Exactly why do Ethernet cables sound different?

Well, there may be a few reasons, but the ones I am aware of is RFI, shielding drain current, twisted pair imbalances, and connector design.

The cable itself might might modulate RF ground currents in the connected devices and this is what we probably hear especially when the connected devices are sensitive audio devices

Texas Instruments amongst many are aware of this and design various hardware design drivers  and provide hardware design guides to help mitigate this.

So i say it's not so much the cable that sounds different, more the interaction of the cable with the connected equipment that sounds different, and this interaction is determined by the properties of the cable.. But this has nothing to do with data frames or bits, and that perhaps is what your son is thinking of, and I agree.

Simon

At this point, I do not see your son's point. Beside being a potential source of RFI, data transmission is itself subject to errors, see for instance http://en.wikipedia.org/wiki/E...ction_and_correction. I am not aware of either first principles or rigorous statistical data on the impact of cables and connections on the quality of audio data transmission over TCP/IP suggesting such impact to be void. Thus, I have no reasons to believe that different wire construction and different connectors do not make a difference. Nor, of course, that they do.

I insist we should be told what the 'special tape' is?

Clearly the resistance to fungus growth is the key factor. I wonder if that will convince Ethan.

I have been wondering why the difference since I heard a Cord demo. No question the hi end Cord cable was better.

Somthing is going on here. It makes no sence. However:

- could the better cable reduce physical vibration fed into the player?

- are the players not properly isolating network traffic from the audio components. So RF noise etc. is getting transferred to the audio components?

Paul

Isolation works well for DC, ie galvanic isolators which Ethernet NICs have are to avoid earth loops and the like.. Alas common mode RF currents will quite happily hop across a galvanic isolator .. It's kind of the nature of RF current. Additionally the carrier frequency of each direction in 100BaseT link is approx 31 MHz...  but the twisted pairs should cancel any emissions out and zero any resultant current, but that assumes they are theoretically perfect.

Good for Ethan. He's doing a useful degree course and one that will stand him in very good stead for his future career.

I have no reason to doubt that different Ethernet cables will make a difference and agree that it's more down to noise coupling into the audio device and this will vary depending on network configuration and the environment where the audio device is placed and also how sensitive the device is to external interference.

Given that most will be sceptical (the bits are bits brigade - they have a point but this is not digital data in isolation) of the effects of a digital cable, whether Ethernet, USB or Coaxial S/PDIF that no magazine (or indeed manufacturer) has done any in depth testing to show why these effects happen and what is actually happening. The perceived audio output has changed to the listener so why is this - ground currents changing, noise getting onto the clock supplies causing phase noise and changes in jitter spectrum etc etc. The only thing i've seen meausure is rise time figures for USB cables but again, what effect does this have on the downstream device. We can discuss the relative theories on why this may happen but until someone actually does some measurements to show the effect digital cables have on the resulting audio output, we can only speculate.

James

PS - Nigel - sit him down and demonstrate that your cables make a difference (if he can hear it). Might be an interesting final year project for him to produce a test setup to demonstrate why they make a difference :-)

Do the manufacturers not shed any light on it, or is it usual marketing techno babble. I mean if they create these cables are they designing them or is it just hit and miss.

If they are designing them then surely they must have an idea of what is going on.

Graeme

 It's a very simple computation: the amount of dollars coming out of a connector is linear in the number of electrons that gets in and in its length. It is quadratic in its diameter.

Ah that's it then problem solved.

Just changed all mine from freebie cat5e cables that I received with various routers to cat6 belkin cables (so still relatively cheap) and the difference was very noticeable. I had to try it myself as there are so many who say its all placebo and there cannot be any difference and others swear there is a difference. Well there definately is a marked difference!

Now I promised I would make a posting to the Chord representative at the recent Cloney Audio event here in Dublin.

Despite my objections that there shouldn't be any difference from the type Ethernet cable used to connect a network switch to a streamer device, there was a clear difference between the Chord C-Stream and the Chord Sarum Tuned ARAY streaming cable. The demonstration was using a ND5XS, from a NAS to a 5-port switch in the room.

And I still can't explain to myself or anyone else why there should be. The data stream in this instance is a communication stream using TCP/IP principles, which under the definition specifies how the data should be formatted, addressed, transmitted, routed and received at the destination with data recovery procedures. I do understand how the four abstraction layers (out of the OSI 7-layer model)  are used and I have no explanation of how the cable used at the link/physical layer can influence the sound of the music being transmitted between two essentially computer devices. Within the protocol stack implemented in both the NIC card in streamer and the switch, the protocol stack will be receiving the data stream into a memory buffer of a few kB, then once it has been completely received, subject to the handshake/header details, the data packet is copied into the buffer for the next layer and so on. Please note, that these memory buffers are all independent of the device's Buffer, and is within the TCP/IP communication process. If the data packet is not correctly received, as per the handshake/header details the data is re-requested etc.

Anyway there was a difference in the track being used, with respect to separation and clearer/cleaner across the whole of the music.

Some explanation was given to the fact that the Sarum Tuned ARAY streaming cable (and also the Indigo Tuned ARAY streaming cable, which also exhibited improvement, just not as marked), uses digital streaming cables for the data signals and not just foil encased twisted pairs, as in the C-Stream and most other CAT 6, 6a, 7 ethernet cables. Maybe this changes the type of RF noise present from the cable and changes the way this noise interacts with the device, I don't know.

Again, I promised the Chord representative that I would make this post, after giving him a hard time about why there shouldn't be a difference. As yet, I haven't changed the CAT 6a cable feeding my streamer from the RPI based UPnP server, but it is on the list.

Simon_in_Dublin

(BTW I was a firmware Engineer in ICL in the mid-90's working on fibre optic based communication between mainframes, after graduating in Electronics)

I can understand an SQ change  (didn't say improvement) with Chord TA cables using data streaming cable configuration rather than twisted pairs.  Then if the TA cable section installed is only that between the player & switch & the other section(s) using ordinary twisted pairs, how does that work?  If it's claimed the SQ gains are still there, how come a short section of data streaming cable does that, is that not implying it's a filter? 

I get the feeling it's not quite as simple as just the Chord TA design;  I can understand optimised twisted pairs (twists per metre) solid v stranded, screened pairs v screened pairs + overall screen, PVC v PTFE - all such as we get between cat5 6 6a 7 & 7a.   I fear this discussion will go on for a few more months - maybe years. & maybe we will never get to the bottom of it.

What does that actually mean. I understand what a standard ethernet cable with its twisted pairs is, but what is a data streaming cable configuration.

Is it a beldon screened twisted pair type ?.

Graeme

It's true that the digital data passed to the streamer are the same (provided there aren't network errors).

However the streamer isn't just a digital device, there is also analogue circuitry in there; and that is subject to all sorts of external influences (including, but not limited to, small amounts of RFI being introduced by the cable acting as an antenna).  This susceptibility to external influence is why I believe the results are so variable - different cables, different streamers, different system configurations and different environments can all play a part.

I am very concerned about the potential serious consequences of the test set out in the second paragraph. I'm really not sure I'd take the risk - he's a very nice young man. 

The question is not so much whether the cables sound different, but rather, if they do, then the technical reason why that might be so. 

Id still like to know what "digital streaming cables" are, rather than twisted pairs. Is this why they sound different

Potentially fatal (or even harmful) tests performed on humans are usually considered unethical (and do attract the attention of the Police).  However the possibility that he could proclaim anything after dying would at least answer the question of life after death!

Sorry Wat, I just couldn't resist that.  This reply is intended purely as humour, please don't take offence, none at all is intended - I've made so many slip-ups myself.

Most are twisted pairs, some aren't.

Chord TA cables are a set of four micro coax cables in an overall screen, most of the others are essentially SSTP or SFTP made to tight tolerances and often with additional outer 'jackets'.

Thanks Huge, never new you could make cat cables from coax

The Ethernet Category-X specifications actually only refer to the electrical properties of the cable rather than the construction - manufacturers are free to use what ever construction they please, so long as it meets the prescribed performance criteria.

Having said that, to the best of my knowledge,  Chord don't specify a particular Category rating for their TA cables.  Chord C-Stream however exceeds Cat7, but there again it's a fairly conventional SSTP construction.

If this is so, it's appalling. Ethernet BaseT lead needs to use balanced conductors and network interface adapters are designed for balanced pairs. A balanced conductor is either usually a ladder line / parallel feed with space / dielectric around or a twisted pair. 

Coax is only effective for unbalanced signals. Sending a balanced signal down an unbalanced lead with out using UNBAL and BALUN transformers will make the cables well out of spec with respect to the poor network interface card, and it will radiate horribly or cause excessive shield drain current or become highly susceptible to interference. Either way it will almost certainly not conform to the Ethernet BaseT physical spec  and I would keep well clear.

Simon

Simon,

From the Chord website...

"The Sarum Tuned ARAY streaming cable is not a conventional streaming/Ethernet cable; it does not rely on twisted pair conductor geometry, it’s a design that has been built to carry data to an audio streamer, making use of the design work used to produce Sarum Digital Tuned ARAY co-axial and USB cables.  It employs an extremely high quality sub-miniature coaxial high speed cable for all critical signal paths."

I agree with your comments about balanced / unbalanced, and probably the reason that I havent seen a CatX class rating for these cables.  I think it may be a case of it working in practice because 1000BaseT Ethernet devices are surprisingly tolerant when working with 100BaseT signals!

P.S. I use a Chord C-Stream; that's a conventional Cat7 SFTP and it does seem to work well with the ND5 XS.  This cable is currently the subject of some on-going experimentation.

Huge, thanks.. Well perhaps such cables should have a non compliance warning if they are marketed to be used for Ethernet. From memory it's the Ethernet RFCs under 802.3 that defines the requirements of the cable for use with Ethernet rather than the CAT specifications themselves.

Also remember 1000BaseT is a different physical protocol and way of working compared to 100BaseT. At the end of the day the NIC needs a balanced connector on the send and receive, anything else and your forcing the NIC to work out of spec and therefore ultimately potentially unreliable operation and unforeseen consequences with switches and NICs. thanks for the rather shocking heads up!

Simon

That is my understanding also.  I believe that the Category specs are essentially a list of test criteria rather than a formal design spec.