First post here… just as well be on such a delicate topic.
A little background as I’m new here. Electric engineer, did some analog electronics design. 35 years of designing wired and wireless equipment. I guarantee that you have used technology I invented and some hardware and software I contributed building. Unless one stays away from mobile phones and wifi :-) Audiophile for 40 years. Enough of this, only proves I’m old.
There is a fundamental difference between IP networking and the other digital technologies that we are used to in audio…. DACs are very delicate machines in the digital section. There’s all sorts of opportunities for noise, jitter, skews and multiple issues potentially affecting sound quality. Folks get that complexity here.
USB and SPDIF are simple “link layer protocols”. This means that those protocols allow for 2 devices to exchange data over a single physical link. The same applies to Bluetooth, AirPlay, in the wireless domain. USB and SPDIF are simple, basic protocols. They do not support FEC, or forward error correction. This opens the door to all sorts of clock, skew, jitter and noise problems. The unfortunate truth is that because of the protocols limitations it’s more dicey to move a bunch of bits from from the streamer to the DAC over those interfaces than it is to move the same bits from tidal/servers to the streamer, more in a second. Note that I’m not talking about noise propagating inside the components from those interfaces. This issue is orthogonal to messing up the bits in transit, it’s a design issue and well made digital interfaces can avoid noise propagation in a component with usual precautions, and costs associated. That would apply to any digital interfaces. Tl;dr - USB and SPDIF are fragile designs optimized for cheapness, fancy cables can make a difference there and discussions will continue to rage on.
Now IP networking is something completely different. There’s 2 pieces to it. A) a link layer protocol, Ethernet, that moves data between 2 devices B) a network protocol, IP, that carry data between multiple switches/routers, possibly over the internet. Ethernet is the protocol that home switches deal with. Ethernet is a fairly simple protocol, but it has FEC and more features not relevant to this discussion. Ethernet layer will correct corrupted packets if possible or drop it if is not. Drops you may ask? Does not matter. The data that travels over it is layered like this:
music data
|
Http, https, some other applications
|
tcp
|
IP
|
Ethernet
IP protocol has its own error correction too, end to end between let’s say a local or internet music sever (Qobuz, tidal) and a streamer. And TCP sitting on top of that will handle retransmissions as needed. In this whole chain, there are zero opportunities for messing anything up. Either the data gets through with 100% integrity or it doesn’t. Please note that there is no timing/noise/jitter-what-have-you issues that apply here. This becomes a problem again once the music data needs to be transformed in USB or SPDIF or fed to a DAC. Complete non issue in the IP network.
An ‘audiophile’ switch cannot do anything useful on the data integrity front that I can understand that a regular switch cannot do. The only thing I can see it can do is possibly keeping the noise low on those Ethernet links. Maybe. They have to use the same Ethernet transceiver chips as everyone else, it’s limited to the quality of board and components around and isolation.
BUT. The target audience for this is not the fellow that streams from a Pi. The only place in the system I can see an ‘audio switch’ maybe making a difference is on the box where the streamer runs and feeds the DAC or convert to USB or SPDIF. I have 3 ways to do that on my system, a Naim ND5 XS2, a Mac mini and the network card on the Bricasti . Unexpectedly, they sound different, I’d even say very different. More unexpectedly, the Naim sounds quite a bit better than the network interface on the M3 (all with Roon). (to me anyways, better clarity, definition). I really think a good quality streamer should do a great job at isolating the noise from the digital interfaces, including Ethernet, and it’s hard for me to see how a fancy switch would make a difference.
So, Maybe, I’m open minded. But until someone credible reports that it makes a major difference in a nice system, I’m not ditching my ubiquity POE switch (Nope, that’s not who I work for). I’m in good Company, Paul McGowan sells cables that cost more than my MA PL200ii speakers and a major ‘usb cables make a difference’ guy, but he refuses to go down the ‘Ethernet cables make a difference’ path. I’m at the same place, I run decent but not outrageous interconnect and MIT speaker cables, Gaia feet under my speakers and a couple of components which many see as snake oil potential, I’m just not there with ‘audio switches’…
