Audio Network Config Discussion Continued

Posted by: Huge on 18 August 2014

Originally Posted by Simon-in-Suffolk:
Huge
 
My comments
 

L2 Unmanaged - passes Broadcast data packets to all active ports, but 'learns' to pass data packets that specify an endpoint only to the port where that endpoint is connected.

 

The switch send the broadcast frames out of all active ports on the same network or VLAN . If a switch does not support VLANs it will send the broadcast frames out of all active ports.

 

A switch learns and constantly checks the association of Layer 2 network addresses to specific ports so it knows which ports to send the frames to. A broadcast frame has a specific broadcast destination address which is understood as a special address by the switch.

 

For many consumer switches multicast frames are handled the same way as broadcast frames

 

A switch is a layer 2 device and handles Ethernet frames and not packets.

 

L3 Managed - allows the network admin to specify which classes of data packet (selectable using both Layer 2 and Layer 3 characteristics) are passed to which port.  Is the programming done by port, by endpoint ID or by either?

 

(I know L2 Managed also exist, but for this use they seem to be a less good compromise - they still need programming, but don't offer identification of packets on Layer 3 criteria).

 

 

Layer 3 refers to the TCP/IP transport layer and works with data packets. A L3 packet is encapsulated within a L2 frame.

A layer 3 function includes routing. Routing allows the flow of  data packets  from one network to another - such as your internet access network to your home LAN or between VLANs on your home LAN.

 

Packets and frames have source and destination addresses. Packets have source and destination IP addresses and frames have source and destination network or MAC addresses. A network or MAC address is only deliverable within the same network or subnet. For the data to pass between networks or subnets the source and destination IP addresses are used and the router handles this function.

Essentially the layer 2 network address is used to deliver the frame to the router and then the router looks at the IP addresses to see where to send the packet.

 

A 'Layer 3 switch' is a combined switch and router.

 

The term of managed and unmanaged largely doesn't define the functionality of the switch but typically refers to the ability to 'manage' the configuration ie bespoke its operation in some way - I think what you refer to as programming. The programming can be applied to many different aspects of the network, frames and packets - and is subject to the capability of the switch

 

I hope that helps

 

Simon

 

Thanks Simon,

 

Most of my errors were in terminology rather than understanding.

 

Most of my work was in layer 7 (but also specifying which L6 and L5 protocols to be used), I learnt just enough of the other layers to be able to explain to the experts what I needed the network design to achieve, and to understand and confirm what they designed.  This is easier face-to-face than via forum posts, so please bear with me.

 

Last thing I did was a desktop virtualisation project in a legally regulated international environment.  The VLAN design for this gets, well 'interesting'; particularly trying to explain what constitutes 'local'.

 

 

I had missed the point about a router being required to cross subnets (which is why of course it's called a router - d'oh!).  I also assumed that L3 managed capability was rather more defined.

 

I did realise that cheapo devices usually treat broadcast and multicast the same way.

 

 

Thanks again I will no doubt have more questions.

The diagrams and content for the guide site are now fleshed out, I'll get them wrapped in the http design layer when my friend can do so.  Then I'll post a uri here so you can review it.

Posted on: 27 August 2014 by Mr Frog

Thanks Simon and Mike - as ever, fantastic wealth of knowledge and advice from both you guys 

I think I'll put loose coils on the Ethernet cables and use cable ties to keep the Ferrites in place

Posted on: 27 August 2014 by Mike-B
Originally Posted by Mr Frog:

I think I'll put loose coils on the Ethernet cables and use cable ties to keep the Ferrites in place

Be careful - when you bend an Ethernet cable too much it distorts the very specific cable twist of the individual pairs in the jacket & can cause performance degradation.

Ethernet cables have specified bend radius which describes how much they can be bent before data transmission is impaired. I advise to look up the cable mnft data sheet on this before you go ahead.

Personally I would not do this to my Ethernet if I had UTP,  but especially so as I have STP.

Posted on: 27 August 2014 by Huge

In terms of ferrite material, there are a lot of different mixes from different companies, and they're often slightly different, also there's no standard numeric description.

 

For instance:

TDK's older PC30 mix had very high initial permeability, but is a specialist LF mix

TDK's general purpose 30 mix has a fairly low initial permeability, but works to higher frequencies

Fair-Rite's 31 has higher initial permeability, but is a bit more limited at very higher frequencies.

Steward's 28 mix is between the two.

 

 

So, in short:

 

All the general purpose wideband mixes will do the job, the other mixes tend to be limited to specialist components, so from 'general' suppliers you'll usually find the general purpose wideband mixes, and they're fine for the job, as Simon says.

Posted on: 28 August 2014 by Mr Frog
Originally Posted by Mike-B:
Originally Posted by Mr Frog:

I think I'll put loose coils on the Ethernet cables and use cable ties to keep the Ferrites in place

Be careful - when you bend an Ethernet cable too much it distorts the very specific cable twist of the individual pairs in the jacket & can cause performance degradation.

Ethernet cables have specified bend radius which describes how much they can be bent before data transmission is impaired. I advise to look up the cable mnft data sheet on this before you go ahead.

Personally I would not do this to my Ethernet if I had UTP,  but especially so as I have STP.

Hi Mike and other Network experts .........

That's interesting regarding bends in Ethernet patch cords etc.

I'm currently using some 'Freebie' Cat5e cables that came with various free routers over the years from my ISP (Virgin Media). Without exception, the Ethernet cables were always tightly folded in the box and packaged with the relevant router. I'm pretty sure that this was also the case when I bought the NAS unit.

That being the case, do I need to bin all the cables and start fresh with new ones or do you reckon that they should be okay?

On a similar note I recently saw some online guides from Canon and Ricoh regarding Ferrites on the Ethernet cables for their printers etc. It showed a Ferrite fitted next to the Ethernet connector plug and another about 10cm along the cable which was actually wound once or twice around the Ferrite ..... which would clearly be bending the cable and on the face of it knackering the cables.

I'm a little confused to be honest re Ethernet. DC cables for SMPS can be wound several times around Ferrites but are we saying the Ethernets can't be? If that's the case, I'm a bit perplexed as to why the cables are already supplied tightly folded from some manufacturers and that some big companies are advocating wrapping/winding them around Ferrites

Sorry to be a pain, it's just that I've spent days laying Network cables in the loft and through walls to get the Network router and NAS away from the HiFi and I simply don't want to bugger it all up by making a mess with Ferrites .... bending cables where I shouldn't, if it will deteriorate the sound quality.

Hope someone can give a definitive response and guidance on this issue. Thank's in advance 

Posted on: 28 August 2014 by Mike-B

Whoa their Mr Frog,  you've misunderstood or I've frightened the Frog pond

Stick to the mnft recommended bend radius

If you can't find that, the industry rule of thumb for UTP is 4x the diameter  so if its 5mm diameter the bend radius must not be less than 20mm. (40mm diameter)

I have an industry paper that puts a good argument for double the x4 minimum radius for for UTP, so based on that if you can set a x8 bend as you go around the ferrite, IMO that would be better. 

 

I doubt your tightly folded Cat5 will be damaged, Cat5 is not that critical as it does not have such a complicated construction & is less liable to be distortion damaged; especially so as freebee cables are invariably flexible wires inside the cable. 

I would be more concerned with Cat6 which is more precisely & carefully manufactured

Doubly so with Cat6 STP & even more so with Cat7 which has solid core wires

 

If you want to have a double pass cable thru a ferrite,  OK go ahead

I would be happy enough to do that with Cat5 with the correct radius,  I would increase the diameter a little bit more with Cat6

But I would be wary with STP & not consider it with my own Cat7

Posted on: 28 August 2014 by Huge

+1 to Mike's answer.

 

Just to add, all the Ethernet cables I've ever had, have been supplied coiled up in a circle 100-150mm diameter.

Posted on: 04 October 2014 by bicela

I've discovered this thread only now, I would write thank you to Huge and Simon and all the others to have shared these informations with new forum users (like me).
Grazie! I'm going to buy TDK ferrites :-)

Posted on: 04 October 2014 by Huge
Originally Posted by bicela:

I've discovered this thread only now, I would write thank you to Huge and Simon and all the others to have shared these informations with new forum users (like me).
Grazie! I'm going to buy TDK ferrites :-)

Hi, you need not be too concerned about the specific manufacturer of the ferrites, just go for the general purpose ones from which ever supplier or manufacturer you choose.

 

I use generic types (from Amazon / eBay) and they work as well as TDK (in fact in my circumstances I have more lower RF interference and the TDK ones are less good here).