"Audiophile Grade" Ethernet Switches - The new generation

[Pim]

@Confused the new Uptone switch will have a 75ohm BNC 10MHz input so it will be possible to sync it with other equipment.

How does one sync packed data? Isn't the whole idea of packet data that it doesn't need to be synced?

Hey dont shot the messenger I only forwarded what was written at Computer Audiophole.

LOL! I thought I'd ask the question first. The shooting can wait. 

Seriously though, I read the thread but didn't read anything to answer my question.

[thumb5]

As you say, Pim, it can't be about synchronisation. I guess the idea of the clock input is that the switch can use a "clean" external clock rather than some internal clock as a reference for the Ethernet physical layer devices. That might in principle reduce clock phase noise passed on to connected equipment (hand-waving at this point).

[Confused]

As @octaviars stated, the Uptone switch has a 10MHz input, so this is for use with a reference clock, not a word clock or master clock. So I guess you could say that it "syncs" with a 10MHz reference clock, but it does not sync clocks in a system in the way a word clock generator might do.

The theory behind all this is exactly as @thumb5 has stated, a reduction in clock phase noise. This is where things get problematic. Nobody has really come up with a technically robust reason why this should influence a packet data which is then FIFO buffered. There are similar arguments relating to clocks used in USB streamers and endpoints, where the clock in the receiving DAC is all that should matter in an asynchronous system, as long as other system clocks are within broad tolerances. So it is very easy to dismiss all this as audiophile foo, and very difficult to come up with a robust technical argument as to why it is not audiophile foo. On the other hand, there is currently a lot of evidence that that these seemingly non critical clocks do influence sound quality. It is fascinating stuff, and the market is already filling up with products that make use of better clocks upstream of the DAC, it is the main reason a ultraRendu is better than a microRrendu or the sMS-200Ultra is better than a sMS-200. People are selling this kit and people are buying this kit, but nobody seams to really know why the clocks should influence sound quality as they do. Fascinating stuff. I look forward to the day all this stuff is properly understood, I wonder if that will ever happen?

[thumb5]

@Confused From what I've picked up, from (e.g.) John Swenson's posts, the point of reducing clock phase noise is mainly to do with its effects on the PHY devices, which might (it is supposed) inject analog noise into the DAC (e.g. via power/ground planes). As you say, it should not have any effect on the timing of samples reaching the DAC because they are clocked out of a ton of buffer memory by a clock that is completely asychronous to the Ethernet.

[Confused]

@Confused From what I've picked up, from (e.g.) John Swenson's posts, the point of reducing clock phase noise is mainly to do with  its effects on the PHY devices, which might (it is supposed) inject analog noise into the DAC (e.g. via power/ground planes).  As you say, it should not have any effect on the timing of samples reaching the DAC because they are clocked out of a ton of buffer memory by a clock that is completely asychronous to the Ethernet.

To add to this, I understand that the concept is that the switch has a "clean" side and a "dirty" side, so I presume that the low phase noise clock is the one doing the work on the clean side. 

So, this all makes perfect sense.  That said, I do not believe anyone has ever measured analogue noise generated by phase noise on PTY devices, let alone established exactly what this does to the DAC.  Plus, the DAC is supposedly isolated from noise, most audio grade Ethernet inputs have galvanic isolation and other means to block noise, so even if the noise exists there are ways to mitigate the damage it might do.  So I feel like these things are being designed on an informed hunch and unproven theory, and maybe a some subjective listening tests and similar.  It would be so much nicer the design was based on proper measurements, testing and demonstrable benefits.

If the Uptone switch does come in at $500 though, I will probably buy one to try.  A case of blind faith driven by a genuine curiosity.  Or maybe curiosity killed the cat, then moved onto liberating my wallet of another few quid....

[octaviars]

@Confused From what I've picked up, from (e.g.) John Swenson's posts, the point of reducing clock phase noise is mainly to do with  its effects on the PHY devices, which might (it is supposed) inject analog noise into the DAC (e.g. via power/ground planes).  As you say, it should not have any effect on the timing of samples reaching the DAC because they are clocked out of a ton of buffer memory by a clock that is completely asychronous to the Ethernet.

I suppose we will have to wait and see what John Swenson can measure after he built the switch. Perhaps it will affect the DAC (I dont know why and how) but if it affects it is it something that can be heard?

It is safe to say that the market is shifting from USB gizmos that is supposed to isolate, re-generate and so on to more gizmos that is supposed to do all sorts of things with ethernet. The companys that comes up with all this stuff goes where there is money to make

[Hifi_swlon]

Has John Swenson ever released any measurements for anything? Haven’t followed forums that closely lately, and certainly not CA, but there always seemed to be talk of ‘working on measurements’ but to my knowledge none were ever shared.

[octaviars]

Has John Swenson ever released any measurements for anything? Haven’t followed forums that closely lately, and certainly not CA, but there always seemed to be talk of ‘working on measurements’ but to my knowledge none were ever shared.

He have shared measurments on CA regarding the "high impedance leakage noise" but as far as I know nothing that have shown how this affect a DAC ........

[Confused]

Has John Swenson ever released any measurements for anything? Haven’t followed forums that closely lately, and certainly not CA, but there always seemed to be talk of ‘working on measurements’ but to my knowledge none were ever shared.

8.25", but then Amir on Audio Science Review started posting that it could not be anywhere near 8.25" as it hardly registered on his oscilloscope, it started a bit of a ruck as I recall.   (for those who don't know, this is a quite hilarious microRendu with iFi power supply joke, or maybe not)


To provide a slightly more serious response to your question, I would say not much.  There has been the odd post where he mentions something he has measured, but he has not published a full set of measurements for anything, or at least not that I have seen.

He has built this though:

https://www.computeraudiophile.com/forum...ent-855985

The claim is that this will allow him to build the prototype boards that he needs for the custom test kit to measure the effects of upstream audio clocks and similar.  It is anyone's guess as to if this will ever bear fruit.

Keeping the thread on topic, the video in the first post per the above link shows a prototype board for the new Uptone switch.

[chrisc]

I acquired a shielded Cat 7 LAN cable, one metre long, with shielded RJ45 plugs.

Three friends came around for a few hours. After some careful listening (and a few glasses of wine, since we live in Cape Town, South Africa), we came to the conclusion that on complex orchestral music, the sound was cleaner. On "commercial" music, there was no difference

The computer (a MacBook) was attached directly via the LAN cable. The previous LAN cable was one I made up using solid Cat5, about 800mm long