Funnily enough, I read the other day someone saying that IGMP Snooping should be turned off to prevent it breaking things like UPnP.
This doesn’t appear to make sense given that it exists (I thought) to help with not hinder multicast traffic (but, then, many things have been invented to make life easier with unintended consequences!).
Not quite actually; IGMP Snooping was designed to help prevent layer2 Switches from having to flood all ports with multicast traffic by snooping the layer3 IGMP packets to know which port are actually participating in multicast groups, and filtering out multicast traffic for all other ports.
It’s becoming less and less important these days for consumers as folks migrate towards unicast video services (i.e. Netflix etc).
But it shouldn’t actually break multicast while doing so 
I guess it’s one of those well intentioned proposals!
It doesn’t, normally.
A proper Switch with configurable IGMP Snooping usually has configurable timeouts to ensure applications that rely on some aspect of multicast doesn’t break. Managed Switches which have it On by default, usually have appropriate timeouts that avoid breaking applications like UPnP.
Unfortunately, it seems the unmanaged D-Link GS-108 used by Ansuz is just rubbish. 
That’s interesting, as I was beginning to wonder whether Ansuz were actually using a managed switch for the innards and had set it up incorrectly.
The Ansuz must sound substantially better than the PhoenixNET if @T38.45 is going to persist in sorting this.
Many thanks for responses- appreciate folks!!
Ok, yes - the SQ is superior, it’s the best switch I had for testing (I had Melco, innuos, Cisco, Ansuz, some cheap other vendors as well…). And yes, it seems that they’re still using the D-Link engine. So the IGMP problem is not fixed (I hoped they would fix it with Gen3.)
I would compare their approach with a DAC vendor- many DAC vendors are using ESS chips but their DACs sound all different because the chip is one part of the chain. Powersupply, filter, clocking etc is very important too. Look f.ex. what Audio Note or Naim are doing (very old chips…).
So here the point that drives me nuts- it seems that one or two folks here got that switch up&running in a dCS setup…
Again- thx! I will send the switch back again…
Interesting re SQ, I’ve always considered the PhoenixNET the one to beat but have yet to hear an Ansuz. Not sure I buy the Tesla coils thing but they’re clearly doing something right to your ears!
Agreed. Far more to a switch than the board. Or there damned well should be if it attracts the audiophile tax!
Whoa. Don’t rush. If this is true, you need to do some sleuthing. Obviously there is something, er, special about the Ansuz which means it doesn’t work with UPnP for you while other switches do; on the other hand, if other folk have the A2 or A3 working in their dCS setups then why doesn’t yours work? There must be something somewhere in your network/topology which is different from theirs. Have you investigated?
If you really do like the switch so much, you basically have to make it work with UPnP or settle for an inferior (to your ears) non-Ansuz switch. I’d personally exhaust the first of these.
Good luck, whatever.
Indeed Nigel, I had a call with a customer (my dealer initiated). He runs AVM/Fritzbox as well with a Mcintosh amp, Bartok and Ansuz A2 switch but he’s using a Mac with upnp service on it- maybe that’s reason why the upnp service/server is always visible (Mac is permanently broadcasting ?)
Anyway- If you can try to listen to an A3 or even better a D3 
I don’t know if this is helpful but I had some trouble initially with my Ansuz D2 Powerswitch but it was due to a failing ATT router. The Powerswitch actually worked very well with the Rossini Apex.
Thanks Lee- does it mean you’re running Rossini + dnla/upnp? What NAS server do you use!
Tx!
Well, unlike analog components, “fine tuning” the PSU and Clocking sections are relatively insignificant to an Ethernet Switch functioning as it should.
Fact is, Ansuz uses the entire D-Link Ethernet Switch board, not just a chip or “engine”, even to the point of pasting a sticker to cover D-link’s PCB labels. It’s just absolutely hilarious 
That said, doesn’t Ansuz have any support channel to properly address this issue? Have you asked them directly about this? (Although, from what I’ve seen, I suspect Ansuz doesn’t have the expertise to properly address or fix this).
@Anupc Looking at the Ansuz website, it would appear they’re using something similar to either a bank of modified inductors or transformers to clean up the power (it’s difficult to tell).
What do you make of these components? I’m not sure I’ve seen anything quite like them (similar, but not the same).
Thanks Anup- I’m in contact with my dealer about that issue again 
.
There’s nothing wrong with using a stock board, as long as you choose a quiet one! There’s a lot a designer/manufacturer can do off-board which can contribute at least as much (more in my case) to the performance of the switch as the circuitry. The sticker is pretty funny though…
But where is than the value adding?
Torben
You may check this here 
THX - but there is no information why this switch would sound better compared to other good standard switches.
Torben
Just listen to it. As I mentioned earlier here I tested Innuos, Cisco, Melco, and some standard switches in an A/ B test. The SQ improvements were quiet obvious. And frankly, they could put pudding in it when it sounds better😀. I have to explain technology in my business but not in my hobby.
Nowhere in the digital domain, as established elsewhere. Which is why the board itself is not the point. Or shouldn’t be.
I am just connecting a router. No NAS although I am considering a Roon Nucleus+ for later in the year.