Small basic ethernet switches and grounding/shield

[numberwrong]

Hi,

 

Another ethernet question....

 

So most of the CAT5/6 cable running down a multicore I've ever used is shielded, which makes sense as you've normally got a long run of power running next to it so you'd want that ethernet cable to be protected with a shield.

 

I was wondering if a basic network switch would make the ground redundant? The really cheap domestic ones have plastic outsides to the RJ45 connectors like this (better ones are metal like this) so, I'm assuming these domestic switches are removing the shield protection all together? or is this metal housing just for looks/durability?

Lets say you've got a metal one, most of them use a 12v DC power supply so there is no physical grounding to earth going on. Do you need to use shielded patch cables so the shield eventually connects to a device that is earthed (ie a lighting desk) for the shielding to work or is the shielding dealt with in the switch/devise and nothing to do with mains grounding?

If your multicore runs from one switch to another do both switches need to be able to 'do' shielding or can just one end deal with it?

Cheers

[mackerr]

Hi,

 

Another ethernet question....

 

So most of the CAT5/6 cable running down a multicore I've ever used is shielded, which makes sense as you've normally got a long run of power running next to it so you'd want that ethernet cable to be protected with a shield.

 

Shielding does not keep induced interference from power cables out of audio or data cables, it protects against RF interference either getting into audio cables, or out of data cables. Protection from induced interference from power cables comes from the twisting together of the signal lines. Category 5 & 6 cables are very effective at avoiding electrically induced interference.

 

Mac

[richardash1981]

I was wondering if a basic network switch would make the ground redundant?

Just because it isn't grounded, it doesn't follow that a screen is worthless. The screen still ensures a uniform electric field inside it (being a conductive cylinder), so all the cores get the same signals by capacitive coupling. This means differential pairs work and you still get a useful improvement over an unscreened cable. Of course, earthing it is better from a shielding perspective.

The really cheap domestic ones have plastic outsides to the RJ45 connectors like this (better ones are metal like this) so, I'm assuming these domestic switches are removing the shield protection all together? or is this metal housing just for looks/durability?

I've not seen the switch with plastic RJ45s on before - even the cheap ones I have got have a metal contact inside the socket to connect to the sides of the (screened) RJ45 plug and provide ground continuity. If there isn't a continuous metal screen on the socket then higher RF frequencies (think mobile phones) will leak through the gaps, hence the use of all-metalclad connectors for higher specification systems (I think they are mandatory for CAT6 and CAT7 cable systems).

Lets say you've got a metal one, most of them use a 12v DC power supply so there is no physical grounding to earth going on. Do you need to use shielded patch cables so the shield eventually connects to a device that is earthed (ie a lighting desk) for the shielding to work

That's generally the idea, and in industrial contexts every desktop PC and server attached is going to be earthed anyway.

If your multicore runs from one switch to another do both switches need to be able to 'do' shielding or can just one end deal with it?

It's likely to work much better if you have ground continuity everywhere you have signal continuity, because then the signals have a sensible path "home" and you minimise issues with ground loops. Long single-end-earthed cables tend to become antennas rather than screening ...

That said on a fixed rig it wouldn't worry me if the grounding was achieved using screened keystone jacks on patch panels to link data cable grounds together without going through the switch itself (still use screened patch to the switch unless it is inside a larger screened enclosure). Would seem to be asking for trouble on a gigging rig however.

[martinw]

The screening provided on most Cat5/6 cables is very limited, and can cause problems which wouldn't exist without it. Ethernet is transformer-isolated at both ends of the link, which gives great protection from ground loops, but if you use a screened cable then you're introducing the potential for a ground loop where there previously wasn't one. At speeds below 10G, there really isn't a good reason to use screened cable.

 

Martin

[numberwrong]

The really cheap domestic ones have plastic outsides to the RJ45 connectors like this (better ones are metal like this) so, I'm assuming these domestic switches are removing the shield protection all together? or is this metal housing just for looks/durability?

I've not seen the switch with plastic RJ45s on before - even the cheap ones I have got have a metal contact inside the socket to connect to the sides of the (screened) RJ45 plug and provide ground continuity. If there isn't a continuous metal screen on the socket then higher RF frequencies (think mobile phones) will leak through the gaps, hence the use of all-metalclad connectors for higher specification systems (I think they are mandatory for CAT6 and CAT7 cable systems).

 

A lot of cheap domestic switches definitely do not have any metal inside the RJ45 connector for grounding. I have both of the netgear switches linked above and the cheaper one in the 'home' range has no metal. My BT home hub also doesn't have any metal for grounding either suggesting a strip of 4x un-grounded chaises mount RJ45 sockets must be a fairly standard component used in many (domestic) switches. I know for sure I've spec'ed Ethernet switches on jobs before only to be given a domestic style switches (which incidentally worked fine). I suppose part of the point of the post it to establish if using these kind of switches should be a real no no or if it's no big deal.

 

The screening provided on most Cat5/6 cables is very limited, and can cause problems which wouldn't exist without it. Ethernet is transformer-isolated at both ends of the link, which gives great protection from ground loops, but if you use a screened cable then you're introducing the potential for a ground loop where there previously wasn't one. At speeds below 10G, there really isn't a good reason to use screened cable.

 

Martin

 

This is interesting, would you suggest that best practice would be to use un-shielded cable? or maybe a shielded cable that isn't actually grounded as this would provide some passive protection (as suggested by richardash1981) and would not be susceptible to a dirty ground?

Would it not be a useful option to be able to lift the ground on a switch designed for our industry?

I'm pretty sure all the ethernet I've seen on a multicore is shielded, whereas patch cables can be a bit of a mixed bag. needless to say, all the jobs I've ever worked on the ehternet has either worked first time, or not worked and been fixed by swapping out cables. I've never had a poor or glitchy connection (yet), just fully working or fully not working.

[bruce]

In practice, it makes pretty much no difference at “normal” network speeds. If you’re running 10 gig, it’s different story.

 

In general, grounding of the shield would be done at the patch panel rather than the end device - the assumption being that you’ve got a long infrastructure cable, with a short patch cable either end.

 

 

 

Useful article from Nexans.

 

https://www.nexans.co.uk/eservice/UK-en_GB/fileLibrary/Download_540239149/UK/files/FTP%20vs%20UTP_V5_GBA_HFO%20DWI2%20NR3_PBE_V2_1.pdf