With regard to termination of ethernet shielding I have been taught always to terminate only ONE end of the shielding to ground to prevent ground loops occurring.

Is this best practice?

and is there any other way of terminating the shield?

Finally.. and I'm sorry for the list of questions but the way I look at this schematic symbol is that the shield connects to pin QE. Is this what others think too?

enter image description here

  • \$\begingroup\$ This is actually interesting question. Looking through the web I see absolutely different points of view: that ground loop is not a problem and that ground loop is huge problem (just put "ethernet ground loop" into Google). I would also want to see answers to your questions. The one thing bothers me with grounding at both ends: for differently grounded devices it simply creates risk of electric shock and equipment damage when plugging. BTW, twisted pair itself is electrically isolated within both network devices being connected. And last - do you have any sample of shielded cable? \$\endgroup\$
    – Anonymous
    Commented Sep 30, 2016 at 20:29
  • \$\begingroup\$ "common mode noise currents do not affect data transmission because, regardless of their voltage magnitude, their waveform is always associated with the profile of the building AC power (i.e. 50 Hz or 60 Hz). Due to the excellent balance of the cabling at low frequencies, common mode currents induced onto the twisted-pair either directly from equipment impedance differentials or coupled from a screen/shield are simply subtracted out by the transceiver as part of the differential transmission algorithm." siemon.com/us/standards/… \$\endgroup\$ Commented Jul 31, 2017 at 19:10

2 Answers 2


Unfortunately junior engineers may have misunderstood advice about Single Point Grounds (SPG)

If a shielded cable is NOT grounded at both ends, H fields are attenuated by LPF and E fields are amplified by HPF effect.

  • There can be real problem grounding any conductor including shields between two units operating off of two power mains in or between buildings. Ground potential differences can be kilovolts during lightning strikes. Ethernet has common mode in-line isolation rated at 1.5kV rms per IEEE 802.3.

The shield is not the issue; Engineers’ knowledge of ground potentials, isolation, and shielding is the issue.

  • Floating one end of the shield only makes the problem worse.
  • Isolation units can be inserted to solve most problems, if high AC ground current exists, or if interbuilding lightning is a risk in high strike areas such as Florida during hurricane season. (100 flashes/s in a 30km radius.) Then building grounds must be improved.
  • a series plastic Y2 Cap that is high impedance at 50/60 Hz but low Z, ESR are 100KHz is a possible solution if inter-building ground fault currents are large. Then any metallic exposed grounds must be insulated >3kV rating.

The PHY of the ethernet has an effective high CMRR balun but may not be enough if RF noise couples to an open ended shield with 1/4 wave effects and high E field coupling.

The shield is supposed to be grounded at both ends.

  • \$\begingroup\$ Than you for the answer, the application is for rail application 25KV dual traction equipment, which switch loads constantly as each carriage goes through a neutral section. would this change your answer. ? \$\endgroup\$
    – hoboBob
    Commented Sep 30, 2016 at 20:55

When I am using the industrial ethernet, the sockets are metalized, and the connectors are metalic. So the shield is clamped on the connector casing and this goes directly into the socket, therefore this equals to the conection of the shield at both ends.
The using of FTP cable implies also using of metal connectors and sockets, so IMO nobody connects shileding separately. When using patch pannels, then the shilding is also connected on the casing - both ends.


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.