Can fiber improve vs lan cable

Theope, UDP and TCP are very different TCP/IP transport protocols. TCP is a connection oriented transport protocol that can recover lost or corrupted data that occur anywhere between the peer session stacks and UDP is connectionless that in itself is fire and forget and has no way of recovering lost or corrupted data. There is also RUDP but I have never seen that used in home audio so we will ignore.

Now the key aspect in the above is the latency of the stream - if the relationship between latency (delay), bandwidth and segment memory permits TCP can recover lost data in the stream. This can be seamless if the resent packet is received before the segment buffer is exhausted, if not there is a loss of data and interruption (audio drop out or video pixelation). If the link has relatively high latency between peers then it will be less robust at recovering lost data for a given throughput and segment buffer size. When the receive TCP segment buffers are full the correctly received segment buffer is assembled and is passed up to the application. This is why there is a delay or latency.

With UDP as it connectionless and there is no confirmation - the UDP datagram pass across and up the to the application stack - bypassing windows and receive segment buffers. UDP is for low latency but at the cost of robustness or reliability. In our home audio cloud streaming and UPnP, TCP, not UDP is used for media transfers. UDP is used for signalling and discovery
 where is loss of a datagram is not a big issue - other than a small performance hiccup.

You mention jitter to the signal - jitter is not meaningful in TCP as the flow can be very un even and even can be un ordered.
UDP can have jitter at the datagram level - where the UDPs datagrams are sent with precise timing - as per carrier grade Voice over IP. But the jitter is for blocks of data rather than sample values.

If you are thinking of sample based jitter - then this is meaningless in these examples, as essentially with TCP a file is being transferred, and with UDP mini file components are being transferred.

Jitter is meaningful when you have a data stream where one of the values sent is time and the other is the sample value. However as the network protocols we are discussing transfer media files or parts of files this is a different paradigm and jitter in this context is meaningless.

Now finally - with respect to the reliability of networks as per my post you referenced
 ethernet including home ethernet is inherently reliable, however host protocol stacks are less so as they are dependent on CPU, I/O and other local system type resources. in some scenarios these can fail to send or receive data when very busy - and so TCP steps in and attempts the recovery of data that was lost by the peer due to temporary lack of resources or bandwidth. This is a far more common use of TCP data recovery in home networks than data corruption or loss on ethernet in my experience.
With internet access it can be more balanced between network and receiving network stack
 but TCP is busier on internet accesses where there is higher round trip latency.

Hello Simon-in-Suffolk,

Thanks for the very thorough reply on my post. I appreciate that very much.

What I’am trying to understand is why listeners experience such a great difference in SQ when they are tinkering with their cables, switches, routers and so on. I take their experiences seriously and believe what they are writing. No hokes pokes on psycho-sound effects etc. That is the part I don’t believe, or take seriously. Especially the Snake-Oil Reply’s, (abbreviated as SOR’s). I think it is a cheap and pointless way of communicating with each other. Our senses work together, so you cannot isolate hearing from seeing etc. It al affects each other, I know, but SOR’s, not done, as far as I am concerned.

Question: The log which you made shows only zeros. How much data, in the end, is really recovered?

When I introduced a switch in my home network the SQ improvement was dramatically. I came across an expression: “The last mile of the Internet”. It stipulates that the internet is very robust till the front door of the end-users. Users connect everything plug and play without any consideration about the consequences. 30 years ago, with every customer, I gave the advice: Buy a Switch. Those days they where very expensive. It was my job, for God sake!! Over the years, changes of jobs etc., I forgot the importance of a decent layout of my own home-network, and plug and played around like everyone else.

I came across that problem by some speed issues on my internet connection. I called for help, and the lady on the other end of the line just, very silent, whispers: I see a 10 Mb connection on your router. OkĂ©, I replied: what’s the problem? Oh, no nothing, its okĂ©, no problems. But her remark worried me. Till the moment I realised that a ISP router cannot work as a switch, (you fool . . . . . .)

That is, imho, a part of the last mile of the internet problem. My suggestion: The ISP should standard provided, besides a Router, a Switch. I think it will prevent us, and them, for a lot of issues.

With the dramatically impovement of the SQ, I believe that package losses occur a lot in that last mile. And that the Quality of Service of a Switch is very important for our hobby and so for a better SQ.

Again, thanks for your informative and extensive reply, and sharing your knowledge!!

Enjoy listening,

Theopa

1 Like

Hi, much of the difference in SQ that one hears through different electronic/network devices/cables is through coupled analogue noise of various types. It’s independent of digital data protocols etc.

As far as final mike
that is a term used to describe local network access technology from an exchange, POP or aggregation point. It is a network physical or link access and is inherently reliable and does not cause issues or problems unless there is a fault.

Data networking via physical links is inherently reliable at the link or transport layer
 I think you might be confusing that with the effects of analogue noise coupling such as the effects of RF interference and ground plane modulation etc which can often be audible.
Remember networking technology uses as its foundation high frequency analogue voltage and current modulation and it is the coupled side effects of these high frequency modulations that you are probably noticing/hearing. Hence in engineering design there is often the importance of decoupling, but decoupling is seldom if ever perfect.

I’m doing something similar to that at the moment: router-ethernet-Isolator-swithch1-fiber-switch2-etherrnet-streamer. I have a second fiber into swithch2 from my Roon NUC. Kit is Acousensce, Cisco, Melco, Sonore. The fiber is damped to stop any acoustic vibration effecting it. You also need to consider fiber cable lenght length and frequency.

1 Like

And how does that sound compared to a simpler copper Ethernet connection?

Because everyone system and hearing is different I guess any comments will be purely academic. All I would say is that it sounds no worse that an ordinary ethernet cable but you would really have to try it for yourself to see if it suits your system and ears. If nothing else you’ll get a degree of lightening protection perhaps.:zap:

I suppose that one reason a fibre optic link might sound worse than an all copper ethernet chain is because the signal has to be processed/converted to optical format and back again.

Hi Simon-in-Suffolk,
Before it is getting semantiek, I don’t believe that the streaming-Internet is perfect. And I was supprised by the fact, that adding a noisy power Supply, (from that switch), in the audiochain boosted up the SQ. Its is not logical, and makes no sense at all.

So I believe that a stream, which at a certain moment has the be realised in our time domain, causes lots of faults behind the front door. A Switch is far more capable of handling streaming data than a (ISP) router. Thats my explaination of the increasing SQ with a Switch, despite the added noise from its power supply.
Besides that: TCP CAN be used for data-recovery, but it is not mandatory to do so. Roon uses it for communication with their Roon Ready devices, not for recovering purposes. And it all takes place behind that same front door
 so there is indeed a lot going on there, in that last mile.
Another tought: Games, interactive. TCP. I cannot imagine how to recover in such a situation data. And why
 you shoot, and the moment is gone
 just like streaming-audio.

TCP in the streaming world is used for communication (interactive) between users/endpoints.

And on the way trough that Internet, frames are lost, and in that last mile it only can go wrong, or with some tinkering, less wrong:-)

Thanks again for your input.

Enjoy listening,
Theopa

Only a question, but I’m suspicious of the noise potential from the SMPS’s that power the convertors. With the consumer grade kits on the market I doubt the supplied SMPS’s are much other than basic. So whilst the fibre link itself may be ‘quieter’, do the SMPS’s add more than the fibre removes ???

1 Like

That’s already happened to my streaming services as I’m on fibre internet access not copper. They sound the same to me with the added benifit of no dropouts and a high speed.

1 Like

There are lots of linear power supplies that you can try as an alternative to SMPS supplied with may of the optical modules. Worth trying to see if it makes a difference in your system. Get on appro or buy- dislike- fleabay.
Like all the debates about hifi tinkering, you have to try it for your self.

Thanks Peter, I was just posing the question, maybe a bit of mischief.
I have no intention of going fibre until Naim (or another streamer) & NAS, includes a fibre connection.

Yes I know, like you I come here most days :scream: There are already fibre equipped streamers out there, just not a Niam one. Perhaps fibre, AES/EBU, I2 S-H and USB ports on any new Naim stand-alone DAC :sunglasses:

When I can get fibre to my house, then it will be tempting to connect to a streamer that takes a fibre cable such as an opticalRendu - or a Naim streamer with optical input as I’m sure will exist by then.

I wouldn’t hold my breathe. But you can get an ND-level streamer already, that has fiber input, in the opticalRendu. And yes, power matters - in theory it’s best practice whether SMPS or LPS to power the FMC’s from a different circuit than the streamer or other front end gear (and espthe receiving FMC from the sending FMC).

But even then it might not matter - in the office I go SFP out on the main Cisco switch to a 10Gtek FMC to the UQ via BJC Cat6a. The FMC is powered with an iFi SMPS into the same socket as the UQ and it sounds just that much better than straight ethernet - subtle but there, and since I already had the FMC and iFi, why not? Plenty of theories out there why fiber might actually be worse, but one won’t know until one tries it for themselves in their system (and if implemented properly, it’s considered almost always better).

1 Like

I’m not sure I could tell you it was better just that I prefer it. But direct fibre to the house at least means 4k films down load in moments not hours which keeps other people away from the hifi and I can stream Radio Paradise FLAC stream at the same time with no dropouts.

Fibre shouldn’t necessarily sound worse than anything else
the variables that affect twisted pair affect fibre too.
The only thing that you definitely don’t have with direct fibre connection is common mode electrical noise, but you have everything else.
The only definite downside of fibre over twisted pair is mechanical/sound pressure induced serial encoding clock jitter. Of course in surveillance applications this is a key benefit
 but not generally with home audio :grinning:

So fibre and twisted pair have pros and cons
 you just need to decide the balance that works for you


I personally don’t recommend direct fibre for audio connections where they will be exposed to relatively high sound pressure.
Exposing fibre to a noisy environment is kind of like running from a low/average quality clock in a switch.
Clearly if there was a soundproofed fibre cable trunking then it would be different.

1 Like

Well that wouldn’t be our place then!

The 912? Borrowed - I havent been near valves for so many years. I only used the line-amp part but I found it very fast, very musical but not necessarily in a Naim way. I have since tried EAR 568L + EAR 534 (with efficient Devore 3XL speakers) which I liked more than I thought I would - but I need more time on this. But I certainly think the 912 + Klimax Twin certainly was something I could live with.

1 Like

After spending a bit too much time up a tower looking at this cabinet for the last few days, I keep wondering what this little Cisco switch would sound like. :roll_eyes:

I need a holiday


1 Like