Few field questions start more arguments than where to ground a cable shield. Ground it at one end and you avoid the ground-loop current that can inject noise, which is why single-end grounding is the classic rule for low-frequency analog signals. Ground it at both ends and you get better immunity against high-frequency interference, which matters for fast digital and communication cables. This page lays out why each answer exists, the physics that makes them conflict, and a practical way to decide which rule applies to the cable in front of you.
Shield Ground One vs Both Ends in one line: For low-frequency analog signals such as 4-20 mA loops, a cable shield is normally grounded at one end only, usually the control-room end, so no ground-loop current flows through the shield and injects noise. For high-frequency interference, grounding the shield at both ends gives better immunity, so fast digital and communication cables often ground both ends. The right choice depends mainly on the signal's frequency and the difference in ground potential between the two ends.
The case for grounding a shield at a single end comes down to ground loops. The two ends of a long cable are rarely at exactly the same ground potential; there is usually a small voltage difference between the control-room ground and the field ground. If the shield is bonded to ground at both ends, that potential difference drives a current along the shield, and at power-line frequencies that current couples noise straight into the signal conductors it is supposed to protect. Grounding the shield at only one end breaks the loop so no such current can flow.
For low-level analog measurements this is decisive. A 4-20 mA loop or a millivolt sensor signal is exactly the kind of slow, small signal that power-frequency ground-loop noise corrupts most visibly, showing up as hum or drift in the reading. The shield still does its job of intercepting capacitively coupled interference and draining it to ground through its single connection, but with the loop open it cannot itself become a noise source. That is why the standard rule for instrument signal cable is one-end grounding, conventionally at the control-room or supply end.
The other half of the rule is that the far end must be genuinely isolated. If the field end of the shield is grounded, even accidentally through a metal gland, a fitting, or a device case touching ground, the loop is quietly re-formed and the noise returns. Single-end grounding therefore depends on carrying the shield continuously and insulating it wherever it is not meant to be grounded, so the intended single ground point really is the only one.
The case for grounding both ends is about high-frequency interference. At high frequencies the mechanism that hurts you at power line frequency, the ground-loop current, matters far less, because a shield grounded at both ends can carry the return current that lets it properly cancel magnetically coupled and radiated interference. A shield grounded at only one end behaves poorly at high frequency, effectively acting like an antenna over part of its length rather than a fully working screen, so for fast signals both-ends grounding gives cleaner results.
This is why the rule flips for cable types where high-frequency immunity dominates. High-speed data links, communication buses, and cables running through electrically noisy, radio-frequency-rich environments generally do better with the shield bonded at both ends. The concern shifts from a slow power-frequency ground loop to fast interference that a single-end shield cannot properly reject, and the both-ends connection is what lets the shield perform across the frequency range that matters for those signals.
Where a ground-loop current at both ends is a genuine worry but high-frequency performance is still needed, engineers use compromises rather than picking one extreme. Grounding one end solidly and connecting the other end to ground through a small capacitor gives a hybrid: the capacitor blocks the low-frequency loop current while still providing a low-impedance ground path at high frequency. Techniques like this exist precisely because the one-end and both-ends rules each solve a real problem, and some cables need a bit of both.
In practice the decision follows the signal. Low-frequency analog instrument loops, the bulk of traditional 4-20 mA and sensor wiring, default to single-end grounding at the control-room end, and the field end is insulated. Fast digital, communication, and network cables lean toward both-ends grounding for high-frequency immunity. When both concerns apply, or the two grounds are known to sit at meaningfully different potentials, a hybrid capacitor termination is the pragmatic middle ground. The governing questions are how fast the signal is and how far apart the two ends' ground potentials are.
Consistency and documentation matter as much as the rule itself. A shield grounded at both ends by accident, or a run where half the shields are done one way and half the other, produces intermittent noise that is miserable to chase. Deciding the convention for a given cable type, carrying the shield continuously through junction boxes, and insulating the ends that are supposed to float is what keeps the intended scheme intact over the life of the installation, especially when different crews touch the wiring over the years.
The reason any of this reaches a monitoring team is that shield grounding lives entirely in the physical layer beneath the control system, and getting it wrong shows up downstream as a noisy or drifting tag. A cloud SCADA platform such as Merobix simply reads the value a controller reports after the field cabling has, or has not, delivered a clean signal to its input. When a remote analog point is jittery for no obvious process reason, an incorrectly grounded shield, most often a shield accidentally grounded at both ends, is a classic root cause to check before suspecting the instrument itself.
For low-frequency analog signals such as 4-20 mA loops, ground the shield at one end only, conventionally the control-room end, and insulate the far end so no ground-loop current flows through the shield. For high-frequency digital or communication cables, grounding both ends usually gives better immunity. The deciding factors are the signal's frequency and how different the ground potentials at the two ends are.
The two ends of a long cable are usually at slightly different ground potentials, and grounding the shield at both ends lets that difference drive a current along the shield. At power-line frequencies that ground-loop current couples noise into the signal conductors the shield is meant to protect. Grounding at only one end breaks the loop, which is why single-end grounding is standard for slow analog signals.
Yes, a common compromise is to ground the shield solidly at one end and connect the other end to ground through a small capacitor. The capacitor blocks the low-frequency ground-loop current while still giving the shield a low-impedance path to ground at high frequency. This hybrid termination is used when a cable needs to avoid a power-frequency loop but still reject high-frequency interference.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.