Which Switch, and does PoE make a difference?

Blythe, good stuff.. there will almost certainly be more detailed plans either created or in the making, but might not yet be in public domain. Interesting your initiative is at district rather than county level as in Suffolk. Not that it matters if the job gets done.

What will be interesting for our village parish is that although the village boundary is small and centred around the church, pub and green, the village parish is  spread out over over a large area and indeed has multiple area dialling codes with outlying houses and farms .. The area dialling code (unless ported) will follow the copper lines back to the relevant exchanges.. All outside the parish.. Now when we get the VDSL cabinet in the village, those that share the area code as the core village  in the parish have a good chance of superfast unless really outlying, but there is a slight question for those served by a different area code... we will have to wait and see.. I should be ok as I live in the village boundary itself.

Nothing is straightforward. I trust your parish/village doesn't have such a complication.

Simon

Simon-in-Suffolk posted:

Mike - I will look at your link - but underlying chipsets spring to mind...

The ubiquitous Broadcom's & whatever other makes.  I think mine has a Broadcom,  but its hidden under a heat-sink for some reason;  whatever I don't think we need to go down that goat trail,  do we ?? .  

Mike, I agree, I read the article and found it facile and therefore not at all useful.. Other than, yes, different network switch products can affect the sound of audio equipment connected..

Yes heat sinking and with larger switches additional forced cooling is nearly always required.. when a switch starts filling up there is a lot of data being switched.. But yes I guess if one was being objective that path would need to be trodden 

Mike, look forward to it..

Simon

Thanks to everyone for your input.

I put the 2960G in this afternoon, and at least all connected devices are talking to each other.

Mike-B, Please can you point me in the right direction regarding purchasing the low noise SMPSs?

I have a number of GS108s I wanted to experiment with

robgr posted:

Mike-B, Please can you point me in the right direction regarding purchasing the low noise SMPSs?

I have a number of GS108s I wanted to experiment with

 

If you search for iFi iPower you might find something that fits the bill. I noticed these a few months ago, and thought they looked like a possibility, and I think Mike's posts were about the same one.

robgr posted:

Mike-B, Please can you point me in the right direction regarding purchasing the low noise SMPSs?

I have a number of GS108s I wanted to experiment with

 

I am a total convert to Cisco Meraki. Not cisco, but cisco meraki. 

Jon, yes I work with Meraki a bit, and it does have advantages in terms of centralised wifi access point management ... but not sure of the relevance to this thread?

Any modern wire speed gigabit ethernet switch should work fine. The switch, assuming it is not malfunctioning or dropping data and the ethernet cable itself as long as it is a non-defective Cat 6E certified cable will not impact the sound in any way. Ethernet is a transformer isolated system and the grounding and isolation of the NDX itself separates the electrical signal from the data stream sent from the Ethernet PHY to the streamer. Don't fall prey to this garbage about fancy ethernet cables, Cat 6E cables are highly shielded and none of the "precious" cables test better electrically using a Fluke DTX or any device that measures cables performance - it is absolutely the domain of con artists. On the other hand, the more expensive switches are generally managed and you can get statistics, hard code speed/duplex setting, potentially do routing and other features. If you want that flexibility go for the higher end switch. Otherwise The NetGear ProSafe 16 is a good switch, so is the Cisco SG-110 if your looking for a 16 port you won't go wrong with either. PoE is only necessary if you plan on using devices that draw power from the switch - This is almost always IP Phones, Enterprise wireless access points and some webcams/security cameras. If you don't plan on using any of those type devices, don't bother with PoE.

Dupree, I think you'll find that many including me have found data switches connected to the other end of the Ethernet patch lead connected to the audio equipment DOES AFFECT the sound. However don't fall prey to assuming this has to do with TCP/IP related matters since as you summarise it is almost certainly nothing to do with It. It is most likely to do with common mode EM currents and almost certainly patch lead transmission line factors.. Texas Instruments provide a good white paper on conducted Ethernet noise related design considerations.

Of course some switch manufacturers go to some extent to minimise such side effects .. for reasons that I suspect are nothing to do with the high end consumer audio market.... In my experience Cisco is one example (according to thier EM device certifications which include voluntary as well as mandatory) and take EM interference seriously which we can take benefit from if we wish.

Also for Naim one only needs to use 100 Base TX switch and not 1000 Base TX.. anfd from a point of view of EM interference the two pairs of a 100 Base TX connection is almost certainly preferable to the four pairs of a 1000 Base TX connection.

Simon

Simon-in-Suffolk posted:

ok yes its worth a punt with one proviso - the 24 port I believe has a fan and so will have an acoustic noise footprint. This may or may not matter - it depends where you put it. The smaller 8 port 2960 switches have no fan and so are quiet. 

Simon makes a good point. As switches grow larger and more complex they get louder. I installed a higher quality/quieter fan in one of the larger POE switches at home in order to reduce the acoustic noise. I was quite surprised at how effective the new fan is. It's gone from being noisy to inaudible. I used a product from an Austrian company called Noctua.

sorry to hijack this thread:  but on the 'ear' website that Mike-B posted, i can see switches daisy chained and paralled with nothing, apparently, conected to the router except the cables for the various switches -- and audio separated having dedicated switches (in fact nas drive separated from streamer for example).

my question is: (a) am i reading this right? is that a good idea? if so, can i do it with GS108 switches? anyone played with optimum configs with these multiple switches?

sorry, lots of noddy questions...

enjoy

ken

Any time you can avoid daisy chaining it is preferable in Ethernet networks. I have been involved with ethernet hardware and designing ethernet networks for over 20 years. In the audio space there is just a fountain of misinformation as to how ethernet networks work and things that will purportedly make networks "sound better". Every switch hop a packet goes through introduces some nominal amount of latency and possibly some nominal amount of jitter - there is no time that it is desirable to add additional switch hops unless it is required to scale. Larger networks usually follow a star topology to avoid these kind of problems. Ethernet itself is a transformer isolated system and electrical noise is isolated at the PHY so connecting your NAS or switch directly is fine. The only possible piece of concern is the actual noise that cheap switch-mode power supplies may introduce into your environment, so maybe an AC filter or one of these purported lower noise wall-wart adapters could be useful. However in the advent of cellphones, tablets, consoles etc, everything has a Switch-mode power supply which are inherently noisy so unless you actually have your system on a separate filtered spur I doubt this would make much of a difference either. The simpler you make your network the better - it will cause less headaches and less time troubleshooting. 

Agree about lots of misinformation on data networks and electronic engineering in the world of consumer audio... including on UPnP TCP media transfer as used by DLNA servers and Naim streamers  - network jitter is a meaningless concept. The media is transferred using the TCP transport method and therefore a system of windows and sequencing is used.. jitter has no meaning.

Now if we were using UDP datagrams as used for VoIP systems etc as used in IP telephony then yes jitter comes into play - but compared to jitter in SPDIF data streams for DACs the values are hugely different and the the network jitter is massively less sensitive... for G711 typically 20mS or less.

BTW the the electrical noise and EM is not isolated by the galvanic isolators (or as you say transformers) - the galvanic isolators are there to avoid ground loops and other issues from DC. Common mode EM currents will traverse along the twisted pairs in an Ethernet lead. Usually not an issue - unless you need very low noise in your ground plane - such as in quality audio circuity and audio digital clocks where care is then needed.

Simon