Why a Shared Common Causes Symptoms That Make No Sense

Why this matters

The hardest control calls are the ones where three devices misbehave at once and none of them have anything to do with each other. A damper sits short of its commanded position, a transmitter reads high, a binary input chatters, and every one of those devices tests perfectly on the bench. Techs replace parts in rotation for two visits and the symptoms move around but never leave.

Almost every one of those calls is the same fault, and it is not in any of the devices. It is in the conductor they all share to get back to the source. A shared return makes every device on it a component of every other device's circuit, and the symptoms it produces are strange in a specific, recognizable way: they point in opposite directions on devices that are wired identically.

Before you put a meter on a shared return

De-energize the panel's line-voltage supply, lock and tag it, and prove dead before opening it, under 29 CFR 1910.333(b)(2), with live-dead-live per NFPA 70E-2021, 120.5. The measurements in this card have to be taken with the control circuit energized to mean anything, which is the narrow troubleshooting allowance at 29 CFR 1910.333(a)(1), taken with meter and leads rated for the circuit's measurement category and available fault current plus the shock and arc-flash protection the assessment calls for. Control terminals and line-voltage terminals share most panels.

Do not lift a shared common while the circuit is energized. Every device referenced to it loses its reference at once, outputs go to states nobody predicted, and on an AC control circuit an interrupted return can leave the supply potential sitting on the conductor you are holding. Isolate first, then lift. If a current transformer secondary shares that terminal block, its secondary is never opened while primary current flows, because an open secondary develops dangerous voltage - short it at its shorting block first.

The call: three complaints that did not belong together

A rooftop unit with a panel-mounted controller and several field devices on a 24 volt DC control common. Three complaints on one ticket, from two visits by two techs.

• The mixed air damper sits noticeably short of where the graphic says it should be, consistently, in both directions of travel.
• A duct pressure transmitter reads high against a reference gauge, by an amount that seems to change with what else is running.
• A binary input flickers between states while the unit is in full operation and is rock solid when the unit is off.

Nothing connects a damper actuator, a pressure transmitter and a contact input. They are different manufacturers, different functions, different points on the controller.

The two obvious explanations, and the arithmetic that killed them

The actuator is failing. Bench it: fed 0 to 10 volts from a hand source with its own return, the actuator tracks the full sweep against a fixed reference mark, within its published tolerance. It also runs correctly on the unit when driven from that same hand source. So the actuator responds correctly to a signal, when the signal is referenced the way the source intends.

Voltage drop in the signal wire. This is the reflex answer and it is wrong by orders of magnitude, which is worth working out once so you stop reaching for it. The actuator's voltage input is high impedance, on the order of 100 kilohms for this class of device, so at a 4.80 volt command the current in the signal conductor is 4.80 divided by 100,000, which is 48 microamps. The run is 60 feet of 18 AWG copper; at about 6.4 ohms per 1000 feet at 68 F for solid copper, rising roughly 0.4 percent per degree Celsius above that and a few percent higher again for stranded conductor of the same gauge, 60 feet is 0.38 ohms. Multiply: 48 microamps through 0.38 ohms drops about 18 microvolts. Eighteen millionths of a volt on a signal where the symptom is worth about a volt.

The signal conductor is innocent, and the arithmetic proves it in one line. The return conductor is a different conductor with a completely different current in it.

What a return conductor actually is

A common is not a place. It is a wire, with resistance, carrying the summed current of everything that uses it.

At the panel end, the common terminal is the source's reference and it is by definition at zero. At the field end, 60 feet away, the same conductor is carrying the return current of every device out there. Current through resistance is a voltage, so the field end of that conductor is not at zero. It sits above the panel end by the product of the current and the conductor's resistance.

Measure the panel's common terminal against the field junction box's common terminal with everything energized and you get a number that should be zero and is not. On this unit it read 0.96 volts.

That number is exactly what the arithmetic predicts. The four modulating actuators, the relay bank and the two transmitters on that common together draw about 2.5 amps measured at the panel; 2.5 amps through 0.38 ohms is 0.96 volts. That arithmetic holds cleanly because this is a DC common. On a 24 volt AC common the same mechanism operates, but the individual coil currents are not in phase with each other, so a straight arithmetic sum of the branch currents is a worst case and the real offset is somewhat lower.

Why the two symptoms pointed in opposite directions

This is the part that makes the call look impossible, and it falls straight out once you know the offset exists.

A signal going out reads low at the field device. The controller puts 4.80 volts on the wire, referenced to the panel common, which is zero. The actuator measures that wire against its own common terminal, which sits at 0.96 volts. It sees 4.80 minus 0.96, which is 3.84 volts. On a 0 to 10 volt span that is 9.6 percent of full scale, so a damper commanded to 48 percent parks at 38.4 percent. Consistently, in both directions of travel, exactly as reported.

A signal coming in reads high at the controller. The pressure transmitter produces its output referenced to its own common, which is that same 0.96 volt point. If it is genuinely producing 6.00 volts, its output terminal sits at 6.96 volts relative to the panel common. The controller measures against the panel common and reads 6.96. The transmitter is honest and the reading is 0.96 volts high.

Same offset, same magnitude, opposite sign, decided entirely by which end of the loop the signal originates at. Once you have that rule, a tech can predict the direction of every symptom on the panel before measuring the next one, which is what makes this diagnosis fast the second time you meet it.

The binary input follows from the same offset. A contact input decides its state by whether a terminal is pulled near the common, and a 0.96 volt pedestal that moves whenever a load switches puts a marginal input near its threshold. It settles when the unit is off because the offset collapses with the current.

The confirming measurement, predicted before it was taken

A measurement that only agrees with a theory afterward is weak. Predict the number first.

One of the modulating actuators on that common draws 0.9 amps measured at its own conductor. If the offset is the product of summed current and return resistance, dropping that one load should reduce the offset by 0.9 times 0.38, which is 0.345 volts, taking the common-to-common reading from 0.96 volts to about 0.61 volts.

Isolating that actuator's supply and re-measuring gave 0.62 volts. Within the resolution of the meter and the tolerance of the resistance estimate, the prediction held, and at that point the shared common is not a hypothesis any more.

The damper's error moved with it, as it must: at a 0.62 volt offset the actuator commanded to 48 percent sits near 41.8 percent rather than 38.4. That second confirmation costs nothing and it closes the loop between the electrical measurement and the mechanical complaint.

What the fix was, and what it was not

Upsizing the conductor is a mitigation, not a cure. Replacing the 18 AWG return with 14 AWG, at about 2.5 ohms per 1000 feet at 68 F, cuts the resistance by a factor of about 2.5, which would take the 0.96 volt offset to roughly 0.38 volts. Better. Still 3.8 percent of a 10 volt span, still enough to matter on a modulating loop, and it grows again the day someone adds another device to that common.

The cure is to stop sharing the path. Give the signal group its own return conductor back to the source, landed at the same point the signals are referenced to, so that no load current flows in the conductor the measurements are made against. That is what a star or single-point return arrangement buys and it is the whole reason the practice exists.

The other cure is to pick a signal type that does not care. A two-wire current loop carries the same current through every element in series, so a difference in reference potential between the two ends changes nothing that arrives. Where a run is long, shared, or likely to grow, that immunity is worth more than the resolution the offset range costs. A sibling card covers the signal types and their trade-offs.

The failure mode this replaces. Three devices had been swapped across two visits, all three tested good on the bench, and the symptom persisted with new parts because the new parts referenced the same conductor. The tell that should have redirected the first visit was the word "consistently" in the damper complaint: a device that is wrong by the same amount every time is being offset, not failing.

Where this mechanism does not apply

Where nothing shares the return. A device with its own dedicated pair back to the source has no other current in its return, so there is no offset to create, whatever the run length. This is why one field device on a long home run behaves perfectly while its neighbor on a shared trunk does not.

Where the signal is differential. A receiver that measures between two dedicated conductors rather than between a conductor and a common rejects a shift that appears equally on both. That rejection has a stated limit, the common-mode range of the input, and beyond it the rejection stops abruptly rather than degrading, so a large offset can push a differential input out of range and produce a reading that looks like a failure rather than an error.

Where the offset is large enough to be something else. An offset in the range of a volt is a shared return. An offset approaching the supply voltage is not this mechanism at all, it is a broken or high-resistance return, and that is a different and more urgent fault: devices downstream of a failing return can find an alternate path through signal wiring and be destroyed by it. Treat a large common-to-common reading as a stop-and-isolate condition rather than a measurement to refine.

References

• 29 CFR 1910.333(a)(1) - live parts de-energized before work, and the narrow conditions permitting energized troubleshooting
• 29 CFR 1910.333(b)(2) - lockout and tagging for work on electric circuits and equipment
• NFPA 70E-2021, 120.5 - process for establishing and verifying an electrically safe work condition
• Standard copper conductor resistance tables for the gauge, construction and temperature used in the arithmetic above
• Manufacturer documentation for each device's input impedance, reference arrangement and common-mode range
• See related: What a Control Signal Type Implies About the Fault; What a Sensor Actually Reports; How to Tell a Command Problem From a Feedback Problem