Which end of the shield on instrument cable needs to be grounded?

[bdn2004]

For example, a 4-20mA pressure transmitter in the field is wired to a PLC with a #16AWG shielded twisted pair. One end of this cable is grounded, the other is left floating.

What is the purpose of this? And which end is supposed to be grounded? Is there a document that says to do this? Is there a Code requirement?

What would happen if both ends were grounded? Neither?

 

[wroggent]

I believe only one end is grounded because if a potential exists between the two grounds (i.e. locations where ground connection is made) then a circulating current will flow through the shield and ground. A current flow through the shield will result in a voltage drop in the shield. There is some capacitance between the shield and the signal conductors; the electric field between the shield and signal conductors will cause charge to accumulate on the signal conductors; this charge will change the voltage on the signal conductors and thus your sent and received signals are slightly different. I think that's how the story goes anyway. A side note: I read a document from a manufacturer of profibus cable that said both ends of the cable should have the shield grounded. The document recommended installing a parallel grounding conductor along with the cable if noise was an issue. The resistance of the grounding conductor was supposed to be something like 1/10th of the shield resistance. I believe the idea was that the majority of circulating current, if any, would flow through the grounding conductor instead of the shield.

 

[Skogsgurra]

1. Use a ground plane.
2. If that isn't possible - bond.
3. Connect shields of both ends to ground. Always.
4. Ignore all fairy-tales about "hum loops" and such.

Waiting for corrective comments - which I do not intend to read

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[marks1080]

Unless there is some specific reason not to, I would recommend bonding both ends of the cable to ground. Make sure your bonding conductors are as short as possible - rule of thumb is that the total length of bonding conductor used to ground the cable shield should be less than 1% of the length of the cable. If using a cable gland to bond the shield to a ground plane, than the length would be zero, which is ideal.

If you intend to only ground the cable shield at one end it should be done at the end where the circuit wiring is ground referenced. This will only give you protection against capactive coupling at low frequencies... this is why it is not recommended.

 

[Skogsgurra]

By bond, I mean bonding the two structures.
The 1% rule is not very good when cables are getting longer than ten or twenty feet. The general rule is that anything longer than 1" (like a pig-tail) should be avoided. And best practie would be to clamp the screen to a "clean" ground rail. Or use a 360 degree gland.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[bdn2004]

What I was really looking for is an engineering standard or a published recommended standard. Is there one?

 

[wroggent]

IEEE 1100 (emerald book) discusses this. If you have a searchable pdf of that document search the term "both ends".

 

[dpc]

I suspect you can find published guides and references that recommend every possible permutation. You've entered the dark art of "grounding". Beware of "clean" grounds, penisular grounds, "isolated grounds", and the dreaded ground loops. I remember GE engineers spending DAYS trying to find an "illegal" ground connection into their governor cabinet, even going so far as to jack the entire cabinet off the ground and disconnecting every wire connected to it. They took their ground loops very seriously. Did it matter in the end? I have no idea, and neither did they.

 

[waross]

The industry standard in the tar patch is to ground only the supply end and insulate the shield and the drain wire thereafter. Many projects in the tar patch are reported in billions of dollars. On an upcoming project the electrical contract alone is rumored to be one billion dollars. All the 4-20 loops will be grounded at the source end and insulated thereafter.
You may find recommendations in the manufacturers literature.
That said, Gunnar may be correct about the actual results.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[Skogsgurra]

I can understand that a large plant, like a tar sand site, has chosen that road. The reason may be unsufficient research (yes, such things are known to happen) or, more probable, that the distances are so large that effective bonding isn't possible. Then, the screen can be in danger if a potential difference occurs and it is best leave it unconnected "out there".

4-20 mA loops are inherently more or less immune to interference and can be hardened in simple ways (HF chokes, capacitors, isolation amplifiers etcetera) and that makes the "Ground only one end" possible. But I wouldn't even think of it in other applications.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[dpc]

Bill is correct, grounding only one end of the shield for analog circuits has been accepted industry practice in the US for 40 years at least.

 

[rcw retired EE]

Our standard is to ground the instrumentation and communication cable shields at one end and float and isolate the shields and drain wires elsewhere. All cables are twisted shielded pairs or triads. Usually the ground point is a common signal ground bus in the main DCS I/O cabinet. When noise problems occur on circuits, many times we find the shield unintentionally grounded or shorted in a field junction box. Isolating the shield seems to cure the problem 80% of the time. Maybe we are just lucky.

I question this practice of single point ground for shields on CT and control wiring in the 230 kV and 345 kV substations. I think the consensus is 50/50. Half say grounding both ends is the only way, others do as we do and say single point grounding works. We haven't been alerted to any problems with the substation wiring, but maybe that's because we haven't had a major fault that would induce the noise.

Many years ago we had noise on multi-pair communication cables for a new distributed control system using FSK communications. Each shielded pair's shield and drain wires were grounded at both ends. I was troubleshooting with a sound powered headset trying to identify individual pairs by talking with another engineer at the other end. I could not tell which pair he was connected to because the signal was just as strong on all 24 pairs. When we isolated the shields to a single point ground the cross talk disappeared and many of the communication error messages went away.

Until experience proves otherwise: I go with single point shield grounds at one end on instrumentation cables. (Almost all communications are over fiber).

 

[itsmoked]

or

Keith Cress
kcress -

http://www.flaminsystems.com

 

[mikekilroy]

excellent graphic itsmoked!

but with true differential signals (one side of sig not grounded) then the single ground point adding circulating ground current voltage drop Vcm is not true.

another reason why most of today's recommendations is ground both sides.

I was taught the prevailing 1960's thinking: ground source side only, to keep noise OUT; ie., electrostatic shielding for small level control signals (60hz hum etc), vs ground both sides to keep noise IN (electromagnetic shielding) for noisy radiating things like motor leads & RF transmission lines.

Used to be we shielded against that awful 60hz hum everywhere and power sigs from SCRs switching in the 60-1800hz arena.

I believe the advent of ultra high speed switching circuits (sub 0.1 usec,, aka 10-100-300+ mhz) is what has caused the majority of folks mfgring stuff needing shielding to change from 1960 thinking to today's recommendation to shield EVERYTHING at both ends. And with stripping pvc back and wire tying exposed shield to backplane ground. Doing the math one can see a short 6-12" little #12 wire pigtail from shield to ground has a 1/4 wavelength in the area of these switching frequencies and their harmonics, meaning their resistance at those freqs are infinite so might as well not bother. Our engineers began recommending all signal and power leads grounded both ends about 10 yeara ago and we never looked back. I think this is new enough that it has not become the new standard everywhere yet, but will. And with all those older engineer books in print, the signal side grounding will still take a long time to go away. My opinion.

http://www.KilroyWasHere<dot>com

 

[mikekilroy]

wish we had an edit button.... dang typos:

from SCRs switching in the 60-1800hz arena.

meant to be:

from SCRs switching in the 60-180hz arena.

http://www.KilroyWasHere<dot>com