Multi Pair Overall Shield Grounding

[Leyden]

On a multi-pair cable where each individual pair has its own isolated shield and there is one overall shield around all of the pairs, why do we ground the overall shield at both ends and the individual shields only at the cable origin. I can't find IEEE 518 right now, but i believe it tells you to do this. And I know 518 is withdrawn.

Thank you,

 

[gnurf]

On a multi-pair cable where each individual pair has its own isolated shield and there is one overall shield around all of the pairs, why do we ground the overall shield at both ends and the individual shields only at the cable origin. I can't find IEEE 518 right now, but i believe it tells you to do this. And I know 518 is withdrawn.

We don't. Or at least not where I'm from. I'm not familiar with IEEE 518 (is it old?), but the only case where grounding only one end of the cable makes sense is to break a low-frequency ground loop. However, grounding only one side may--depending on cable length (measured in wavelengths) among other things--create radio-frequency (read: high-frequency) interference problems. The outer shield, which is grounded in both ends, aims to tackle this problem. So you get the best of both worlds, of sorts.

Edit: Expanded the answer slightly.

 

[jim hardy]

There are varying schools of thought and various approaches to measuring signals that dictate different grounding schemes.

This book is a pretty good introduction.
https://www.amazon.com/dp/0471029920/?tag=pfamazon01-20

The school of thought i grew up on said "It makes no sense to ground a shield if the signal itself is not grounded. Enclose the whole measurement system in a Faraday cage connected to the low side of the signal at the point of measurement.".

This works exceedingly well when there's significant common mode voltage present.
Our plant computer used three pole relay multiplexers with floating guarded amplifier. It worked beautifully ,

 

[jrmichler]

I once worked in a paper mill, where many miles of low voltage control signal wiring had to coexisting with more miles of 440 and 4160 volt three phase wiring. The mandate from the electrical engineers was simple: "All control wiring will be in shielded twisted pair inside rigid galvanized conduit. Control wiring will not share cable tray with high voltage wiring."

It worked. I do not recall any cases where interference was a problem.

 

[Leyden]

We don't. Or at least not where I'm from. I'm not familiar with IEEE 518 (is it old?), but the only case where grounding only one end of the cable makes sense is to break a low-frequency ground loop. However, grounding only one side may--depending on cable length (measured in wavelengths) among other things--create radio-frequency (read: high-frequency) interference problems. The outer shield, which is grounded in both ends, aims to tackle this problem. So you get the best of both worlds, of sorts.

Edit: Expanded the answer slightly.

ieee-518 1982

i'm talking about instrumentation cable, I am not sure what other applications this type of cable is used in. you don't ground both ends of shield? for instrumentation cable? i know shielded power cable gets both ends, but every standard i have read on the subject says ground the individual shields only at one end.

thanks

 

[Leyden]

There are varying schools of thought and various approaches to measuring signals that dictate different grounding schemes.

This book is a pretty good introduction.
https://www.amazon.com/dp/0471029920/?tag=pfamazon01-20

The school of thought i grew up on said "It makes no sense to ground a shield if the signal itself is not grounded. Enclose the whole measurement system in a Faraday cage connected to the low side of the signal at the point of measurement.".



View attachment 230129

This works exceedingly well when there's significant common mode voltage present.
Our plant computer used three pole relay multiplexers with floating guarded amplifier. It worked beautifully ,

the signals are grounded, i ordered the book. so are you in the ground both ends of the overall shield group?

thanks. you are very helpful on this site

 

[jim hardy]

the signals are grounded, i ordered the book. so are you in the ground both ends of the overall shield group?

I hope you find Morrison's book helpful.

EDIT Short answer - I'm in the 'single point ground' group.
Long Answer - read on and yawn. (I'm the official PF bore)
.......................
Well, as a maintenance man you sort of get what the designers gave you.

Since our plant computer was a DIY project we had the latitude to do it the way we wanted.
I prefer shield tied to low side of signal and signal grounded only at point of measurement, per Morrison, and we wired the computer that way .
Our multiplexers used three pole relays to switch signal hi lo and shield to ADC's hi lo and guard lines.
ADC and its adjustable gain amplifier floated at common mode potential of each signal being measured, most of which were very near zero by our choice of where to tie into the individual current loops...
The relays were mercury wetted so had essentially infinite life. Indeed many of them clicked away once per second from 1972 to 2002 , accruing about 10^9 operations..
So our computer shielding was 'grounded' per that sketch i made which is what Morrison recommends in that book you ordered.
It really performed well reading thermocouples

Most instrument system grounding schemes don't agree that closely with Morrison's suggestions.
Our process instrumentation, 4 to 20 milliamp current loops, had shields grounded at one end only - where they entered the electronic racks in the control room.
That worked okay. But we spent a lot of hours at startup chasing 60 cycle noise out of the system mostly from accidental grounds that made loops. There's a lot of 60 hz magnetic flux running around a power plant and any conductors that encircle it will see induced voltage. Single point grounding helps there.
The impedances of those 4 to 20 milliamp loops was low enough, and the signal levels high enough, that capacitive coupling didn't cause much trouble.

So i guess that's a long winded answer to your question. Broke my own rule - explained "Why " before "What".

Brief answer is - i prefer single point grounding.
When you get into the field you will see a hodgepodge of grounding approaches . Some work better than others.

The basic thing to keep in mind is - you want to minimize voltage between shield and signal lines so that no(as little as possible) current flows through shield to signal capacitance into those signal lines.
"Driven Shield" becomes obvious once you realize that. But my first encounter with a "Driven Shield" sure baffled me. When i read Morrison's book the light came on - "So THAT"S what they were doing !"

Nothing is that simple, though.
Perhaps somebody with radar or radio experience will add some words about shielding Radio Frequency systems. It is my understanding that in Radio World you should ground your shield every quarter wavelength ?

Sorry about the ramble. Just i think if you have a grasp of the fundamentals it helps you solve problems when you get into the field. So i try to introduce them.

old jim

 

[Leyden]

I hope you find Morrison's book helpful.

EDIT Short answer - I'm in the 'single point ground' group.
Long Answer - read on and yawn. (I'm the official PF bore)
.......................
Well, as a maintenance man you sort of get what the designers gave you.

Since our plant computer was a DIY project we had the latitude to do it the way we wanted.
I prefer shield tied to low side of signal and signal grounded only at point of measurement, per Morrison, and we wired the computer that way .
Our multiplexers used three pole relays to switch signal hi lo and shield to ADC's hi lo and guard lines.
ADC and its adjustable gain amplifier floated at common mode potential of each signal being measured, most of which were very near zero by our choice of where to tie into the individual current loops...
The relays were mercury wetted so had essentially infinite life. Indeed many of them clicked away once per second from 1972 to 2002 , accruing about 10^9 operations..
So our computer shielding was 'grounded' per that sketch i made which is what Morrison recommends in that book you ordered.
It really performed well reading thermocouples

Most instrument system grounding schemes don't agree that closely with Morrison's suggestions.
Our process instrumentation, 4 to 20 milliamp current loops, had shields grounded at one end only - where they entered the electronic racks in the control room.
That worked okay. But we spent a lot of hours at startup chasing 60 cycle noise out of the system mostly from accidental grounds that made loops. There's a lot of 60 hz magnetic flux running around a power plant and any conductors that encircle it will see induced voltage. Single point grounding helps there.
The impedances of those 4 to 20 milliamp loops was low enough, and the signal levels high enough, that capacitive coupling didn't cause much trouble.

So i guess that's a long winded answer to your question. Broke my own rule - explained "Why " before "What".

Brief answer is - i prefer single point grounding.
When you get into the field you will see a hodgepodge of grounding approaches . Some work better than others.

The basic thing to keep in mind is - you want to minimize voltage between shield and signal lines so that no(as little as possible) current flows through shield to signal capacitance into those signal lines.
"Driven Shield" becomes obvious once you realize that. But my first encounter with a "Driven Shield" sure baffled me. When i read Morrison's book the light came on - "So THAT"S what they were doing !"

Nothing is that simple, though.
Perhaps somebody with radar or radio experience will add some words about shielding Radio Frequency systems. It is my understanding that in Radio World you should ground your shield every quarter wavelength ?

Sorry about the ramble. Just i think if you have a grasp of the fundamentals it helps you solve problems when you get into the field. So i try to introduce them.

old jim

Thanks!
that's not a ramble at all, i greatly appreciate it.

one thing that's standing out to me though is you didn't really reference the fact that i am using two shield's
the inner shield i agree only ground where signal is grounded(only one end grounded), but then i also have an outer shield independent and not connected to inner shield except where signal is grounded and the outer shield is also grounded at other end of cable. what i have been reading is the inner shield helps with low frequency interference while the outer shield helps with higher frequency. I'm still researching this as you can tell but that is what i hav been finding, nothing crystal clear though seams like a lot of opinions and theories rather than actual experimental data.

thanks

 

[jim hardy]

My plant was consistent . All shields were grounded at the entrance to the instrumentation racks.

We had two instruments that used dual shields. I'll describe them.

One was the reactor neutron monitors. A couple dozen cables carried signals from neutron detectrs adjacent the reactor to the instrument racks in the control room, a couple hundred feet.
Both shields of each cable were connected to chassis at the back of the instrument rack where the triaxial connector plugged in. (Amphenol RF series, 53175 to best of my alleged memory)
The long long cables running out to the reactor ran in dedicated 4 inch steel conduits .
That conduit was mounted with insulated clamps all the way to the reactor and grounded only at the control room end, to the bottom of the instrument rack. So it made an effective third electrostatic shield and being steel provided considerable magnetic shielding as well.
The outer shield of the triaxial cable ended at the reactor.
The inner shield of the triaxial cable became the outer skin of the neutron detector.
The detectors themselves were mounted on ceramic insulators.
So we had Morrison's "Faraday Shield" surrounding the detector and measurement system, just it was grounded not way out at the point of measurement but instead at the entrance to the instrument. It worked well.

The other instrument was a Foxboro magnetic flowmeter. It worked on the principle of voltage induced in a moving fluid by an AC magnetic field. Not unlike Tom Clancy's "Caterpillar Drive" in Red October.
It produced a zero to three or four millivolt AC signal at line frequency that had to be delivered faithfully to a receiver in the control room instrument rack , again a couple hundred feet away, traversing an electrically noisy power plant environment all that distance !
I didn't believe it could possibly work.
As best i recall ( i studied it because it was such a marvel , but this was 1971 and details fade - i'll do my best)
That instrument's cable had two signal conductors each with its own driven shield. Of course driven shields aren't grounded.
Those two shielded signal cables were surrounded as a pair by two more shields.
The inner of those two shields connected the flowmeter's internal shield to the receiving amplifier's guard .
The outer of them connected to the flowmeter's case to intercept capacitive ground currents keeping them out of the inner shield. (I think it was grounded only at that far end so as to not make a ground loop but i could be wrong)
That leaves only the driven shields.
Each of those was driven by a unity gain follower in the receiver to the same voltage as the signal wire it surrounded - thereby achieving Morrison's goal of no voltage between the signal wire and its shield.
Voila !
Amazingly it worked !
I admit though it was sophisticated , employing synchronous demodulation before i'd ever heard the term. So It had an adjustment for phase difference between AC power at the flowmeter and at its receiver. A testament to the designers (and physicists) of early 1960's.

Anyhow - there are the two examples of multi-shielded signals in my experience. Both were single point grounded.
My second one sounds rather like one you described

but then i also have an outer shield independent and not connected to inner shield except where signal is grounded and the outer shield is also grounded at other end of cable.

and i can't quite remember what they did with outer shield at receiver.

Will search for an old Foxboro instruction manual.

I hope above helps.. Really it's only moral support, as advice goes it's overpriced at two cents.

"Believe in your basics" is the good advice. .

old jim

 

[jim hardy]

and i can't quite remember what they did with outer shield at receiver.

Will search for an old Foxboro instruction manual.

Well it looks like my memory probably swapped something.
This is a newer Foxboro manual
but it still uses driven shields plus two more
outer not connected at transmitter end .
http://apps.watersurplus.com/techlib/Foxboro/Foxboro_Transmitter_IMT25_instr_D1297.pdf

So i think
My memory most likely had it connected at wrong end. (That's not unusual).
It was probably grounded at receiver end like yours.

Thanks for your patience.

I hope this triggers some technical discussion and perhaps reference to standards ?

old jim

 

[Dullard]

There is another (common where I live) issue with grounding a shield at both ends: A lightning ground-strike near the facility may turn the shield into a filament, due to ground potential differences. I just dealt with one of those 2 weeks ago.
In my experience, the most annoying thing about shields grounded at both ends is that they can add more noise as they subtract, but intermittently - that's often on 3rd shift when someone operates a piece of equipment not usually present, or grounds their welder... Single-point grounding doesn't completely eliminate this, but it certainly reduces susceptibility. The quality of the facility ground system matters (a lot). The 'ground' required for NEC compliance (safety) is often not god enough for instrument shielding, where higher frequencies are an issue. Sometimes terminating a shield at both ends 'works better' - it really depends on the nature of the noise and the sensitivity of the shielded signal.

 

[Leyden]

?518? "1) The twisted pair shield should be grounded at one end only. This ground shall be the instrument ground . It is critical that this shield be held to the instrument input reference ground potential. 2) The overall shield must be grounded at both ends, and at appropriate points along the cable. The rule of thumb is to ground at internal of 1/8 the wave length of the worst case expected radio frequency interfering noise. This may prove to be either impractical and/or undefined, thus the recommendation is to provide multiple grounds by grounding the outer shield at convenient locations where the shield(s) are exposed such as at pull boxes, junction boxes and termination cabinets. Failure to ground at both ends and, to a lesser extent at intermediate points, renders that outer shield useless and in some cases can cause interference/false operation. The outer shield is to be grounded to the safety ground not the instrument ground. 4) The use of metallic conduit, continuously welded metallic jacket, or covered trays is recommended as a means of providing additional grounded shielding. Note: The above shielding recommendations assume the use of instrument control/signal wire that is twisted pair, double shielded (each pair shielded with an overall cable shield). The shields shall be aluminum Mylar tape with complete coverage"my thinking on the outer shield grounded at both ends is, its better to run a cable in its own conduit than running it in open. the inner shield is only grounded where signal is grounded so that will protect the signal from circulating ground currents on the outer shield, the outer shield grounded at both ends will protect from other interference. i know mylar thin shield is not as good as rigid conduit but i think it works well for high frequency if grounded frequently. any thoughts on this?IEEE 1100 IEEE Recommended Practice for Powering and Grounding Electronic Equipment
i can't copy the text in my pdf
says grounding at one end does not protect signal from near field EMI, i believe they are talking about long runs of cable.

im trying to find IEEE 1143 IEEE Guide on Shielding Practice for Low Voltage Cables
and 518 (withdrawn(i think superseded by the above two))IEEE Guide for the Installation of Electrical Equipment to Minimize Electrical Noise Inputs to Controllers from External Sourceshttps://www.emcstandards.co.uk/cable-shield-grounded-at-one-end-only
https://www.celeramotion.com/sites/default/files/TN-1202_Grounding_and_Shielding_Recommendations.pdf

please remember i am talking about a cable that is shielded with two independent shields, inner is grounded where signal is, outer is grounded at every cable junction.

whats your thoughts on this? i appreciate anything even if your not sure.

 

[jim hardy]

whats your thoughts on this? i appreciate anything even if your not sure.

There are times you have to choose how far short of theoretical perfection you are going to stop in order to accommodate the practical.

I have seen the results of welding current trying to return through a shield. (Somebody mentioned that earlier.)
Don't design that trap into a system for some unfortunate workingman will fall into it.

My control room's instrument ground was kept separate from the plant's equipment ground specifically to keep electrical noise out of the control room. .
Instrument ground was a red cable in the concrete floor served by its own two dedicated ground rods , so was connected to the plant equipment ground mat only by the moist Earth eighteen feet below the plant. It was red to be recognizable as instrument ground.
In my magnetic flowmeter example above the meter body is bolted into process piping which is grounded locally to plant equipment ground some hundreds of feet from the receiver in the control room.
Grounding both ends of that flowmeter's shield would have tied control room's red instrument ground to that distant equipment ground making a huge loop. I wouldn't do that.

Your Celeramotion folks seem to agree

I never worked in a radio or radar installation so can't advise you there.

 

[Leyden]

i thought about this when i was writing and i guess i should have clarified, i said outer shield grounded where inner is grounded but i should have clarified i was misleading, it will not be landed on the instrument ground it will be landed on the safety ground/chassis ground/ common Earth ground. the inner shield will be connected to a cleaner dedicated ground. but both are in the same cabinet or room. this cable will be ran in a well grounded and bonded cable tray which will have additional bonding jumpers to the equipment the cables are running to.

with the welding comment, there are extra grounding and bonding requirements in the field to help prevent this and of course there is the rule the ground lead be clamped on the work being welded.

thanks again, i know i will not likely come to a perfect answer, but I'm going to try to come close. i know the only way to know if it is going to work in the scenario is actually doing it.

Thanks

 

[Leyden]

Your Celeramotion folks seem to agree
View attachment 230370

.

did you see the part after that though? it recommends both ends but with capacitor, and i think those are for single shielded cable, but it doesn't say it either way so i don't know. page 3 figure 2 shows outer shield grounded at both ends and inner only grounded where signal is grounded

i should have also added that this is for an industrial environment with many motors, PWM drives, lighting, welding, radio(not a radio station but typical industrial radio communication) and possibly other noise sources I'm trying to think of. my thinking on leaving out some of these details was to try to avoid cluttering my posts and deterring comments.
thanks,

 

[Leyden]

i guess i should also add, the shields have a parallel drain wire ran with them, i will look up the sizes but i think they are the same size as the signal conductors or else very close.

 

[jim hardy]

it will not be landed on the instrument ground it will be landed on the safety ground/chassis ground/ common Earth ground. the inner shield will be connected to a cleaner dedicated ground. but both are in the same cabinet or room. this cable will be ran in a well grounded and bonded cable tray which will have additional bonding jumpers to the equipment the cables are running to.

It's going to work fine, i believe.
If you get noise try to eliminate it at its source.
These are good for snubbing relay coils
http://www.cde.com/resources/catalogs/Q-QRL.pdf

An old fashioned pocket transistor radio set to AM and tuned between stations is handy for localizing RFI.
I roamed the power plant with one long enough to recognize when any individual motor's 'voice' changed.
Rainy weather makes questionable insulation show up. You hear its corona on the radio, but it's only qualitative . . Good enough to schedule an insulation check on it though.

Good luck in your industrial career. You seem a practical type who'l be an asset .

regards,

old jim

 

[Leyden]

1100
-says best to ground both ends except sometimes with low frequency common mode dc signals. cable shields can carry unwanted ground current, use a blocking device except at one ground termination (silicon avalanche diodes or ac capacitor rather than rectifier diodes)
-long runs should have shield grounded at multiple points, or use SPD to ground everywhere but source
-electromagnetic shielding ground at both ends1143
-grounding both ends allows a counter current that cancels out interference

-when grounding both ends consideration should be given to the shield thickness, need shield thick enough for voltage gradients from faults, 0.19mm aluminum foil shield is usually thick enough

518
-it may be neccesary to ground the second overall shield at multiple points

-interesting side note, half of unused pairs and shields should be terminated at one end and the other half at the othermy summary
-grounding at one end provides electrostatic shielding, low frequency shielding.

-for grounding at both ends without an extra device you need shield of some thickness(probably 0.19mm for aluminum 0.15mm copper) and if its thick enough to withstand voltage gradients under a fault it is likely better to ground at both ends. the permeability of the shield is important for electromagnetic interference so thickness will play a role here as well. normal thin aluminum shields probably won't do much for electromagnetic shielding.

-grounding the shield at both points will still retain the electrostatic shielding ability

-don't leave shield floating it will enhance electrostatic field coupling

 

[jim hardy]

Be careful about grounding both ends.
We had a huge cable that passed through metering current transformers for a diesel generator.
Ir was covered with a thick armor sheath made of lead(noun metal, not verb precede)
For years we wondered why that particular diesel's exhaust manifold was red hot at full power. And why it needed frequent overhauls.

When we found somebody had grounded the armor sheath at both ends, making an additional conductor through the current transformer, we said "Uh oh".
Ungrounding one end of the sheath raised current indication perhaps twenty percent. And megawatt indication as well...

We had been abusing that poor engine for years.

Moral - beware of making ground loops with your shields..

 

[Leyden]

Be careful about grounding both ends.
We had a huge cable that passed through metering current transformers for a diesel generator.
Ir was covered with a thick armor sheath made of lead(noun metal, not verb precede)
For years we wondered why that particular diesel's exhaust manifold was red hot at full power. And why it needed frequent overhauls.

When we found somebody had grounded the armor sheath at both ends, making an additional conductor through the current transformer, we said "Uh oh".
Ungrounding one end of the sheath raised current indication perhaps twenty percent. And megawatt indication as well...

We had been abusing that poor engine for years.

Moral - beware of making ground loops with your shields..

Interesting, thanks. I've never heard of that problem. although i would think your actual root problem may have been the ground current the extra current was from, it sounds like it may have been a decent amount, whether a ground fault or some unintentional induction.

 

[jim hardy]

i would think your actual root problem may have been the ground current the extra current was from, it sounds like it may have been a decent amount, whether a ground fault or some unintentional induction.

Yes, induction. The sheath made one shorted turn around the CT core, in effect another winding.

 

[Leyden]

Yes, induction. The sheath made one shorted turn around the CT core, in effect another winding.

ahh, i see(kind of). I've got to wrap my head around it a bit, we run all of our shield drains for MV cable through the ct's and the shields are grounded at both ends, I've never heard of a problem(which doesn't mean much). I'm trying to figure it out but i wouldn't think one turn around the CT would get that much current, maybe back in the day they used smaller ratios? i see ct ratios of 600:5, 1500:5 maybe 200:5 on new installs (current running through cable in ct:ct secondary)

edit: never-mind, the shield goes through and then the drain back through. and much equipment has the ct's on the factory wiring before we get to the shielded cable anyway

 

[jim hardy]

edit: never-mind, the shield goes through and then the drain back through.

yep,
if shield current flows through the donut hole twice,
one direction on the shield and other direction on the drain,
they cancel.

How the simple things of the Earth confound the mighty, eh ?

old jim