Does Ethernet Cable makes sound difference like an USB cable?

[dbastin]

Oh, in hindsight my statement 'let this mess with your brain' could've read like I was being rude, so I am sorry if it sounded that way.  I was intending to be humorous.

 

And I was speaking to any reader, not just you @Cruncher. 

 

I need to clarify, both Wireworld cables are Platinum series 7 Cat 8 except one is 1m and the other 50cm.

 

@Cruncher, are you implying that there is a technical explanation why they may not sound the same in the situation I described?  And the NEXT measurement might help explain why?

 

I dont know what NEXT is, or how to measure it.  In the absense of my abiity to take technical measurements, I resort to experimenting and listening, and not in a scientifically controlled way - I just change only ond thing at a time, and listen for quite a while to become accustomed to any change.

 

I'd be interested in measurements if they could reliably predict how cables will sound. 

Edited April 13, 2020 by dbastin
typo

[Guest rmpfyf]

1 hour ago, dbastin said:

Oh, in hindsight my statement 'let this mess with your brain' could've read like I was being rude, so I am sorry if it sounded that way.  I was intending to be humorous.

 

And I was speaking to any reader, not just you @Cruncher. 

 

I need to clarify, both Wireworld cables are Platinum series 7 Cat 8 except one is 1m and the other 50cm.

 

@Cruncher, are you implying that there is a technical explanation why they may not sound the same in the situation I described?  And the NEXT measurement might help explain why?

 

I dont know what NEXT is, or how to measure it.  In the absense if my abiity to take technical measurements, I resort to experimenting and listening, and not in a scientifically controlled way - I just change only ond thing at a time, and listen for quite a while to become accustomed to any change.

 

I'd be interested in measurements if they could reliably predict how cables will sound. 

 

Two cables, same type, different length, slightly different sound. 

 

Given the way Ethernet works, jitter is the only answer.

 

You can change whatever you want downstream, it's likely a ton more expensive than sorting jitter performance closer to the DAC IC.

[Cruncher]

@dbastin I think there is a technical explanation on why you hear a difference between cables. This is what I am trying to figure out.

 

When I started my career I did some cabling work and a lot of troubleshooting. The links that gave the most grief often had sub-par cabling. Cables that met the spec were ok. Cables that exceeded the spec I found the most reliable over a wider variety of circumstances. We had equipment in electrically noisy and hostile environments.

 

Category 5/6/7/8 are standards. To meet the standard you need to meet a variety of conditions. Near End Cross Talk (NEXT) and Far End Cross Talk FEXT) are two of several test criteria. My working theory is these are making the difference. All commercial cable suppliers will provide test reports. Where I used to work I had test gear. I found test tools invaluable and used then to rule in or rule out any theories we may have had.

[Steffen]

5 hours ago, dbastin said:

Well, let this mess with you brain ...

 

Firstly, to be clear, my experience is that some ethernet cables with unobtainium conductors do make a difference - I have some.

So, when you unplug the Ethernet cable, does the sound improve or get worse?

[Guest rmpfyf]

2 hours ago, Cruncher said:

@dbastin I think there is a technical explanation on why you hear a difference between cables. This is what I am trying to figure out.

 

When I started my career I did some cabling work and a lot of troubleshooting. The links that gave the most grief often had sub-par cabling. Cables that met the spec were ok. Cables that exceeded the spec I found the most reliable over a wider variety of circumstances. We had equipment in electrically noisy and hostile environments.

 

Category 5/6/7/8 are standards. To meet the standard you need to meet a variety of conditions. Near End Cross Talk (NEXT) and Far End Cross Talk FEXT) are two of several test criteria. My working theory is these are making the difference. All commercial cable suppliers will provide test reports. Where I used to work I had test gear. I found test tools invaluable and used then to rule in or rule out any theories we may have had.

 

Pays also to remember that these things are developed for different purposes. CAT8 isn't supposed to run mega distances, f'rinstance. 

 

People super worried about about whatever crap can get picked up along a wired connection should just go optical, work on good switching at one end, get a good receiver at the other and tune packet sizes/network traffic/IRQ priorities.

 

Conductors out of precious metals, matched length conductors, etc... it's amusing to watch the audiophile world throw the usual abstracts at it and charge epiphanic amounts. Back over in network land-ia this stuff was solved many years ago for differing applications. 

 

Admittedly there are many that are not able to run optical for a variety of reasons, though it'd pay to assess the problem and deal directly - what's the problem?

 

1 hour ago, Steffen said:

So, when you unplug the Ethernet cable, does the sound improve or get worse?

 

We're on the verge of someone making a programmable relay that physically open-circuits the Ethernet line during playback. A plugin for whatever package you'd use for playback automates coping a track over, stuffing it into memory, disconnecting the network, playing back and reconnecting the network.

 

Now, @Steffen  if you silver plate it at some point and give it a fancy name I'd think you could retail it for $10k easy  

[Steffen]

4 minutes ago, rmpfyf said:

Now, @Steffen  if you silver plate it at some point and give it a fancy name I'd think you could retail it for $10k easy  

Sorry, I’m rather busy bringing my Cable Direction Diviner™  to market. It will feature pairs of sockets for RCA, balanced, coax, Toslink, RJ45, USB, power (AU) and banana, and LEDs indicating which direction the cable under testing should be used. It’s early days, but I can promise it will be under $30k, and I’m confident it will fly off the shelves.

 

Now, cable direction depends on many factors, including the hour of day, barometric pressure and sunspots, so don’t be surprised if the device indicates a different direction for a given cable each time you use it. 

[dbastin]

11 hours ago, Steffen said:

So, when you unplug the Ethernet cable, does the sound improve or get worse?

That is a curious experiment because it only works for a few seconds.  To be able to listen for longer (my strong preference), a cable needs to stay plugged in (even if I use fibre all the way before that).  On that basis, the cable I am using is the best sounding I've had.

Edited April 13, 2020 by dbastin

[Guest rmpfyf]

2 hours ago, dbastin said:

That is a curious experiment because it only works for a few seconds.  To be able to listen for longer (my strong preference), a cable needs to stay plugged in (even if I use fibre all the way before that).  On that basis, the cable I am using is the best sounding I've had.

What's your endpoint - if a PC you should be able to copy to ram and play from there.

[dbastin]

1 hour ago, rmpfyf said:

What's your endpoint - if a PC you should be able to copy to ram and play from there.

I have an Antipodes Server and Devialet Pro as endpoint, using Roon.  Neither have optical ethernet input.  Separating server and renderer achieved big gains, Roon gave further gains (and ut relies on ethernet) so this set up requires ethernet to get audio data from server the endpoint.  So your suggestion wont work for me.  And, I am not permitted to save Tidal to a drive, which is probably encrypted.

[Guest rmpfyf]

20 minutes ago, dbastin said:

Separating server and renderer achieved big gains, 

No surprises there

 

20 minutes ago, dbastin said:

Roon gave further gains (and ut relies on ethernet)

There's better than Roon IMHO, and it doesn't 'depend' on Ethernet any more than it does any other network connection

 

20 minutes ago, dbastin said:

And, I am not permitted to save Tidal to a drive, which is probably encrypted.

Encrypted no (zero advantages for an audiophile rig), and what you'd be doing is no less legit than buffering a track locally (in essence what it is) - so no issue there

 

If you're getting audible differences on the cable between the two machines I'd not be terribly impressed with the renderer IMHO at that price. Clearly whatever it does to isolate noise and reclock signals is affected by what's upstream. The only other possibility is much worse so let's not explore that. Half the point of separating both renderer and streamer is jitter reduction.

 

I would (very simply) create a test rig with ease - any PC, pref fanless, a copy of anything that'll play from RAM, and run your A/B cable tests there. 

 

If reticulating network to your server from your main network I'd run SFP/SFP+ on decent switches either end to deal with as much as you can.

Edited April 13, 2020 by rmpfyf

[Pim]

On 12/04/2020 at 8:13 AM, pwstereo said:

Imagine how bad streaming should sound with all the different Ethernet cables and switches and routers and what-nots between the server and your playback equipment. I bet those data centre, telco,  and ISP techs aren't selecting the finest gold/platinum/diamond Cat 8 super cables when they making patch changes or installing new equipment.

 

That's because Cat.8 doesn't exist

[BugPowderDust]

On 05/07/2020 at 1:36 PM, Pim said:

That's because Cat.8 doesn't exist

Really? ISO/IEC 11801-1, Edition 1.0 2017-11 says otherwise.

[davewantsmoore]

On 23/03/2020 at 1:16 PM, gwurb said:

This may have something to do with higher data throughput.

If the physical interface is 1gbps ... then there is no "higher data throughput".

 

... in fact.  It offers no higher data rate for 10gbps ethernet either.

 

[davewantsmoore]

On 23/03/2020 at 5:40 PM, gwurb said:

What makes you say its unlikely to be about priority of traffic?

Because it's trivial to re-create these conditions on purpose (packet loss, latency, etc.) ..... and see (hear) that it has zero effect.

[gwurb]

3 minutes ago, davewantsmoore said:

If the physical interface is 1gbps ... then there is no "higher data throughput".

 

... in fact.  It offers no higher data rate for 10gbps ethernet either.

 

You won't get maximum throughput out of an interface with every cable. There may have been a problem with the first cable. "higher data throughput" could indeed have occurred with the new cable, or maybe it didn't. Either way it was a guess on my behalf.

1 minute ago, davewantsmoore said:

Because it's trivial to re-create these conditions on purpose (packet loss, latency, etc.) ..... and see (hear) that it has zero effect.

So I am back to:

1) Network player needs to not experience buffer underrun due to cables that act as a bottleneck, or due to packet flow issues

2) Check to make sure that the cable is allowing sufficient speed of communication

3) On busy networks QoS may help with buffer underrun, if in fact packet flow is the issue

4) Make sure that the interfaces themselves have no problems

4) If everything that I mentioned above is ok then SQ changes are a placebo effect

[Ittaku]

4 minutes ago, gwurb said:

You won't get maximum throughput out of an interface with every cable. There may have been a problem with the first cable. "higher data throughput" could indeed have occurred with the new cable, or maybe it didn't. Either way it was a guess on my behalf.

You misunderstand. CD resolution audio requires 1.4Mbits per second. If you have a gigabit ethernet connection, you have more than 600x the bandwidth you require. Even if you were using a coat-hanger instead of a fancy ethernet cable, you will not run out of bandwidth. 96/24 highres audio is still only 9.2Mbits per second

Edited July 10, 2020 by Ittaku

[davewantsmoore]

On 12/04/2020 at 6:26 AM, Cruncher said:

I struggle when I read material from companies that selling audiophile Ethernet cables and products. I see a lot of discussion on Ethernet corruption, packet loss, delay and “Ethernet noise” ( I need someone to show me this and explain it to me) and how bad it is and how it can interfere with the clocks in DACs and wreck audio etc etc but I never see and data.

The person who can show you this, is someone who can actually do the experiment.

 

ie. inject controlled amounts of noise, and show the evidence that the DAC, or analogue amplifier, is being affected.... of course, part of the whole design process of audio equipment (especially digital equipment) is to minimise this.

 

I will leave you to speculate why this doesn't happen (there are many, and often conflicting, reasons).

On 12/04/2020 at 6:26 AM, Cruncher said:

From the stats you can see I have transmitted 14 Trillion packets with not a single CRC error. 

Yes... digital transmission is very very very robust.

 

Another issue that is possible is that electrical noise entering via the transmitting device(s), or radiated into the cable itself .... is entering the DA converter, and messing with the DA conversion.

 

It's fairly trivial to demonstrate this in a controlled test.... inject noise... show analogue response differences.    I've seen some instances of that being done......   but I'll leave you to speculate on why it is not more common.

 

On 12/04/2020 at 6:26 AM, Cruncher said:

If ‘Ethernet Noise’ is messing with your DAC’s clock that is simply poor design

Yes, DA should be very well guarded against this, if the product is to have any hope of practical high performance.

[davewantsmoore]

On 13/04/2020 at 10:23 AM, dbastin said:

That is a curious experiment because it only works for a few seconds.  To be able to listen for longer (my strong preference), a cable needs to stay plugged in (even if I use fibre all the way before that).  On that basis, the cable I am using is the best sounding I've had.

Some streamers (perhaps not at the regular UI, but) can be configured to cache a larger amount of data...... which will let this test run for an appropriate amount of time.

 

As someone who has tried it (in a number of different permutations) .... I could tell you there is no change.

 

What normally happens then is a pat on the back about how I'm fnancially better off, with the inference that my systsem isn't up to it, and my ears are painted on.    .... which is why I tend to stick to the theoretical/evnidence side of the chat.

[davewantsmoore]

16 minutes ago, gwurb said:

You won't get maximum throughput out of an interface with every cable.

Nonsense.

I have spent my whole life working in IT/datacentres.

 

If the cable doesn't get the maximum rate it is "broken" and goes in the bin.

 

16 minutes ago, gwurb said:

There may have been a problem with the first cable."higher data throughput" could indeed have occurred with the new cable, or maybe it didn't.

Grade-A, nonsense.

 

If either cable didn't transmit 1gbps.... then it is defective to cat5e.

 

[gwurb]

On 21/03/2020 at 10:22 PM, I'mInterested said:

Hi, I was pleasantly surprised by the difference in going between blue jeans cat6 and the Cardas clear cat7. After some settling in I found the music had a whole lot more “body” with the Cardas. So happy I gave it a go.

 

13 minutes ago, Ittaku said:

You misunderstand. CD resolution audio requires 1.4Mbits per second. If you have a gigabit ethernet connection, you have more than 600x the bandwidth you require. Even if you were using a coat-hanger instead of a fancy ethernet cable, you will not run out of bandwidth. 96/24 highres audio is still only 9.2Mbits per second

No misunderstanding. I have zero knowledge of how good or bad the network is setup for that comment. No idea about traffic flows, no idea about cable quality. It is improbable that there were buffer underruns, but I am willing to accept the possibility that they could have been occurring thus a change in cable maybe made a difference.

 

I am back to if no buffer underruns are occurring then it's placebo.

[gwurb]

Just now, davewantsmoore said:

Nonsense.

I have spent my whole life working in IT/datacentres.

 

If the cable doesn't get the maximum rate it is "broken" and goes in the bin.

 

Grade-A, nonsense.

 

If either cable didn't transmit 1gbps.... then it is defective to cat5e.

 

Let me re-write it for you:

There may have been a problem with the first cable == the cable was defective

[Guest rmpfyf]

It's relatively easy to test that the data got there. If it didn't, cable >> bin.

 

The only audible differences are down to jitter. Which has many potential sources.

[davewantsmoore]

 

 

2 hours ago, gwurb said:

Let me re-write it for you:

There may have been a problem with the first cable == the cable was defective

OK.... You could have just said that to the other guy....  instead of inferring that "Cat7 has a higher data throughput than Cat6... so that might be it" 

 

[davewantsmoore]

On 12/04/2020 at 9:58 PM, Cruncher said:

Category 5/6/7/8 are standards. To meet the standard you need to meet a variety of conditions. Near End Cross Talk (NEXT) and Far End Cross Talk FEXT) are two of several test criteria. My working theory is these are making the difference. All commercial cable suppliers will provide test reports. Where I used to work I had test gear. I found test tools invaluable and used then to rule in or rule out any theories we may have had.

The ethernet leads that I use in/around my audio system all pass with flying colours (BJC).

 

In our data centre, it is much harder to pass these specs considering lengths, closeness to other conductors (in large installs)  and surrounding environments.