1
\$\begingroup\$

I have a question related to shielding and grounding of ethernet cabling in industrial environments, offices and datacentres.

What I assumed as correct so far (please correct me if I’m wrong):

  1. In Consumer PCs the ethernet shield is connected to the case via the IO-Shield. The case is grounded over the PE-conductor in the PSU.
  2. On a commercial server it is the same (only had one to check).
  3. The ethernet standard specifies signal transformers at each interface to achieve a galvanic isolation between endpoints.
  4. For a cable shield to be effective (at high frequencies) and not act as an antenna, it should be grounded at least at both ends. For very long cables more ground connections may be necessary. To break ground loops this connection may be capacitive at one end.
  5. The galvanic isolation on the signal line is, I presume, there to protect the sensitive electronics from a potential voltage difference between the two endpoints. But if you use shielded ethernet cables, then you will create a low impedance connection between the two endpoints and a current will flow through the shield.

Now to my question: Industrial interfaces like EtherCat have a capacitive coupling of the shield to the case. And I assume in datacentres and the likes the grounding system is tightly controlled so no potential differences can arise. But how is this handled in office buildings? Is it just assumed that the grounding system is good enough so that there are no, or only very small, balancing currents on the shield between the two endpoints? Or are these currents just no problem in these practical applications?

\$\endgroup\$
  • 1
    \$\begingroup\$ ad 4) This is not correct. A shield is preferably connected to the earth in one end, only. \$\endgroup\$ – Marko Buršič Feb 27 at 11:49
  • \$\begingroup\$ I disagree. To shield against E-Fields at lower frequencies (l < λ/10) a single-point termination of the shield is recommended (as in audio applications). Against E-Fields at higher frequencies (l > λ/10) both ends or even multiple points along the wire should be grounded (Source: Grounds for Grounding – Elya B. Joffe, Kai-Sang Lock, Chapter 7.3.5). If grounded at only one end, the shield may act as an antenna and might increase the problems. \$\endgroup\$ – LouisRast Feb 27 at 12:21
  • 1
    \$\begingroup\$ If the network cable is shielded, the shield is definitely connected to protective earth ground at the patch bay. The work area connector end of the cable shield is typically not grounded. It is unclear to me if the work area connector provides connection to shield or not, but if it does, then the patch cable to PC should be UTP so it does not connect PC chassis ground to work area connector shield. You really do not want to connect PC chassis to earth via network cabling as that would make a current loop. Even worse if the PC is on ungrounded mains socket and only ground is via network cable. \$\endgroup\$ – Justme Feb 27 at 12:26
  • \$\begingroup\$ @Justme In fact this is also my thought. The cable is grounded at patch panel, the other end - wall socket is not grounded. You do conenct the PC with UTP patch cable, so it really is grounded in a single end. \$\endgroup\$ – Marko Buršič Feb 27 at 13:29
  • \$\begingroup\$ @MarkoBuršič Yes, but I don't know what would happen if you had STP patch cable - will the wall socket keep the shields disconnected? \$\endgroup\$ – Justme Feb 27 at 13:34
2
\$\begingroup\$

In Consumer PCs the ethernet shield is connected to the case via the IO-Shield. The case is grounded over the PE-conductor in the PSU.

And it does NOT assume shielded Ethernet cable.

For a cable shield to be effective and not act as an antenna, it should be grounded at least at both ends. For very long cables more ground connections may be necessary. To break ground loops this connection may be capacitive at one end.

Not correct. This is not only a grounding issue from the protection point of view, it is also about integrity of the signal. Current flowing through the shield will cause problems with the signal in the wires inside it - as the current source is very close and can be of considerable value. Some time ago I was researching grounding and long transmission lines (but in audio), and there were a number of recommendations on how to minimize noise in the differential signals, I clearly recall one saying that if using grounded cable it must be connected to chassis only at one side. Unfortunately can not find this document right now.

But how is this handled in office buildings?

This is too broad question: every country may have its own requirements for the electricity in the buildings, and thus the grounding circuits for the devices being connected to the mains.

I assume in datacentres and the likes the grounding system is tightly controlled so no potential differences can arise.

Ideally it must be this way. In consumer non-datacenter environment variations are possible, but you must know one thing: if there's a rule for grounding, it must be strictly followed. As an example, the worst situation is you bringing non-grounded device (having virtual ground) to the fully grounded environment.

Or are these currents just no problem in these practical applications?

Yes in office environments. No shielded cables are required. Using shielded cables can actually worsen the environment.

| improve this answer | |
\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.