6
\$\begingroup\$

When installing cabling for Ethernet purposes, the question often arises: does one want/need simple UTP or any of the shielded variants, i.e. F/UTP, S/UTP, SF/UTP, U/FTP, F/FTP, S/FTP or SF/FTP, often labelled FTP or STP?

Most sites with information on the topic are from cable manufacturers or distributors, and all the shielded variants are of course deemed superior (and they are more expensive), but with very little justification. Beyond the additional cost of the cable (and sockets), shielded variants also require grounding, which complicates installation.

Cat6E, which supports up to 10G Ethernet, only requires UTP. Wikipedia states that:

Cable shielding may be required in order to avoid data corruption in high electromagnetic interference (EMI) environments

Are there real, objective, guidelines/rules on when one needs more that UTP for Ethernet? What are "high EMI environnements"? Is this something you are likely to find in regular offices (I've seen vague references to lighting equipment), or is this just for some extreme cases one is unlikely to encounter outside of industrial premises or research labs?

In my case, I'm interested about "run of the mill" 100 Mbits/s or Gigabit Ethernet, nothing fancy like 2.5G, 10G, 25G, 40G (I understand the latter two require various forms of shielding).

Some places also indicate (without justification) that shielded cables are required/better for PoE applications, though I fail to understand the relationship.

\$\endgroup\$
7
  • \$\begingroup\$ Do you understand why twisted-pair is needed and that twisted pair also cuts down noise in high EMI environments and reduces cross-talk? \$\endgroup\$
    – Andy aka
    Commented Feb 17, 2023 at 11:59
  • \$\begingroup\$ @Andyaka Yes, I understand how twisted pair reduces EM interference, and that's why I fail to understand what exactly the additional shielding protects from, and in what circumstances it's actually useful/needed. If it were never needed I suppose the additional types of cables would not even exist, so there must be cases where it's useful, but I would like to undesrstand what those cases are. \$\endgroup\$
    – jcaron
    Commented Feb 17, 2023 at 13:06
  • 1
    \$\begingroup\$ Twisted pair reduces the M part of EMI i.e. it reduces the induction of noise voltages due to magnetic interference. A shield reduces the E part of EMI i.e. it reduces the injection of currents into the wires due to capacitive coupling. Does that help? Is that the answer you are looking for? \$\endgroup\$
    – Andy aka
    Commented Feb 17, 2023 at 13:17
  • 1
    \$\begingroup\$ @Andyaka I'm more interested in what are the likely sources of interference which the shield would protect against (but the TP wouldn't), and whether those can be expected to be found in "regular" office environments or not. All the information I have found is very subjective, the various shielded versions are "better", but there's really nothing quantitative about what kind of source of interference (and/or parameters, like size, voltage, power...) would require one. \$\endgroup\$
    – jcaron
    Commented Feb 17, 2023 at 13:25
  • \$\begingroup\$ The general advice I've always heard is that for normal use shielded TP is not necessary. I've primarily used it around industrial equipment and especially electric motors. I'm not sure if that still applies looking at 5/10g Ethernet. \$\endgroup\$
    – user1850479
    Commented Feb 19, 2023 at 15:25

1 Answer 1

Reset to default
4
+100
\$\begingroup\$

The problem with EMC/EMI is that it is hard to state rock-solid limits for power/voltage/magnetic flux as the resulting coupled-in signal strength is very setup dependent. However, there is interesting research on "no invasive" attacks on ethernet, e.g., https://download.hrz.tu-darmstadt.de/pub/FB20/Dekanat/Publikationen/SEEMOO/wisec2016-trust-the-wire.pdf figure 7 & 8. - while this is "the wrong direction" (i.e., detecting the emitted signals from the cable), it will give you a good idea about how much the shielding will dampen an external source. I am sure you will find plenty of other relevant publications if you use this one as a starting point.

Regarding potential EMI sources, there are standards all equipment (should) meet. For example, the generic IEC 61000 defines allowed limits for residential and industrial/commercial space for the maximum permitted emission and minimal required tolerance for immission. There are sometimes specific limits defined by standards for other applications like automotive, aerospace, military, and medical. If you know your surrounding and the installed equipment, you can look up the limits they (should) comply with and deduct the maximum expected interference strength. Then you can either simulate the interference with the adequate equipment or design your system by "simple" SNR calculation. In my experience, there will be a range where everything might work fine most of the time. However, you will sometimes notice decreased performance or erractic behavor due to sporadic immission leading to packet collision. The same apply for too long cables.

In the end the performance is a result of the achieved SNR. If your S is high enough you dont need to worry about N and vice versa. Considering your statement that you want to run only 100M/1000M and are likely in a standard office space, UTP should be fine.*

*I work in an office where we noticed issues with some of our oscilloscope readings. Later we realized that a neighboring company tested smart-grid stuff frequently in their lab - most likely, they did not adhere to residential limits.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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