Ethernet System for Audio: putting it all together

[Grizaudio]

post removed.

Edited August 15, 2022 by Grizaudio
incorrect information.

[Assisi]

1 hour ago, Grizaudio said:

 

CNC case, OCXO, LPS. LHY is a subsidiary of Jay's audio. 

https://www.beatechnik.com/product-page/lhy-sw-8

Anyone compared one of these v's renolabs? 

 

Looks decent:

 

It will be interesting as to whether there are any available at the moment.  Some audio manufactures are have difficulties getting Aluminium for the CNCd enclosures for their products as Russia is one on the world’s major suppliers of the metal. 

John

[Cardiiiii]

1 hour ago, Snoopy8 said:

Patience, young grasshopper! 

 

Never a turntable. I made a conscious decision not to go there.

----

 

It is something that is not new and not Ethernet related (hence new thread). It may disappoint some of you but it made me go Wow! It was the sound I was looking for.  I will post probably tomorrow...

 

Bet it's a bottle of wine

[AccuTidal]

5 minutes ago, Cardiiiii said:

 

Bet it's a bottle of wine

Something like this will motivate?!

[bob_m_54]

40 minutes ago, Assisi said:

 

It will be interesting as to whether there are any available at the moment.  Some audio manufactures are have difficulties getting Aluminium for the CNCd enclosures for their products as Russia is one on the world’s major suppliers of the metal. 

John

I think China and India have it covered.. Aluminium Production by Country

[dbastin]

19 hours ago, Grizaudio said:

 

CNC case, OCXO, LPS. LHY is a subsidiary of Jay's audio. 

https://www.beatechnik.com/product-page/lhy-sw-8

Anyone compared one of these v's renolabs? 

 

Looks decent:

Could we put discussion about the switches on this thread ...

 

l

[Cardiiiii]

My router's current power brick is rated at 12V 3.3A. Would I be doing any damage if I use a LPS rated at 16V 4A? Thanks.

[AccuTidal]

36 minutes ago, Cardiiiii said:

My router's current power brick is rated at 12V 3.3A. Would I be doing any damage if I use a LPS rated at 16V 4A? Thanks.

Certainly will. Try to stay within the voltage tolerance of 0.5vdc plus or minus. 

[Snoopy8]

Going back to OP’s request for people to share how they have put their ethernet systems together, and why.

 

I have been on a journey of simplifying my HiF since early 2019.  The Phantoms have now given way to the Kii Three. In all this time, I have been using a standard home network, with all music on a NAS. 

 

The Uptone EtherRegen (ER) was the first Ethernet switch that caught my somewhat skeptical attention. I was curious to try it, knowing that there was a money back guarantee.   However, the ER surprised me by the SQ improvement.  And then, the “rabbits started breeding!” 

 

 

A second switch came, with potential for more, and fibre optic with different SFPs (small form-factor pluggable) were added to the mix.  Not shown in the diagram were an external clock and quite a few LPS (linear power supplies).  Needless to say, there were many boxes and a web of cables. 

 

To make things worse, added a JCAT card to the NUC (which had to be housed somewhere to protect it) plus its own LPS.  All my previous simplification efforts were unraveling. 

 

 

After much re-thinking, testing and listening, found that the Renolabs was good enough to replace the “rabbit warren” and still deliver excellent SQ.  The Renolabs connects everything now, including the NAS and Orbi. I had “simple” back.

 

Earlier, I may have created unrealistic expectations about a device that gave even an bigger SQ jump than the Renolabs. It was the Innuos Phoenix USB Reclocker, more info in this thread.  

 

However, having a smaller SQ improvment (than the Phoenix USB) does not diminish the message that investing in the Ethernet can bring SQ improvements. Not everyone agrees about the improved SQ but I only ask that you approach Ethernet for audio with an open mind.  You may discover some new and fruitful ways to improve SQ....

[Grizaudio]

On 15/08/2022 at 8:55 PM, bob_m_54 said:

I think China and India have it covered.. Aluminium Production by Country

Plenty available FYI 

[Grizaudio]

20 hours ago, Snoopy8 said:

Going back to OP’s request for people to share how they have put their ethernet systems together, and why.

 

I have been on a journey of simplifying my HiF since early 2019.  The Phantoms have now given way to the Kii Three. In all this time, I have been using a standard home network, with all music on a NAS. 

 

The Uptone EtherRegen (ER) was the first Ethernet switch that caught my somewhat skeptical attention. I was curious to try it, knowing that there was a money back guarantee.   However, the ER surprised me by the SQ improvement.  And then, the “rabbits started breeding!” 

 

 

A second switch came, with potential for more, and fibre optic with different SFPs (small form-factor pluggable) were added to the mix.  Not shown in the diagram were an external clock and quite a few LPS (linear power supplies).  Needless to say, there were many boxes and a web of cables. 

 

To make things worse, added a JCAT card to the NUC (which had to be housed somewhere to protect it) plus its own LPS.  All my previous simplification efforts were unraveling. 

 

 

After much re-thinking, testing and listening, found that the Renolabs was good enough to replace the “rabbit warren” and still deliver excellent SQ.  The Renolabs connects everything now, including the NAS and Orbi. I had “simple” back.

 

Earlier, I may have created unrealistic expectations about a device that gave even an bigger SQ jump than the Renolabs. It was the Innuos Phoenix USB Reclocker, more info in this thread.  

 

However, having a smaller SQ improvment (than the Phoenix USB) does not diminish the message that investing in the Ethernet can bring SQ improvements. Not everyone agrees about the improved SQ but I only ask that you approach Ethernet for audio with an open mind.  You may discover some new and fruitful ways to improve SQ....

 

I'm definitely open to trying a modified switch. 

Also considering the Phoenix. 

 

Which gave them biggest improvement in your system Snoopy? 

I would also question the necessity for the XE with the Phoenix, but you did say you heard an improvement leaving it in signal. 

My Pi4 sounds better than my NUC in my system. The NUC sounds more on edge. Just my opinion.  

Out of the box the Pi4 has less glare. 

 

Edited August 17, 2022 by Grizaudio

[bob_m_54]

1 hour ago, Grizaudio said:

Plenty available FYI 

No, I don't need any aluminium at the moment, but today I did spend $210.00 for a piece of 1.5mm stainless, 920mm X 460mm with most of the middle cut out. 

[Snoopy8]

55 minutes ago, Grizaudio said:

Which gave them biggest improvement in your system Snoopy? 

I would also question the necessity for the XE with the Phoenix, but you did say you heard an improvement leaving it in signal.

The Phoenix easily beats the JCAT XE.  I too, thought the JCAT could be made redundant!  However, it was cleaner and there was more detail with the Phoenix + JCAT than the Phoenix with standard USB port.  Hence, I am running both.  It re-confirms Darko's findings that the both the source feeding the Phoenix and the Phoenix itself can influence SQ.

 

56 minutes ago, Grizaudio said:

My Pi4 sounds better than my NUC in my system. The NUC sounds more on edge. Just my opinion.  

Out of the box the Pi4 has less glare.

If you prefer the Pi, stick with it.

 

I have been happy with the NUC running SnakeOil for quite some time.  I am delighted with sound of the NUC+JCAT+Phoenix and have no desire to look at the Pi nor something else. 

 

At the end of the day, it comes down to personal taste...

[Grizaudio]

14 minutes ago, Snoopy8 said:

The Phoenix easily beats the JCAT XE.  I too, thought the JCAT could be made redundant!  However, it was cleaner and there was more detail with the Phoenix + JCAT than the Phoenix with standard USB port.  Hence, I am running both.  It re-confirms Darko's findings that the both the source feeding the Phoenix and the Phoenix itself can influence SQ.

 

If you prefer the Pi, stick with it.

 

I have been happy with the NUC running SnakeOil for quite some time.  I am delighted with sound of the NUC+JCAT+Phoenix and have no desire to look at the Pi nor something else. 

 

At the end of the day, it comes down to personal taste...

Snoopy thanks very much for commenting. 
 

I think you have convinced me to try the Phoenix before the network switch. 
 

I read your Phoenix review also, seems to sit well with most other Phoenix reviews. 
 

I agree regarding preferences, it’s a very subjective game…..but I always found the native USB on the NUC to sound a bit hard/digital. The Pi sounds  more relaxed. I don’t understand why Darko rates it higher than the Pi on the Phoenix. Will need to try for myself. I think the extra perceived detail is actually higher noise. just an opinion. 
you can see this on the archimajo review of the HS01. 
 

I also didn’t have luck with multichannel and NUC/ROCK usb output. The Pi worked perfectly with Ubuntu. 
 

 

[Cardiiiii]

Anyone try the Telos Macro G RJ45?

[dbastin]

On 19/08/2022 at 6:55 PM, Cardiiiii said:

Anyone try the Telos Macro G RJ45?

As an indication of the potential and quality of Telos, I have the Telos Macro Q (USB) and it is very effective.

[dbastin]

I recently came across these posts about cabling which is really worth considering.  I am pretty sure I rad a similar post elsewhere by someone with similar experience.  It contributes to the 'everything matters' but perhaps highlights attention to smaller details may at least reduce the potential for problems that are assumed to be elsewhere.

 

[Hydrology]

https://darko.audio/2022/08/expert-opinion-mark-jenkins-goes-deep-on-digital-audio/

[Assisi]

For me a very useful discussion that provided much knowledge and wisdom to contemplate.

John

[DanFi]

Has anyone compared a well designed SMPS to a LPS on all digital devices such as a switch? 
 

i.e iFi iPowerX vs GIESELER for instance. 
 

im reading it could have an effect on speed of the switch but this may be irrelevant for audio quality 

[Stereophilus]

I've tried iFi iPower's vs uptone JS2 vs Gieseler on all sorts of digital devices from DDCs to switches to FMCs.  The LPS' will usually sound better, if there is a difference at all. I would caution that how you connect your power supplies also matters, as does using multiple outputs from 1 device.  

 

Also note that the iFi devices use technology to suppress power supply noise in the audible band.  They do not do much for switching frequency noise.

[dbastin]

On 02/09/2022 at 7:18 AM, DanFi said:

im reading it could have an effect on speed of the switch but this may be irrelevant for audio quality

I don't think psu affects the function or speed of the switch.  But it seems to me noise getting into a switch or router can have an impact that is audible.  

 

Ifi PowerX does not seem to reduce noisetransmitted through its output as much as LPSs I have tried.  A downside of ifi Power X could be the DC wire.

 

In short, ifi Power X is better than stock smps but there are lower noise options.

 

And try to power each switch from seperate power sources where possible ... especially power your streamer independently and not shared with any switches.

[dbastin]

 

On 29/08/2022 at 5:30 PM, Hydrology said:

https://darko.audio/2022/08/expert-opinion-mark-jenkins-goes-deep-on-digital-audio/

That is a thoroughly worthwhile and informative interview.  It is not short, but hang in there.

 

I think Darko is far more knowledgable about this subject than he implies by his questions.  His approach seems similar to an investigative journalist ... probing for answers to awkward questions on behalf of those with 'general' knowledge.

 

Mark Jenkins carefully, calmly and clearly answers all, and hints more insight into Antipodes designs.  I feel most importantly he encourages experimentation and inquiry.  His observations seem to support the notion that everything matters, even if it can't be measured or proven. 

 

In relation to ethernet, some key points are:

• the data is always  correct
• the signal carrying the data can carry and transmit noise and cause distortion
• there are many types of distortion and noise, even on a pcb and every trace or wire is both an antenna and transmitter of noise that causes distortion
• 'cleaning' the signal of noise transmitted to the endpoint reduces workload at the endpoint and thus reduces noise created during the endpoint process of converting Asynchronous signal to Synchronous signal
• Minimising noise closest to the DAC chip is a key objective
• Reducing opportunities for congestion in the network can reduce workload in each device (eg. minimal resending of data packets), particularly important at the server and endpoint
• While DACs will reclock, reclocking prior and external to a DAC reduces reclocking workload on components in the DAC, which in turn generate less noise and distortion affecting the DAC output (but it depends on the DAC, they are not all equal, some benefit less)
• some patterns of noise and distortion are preferable in terms of impact on music enjoyment
• Beware of endpoints located adjacent to DAC chips, the endpoint processor noise can impact the DAC
• System synergy is partly a result of components (and cables) having compatible or complimentry noise patterns rather than undesirable intermodulation
• Galvanic isolation  (USB, ethernet) usually deals with noise sources other than thecsignal itself
• Galvanic isolation is not all equal In its effectiveness and can/will permit noise and the impact of noise on the signal to pass (ie. distorted square wave)
• LPSs are not the ideal, but SMPS that perform better are very costly

Edited September 4, 2022 by dbastin

[Assisi]

@dbastin

 

Thanks Dale for the thorough and useful summary of the interview.  There is much to take in with the interview.  Mark Jenkins responses to questions and explanations are most valuable.  Your summary clarifies the complexity of many aspects of the setup of the audio source in relation to the use of Ethernet.  When the network is done right the outcome is sublime. 

 

John

[dbastin]

18 hours ago, dbastin said:

• 'cleaning' the signal of noise transmitted to the endpoint reduces workload at the endpoint and thus reduces noise created during the endpoint process of converting Asynchronous signal to Synchronous signal
• Minimising noise closest to the DAC chip is a key objective
• Reducing opportunities for congestion in the network can reduce workload in each device (eg. minimal resending of data packets), particularly important at the server and endpoint
• While DACs will reclock, reclocking prior and external to a DAC reduces reclocking workload on components in the DAC, which in turn generate less noise and distortion affecting the DAC output (but it depends on the DAC, they are not all equal, some benefit less)

 

Just some after-thoughts and food for thought about these points ...

 

The benefit of sample rate conversions prior to the DAC might not be the new sample rate but moreso the fact that the digital info has been reconstructed with filters applied in the process, so the output is far less imperfect.  For instance, I use roon to upsample to 88 or 96kHz (depending on the native rate), and my Devialet upsamples everything to 192kHz, and so this is less work for the DAC than upsampling 44kHz to 192kHz.  Perhaps I should try using roon to upsample everything to 192kHz so the DAC has even less work to do.  BUT, this means the Devialet's renderer will be working harder to process the higher sample rate, and generating more noise.  Experimentation is called for here.

 

To reduce opportunities for network congestion, possibilities include:

• using multiple switches between the nbn and audio endpoint could maybe reduce transmission errors with each consecutive switch, reduce workload on each consecutive switch and reduce processing distortion
• pushing the busiest switching workload to a switch that is not in the chain to the endpoint (eg. from the router, connect a switch to serve the needs of the household traffic and then separately to that connect the audio part of the network to the router)
• the last device that sends data to the endpoint should have no other traffic

In my case I have:

• a fibre connection from the router to a switch that serves the household, therefore off-loading switching from the router and keeping router noise lower (and by using fibre it reduces the opportunity for noise from the household to affect the router's processing and passing on of data to the hifi).
• a very low power wifi access point (WAP) dedicated to connecting to the endpoint (Devialet) at only 100Mbps.  I will find out if the Mikrotik OS the WAP is using allows to reset that to lower rate (say 50Mbps) to reduce its workload and noise that could distort that last hand-over of data and also slow the rate of processing by the endpoint.

This could be why, when roon does an update that makes its traffic more efficient, I can here a difference.

 

I can't reclock the synchronous signal because its internal to the Devlalet, unless I install a new streamer/renderer and output USB or i2s to a reclocker and then send SPDIF to the Devialet.  That is a considerable extra expense and I really do have an excellent sound now even before the tweaks in the dot points above.

Edited September 5, 2022 by dbastin
edited text in italics to address concerns in subsequent post