I‘m no professional Network guy but my understanding is not based on the digital „0“ and „1“ but on the analog signal/errors which may influence the system and have also impact on timing.
Further to my post #19, all I can say is that to my ears, in my system, I have concluded that the Innuos switch is a significant improvement in all respects on the EE8. I would compare the effect to a component upgrade such as a pre-amp or even power amp. I have no idea how it works. It could be that it is simply removing the noise and interference in the chain and providing a cleaner signal to the DSM? Maybe there is a particular synergy with my Linn Exakt system. It also happens to fit very nicely in the half-width rack (I'll be ordering one in black).
Analog signals are not transported thru the network, its all digital and the network is just a physical medium to move data from one point to the other, while TCP takes care of ensuring that what was sent in a data packet gets delivered to the other end without errors. If the data packet is not acknowledged by the receiving end it will be resent. Timing issues within the streamer is another issue but it is not network related.
You may be referring to jitter and this is not caused by the network. Jitter is caused by errors while reading/writing the data packet payload and usually is caused by insufficient buffer space or performance issues while reading/writing data packets. For example, a proper 8 bit packet would look something like 0,1,1,0,1,1,0,1 but a packet with empty bits would look something like 0,1,1,,1,1,,1 (see the consecutive commas and missing values?, that is jitter) and this is not caused during data transmission.
This is an oversimplification but hope it helps you understand my point of view.
As I mentioned before I only foresee issues with severely congested networks or networks suffering massive packet loss, causing gaps in playable music. Nothing more, nothing else.
Additionally, every Ethernet interface is galvanically isolated as per the 802.3 standard (since its inception back in the early 80s) to ensure noise does not travel across the network cable, and cables are twisted in pairs to ensure noise is cancelled if any external electrical interference induces electrical noise in it, so, if any 'audio grade switch' vendor claiming galvanic isolation and 'noise cancellation pixie dust' in his brochures/ads he is just wagging the bone to the dogs.
Many of the real time streaming technologies don't use TCP. For reasons to do with real time control and better timing of media samples a bunch use UDP and some use RTSP and a few other lesser known protocols.
IMO, the switch could be injecting noise via the power lines. SMPS are notorious for that. With that in mind, a switch engineered to be less noisy might show some benefits.
@vicdiaz :
It seems that I can't explain very well what my understanding is (but maybe also because I'm not native English).
The digital signal is transmitted "analogue" via Ethernet. In addition to the digital signal, other analogue interference is also transmitted. Hardware and software/protocols are used to optimize this analog transmission in such a way that, in the end, digital information is created that is as "clean" as possible.
Hope it helps...
Maybe other forum member could explain it better, what I mean.
It seems that I can't explain very well what my understanding is (but maybe also because I'm not native English).
The digital signal is transmitted "analogue" via Ethernet. In addition to the digital signal, other analogue interference is also transmitted. Hardware and software/protocols are used to optimize this analog transmission in such a way that, in the end, digital information is created that is as "clean" as possible.
Hope it helps...
Maybe other forum member could explain it better, what I mean.
I have also been experimenting with network cables, routers and wireless configurations, following the advice on the Lejonklou forum.
Like vicdiaz, I'm an IT professional, and my job relies on those 1s and 0s traveling without loss or corruption. I don't think the data is the issue.
However, changes in the network components are audible, even in my low cost home-assembled system.
I've followed the Lejonklou recommendations for power and network connectivity, and to my ears, in my system, it is the most enjoyable configuration I've tried.
I have no idea *how* changing a network cable can affect musicality, but I've heard differences between closely specified cables from the same manufacturer. Not what I expected.
I have not spent a great deal on network components, since my source is low cost, so optimising your network does not have to be expensive, but it does seem worthwhile. It is similar in scale to my ears to optimising the support.
Like vicdiaz, I'm an IT professional, and my job relies on those 1s and 0s traveling without loss or corruption. I don't think the data is the issue.
However, changes in the network components are audible, even in my low cost home-assembled system.
I've followed the Lejonklou recommendations for power and network connectivity, and to my ears, in my system, it is the most enjoyable configuration I've tried.
I have no idea *how* changing a network cable can affect musicality, but I've heard differences between closely specified cables from the same manufacturer. Not what I expected.
I have not spent a great deal on network components, since my source is low cost, so optimising your network does not have to be expensive, but it does seem worthwhile. It is similar in scale to my ears to optimising the support.
Last edited:
This topic has been flying around Hi-Fi forums for over a decade. For every 50 people who try it we get 50 different answers. The testing evidence tends to point to little/no improvement. But this is the Linn Forum so "testing" is a dirty word. I would suggest if you are going to shell out £1,000 on a switch think of upgrading your kit by the same amount instead, at least that is logical.
Different switches and cables will emit different levels of EMI/RFI - it is measurable. All switches regenerate the signal so the final switch in the chain is the one that may send these nasties towards your streamer (they do not regenerate noise, but may add some of their own). Do these nasties have an audible effect? You can get nice quiet switches from HP, Cisco, Ubiquiti etc., You can get excellent cables from Connectix and BJC. I would avoid metal connectors and use plastic RJ45s to avoid compromising the galvanic isolation inherent in Cat 5e.
Before parting with hard earned cash on a boutique switch, I'd try something like the Cisco CBS350-8P-E-2G Switch.
However, when I listened to a boutique switch at a demo I heard no improvement. It sounded every bit as good as the Netgear it was being compared with. The only way to find out if a boutique switch is worthwhile for you is to listen in your system. It is most likely noisier than one from HP, Cisco or Ubiquiti, but you may still prefer it.
Before parting with hard earned cash on a boutique switch, I'd try something like the Cisco CBS350-8P-E-2G Switch.
However, when I listened to a boutique switch at a demo I heard no improvement. It sounded every bit as good as the Netgear it was being compared with. The only way to find out if a boutique switch is worthwhile for you is to listen in your system. It is most likely noisier than one from HP, Cisco or Ubiquiti, but you may still prefer it.
That criticism is not justified. Right from the earliest days of the LP12, Linn encouraged listening and making comparisons. The Linn dealers that I have dealt with, always offer customers the opportunity to listen to new equipment or upgrades with their own systems. In the past couple of years, they've lent me a Next Generation DSM, Klimax Organik Exaktboxes, a pair of Exakt Akudoriks and a Melco S100 Switch to try at home without obligation.
No analog or 'analogue' information is transmitted via Ethernet. It is converted to digital by an ADC before it makes it to the Ethernet interface. EMI/RFI interference is due to switch proximity to the streaming device, and that will affect sound for sure. That is why I use long cable runs and keep my networking devices away from my audio gear.
I've tried several dozens of networking switches trying to find any differences (from cheap $20.00 non-branded switches to a Juniper Data Center grade core switch) and the only difference I found was that the fans in the Juniper were a little bit noisy for my taste. I used Lejonklou's evaluation method (and I truly believe in the 'tune method') and found no difference in terms of musicality.
Unless the switch has something inherently wrong and it is a radio antenna emitting RFI/EMI and its power supply is screwing up the AC being fed to the rest of the system, I don't see or hear any advantages in spending huge amounts of money in networking gear.
I spent more than 16 years working in IT for several banks, including one of the largest Canadian banks in the world, and the only reason they spent big bucks in networking gear is because they are looking for reliability and redundancy, not packet loss because that is a given in all the switches in the market today. I've seen mainframes running core systems apps connected to a core switch using a $100.00 Netgear switch for weeks while waiting for parts to repair the mainframe switch, no one complained about it, and we are talking about millions of dollars in losses if that switch failed or started dropping packets at all.
Last edited:
For those who mentioned that the difference could be measured, here is a video comparing two 'audio grade' network switches with an off-the-shelf SOHO switch. No quantitative difference at all, except for the price.
Last edited:
- May 20, 2014
- 3,516
- 4,525
- 148
- HiFi Trade?
-
- No
This is my understanding too. Binary data is sent as a series of, IIRC, +5v pulses representing “1” and 0v representing “0”. At the receiving end of the connection, anything below a threshold, usually 2.5-3v, is classed as a “0” and anything above threshold is a “1”. This is Pulse Code Modulation at work.
‘troll
@vicdiaz : it‘s clear that Ethernet transmit digital information but it‘s far away from being „0“ and “1“ only. That is what I mean with „analog“.
We did a lot of blind tests with switches. And sometimes it‘s not easy to find out whitch switch is the best (especially in the long run), but the differences were huge, really huge!
But if you don‘t hear any differences it‘s fine. In consequence you can save a lot of money (switch, Linear Power Supply for the Switch, external Clock for the Switch, Ethernetcables etc. etc.).
Let‘s enjoy the music.
We did a lot of blind tests with switches. And sometimes it‘s not easy to find out whitch switch is the best (especially in the long run), but the differences were huge, really huge!
But if you don‘t hear any differences it‘s fine. In consequence you can save a lot of money (switch, Linear Power Supply for the Switch, external Clock for the Switch, Ethernetcables etc. etc.).
Let‘s enjoy the music.
Last edited:
Page 5 of this thread was quite interesting to me, as apparently there’s a business in selling fancy switches to Hifi brands to resell. Post #92.
https://www.hifiwigwam.com/forum/threads/do-i-really-need-an-audio-grade-network-switch.106428/
https://www.hifiwigwam.com/forum/threads/do-i-really-need-an-audio-grade-network-switch.106428/
Amir is measuring distortion in an audio context. I was referring to measuring RFI/EMI contributed by network components; something we had to for a customer requiring high security levels (TEMPEST/EMSEC).
In general, to send data between two devices, we need electricity and an encoding scheme (e.g. when we send a 1 we could send a high voltage and when we send a 0 we could send a low voltage. As long as two devices use the same encoding scheme, we can exchange data). However, we are using electricity through a wire so we'll get EMI (electromagnetic interference). Twisting cables helps cancel EMI between wire pairs (crosstalk) - please click here for an outline of measurement. I'd suggest a switch with Power over Ethernet so you do not have one of those nasty looking wall wart things around or indeed another PSU.
The reason I doubt it affects Linn DS(M)s is they contain Field Programmable Gate Arrays and the RFI given off by these wonders will dwarf anything coming from a decent switch. And yet, Linn copes with it ... DS(M)s sound very good indeed. So if Linn can do this for a device at the heart of its architecture then it can cope with a few nasties shoved up its Ethernet ports - remember, Ethernet has galvanic isolation so (in theory) doesn't pass on noise.
And if you are lucky enough to have a Next Generation KDSM you can buy a nice switch that has fibre out and connect that way and forget all about EMI/RFI - make sure you use audio grade SFPs (you can tell they're audio grade as they are an order of magnitude more expensive than standard SFPs).
At the end of a Perfect Day though all you need to do is have a listen in the comfort of your own home in your own system. Just slip on that prized copy of Argent's "God Gave Rock and Roll to You" and ask yourself "does it sound better?". if your answer is No then you're right and if your answer is Yes then you're right. There is no wrong answer and remember it's better to be happy than right [Ref: Douglas Adams].
Vic I fear you are flogging a dead horse. The world is divided into those who believe in audiophile switches, power supplies, ethernet cables etc and those who don’t. (I don’t). As far as I know no one on one side has ever been converted to the other. Those who believe can rest happy that they have improved their system, those who don‘t can rest happy that they have not spent unnecessary money on something that makes no difference.
Let it end there.
Let it end there.
- May 20, 2014
- 3,516
- 4,525
- 148
- HiFi Trade?
-
- No
Belief one way or the other doesn't help much with the original question. As always, better to listen for yourself and determine whether any such item is beneficial or not.
Similar threads
