Why shield should be at one end in cabinets


Thread Starter


Instrument shield connection to ground, Why we should connect at one end of the cable, especially internal cables? If the cable going From the DCS to ESD, MCM, IRP cabinets which side we need to connect the shield?

Please explain me detailed.
Sudheer... I suggest you search the Control.Com Archive for like threads.

You will find that most posters suggest grounding the shield at the panel-end, rather than the instrument-end, and never both-ends simultaneously!

Regards, Phil Corso

Curt Wuollet

The why is that it's usually "quieter" that way with the shield carrying only induced currents from noise sources rather than potentially carrying currents from differences in ground potential. It also allows common mode rejection of noise from from such differences and other such subtleties. I have seen exceptions to the rule but they are rare and grounding is so poorly understood that unless you have a good reason, it's a good way to do things.


cww who has seen grounding cause a lot of weird problems.
cww is correct. Induced voltages caused by external influences on low-voltage analog signal wires can cause a LOT of unusual, intermittent, nuisance and extremely frustrating problems. The shield itself captures the stray external influences, and the bare shield drain wire which is in contact with the shield should only be grounded (earthed) at one end of the cable. That way all of the captured signals are "drained" to ground at one place which provides the protection from induced voltages on the conductors inside the shield.

If the shield drain wire were grounded at both ends of the length of cable, then as cww has said it provides a path for current to flow between the two ground points. If the ground points are not at the same potential (and many times they are not!) then current can flow in the drain wire, which is in contact with the shield, and cause more problems by inducing voltages on the wires inside the shield. Even small voltage differentials between two locations can cause unwanted current to flow in the shield drain wire, which is in contact with the shield over the entire length of the cable. And, it's not always so much the fact that current is flowing, it's the fact that the current can vary greatly and can be very intermittent--and that's what causes the "weird" problems we all encounter from time to time.

Induced voltages can be caused by many things--but mostly by cables which are not properly segregated from each other based on the voltage- and current levels carried by the conductors under normal conditions. Most often when low-level analog signals are run over long distances in parallel with higher voltage and higher current-carrying wires then induced voltages in the low-level analog wires occur which can adversely affect the control system using these low-level signals for control and protection.

If multiple sections of cable are used for low-level analog signals which pass through multiple junction boxes, then extreme care must be used to terminate one end of the shield drain wire of each section of cable. The shield must NOT be continuous from the device to the control system input- or output terminal in the case where multiple sections of cable are used to connect the device to the control system. Rather, each section of cable must have its drain wire terminated at one end only, so that the captured stray influences are taken to ground at one place only.

And, as cww said, grounding is a very poorly understood concept. It seems easy enough, but in reality it's not very well enforced during construction or afterwards. Plants with functional earth systems and protective earth systems can--and quite often do--have unintentionally mixed grounds which can--and does--lead to voltage differentials between the two ground systems which cause lots of nuisance and strange problems.

Again, if a shield drain wire is terminated at both ends there is a very high likelihood that it will actually cause more problems than it is intended to solve, by becoming a current carrying conductor and inducing unwanted voltage on the conductors it is meant to protect from stray influences. So, as cww says, unless you have an extremely good reason for grounding the shield drain wire at both ends (better than, "I think it should be grounded at both ends!") it is best to ground the shield of any section of shielded cable at only one end.
Thanks for the info.

I am particularly interested in why cannot we make our shield continuous when it passes thorough multiple junction boxes. Can you explain please? I have a similar situation where I have a multiple section of cable but the shield is continuous and grounded at Main Control Cabinet only?

What you are describing is entirely possible--and would be correct as long as the shield is grounded (earthed) at one end only.

The problem is that most electricians are not knowledgeable in proper grounding practices, and tend to ground shields at BOTH ends (usually because there are two different electricians working at each end of the cable and there is no coordination between them and a lack of supervision).

Yes; a "shield" can be made continuous through multiple junction boxes using several individual cables--BUT the caveat is still the shield must be grounded at one end only and open (floating) at the other end. In several cases where I've encountered intermittent noise problems I've found "continuous" shields--BUT someone had added a ground at the junction of the cable shields in one or more intermediate junction boxes, again likely because of lack of supervision and a false belief that a shield drain wire must always be grounded in every junction box--which is not true. Each segment of a shield which is NOT continuous must be grounded--BUT only at one end. OR, if the shield is truly continuous through multiple cable segments and junction boxes then it must only be grounded at one end.

Shield drain wire grounding is very simple--but proper supervision, and an understanding of the concept by the electricians and proper supervision and oversight of them is required to ensure it properly followed and implemented. Again, most electricians think the shield drain wire at the end of the cable they are working on must be grounded--without realizing their colleague at the other believes the same thing! So, that's usually why one end of each segment is grounded when multiple segments are used, because someone has to go back through the junction boxes and lift one end of each segment's drain wire.

Hope this helps!

Curt Wuollet

You can do whatever you want with your shields:^). The idea behind grounding one end only is that the shield will then only conduct induced currents to ground and prevent induced voltages on the signal wires. Thus shield. If you ground it multiple places and there is a difference in ground potential at those places, your shield is now carrying current. And your shield is in the very best position to induce currents in your signal wires because it completely surrounds them and has a lot of distributed capacitance to the signal wires. So the idea is not to turn your shielding into a noise injector. If you _wanted_ to induce noise in some wires, the best way would be to wrap them in foil or braid and drive the wrapping. That's why an ungrounded shield can be worse than none at all, and carrying current through the shield is a bad idea.

And yes, sometimes you get lucky and it works fine grounding at every box. But then grounding conditions change and once a month, you get errors that go away Or when that big machine turns on or when the arc furnace up the road leaks a little, or........... It's simply more predictable if you don't provide a path for current in the shield. Don't feel bad, grounding is the least understood topic in electronics, I see evidence every day.

Agreed. However it needs to be understood that the foil wrap on a single twisted pair is conductive on one side and insulative on the other. I have seen many junction boxes where individual pairs from a multi-conductor cable are exposed for preparation to be terminated into terminal strip. if the individual foil wraps and not taped or heat shrink covered, this may produce a foil to foil connection that compromises the integrity of ground at one end only policy. Grounding shields is the least respected and most violated aspect of many installations I have seen.

> The shield must NOT be continuous from the device to the control system input- or output terminal
> in the case where multiple sections of cable are used to connect the device to the control system. Rather, each section
> of cable must have its drain wire terminated at one end only, so that the captured stray influences are taken to
> ground at one place only.
The electricians only wire what is on the drawing and more than a few engineers have made grounding mistakes.

> The electricians only wire what is on the drawing and more than a few engineers have made grounding mistakes.

I will wholeheartedly agree with the second of the two statements--but not with the first. There are few drawings used in industrial projects these days, and in my experience drawings aren't generally given to electricians. Mostly they just get a table of terminations (sometimes called a "cable schedule") and sometimes the person(s) creating the tables/cable schedules from the drawings include shield drain wire information (which may or may not be correct!) and sometimes they don't (my experience is, it's not included). So, again, left to their own devices (and a lot of craft people are these days) they will do what they did on the last job and previous jobs before them and just ground the shield drain wire at their end of the cable without thinking about what their colleague is doing at the other end of the cable.

Old Guy said it best:

> Grounding shields is the least respected and most violated aspect of many installations I have seen.

Few truer words have ever been spoken about most industrial installations these days.

Here's a drawing which is how I usually find most electrical installations prior to start-up/commissioning:<pre>
Control 0---\/------------\/---0---\/------------\/---0---\/------------\/---0 Field
System 0---/\------------/\---0---/\------------/\---0---/\------------/\---0 Device
0--------------------| |------------------| |------------------|
| Segment 1 | | Segment 2 | | Segment 3 |
--- --- --- --- --- ---
- - - - - -
GND GND GND GND GND GND</pre>There are three cable segments, each passing through a junction box; terminals are represented by "0". Each cable segment is a TSP (Twisted, Shielded Pair), and the twisting of the two twisted signal wires/conductors are represented by the "crossing" symbol. Each TSP has a foil shield and a bare shield drain wire which is in contact with the foil shield--represented by the lowest "straight" line of each segment.

The drawing above <b>IS NOT</b> a valid cable installation--and would most likely result in lots of nuisance and intermittent low-level signal problems particularly if proper cable segregation practices are not followed. (And I would add to Old Guy's statement that the other most violated and disrespected installation practice is cable level segregation--and it's virtually impossible to correct after installation.)

So, what usually happens is that we end up going through each low-level junction box and each circuit ends up looking like this:<pre>
Control 0---\/------------\/---0---\/------------\/---0---\/------------\/---0 Field
System 0---/\------------/\---0---/\------------/\---0---/\------------/\---0 Device
0-------------------- |------------------ |------------------
| Segment 1 | Segment 2 | Segment 3
--- --- ---
- - -
GND GND GND</pre>The above <b>IS</b> a valid and acceptable shield drain wire grounding scheme--<b>each TSP segment's shield drain wire is grounded at one end only.</b>

Now, here's another perfectly valid and acceptable (any my personal preferred) shield drain wire grounding scheme:<pre>
Control 0---\/------------\/---0---\/------------\/---0---\/------------\/---0 Field
System 0---/\------------/\---0---/\------------/\---0---/\------------/\---0 Device
| Segment 1 Segment 2 Segment 3
GND</pre>In the drawing above, the shield drain wire is made continuous along the entire length of the circuit, and is grounded <b>at only one end.</b>

Hopefully, three poor drawings are worth a thousand three hundred and fifty-four words!
Hello Old Guy,

Absolutely agree with your answer. However, what is the reason that you have "The shield must NOT be continuous from the device to the control system input- or output terminal."

If I chose to ground my shield at the Control System Cabinet only, why can't I have my shield continuous? It adds to more cost by adding the grounding blocks in the din rail of the junction boxes. If I carefully make my shield continuous and terminate it at the Control System Cabinet end only, instead of terminating it at each section, what difference would that make? By termination, I meant only one side of the shield?
Typo error somehow made with reply.

Correction as cited pertaining to shields being continuous.

good catch!!

Old Guy.
I know this post is a bit old but I have a relevant situation that I know applies here. I have an instrument with a Siemens PLC controller using RS485 serial communication to talk to 12 devices (pumps with stepper motors and chopper drives in them). We have failed our EMI testing and the culprit we think is improper grounding. Since I am the engineer who grounded the wire I would like to understand what went wrong...

My grounding scheme for my twisted pair shield is just like the second drawing above (http://www.control.com/thread/1373726311#1414703336)

with 0----0----0----0---GND.

Grounded at only one end of a 12 segment. I just had an EMI consultant take a look at our machine and immediately he knew the problem by looking at the grounding scheme. He told me the shield needed to A) be grounded at both ends and B) be grounded at each segment. The guy is a relatively older gentlemen who certainly has a lot of experience with this sort of thing and I don't doubt he is correct. When I told him that I have always heard you want to ground only one end of the drain he simply told me that is for "audio cables only".

In our case our instrument is emitting too much radiation and we are trying to contain it/reduce it. I maybe over-simplying the situation but, is the reason for "both-end grounding" because we are trying to reduce our emissions? If so, is "one-end grounding" to eliminate the induction of noise on the transmission line? Thanks for any help!
The industry "convention" is that the shield drain wire should be earthed/grounded at the source of power, which is usually the control system end of the wire--but not always. And, that's just a convention. Which is especially convenient for sites which use a protective earth grid and instrument earth grid--because the instrument earth grid isn't usually run out to/available at the field decive/instrument locations.

You can earth/ground the shield drain wire of ANY segment as long as it's only grounded at one end. And, it can be earthed/grounded in a junction box, as long as you know the earthing/grounding lug/bar has a very low resistance to ground. (Many times painted junction boxes are not properly grounded, or are poorly grounded using star washers which were improperly installed.)

You can also earth/ground the field device end of the shield drain wire at the field device instrument location--again, knowing that the earth/ground lug has a very low resistance to ground. The problem with doing this is that if the plant uses the dual earthing/grounding system there may not be an instrument earth/ground location available at the instrument/device location.

I have some very serious concerns about what you were told; it just doesn't jibe with anything I've experienced or seen practiced in industrial process plants, including power plants, for shield drain wire earthing/grounding.

Something I learned a few years ago is that the main microprocessor or most controllers can be VERY noisy and emit lots of EMI just by themselves. Using a filter on the power supply of the controller designed to limit the amount noise which is reflected back on to the power supply wiring can do a lot to help mitigate EMI problems. I've also seen Ethernet network switches emit a LOT of EMI, especially when people try to use consumer-grade switches instead of industrial-grade switches.

Are you running signal wiring in the same conduit/tray as the chopper drives? How much current are the chopper drives drawing, and is it smooth/continuous current, or choppy?

Lastly, are you sure the Siemens PLC controller is properly earthed/grounded?
Yes, there are no end to grounding and shielding questions. And yes, the general rule is that you want shields grounded at one end to prevent circulating currents and induced currents in the shielded wiring. BUT, you need to consider exactly what it is you are trying to do with the shielding in a particular situation. And that is different when you are making noise rather than trying to reject noise.

For example, when you are trying to get a high impedance analog signal from place to place, you want the shield to intercept noise fields and drain the noise induced without causing more or producing the condition for that noise to couple into the shielded wiring. You want to drain them to the reference for the signals and a low impedance point. This is generally the sending end. Any noise on the shield is best tolerated there, if you want to think of it that way. Leaving the receiving end flying prevents currents from differences in ground potential from creating currents in the shield which would induce noise. With no complete path, noise on the shield will not be injected into the signal ground on the receiving end. These are not absolutes, but in most situation will work best.

If you are making noise, the situation is almost entirely different. First of all you are not worried much about outside noise getting in or being induced into the receiving end. Except in very unusual circumstances you are not particularly worried about currents from differing ground potentials. What you are trying to do is make a complete Faraday cage around your noise antenna from one end to another and keep the noise inside the shield. The power levels are such that it's very likely that your shield will carry currents from imbalance and often components of the "noise" are stuff that you want to get to the other end. Take one of our most popular and noisy sources. For a VFD, if you use shielded cable on the output, you "ground" both ends and insulate from everything in between. I put ground in quotes because is a fact and the code but really what you are trying to do is complete the return for radiated noise back to the source. So the connections to the frames are the key. The shield has finite impedance so it will likely carry measurable voltage at any distance from the source. That is why you insulate it to prevent injecting noise in the grounding system of other cabinets it goes through. On the motor end, you connect it to the body to complete the shielding of the antenna. The VFD is well aware of the noise, the motor, for the most part doesn't care if it's frame is a little lively. So your goals are different. Also. even if you were concerned about the ground loops and external noise induction, that's _your_ problem. The authorities make the goal preventing radiated noise as much as possible, so you do what that takes. And sometimes it takes attention to impedances and frequencies and all the impolite details of electrical physics. It may take clamping the shield directly to the frame rather than using a wire, for example, because a wire has higher impedance. A foot of wire might well render the shielding moot. So details are important as well as keeping in mind what you are trying to accomplish.

I hope this makes sense. I shouldn't write before coffee.

"In our case our instrument is emitting too much radiation
and we are trying to contain it/reduce it."

Given you rather minimal description, just which end is electrically noise, and what sort of noise?

Signal grounding aside, the devices at each end of the signal link should be grounded.

So what sort of EMI measurement are you using and with respect to which ground.

The cases at each end should be electrically grounded so safety.

The issue of shield grounding has to do with minimizing electrical ground loops through the signal shield. This is important, as then the signal wiring has only to deal with the common mode and induced noise.

The common mode noise must be within the CMRR of you devices, and the wiring must be segregated from non-signal/instrument wiring.