Nordost Network Switch

Hi Chris - I don’t think so per se… with respect to twisted pair the reason 100 mbps is preferable in my opinion is that only two twisted pairs are used - with 1gbps 4 twisted pairs are used - and that therefore is potentially why 100 mbps is preferable

In a way higher frequencies with respect to the audio band are preferable as they are less likely to interact with audio frequencies and audio frequency electronics and are easier to filter.

1 Like

Hi Simon, I am doing this myself, keeping it simple. I have decided to purchase a Lumin P1 streamer/Dac which I should have in a couple of weeks. Two boxes, the Lumin and my Nagra integrated will be all I will use.
I have a question for you about optical fibre.
I use a Cisco 2960 as per your recommendation a few years back. The Lumin has a regular Ethernet input, plus a separate fibre input.
Is it possible to insert one of those fibre sfp modules into the 2960 to try this input on the Lumin?
If so which port would I use?

2 Likes

The 2960 has like a billion in the range but if it has fibre there will likely be two, use one of them.

1 Like

Yes you can if your 2960 has an SFP module transceiver port. Not all models do, but some have a module port or two on the uplink ports … if it’s there it will be the separated ports on the right hand side.

Ensure your fibre mode and type is the same at both ends of the link.

Multimode LC Duplex SFPs with OM1, OM2 or OM3 fibre would be good, subject to your streamer end.

You will also get the benefit of being able to check the SFP module on the 2960 to see if you are getting data errors on fibre… this is more likely to occur, though still unlikely, compared to twisted pair over short lengths… oh yes do check the SFP module is supported by your Cisco switch.

1 Like

Thanks Simon, and Garyi.
I also got some info on the Hifi Corner under the NEXIT thread. I got lucky and managed to write down a bunch of links to parts (Yellow-banana) suggested before Richard removed them.:grin:
On the other side of this, a guy that has a P1 on another forum said he tried fibre for a while, but ended up switching back to copper Ethernet.
He said the music lost it’s “groove” with fibre.

2 Likes

Well it’s unlikely to be the use of fibre per se… but considerations… if using cheap media converters… it is possible the serialisation clock is less stable that that of the connected switch so the fibre link would be potentially noisier than twisted pair in the host streamer in a detracting way.
Alternatively there may have been an element of common mode noise in the setup prior to the use of fibre, that the listener learned to enjoy… it might have given a sheen or edge to the sound to make it more exciting (this is all very subtle… but it can be there) and when removed it sounded like something was ‘missing’.

But yes use of fibre doesn’t make things automatically subjectively preferable

There is no harm in playing around to explore… but my view is decouple precision transport from the DAC function as much as possible, and then upstream network considerations become largely irrelevant. I find the current digital signal output on the current Naim streamers excellent for this purpose.

1 Like

Do music streaming formats and protocols contain error correction features? If yes, won’t that take care of data transmission errors? Don’t the streamers themselves buffer and re-clock the incoming signal to remove timing errors? Re EMI, surely there is good isolation of the Ethernet circuitry from the audio components? I’d be most interested to hear the views of a network expert and analogue designer on these issues because I struggle to understand how Ethernet switches could have an audible impact on perceived audio quality. I may not have super high-end kit but my ND5 XS2 is fed by a cheap powerline adapter connected over mains to a commercial switch and a home grade pfsense router and I never hear any glitches, just enjoyable music.

1 Like

No not generally… streaming uses httphttps data transfers.
In TCP/IP use any error correction at the’network’ layer is performed by the transport layer. Here any data loss / corruption that is detected, is requested to be resent within a small time interval.

Direct protocols SPDIF have no error correction, but can can detect errors, and discard the errored frame.

There are no timing errors in TCP streams… that affects the payload… the payload is reconstructed into a partial file by the streamer and then a new click serialises that’s data into a stream, typically using a precision clock.

Some people get confused by network serialisation clocks and sample clocks… they are completely different and have no connection. The only commonality is the word ‘clock’.

However network clocks that are relatively unstable act as a tiny FM transmitter, transmitting noise across the network twisted pair links. This noise can couple in the host and create noise modulation I’m ground planes or powerlines or through EM radiation. This noise can modulate precision clocks and / or analogue electronic stages… all extremely subtle… but clearly audible to many.
This is how switches can affect audio.
I realise if not an electronics or digital electronics engineer, this can be not that intuitive… but my dissertation many years back in computer engineering had me delving here… so it all feels quite natural to me…

You are not alone

In my extensive experience they do not make a difference. You have to understand the people that choose to spend this sort of money on a switch, its certainly not about music.

2 Likes

So zero data loss to the streamer and no timing issues which leaves only EMI effects. This vulnerability to known about, so one assumes designers isolate sensitive components through physical screening, physical distance or additional filtering? Aren’t NAIM renowned for attention to detail? Which brings me back to, how can there be a perceivable residual effect? My network setup is about as near to the coat hanger connecting the speakers scenario, and I could always do a blind test by streaming direct from a USB stick, but I bet my old ears wouldn’t hear a difference. (p.s. please don’t tell me you can buy expensive “audiophile” USB sticks…please!)

My advice would be to use genuine Cisco SFPs for the switch. They are much more expensive that the various non-OEM brands, but like the switches, you can buy them for next to nothing on ebay.
If your switch hasn’t been configured in some way by a previous user it will then be ‘plug and play’ simple.
I’m sure Lumin can advise you regarding a compatible SFP at the other end.

1 Like

Well to an extent… from a digital encoding perspective on a TCP based network there is unlikely to be data loss, however digital encodings only exist in the world of analogue encoding in terms of scalar quantities and time.
When one switches and handles digital encodings there will be analogue interactions. It is this interactions that can (in a closed system will) interact with other functions… albeit at varying levels from not being measurable to being quite noticeable.
All electric flows have electro magnetic fields associated with them, and changing current will always have a reaction. I wouldn’t call it interference… as these flows are generated by the system itself and is a consequence of natural physics … not an external source of electro magnetic fields. I would call it ‘real world’ vs ‘architecture models’.

So yes to get behind why ‘digital’ interactions are audible in a system, typically one needs to the analogue domain.

This was an area where Naim put many resources into improving in the current streamers( including using LVDS, improved screening, improved ground and Powerline decoupling) … it’s not perfect, what ever is?, but it is significantly improved in my opinion over the first gen Naim streamer products.

Just a point that may come up … when one has a digital bit stream, there are three values… true, false and time, so time becomes as important as the binary value in such scenarios.

As far as audiophile USB sticks… I am sure they must exist somewhere… certainly on my venerable Naim DAC different vendor usb sticks sounded different from each other slightly, and the front USB port sounded slightly different to the rear port… this will be nothing to do with the ‘digital’ data.

In my experience when using the first gen streamers I did find network changes and even UPnP server types were slightly audible … whether using digital or analogue outputs.
With the UPnP server differences, I analysed a cause of this, and volunteered my analysis, (logs, test streams etc) it to Naim at the time when they were developing the current streamers, and was later told how they had mitigated this in their new designs.

Anyway with the current Naim streamers feeding a separate DAC using the digital signal output, I can’t hear any effect of differing network components and set up at all…. as long as it provides the throughput…. and when you get to the point of achieving this strangely you cease obsessing over it :grinning:

1 Like

“I can’t hear any effect of differing network components and set up at all…”

So how can you explain that so many users elsewhere, with the best servers/ transports available ( like Taiko Extreme, Antipodes k50…) connected to the best external dacs ( DCS Vivaldi, MSB …) still hear a noticeable improvement when using costly audiophile switches and costly linear ps ?

3 Likes

How do you explain some users with top Naim systems hearing a difference after updating the app?

And of course a million things in the history of hifi where people were convinced about differences. We have discussed many of them from magic marker on CD edges to the device sold by Van den Hul that is not connected to anything and “communicates directly with the brain”.

As inconvenient as it is, the brain wants to experience differences and patterns, and it invents some if it can’t find any. That’s because it evolved this way when looking out for a sabretooth tiger in the bushes, and a false positive was preferable to a false negative. And everybody knows and accepts this in all other domains, just not in hifi.

Edit: OK I am violating my own rules again :face_with_hand_over_mouth:

3 Likes

Only read this post edit, and will understand if you don’t want to point out what you’ve edited out. But I’m just curious to hear what rules you’ve set yourself around this.

1 Like

Trying to be objective, “noticeable” would surely mean a) differences were measurable using conventional test equipment at the output transducer (speaker) that are detectible by anyone with average hearing and/or b) there was a plausible scientific explanation for any perceived differences that could be measured but were outside the range of normal human hearing. Does anyone claiming to hear improvements ever come up with objective evidence?

3 Likes

Just to try and shut up about it and let people hear what they want and have their fun. I’ve given up essentially. I’m doing pretty well with it actually, but sometimes it is hard and had just the same discussion with FR in the other currently ongoing thread about the Synergistic Research switch, around the same price range and with a carbon fibre chassis as if it was going 330 kph around Silverstone

I think the issue is that the other side isn’t always about live and let live either. And there’s more of them…

But I understand it gets tiring. I’ve also been trying to keep quite a lot.

2 Likes