Should a Shielded Cable Be Grounded at Both Ends for Optimal Performance?

[likephysics]

What is the correct way to terminate a shielded cable?
Should it be connected to Earth ground at one end? What about the other end, leave it open?

I was playing around with Radiated immunity tests -one end(close to EUT) of the cable was connected to Earth and other end was left open. Interference was horrible. No better than a unshielded regular wire. I grounded the other end (to the same ground point) and a lot of the noise was gone. It was a short cable, say 3meters, sheilded (non-twisted).

A lot of articles I have read, says ground the cable at one end. Otherwise it will result in a ground loop.
So what do you do at the other end.
How come the signals that interfere do not take the grounded path and still couple to the center conductor.
Equivalent ckt diagram would help a lot.

 

[berkeman]

How to best terminate and ground shielded cable depends on the application and on the environment. If you have quiet Earth ground available all along the length of the cable, then tie off the shield to Earth ground at intervals less than the wavelength of the external interference (or internal noise source, if the problem is in that direction).

But it is rare to have lots of quiet Earth ground connections available, and you often have to worry about 50/60Hz ground currents. So then a good alternative is to just use AC ground ties with resistive bleeders at each end of the cable. See page 4-6 in this document, for example:

http://www.google.com/url?q=http://...ds-cse&usg=AFQjCNE7CLJvhbZlxq1TguE9lNWBhyGXyw

When you have a lot of common-mode noise trying to couple into your shielded data cable, you can use full terminations at the end of your doubly-terminated bus where the terminations take into account all 3 characteristic impedances of the cable. That includes the differential Zo of the twisted pair in the shielded cable, the common-mode Zo of the twisted pair to the shield, and the common-mode Zo of the shield to the outside environment. The first two Zo numbers will be determined by the cable itself. The 3rd Zo will depend on how the cable is routed with respect to grounded metal surfaces. See the termination on page 3-7 of this document, for example:

http://www.google.com/url?q=http://...ds-cse&usg=AFQjCNGVJxC71nmBjlRYZ9IGNGU1LMEngQ

.

 

[likephysics]

Thanks Berkeman.
How does noise still get coupled when one end is open?
Is it because of the capacitance between the shield and the center conductor?
or is it because of the magnetic field from the ground current?

 

[jim hardy]

earlier editions of this text served me well.
https://www.amazon.com/dp/0471245186/?tag=pfamazon01-20

he suggested the shield be tied to low side of signal at its source. that way capacitive current flows in the shield not the signal conductors - you've built a Faraday cage around them. sort of same idea as a driven guard...

 

[berkeman]

Thanks Berkeman.
How does noise still get coupled when one end is open?
Is it because of the capacitance between the shield and the center conductor?
or is it because of the magnetic field from the ground current?

I believe it is mostly capacitively coupled from the sheild to the inner conductors.