Not to hijack a thread, but... if you are skeptical but curious about Ethernet switches, you can experiment for $70. Might be fun and enlightening.
My switch story:
I heard a difference this summer in a switch test.
First, I do not understand how low packet loss could be an issue given protocols and buffering techniques used by major streaming services (see TIDAL behavior screen shot attached from a capture -- of course if peoples systems have intervening streaming to other devices who knows). TIDAL appears fragmented, like HLS which someone said TIDAL uses, which allows lots of time for drops and retransmitts of those fragments. Experiment with connecting/disconnecting your streamer for short intervals (assuming its software doesn't get upset) and see it keeps playing. Please share details of how low packet loss/jitter could cause issues (I like to learn to improve my setup, and its part of the hobby) -- privately is OK too. Of course high packet loss could cause disruptions.
Second, I do believe there could be other interference issues causing audible problems and wanted to try eliminating those inexpensively. Starting with an inexpensive shielded cable given all the crazy going on in my cabinet. As we all know, a shielded cable without ground is also called an antenna -- and I suspect that lack of cable grounding is why people often get different performance out of many high end Ethernet cables. To introduce a shielded cable easily, I needed a different switch than my consumer grade dlink.
So I decided to purchase an old used Cisco WS-C3560-8PC-S off ebay ($60), and a shielded cable off Amazon ($10). Why the 3560? 1) It is widely available and inexpensive used. 2) It properly grounds shielded cables. 3) It is engineered and supplied with a power cable such that no additional ground is needed (most consumer switches are not -- if there isn't a ground cable going in on the power cord, and you haven't attached to a supplied ground on the switch, then it is not grounded). 4) In hopes it was well engineered (enterprise class, additional POE attention paid, internal power supply, fanless, and I have had these in some harsh environments over the years). 5) It has a management interface and port mirroring that I situationally believe... mostly.
The two scenarios:
A) dlink router --> dlink switch --> streamer
dlinks on house power, streamer on PS Audio P15 regenerator
B) dlink router --> dlink switch --> 3560 --> streamer
dlinks on house power. 3560 and streamer on different P15 banks.
B sounded better than A, so I kept it since then.
Why? I don't know. Perhaps I'm just suggestible (I hear differences with power cables). Would a different cable work better? I might try one at some point. But for electrical interference, not packet loss. Would fiber isolation help? It could reduce noise (if it exists) into the 3560, and I might try it at some point (the switch has an SFP port). Would clock chips on the switch matter? I don't understand how in the context of streaming protocols (until explained). Could poor clocks add interference? Perhaps. Will I try an EtherREGEN? Not just now (chasing other improvements), perhaps in the future. Some things people talk about in the EtherREGEN context do not make sense to me (does not mean they are not correct, but give me pause).
Despite mirror ports and switch counters, I don't know what the Innuos streamer ultimately receives on its interface. On a MacMini off the dlink, I don't see errors at 20Mbps (iperf/UDP tests) on the internal network, but do over 200Mbps. I didn't try to engineer the whole path as I can't look at the whole Internet -- so that last part doesn't really matter for drops as long as they would be low level.
Its hard to work with my cabinet (I have to pull gear out to change cables), so I don't want to do a bunch more testing. And no, I did not try the dlinks off the P15.
To the greater glory of this shared hobby!
|