Ethernet Cables
Posted by: tyler on 03 March 2014
I need cables for my office, which cables are best for use?
Anything that lowers the noise floor results in more music... For example 24 bit audio implies 120dB resolution, or 1 microvolt against a 1 Volt reference signal. It wouldn't take many microvolts of noise to reach the verge of audible impact, which explains why Naim have been so obsessive around electrical noise within the NDS.
Cables, power supplies, grounds, cable routing ... everything helps.
Much as I believe in the virtues of good ethernet cables (my last run to ND5XS is AudioQuest Vodka), I can't agree that an ethernet cable's SNR has anything to do with "more music", because there is no music flowing in that cable, just ethernet packets 
The only explanation so far why some ethernet cables sound better than others has to do with the amount of work the receiving circuit has to do to process those packets.
It's a similar phenomenon to why many listeners report that FLACs sound better when transcoded to WAVs at a server level - less decoding work for the streamer's CPU.
At least I haven't heard of any other sensible explanation...
Gajdin, if you can't believe SNR has an effect on the cable performance - end result is music ....... this might be on interesting short read for you.
It might also explain why your cable experiences were as they were.
If not, well its an interesting read anyhow
Mike, as a former recording engineer I know what S/N is very, very, very well
This (or the article you quote) in no way explains why a lower noise level in a UTP cable would affect the quality of music encoded in the packets of the ethernet transmission in that cable.
There are, however, other phenomena that could explain why different ethernet cables "sound" different, and I mentioned them in my post above.
True, but who said FLACs are the only things flying through a home network Mine carries data to and from 6 computers, a NAS, 2 TVs and 3 audio streamers... Oh, yes, and a 250mbps internet connection.
True, but who said FLACs are the only things flying through a home network Mine carries data to and from 6 computers, a NAS, 2 TVs and 3 audio streamers... Oh, yes, and a 250mbps internet connection.
If you feel you need bandwidth for different things then that of course is a different matter entirely to what is needed for audio.
If, one day, all your TV's and streamers were active at once you might used as much as 2.19 MB/s:
3 streamers = 3 x 0.13 MB/s = 0.39 MB/s
2 TV’s = 2 x 0.9 MB/s = 1.8 MB/s
This would leave about 80% of the throughput on a simple 10 MB/s network redundant. If those 6 computers are employed in a grid computing effort, this may present an issue.
3 streamers = 3 x 0.13 MB/s = 0.39 MB/s
2 TV’s = 2 x 0.9 MB/s = 1.8 MB/s
This would leave about 80% of the throughput on a simple 10 MB/s network redundant. If those 6 computers are employed in a grid computing effort, this may present an issue.
It's enough if 4 of my 6 computers are downloading something from the Internet at 250mbps and my gigabit ethernet is filled
I do understand the feeling, I missed the physical contact with the disk at first, especially in case of Naim where the swinging drawer and the puck add a bit of a ceremony and respect for the record, almost like cueing that needle...
But once I got this whole streaming setup to work I actually feel closer to the music, because it's so much easier and faster to switch between the records and appreciate what's similar and different between artists (and I mean all artists, not only those who compose and play, but also the ones who play the instrument called a recording studio).3 streamers = 3 x 0.13 MB/s = 0.39 MB/s
2 TV’s = 2 x 0.9 MB/s = 1.8 MB/s
This would leave about 80% of the throughput on a simple 10 MB/s network redundant. If those 6 computers are employed in a grid computing effort, this may present an issue.
It's enough if 4 of my 6 computers are downloading something from the Internet at 250mbps and my gigabit ethernet is filled
This is incorrect. You have a 250mbps connection as you said above. Your 4 computers have to share that connection. They don't each get it. Therefore the internet speed remains slower than the gigabit ethernet you have.
Furthermore, in the [unlikely] situation where you have....
- 3 x audio streamers all streaming FLAC's at once
- 2 x TV's both streaming HD at once
- 4 x computers downloading from the internet at max speed - with no management on your part
......Cat 5e and a gigabit network will be sufficient.
If you consider thoughput in simple terms, as we have, along with the logical and physical decoupling inherent in your network, it helps define what, if anything, merits further consideration and investment.
It's enough if 4 of my 6 computers are downloading something from the Internet at 250mbps and my gigabit ethernet is filled
This is incorrect. You have a 250mbps connection as you said above. Your 4 computers have to share that connection. They don't each get it. Therefore the internet speed remains slower than the gigabit ethernet you have.
Furthermore, in the [unlikely] situation where you have....
- 3 x audio streamers all streaming FLAC's at once
- 2 x TV's both streaming HD at once
- 4 x computers downloading from the internet at max speed - with no management on your part
......Cat 5e and a gigabit network will be sufficient.
Gosh, I wasn't thinking. Of course the internet connection is 250mbps total for all machines.
However, it's enough if I'm copying files from one of the PCs to the NAS, Windows reports that on large files I get ca. 90MB/s, that's 720mbps (probably more with the control bits etc.) If my son tried to do the same at the same time from his computer, we would bump against the gigabit network bandwidth limitation. So just 2 of my 6 computers can fill in the available bandwidth!
And I copy large files a lot - like when I come from downstairs with my band's demo recordings, that's 24/88.2 multiplied by 14 tracks, so 14 large WAVs per song, multiplied by several takes...
Gajdzin, reading your post, can I just suggest that when using a switch on an Ethernet LAN, the port speed defines the maximum throughput on that port (this is always a power of 10, ie 10, 100, 1,000 or 10,000 Mbps ). So if you and your son are transferring from different PCs to the same NAS on the same port, then the NAS port speed becomes the limiting factor, such as 1000Mbps.
if however you and your son transfer data to different devices on different ports on the same switch then you should not impact each other which means you could both transfer at 1000 Mbps . Yes ultimately switches have what is known as a back plane bandwidth, which is the bandwidth of all the current switched traffic. This is typically a percentage of the number of switch ports times the maximum port speed. In the home LAN I doubt this would be a limiting factor unless it's a very poor quality switch.
Simon
Yes, that makes sense, but what I was trying to show is that it's not difficult to completely fill the bandwidth of a gigabit ethernet line. I agree that if we are both using NAS it's the NAS port throughput that would be the limiting factor, but imagine a different scenario: I copy a terabyte of music data from my recording laptop to my mixing desktop, while my son copies movies from the NAS to his laptop in his room, for watching them when we think he's asleep. If both transmissions go through the same cable between the study (where the desktop and the NAS are) and the other rooms, that single ethernet cable won't allow maximum speed for both transfers at once, even if the switch was a Cisco 9000 series monster with 10 gigabite ports.
It's just that Mark was trying to make a case that we only use a fraction of a 1000baseT cable's throughput and are unlikely to need anything faster at home, while I'm trying to show that it's very easy to use up its entire bandwidth with just a few ordinary home computers. Which is why I hope one day 10GBASE-T will enter home computers and switches, and which is why I recommend anyone wiring their house to go for Cat6, as Cat5 doesn't do 10gbps.
PS. these are not entirely theoretical situations - occasionally I copy big WAVs from the laptop to the desktop and Windows reports only, say 55MB/S instead of 80-90. When I go to check what the family is up to, surely enough I find my wife copying a movie from the NAS to the TV PC's desktop, so that she can watch it at night without waking the NAS (while her MacBook is backing up data to NAS, to make matters worse). Voila - the main home cable is full. Or a port in a switch, it doesn't really matter which, because both scenarios show that gigabit ethernet may not be enough in normal, home use cases.
Gajdzin, reading your post, can I just suggest that when using a switch on an Ethernet LAN, the port speed defines the maximum throughput on that port (this is always a power of 10, ie 10, 100, 1,000 or 10,000 Mbps ). So if you and your son are transferring from different PCs to the same NAS on the same port, then the NAS port speed becomes the limiting factor, such as 1000Mbps.
if however you and your son transfer data to different devices on different ports on the same switch then you should not impact each other which means you could both transfer at 1000 Mbps . Yes ultimately switches have what is known as a back plane bandwidth, which is the bandwidth of all the current switched traffic. This is typically a percentage of the number of switch ports times the maximum port speed. In the home LAN I doubt this would be a limiting factor unless it's a very poor quality switch.
Simon
Yes, that makes sense, but what I was trying to show is that it's not difficult to completely fill the bandwidth of a gigabit ethernet line. I agree that if we are both using NAS it's the NAS port throughput that would be the limiting factor, but imagine a different scenario: I copy a terabyte of music data from my recording laptop to my mixing desktop, while my son copies movies from the NAS to his laptop in his room, for watching them when we think he's asleep. If both transmissions go through the same cable between the study (where the desktop and the NAS are) and the other rooms, that single ethernet cable won't allow maximum speed for both transfers at once, even if the switch was a Cisco 9000 series monster with 10 gigabite ports.
It's just that Mark was trying to make a case that we only use a fraction of a 1000baseT cable's throughput and are unlikely to need anything faster at home, while I'm trying to show that it's very easy to use up its entire bandwidth with just a few ordinary home computers. Which is why I hope one day 10GBASE-T will enter home computers and switches, and which is why I recommend anyone wiring their house to go for Cat6, as Cat5 doesn't do 10gbps.
PS. these are not entirely theoretical situations - occasionally I copy big WAVs from the laptop to the desktop and Windows reports only, say 55MB/S instead of 80-90. When I go to check what the family is up to, surely enough I find my wife copying a movie from the NAS to the TV PC's desktop, so that she can watch it at night without waking the NAS (while her MacBook is backing up data to NAS, to make matters worse). Voila - the main home cable is full. Or a port in a switch, it doesn't really matter which, because both scenarios show that gigabit ethernet may not be enough in normal, home use cases.
Your 3 x audio devices and 2 x TV's, all streaming at once, use minimal throughput.
The additional scenario you outline, multiple computers all copying terabytes of music or film all at once from one to another at the same time is not representative of most home networks and best managed by a switch as Simon said.
Regarding the back-up mentioned, perhaps consider an incremental back-up? This will reduce load to almost nothing. Unless, that is, you're making multiple redundant back-ups of the terabytes of music and film which are simultaneously copied everywhere! Defeats the object of the NAS. Better software and network management might be the simplest solution.
Finally, your example of copying a WAV and getting 55 MB/S instead of 80 MB/S suggests this process may take about 2 seconds rather than about 1.5 seconds. When copying a whole album you may lose as much as 5 seconds. Food for thought?
I do understand the feeling, I missed the physical contact with the disk at first, especially in case of Naim where the swinging drawer and the puck add a bit of a ceremony and respect for the record, almost like cueing that needle...
But once I got this whole streaming setup to work I actually feel closer to the music, because it's so much easier and faster to switch between the records and appreciate what's similar and different between artists (and I mean all artists, not only those who compose and play, but also the ones who play the instrument called a recording studio).In fact, i feel very confortable with the dematerial system ;-) My point was onthe renderof my system, since i have the NAS, i feel like a change. I already invest for a Cinnamon and i have to say that it definitely works! It is even better than before the NAS arrived. Very worth the price.
Important to notice, these cables do not change the "Naim sound" ! great product !
I do find this discussion interesting, as I'm working on a prototype for a much cheaper design of audiophile ethernet cable.
Incidentally, from my network stats, no matter whether I use Cat5e or Cat6a, I currently have zero packet loss, so I'm not seeing to increase the integrity of the electrical signal transmission!
Hi Hugh, I'll be happy to be your beta tester
I think you should consider more physical factors that may affect sound quality of the network streamer, because until we know exactly WHAT is causing the differences, we have to consider several possibilities. Packet loss is only one. RFI/EMI is another...
Gajdzin - assuming the shape and integrity of the signal on the differential twists affects the 'work' and timing undertaken by the receiver - and this in turn produces audible side effects either directly or indirectly - then cross talk and cable dielectric of the twisted pairs has an impact. Also the design and characteristic impedance of the connecting plug has an impact.
Some interesting design formula here
Good point, Simon. I would kind of consolidate these things into one big category, for which you will no doubt find a better and more technically sensible name - in the interim I'll call it "everything that makes the streamer work harder to convert the ethernet signal into correctly tied 0s and 1s".
That category could contain many components, probably well described in relevant literature on ethernet transmission... hopefully. There's no saying we won't find an undescribed components, just like jitter was not something that early digital audio designers were much aware of, until the audiophile community convinced them that digital sources DO sound different, which made them go back to the lab and start measuring things they never thought relevant enough to measure before...
There are other categories, though, as you very correctly pointed out in the past in other threads:
- EMI/RFI
- Physical vibration
So we are down to THREE categories. Unless somebody here can think of a fourth one