I'm assuming that screen is the same as the shield wire in a twisted-shielded pair.
One important factor as you know is that each shield should only be grounded at one point to avoid circulating currents.
I believe the cabinet end (as opposed to sensor end) is chosen by convention. I'm not sure if there are any advantages of choosing one end over the other.
Hi electricpete, it is usually assumed that the earth potential between the sensor end and the cabinet end will vary, and that the cabinet is earthed to the building grounding system. Thus, a less variant ground, and thus less common-mode noise on the shield. Further, by everyone using the same convention, cables don't end up with both ends tied to ground.
Just a couple of cents worth for whatever that might be worth.
GE Fanuc says that grounding the shield for input devices at the device rather than the panel is preferable (since it removes the currents due to local area noise before they get to the cabinet), but goes on to say that shields for input devices can be grounded in the panel in less noisy environments. They give a pretty good explanation and a variety of examples in the manual referenced below.
Ref: "Series 90-30 Programmable Controller I/O Module Specifications," pp. 59-71
A major problem is that there is often no convenient place for a ground connection in the field, where there is in the panel. Can you imagine ground wires and grounding points all over your facility? I ground shields for both analog inputs and outputs inside my panels, and I've had no problems with noise to date.
From IEEE Std 1100-1999, "IEEE Recommended Practice for Powering and Grounding Electronic Equipment":
4.8.1 Electrostatic shielding
"In order to be effective, shields must be grounded via low-impedance paths at the frequencies of interest. Long grounding conductors and long (single-grounded) shields exhibit reduced effectiveness at high frequencies due to inductive reactance in the grounding conductor or shield (e.g., +jX is randomly being placed in series with -jX). Therefore, very short grounding/bonding leads must be used, and they must be connected at the nearest equipment ground. Long shields need to be grounded at multiple locations along their length. Cable shields must be either grounded at both ends or grounded at one end and grounded via an SPD at the opposite end."
4.8.2 Electromagnetic shielding for EMI
"... The following generalizations are also pertinent:
...e) Enclosing the signal conductors inside of a shield and then grounding the shield at both ends. This is a key concept for protection of the contained conductors from the H-field effects produced by nearby lightning and other surge currents."
4.8.2.1 Cable shields grounded at both ends
"The "golden rule" of cable shielding requires that the shield on a cable only be grounded one time and at one end only. This "rule" has been established in order to prevent conductive "ground loops" from being established that would cause unwanted current to flow in a shield that is grounded at more than one place, e.g., at each end. The problem is that this is not a valid "rule" except sometimes when dealing with dc through LF signals (particularly analog signals) and where the signal circuits are not connected in the differential mode."
From IEEE Std 518-1982 (I apologize for the out of date copy), "IEEE Guide for the Installation of Electrical Equipment to Minimize Electrical Noise Inputs to Controllers from External Sources":
6.4.2.7 Practical Noise Immunization Techniques and Intercconection Wiring Practices.
"(10) All electrostatic shields should be terminated. An electrostatic shield should be terminated at one point only. The shield termination point should be at the same electrical potential as that to which the signal is referenced (typically amplifier power supply common).
(11) If a significant potential difference exists between a shield termination point and equipment safety ground, special electrostatic shielding techniques may be required."
Clearly not a simple subject, given the conflict between 1982 and 1999 IEEE standards regarding single-point grounding of electrostatic shields. This subject is certainly worthy of more than a pat answer. I encourage simpsy to investigate the standards cited above, as well as other relevant standards and material.
lots of good info. thanks for reminding us it's not a simple subject redtrumpet.
if we restrict the discussion to instrument loops consisting of analog signals (4-20milliamp, 0-1v, 1-5v,0-10v etc), I believe the standard practice is that shields are only grounded at the panel.
It appears from the references cited above that different practices may sometimes be applied for digital circuits (I was unaware of that).
Good point, electricpete, on the instrument loops. I just re-read the original question and it specifically referenced instrument cables. The references I cited concerning grounding of both ends of the shield are of concern only at much higher frequencies.
On long cables, the difference in ground potential at each end of the cable can be large enough that a substantial current will flow, possibly even enough to damage the cable.
In addition, as mentioned elsewhere above, by breaking the path and stopping current in the shield, common mode noise is reduced in the signal conductors.
Some instrument manufacturers (including GE/Fanuc) provide separate shield in/out terminations for shields. The shield out terminal is solidly bonded to ground, but shield in is capacitively coupled. This is a more expensive way to handle things but arguably is a good compromise.
Grounding at the Panel end of the signal cable is not something that is strictly adhered to. I do recommend it though. I have seen cases where people grounded at the instrument end of the cable, but failed to recognize that what they thought was a legitimate ground was actually isolated from ground. For example, I saw RTD and pressure transmitters with a ground terminal available at the instrument, but the instrument was isolated from true ground by Teflon tape.
Excellent discussion!!! GE Std for Speedtronic Turbine Control Panels is to ground at the ground bar @ the control panel, while from xnuke's posting seems like another GE business recommends otherwise.
Peebee mentioned the separate ground bus bar at the control panel.
I cannot go into the code details, let me add a couple of comments (from a field guy):
Having all the shield grounding grouped together at the control panel:
1. Makes life real easy to troubleshoot shields that may have more than the ONE GROUND ONLY covenant... i.e. by lifting the shield ground bar it should become isolated from ground (OL indication in the multimeter), if there is a ground present on the bar...tough luck, you have to check all the shields landed on that bus.
It really saves a lot of troubleshooting time specially going from JB to JB (Junction Box, not the whiskey mind you... that comes after you've found the sucker!!)
2. If the shield is broken somewhere between the device and the control panel... the control panel is still protected against spikes like those caused by lightingbolts... i.e. better to change a $1500 transducer than a $5000 electronic control card, besides... perhaps you can surely live without one instrument...but more likely than not you may trip the equipment (or loose redundancy) if the control card gets smoked.
3. During initial installation it is much easier to check that all the shields are landed for the same reason explained in 1. above.
If I summarize what I have read above, it sounds like there are two different classes of reasons for choosing the panel end over the field-device end:
#1 - convenience
#2 - better emi protection in some cases (?)
Maybe this is off topic, but in support of the ‘both ends’ approach—an oldtimer in instrumentation design and installation used to explain that is important to separate electrostatically-induced noise versus electromagnetically-induced noise. He stated that the single-ended shield bond is best for electrostatically induced problems, but for electromagnetically induced effects, shields need to be bonded at both ends so that shield current could offset and minimize undesirable effects, as if they were {‘air core,’ I guess} transformer windings.
There is some interesting common-mode ‘noise’-mitigation components for objectionable 60Hz steady-state and transient interference, along with related solutions for common-mode or 'lateral' ground-potential rise for phone/data lines up to several kilovolts. It is a lot of text to review, but there is some interesting information that seems to well support the double-grounded approach for some situations.
The product of interest is an “induction neutralizing transformer” with the primary application in communication circuits.
if I'm not mistaken I have seen a neutralizing transformer used in conjunction with a pilot wire protection scheme for transmission line between generating plant and it's switchyard (about 1/2 mile). I couldn't tell you what it does or why it's there.
From my perspective, the magnetic concerns are primarily addressed through twisting of the pair. There is no area between the pair for differential mode (opposite direction in each signal lead) em voltage. I can't really see common mode magnetic induction (same direction in each lead) as a problem.... where is the loop path of concern?
I should point out that I'm no emi expert... I'm just trying to follow the conversation.
Suggestion: Normally, the shield is grounding on the "sending end," where the signal (e.g. voltage) is expected to be a "source," and the "receiving end," which is expected to be a load, is usually ungrounded. If needs to be (according to rules posted in the above postings) both ends are grounded (to prevent a relatively high voltage at the receiving end). This means that the instrument loop shield grounding needs to be assessed on a case-by-case basis. Please, notice that some instruments have power source/supply in their associated instrument cabinet and others have their internal power supply; therefore, they are not grounded in the instrument panel, cabinet or on the instrument rack. The thorough intrument understanding and schematics of instrument loops are necessary to avoid the instrument loop malfunctions.
