I'm about to do cabling in a server rack that consists of two switches and forty 1U servers. Since every server has three ethernet cables attached we'll have around 120 cables in a single rack.

  1. Is putting unshielded RJ45 connector on a shielded Cat6 ethernet cable a good idea? Since the RJ45 connectors are very close to the endpoints, does it really matter (EMI-wise)?

  2. How important is grounding of that cable? Should I go extra mile and put RJ45 connectors with metal jackets that connect to the external metal shielding of Cat6 cable?

Thank you in advance.

  • \$\begingroup\$ how many "miles" ;) long are the cables and do they share a common rack chassis gnd? \$\endgroup\$ Commented May 15, 2017 at 21:40

1 Answer 1


Since you have STP, you should terminate them to gnd for RF improved performance.

Signal Integrity and Group delay times vs f for CAT6 signals are improved with a grounded shield rather than,a floating shield that couples the leakage of all signals to the ground. But this property is depending on cable lengths up 50m or 100m in future drivers.

Avoid cable kinks and shape bend radius smoothly > 4Dia and avoid shield strip > cable dia. at ends.

While shielded components do cost slightly more than UTP, the increased performance and reduced testing time can ultimately lower the total cost of ownership.

The next generation cabling systems that perform beyond 10 Gbs transmission speeds will require shielded cabling and components, which makes STP systems more future proofed than UTP.

The result is similar lifetime costs for both STP and UTP systems.

  • \$\begingroup\$ Are both ends of the shield to be grounded? I thought the big point of ethernet being transformer coupled was no shared ground? \$\endgroup\$
    – Colin
    Commented May 16, 2017 at 8:18
  • \$\begingroup\$ Au contraire , a similar question ignored my answer to that. and with Cat6 using 500MHz BW and up the 3dB loss point affects performance. so shielded CAT6 is essential in future upgrades and helpful at present max speed. for broadband impedance control. I showed that the terminators must be coupled to chassis ground with 1000pF so DC or AC line ground currents are not possible. but an AC "RF" ground essential for CAT5 but CAT6 standard terminates to metal plugs to shield (STP ) \$\endgroup\$ Commented May 16, 2017 at 10:49
  • \$\begingroup\$ Ah, so the shield is capacitivly coupled to ground at both ends? \$\endgroup\$
    – Colin
    Commented May 16, 2017 at 10:49
  • \$\begingroup\$ optional now unterminated but better terminated.... , but must be terminated for better Return loss up to 500MHz spectrum used to encode 10Gbps data rate \$\endgroup\$ Commented May 16, 2017 at 10:51
  • \$\begingroup\$ 35 yrs ago , I did a spectrum analyzer study one day only on CATV up to 300 MHz from train traction motor EMI near a street with ingress noise problems. Problems with shield earth grounds from coax earth ground >>1 Ohm affected shielding quality was the root cause for interference. Ethernet is no longer simple biphase but multi phase amplitude like modems at > 1Gbps so group delay distortion is critical for equalization and shield term and standing waves on shield affect this plus crosstalk and EMI and signal integrity \$\endgroup\$ Commented May 16, 2017 at 11:01

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.