Is there standards/guidelines published for grounding the shield when using shielded cable for EMI? I have heard arguments to ground at one end only; others say to ground both ends and use
capacitor to combat ground loop problems; still others say use resistor instead. Which is correct/best method? Is it based upon the signal frequency? I realize that much depends upon the
application but some rule-of-thumb guidelines based upon time proven testing/experience would make for easier design decisions.
Senior Controls Engineer
Gem City Engineering
(937) 223 - 4600 ext. 1456
I ground shielded cables only on one end ussually to the grounded terminal right next to the analog inputs when available, else to the panel ground.
It is true that ground loops will occur if you ground both ends.
I believe that using capacitors or resistors degrade the grounding and noise reducing capacities. This is because of the impedance to ground becomes higher, and thus allows more
voltage fluctuations in the shielding...
Books have been written on this subject and a proliferation of rules, "old control eng. tales", Etc. exist. General principles: Shielding protects the inner conductors from Magnetically induced voltages and from Electrostatic field
voltages. The shield need not be made of magnetic material, (Although it works much better at low frequencies e.g. 60 Hz if it is steel conduit), as the AC Magnetic field induces eddy currents in the shield which in turn oppose the
coupling magnetic field, (Lenz's Law). With the shield of the cable grounded, any electrostatic field in the area sees the shield as a ground and therefore does not penetrate beyond it. The shield of the cable also acts as one wire and
the enclosed wires runs parallel with it. These two conductors form an air-gap transformer with a turns ratio of 1:1. They are also capacitively coupled with a capacitance of pF/ft times the length, in feet. Grounding both ends of the shield to separate points that, e.g. differ in "Ground Potential" by 10 mV will couple 10 mV into the enclosed wire. This is very simply corrected by grounding ONE end only of the shield. Does it make any difference which end? No. A common "cook-book" rule is to always ground at the signal origin end. My own experience tells me to ground at the "Cleanest" end, (i.e. the end with the least noise present at the ground point), this way the least noise will be capacitively coupled into the inner conductors. I like to ground at the "lowest signal level point" e.g. close to the input to an op amp. Foil shield is better than braid, (100% coverage vs 85%). Where the magnetic field has sufficient strength to "Cut" the conductors, (and the shield), little improvement is possible, (apart from the twist in signal pairs, within the
shield). A thickwall shield, (conduit or even cast iron water pipe), provides a high permeability path that "short-circuits" a great deal, (ideally all), of the magnetic field thereby providing better magnetic shielding at low
frequencies. At RF the wall thickness makes little difference - here surface effects prevail.
The integrity of the shield must be maintained from the signal source to its destination, i.e. where the cable passes through connectors, the shield must also brought through the connector and not grounded at the intermediate connector(s). At high frequencies, standing waves require the elimination of 1/4 wavelength lengths of cable and it may be necessary to bond shields every 1/10 wavelength or so, so as not to end up with very high impedance, with high voltages, at the cable terminations.
Hope this helps,
Tony Firth, Electrical Eng.,
Quester Technology Inc., Fremont, CA.
Actually, I have found it generally better to ground at the transmitter end of cables feeding a multi-input panel. A not uncommon problem is
shield jackets getting nicked in field runs and shorting to local grounds. If they have been panel-end grounded, this results in ground loops through the panel to every ground on every cable. If your cables are all perfect, no problem. However, I have seen, and fixed, problems from this kind of thing in at least 6 systems over the years. I have never seen a problem due to shields grounded at the xmtr end.
Tony Firth wrote:
...............This is very simply corrected by grounding ONE end only of the shield. Does it make any difference which end?
Dear Mr. Baird,
Although I was told never to connect both end to ground, once, in a crane, I had a proximity switch which insisted to produce randomly short pulses. I found that happen everytime one of the AC motors contactor was operated. The only way of getting out of that was to connect BOTH ends to ground. No problems since them! On the other way, I have a friend who was working for the Sao Paulo
subway company. He had a similar problem in a communication line when they switch over power lines. He connect both end to ground and fixed the problem, but, the current that circulated
across the shield was so huge that melted the cable isolation. May be a capacitor in one side could have better solved the problem. So, as you can see, every application has its best way to do.
(Every problem always has, at least, one solution. The problem is to find it!)
Ricardo Ribeiro wrote:
>Although I was told never to connect both end to ground, once, in a crane, I had a proximity switch which insisted to produce randomly short pulses. I found that happen everytime one of the AC motors contactor was operated. The only way of getting out of that was to connect BOTH ends to ground. No problems since them! ...<clip> <
If you read the literature from Allen Bradley drives (and many other manufacturers), it states that to conform to EMC regulations you must connect both ends of the motor cable shield to the cable ground conductor and the drive frame/motor frame terminal. Associated control cables should have the shield connected both ends to signal common.
Normal installation methods recommend motor cable shields are connected both ends, but control cable shields are to be connected at the drive end only. This is the usual method which most manufacturers recommend installation.
Dairy Containers Ltd
Some words of caution regarding generalizations:
Grounding both ends of a shield on motor feeders could affect ground-fault relaying. A circulation current will cause erratic operation in protection circuits that use zero-sequence CT's.
Intrinsic Safety Circuits.
Grounding both ends of a shielded instrument cable could cause erratic operation of IS Barriers (less of a problem in Europe installation, but that's another thread).
Crane Proximity Switch Problem.
The transient was probably generated by de-energization of the contactor's coil or consequential de-energization of an electromagnetic device like a solenoid, control relay, brake-coil, etc. Although the phenomena is present in all circuitry, it is exacerbated in DC control circuitry.
Transient Suppression Techniques.
As an example, install diodes (if DC) and MOV's (if AC) in parallel with all coils of the type mentioned above. This would probably have been the
more technically correct fix. By the way, even a small 1-2 watt relay can generate a 1,500 volt "spike" which can "couple" into seemingly
Even transient suppression devices have a "useful" life. Like the proverbial "straw that broke the camel's back", these devices "wear out"
and should be replaced periodically.
Phil Corso, PE
I wish to say that probably there will be a more technically backed sort of answers, but mine is based on my practical experience after facing the same doubt years ago as Instrumentation engineer. I read a lot of white papers, articles and books about this topic. Let me say that there is not complete agreement among the statements of all this writings (see IEEE Transactions on Energy
Conversion, Vol. 8 No. 1, March 1993) but I would say the following can be considered basic: *If you try to ground shield for EMI on cables carrying low level analog or low frequency signals do it at one end only, preferrably near the source. This to avoid the so called "ground loops". Additionally take care of the wiring to the "ground": One practical approach is to concentrate all the ground shield leads in an isolated bus in the marshalling cabinet near the
DCS or Control Panel. Then run one cable from this point all the way down to the only "ground reference point" where all the grounds of the plant should be joined together, and I mean not only the EMI but the electrical safety grounds. I would recommend the reading of the IEEE std 142 (I have 1991 version) to get more information. *If you are speaking about shielding against EMI in cables that carry medium or high frequency signals, yes, you should ground not only at one end but probably at some more itermediate points depending on the frequency and the lenght of the
wires. I can't say more than this about this but I'm sure some one will come at rescue passing to us sources for information.
I apology for any grammar or style mistake. English is not my native language.
There has been some very good replies to this already. I agree that only one end of the shield (DCS/source) should be connected. If through experiment you find that connecting the field end helps, try isolating (insulating) the field device from local ground currents. If you leave both ends grounded (even with caps or resistors) expect to get problems as field conditions change, or after a welder strikes an arc in the
i fully agree with this . screens should alway be connected to ground at one end, i.e marshalling end. on the instrument side , the screen wire should be trimed and insulated .
I am not sure what is the question!
But I have seen various designs for various industry that actually tie both ends of a shield to:
1. In printing industry, some of the german printing presses purposely ties both ends of the shields to to ground. But they do have a ground detector in the main console, for alarming the user of a ground existing in the system. I am not sure why.
2. In a SCADA systems for monitoring Utility Power Transformer, the design ( from canadian outfit) of an oil sampling sensor specified to connects one end of the shields to ground and the other end of the shield to a terminal.. When I asked the manufacturer what is this terminal tied to? His answer was to ground through a capacitor!! They also insisted on mounting ferrite( Theses are hollow cylindrical made of steel) on the analog low voltage communication for noise suppression!!
I am not sure what all this is telling us but the idea of only connecting only one end of the shield to ground is not the only acceptable method.
Which manufacturer is this? Is this the regular connection to a standard PLC analogue module, or is there a device between the sensor and the PLC?
If it's a capacitive tank level probe, or something like it, then yes, the metal of the tank becomes part of the circuit because it and
the probe form the two parts of a big capacitor. The material being sensed is the dielectric media. These can be problematic because of ground loops, which would necessitate the RLC circuit to filter out all possible noise frequencies.
I shy away from such devices because of the pain in the butt that they are.
Anthony Kerstens P.Eng.
Adequate cabling, good connections and very important "Technical Intrumentation Groud", then shields may be connected. Where ? the practice
is at single point in the Control center (Panel,room ...). But yes, eventually for some physical principle in the measuring device, the shield must be grounded at the measuring device.
I have seen shields fusing each time the man attempted grounding in the field (ground loop).
The mistake originated from the drafting room
(not cultured in Instrumentation).