Not block, common mode noise can capacitively couple through to the streamer and pass down through earth. DC offsets as it were are galvanically isolated by the port, but these wouldnāt cause interference.
So yes best use a mains earthed switch directly upstream to āsinkā any common mode noise coming through other connected cables.
Further Naim learnt from their Gen 1 streamers and significantly decouple the network streaming transport module from the rest of the streamer.
1 Like
@Simon-in-Suffolk Perhaps you can help me here, as you know far more abut this than I do 
.
I have an Aruba 1930 switch in my comms rack (in a separate room to the music room) to which I connect all my Ethernet devices/cables routed around my house. This switch has a built in power supply, so the power cable provide an earth connection and runs from an APC SmartUPS in the rack. My Vodafone router and WatchGuard firewall (cascaded, with the WatchGuard in a DMZ of the Vodafone router) are also in the comms rack and are powered by the same UPS, although both have DC power supplies.
The Aruba switch supports IGMPv3 and is configured for Multicast on the VLAN used for mobile/media devices.
My NSC 222 in the music room is connected to the Aruba switch by Ethernet (RJ45 wall socket). All cabling is Cat 6 or 6A.
So, from what you are saying, it would appear I have largely mitigated any common mode noise, but is there anything more/better I can do?
All sounds pretty good to meā¦ and if you have no issues - then congrats
Thanks Simon. As Iām sure you have guessed my network is set up for IT use primarily (I also run a blade server and two NAS boxes for photo storage). The use for streaming is therefore additional rather than primary, so I just wanted some reassurance that it would OK for streaming audio as is. Thanks again.
Hi there MrFixit et al,
In principal, any network that works for IT type applications ie. Has the ability to deliver a working data stream will be OK for audioā¦.in exactly the same way a boom box ghetto blaster is OK for audio. To use an alternative word to OK, substitute āadequateā in that audio will work and music will play. My posts arenāt intended for people who are perfectly happy with āOKā or āadequateā. In my experience, digitally based streamed music can achieve quality levels that utterly blew away my high-end analog systemā¦.blew away in regards to noise, micro and macro dynamics, PR&T, rhythmic āresolutionā, neutrality, naturalness, listener involvement, emotional response (joy, excitement, atmosphere), dimensionality and spatial resolution, timing etc. etc.
A bog standard, non-optimized network will play music, but only a fully optimized network will create mind blowing experiences that make you feel like you do at a live eventā¦.where you want to clap and cheer at the end of the performance as a means of releasing the emotional āpressureā that the music creates. You may have read in various articles about a hi-fiās ability to communicate the musical message. This is precisely what optimising the network brings you. Optimising is about getting EVERY stage of the network to operate as perfectly as possible, because every stage (modem, router, bridge, switch etc) adds its own character to the final result. Thatās the difference between a network built to transmit data adequately and one built to produce the finest music possible.
10 Likes
Can you say what your best system consisted of in terms of naming the key components, inc. cables.
1 Like
Youāve convinced me. Some recent changes to my system has done exactly what you describe so eloquently. Iāve had some of the best evenings of music that Iāve ever had with these changes. Itās still not quite perfect - I think my LPS needs improving as it sounds better if I switch it off for a while and restart it. Itās fine for a while but as demand gets greater it seems to run out of grunt or itās maybe itās slightly unstable as things can just go off a bit. If I unplug it and switch it back on it seems to correct this issue. I really appreciate your input as it certainly pushes the envelope of audio performance for me!
1 Like
So how should I go about optimising my network for audio? And how will I know when it is? Iād appreciate any advice as Iām pretty clueless. Just trying lots of kit thatās marketed as being able to improve SQ over a network seems to be a bit of a journey and full of potential mis-steps and expense.
Thereāre pretty inexpensive steps that also easy to adopt logically. Basically you have to isolate your endpoint and routers. That means dedicated network line for audio only, putting routers/switches in a silent mode and mains isolation by whatever approach is available and works for you.
2 Likes
Hi Simon,
What Naim set out to achieve was to build a system that is as independent of the quality of the network as possible. Essentially this is a similar goal to DAC and Server manufacturers, who design their systems to be as independent as possible to the vagaries of the network.
What I have found is that itās not only the Server, streamer and DAC that are sensitive to the quality of the network.
If for example I replace a cheap SMPS on a switch with a high quality LPS I hear a major jump in sound quality performance. Similarly if I replace an ISP provided, Intel Puma-based router with a high quality Broadcom-based, tri-band unit and dedicate the extra 5GHz channel exclusively to audio, I get a major uptick in SQ performance. In order to hear the uplifted performance of the router, the improvement has to pass through the rest of the network, including any fibre optics, bridges, switches and servers it may include. This can only happen if the improved input to a network module results in an improved output. I.e if better in = better out. And this has to remain true for every single network component. If it wasnāt true and a component wasnāt sensitive to the quality of its input, you wouldnāt hear any improvement if even one component in the chain failed to pass on the improvement. What this in turn means is that the quality of every component on the network matters. And this holds true even when the transmitted bitstream was already bit perfect before the upgrade. If that is the case, then clearly the bit stream itself isnāt changed by the improvementā¦so if the bit stream itself isnāt changing, what is actually causing the improvement in sound quality that can be very clearly heard and indeed easily reversed simply by reversing the upgrade? Bear in mind that things like vibration control, cable quality, power supplies, DC cables, oscillator accuracy, switch quality all have similar effects in that their improvements can be heard as improvements in sound quality without changing the bit structure from what was and remains ābit perfectā. And in the same way that sound quality is improved by better network components, the same is also true when those network components have less to do. Less traffic means less work, less power supply demand, less noise, less interrupts, fewer errors, less latency etc. and like all the improvements mentioned above, the output of every network component seems to be improved when its workload is reduced. The earlier on in the network an improvement happens, the more times its compounded by every downstream component, so the earlier you can reduce unnecessary network traffic, the more components benefit, the greater the impact on SQ becomes.
2 Likes
Other than the network segment between switch port and streamer - there is no real concept of āearlierā on in the network. Naim use UPnP media transfer flows using TCP that are independent of network packet timing in their current streamers, so there is no impact to SQ. If there was here one would simply enable DSCP and configure it - but it now inconsequential so I donāt bother.
The flow is very non linear in terms of throughput - and on a typical home networks it appears as bursts of transfer of packets with differing time between the bursts, like flow burst, pause, flow burst, pause, flow burst, pause etc - rather than a steady flow.
Now with the Gen 1 streamers I had deterministically discovered that the inter frame timing consistency from flows from the UPnP server did give a distinct sound signature - to the point you could argue that some UPnP media servers sounded better than others, and even home networks where there was congestion (extremely unlikely) could make a difference. I shared my findings and engineering traces with Naim development whilst the next gen streamers were being developed- I built quite a good relationship at that time - and the approach was changed with the later streamers to the point such inter frame timing consistency became inconsequential, and the way media spooling occurred was changed. A lot of the improvement here was also possible because of the more effective and efficient streaming transport module that Naim adopted.
But the key point here these aspects were not about data integrity (for the most part) - but the noise signature caused by the processing of the digital streams.
This last aspect is fascinating, as these tonal signatures occur from several aspects of the streamer other than from the transport module. To the point Naim development, so I was reliably informed, had developed a firmware model where the code execution timing could act as a kind of parametric eq on the resultant DAC output through the noise signature created. This is controlled and optimised when key parts of firmware are optimised to make the resultant audio sound pleasingā¦ to the Naim in house āearsā.
5 Likes
Hi JimDog,
Let me start out by emphasising that network optimisation is beautifully scalable and can fit with pretty much any reasonable budget. I am lucky to have had a reasonably generous annual budget and because the results I was getting were so worthwhile I dedicated that entire budget over 5 years to the network and server. In over 55 years of hi-fi, chasing the Holy Grail of music that makes me feel exactly the same way that great live music does, optimising the network of a digital streaming system was the first time I actually achieved that goal, despite some very beautiful analog, fully active and vacuum tube based systems over the years.
In the end the results I got were certainly at the lunatic fringe end of network optimisation, but the total amount I spent was probably not as much as a high-end turntable, speakers or amp. It was only āa lot of Ā£Ā£Ā£Ā£ā when looked at in the context of the network. If the hi-fi as a whole was considered, it was no more than many audiophiles would spend annually. What this focussed expenditure illustrates is just how productive and rewarding I found concentrating on the network to be. My system just kept getting better and better, with absolutely no ālaw of diminishing returnsā. For sure the upgrades gradually became more expensive but by the same token, the returns just became more jaw dropping, at least in line with and IMO, exceeding what the same Ā£Ā£Ā£ could have bought when applied elsewhere in my system.
My system started with an Innuos Zenith MkII SE server fed from a Virgin Media Superhub 3 via 15m of ethernet cable, which sounded very nice indeed
My first upgrade was to add an AQVox SE switch, powered by a wall-wart SMPS, which brought a small improvement. I then replaced a Chord āCā USB cable with a USB retimer, powered with a Ā£500 SMPS which robbed the system of most of its previous magic. As soon as the Chord cable returned, so did the magic. It turned out it was the SMPS that was impacting system performance, so I replaced the SMPS on the AQVox with a Sean Jacobs DC2. This brought such a huge improvement I immediately returned it for a DC3 with Mundorf caps, which brought another huge jump in performance. I then added galvanic isolation by way of a TPLink RE650 Extender/Bridge, then upgraded the router to a TPLInk AC5400 Tri-band. I used a Meicord Ethernet cable between Virgin Modem and Router, placed the router on an Atacama base, upgraded the Chord C USB cable to a Synergistic Research Atmosphere X Ref. I modified the Bridge to remove the 230V plug and SMPS, added an anti-vibration wall mount and powered it with a DC3. I added a dual rail DC3 to the Modem and Router. I added a Melco S100 switch with Pink Faun OCXO mods, powered by a dual rail DC4. I upgraded the Innuos Zenith MkII to an Innuos Statement, to which I added Chord umbilical cables. I upgraded the Meicord WAN cable to a SR Active SE and upgraded all the DC3s to mini DC4s. I added an Innuos PhoenixNET with SR BLUE HC PC and upgraded the Statementās PC to a SR Atmosphere HC. I added a second Finite Element Pagode MR for the switches and power supplies and replaced all the PS DC cables with Mundorf Silver/Gold JSSG360 screened units. The Innuos Statement was upgraded with a Next Gen power supply and all the DC4 and Mini DC4 rails were upgraded to ARC6 supplies. Finally the SR Atmosphere Ethernet and USB cables between the PhoenixNET and Statement and Statement to DAC were upgraded to SR Galileo SX.
Then I woke one morning to find Iād lost hearing in one ear so the entire system has now been soldā¦one piece at a time.
The most important things to bear in mindā¦.not once during this whole sequence of upgrades was I in any way dissatisfied with the sound. From the first SJ DC2 power supply, the system delivered wonderful music. All upgrades were on the basis of; āif itās this damned good now, what ifā¦ā¦ā
Every single upgrade bar the USB retimer delivered what I considered to be excellent value for money
Every upgrade made a fundamental difference in that the music created greater joy, listener involvement and emotional response. Many upgrades are cosmetic, making a small uptick in certain aspects but barely moving the needle in terms of long term listener enjoyment. All these upgrades made a long term impact on the enjoyment of all types of music and all recordings, allowing me to enjoy music and discover new music that previously had not appealed.
Probably the 2 most amazing thingsā¦.how live, spontaneous, spectacularly 4 dimensional and utterly captivating full-fat streaming sounded and how close to an actual concert experience Swiss Radio Classics 128kbps MP3 transmissions of recorded concerts could get me.
5 Likes
Just a couple of questions.
What is your LPS powering?
Whatās the longest youāve kept the LPS switched on before powering down?
How old is the LPS?
The LPS is powering my EE8 switch. Itās not an ultra expensive LPS. I bought it off a friend who upgraded to a better one. Up until a change of Ethernet cable Iād been using the standard SMPS with great success. However, the new Ethernet cables showed me the shortcomings of the SMPS. If I power down the LPS for about three minutes then switch it back on everything is great for a while, then it can just suddenly go off and sound a bit naff. As I said, if I switch it off then on again it seems to rectify the problem. I have an itch for a Farad 3 to do the donkey work.
2 Likes
Instead of āearlier on in the networkā, substitute, āthe farther upstreamā. A network is nothing more than a series of modules designed to move electronic data following certain prescribed (standardized) protocols. There are 2 aspects to the quality of that electronic data. The data integrity and its physical implementation. Bit perfect takes care of data integrity by virtue of the bits having the exact same pattern as the original. The steps Iām suggesting influence the physical layer of transmissions, ensuring that elements like noise, jitter, errors/retransmissions, voltage transitions, latency, EMI, RFI etc are all minimized. In the same way a better power supply improves a streamer, a server or a DAC without any alteration of the bit pattern, all Iām saying is that my experience and that of others who have done similar experiments is that the same is true for the balance of the network. The more perfect the physical implementation of bits or parcels of bits, the better the sound quality based on better in = better out. Many manufacturers have designed servers, streamers and DACs with a goal of minimizing the effects of the network. A pioneer in this endeavour, Taiko Audio eventually ended up designing routers and switches specifically for high-end audio due to the fact they could achieve such massive improvements in the final sound quality via this route. They found that something as simple as an RJ45 network cable socket can be substituted to produce less noise, because everywhere on a network where work is done, noise is produced. Reduce that work and you reduce the noise. Its that simple
2 Likes
Hi Geko,
Iām not an EE but it sounds like you may have a problem with the power supply. I would have an engineer check its short and long term stability under various loads to make sure its functioning properly.
Cheers Blackmore, yes Iāll get it checked but, like most things in life, you tend to get what you pay for. I will probably try to get a listen to both the Farad and the Phoenix Net again as they did impress me the last time I had them on loan and with a far more transparent ethernet cable they should be pretty epic. Thanks again for your input.
2 Likes
Do a blind listening test, there is no difference between standard Ethernet and so called āaudiophileā cables.
If they made a difference, why do Naim, Linn, Rega etc not produce them? Indeed Linn only recommend bog standard Netgear switches.
2 Likes
Which high end cables and switches have you tried?
1 Like
Standard Cat 5e, and a Chord company cable at a local dealer. No difference.