Networking Logic - Intro

Blackmorec

Newbie
Wammer
Feb 6, 2019
90
65
23
About 5 years ago I started setting up and optimising a system specifically for local and remote streaming. I finished the complete optimisation in February ‘23, 4 weeks prior to entirely losing hearing in one ear, which resulted in my tearing the whole system apart and starting to sell all the components. But before it disappears entirely I wanted to share some of the logic I followed. Why? Because it resulted in the finest and most exciting musical presentation I have ever heard in over 50 years of chasing the hi-fi dream, including some genuinely top end vinyl, massive horn loudspeakers and all-tube amplification and CD replay.
While building my system I developed a logic for the network, which I tested multiple times and found entirely robust, with each improvement providing genuinely unexpected results that delivered hugely increased realism and more important, bags more of the emotion and musical involvement that we audiophiles crave. When I explain what I did, I hope you’ll see the inherent simplicity and logic that explains why certain things are true and give the results they do. This is a long story, so It’ll probably take a few posts and some reading stamina on your part. but I believe that if you are indeed chasing the audio dream it will be worth while reading. Just to mention at this juncture, I’m not on some sort of ego trip, rather this is part of my healing from a profound loss. I’ve lost my hearing, I don’t want everything I learned to be part of that. In losing my hearing I genuinely lost the keys to paradise. It would be a shame if they disappeared without a trace
Part of an audiophiles’ life is ‘upgrading’ and that is usually driven by the desire to get greater access to the intense emotions and feelings that music can generate. Music has the ability to stimulate our subconscious pleasure centres and the better the sound quality gets, the bigger and more intense those emotions And feelings become. Humans are pleasure seeking animals, often to the detriment of the organism (drugs, alcohol, overeating etc). Those pleasures are very addictive…the greater the pleasure, the stronger the addiction. When the finest music is presented in all its glory, it has no less power to create addiction. That is what this is about…..making music that is utterly addictive.
So let’s get into it by establishing a foundation. I would like to start by making a controversial statement. If you goal is to achieve the finest sound quality, a perfectly functioning, fault-free network is not enough.
Networking was developed by various standards committees to move data quickly, cheaply, reliably and seamlessly between devices. During that development, not a single thought was given to the resulting audio quality that may result. It was and still is assumed by the majority that all one needs in digital is bit perfect transmission of data but that simply isn’t true. A few top companies like Innuos and Taiko have discovered this untruth and set out to remedy the situation, with often spectacular results. I am not a network expert (although I was closely involved in the networking standards evolution) and most of what I have learned has some from observation and the generosity of people who are indeed experts in the fields of both network AND Audio.
So, the logic!
Let’s say that I have a simple network of ISP provided modem/wireless router, a Wi-fi -ethernet bridge, maybe a switch, a server or streamer and finally into a DAC. And let’s say that I decide to improve on that really crappy ISP device by adding a better power supply instead of the cheap-as-chips wall wart SMPS that the ISP provides. What happens when I do that? Well if the new PS is indeed superior, I hear an uptick in sound quality. Nothing revolutionary there, BUT such an improvement has a profound implication. In order to hear that improvement, it has to pass through my entire network, so first, the output from the ISP device has to have been improved, which means that the input to the wifi-ethernet bridge has improved, which means its output has improved and so on. The improvement ripples through the network in a better in = better out basis. Just like the rest of hi-fi where a better input results in a better output, an axiom I have never seen disproven. For an improvement early on in the network to be heard in the music, that initial improvement must pass through the entire network, and the only way it can do that logically is better in = better out. Some may claim that they can’t hear a better power supply, and I’ll get to that, so bear with me into part 2 (I warned you this will be long)
 

Blackmorec

Newbie
Wammer
Feb 6, 2019
90
65
23
Networking Logic - Part 2

So in part 2 I want to consider what a network data stream actually is and how that affects how we treat it. In part 3 I’ll consider the network itself.
So again to remind you, I’m not an IT or networking guru, simply someone who knows and talks to some very knowledgable people, has 50+ years experience with hi-fi and 45+ years with building and supporting complex systems.
So let‘s start by comparing an analog signal to a digital signal. An analog signal is essentially a voltage representation of the music we want to listen to. It has three ’characteristics’, namely: frequencies, amplitudes and phase. The analog signal is used directly to drive an amplifier which makes a ‘larger‘ more powerful version of the same signal. Any contamination along its journey from source to amplifier and speaker becomes an integral part of that signal and once added, cannot be removed because noise can’t be differentiated from signal, so the secret of good analog is cleanliness. Add nothing, because whatever you add becomes a permanent feature of the signal such that what you eventually hear is signal plus contamination.
Digital is very different. Digital still has a voltage component but essentially the original analog voltage is encoded into a time based ‘representation’ of the initial voltage signal. The basis of a digitally encoded signal is the bit. The bit is nothing more than a slice of time with a binary state…..on or off, up or down, high or low, 0 or 1. In order for the digital signal to be able to move along wires, we encode the binary state as a voltage….0Volts representing off, down, low or 0 and 2volts representing on, up, high or 1. So a digital stream passing through a circuit or down a wire is nothing more than a voltage that switches between 0 and 2 volts on a fixed time base. Given sufficient number of ‘switches’ we can encode frequency, amplitude and phase. We can pack those bits together as ’parcels’ and send those along a wire as ‘bursts‘ of information. But here’s the key thing. As long as we have the sequence of ‘1s’ and ’0s’ we can accurately rebuild the signal AND in doing so, completely exclude any contamination, as long as it doesn’t alter that sequence of ‘1s’ and ’0s’.
In summary, a digital stream is nothing more that voltage in a wire that switches between 0 and 2 volts on an agreed time base.
When we have the correct sequence of ’1s’ and ‘0s’ we can retime them and we can rebuild their voltage representation and in so doing, leave behind any noise and contamination. Now we’ve already discussed that better in = better out, so how does that translate here? Well first we need to discuss the term ‘better’ and how it applies to the digital signal. The digital signal has certain ’characteristics‘….its time base, so timing accuracy and variability, its voltage pattern, so how accurately and stably the 0 and 2 volts are delivered and how quickly it can switch between the two and the amount of accompanying noise. So when the input to any network
device is improved, we can see that the improvements are all to the ’Physical Layer‘.
So, from this section we can conclude or theorise that improvements to a network’s Physical layer are what ultimately contribute to improved sound quality. So far, this is theory, so we need to test that theory and see if it holds up….which brings us to part 3
 
Last edited:

Blackmorec

Newbie
Wammer
Feb 6, 2019
90
65
23
Networking Logic - Part 3

Here I’d like to discuss the network itself and how we can use it to dramatically improve our music listening experience.
As I’ve already stated, the network is a convenient and reliable way to move digitally encoded data files from one location to another. To achieve that goal, the network requires several different modules in order to fit into an existing environment and topology. Your hi-fi may be remotely situated vs. the incoming network stream, it may require different data formats, it may be part of a very simple or extremely complex infrastructure etc., so practically every network will be different. In order to manage all those variances we need an approach that can cope with any and all situations.
So, I’ve already said that a perfectly functioning network is insufficient for achieving SoTA audio. This section will explain why and what to do about it.
A network is nothing more than a set of different processes linked together to form a ‘whole’. There will be modems, routers, bridges, extenders, switches etc, and it may be based on ethernet cable, wi-fi, fibre-optic links or a combination . From an IT perspective, the network’s only function is to deliver data files reliably, quickly and generally speaking economically to the end point where they are ‘converted’ into something useable by humans. In general, properly set-up networks are good at doing this. ‘Properly set up’ infers sufficient bandwidth (speed) to avoid bottlenecks, low latency and wi-fi with sufficient signal strength to avoid drop-outs. Ideally a network is able to transmit files ’bit perfectly’ such that the bit structure of what is ‘sent’ is identical to what ‘arrives’ . So where does audio vary from this? Essentially in one main area, namely the physical layer that we discussed previously. In networking the physical layer needs to be good enough to get the job done. In audio it needs to be as good as possible because anything that impacts the physical layer impacts sound quality and that’s true across the entire network. Getting back to the initial axiom better in = better out, you can see that one network component impacts the next, For example, the better a bridge’s input, the better its output. If that then goes into a switch then the bridge‘s better output will impact the switch’s output and so on. So if we accept that improving an input/output entails improving the physical layer, then in audio, the role of the network in addition to delivering files is to clean and improve the physical layer of the data stream. This is where we start getting into architecture and topology. Let’s go back to that initial ISP modem/router power supply upgrade, where we substituted the SMPS for a LPS. The result of doing that is an improved physical layer of more stable voltages and less noise going into the next component. But if the next component has a noisy SMPS, much of the gain made at the modem/router will be lost. Similarly if we implement a really good clock oscillator with a spec of 3ppm in a component then feed its output into the next component with clock spec of 100ppm, we lose most of what we just gained. This is exactly why when some people add a power supply or a clock, they hear very little difference, because the improvement is lost further down the chain. When we build a network for audio, the goal is to use the network to refine and clean the physical layer, so its essential that every stage along the network has the capability to further refine the physical layer associated with its input. In that way you build a cascade of improvements, where the router improves the bridge, the bridge improves the switch, the switch improves the server and the server improves the DAC. In that way, an initial improvement to say the modem/bridge is compounded at each subsequent stage such that a minor initial improvement becomes quite major by the time it arrives at the DAC. Ideally for audio, the network takes a signal with an adequate physical layer and refines it and refines it until it is as close to perfect as possible when arriving at the DAC.
In addition to the above, there are several other aspects where attention to detail pays off. Power supplies are the most important elements of a network, as they all strongly influence the final music’s presentation. Vibration control is another area that pays dividends and cables of course.
I’ve said that one of our goals is to reduce noise and some of the worst noise in an audio network is other non-audio related traffic. Traffic is noise, the more traffic, the more work required and the more work, the more power is needed and all that produces increased noise. Ideally the audio network should be a quiet backwater with very little activity, especially during music replay. Most households will have a single incoming broadband feed, which has to do everything including audio and video streaming, games, support of mobile devices and a host of other things from smart plugs and lights to networked fishtanks or burglar alarm control. The first consideration in an audio network is how to separate the audio related traffic from everything else and the earlier you can do that the better. I use a 3 band router, with one 5GHz band dedicated solely to audio, so that the router is the only point of my audio network that sees any non-audio traffic.
Another important aspect is isolation. Typicall we’ve no idea what kind of electrical noise is coming into our homes with our broadband supply, so some form of galvanic isolation between the modem and the rest of your hi-fi is a really good idea. It may be wi-fi or fibre-optic; both work extremely well. Wi-fi has a bad reputation, but there’s absolutely nothing wrong with the technology, rather its the implementation where most problems lie. Of course if your house has lots of thick brick walls or steel beams and girders in the structure, then wi-fi may not be ideal, but for all other cases it could be a really good solution. While isolation is a good thing, conversion of volts to radio waves or light pulses and back are noisy processes so ideally should be carried out sufficiently far from your hi-fi, unless your hi-fi has been optimised for a particular format, for example switches are often set up to accept fibre optic input.
There are several really positive aspects of an optimised network. First the better the physical layer gets, the better the sound quality becomes. I spent 5 years optimizing everything and never reached a point of diminishing returns. The upgrades did gradually become more expensive, but the improvements in sound quality also became greater. The second is the scale and nature of the actual improvements. Every upgrade I did was IMO very much worth the money I spent. Often upgrades are cosmetic, increasing some aspects of the replay but not really moving the needle of listener enjoyment and involvement. When you improve the network’s physical layer, everything gets better, so the hi-fi aspects change, but way more important than that, the level and intensity of joy and emotion goes up by large increments and believe me, that’s what we’re all after. Sound that makes us feel so damned good its utterly addictive. Finally, this whole thing is entirely scalable. Do a little, do a lot, dabble or like me go completely overboard, whatever you choose, you’ll get the rewards, however be aware that once you start, you may find it difficult to stop.
 
Last edited:

JustinTime

Wammer Plus
Wammer Plus
Jan 18, 2023
699
1,397
148
HiFi Trade?
  1. No
Firstly, I was very sorry to read in your first post about your hearing loss. It feels like getting all this down on paper, so to speak, is an important part of coming to terms with this. Thanks for taking the time, and very best wishes for your continuing adjustment to your new reality.

When you improve the network’s physical layer,
It’s not clear how you/anyone can do this. Could you explain what you mean by “improve”? My understanding is that the bits either arrive in the right order or in a terrible network (dodgy out of spec cable?) rarely might not. Timing is irrelevant because of the way ethernet works. So all we’re left with is noise, which you mention in connection with power supplies. Is reducing noise what you mean by improving the physical layer?
I’m not arguing that you didn’t hear the improvements you described, just trying to understand what might have caused them. Thanks
 

tuga

. . .
Wammer
Aug 17, 2007
14,341
7,001
173
Oxen's ford, UK
AKA
Ricardo
HiFi Trade?
  1. No
It’s not clear how you/anyone can do this. Could you explain what you mean by “improve”? My understanding is that the bits either arrive in the right order or in a terrible network (dodgy out of spec cable?) rarely might not. Timing is irrelevant
Timing is indeed irrelevant in networking but there’s noise travelling along the copper network with the data which if not dealt with will affect the DAC’s clock and D/A stage.
This noise is either generated by the processing inside the streamer and the switch and the router and can be minimised by choosing clean PS, minimising processing and filtering.
Standard unshielded cables cables should be used to take advantage of Ethernet’s galvanic isolation.
 

tuga

. . .
Wammer
Aug 17, 2007
14,341
7,001
173
Oxen's ford, UK
AKA
Ricardo
HiFi Trade?
  1. No
@Nestor Turton you've probably read that on the design of hi-fi electronic equiment particular attention is given to grouding/noise.
Since the noise-level requirements are far more laxed in computer processing (DSP) and networking once we mix hi-fi and IT one is potentially opening a door to noise which as mentioned will affect the clock and D/A stage. Remember that if you are using Ethernet your DAC is physicaly linked to you computer/streamer and your router. This is not noise that will show itself as raised noise-floor in measurements made at the DAC's output but will result in higher jitter and also intermodulation above the audio band. That is why it is difficult to trace the effects of network noise in typical measurements which are generally bad-limited to the audible frequencies.
USB interfaces in DACs have evolved in the past decade to address the issue as well as possible but with many DACs a USB filter (such as the Intona) is still recommended.
For those of us using computers as streamers it makes sense to reduce the processing to the essential required and to decouple the computer from the DAC either by using a USB filter or a low-processing network bridge.
 

Blackmorec

Newbie
Wammer
Feb 6, 2019
90
65
23
Hi JT,
Thanks for your response. Up until 5 years ago I was a dyed in the wool vinyl lover with a history of high-end turntables. I’d bought numerous CDPs and the only one I could enjoy living with was the Balanced Audio VKD5SE with its vacuum tube analog output stage. Then I bought an Innuos Zenith MkII SE and suddenly digital started sounding much better. I read about Innuos’s engineering and talked to them and a few IT trained audiophiles whose ears I trust. Innuos were using 6 key principles to get better sound: Power supply quality, EMI suppression/elimination, improved clocking, vibration management, network traffic management and improved cabling (which in the next iteration became elimination of cabling).
My IT friends, who had Zeniths were recommending the use of a network switch from a German company AQVox, so I purchase their SE model and put it into my system, which took ethernet feed directly from a Virgin Superhub 3. The switch made a nice difference, not huge. I then got to wondering what would happen if I applied the same principles to the network that Innuos use on their server so I purchased a Sean Jacobs DC2 to replace the AQVox wall wart SMPS. This was my first jaw dropping moment. It sounded so damned good that I sent it back to Sean in exchange for a DC3 with Mundorf caps. Another jaw dropping moment. And so it went, through my whole system. In 50+ years of hi-fi, including some really top class components I have never come across an area that continually gives such profound improvements as does the network.
So what concrete steps can you take:
1. Add a decent router to replace the ISP provided combination modem/ router. Most ISPs force you to use their device for the incoming stream, so all you do is switch that device to ‘Modem only mode’ and use a short Ethernet cable between the WAN output of the Modem and the WAN input of the new router. The router you buy should be based on the Broadcom chip set. Many of the ISP routers are based on the Intel Puma chipset which suffers some serious flaws like firing off diagnostic routines when workload becomes excessive.
3. Make sure you have sufficient performance on your network so that everything runs without bottlenecks. For example, if you see buffering quite often when streaming video, you need to fix that if possible. A network needs a lot more than just enough bandwidth. The nice thing about a fast network is that the server‘s user interface feels like its hardwired with almost no latency.
2. Try to separate your audio stream from the rest of your household network traffic. I did this by purchasing a 3 band router and dedicating 1 5GHz band solely to audio. In this way my audio related network only has to deal with audio traffic.
3. Replace cheap and very noisy wall wart switched mode power supplies with decent, well regarded linear power supplies
4. Plan to have some form of galvanic isolation between your incoming broadband and your hi-fi system. This may be wi-fi or it may be fibre optic…either way, it should be well done and optimised for best sound quality.
5. Minimise vibration for all network devices and power supplies. I use Atacama Evoque 35/40 bases under modem and router and they work well. I use Finite Elemente later in the stream and all power supplies sit on Isoacoustic‘s Gaia footers
6. Use a decent switch to filter and retime your data stream. You say that timing doesn’t matter but that’s not my experience. Both Innuos and Pink Faun for example use improved clocking as an important part of their strategy. In my system every part of my network seems to work better (produce higher quality sound) when clock accuracies are improved. What’s important is that the clocking becomes more accurate as the stream progresses through the network. No point having accurate clocks early in the network be cancelled out by poorer clocks downstream. For me, I think the DAC+ rest produces much better quality music when all the clocks in the network are vibration isolated, have really good, super low noise power supplies and the ones coming later in the chain are temperature controlled.
7. Use decent quality, very well screened DC cables between power supplies and their devices. All my DC cables were Mundorf Silver/Gold with John Swenson developed JSSG360 screening, built by my IT expert friend ‘Nenon’. Their contribution was extraordinary.
8. Use decent quality ethernet and USB cables. I used Synergistic Research with all screens star wired to a common earth point, so all screens shared a common ground potential.
9. Use a good quality server, preferably one that minimises network comms during replay and offers a really great software and user interface.
10. Arrange your network as I’ve suggested such that power supplies, vibration control, cables etc. increase in quality along the network.
11. Pay attention to power cords on all supplies, as they make their own contribution to the final sound.
12. Where possible, use a dedicated, high quality mains supply for items following the galvanic isolation. I use a solid copper Henson block to split off phase and neutral after the meter to a Hagar Consumer Unit with Doepke RCCD switch and 2 Gigawatt circuit breakers. In-wall cable is Acrolink 7N P4030II to a total of 4 Furutech Gold duplex sockets. Earth is taken directly to the incoming supplier earth. All fully code compliant. If you’re forced for whatever reason to use house mains, then use a single socket and a decent mains distribution block, which is what I used for the modem and router supplies.
One of my future moves was going to be taking a dedicated mains supply to the modem/router power supply, but not any longer obviously.

In summary, you are removing noise, preventing noise, increasing timing accuracy, minimizing vibration, minimizing cable losses and influence and improving DC power.

I would assume that a well set up audio network with minimal other traffic should be ‘bit perfect’ unless there are some capacity issues or set up faults. Any weak links that have the potential to cause losses need to be addressed. That’s a basic need to start with. From there on, non of the above enhancements cause any alteration to the pattern of ‘1s’ and ‘0s’ in the bit stream. These enhancements refine the basic structure of the already bit perfect stream, because the more perfect the structure, the better the sound coming out of the DAC. It’s why adding LPSs helps. The better the power supplies, the better the soundquality. Sound quality….how the bits are interpreted….is very much a function of how high quality the bitstream….how accurate the timing, how much noise with the stream, how the voltage wave looks inside each bit of time (ie how fast and precise the polarity switches). The more perfectly the bit stream follows the specification for timing, voltage polarity switching, lowest possible noise and the lower the vibration along the entire network, the better the sound quality.

Every upgrade of any power supply can be heard and sometimes the impact is jaw dropping (that phrase again). So here’s a question, if the bit stream was already bit perfect before the supply upgrade, the power supply can have no effect and indeed should have no effect there. So the only place that influence can happen is in the bit structure. The power supply builds the bit and imparts its identity with it.
So there’s a lot of information in the bit stream that‘s not related to the actual bit pattern but to the bit structure, the fabric. That’s the difference between IT functions and Audio. In IT the bits are converted into something with no latitude for interpretation. It either is, or it isn’t. In audio the bits drive the DAC and the DAC reacts to the structure of the bits which influences the DAC’s interpretation. It’s again why you hear differences. Just to mention, I used the same power supply design, topology and implementation throughout my entire system, which keeps things simple from a noise spectrum perspective.

Photo one shows the SMPS I removed from the RE650 TPLink shown in photo 3. The photo is about twice its actual size. The RE650 unit has its 3pin mains plug and internal power supply removed, a 2.5mm barrel connector fitted, wired straight into the circuit board with Mundorf Silver/Gold cable. The unit is suspended off the wall using 4 slack o-rings. Power comes from a MiniARC6-DC4 supply via a Mundorf JSSG360 cable. Output is via an SR Atmosphere X Ref ethernet cable
Photo 2 shows the TPLink Archer AC5400 tri-band router, also powered by a Mini ARC6-DC4 Mundorf cable and sitting on an Atacama base.

In the past I have spent money on a wide variety of great components but for the past 5 years, my network got all my discretionary £££ only because I was getting such huge upgrades in sound quality. My vinyl set-up went somewhere along the way because it frankly became irrelevant (I have a really great little collection of mint or near mint classical performances of the great works for sale if anyone is interested 😊) If I thought I could have got more from upgrading an amp, speakers, cables or anything else, that’s what I would have done. As it was, upgrading the network gave me more and more of exactly what I craved. To be utterly, absolutely, thoroughly involved with and enchanted by the music. With each upgrade I moved further down that road.

Finally, on A personal note, I write so much because at the moment this is what I do. Write about hi-fi and share what I found by obsessively refining my network.
I found audio Nirvana, now lost. To help others get to the same place would be a consolation. IMG_6310.jpeg IMG_0741.jpeg IMG_0535.jpeg
 

tuga

. . .
Wammer
Aug 17, 2007
14,341
7,001
173
Oxen's ford, UK
AKA
Ricardo
HiFi Trade?
  1. No
Photo one shows the SMPS I removed from the RE650 TPLink shown in photo 3.
Wow, pimp me bloomin' RE650!!! o_O

That's quite a trick you pulled there. I hadn't even imagined such a thing but I suppose it might be worth considering if I were to plug my network bridge directly without going through a switch.

I must confess that I am tremendously sceptical of vibration control being able to produce any benefit in networking devices. Same regarding "improved" clocking of data transfer.
 
  • Like
Reactions: JustinTime

JustinTime

Wammer Plus
Wammer Plus
Jan 18, 2023
699
1,397
148
HiFi Trade?
  1. No
Thanks for sharing so much further info. I have to disagree with some (but by no means all) of what you say, but am grateful to you for taking the time to spell it out.

1) Not convinced. You might get faster response from devices on the network but this doesn't/can't translate to sound quality by any mechanism I'm aware of. Buffering for example doesn't affect sound quality.
2,3,4) Agreed
5) I need to experiment but confess to a mental block re how vibration can impact solid state componets. If it does, I'd imagine it will have more impact once we're talking after the streamer rather than with ethernet
6) Any galvanic isolation thing (switch, optical, filter, whatever) should be only a short cable from the streamer.
Timing is a complete red herring as ethernet is asynchronous: clock accuracy in and after the streamer is a completely different thing (can impact sound quality) from clock accuracy on ethernet (can't).
7) Agreed
8) Be careful with confusing "good quality" with expensive and/or with shielded as shielded ethernet cables can negate the galvanic isolation you mentioned earlier.
9,10,11) Reserving judgement
12) Sceptical

None of the above should detract from your contribution to the forum, it's just some other views on stuff you've mentioned.

Justin
 
  • Like
Reactions: tuga

Blackmorec

Newbie
Wammer
Feb 6, 2019
90
65
23
Wow, pimp me bloomin' RE650!!! o_O

That's quite a trick you pulled there. I hadn't even imagined such a thing but I suppose it might be worth considering if I were to plug my network bridge directly without going through a switch.

I must confess that I am tremendously sceptical of vibration control being able to produce any benefit in networking devices. Same regarding "improved" clocking of data transfer.
Hi tuga,
here’s the thing. In a standard RE650 one gets vibration from the wall, vibration from the AC/DC rectification, vibration from the transformer and noise from the cheap-as-chips smps which likely puts a lot of hf noise back into the mains. My mods dealt with all those issues. I ran the unit with 2.4GHz and polling switched off and the only client was the server, so the only traffic was audio related. The standard RE650 sounded really good, until it was compared with a modified unit, which made quite clear the sonic effects of the above shortcomings.
I can respect that you are sceptical about vibration control but consider that manufacturers already recognise the benefits of fitting vibration control footers to devices like servers and power supplies. It wasn’t long before the lunatic fringe took the same footers, applied them to switches and found the increase in sound quality quite a compelling proposition for the money. In my case, the lesson came from my router, which originally sat on the carpet, looking untidy. I happened to have an Atacama half width rack from an SoTM experiment so I used the base under the router. By the next listening session I’d completely forgotten about the table, as it was only there for cosmetic reasons. I clearly heard an improvement in detail recovery, so basically the ability of my ears to resolve more discreet information from the sound. It took me quite some time to realise that it was the table that brought the improvement, because it was the only thing to have changed. Was it a huge difference? No. But it was enough to grab my attention that something had improved. Sometime midway through the development of my system I aquired a second Finite Elemente rack for the post bridge switches and their power supplies. The effect of the better rack was clearly heard, as was the addition of Cerabase footers to the rack.
The point is, as I kept adding all these individual gains, one after another, the system just got better and better, often in areas I wasn’t expecting. For example in the amount of spatial detail resolved, or the amount of timbral information at very low amplitudes, or the sheer physicality first of the bass then latterly with all of the music. The more improvements, the more the music could communicate the feelings and emotions the composer intended to arouse And the more convincing and involving the system became.
 
Last edited:
  • Like
Reactions: tuga

Blackmorec

Newbie
Wammer
Feb 6, 2019
90
65
23
Thanks for sharing so much further info. I have to disagree with some (but by no means all) of what you say, but am grateful to you for taking the time to spell it out.

1) Not convinced. You might get faster response from devices on the network but this doesn't/can't translate to sound quality by any mechanism I'm aware of. Buffering for example doesn't affect sound quality.
2,3,4) Agreed
5) I need to experiment but confess to a mental block re how vibration can impact solid state componets. If it does, I'd imagine it will have more impact once we're talking after the streamer rather than with ethernet
6) Any galvanic isolation thing (switch, optical, filter, whatever) should be only a short cable from the streamer.
Timing is a complete red herring as ethernet is asynchronous: clock accuracy in and after the streamer is a completely different thing (can impact sound quality) from clock accuracy on ethernet (can't).
7) Agreed
8) Be careful with confusing "good quality" with expensive and/or with shielded as shielded ethernet cables can negate the galvanic isolation you mentioned earlier.
9,10,11) Reserving judgement
12) Sceptical

None of the above should detract from your contribution to the forum, it's just some other views on stuff you've mentioned.

Justin
Hi Justin,
Regarding your point 1 about buffering. As an example, let’s take an Innuos Server. Before it plays a track, the track is loaded into then played back from RAM. As the RAM is feeding the server, all network activity, polling etc can be shut down to minimize noise within the server. Then there’s speed. Firstly, the whole UI operates more instantaneously, second, high speed cuts down on latency. Third, speed cuts down on the amount of time required to load tracks. Fourth, speed ensures fewer interruptions and less opportunity for data losses. Fifth, speed insures no bottleneck and delays, no long buffering times.
Regarding clock accuracy, try fitting a Pink Faun Ultra OCXO clock to a Melco S100 switch and tell me it doesn't make a difference. That clock is accurate to 3ppb and gets its own ARC6-DC4 5V rail. The difference is heard in terms of purity, transparency and a feeling of effortlessness, rhythmic flow and simple musicality.
If you look at an Innuos PhoenixNET, a major element of the fidelity increases it generates come from retiming the incoming and outgoing ethernet stream prior to the server.
Point 5 I discussed in another recent post.
Point 6 Galvanic isolation. You say it should be only a short cable away from the streamer/server. I would say that very much depends on how good and how noisy the conversion. Volts to light or radio waves and back are inherently noisy processes and very dependent on hardware and power supply. For example, I would not like to feed my server with the output from my Wi-Fi-ethernet bridge, even my modified one, without first putting the resulting stream through a clean-up and retiming device with a high accuracy clock. The bridge output is good, but it can be made a lot better. Galvanic isolation has its own sources of noise, so its not a universal panacea. Why it works is that it isolates more noise than it creates, but it still creates noise, which when removed, enhances sound quality.
Cables can be screened without creating continuity between components, for example, any screened cable with plastic RJ45 plugs.
Finally not everything that is expensive is good but everything that is good is unfortunately usually expensive….such is the nature of the hi-fi market.
Again thanks for the discussion 😊
 
Last edited:

tuga

. . .
Wammer
Aug 17, 2007
14,341
7,001
173
Oxen's ford, UK
AKA
Ricardo
HiFi Trade?
  1. No
here’s the thing. In a standard RE650 one gets vibration from the wall, vibration from the AC/DC rectification, vibration from the transformer and noise from the cheap-as-chips smps which likely puts a lot of hf noise back into the mains. My mods dealt with all those issues. I ran the unit with 2.4GHz and polling switched off and the only client was the server, so the only traffic was audio related. The standard RE650 sounded really good, until it was compared with a modified unit, which made quite clear the sonic effects of the above shortcomings.
I don't know much about IT but I think that my RE650 is only connecting the router to the switch (I stream locally); if that is the case then I have no audio data passing through it because it is read from a USB SSD, DSP'ed in a (headless) streaming desktop computer and sent through the switch to the network bridge which links to the DAC with USB.
I need an RE650 to wirelessly connect the streaming computer to the laptop I use to operate it remotely (I used to have a dedicated wired-only network but I got fed up of having to plug my laptop evry time I wanted to listen to music and also the long cable lying around on the floor).
 

Blackmorec

Newbie
Wammer
Feb 6, 2019
90
65
23
I don't know much about IT but I think that my RE650 is only connecting the router to the switch (I stream locally); if that is the case then I have no audio data passing through it because it is read from a USB SSD, DSP'ed in a (headless) streaming desktop computer and sent through the switch to the network bridge which links to the DAC with USB.
I need an RE650 to wirelessly connect the streaming computer to the laptop I use to operate it remotely (I used to have a dedicated wired-only network but I got fed up of having to plug my laptop every time I wanted to listen to music and also the long cable lying around on the floor).
Hi Tuga,
If you could list the actual hardware involved including how it is powered and connected that would be helpful to picture your set-up. I use an IPad to control my system. It is linked to my router via the router’s 2.4GHz wi-fi. Commands from the iPad go first to the router, which then send it out via the hi-fi’s dedicated 5GHz band. So, the hi-fi sees nothing of the household network and the hi-fi related network only processes audio related data, including commands and responses from and to the iPAD.
Is there a reason you don‘t use a service like Qobuz? For me that was a dream come true. Fully integrated into my server‘s library management, millions of albums and extremely high fidelity for £15.00 a month. In terms of fidelity, I eventually couldn’t tell a locally from a remotely streamed file.
 

JustinTime

Wammer Plus
Wammer Plus
Jan 18, 2023
699
1,397
148
HiFi Trade?
  1. No
Regarding clock accuracy, try fitting a Pink Faun Ultra OCXO clock to a Melco S100 switch and tell me it doesn't make a difference. That clock is accurate to 3ppb and gets its own ARC6-DC4 5V rail. The difference is heard in terms of purity, transparency and a feeling of effortlessness, rhythmic flow and simple musicality.
Difference compared with what clock powered by what psu/rail?
 

tuga

. . .
Wammer
Aug 17, 2007
14,341
7,001
173
Oxen's ford, UK
AKA
Ricardo
HiFi Trade?
  1. No
Hi Tuga,
If you could list the actual hardware involved including how it is powered and connected that would be helpful to picture your set-up. I use an IPad to control my system. It is linked to my router via the router’s 2.4GHz wi-fi. Commands from the iPad go first to the router, which then send it out via the hi-fi’s dedicated 5GHz band. So, the hi-fi sees nothing of the household network and the hi-fi related network only processes audio related data, including commands and responses from and to the iPAD.
Is there a reason you don‘t use a service like Qobuz? For me that was a dream come true. Fully integrated into my server‘s library management, millions of albums and extremely high fidelity for £15.00 a month. In terms of fidelity, I eventually couldn’t tell a locally from a remotely streamed file.

Here's is the network layout:

FetK31Z.png
 

StingRay

Legend Wammer
Wammer
Apr 27, 2016
13,040
8,521
163
Suffolk coast, UK
AKA
Ray
HiFi Trade?
  1. No
Hi Tuga,
If you could list the actual hardware involved including how it is powered and connected that would be helpful to picture your set-up. I use an IPad to control my system. It is linked to my router via the router’s 2.4GHz wi-fi. Commands from the iPad go first to the router, which then send it out via the hi-fi’s dedicated 5GHz band. So, the hi-fi sees nothing of the household network and the hi-fi related network only processes audio related data, including commands and responses from and to the iPAD.
Is there a reason you don‘t use a service like Qobuz? For me that was a dream come true. Fully integrated into my server‘s library management, millions of albums and extremely high fidelity for £15.00 a month. In terms of fidelity, I eventually couldn’t tell a locally from a remotely streamed file.
Qobuz is less than £15 pm, only £10.83 with annual sub. But I agree streaming is what I use all the time now, although I use Deezer which is much cheaper.
 

tuga

. . .
Wammer
Aug 17, 2007
14,341
7,001
173
Oxen's ford, UK
AKA
Ricardo
HiFi Trade?
  1. No
Is there a reason you don‘t use a service like Qobuz? For me that was a dream come true.

HQPlayer has Qobuz integration so I could use it.
We have a Spotify family account which the kids prefer because they can share playlists with friends, and which I use for casual listening as well as to explore new music (I will buy the CD or download). But if it weren't for the kids I wouldn't have a streaming service.
 

tuga

. . .
Wammer
Aug 17, 2007
14,341
7,001
173
Oxen's ford, UK
AKA
Ricardo
HiFi Trade?
  1. No
Qobuz is less than £15 pm, only £10.83 with annual sub. But I agree streaming is what I use all the time now, although I use Deezer which is much cheaper.
Qobuz's family subscription plan costs £18/month, only £1 more than Spotify which is lower resolution (only 320kbit/s).
 

StingRay

Legend Wammer
Wammer
Apr 27, 2016
13,040
8,521
163
Suffolk coast, UK
AKA
Ray
HiFi Trade?
  1. No
Qobuz's family subscription plan costs £18/month, only £1 more than Spotify which is lower resolution (only 320kbit/s).
Not keen on Spotify soundwise and you can get Deezer for £5.99 pm which is much cheaper and cd quality. I don't know about families.
 
  • Like
Reactions: tuga

Latest posts

Forum statistics

Threads
113,444
Messages
2,451,263
Members
70,783
Latest member
reg66

Latest Articles