Home network - basic hygiene to make your music server sound its best

Hi all. I have just joined the forum and it’s great to see the strong community spirit, generous knowledge sharing and all round civility & respect you have for each other.

I’m exploring how to improve my home network configuration and keen to understand how K50 & Oladra owners have set theirs up. My current baseline configuration is:

• Fibre to the home ONT>ISP supplied/branded modem>Tp-link Deco X20 Mesh Wi-Fi 6 unit (ethernet out)>Netgear GS108 network switch>K50.

• All my ethernet cable are mass market ones: cat 6a shielded.

As initial “good hygiene” steps, I have:

a) Separated the mains line feeds for a) home network components and b) 2 channel set audio up.

b) added a PS audio dectet AC power conditioner for the networking devices

c) disabled the wifi functionality on the ISP modem limiting wifi to just the mesh wifi set up;

d) replaced all OEM switch mode power supplies (SMPS) with 12V ifi ipower v2 & X models (lots of debate abound about SMPS vs linear power supply (LPS) and quality/expense of each relative to SQ gains).

While cumulatively the above have yielded solid but modest improvements, it’s good to have some basic hygiene sorted as a platform/good baseline against which to test the outcomes of further exploration.

I have already done a fair few other experiments (with power supplies, fibre media converters, Net network switches……etc) and plan a few more. So if this new topic helps facilitate further knowledge sharing of value and if this thread takes off, I will share results.

Distilling my lines of enquiry to the fundamental questions I’m trying to answer:

  1. To what extent are music servers impervious to the quality of network inputs/connections (ie diminishing/obviating the need for heroic/costly upstream network optimisations)?

  2. Can an ethernet/fibre combination ever better an all ethernet set up using (in both scenarios) high quality audio grade network components?

  3. High quality audio grade power supplies: which is better suited to networking devices/applications: SMPS or LPS?

Apologies for the initial long winded post and thanks for taking the time to read it.

1 Like

Here are some relevant threads on this forum. Plenty more on this somewhat controversial subject onllne elsewhere.

In addition, in direct response your your queries here about basic hygiene:

1. To what extent are music servers impervious to the quality of network inputs/connections?

It seems most are not impervious, some are better at rejecting or filtering out the unwanted stuff that comes in with wired ethernet signals.

2. Can an ethernet/fibre combination ever better an all ethernet set up using (in both scenarios) high quality audio grade network components?

I think fibre is a critical and very cost effective way to create an absolute noise barrier to isolate your hifi from both the rest of your network and also the outside world and things you can control/change (such as the device that terminates your fibre to the premises). FIbre, and SFPS in particular, are a rabbit hole in themselves and it seems it is easier and less audible putting fibre as far upstream from your hifi as you can.

3. High quality audio grade power supplies: which is better suited to networking devices/applications: SMPS or LPS?

Whichever generates the least noise.

In relation to some of the things you have already:

All my ethernet cable are mass market ones: cat 6a shielded.

Shields are problematic because they are typically connected at both ends to ground and so can pass on noise. As a general guide I suggest:

  • long runs should be fibre
  • cables closest to the hifi endpoint are best to be ‘audiophile’ variety
  • other cables, such as upstream of fibre, Cat 6 or 6a UTP.

a) Separated the mains line feeds for a) home network components and b) 2 channel set audio up.

Good, ideally do not power network devices that are enroute to your hifi with the same power supply.

c) disabled the wifi functionality on the ISP modem limiting wifi to just the mesh wifi set up;

The ISP modem will be a weak point. Get a better router, preferbly enterprise grade. I use Mikrotik, and that is another story. Connect the Mesh router vais LAN rather than WAN so it will still mesh.

d) replaced all OEM switch mode power supplies (SMPS) with 12V ifi ipower v2 & X models.

Good start, others are lower noise LPSs and they will make more dffference closer to the hifi endpoint.

I have already done a fair few other experiments (with power supplies, fibre media converters, Net network switches……etc) and plan a few more.

This is essential, every combination of devices and cables etc is unique, and so are your circumstances and preferences.

1 Like

I did find a benefit from adding a span of fiber. What I have learned from that though is that there is no such thing as an absolute noise barrier. Harm being done upstream of the span can still be heard, sadly.

What fiber can do is offer some isolation from leakage currents. Buy it needs to be done right as a sub-optimal fiber setup can actually introduce noise. I agree with that part about the rabbit hole. It takes some effort and expense to do it right.

I think one potentially elegant way of gaining the benefits of fiber without the hassle is to go with something like what follows. I would want to compare it though to the Muon Pro filter.

I too benefit from a run of fibre to replace many metres of Cat 5e UTP. And yet I still notice changes upstream of that fibre. Getting a bit beyond basic hygiene here, but it seems to me interference can get into every ethernet device and disturb its function in a way we can hear it in hifi. The noise is not being passed on via fibre, but for instance it could be poor phase, jitter, or something.

When doing fibre it is best to use high quality hardware, because a crap FMC on the downstream end of fibre will likely sound less good. For example, my fibre has EtherRegen on one end and a Mikrotik 10G/SFP+ switch/router on the other end. All wired cables are as short as practical.

@dbastin2605 Awesome steer & information. Thank you. Glad I shared my questions and received your answers.

I can confirm your steer on crap FMCs. I found while (quite noticeably) soundstage widened with slightly more depth, imaging improved, & background dead silent…HOWEVER timing/coherence was very off, tonality less natural, sound thinner/more edgy…and so I never relaxed into the music and ultimately it was fatiguing even over shorter listening session.

I’m keen to find a way to make fibre work if only for the electrical isolation advantages/safety (ie without compromising SQ or even improving it). Will report back when I find something I am happy with.

This is similar and literally just released by Telegaertner Japan (who make the M12 switches).

1 Like

While this is far from the basics, the discussion here is really all about that … Optical Network Configurations - Page 124 - Networking, Networked Audio, and Streaming - Audiophile Style

They get down to what SFP makes and models sound best and even some discussion about compatibility with different devices. The essence is:

  • use really good hardware like SFP+/10G switches, Sonore Optical Module, EtherRegen with really good power supplies
  • use SFP+ modules if you can afford it (Finisar are popular), even if you run them at only 1G speed
  • choose SFP modules that are compatible with your devices
  • use quality single mode fibre
  • be patient

My two cents: the better the server (and system), the more that everything matters.

My reference for optimizing my network had been a comparison to the sound when disconnecting the network. Several years ago that lead me to an 8m span of fiber, with a Sonore opticalModule upstream and an EtherRegen downstream. When comparing this to a span of copper or to no network connected, what I heard with the fiber span was closer to the sound with the network disconnected - but only with Finisar SFPs.

It used to be that I would occasionally power down the opticalModule to kill the network while listening, as this would improve sound quality. I stopped doing this as the magnitude of the improvement shrunk. Using a REF10 SE120 to provide a reference clock to the EtherRegen played a part in that. It’s crazy but improvements I make to the network even improve the sound of music played off the K50’s hard drive. I recently upgraded the DC cable powering my EtherRegen and the improvement wasn’t difficult to hear. Everything matters.

A great answer was provided by Emile Bok of Taiko Audio: “So anyway, there are definitely merits in exploring Switch Mode Power Supplies for audiophile applications, the benefits over Linear Power Supplies are obvious, if implemented correctly, it can be much lower noise then a Linear Power Supply for higher current applications, have a much smaller footprint and no massive heatsinking requirements, BUT they should be designed to not have any switching “noise” anywhere near the audio frequency range, or even better yet no measurable residual whatsoever.”

If you can find such a SMPS then it would likely be a great choice for powering a network device. But can you actually find one? And how expensive would it be? I don’t have those answers. My point is simply that implementation is what matters most in this hobby.

1 Like

Hi MusicLover

Why do you have network switch between your Deco mesh unit and K50?

I have a Deco mesh modem/router, wifi bridge to a Deco X20 mesh unit, ethernet out direct into my CX.

Ethernet connection to CX is with a Quiescent cable and an Entreq Macro is connected to the second ethernet socket on the Deco X20 which is also powered with a iFi low noise PS.

I’m very satisfied with this set-up no problems running Roon or streaming 24bit Qobuz and SQ is great. The Quiescent cables (I also have one between the CX and EX) made a very noticeable improvement. The Entreq Macro not so much which makes me think the Quiescent cables are delivering most of the SQ gains to be had. They’re not cheap but it’s not so hard to spend more!


Hi Andy. Good question. Let me preface my answer by briefly explaining the back story to my experiments.

I have a modern tube pre-amplifier that began putting out high pitched EMI/RFI interference type noise a few months ago (ie a real world problem in my chain). By process of elimination I narrowed it down to some changes I had made to my network & power configurations. (Note: data transmission performance either over wifi or ethernet has never been an issue for me soI can’t comment on your CX being connected to the Deco X20 unit).

My experiments have therefore been focused on cabled network components & their power supplies…and I am trying to figure how best to minimise the extent to which individual components either filter &/or generate &/or pass on this type of noise. (Note: I plan to move to cable experimentation after that - not sure this is the right order to approach things but it’s the one I have chosen).

I can’t explain scientifically (or with measurements) why having a network switch between the router and server is optimal, but my experimentation has led me to believe that:

  1. Power supplies really really really matter (I can’t overemphasise this).
  2. As does the quality of network components being used.
  3. Having the highest quality network component and power supply closest to/feeding the music server has consistently worked best.
  4. It has been easier for me to find/make the network switch the highest quality component in the chain as Routers & Wifi access points tend to be generic/mass market components. (Note: conceptually I also like the idea of components doing fewer things/dedicated single tasks on the assumption that there may be more risk of EMI/RFI noise generation with more activity at each individual point of the chain… I’m not stating this as fact).

More generally, I have found that while the network/digital feed into music servers can have a profound impact, the differences are:

· less obvious at the lower quality end of music server spectrum;
· then become increasingly noticeable through the mid-tier level; before
· beginning to diminish again through the very high end tier.
I.e. the impact for me has not been linear but more of a bell shaped curve.

This being the case, there comes a time (post basic network hygiene) when one has to confront of the question of: do I get bigger bang for buck by spending $s on a music server upgrade vs ongoing network improvements?

I think that’s an impossible question to answer definitively. So personally I try and force myself (often unsuccessfully) to be guided by some basic principles:

  1. Is the level of ongoing network spend proportional to the cost of the music server? Having a $100 router with cheap OEM switch mode power supply feeding a K50 is questionable. Likewise having a Telegaertnet M12 switch plus Opt Bridge optical Isolator (>US$8k in total) feeding a single (without EX) CX would also be questionable also. Where one fits on the spectrum between those two bookends is very personal.
  2. Simplicity & elegance. Eg while cascading multiple switches has improved SQ for many, I find the resulting clutter unappealing, especially since the gains for me have been quite modest. Of course whilst this may limit incremental spend accumulating into big $s, elegant & effective single box solutions can also be quite expensive…. though easier to confront upfront vs the gradual ‘wear & tear” on one’s wallet (think about frog being boiled slowly in the pot :slight_smile:).
  3. Does a change “just sound RIGHT”. I find when something isn’t quite right soundwise, my analytical brain is switched on & turns into overdrive (I simply can’t help myself). When the sound is “just right” that part of my brain switches off and I simply just get immersed in & enjoy the music. As examples, my tube pre-amp… & turntable… felt “just right” when I inserted them into the chain. Ditto the Oladra when I heard it (I don’t own one).

Everyone’s personal “it just sounds right” is different. Because I listen to a lot of classical music (including live performances), my own preferences/sensibilities are highly tuned to:

· Musical flow & cohesiveness (by that I mean timing/rhythm…& the sum of the parts is greater than the individual elements… ie instruments/voices etc…all in an emotionally engaging context).
· Timbral accuracy (eg easy to discern the difference between say a violin & viola in a complex, say orchestral piece of music);
· Tonal balance (eg on a grand piano the spread of tone across & the relativity between its 88 keys…with good gear I can especially hear realistic weight/darkness in the low end)
· Transient responses that can be as dynamic as they need to be (eg again esp with large scale orchestral pieces).

Had some spare time this Sunday morning (and it’s raining outside) so I thought answer your question very fully :slight_smile:. Cheers, Alex


Just remembered the one other important aspect my ears are tuned to are… the correct placement of the different orchestral sections ( strings, woodwinds, brass, percussion etc…and even within those the sub sections eg violins, cellos… etc). Left to right, front to back with adequate spacing (ie not congealed) within a realistically sized soundstage. Listening out for this and the others noted earlier has really helped me quickly separate out the wheat from the chaff during my experimentations.

@Musiclover you seem to be on a pretty good track. Here is a few quick comments that may help.

I guess it would be best to start with listening to the server that is the next step up in one’s system. Then knowing what that benchmark is and its cost will guide decision making about improving the network.

I’d say a decent network won’t be wasted if one does decide to upgrade, even if some of that is stripped back.

That is one benefit of network inprovements, it is several incremental bits and bobs that are acquired and can be sold seperately… can’t do that when upgrading a server.

Finally, it is quite possible a totally noiseless network is either not possible or would take a gazillion things to achieve. This is not unlike the rest if a hifi system. It’s owner needs to determine when is enough to be satisfied with the result. Because the way things are going, there will be plenty of new products claiming to be the next best solution.

I agree. The next step up from basic hygiene wouid be improving the router and isolating wifi and other devices from the router.

For example, I use a Mikrotik CRS 305 as a router. It is SFP+ / 10G soec and very low jitter even at lower speeds.

From that I run fibre to other devices so it is not receiving or transmitting any electrical interference/noise.


Thanks for the very full reply. On your bullet points….

1 completely agree. A Sonos ZP90 taught me that several years back when first dabbling with streaming.

2 for practical reasons I choose a wifi link to my system. Before going mesh with the Deco’s I used a TP-link travel router as the receiver. Obviously that is pretty much the inverse of your principle although I did use an iFi PS on it and plug an Audioquest jitterbug into it to try and minimise the adverse effects on SQ. I chose to look into cables at that point which is when I came across the Quiescent cables and discovered how much they lowered the noise floor.

3 I wouldn’t disagree but as noted I have to compromise for practical reasons.

4 It would be interesting to compare the Quiescent cables vs something at a broadly similar cost like an EE switch to see how they fare. The nice thing about a cable is it doesn’t need yet another PS.

Also I agree on sounding “right” in a holistic musical sense and that can be achieved with a number of different “flavours” to the presentation. IRL not all orchestras sound exactly the same after all!

Happy listening


thanks, that very reassuring. To be honest I was quite lost at sea for a while. I’ve found typical Hifi retailers (ie no networking/music streaming specialists) either haven’t kept up with this fast moving/developing category &/or just want to sell gear that they stock/brands they represent.

yes have done that and will share my experience/final decisions once I completed this exercise (likely end of Month).

this was also a great suggestion which I followed up on. Some pretty interesting results which has me asking questions about the SSD I’m using…but there is separate thread for that which I’ll contribute to when I’m finished with this exercise. With the benefit of hindsight I should problem have started with Kenny’s suggestion once my preamp issue had been resolved.

This advice is pure gold and I hadn’t thought about approaching the problem this way. Much easier (& cheaper) to isolate busy/noisy network components from the router (and connect streamer to router) than it is “throwing the kitchen sink” at isolating the streamer downstream from the chain of network components as I’m currently trying to do. Pure genius :slight_smile: thank you.

On my buy list straight away (and it’s priced reasonably :slight_smile: ). It’s out of stock everywhere in my country so seems maybe very popular or perhaps there’s a newer version.

Thanks so much “dbastin2605” for your guidance/knoweldge sharing

Rather more succint with this one Andy :slight_smile:

I suspect using a wifi end point can be a valid & very effective way isolating from other cabled noisy network components. If you haven’t already, upgrading the power supply on the Deco X20 wifi receiving unit. You could also experiment with a ~US$ 100 passive LAN isolator - Pink Faun, Jcat & iFi make/sell such devices. Burn them in for at least 3 weeks through continuous use before you draw any conclusion.

I hear you! :slight_smile:

Yep flavouring (& fine tuning) is part of the joy of the process. I have however had one very costly experience where I flavoured with a tube component, which yielded a very (overall) pleasing euphonic sound but the piano forte ended up sounding like a key board! So now I try and make timbral/tonal authenticity & purity the first hurdle to get over/the one thing not to be compromised with changes!

That is absolutely correct. Period instrument orchestras vs modern instrument orchestras. Full sized orchestra vs reduced size. Large stage &/or large concert hall vs small stage &/or small concert hall. Front row seat, middle or last row seat. Luckily when it comes to how the the different sections of an orchestra are organised, there is almost always a consistent logic which can be clearly heard on good gear.

The point you correctly make doesn’t end there…there are very clear timbral differences between say a Stradivarius violin and a Guarneri. Often times concert musicians don’t have expensive hifi set ups at home because they understand and know what to look for and no music reproduction system can capture the nuances at their level of understanding/ appreciation. Long story short maybe best for us not acquire such deep knowledge so we can continue to enjoy our music reproduced at home :slight_smile:

Btw I keep remembering things I look for when doing listening test. What reviewers call “inky black background” is one of them. I never really understood what this phrase meant until I hear an Innuos PhoenixNet and then later an Oladra. Musical details & nuances are allowed to emerge from such a background that one has never heard before. And the space between notes (which I think Beethoven is quoted as saying is an important part of the music) that you hear is an absolute thing of beauty. AMEN!

I didn’t realise Beethoven said this, but I think the space between notes defines their length, which h is a big element of musical timing. It is one thing to properly hear the start a note, another for its attack and decay to be as intended, and to hear the length of notes and changes in the note through its duration accurately completes the equation of the overall timing, energy, emotion etc of each note in the musical passage. Of course multiply that for each instrument in the ensemble.

It seems a key factor is the impulse response of hifi gear. Jitter and ‘noise’ affect impulse. Blackness and ‘calm’ seem to represent the absence of stuff tgat is not part of the music, it is extra ‘pollution’.

I agree blackness is a very important outcome. I use Dual mono Devialet Pro renound for super low distortion and super black noise floor. I connect to it via wifi for total isolation. The wifi access point is dedicated to that connection only, power by battery,connected to my EX dedicated direct out.

The only wired connected to the Devialets are power, speaker and ground. All unused connections a capped.

The Devialet renderer is a purpose designed board rather than repurposed computer main board.

So I fully support the use of wifi to the end point.

Here! here! Butcas you say it might be impossible.

Don’t rush yourself, these things take time to work through such as burn in and settling

You might be better off with no internal SSD. It is another thing that could be detrimental. You couid use externally SSDs galvanically isolated from the server. I know someone who uses Waversa SmartHub 3 for not only ethernet but also to isolate and reclock external drives.

The CRS 305 is a bit if a hassle to set up, but worth it. I can help with that. It is a switch that can be a router, depending on the firmware. I think in hifi terms it delivers value that is multiples of its purchase price. There are other Mikrotik CRSs, but tgexSFP+ / 10G is the important bit. There are also Mikrotik routers, and Ubiquiti Dream Machine which is a gateway so has more functions.

Try to buy from someone who provides support to help set up.

I haven’t gone as far as many others as far as optimizing my network. But the little I’ve done has me wondering if this is really the right place to invest dollars to gain improvements in the areas you mentioned. In my experience, our rooms do the most harm to these things. No single audio upgrade has done more to improve the sound of orchestral music - and I specifically mean every area you mentioned - than placing a stacked pair of 16” ASC Isothermal tube traps in each of my front corners behind the speakers. I can’t think of anything else I could spend $3000 on to get that level of improvement.

Of course doing one doesn’t preclude one from doing the other. I just mention it because the gains were exactly in the areas you mentioned. Plus I can’t imagine our networks even doing a fraction of the harm our rooms do to the music. Just my two cents.

1 Like

My experience too Kenny. I have Vicoustic bass traps. Room treatment is much more important than any of this stuff. The fist install I had done (which also included diffusers) actually deadened the acoustics in my listening space so some of it had to be wound back. So like everything there is a sweet spot. But absolutely bass traps is a must if at all possible.

1 Like