Jump to content

Recommended Posts

Posted (edited)

This thread is for ...

 

people to share how they have put their ethernet systems together, and why.  Let’s help each other understand how to assemble ethernet hardware to improve sound quality.

 

Please be constructive rather than critical.

If you wish to argue that network quality makes no difference to sound quality, please do so in other threads such as some I have started for that purpose.
 

Back to basics ...

 

It is not breaking news that more and more people are using computers of some sort as a source of digital audio data.  These computers range from mass produced PCs, laptops and NUCs, to more DIY oriented Raspberry Pi and 'the Computer Audiophile' systems, to bespoke servers and players purpose built for audio performance.

 

Darko explains this for beginners here ... https://www.google.com/url?sa=t&source=web&rct=j&url=https://www.youtube.com/watch%3Fv%3DtT-qby5q3V0&ved=2ahUKEwj05fD3mLPyAhX38XMBHYaiBa0QwqsBMA56BAgwEAE&usg=AOvVaw0506eb9WGy7QCcwJJb_bt8

 

The anatomy and architecture videos here also explain how computers are used.

https://antipodes.audio/portal/guides/

 

Typicaly there is a server and renderer, often in one device.  The term streamer is interchanged with renderer and both are also referred to as endpoint.  Roon refers to server as Core.

 

Most of these rely on ethernet to transmit the data to the renderer/endpoint, the data source could be computer disk drives, network attached server, internet streaming services, or a combination.  Ideally the renderer design would be immune to noise, however that is not common.

 

Noise is the same kind of noise that degrades power and is addressed by power filters, conditioners and cables.

 

The way that data is transmitted can have quite an impact on the resulting sound quality.  Even generic ethernet hardware can sound quite good, especially with some basic improvements, and like many things in the audiophile world, more technologically advanced hardware usually improves the sound quality.

 

Importantly, if your audio gear is connected to the ethernet network for any reason, such as just control of roon, that connection can degrade the sound quality even if no audio data is coming over than connection.

 

In my experience, the expenditure on improving ethernet yields great value for money and is as worthwhile as power conditioing and cables.In my view it is worth treating the ethernet system as a component level expense, and ethernet cabling warrants the expense of digital cable (coax, USB) or analog interconnects.

 

Ethernet for audio has several parts:

  • Data Transmission – Cables, fibre, wifi
  • Switches, isolators and routers
  • Power supplies, AC cables and DC cables
  • Master Clocks
  • Tweaks

I have created threads on some of these parts, if you haven’t come across them search ‘Ethernet for Audio’.  I have explored ethernet for audio quite a lot, so I will kick off with some lessons I have learned to help others put together their own ethernet system.

 

Assumption

 

There is noise in your audio system that is unintended and it diminishes sound quality.  Overall, I suggest it can be assumed there are 2 types of noise to address:

 

Very high frequency ‘noise’:

  • effects the performance of electronics in the ethernet system
  • effects audio gear to the extent it diminishes sound quality
  • Is created by devices in the ethernet system
  • exists in the environment and infiltrates devices and cabling
  • can be transmitted by cabling from one device to the next,

Phase Noise of the clocks used by ethernet devices is reduced by using higher accuracy clocks (either internal or external clock).

 

So the 3 aims are:

  1. isolate from noise penetration and transmission
  2. minimise noise production
  3. reduce phase noise

 

Power:  power supplies, cables and conditioning are important foundations, perhaps the most important.  Even generic IT ethernet hardware benefits from good power hardware and in my experience can produce very good sound quality.  However improving power hardware can become costly (and messy) quickly so it is worth consdering from the outset how to consolidate the power hardware.  Also, power hardware vary in their sound character/qualities and this applies as much to ethernet as to other DC powered audio devices.  For instance, read this … https://audiobacon.net/2021/08/06/the-worlds-best-audiophile-linear-power-supplies/ 

 

Data transmission:  Next most important is the choice of ethernet data transmission, especially the last link to the endpoint.

 

Wire cables are the most common – typically copper, silver plated copper and sometimes more exotic metals.  Some cables reject noise (eg. sheilding), some even filter, absorb or shunt the noise to minimise the degree the noise gets onto the data signal to be transmitted. An ethernet cable can have an impact on sound quality similar to power or an audio grade ethernet switch.  Equally, a poor cable can diminish the benefits of good power or switches. 

 

Typically the cable connrcted to the streamer has the greatest impact on sound quality, and each cable further upstream from the endpoint has incrementally less impact on sound quality.

 

Fibre is very effective, because It can not carry or collect noise during transmission and creates noise isolation between devices.  The benefit of using fibre to sound quality can be far greater than its relatively low cost, particularly over longer distances.  I suggest implement at least one link of fibre, however be mindful that the final link to the endpoint is most sensitive and Small Form Factor (SFP) modules can have quite an effect on the sound quality, and there are many to choose from.

 

Wifi can also be very effective, however the wifi hardware should be reasonable performance and unobstructed transmission is best.  Wifi to an endpoint can be very good because the endpoint is not receiving noise from other ethernet devices, however the endpoint ought to minimise the noise generated by its wifi transceiver, which is not a given in every endpoint.  In my experince wifi to the endpoint can perform as well as $1000s worth of wired ethernet hardware.

 

Isolation:  All ethernet is required to incorporate isolation transformers adjacent to the RJ45 ports (sometimes integrated into the ports). I gather these are principally for isolation from electrical faults rather than designed to isolate from the noise that impacts sound quality.

 

For sound quality, isolation can be fibre, wifi (as above) or purpose built products. There are some relatively low cost isolators that use compounds or electronic filters, for example SOtM, JCAT and Acoustic Revive.  There are also more costly devices that isolate using galavanic approach (eg. EtherRegen) or fibre (eg. Gigafoilv4), and as these need power their cost to implement is greater. 

 

However, even if you use 1 layer of isolation, improvements upstream of that are still beneficial.  I have had 5 in series and each added an improvement, although each was incrementally less improvement.  Even changing power to each isolation made a difference.  And this applies to fibre and wifi too!

 

Switches:  The adoption of generic switches for audio has been evolving from various modifications to generic switch hardware to bespoke hardware intended specifically for audio.  

 

It is tempting to usea switch like a multi outlet power board.  That would usually be fine, however in audio be mindful that everything connected to a switch can injection noise into the ethernet system and that is distributed by the switch.

 

It has been discovered that daisy chaining 2 or more switches in seriescan also improve sound quality, and that different devices impart a different flavour on the sound.

 

Switches designed for audio have different ways to reduce noise and improve sound quality, but the end result is what you hear.  Some switches designed for audio that have 2 banks of RJ45 ports have galvanic isolation between those banks.  More advanced ethernet systems employ more than one isolation or audio grade switch daisy-chained in series which does typically improve sound quality, although each extra device typically has incrementally less benefit than the previous.

 

Some servers incorporate a switch or bridge to enable direct connection in the series to the rednerer/endpoint.  Antipodes, Innuos and Melco are examples.  These usually have very low noise power supplies, keep system noise to a minimum, and sometimes good clocks, which benefits the switch performance.  I have found that 1 or more isolation between the server and endpoint is most effective.

 

Router quality can improve sound quality.  Ubiquiti EdgeRouter X and EdgeRouter X SFP are good performance, good sound quality, accept a range of DC voltages, not difficult to set up not expensive.  They also incorporate a switch chip, rather than burdening the processor with software based switching.  As such these are commonly recommended.  Again, sound quality can be improved by powering the router via good power hardware.

 

Clocks:  Some ethernet switches incorporate high performance clock already, such as Paul Pang, Innuos and Renolabs.   Some ethernet switches accommodate external 10MHz master clocks enabling use of a higher performance clock with lower phase noise than built into the switch.  Some people claim using a master clock yields greater benefit than the switch itself.  But this is not top priority because a clock needs a good foundation network to bring out the best of the ethernet system 

 

Tweaks can provide some gains to be achieved, for example placing a mass on ethernet switches, careful cable routing, grounding ethernet devices, placing quantum chips or the like inside switches and routers, putting power supplies on good physical isolators.  These individually make small differences, but contribute a cumulative improvement that might otherwise cost a great deal to achieve.

 

A Foundation

If starting fresh, I suggest the following foundation to build upon, starting from where your network connects to the world:

  • nbn box to router - shielded ethernet cable
  • Router - Ubiquiti EdgeRouter X SFP with LPS such as ifi ipower
  • Insert Wifi Access Point (WAP) into router for general wifi access (TP Link and Mikrotik are low cost)
  • fibre from router to a switch (ie. to isolate from the network) (Or wifi to a 2nd WAP, but this adds a device that could generate noise)
  • a switch with SFP port, powered by LPS such as ifi ipower, preferably better grade LPS.  Ubiquiti EdgeRouter X SFP has a switch chip and can be used as a switch only, there are also other options such as Mikrotik and Ubiqiuti switches.
  • Switch to streamer - the best cable in your network
  • Switch to server - 2nd best cable, but ideally equivalent to streamer cable.

 

Next Step

Enhancing from this priorities would be improving:

  1. power hardware (power supplies, power cables)
  2. ethernet cables
  3. isolation
  4. switch quality
  5. tweaks
  6. grounding

As a reminder, hardware closest to the endpoint tends to give the greatest benefit to upgrades/expenditure

It is possible to spend $1000s in this realm, but if you are not satisfied then go ...

 

Deep in the rabbit hole

  • daisy-chain 2 or more switches/isolators
  • external clock (large range of costs here)

 

Somewhere in this journey it is worth reviewing the endpoint.

  • would a stand alone streamer give better sound quality than the integrated streamer?
  • would a DAC with an integrated streamer be better than what you currently use (ie. DAC is better, or integration of streamer is more beneficial than separate streamer)

Keep perspective of expenditure, somewhere in the ethernet journey it may be more beneficial to give some attention to another part of your audio system

 

As an example, here is my current ethernet system.

  • Endpoint = Devialet Pro
  • Isolation = Wifi Access Point (WAP) (TP Link)
  • Isolation = 2 x Uptone Audio EtherRegens (ER)
  • Isolation = single mode fibre link
  • Server = Antipodes EX
  • Router = Ubiquiti Edge Router X SFP
  • Power supply = Wyred4Sound PS 1 (4 outputs)
  • Power cables = Synergistic Research Atmosphere Level 3 Digital
  • nbn FTTP connection box

The simple arrangement is ...

nbn > ER1 Side B - Side A > fibre > router > EX > ER2 Side A - Side B > WAP > --- > endpoint

 

I use the following wired cables which I arrange for the best combination of their sound character.

  • Synergistic Research Atmosphere X Reference (grounded) x 2
  • Shunyata Research Sigma
  • JCAT Signature Gold

All devices (except nbn box) are:

  • powered via power conditioner
  • grounded to Synergistic Research Ground Block
  • have Synergistic Research ECTs inside them
  • on Synergistic Research Tranquillity PODs

The PODs are sitting on the server, power supply and Devialet so benefit those  mainly. 

 

I suspect the nbn box would provide benefit from better power, however I haven't tried that (yet) because I have FTTP and it appears to be a 3 voltage power supply.

Edited by dbastin
Tweaks, mods, enhancements ...
  • Like 14
  • Love 1
  • Thanks 1
Posted (edited)

Reserved for more info

 

 

 

 

 

 

 

 

 

Edited by dbastin
  • Like 2
  • Love 1
  • Thanks 1
Posted

Very comprehensive summary Dale.  Thankyou.
 

I do think grounding is a semi critical element to the overall equation, although probably system dependent.  Most cable shields are ground connected and provide inadvertent conduits for noise conduction between components.  In addition power supplies can inadvertently circumvent signal isolators, depending on how they are connected.

  • Thanks 1
Posted

Clean power and clean ethernet are essential to making the Devialets sound fantastic- they transform from good to genuinely believable.

I found a good Ethernet cable for the long run to my Mac from the router was helpful- not the biggest yield, but they all add up, as you say.

Sounds like you've got a superb, and carefully tweaked, setup. I look forward to hearing from others as to what they have come up with- good thread idea.

  • Thanks 1
Posted (edited)

Yes, I found significant benefits in using Supra CAT8 cables throughout, also linear power supplies on all pieces of network gear,  media PC's and endpoints.

 

The below diagram is a little out of date as I now run a Soundaware A1X streamer > VTL Reference D/A Converter in the "Second System"

 

I'll also add an Eno Ag filter between the TP-Link Deco X60 and Media PC in my main system soon.

 

Likewise I've some "fake" Nordost Odin CAT8 cables on the way to try. They interested me not due to being silver plated copper,  though rather due to each wire strand being individually shielded.  Will be interesting to see what they do. The Supra CAT8 cables were a nice step up from the no name stuff I previously ran.

 

1932485039_SystemTopology.thumb.png.ba4811e3753c45bfe3dd0f6f47d3f610.png

Edited by MattyW
Posted
16 hours ago, Stereophilus said:

I do think grounding is a semi critical element to the overall equation, although probably system dependent.

 

I agree, grounding an audio system can provide a considerable improvement.  And attention to the details of grounding ethernet devices is important.

 

In my case, the router, EtherRegen, power supply and even 2 ethernet cables are grounded to a Synergistic Research passive Ground Block (which sits on a Tranquillity POD), which is connected to an Earth dedicated to my audio gear via Shunyata Noise Isolation Chambers in a Cyclops.  Even using the basic ground cables is quite beneficial, and I have managed to acquire bit by bit a whole set of Synergistic Research HD Ground Cables from the used market, and they are well worth the improvement.  Even upgrading the wall plug and also cleaning and treating the ground wire terminals with Detoxit and contact enhancer provided a quite noticeable improvement.

 

In relation to your points, for those who are not familiar:

  1. noise can be conducted from one ethernet device to another, and to the endpoint, via cable shields, so some people make cables where the shield at one end of the cable is not connected to the connector body or contacts..  There is much chat about this in relation the Ether Regens.
  2. noise can be conducted from one device through the ground of its DC cable and power supply to another device using the same power supply, thereby degrading any isolation that may be between those 2 devices.  Of course, this is prevented by using fibre or wifi, or probably by providing a more attractive path to Earth by grounding each device

Basically, noise shunted to ground in devices should flow via the path of least electrical resistance to Earth (as in Mother Earth, the lowest resistance there is).  The least resistant path is achieved by using highly conductive wire (eg. heavy gauge copper like Shunyata, or highly conductive like Synergistic Research silver).

 

16 hours ago, ICUToo said:

I found a good Ethernet cable for the long run to my Mac from the router was helpful- not the biggest yield,

Consider enabling wifi in your mac and isolating that noisy thing from the network via wifi.

  • Like 1
Posted (edited)

@dbastin,

A useful and interesting opening post to this thread.  Thanks for the summary.  I have done much of what you propose.  For me it is important for those who have any Network Ethernet digital aspects to their system to appreciate one fundamental perspective.  Quality and synergy in the components are a must.  If you get it right the listening outcome will be enhanced than if you just treat the network as not important.  For me, my investment in my Network and the switches cables etc is worth every one of the many considerable cents I have spent.  The Network and its bits are definitely the contributory source of my final listening outcome.  If you get that right you are well on the way to quality. 

 

 

For me mac books and Pcs etc are a no no.  A dedicated audio server is a must.  The only thing that I have not tried is master clock.  Just more cables and I have more than enough now.  May be one day

 

John

Edited by Assisi
  • Like 1
Posted

@Assisi sadly a media PC is a must for me purely because I've zero chance of my wife learning how to change inputs on my Mutec MC-3+USB. I know as I've tried.  If sound doesn't work when she wants to use it, I'll recieve an angry phone callat work. Just not worth it for me. Sound quality is not always the only consideration. :(

  • Like 1
  • Love 1
Posted (edited)
1 hour ago, MattyW said:

I've zero chance of my wife learning how to change inputs on my Mutec MC-3+USB

I understand.  The phone call to me this evening  was:

"My gmail doesn't work.  How do I connect to my gmail account?

You do this I said.....

John

 

Edited by Assisi
  • Like 1
Posted

I'm taking baby steps. Have finally got rid of my below average Optus 4G internet and have got the fastest Opticomm plan available. Gosh why didn't I do this sooner.

 

I have connected my CXN to the router via ethernet for the first time. For a start, no more skips on Tidal especially when streaming MQA files. We are off to a good start.

 

Have two pairs of Audioquest Forest Ethernet cables on the way to sit between wall & router and router & CXN. Have also got a LPS for the router itself on the way.

 

I will report back with any differences if I hear any,

Posted

I have a relatively simple system, effectively Commercial grade as my sons are/were electricians, one has changed industry but still sets up and maintains our house Network, whilst the other is  Project Coordinators managing a team doing Electrical and Network cabling for a number of  data centres.  I also worked in IT for over 40 years and did a fair bit of Network planning as part of my Technology work so I find these discussions interesting.  We have had some good debates at home about the merits of audio power and the boys indulged with a dedicated circuit to my music room and helping me wire up some heavier wall plates s well as QA'ing my assembly of Furutech plugs/power cables.  Our theory so far is good bandwidth, using commercial grade equipment and ensuring clean power.  As an aside we built this house so had the luxury of designing (over engineering) all the electrical and cabling work, we also installed Digital lighting control with some sensor capability, which uses Ethernet switching as you would see in modern offices.  So we have 2 wired Ethernet backbone Networks in the house, plus WiFi.

 

All the Comms equipment is on UPS Power so I can assume clean power.  My youngest talks to me about buying a whole house electrical filter system .. and a 3 phase generator with battery and auto cut over which he will install  .. one day 😄

 

Up until late last year we had 2x ADSL2 plus a 4G Wireless Internet modem multiplexed to guarantee bandwidth to each of us, me with Music, boys home business, and lots of gaming (too many Ping complaints !!!) and then finally after a long wait we have NBN, a story in itself because on their advice for Fibre to the house we ran 50m conduit (x2 in case we needed power as well) to the street in preparation but they constantly morphed their implementation on the advice of others (according to KRudd) to cripple the NBN bandwidth so that option is only now available at exorbitant cost but at least the long wait for us meant we got FTTC.  With extended family members working from home during lock down we need guaranteed high bandwidth which is great for my Music and Video Streaming as well.  After research (Whirlpool) I pay a premium for Aussie Broadband including an additional fee for 40Mbps upload giving us 100Mbps/40Mbps.  International speed test shows I am getting 93Mbps down and 36 Mbps up at the moment so that pretty good.

 

The house is wired with 2 Cat 6e's to each room, head end is Cisco and Netgear Switches which are rack mounted, with UPS as i mentioned.

 

Wireless I upgraded to an Orbi AX6000 Tri-band Mesh WiFi 6  supplying 2.4Ghz and 5Ghz with a unit at each end of the house, also on UPS.  Wifi coverage showing at the moment for my Living room (Node 2i) is 229Mbps and the Bedroom (another Node 2i) showing 648 Mbps.  I am very impressed with Mesh WiFi 6, replaced a plethora of extenders that at times were struggling, I currently count 26 devices attached, Node 2i's, Laptops, various Tablet devices, Cameras, Nest (smoke sensors),  Google and Alexa's devices, and Mobile Phones.  It is surprising how many devices connect nowadays but we have 5 adults living in the house at the moment.

 

The Lumin U1 Mini (with SPower LPS), Synology NAS running MinimServer, desk top computers, and 3 Android TV's are Wired Ethernet.

Posted
8 minutes ago, Rosco8 said:

All the Comms equipment is on UPS Power so I can assume clean power.

UPS power can be (is usually) some of the dirtiest.

 

.... but I don't really think that the power into your comms equipment is of much concern.    It's about what goes into your audio devices..... and this means anything that is getting out of your comms equipment via the mains, or via the ethernet cable... and getting to the audio devices.... is what is relevant here.

 

 

 

  • Like 2
Posted (edited)
42 minutes ago, davewantsmoore said:

UPS power can be (is usually) some of the dirtiest.

It does depends on the UPS .. Sine Wave i thought was good ??

 

UPS with Pure Sine Wave 

The biggest difference between a regular UPS or a UPS with sine wave technology is the quality of output AC waveform being produced from the unit. 

AC power naturally has a perfect sine wave curve to it’s waveform, while DC power from a UPS battery produces a choppier wave curve. Lower end UPS units produce a ‘modified sine wave’ which looks like a staircase as opposed to higher end UPS units which can produce a pure sine wave which looks like a natural curve.

The pure sine wave tech provides an even cleaner source of power to all the gear plugged into it. Sine waves add more sensitivity and overall smoothness making it ideal for getting a clean audio recording. 

Edited by Rosco8
Posted

Well what do you know....  The fake Nordost Odin CAT8 cables are actually rather excellent even before burn in. I kind of suspected they had potential for this due to the individual shielding of each wire though it was still a gamble.  Be interesting to see what burn in brings. Very vivid and some detail has come to the forefront that wasn't audible with the Supra CAT8.

 

I've always disliked this wire in any other application (SPDIF, RCA, USB) though for network use its rather excellent. 

Posted
Just now, Rosco8 said:

It does depends on the UPS .. Sine Wave i thought was good ??

It does....   they can be terrible (because most devices they'll power don't care) ..... often the ones which advertise "pure" .... are just "not terrible"....  and claims using the word "sine wave" are marketing fiction.

Posted (edited)
1 hour ago, shailesh said:

@MattyW do you mind sharing the link to the fake Odin ?

 

Sure thing. I've 2x 0.75m cables. 1 on my server and 1 on my media PC. I've ordered a longer one for my Soundaware A1X. Everywhere else has Supra CAT8.

 

https://a.aliexpress.com/_mNz94iL

Edited by MattyW
  • 3 months later...
Posted (edited)

Hi, Do any SN users run streamers via network bridge direct to servers/Roon cores. 

This functionality would require Win10/Linux OS, but it would be great if Roon Rock V2 would include this ability.  

 

What SQ improvements are noticed if any?

 

I.e.:

https://community.roonlabs.com/t/bridged-ethernet-any-pros-or-all-cons/44108/3

 

 

Edited by Grizaudio
  • 3 months later...
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...
To Top