Ethernet Cables

Yeah, my first IBM PC had 640k and I remember buying a special memory card (long type) to boost it to 1MB. MS-DOS couldn't directly access anything above 640k (I think that's where Bill's remark came from), but that remaining 384k came extremely useful as a RAM disk. I would copy applications there with autoexec.bat at they loaded much faster than from a hard disk.

Those were the days, my friend,

Bill thought they'd never end...

Gajdzin, regarding future proofing, I wouldn't go copper, I'd go fibre. Not really consumer grade at the moment, so my suggestion use trunking/conduit so you can replace the cables at some future point. 

For infrastructure in typically electrically low noise environments such as homes, Cat5e is fine and good for upto  1Gbps, much higher then as I said I would look to fibre unless using relatively short patch leads. Another option is Ether channeling the copper which involves using bonded pairs of Cat 5e or other into supporting devices such as switches, again not really consumer grade currently, so again trunking is the best option.

Using shielded (STP) or Cat 6 type cables seems to have advantage near audio equipment as the extra shielding helps reduce the RFI. But there is a lot of mumbo jumbo spoken about Ethernet patch leads and networks, and this seems to be proportional to the rise in popularity of home networks, I am afraid some marketing departments do seem to play on people's ignorance, perhaps not surprising Or new....

Simon

A good switch handles more throughput. I am not at any stage discussing whether music 'sounds better' I am talking about your network working reliably.

I have not rebooted my router since I purchased it nearly 4 months ago, its rock solid and the fastest (or there abouts) on the market, it cost 200 quid.

I would direct money at the router and switch, 60 quid will be nearly enough to wire the average house. Then get a really good router and a good switch if needed.

I stand by my comment that for the average house, Cat7 is just not needed. Cat5e will carry a 4k hidef stream no problem. Unless you have 7 kids all streaming hidef 4k video at once, cat7 is overkill to the extreme. It seems to be around 300 quid with vat for 305m, verses 60quid or less for cat5e, for exactly the same outcome.

garyi is right–I've seen properly-run drops of older Cat3 telephone wire (with good terminations) sweep well beyond what it should; Cat5e should be more than sufficient, especially if you do end up bonding a pair together (with managed switches at both ends) as Simon mentioned.

Conduit, as mentioned before, allows you to change your mind later....

Based on all gigabit switch tests I found (e.g. here: http://www.smallnetbuilder.com...showall=&start=1), actually... it doesn't   The differences between various brands and price points appear to be negligible in terms of performance.

Thanks, Simon. Always good advice from you Yes, I'll lay it in a conduit this time (the current lead is glued to the wall where it meets the ceiling all along the corridor and painted over - not very pretty nor safe - for the cable). Still I'm curious why nobody is recommending Cat7. Its not much more expensive than Cat 6e and surely even better shielded...? And still waaaay cheaper than the AudioQuest Forest etc.

I upgraded all my cables to Cat7 at such relatively low cost that I don't see the point in not using the best shielding. I heard a difference between them and Cat5 (but I didn't try any cables in-between. 

A good router is more important by far.

This is the second time in this thread that the question of router quality comes up. What is a "good router"? How can I say if it's good or not? BTW, in many cases the traffic between the data source (e.g. NAS) and the streamer is carried by a switch, not a router. A non-managed switch is a pretty simple beast. Tests in professional computer magazines show that there's almost no performance difference between switches, even the cheapest ones. So what should we look for?

I use an Apple Time Capsule (basically an Airport Extreme) and 100' of Cat6 cable between that and my ND5XS. I have never experienced dropped audio with that all the way to 24/192. No extra amount of shielding and no other switch is going to change the sound in anyway. It either works or it doesn't.

How would I use a switch when connecting my mac, my qute and the internet? A switch would be better than a router, I presume.

I don't know what a good router is, I had a bad one I installed a linksys e900 (simple yet reliable) I was suffering glitches with (I forget what I had before). Linksys does like to be reset every 10 days or so.

That's what they were saying about the sound of CD transports 20 years ago (before they discovered jitter)

I have a bit of a fettish for routers. My set up is vigin media into one of there terrible superhubs set to modem only.

To this a router, and from the router to a 24 port netgear switch. There are at any one time around 30 things on my network on wireless and wired.

I have had lots of routers, draytek, asus, apple etc.

I use the Asus RT AC68U.

Before this I had an apple new style extreme

Before this an Asus RT N66U

Now logic dictates I suppose that as all of my wired things come off the network switch the router should make little difference.

However with the apple in the mix, my throughput speed halved compared to the Asus before it on wired. So what ever is happening the router is an influence.

Asus make great routers. Apple make reliable but slow routers. Draytek make great routers for the geek, you can do all sorts to them.

I would avoid anything including routers by belkin and Dlink, but thats just my personal experience. I have not tried billion routers, but they are supposed to be good too.

The ethernet cable won't affect the audio quality for the simple reason there is no audio on the ethernet cable. It is all packetized raw data. The NIC in your client device reassembles the data back into audio. If something is missing the NIC requests retransmission. If that happens too much there is a problem with your network (or hardware, or cable). Like I said, the ethernet cable either works or it doesn't work. There is no "audiophile" ethernet cable. That's snake oil.

That's precisely what everybody was saying about the CD transports 20 years ago. There is no audio in the digital cable between the transport and the DAC.

That was before they discovered jitter...

There's several reasons that are consistent with our knowledge of physics and can explain why one ethernet cable can "sound" different than the other, such as:

- the amount of transmission errors that the decoding device has to deal with, increasing its workload, power consumption, etc. - all analogue phenomena that may affect the performance of its audio circuits

- RFI picked up by the ethernet cable

If you have problems with the cable (i.e. dropped packets, RFI, etc) then that results in dropped audio, not diminished sound quality otherwise. When you get dropped audio you then have to deal with getting a cable that works. But once you have an adequate cable (and router/switch), no other will improve the sound quality.

So let's think about "packetized raw data" from the physical layer perspective.  It's a big collection of 0s and 1s.  A 1 is defined as a certain voltage being reached (or exceeded) on the cable.  So a 1 is represented by a square waveform.  0s and 1s are therefore represented by a high frequency signal and therefore could potentially be affected by RFI.

Part of the problem here is that reasons can't always be given for why product (a) sounds better than product (b).  It took some time for science to be able to explain why multi-stranded speaker cables sounded better than the simple bellwire that preceeded it.

I was more thinking about the RFI from the ethernet cable entering the streamer and affecting its analogue circuits (where there IS "music" as even RaceTripper will agree, I'm sure )

But it's important to remember that those "0s" and "1s" in a cable are just voltage swings. They are not discrete, it's actually an analogue signal if you look at it on an oscilloscope. Now the DAC has to observe that voltage rise and decide where the "0" ends and "1" starts. If its guess is a bit off you get timing errors. Otherwise known as jitter.

I have never heard even the tiniest difference in audio between different ethernet cables (and I have swapped them around for various reasons). I bought a 100' CAT6 cable from Monoprice for $14. It has never dropped audio and it runs past electrical outlets, irons, lamps, record cleaning machine, and other stuff. This one works and I do not believe for a second any other cable will sound better. We are not talking about speakers cables and interconnects that carry digital or analog audio signals. 

I do not believe there is a scientific reason for ethernet cables to sound different for streamed audio when using TCP/IP as a transport for data. If you can demonstrate to me there is then I will be open to changing my POV. A scientific paper or a white paper that is not produced by anyone with commercial interest in ethernet cables will do.

 A simple Cat 5E cable will run 100Mbit/s, and 1Gbit/s which is more then enogh for DSD, 32/384, or any other audio signal you can imagine. If you want, go for Cat 6A, and you'll be able to run 10Gbit/s one day. This has no influence on the audio (it has more then enough bandwidth as-is), but if you're cabling the office... it's nice to do. 

 Now, streaming is TCP over IP over Ethernet. It's buffered, and has enough time for retransmission if needed. It carries no clock. While IP networks do suffer from jitter, and network engineers running metro-wide links do care deeply about it you will not have ANY problems with properly done LAN. Ever. The only thing that might be a problem network-wise, are loose connections. 

 Audio wise, I could imagine ground loops, and EMI. These might be a problem, but I've never saw them as such. Use plastic connected (isolated) patch cords on equipment end, and ground them on the patch panel (as you should) it you worry about this. 

>>Like I said, the ethernet cable either works or it doesn't work. There is no "audiophile" ethernet cable. That's snake oil.

Ethernet cables carry high frequency analogue signals. They do vary the reliability of the analogue  encoded digital signal. Therefore they operate across a range of conditions with gradual degradation - its definitely not a case of either they do or do not work. This operation is affected by electrical noise, cross talk, length/loss etc, but note the home environment is typically benign here.

Ethernet cables effectively convey frames - which incorporate packets - but it is the frame that is discarded if there is an error.

However as the media audio that we use in the applications we discuss on this forum use TCP, any issues are managed by the transport layer packets in conjunction with the peer protocol machines, and the associated frames are resent if required, However I suspect in the home environment you are more likely to lose a frame/packet in the host TCP/IP software stack rather than in the cable itself.

However NICs and attached Ethernet cables can be inherently electrically noisy, and this noise will couple and radiate. I posted earlier on this forum a Texas Instrument design guide for mitigating Ethernet noise coupling through cable design and interfacing. This is at least one of the interactions many can  'hear' with Ethernet cables - we hear the interaction of the electrical noise on our system. And the higher the Slew Rate of our system the more sensitive it will be to RFI.

Finally I agree there is a lot of mumbo jumbo / snake oil spoken about audiophile Ethernet leads which insults those of us who understand and who do or have designed around these issues using appropriate sound engineering techniques to mitigate EMI. A well designed Ethernet lead using shielded pairs with a quality twisting with low cross talk and a quality interface (one of more important aspects) connected to a low noise switch will improve on EMI and have less interaction with connected audio equipment. These cables don't need to be expensive.

Simon

Thank you Simon. I appreciate your voice of reason and common sense. 

I am certain simon will agree any cat5e competently made from a regular supplier will do the job, at around 60 quid for 320 metres.

I should imagine the IT community must be pissing themselves at the audiophile community.