Meridian Unplugged - Powered by MrTechGuy.co.uk Forums Discussion HiFi & HT Chat Ethernet switches

It seems this is a white paper in the marketing sense rather than the informative sense. It is lacking any empirical data or interpretation thereof.

I do not doubt that different devices, no matter how apparently benign their role, can and do introduce audible variations into complex audio systems. And I have no experience of comparing Ethernet devices so am not in a position to comment as to whether I have heard any differences.

But that paper seems to ignore the rather fundamental fact that any digital audio device knows exactly what the clock rate is (e.g. for CD it is 44.1kHz) and thus (i) the DAC can reclock and regenerate the signal with relatively little jitter and (ii) if it cannot understand whether the voltage has crossed the Rubicon it can ask for it to be resent. If it does not do those things well it is because it is not a great DAC rather than the signal is jittery or smeared as a result of an Ethernet switch which, as I understand it*, is asynchronous. So the EtherREGEN switch may do something good but I can not see how it relates to the argument presented. Unless he is saying that the very asynchronicity of the Ethernet signal introduces enough jitter-inducing noise as to introduce (audible) distortions through the DAC mechanism? In which case it is an electronically badly designed DAC which is susceptible to any asynchronous signal no matter how it has been pampered as it is routed around the network.

* I usually avoid commenting on technical things as I have a vestigial understanding of the inner working of digital audio - if any of my comments above are wrong I am very happy to be corrected and educated.

Flame suit donned

Jeremy,
Thanks for the posting. I think it's a useful addition to the conversation.

That said, the benefits posited in the paper are, for me, unconvincing.

Regarding ethernet signal jitter, we have to remember that ethernet is true digital, not quasi-digital like SPDIF:
- Ethernet brings us correct or corrected data effectively 100% of the time. Voltage thresholds are highly relevant to DAC's since they span the digital/analog gap, but are irrelevant to true digital (such as ethernet). Any digital domain device that is sending out voltage levels so spurious that they flip a bit is effectively like a 1970's phone modem (i.e. not seen anymore).
- A network connection is on a different clock than a DAC.
- The paper seems to be saying that anything less than 100% buffers are 100% susceptible to inbound jitter, but if we look at Meridian's triple buffering (which I assume to be low pass, high pass than band pass, all on different PLL's) it strains belief that substantial jitter is coming through. If that were the case, I would expect the initial PLL to fail lock from the beginning.

Regarding ground plane noise, this is a more interesting argument, but I have trouble with this one as well. I assume that Meridian (like most companies at this price range) use hierarchical layers of modified-Sulzer style regulators (doubled for balanced circuits) that are very effective at filtering out this kind of noise. I studied and used them decades ago, and I don't think they've changed much since because they so robustly work.

In the test/review published by 3dit0r, the tester measured higher, rather than lower noise on the data output of the "audiophile" switch. No mention of ground noise measurements was made, but if noise on the data stream is higher (remember we're still talking pure digital in terms of signal here), it harms the argument of lowered ethernet signal jitter according to the author's own argument.

Putting all of this aside, I have to admit that my technical level was never super deep, and has diminished significantly.

None of what I've said negates the possibility that there are objective audible differences here. I just haven't read anything that convinces me to cross the bridge of skepticism.

Enough folks have reported benefit that this feels like an itch that we have to keep scratching until we learn something definitive.

Having read some of his in-depth posts on this topic, I believe John knows what he is talking about, however this paper does not go into the same level of depth (I suspect to cater to a broader audience).

In the paper he tries to answer your question by saying:

'Noise on the ground-plane of the DAC causes effective jitter in the DAC chip, which modulates the audio signal coming out of the DAC. Note that the actual clock signal does not change at all. It is how the clock circuit inside the DAC chip sees the clock signal that matters...

We have identified two primary external mechanisms—present in various digital interfaces—which can cause ground-plane noise on a DAC’s internal ground-plane: 1) Jitter carried through digital data and, 2) leakage current.'

In other words, noise introduced by external digital clocks can cause ground plane fluctuations within the DAC, which, in turn, causes jitter on the how the internal clock signal is read. Which is supported by what Meridian said a few years ago when launching their Anniversary Edition and the reason for redesigning the ID40 input card.

Hi Akimo

While I can't explain why ethernet signals can make a difference, I can hear a large difference between, for example, the SoTM switch and the switch built into the Antipodes. I can also hear the difference between two SoTM's connected via fibre optic and the same two connected via CAT 6 (the former, incidentally sounded very smooth, but not as engaging).

I can't pretend to be an expert, but IMHO John makes (and has separately made additional) valid points worth exploring. Noise on the ground plane would sound like a good starting point. I honestly don't think John is fabricating a solution to fit market demand, I have every reason to believe he spends a lot of time thinking about solving noise in digital systems.

I think I've mentioned before, though, that this is the icing on the cake. There are likely other upgrades that will make a greater difference.

J.

I'll keep an open mind.

Neither my tech background nor my hearing are good enough to rule anything truly out.

Thanks for keeping an interesting discussion moving along productively.

I cannot speak for the EtherREGEN since I have not tried one yet, but I can speak for UpTone. I own their Mac mini conversion kit powered by their JS-2 LPS. Alex Crespi supported me during and after the installation and his products made a tremendous positive impact on my system. I doubt that they would offer anything that they didn’t believe in.
David

I re-read the paper, especially the explanation of the PHY Ethernet receiver processing. Combined with your summary above I now understand the theoretical framework of his argument.

I assume that only one EtherREGEN is needed, connected between the network and the audio end-point, to achieve this protection?

Thanks.

Hi Gianni, I believe their view is only one EtherRegen is needed.

I have changed from a Linksys SD205 -many years good service- to the Uptone EtherRegen. I am convinced, it is a keeper.

From a Devolo 1200 Ethernet out to the “A” side of the Uptone. My Roon Rocked MC200 by Mr techguy is on A side as well. From B side, crossing the mote as Uptone describes it, it connects to ID41 in a 818v3. I did have some stability issues in the beginning but after a few times powering on off it is stable.

The stereo image is rock solid. I hear more detail and it sounds more natural. I will not bore you with audiophile extravagance.

I am not trying to convince you. Just try it out yourself in your own home and decide if it is worth the money. I ordered direct from Uptone. Alex Crespi is a very kind and approachable person in case you have a question.

Cheers!

My interest in this topic is its relevance to AV streaming, rather than pure audio.

Is there an opinion as to whether such a switch will benefit AV?

Has anyone tried the Chord/English Electric switch?

My foray into HDMI cables suggests that there are good gains to be had in AV, but I suspect there would be more immediate gains to be had from improving the power supplies in your HDMI sources (such as your Sky box) and the cables before improvements in the ethernet network were obvious.

Even before this step, though, I'd predict good all-round gains to be had from decent power conditioning driving the TV and HDMI sources. Borrowing an AudioQuest 1200 would be interesting...

Hi Jeremy,

Interesting what you say re HDMI interconnects. I currently use Chord “active” HDMIs. Not cheap and cheerful, not expensive, but solid and well made. Interestingly, Chord, having got out of pricier HDMIs, have introduced hybrid optical/copper wired HDMIs. Look forward to them being reviewed.

Re interconnects, I have invested in Chord Signatures. Think they are excellent (as no doubt equivalent AudioQuest ones are).

Re power supplies, I have chosen a different route, installing multiple MK K1826 WHI mains power sockets that claim to filter RFI (both ways) plus provide surge protection. In addition, I use Chord’s basic shielded power cables in anticipation of reducing locally broadcast RFI. Thinks it helps!

The Uptone Audio paper you refer to, although not specifically saying so, perhaps gives an insight to what causes “digital harshness”. Lots of food for thought.

I have definitely reduced, if not substantially eliminated, harshness in my audio only setup, but have yet to succeed in my AV system, which also suffers from muffled sound. But not sure if this mainly poor miking by the broadcasters.

Look forward to feedback on Ethernet switches (just to think about!).