Right. I respect and learn from all my fellow dCS community, but I am with Anup on this one. I’ve listened to a lot of systems. Demo’ed several very high end USB servers well above the Innuos in cost, fiddled with the Pro USB on my MSB Select II DAC. My experience is the exact opposite of Ian’s. I have yet to hear a USB rig best a properly set up Ethernet system.
If you hear that, then yes, I would agree something needs to be improved. By all indications, the Ethernet input on the Bartok should be superior to the USB. So, why did it sound worse? I believe you that it did sound worse to you, but I would want to understand why—because it shouldn’t. That’s the question that should be asked. Perhaps the dealer connected the Bartok to the network with a shielded Ethernet cable? Or has a noisy switch, or is running things right off the combo modem/router/wifi box? Who knows.
We know from comparisons of similar equipment from the same manufacturer—e.g., the PS Audio DirectStream and its Network Bridge and Direct Memory Player, all three of which I have owned and have high-value SQ—that the same company’s implementation of Ethernet and say, I2S between the transport and the DAC, can produce audibly different results. But that’s not because a protocol is inferior or spinning discs are superior; it’s because of how well it is executed in the DAC itself.
I agree, but Melco specifically assert in their S100 data sheet that “data integrity” is one of the reasons to invest in the S100, and they state more than once that traditional IT doesn’t value what makes music special as much as it values speed and throughput. And the Lowbeats review Pete referenced also casts a lot of aspersions on “data integrity” in the standard Ethernet environment. And it’s almost all incorrect and definitely misleading.
I hope you know I am not asserting “if you can’t measure it, you can’t hear it.” Never. But we should be able to explain it, even if only conceptually. I can get tempted by just about anything that offers a real prospect of improving my system’s SQ, or that provides a realistic engineering explanation for how it helps prevent a problem from arising in the first place. Optical is a good example, and there are a number of devices that help deploy it with minimal risk of RFI/EMI injection in the conversion process. It may not be perfect, but it can be an improvement or a preventive device.
In that same vein, I want to be able to understand how/why something improves SQ in my system. And with all due respect to Pete, that Lowbeats “review” sure doesn’t provide any basis for understanding the claimed improvements. The reviewer asserts this in his conclusion:
Another important argument: The advantages of the Melco are always felt to the same extent, regardless of which network player you use. So unlike, for example, high-quality cables, the effect and tonal benefits of which depend heavily on the connected components, the S100 retains its value, even if something changes in the playback chain. It is the foundation for a truly audiophile music streaming chain.
Hmmm. Makes everything better no matter what you connect or change [better even than a direct cable without the switch, according to this reviewer]? Are we to take such a statement seriously? Something is so “universally wrong” with Ethernet for music purposes that this box makes it better in every system the same way and to the same extent?
Give me a reason—doesn’t even have to be measurable, merely logical—for believing that this box can make some change somewhere in the various OSI layers that might have audible effects, and I’ll plunk down $2k on a sale-or-return basis without any hesitation. But tech-babble rigmarole about how music digital is not really 1s and 0s like other computer information, or how unsquared or unfocused waves can affect the timing in the DAC, is not helpful. Anyone who pays attention here knows the story about how 1s and 0s are transmitted through Ethernet physically. And a switch can only send what it receives. It receives and transmits packets. It doesn’t make them bigger-better-stronger. TCP error correction occurs at the endpoints, not in between, and at a layer well-removed from the physical network. If enough data is lost or damaged so that the endpoints cannot fix it, neither will a switch. And if all the packets do get through, then that is all the information the DAC requires or can use. So long as the switch does not inject some new [even euphonic] noise into the line, it’s job is done.
I am also not suggesting that we know everything there is to know about Ethernet [though its definitely not some babe in the woods compared to USB, whether for music or anything else]. Anup’s point about the 50+ year evolution of Ethernet also suggests to me we will keep learning. [I’m a big Ravenna fan, and had rather hoped it might take off by now. Oh well. :shrug: ]
If you hear a switch or any box making a difference, you have to ask yourself at least two questions:
- What is the change I am hearing?
- What is being removed/improved/added, e.g., noise, better clocking, etc.?
- How is that change effected?
Another way to look at that is: what was my system not doing as well as it should have been before I inserted this new box? These are valid questions, and the S100 or other device may be a valid solution if there was a problem it can address. But for anyone already with a proper network, it’s a solution in search of a problem.
P.S. To answer Ian’s very first question in the OP, until Melco makes a Roon server, I will stick with the Roon Nucleus+. I’ve looked at the Innuos line more than once; the ZENith has the integrated CD ripper that the Nucleus does not, as well as more music software options. But I require neither. I’ve also looked at the Silent Angel Rhein Z1 Roon Music Server, but have not been persuaded [mostly because of the marketing blather, but partly because of their network switch which was not credible to me].