Grounding screened control cables

TWControls,

Technically and theroetically, whether or not to ground one or both ends of a cable depends on the frequency of the currents. Low frequencies = ground one end. Higher frequencies = ground both ends.

Grounding both ends of a cable shield can cause problems only if the facility grounding system is not up to par. If there are parts of the plant that have poor equipment grounding conductors, then grounding both ends of a shield can create the missing ground (that should have been achieved by normal grounding methods). That is the main reason that traditionally it has been considered bad practice.

For over 100 years, the US National Electric Code promoted poor building grounding with the idea that an earth connection could be achieved with a wire bolted to a water pipe, or a single rod driven into any type of soil. This simply is not true and caused many problems with things like how to ground cable shielding. Now that many local electrical Code Authorities are adopting the Ufer Grounding System, those problems will almost disappear.

Lancie1 - Maybe I should ask this. I understand what you have stated and thank you. Now may I ask what high frequencies can do to cables that are grounded on a single end?

And the grounding at our facility if awful, so we are a single ended ground but I am always looking for more reasons to give these guys for reasons they should install proper grounds

TW,

I think it is not a black-and-white issue, whether to ground at both ends. Grounding both ends is supposed to be better for removing high-frequency noise. But if it also creates a high-current ground-return loop with potentially dangerous voltages, then nothing will be gained, unless the grounding problem is fixed first by adding a grounding wire somewhere in the plant.

Then what would be an example of high frequency noise and what would be an example of low frequency noise?

I think we need to put it into the perspective that we are not grounding to prevent electrical shock (atleast in this thread). So what are we doing? Draining errant electrical noise/ induction. So I would drain to a one-ended system as mentioned in a previous post.

I'd also be interested to hear from the people more involved in the intrinsically safe arena to see how this applies there.[font=Verdana, Arial, Helvetica, sans-serif]

[/font]

I guess now I am trying to understand the high and low frequency noise. I am taking a wild guess and saying that high frequency noise comes from digital signals and low frequency noise comes from analog signals. This is going off of people saying to ground one end of an analog signal and both ends of a digital signal

There is the need to consider what kind of noise you are trying to eliminate. Single point grounding, is most used with Analog signals, especially those going to a differential type input. Differential inputs are excellent at ignoring 'Common Mode' noise, and often list their specs indicating CMRR (Common Mode Rejection Ratio) in db.
Common mode noise is either inductive or capacitive coupling to both the 'signal' and the 'reference' leads in a circuit.

Digital type inputs, on the other hand, generally have an unfavorable CMRR, mainly due to the speed at which they can switch states. Slight imbalances, or fast spikes, can couple to one channel, and will be detected. This applies mainly to communications type signals, RS-232, RS-422, RS-485, on which almost all communications protocols are based, from 9600 serial comms up to gigabit+ Ethernet.

For digital signals, the shield is usually earthed at each end in order to provide the most complete shunt path for spike noise to earth so that it doesn't impress upon the conductors. This is most needed with RS-232 type signal levels.

High quality Twisted-pair wiring (Twisted-Pair based Ethernet being the most common example) doesn't specifically need shielding, as the cables and terminations are designed to very close tolerances to ensure that Common Mode, low frequency, noise of a reasonable level is impressed on each pair of signal leads identically, and therefore rejected by the receiver. Also, with higher speed communications links, most transceivers don't reference 'common', but rather look at a switching of states between A / ANOT differential signals (one is +5, one -5, or they reverse, so one is -5, one is +5, for example)


Also, for purely digital devices, there is an overall assumption that the communications trancceivers are at basically the same zero potential level, so there is little concern for having the shield actually carry a current. With analog devices, there is a serious problem with the shield carrying a current, as then the shield current will couple to the signal conductors. It's difficult to carry current with only one point of earthing.

Continuing...

Low frequency coupling (50/60Hz) is very VERY detectable by analog signal receivers, and that is exactly the type of coupling you will see if you rely on a cable shield to earth a field device 200 feet away; you essentially create a transformer with a primary (the shield), and one or two secondaries (the signal lead, and possibly the reference lead).

By only earthing the shield at one point, you eliminate the transformer effect, and are left only with capacitive coupling. With almost all shielded cables out there today, induced low frequency noise from shield to conductor capacitive coupling is not an issue, due to the relatively low impedances involved on the transmitter and receiver ends. Inductive coupling though can easily overpower the signals.

Extremely high rise time noise though is a plague to both digital and analog systems. The dV/dT of an AC VFD, or DC Drive armature lead, can punch through almost anything less than 100% shielding. (and thanks to Murphy, always destructively).

For intrinsically safe (IS) circuits the shield should be grounded at only one point (typically in the safe area). If a grounded IS circuit is in a shielded cable the shield should be grounded in the same place as the IS circuit.

Thanks for all the feed back.

To my knowledge, IS circuits a gennerally bonded at the barrier. If you are using a Zener barraier type. Resistance to earth has to be less then 1 ohm acording to CSA. Canadian Electrial code. I think NEC is the Same, as well a IEC. It is necessary for the protection method. So you would have a good ground reference in the panel anyways. But to my understanding if you are using galvanic isolation (ie. opto coupler idea) There is no ground reference necessary at that point. It's used in places where you cannot get less then 1 ohm to ground. In this case you would have to make sure that you can get a good reference to plant ground for your screen. You would bond the screen at the safe side. Ie. in the panel that houses the barrer/isolator. In agreement with Benjo. I stand to be corrected. Again this is to my understanding.

Thanks rdrast for the explination. I can use that as an example when i talk to my students about differential Op-Amps. Good application.

Although i do find this confusing, I think i misunderstand you

For digital signals, the shield is usually earthed at each end in order to provide the most complete shunt path for spike noise to earth so that it doesn't impress upon the conductors. This is most needed with RS-232 type signal levels.

Click to expand...

The spike naturally occures on the signal cable due to the currnt no current, on/off type of fast frequency signals. It is already on the conductors. How does the screen or drain eliminate this? I thought the screen is used to eliminate interfearence both EMI and RFI from outside sources or from conductor pair to conductor pair? I thought that line filters were used to eliminate these spikes electroniclly?

Thanks again for all the responses.

I have asked many of my friends who are industrial electricians as well and they agreed with the statement that "with cables used for analog signals the screen should only be applied to ground at one end."

Is there anyone on here that works with IEC, CSA or or NEC Standard third party approval agency that could maybe shed some light on this. What do the standards exactly state?

Don't forget we are not talking about bonding of cables or equipment exactly. Only noise elimination due to EMI/RFI signals.

TW,
I consider 0 to 1000 HZ to be Low, Above 1000 to be High, but note that one transistor switch can have a "frequency spectrum of the emitted disturbances (breakdown wave-front) ranging from several kilohertz (kHz) to several megahertz (MHz)."



From the Square-D "Electromagnetic Compatibility" manual, here is what that manual says about shielding connected at one end only, and then for both ends:

Shielding connected at one end only

Click to expand...

• Ineffective against external disturbances in HF electric field,​

• Can be used to protect an insulated link (sensor) against LF electric field,​

• Shielding may act as antenna and become resonant​

==> in this case interference is greater than without shielding!​

• Makes it possible to prevent (LF) buzz,​

==> buzz is caused by flow of LF current in the shielding.​

==> this is dangerous and illegal - IEC 364​

Shielding must therefore be protected against direct contact..​

​

​

A large potential difference may exist at the end of the shielding that is not earthed.​

Click to expand...

​

Shielding connected at both ends

Click to expand...

​

• Very effective against external disturbances (high frequency (HF), etc.),​

• Very effective even at resonance frequency of cable,​

• No potential difference between cable and frame connection,​

• Makes it possible to co-locate cables carrying signals of different classes (assuming satisfactory connection (360​

​

°) and equipotential bonding of exposed conductive parts (interconnection, etc.),​

• Very high reducing effect (high frequency (HF)) - is 300,​

• In the case of extremely high-frequency (HF) signals, may induce leakage currents to earth for long cables > 50-100 m.​

​

Very effective​

Because LF and HF equipotential bonding of the site is a golden rule in (EMC),​

shielding is best connected to frame earth at both ends.​

Shielding loses its effectiveness if the length of the cable is too long.​

It is advisable to provide a large number of intermediate connections to frame earth.​

​

​

Click to expand...

​

​

​

​

​

​

very interesting! I'm going to have to look into this a little deeper. It do make sense what you are saying. Thanks for the information.

It seems like there are alot of opinions out there, Looks like the negative aspects of bonding the screen at one end out weighs the benifits.
Interesting. I may have to adjust my thinking on this matter. Thanks for all your comments.

for some cases, say if you have a 4-20mA transmitter in one area and it's connected to the plc input in another area, ground both areas. it's a seperate gounding anyway.

regards

Sherine T.

Just thought i'd let you guys know that I sent an email to CSA inquiring about this issue. I will post their reply when i get it.