A 4-20 mA current loop is supposed to be immune to a lot of electrical trouble, which is exactly why a ground loop on one is so confusing when it appears. Two points in the circuit end up grounded at slightly different voltages, and the difference drives a stray current through the signal wiring that shows up as an offset or a restless, noisy reading. This guide explains how to recognize a ground-loop symptom on a live analog signal, why a signal isolator cures it, and how it differs from an open or shorted loop fault.
4-20 mA ground loop in one line: A ground loop on a 4-20 mA signal occurs when the circuit is grounded at two or more points that sit at different electrical potentials, so the voltage difference pushes an unwanted current through the signal conductors. That stray current adds to or disturbs the intended 4-20 mA, producing a steady offset or a noisy, drifting reading that mimics a process problem. Breaking the extra ground path - typically with a signal isolator - removes the loop and stabilizes the reading.
A 4-20 mA loop is meant to have exactly one ground reference. When a second ground is introduced - the transmitter case grounded at the field, the receiver grounded at the control room, plus a shield or a conduit tying things together - the two grounding points are rarely at precisely the same voltage. Real-world earth is not a single equipotential surface; nearby motors, variable-frequency drives, and power distribution raise or shift the local ground potential from place to place. That difference in potential between the two ground points acts like a small battery in the circuit, and because both points are tied into the signal loop, it drives a current around that unintended path.
The receiver reads current, and it cannot tell the difference between the transmitter's intended signal current and the extra current the ground potential difference injects. The two add together, so the measured value carries an error. If the ground difference is steady, the result is a constant offset - the reading sits a bit high or low across the whole range. If the ground difference fluctuates, as it does when large loads switch on a shared power system, the injected current fluctuates too, and the reading becomes noisy, jittery, or drifts in step with plant electrical activity. The signal that was supposed to be robust is now carrying an electrical artifact that has nothing to do with the process.
The symptom pattern separates a ground loop from the other classic loop faults. An open circuit - a broken wire, a dead transmitter - takes the current to zero or downscale and holds it there; the signal is gone, not corrupted. A short across the loop drives the reading full-scale or erratic in a hard way. A ground loop is subtler: the signal is still present and still tracks the process, but it carries an added offset or an overlay of noise. The reading is wrong or restless, not absent, which is what makes it easy to mistake for a genuine but puzzling process behavior.
Several clues point to a ground loop specifically. The noise or offset tends to correlate with plant electrical activity - it worsens when large drives or motors run and eases when they stop - because those loads are what shift the ground potential. The problem often appears or changes after wiring work, a new ground connection, or equipment added nearby, since that is what creates the second ground. And it frequently affects signals sharing a route or a common ground more than isolated ones. A useful confirming test is to temporarily lift one suspected ground connection under controlled conditions and watch whether the offset or noise disappears; if it does, the extra ground path was the culprit.
The definitive cure for a ground loop is to break the unwanted ground path while still passing the signal, which is exactly what a signal isolator does. An isolator galvanically separates the input side from the output side - typically through an optical or transformer coupling - so there is no direct metallic connection for the ground current to flow through. The 4-20 mA information passes across the isolation barrier faithfully, but the two grounds are no longer tied together, so the potential difference has no closed loop to drive current around. The offset and noise vanish because the mechanism that produced them has been physically interrupted. Good grounding practice - a single, well-chosen ground reference per loop - prevents ground loops in the first place, and isolators handle the cases where a single ground is impractical.
For a monitored operation the payoff is not just a stable reading but avoiding a wrong conclusion. A ground-loop offset can look like a slow calibration drift, and ground-loop noise can look like a genuinely twitchy process, so the fault wastes time being chased in the field or, worse, prompts a reaction to a value that is an electrical artifact. In a cloud SCADA context, where the same analog signals feed remote monitoring, trends, and alarms across many sites, that matters even more, because an operator judging a distant well or facility relies entirely on the number the loop delivers. Clean, isolated signals are what let a remote value be trusted, and recognizing a ground loop for what it is - an installation issue solved with proper grounding and isolation, not a process event - keeps the field team fixing the right thing. This diagnosis complements the component-level pages on the 4-20 mA loop, the signal isolator, and instrument grounding by framing the ground loop as the live failure a technician has to catch.
The strongest clue is that the signal is still present and tracking the process but carries a steady offset or a jittery overlay, and that the disturbance correlates with plant electrical activity - worse when big motors or drives run, better when they stop. Ground loops also tend to appear or change after wiring or grounding work. Temporarily lifting a suspected extra ground under controlled conditions and seeing the noise disappear confirms it.
An isolator galvanically separates the input and output sides of the loop, usually through an optical or transformer coupling, so there is no direct metallic path for the ground current to flow through. The 4-20 mA information still crosses the barrier faithfully, but the two grounds are no longer connected, so the potential difference between them cannot drive a current around the circuit. Breaking that closed path removes the offset and noise at the source.
An open loop - a broken wire or dead transmitter - takes the current to zero or downscale and the signal is simply gone. A ground loop leaves the signal present and still following the process, but corrupts it with an added offset or noise from stray current injected by a ground potential difference. In short, an open loop is a missing signal, while a ground loop is a present-but-contaminated signal, and they call for completely different fixes.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.