This thread is intended to discuss the subject of optimizing a network to obtain best sound quality. Some may wonder whether such a thing is possible, or whether optimization makes any difference and that’s ok. Read, participate, argue, but do it with kindness and politeness. I’m happy to have my ideas questioned but if you want to state that something isn’t a particular way, please provide counter arguments and not just denials. I am first and foremost a system developer and not a network or IT expert. My measurements are purely subjective, but I do have a lot of experience listening to some interesting hi-fi (Extremas, Guarneri Homage, Avantgarde Trios, BAT REX system, Naim 6 pack, Isobariks and SBLs, Magico S1 MkIIs, Devialet 440 Expert Pros, Innuos Statement) and in building and developing the systems with optimum mains supplies, single vendor cable looms, room treatment, vibration control etc.
Over the years of using this kit the goal was always to get the best from each systems in whatever way improved and increased their performances. My most recent system was built for local and remote streaming of my own DC rips and files and from Qobuz and I actually found the sensitivity of SQ to the network quite by accident when I added what I thought would be an improvement that turned out to be a substantial downgrade. The item in question wasn’t the problem, but its SMPS-based power supply was in that it robbed my system of a lot of the magic that I was loving. After 45 years of trying to achieve utterly satisfying sound I am pretty picky. The problem is that any even slightly identifiable faults really irritate. I want to be involved in the music 100% with no distraction and no effort and that takes a pretty sonorous system to achieve.
So i removed the offending component, but it got me thinking about another SMPS i had in the system, one powering a switch in the network. I first tried replacing it with a Sean Jacobs DC2 and the results were so good I sent it straight back and got a DC3 with Mundorf capacitor upgrade, which uplifted the sound further. So please was I with the results that I added another dual rail supply to my modem and router and was again rewarded with a substantial uptick in SQ that IMO was well worth what the supply cost. One of my biggest surprises was substituting some already good Neotech DC cables with Mundorf Silver/Gold cable ina JSSG360 screened format. They literally made the system holographic in its presentation.
So by now I can already feel some bullshit detectors going off, so let me digress a little.
The last 2 stages in processing music cannot be measured because they are in your head. Your loudspeakers create 2 channels of carefully prepared stereo signal. These are essentially mono signals from the source that are processed in stereo in order to either preserve or impart spacial information, which is encoded as a range differences in phase (timing) and amplitude. This creation of soundwaves is the last roll your hi-fi plays in creating music and your brain takes over the processing of the soundwaves to first convert them into nervous impulses then to convert them into conscious music, with all the elements we recognise and cherish like melody, rhythm, tone, spacial definition, ambiance, timing, energy, drive and so on. All the musical attributes are created by the brain and if we look at what the brain is doing, its using the differential between the signals reaching each ear to build all these wonderful qualities listed above. If you think about it the differences in amplitude between the same signal reaching each ear are fairly small, ditto the phase change and timing difference, so not a lot has to change in the signal to make quite a difference in the brain’s processing. Said differently, the brain and therefore the signal are very sensitive to very small changes. And what seems to be happening in the case of the network, is that changes to the way the data stream is generated, carried and processed, for example by using better clocks, power supplies with little noise, superior cables etc. make enough difference to the way the bits are processed in the DAC for your ears to clearly pick up probably very small but equally very important changes to the sound waves, resulting in the brain being able to construct a more information rich conscious awareness of the music.
So that’s the background….what specifically have I found to make a difference? Power supply quality (noise, ripple, impedance), DC cable quality and screening, ethernet cable quality and screening, vibration control, clock timing accuracy and stability, network traffic, network isolation and network topology and finally the router (esp. its chip set).
Over the years, my upgrade philosophy has concentrated mainly on the hi-fi hardware…for example a very nice Linn LP 12, Naim 32, 250, Isobarik system ended up with a six pack of amps, NAXO SuperCap, Aro, CDS etc etc. but with this most recent streaming system I am finding that the most productive area for upgrade with by far the most bang for the buck is the network or rather, the elements from which the network is build so this is the topic i wanted to share 
