Ethernet Routers for Audio

[Assisi]

15 hours ago, dbastin said:

Do you have examples of these?

Few that come to mind.  There must be others as well

·         The quality of the internal components, wiring, clocks and sockets

·         CNC construction of the device box including separate compartments to section off and isolate components and prevent some resonance getting in and internal interference

·         Internal linear power supplies

·         Battery power

·         Cable quality, construction and plugs

Not all network devices and cables would include all of the above but they would feature in some ways in the better more expensive devices and cables.

 

 

15 hours ago, dbastin said:

The fastest home plans seem to be 100Mbps.

 

I read about a enterprise plan being able to reach almost 1000Mbps (1Gbps).

 

With my comment about the NBN I am not talking about NBN speeds at all.  My comment is what I understand about internal network activity, control and interrupts and the probable benefit for the audio signal.

 

15 hours ago, dbastin said:

See attached imagine of inside CRS 305 and its largish size clock to the right of the heat sink.

 

There are clocks and there are many clocks.  You mention the one in the CRS 305 as being “largish”.  Size of itself is not necessarily an indicator of quality.  Some clocks chips alone probably cost many times more than the CRS 305.

 

 

A SNA member a couple of days ago created five convolution filters for ROON for me  That evening I wanted to compare the outcome.  Short time listening and power failure.  Due to three power failures and other factors, I turned my system off.  I did not get to listen again for 36 hours.  All components including switches were cold.  I turned the system on and the outcome was quite disappointing.  Now I have never experienced any noticeable improvement when the amps etc start cold.  After a few hours and I was listening to the best I have heard from my system with the filters running.  Fantastic.  My assumption was that the clocks in the switches needed to warm up as they are all OCXO hence the initial SQ disappointment.  Normally I never turn the network including the switches off.  Quality clocks are very important.

 

John

 

 

 

 

 

 

 

Edited January 30, 2022 by Assisi

[dbastin]

3 hours ago, Assisi said:

Few that come to mind.

Oh, I assumed you were referring to things that can be done with regular network devices, cables etc that implement mitigation strategies.  I am quite aware of the various things audio grade switches employ, again only 1 router employs any of these.

 

3 hours ago, Assisi said:

My comment is what I understand about internal network activity, control and interrupts and the probable benefit for the audio signal

I gather that because CRS 305 and other SFP+/10Gbps hardware are designed to handle such high speeds (ie. 10x the fastest nbn home plan) which may require better parts, clocks, grounding, and better at the things you mention erc that apparently benefits audio a lot.  The clock may not be super good, but likely to be much better than typical domestic routers. The reference to the clock size is that clocks I have seen in audio gear tend to be larger generally.

 

https://audiophilestyle.com/forums/topic/24002-optical-network-configurations/page/59/

 

I'm not justifying CRS 305 specs, or comparing to audio grade switches, but as a router the CRS 305 is very satisfying, even in a cold system, and of course much better still when warmed up and its not only the network clocks, there are clocks in the server, DAC, etc

 

I think what this reiterates is that every device (and its clock) makes a difference.

 

Moreso, without a good performing router, perhaps the audio grade switches are compromised.

 

I think the next weakest link is the nbn Network Termination Device, which is probably cheap and 1Gbps tech.  I see the business class NTDs can do 10Gbps line rate and more, for example ...

 

 

And the switches in the nbn network are probably well above 10Gbps, perhaps in Tbps realm.

 

I gather the clocks AfterDark recycles are from telcos.

Edited January 30, 2022 by dbastin

[Carrera1963]

On 29/1/2022 at 5:09 PM, Assisi said:

An important aspect of a quality switch is the clock.  Jitter and Timing!  It all makes a difference. 

 

Only relevant to real-time protocols where buffering & error correction isn’t available. 
 

we’re not doing ultra-low-latency high-frequency trading!

 

if you’re streaming (buffered) audio from the Internet, the datagram that hits your streamer is always identical to the one that left the server.

 

 the only way a switch or router can impact sound quality of a buffered stream is if it’s causing or propagating electrical noise (or there’s insufficient bandwidth due to congestion or errors causing buffer starvation)

 

Edited January 30, 2022 by Carrera1963

[Assisi]

48 minutes ago, dbastin said:

Oh, I assumed you were referring to things that can be done with regular network devices, cables etc that implement mitigation strategies.  I am quite aware of the various things audio grade switches employ, again only 1 router employs any of these.

 

I did say audio switches.  With the strategies I mentioned there is an expense and therefore that is reflected in the price of the device but also probably in the outcome.  As in all audio, I do not think that there is is a low cost benefit outcome equivalent to a higher cost one.  Everything is relative.

 

51 minutes ago, dbastin said:

 

Moreso, without a good performing router, perhaps the audio grade switches are compromised.

 

I think the next weakest link is the nbn Network Termination Device, which is probably cheap and 1Gbps tech.  I see the business class NTDs can do 10Gbps line rate and more, for example ...

I do not have the answer.  The NBN box and its relative speed is minuscule in the grand scheme of things.  As I have said previously maybe you need to start again with a serious audio switch that is affordable to you and forget about the permutations and combinations of FMC, sfp, optical, Ethernet and  Etheregen etc.  In the end I predict it will be better outcome for you.

John

[Assisi]

1 hour ago, Carrera1963 said:

Only relevant to real-time protocols where buffering & error correction isn’t available. 
 

we’re not doing ultra-low-latency high-frequency trading!

 

if you’re streaming (buffered) audio from the Internet, the datagram that hits your streamer is always identical to the one that left the server.

 

 the only way a switch or router can impact sound quality of a buffered stream is if it’s causing or propagating electrical noise (or there’s insufficient bandwidth due to congestion or errors causing buffer starvation)

To me you are coming at this matter from a perspective that is not orientated towards the audio requirements of a network.  I cannot to explain why what you are trying to say is not relevant to me.  My understanding is that jitter and timing is critical in the quality of audio digital performance.  I do wonder whether you are aware of the fundamental importance of jitter and timing with audio. 

 

When I mentioned my previous NBN experience I did not stream at all.  When I tried my first couple of audio switches, I did not stream at all.  Everything I listened to came from a quality NAS.  I resisted steaming for a long time even though I had components that I could use to stream.    I now stream and the quality is seriously excellent and I am sure it is because of the network and switches.  So, what you are saying is not appropriate to me.  I upset and annoy some posters when I say go and try and experience what I can experience.  So reluctantly in this case again I say try it.  Forget the theory.

John

[dbastin]

2 hours ago, Carrera1963 said:

 the only way a switch or router can impact sound quality of a buffered stream is if it’s causing or propagating electrical noise (or there’s insufficient bandwidth due to congestion or errors causing buffer starvation)

Firstly, if you haven't already, I suggest read about phase noise.

 

Also, I think I demonstrated above/earlier that even changes upstream of fiber isolation (changing cables, exchanging FMC for EtherRegen) can make a difference to sound quality which seems contrary to your claim, and Assisi's (John)

4 hours ago, Assisi said:

NBN box and its relative speed is minuscule in the grand scheme of things. 

 

I would've thought similarly about the router too.  However my recent experience with CRS 305 indicates a eorthwhile improvement could be gained if the nbn box was also 10Gbps spec.

 

I suspect it could be very small of followed by a series of high performance OCXO clocked audio grade switches.

 

As I said earlier, there are many things that don't make sense.  We all need to ...

4 hours ago, Assisi said:

Forget the theory.

... and create and test new hypotheses.

 

4 hours ago, Assisi said:

As in all audio, I do not think that there is is a low cost benefit outcome equivalent to a higher cost one.  Everything is relative.

I generally agree.  I am not looking for a low cost alternative switch in this topic, here I am exploring routers.

 

If Wrouter was just a router (rather than more an all in one with 2 banks of switch and a renderer), we could consider that, but its prices buys a lot more than a router.

 

So we are left to consider The Linear Solution hot-rodded router/modem/ switch/wifi, and standard network devices.

 

As for a switch, I am hoping someone with invent one that has excellent isolation (like ER, Gigafoil) excellent clock (or provision for master clock), excellent chassis, provisions for battery power, resonance control, etc, etc ... and fibre input.  I might be waiting a long time, so may DIY putting together existing products.

 

Until then you may be right in implying I am tinkering about - however this tinkering produces an excellent sound.

[Assisi]

8 hours ago, dbastin said:

I generally agree.  I am not looking for a low cost alternative switch in this topic, here I am exploring routers.

 

If Wrouter was just a router (rather than more an all in one with 2 banks of switch and a renderer), we could consider that, but its prices buys a lot more than a router.

 

So we are left to consider The Linear Solution hot-rodded router/modem/ switch/wifi, and standard network devices.

 

As for a switch, I am hoping someone with invent one that has excellent isolation (like ER, Gigafoil) excellent clock (or provision for master clock), excellent chassis, provisions for battery power, resonance control, etc, etc ... and fibre input.  I might be waiting a long time, so may DIY putting together existing products.

 

I use the Wrouter just as a switch.  I am going to investigate having a router including wifi connected to the NBN box to separate the audio from the rest of the network. 

Most or all of us with our audio are on different journeys with an intention of reaching a similar destination.  The best sounding outcome.  To me there is as strong relationship between price, component quality and outcomes (destination).  The law of diminishing returns will always prevail.  It is possible to achieve a very decent outcome with lower priced components.  Based on what you say this is the case with CRS 305.  However, I suggest a device designed/engineered for audio purposes whilst more expensive will probably provide an even better final outcome.

The ingredients above that you describe for a switch will have a definite price attached.

8 hours ago, dbastin said:

Until then you may be right in implying I am tinkering about - however this tinkering produces an excellent sound.

I also tinker.  Not much with the network though.  It seems to me that with your tinkering you make a number changes in short periods of time.  I do think that is not a good thing to do.  When moving stuff around, there needs to be a reasonable time period for the things to settle down again.  Often several days.  Only then will you be able to detect whether there is a actual benefit or not.  You say the tinkering produces excellent sound.  That only means some thing to you and relative to what?  In some cases maybe the expression “You can't make a silk purse out of a sow's ear” is applicable.

John

John

 

[Hydrology]

15 hours ago, Assisi said:

 So reluctantly in this case again I say try it.  Forget the theory.

As a college qualified Electronics Engineer, I agree 100% - forget what you think you know. What’s the cost to just try something to satisfy your curiosity?

[Carrera1963]

On 30/1/2022 at 8:49 PM, Assisi said:

To me you are coming at this matter from a perspective that is not orientated towards the audio requirements of a network.  I cannot to explain why what you are trying to say is not relevant to me.  My understanding is that jitter and timing is critical in the quality of audio digital performance.  I do wonder whether you are aware of the fundamental importance of jitter and timing with audio. 

Yes, admittedly I’m coming at this from an elec eng & networking background with experience of broadcast digital video & audio environments, which is why I’m saying that for a buffered audio stream if there’s something impacting sound quality it’d be electrical noise/interference rather than a fancy switch or patch lead upstream.

 

If you’ve got a combination of switches/cables/whatever that you’re happy with, great - I’m happy with mine (except the bloody OLED TV with a 100Mbps Ethernet port!)

[Assisi]

14 hours ago, Carrera1963 said:

Yes, admittedly I’m coming at this from an elec eng & networking background with experience of broadcast digital video & audio environments, which is why I’m saying that for a buffered audio stream if there’s something impacting sound quality it’d be electrical noise/interference rather than a fancy switch or patch lead upstream.

 

If you’ve got a combination of switches/cables/whatever that you’re happy with, great - I’m happy with mine (except the bloody OLED TV with a 100Mbps Ethernet port!)

A few years go when I first tried audio switches and cables it was just for music stored on a NAS.  No streaming at all.  I then tried streaming and liked it.  The benefits that I attain from the network switches and cables and filters to me has nothing to do with buffered audio stream and everything to do with how the devices and cables manage what goes through.

 

I understand that you are happy with your present setup.  You might be a bit challenged and surprised by the outcome if you tried an audio switch instead of the switches that you use. 

 

I do not use the network for Video.  Just free to air and a Bluray player.  I consider that for audio, 100Mbps is better than higher speed.

John

[dbastin]

22 hours ago, Carrera1963 said:

Yes, admittedly I’m coming at this from an elec eng & networking background with experience of broadcast digital video & audio environments, which is why I’m saying that for a buffered audio stream if there’s something impacting sound quality it’d be electrical noise/interference rather than a fancy switch or patch lead upstream.

How would a novice with no expertise in identifying and eliminating noise/interference work out the source of the noise problem and how to fix it?

 

If an experts advice is needed, then what would that cost?

 

How does the noise and interference get into the network to spoil sound quality?

 

What would be the typical problems and solutions?

 

For instance you mentioned a dodgy fridge and light.  How does noise from the get into those network?  And what could be done to stop that from happening?

Edited February 2, 2022 by dbastin
typo

[Carrera1963]

1 hour ago, dbastin said:

How does the noise and interference get into the network to spoil sound quality?

It’s not “in” the network - I’m saying that if there’s an electrical noise issue that’s improved by changing a component or cable, then it’s worth looking for the source of the issue.

 

A common electrical issue in audio is a ground loop - in theory you could have a ground loop via a network cable (maybe a shielded cable with metal plugs). Changing the cable could make the problem go away - though more likely “upgrading” to a “better” shielded cable could introduce the problem in the first place.

 

or you could have a signal cable (eg RCA analog audio) running next to the power cable for your fridge, and noise/interference to the audio signal could happen every time the compressor switches on/off.

 

These types of interference can easily impact analog signals by adding noise. As someone mentioned earlier, an oscilloscope  would be the tool to try and check signals for electrical noise.

 

The network (IP packets) would not be impacted by electrical noise in the same way & therefore are not susceptible to outside interference in the same way. The packets either get there or they don’t, so as long as the data is delivered it will always be identical quality.

 

Wifi signals are RF and susceptible to RF interference, including things like microwave ovens!

 

If someone has audio noise issues and improves it by replacing a network component or cable, it suggests the analog audio path is/was being impacted by electrical noise, rather than the IP network signal being improved by the change.

[dbastin]

On 02/02/2022 at 5:36 PM, Carrera1963 said:

It’s not “in” the network - I’m saying that if there’s an electrical noise issue that’s improved by changing a component or cable, then it’s worth looking for the source of the issue.

 

A common electrical issue in audio is a ground loop - in theory you could have a ground loop via a network cable (maybe a shielded cable with metal plugs). Changing the cable could make the problem go away - though more likely “upgrading” to a “better” shielded cable could introduce the problem in the first place.

 

or you could have a signal cable (eg RCA analog audio) running next to the power cable for your fridge, and noise/interference to the audio signal could happen every time the compressor switches on/off.

 

These types of interference can easily impact analog signals by adding noise. As someone mentioned earlier, an oscilloscope  would be the tool to try and check signals for electrical noise.

 

The network (IP packets) would not be impacted by electrical noise in the same way & therefore are not susceptible to outside interference in the same way. The packets either get there or they don’t, so as long as the data is delivered it will always be identical quality.

 

Wifi signals are RF and susceptible to RF interference, including things like microwave ovens!

 

If someone has audio noise issues and improves it by replacing a network component or cable, it suggests the analog audio path is/was being impacted by electrical noise, rather than the IP network signal being improved by the change.

 

By "Network" I mean the physical gear - the router, switches, FMCs, cables.  I realise the electrical/EMI/RF noise can not get in the data.

 

So what you are suggesting is:

• keep network gear and cables away from 'noisy' appliances.
• pay attention to break common ground connections between network gear, cables and hifi gear.

The latter may be partly why removing 2 ERs between my CRS router and WAP (see previous post) improved the sound in many ways.  The ERs LPSs were powered from the same power conditioner as the endpoint (Devialets) - the outlets share the same ground buss and there is no conditioning between the conditioners outlets.

 

Fortunately I have a 2nd power conditioner so will put that back into service to power just the router and server so that any ground connection between those and Devialets gets isolated or at least 'treated' by the conditioners.  This 2nd conditioner has multi-stage filters that reduce power supply-generated noise and prevent power line noise from one component contaminating the other adjacent electronic components.

 

Stay tuned for the result.

[El Tel]

With regards to treating audio as a specialist application of networking, what purpose do the clocks in the switches and routers provide?

[dbastin]

48 minutes ago, El Tel said:

With regards to treating audio as a specialist application of networking, what purpose do the clocks in the switches and routers provide?

This is a good start.

https://audiophilestyle.com/forums/topic/58648-discussion-of-the-uptone-jswenson-etherregen-white-paper/

https://audiophilestyle.com/forums/topic/56501-audiophiles-giving-too-much-credits-on-clock-in-network-applications/

[El Tel]

16 minutes ago, dbastin said:

This is a good start.

https://audiophilestyle.com/forums/topic/58648-discussion-of-the-uptone-jswenson-etherregen-white-paper/

https://audiophilestyle.com/forums/topic/56501-audiophiles-giving-too-much-credits-on-clock-in-network-applications/

Got about 3 pages into the paper in the first link and it's so full of information that is contradictory to the mechanisms that Ethernet uses to prevent such issues. I'm not sure they demonstrate a full appreciation of the asynchronous nature of Ethernet (8B1Q4 and 4D-PAM5 encoding which carry the timing within the data stream without needing to synchronise clocks at either end) or how data is reassembled into an analogue stream by a DAC chip (the timing is pre-determined by the sampling rate). Mention of ground-planes is a red herring if we use the proper unshielded cabling for the job - that's why the newer CAT7 & CAT8 standards need to be understood well and not selected for use in domestic audio lest they introduce interference from upstream equipment.

 

When all things are taken into consideration, and I'm including data buffering as a standard here, I'm not sure there is much to be learned in that paper.

 

I was hoping to find something that had eluded me to date. Thanks for the links though. Always happy to readjust my views and opinions with new information.

[Assisi]

On 04/02/2022 at 6:19 PM, El Tel said:

Got about 3 pages into the paper in the first link and it's so full of information that is contradictory to the mechanisms that Ethernet uses to prevent such issues. I'm not sure they demonstrate a full appreciation of the asynchronous nature of Ethernet (8B1Q4 and 4D-PAM5 encoding which carry the timing within the data stream without needing to synchronise clocks at either end) or how data is reassembled into an analogue stream by a DAC chip (the timing is pre-determined by the sampling rate). Mention of ground-planes is a red herring if we use the proper unshielded cabling for the job - that's why the newer CAT7 & CAT8 standards need to be understood well and not selected for use in domestic audio lest they introduce interference from upstream equipment.

 

When all things are taken into consideration, and I'm including data buffering as a standard here, I'm not sure there is much to be learned in that paper.

 

I was hoping to find something that had eluded me to date. Thanks for the links though. Always happy to readjust my views and opinions with new information.

 

In a post above the one quoted you mentioned Clocks.  They and other aspects of networking for audio are very important.  A number of posters on SNA come from the perspective of extensive networking experience, theory understanding and knowledge.  The experience etc is probably limited or non-existent in the context and requirements of audio networking and streaming.  In fact, for someone like me to say what I do, is probably just a little bit challenging to the theory.

 

 

I know very little about the technicalities.  I do not need to know much to make things work for me.  My understanding is that the extensive networks around the world obey the TCP/IP rules, standards and protocols.  Without the rules etc the extensive universal wired connectivity would fail.  The rules etc were developed so that all the packets arrive on time and intact at the specific destination.  It definitely works.

 

However, the rules etc were not developed so as to comply with today’s requirements of high level quality audio streaming.  With an audio network in your home, the timing and how intact the 1s and 0s are, is absolutely critical to the listening experience.  What happens before the audio stream reaches you house network is not as important as after the stream reaches your own network.  The house network switches with serious clocks, quality cables and power supplies from my experience can significantly improve the streaming playback SQ experience.  It is about the Time Domain.  The Frequency Domain happens later.

 

Quality clocks deal and improve the timing and jitter situation.  Components in and the design of the switches and power supplies can deal with interference and resonance.  For example, the port lights in some switches do not exist or they can be turned off.  The flicking lights can cause interference.  The potential for resonance and interference in the network can be in many places.  You and others may question and ask why quality switches can improve the outcome in the last few Metres.  To me it is analogous to the question as to why the last few Metres of a quality power cable can improve the listening outcome as the power has already travelled 100s or 1000s of Kms.  They just do.  I just appreciate the outcome.  Everything matters.

John

 

[BugPowderDust]

13 minutes ago, Assisi said:

In a post above the one quoted you mentioned Clocks.  They and other aspects of networking for audio are very important.  A number of posters on SNA come from the perspective of extensive networking experience, theory understanding and knowledge.  The experience etc is probably limited or non-existent in the context and requirements of audio networking and streaming.  In fact, for someone like me to say what I do, is probably just a little bit challenging to the theory.

John, that's actually a bit obnoxious sounding.

 

You know my background, and @El Tel has decades of similar experience. What he says above is precisely correct in both an audio and video context when streaming data. The clock in your DAC does nothing at the packet or datagram layer. Everything that happens in ethernet land is clocked asynchronously, but super reliably. 

 

But what you don't know is what else people like us have done with our personal and professional time.  Personally, I've been involved in audio production for decades, have put on many shows with international acts that I've toured in Australia, as well as having been involved in the design and implementation in some of the biggest networks in this country which stream more audio AND video than you could conceive which led to Telstra buying my company for the expertise and capability.

 

I think you need to wind it back in a bit.

[El Tel]

Pondering whether to respond on this or not. But what the hell... I'm still not seeing a compelling argument here.

 

TCP/IP does not provide the mechanism of timing; it exists lower down the OSI layer than that. Ethernet itself provides the mechanisms, not the upper layer protocols that deal with routing, session integrity, and upwards into presentation and application layers of the OSI model.

 

To draw the analogy of power leads in the last few metres versus grid power generation and transmission is actually a false equivalence. They refine and filter out unwanted interference, do not change the state of the electricity.

 

Allow me, if you will, to use a different analogy: The passage of data from a stream through to being a post-DAC-processing audible event should be thought of as being a state change. Power filtering is not a state change. Data goes through a process where its fundamental state is altered once it is recompiled by the DAC into an electrical analogue signal. Think of it as a kettle where electrical energy is converted into heat energy; as long as the electrical input conforms to basic standards, its quality can be hugely variable and it does not impact the water negatively. It just boils the water and cannot, by definition of its changed-state affect the water - it just boils it.

 

There are huge amounts of misinformation and sometimes wilful misunderstanding in this sub-component of audio. It's directly analogous to both sides of the highly-charged and contentious vaccination opinion argument. The same way as vaccines are concerned, in the case of Ethernet (asynchronous timing, galvanic isolation, error correction, TCP/IP, etc etc), the science is already settled.

 

Please, go spend your money on whatever you desire. Go nuts. Knock yourself out. But perhaps we should really understand what it is to challenge the integrity of the meticulously documented evolution of decades of scientific work into globally adopted industry standards that are implemented across billions of end-points on the planet? There might be a slightly different set of conclusions to be had if genuine, non-psychoacoustic differences are heard.

 

Cheers

El Tel

[dbastin]

It is very hard to change strongly and deeply held beliefs.  The mind is a very persuasive manipulator of perception.  I wonder if objectivity is even really possible.  But that can be debated elsewhere - here we talk about using routers in a network for hifi audio.

 

My experience with the CRS 305 is real to me.  It improves the sound so much.  Perhaps the network experts can share insights into why 10Gbps/SPF+ specifications could improve sound quality.

 

Picking up a point by @Assisi, it occurs to me the speed and volume of data being handled in the network outside your home requires quite high spec hardware.  I suspect that my home nbn connects to 10Gbps or higher grade hardware in the nbn network.  The CRS 305 gives an insight into how the audio data on the www is handled.

 

But when it reaches my property, the nbn Network Termination Device (NTD) is 1Gbps max., given that is the max nbn will deliver wholesale.

 

So those of us brave (or stupid) enough can put aside the theory and beliefs and attempt to improve our home network after the nbn NTD.  And, in my experience every piece of network gear, cable, etc matters (unfortunately for my finances).  I don't need to know why it should or should not.

 

What would be beneficial is for those with networking expertise to apply that to try to find cost effective solutions to improve sound quality ... focus on what could be possible rather than what should not be possible.  Then explain it in layman's language, because the technical language is only of value to those who know enough to decipher if it is valid.

[El Tel]

18 minutes ago, dbastin said:

It is very hard to change strongly and deeply held beliefs.  The mind is a very persuasive manipulator of perception.  I wonder if objectivity is even really possible.  But that can be debated elsewhere - here we talk about using routers in a network for hifi audio.

 

My experience with the CRS 305 is real to me.  It improves the sound so much.  Perhaps the network experts can share insights into why 10Gbps/SPF+ specifications could improve sound quality.

 

Picking up a point by @Assisi, it occurs to me the speed and volume of data being handled in the network outside your home requires quite high spec hardware.  I suspect that my home nbn connects to 10Gbps or higher grade hardware in the nbn network.  The CRS 305 gives an insight into how the audio data on the www is handled.

 

But when it reaches my property, the nbn Network Termination Device (NTD) is 1Gbps max., given that is the max nbn will deliver wholesale.

 

So those of us brave (or stupid) enough can put aside the theory and beliefs and attempt to improve our home network after the nbn NTD.  And, in my experience every piece of network gear, cable, etc matters (unfortunately for my finances).  I don't need to know why it should or should not.

 

What would be beneficial is for those with networking expertise to apply that to try to find cost effective solutions to improve sound quality ... focus on what could be possible rather than what should not be possible.  Then explain it in layman's language, because the technical language is only of value to those who know enough to decipher if it is valid.

Sorry for what is about to come.

 

Upstream nbn hardware is of variable quality and speed.  Some places don't see much more than 20Mbps which arrives on premise via a single run of twisted pair. It's an anathema that the transmission medium improves the further upstream you are.

 

I can't explain why your subjective opinion on the professed improvement has weight. Unfortunately, the burden of proof is not on those who are just reading the standards and have decades of experience with them.

 

Fact is, we're dealing with cached/buffered audio, not real-time like voice-over-IP. Try this: start a listening session over your preferred streaming service and then pull the cable between your router and the first switch it is connected to. You will not interrupt the stream or hear any difference in the output until the buffer has emptied. Then the only sound you will hear as the buffer empties, is the tumbleweed of realising there is no impact to the quality of the continuation stream until it ends.

 

 

[dbastin]

4 hours ago, El Tel said:

Try this: start a listening session over your preferred streaming service and then pull the cable between your router and the first switch it is connected to. You will not interrupt the stream or hear any difference in the output until the buffer has emptied.

Would you believe me if the result was not what you expected?

 

How do we know you have legitimate expertise in this subject?

 

I am not stirring, just pointing out your subjective opinion is no more proven than mine.

 

Let's not degrade this debate.

 

When I try something that works well for me, I share it in case others want to try it to see if it works for them too.  That is it. I am not here to convince anyone. 

 

I am confident most people who have looked into ethernet for audio such as threads like this will have become aware of the technical arguments you refer to.  But if you feel the need to educate, start a thread about that.

 

However if your experience using different routers for audio may benefit others, this is the thread.

 

Incidentally, I am pretty sure I could tell which of my cables had been unplugged in your test.  In fact, if I have audio playing from my server directly to my endpoint, I am confident I could detect if a cable was unplugged further upstream, even though no data is coming from uostream, and probably which one (maybe except fibre).  Not saying am superhuman, just very familiar with changes in my system.

[El Tel]

6 hours ago, dbastin said:

Would you believe me if the result was not what you expected?

 

How do we know you have legitimate expertise in this subject?

 

I am not stirring, just pointing out your subjective opinion is no more proven than mine.

 

Let's not degrade this debate.

 

When I try something that works well for me, I share it in case others want to try it to see if it works for them too.  That is it. I am not here to convince anyone. 

 

I am confident most people who have looked into ethernet for audio such as threads like this will have become aware of the technical arguments you refer to.  But if you feel the need to educate, start a thread about that.

 

However if your experience using different routers for audio may benefit others, this is the thread.

 

Incidentally, I am pretty sure I could tell which of my cables had been unplugged in your test.  In fact, if I have audio playing from my server directly to my endpoint, I am confident I could detect if a cable was unplugged further upstream, even though no data is coming from uostream, and probably which one (maybe except fibre).  Not saying am superhuman, just very familiar with changes in my system.

Over 30 years in tech (from hardware engineer, wireman, project lead right through to COO of a successful electrical engineering and IT consulting company), including industry certifications, qualifications, society memberships, white paper submissions, circuit design for a number of building management and fire detection systems. Also extensive experience in real-time audiovisual broadcast/distribution system design and implementations including both unicast and multicast.

 

I agree, let's not allow this to degenerate.

 

I've tried specialist audio switches in various setups including commercial environments. They did nothing. I've been privy to the results of internal inspection and teardown of equipment making claims such as this and seen no sub-component differences between a $400 switch and a $40 switch.

 

As I said, if you want to spend on this stuff, knock yourself out. I'm satisfied from both experience and my modest professional expertise that this is a cul de sac.

 

I'm done here.

[BugPowderDust]

I love how people playing "lego" with of the shelf equipment (which they often can't get working), somehow are the oracle's of this new sound frontier and have gleaned hidden knowledge that millions of others have yet to learn, despite decades of real world experience in precisely these technologies and the same application.

 

I've tried to be open to new experiences but I've not heard anything that convinces me of anything WRT audiophile networking yet and the scientific measurements of same tell me there's nothing to see here either. All of this is precisely what the theory of IP networking would tell us anyway.

 

@El Tel most of us usually smile and exit these threads in the same manner as you have.

 

 

[BugPowderDust]

13 hours ago, dbastin said:

What would be beneficial is for those with networking expertise to apply that to try to find cost effective solutions to improve sound quality ... focus on what could be possible rather than what should not be possible. 

Start here: 

 

This is Dante specific but most of the same concepts apply. In many of your networks though, you won't be using Dante but locally streaming via TCP/IP using traffic distribution protocols like multicast. You won't need to synchronise multiple endpoints, and if you did, you're likely to do it using a protocol like RAAT from Roon to do so. Dante is for large scale audio distribution in live or event type environments. Please don't confuse it for streaming one stream from your Raspberry Pi to your DAC.

 

Note from the intro: "Using Dante does not require you to be a networking expert. The majority of the time you can safely leave settings at defaults and get great results".

 

This is great advice.

 

This advice is the same in many networks unless they are heavily over subscribed or under specified in terms of performance. In a home network, it would be rare to see so much traffic or congestion to require anything more than basic VLAN segmentation to isolate your audio traffic from your non audio traffic (you should be doing this to keep IoT devices off your LAN network anyway).

Where you route traffic from the various VLANs together you would have a router or a firewall.  If you tag audio traffic with DSCP or CoS markings on the inbound port, you can get the firewall to prioritise this traffic first as it traverses the firewall. This means audio first, everything else when capacity exists.

 

This is basic quality of service functionality invented first for IP telephony on the late 1990s. I was configuring protocols like class based weighted fair queuing for Cisco IP telephony platforms when I worked there at the turn of the century after their acquisition of Selsius. This is the same config concept we applied when building networks like Foxtel and BigPond's streaming networks and it's what I'd apply if I thought my home network was in need of it.

 

Note most of the audiophile gear you buy can't be configured to do any of this, as it doesn't have a management plane interface.