so difficult to understand.
- Status
- Not open for further replies.