← Smart Buildings Studio
Concept Explainer · Smart Buildings

Common-Mode Noise & Differential Measurement

Why BAS sensor wiring runs two wires instead of one — and what actually gets canceled out when it does.

Run a sensor cable anywhere near a VFD, a motor, or a bank of fluorescent or LED ballasts, and it will pick up electrical interference — that part is unavoidable. What's avoidable is having that interference wreck the reading. Whether it does comes down almost entirely to one design decision made long before the cable was ever pulled: was this signal measured single-ended, against a shared ground reference, or differentially, between two dedicated conductors? That one choice determines whether the noise a long cable run inevitably absorbs ends up in the final number or gets thrown away before anyone sees it.

The Setup

What "common-mode noise" actually is

A cable running parallel to a VFD output, a motor lead, or ballast wiring sits inside a changing electromagnetic field. That field induces a small unwanted voltage onto the cable's conductors — capacitively, inductively, or both. Critically, if two conductors are routed the same way, the same distance from the same source, they pick up nearly the same induced voltage, riding on top of whatever real signal is on the line, measured the same way relative to earth or chassis ground on both wires at once. That's what "common-mode" means: a disturbance common to both conductors, not a disturbance between them. The real signal, by contrast, exists as a difference between the two conductors. Whether a measurement scheme can tell those two things apart is the whole story.

Single-ended measurement

Corrupted
VFD / MOTOR(noise source)SENSORsignal wire — large induced loop, picks up common-mode noiseground reference — low impedance, stays comparatively flatBAS AIsingle-endedsignal wire (signal + noise)ground reference (~flat)=reading at receiverthe noise has nowhere to be subtracted out — it lands directly in the measurement
Noise coupling
Asymmetric
The signal wire and the ground reference aren't routed identically, so they don't pick up the same induced voltage.
Resulting reading error
Direct, uncanceled
Whatever common-mode voltage lands on the signal wire is read as if it were part of the real signal.

Differential (twisted-pair) measurement

Noise Rejected
VFD / MOTOR(same noise source)SENSORtwisted pair — both conductors share the same routing & loop area+differential inputwire A (signal + noise)wire B (matching noise)=recovered signalnearly identical common-mode noise cancels in the subtraction — the real signal survives
Noise coupling
Nearly symmetric
Both conductors run the same path, the same distance from the noise source — so they absorb nearly the same disturbance.
Resulting reading error
Reduced 60–80 dB
Typical differential input / RS-485 transceiver CMRR — a 1,000× to 10,000× reduction in the noise that survives to the reading.
Why this works

A differential receiver never measures either wire against ground. It only ever measures one wire against the other.

That single fact is the entire mechanism. Since the real signal is deliberately carried as a difference between the two conductors — e.g. +signal/2 on one wire and −signal/2 on the other, referenced to a BAS analog input configured for differential sensing, or the two legs of an RS-485 pair — while common-mode noise shows up as nearly the same disturbance added to both wires, subtracting one wire from the other keeps the difference (the real signal) and cancels the sum-in-common part (the noise). A single-ended input has no second wire to subtract against; it only ever has "signal wire vs. ground," and ground was never coupled to the noise source the same way the signal wire was — so there is nothing to cancel it with. This is exactly why long BAS sensor runs near VFDs and motors specify differential/twisted-pair wiring rather than a single signal conductor referenced to ground.

Common misconception
"Shielding the cable is enough — wiring topology doesn't really matter."

Incomplete, and it causes real field failures. Shielding is a genuinely important defense — it reduces how much noise couples onto the cable in the first place, mainly by intercepting electric-field (capacitive) coupling before it reaches the conductors. But shielding and differential wiring solve two different problems. Shielding reduces the amount of noise that gets onto the pair. Differential measurement rejects whatever common-mode noise gets onto the pair anyway — magnetic (inductive) coupling in particular passes through most shields largely unaffected. A cable that is beautifully shielded but wired single-ended can still pick up enough inductively-coupled common-mode noise near a VFD to corrupt the reading, with no differential receiver downstream to cancel it. That's exactly why BAS design standards call for both proper shielding anddifferential wiring / twisted pairs for sensitive analog signals on long runs — they're complementary defenses, not substitutes for each other.

Related Concept Explainers
Drift, Offset & Span — The Three Ways a Sensor Silently Goes Wrong
Read it →
Why 4–20mA Current Loops Resist Noise Better Than Voltage Signals
Coming soon

Common-Mode Noise & Differential Measurement — Concept Explainer

Explains why BAS sensor wiring near VFDs, motors, and ballasts uses two dedicated conductors measured differentially rather than a single signal wire referenced to ground — and exactly what happens electrically that lets the differential scheme cancel out common-mode interference that a single-ended input cannot.

Why Long Analog Runs Near Electrical Noise Fail Single-Ended

A single-ended input measures one signal conductor against a shared ground/reference. Any voltage induced onto that conductor by a nearby VFD, motor, or ballast — capacitively or inductively coupled — is indistinguishable from the real signal, because the input has no independent way to know how much of what it measured was noise. The ground reference typically has much lower impedance and a different physical routing than the signal wire, so it does not pick up a matching amount of the same disturbance, leaving nothing to subtract the noise against.

How Differential Sensing Cancels It

A differential input (a BAS analog input configured for differential sensing, or an RS-485 twisted pair) measures the voltage difference between two dedicated conductors instead. The real signal is carried as that difference. Common-mode noise, by contrast, is a disturbance that appears nearly identically on both conductors — because a twisted pair keeps both wires at essentially the same physical position relative to the noise source along the entire run. Subtracting one conductor from the other preserves the differential signal while canceling the noise term, which is common to both. The degree of cancellation is quantified by the common-mode rejection ratio (CMRR) of the receiving circuit, typically 60-80 dB for BAS-grade differential inputs and RS-485 transceivers — a 1,000x to 10,000x reduction in surviving noise.

Shielding vs. Differential Wiring — Two Separate Defenses

Shielding and differential/twisted-pair wiring are complementary, not interchangeable. Shielding intercepts electric-field coupling before it reaches the conductors, reducing how much noise gets onto the cable in the first place; it is comparatively less effective against magnetic (inductive) coupling, which is exactly the dominant coupling mechanism near VFDs and motor conductors carrying high dv/dt or di/dt. Differential measurement rejects whatever common-mode noise does make it onto the pair, regardless of coupling mechanism. BAS wiring standards specify both for sensitive analog signals over long runs precisely because neither one alone covers the full noise environment near variable-frequency drives and motor starters.

Frequently asked questions

Is a 4-20mA current loop the same thing as a differential voltage measurement?

Not exactly, but it shares the same underlying advantage. A 4-20mA loop measures current, which is identical at every point in a series loop regardless of voltage drops or induced voltages along the wire, so common-mode voltage noise on the loop wiring does not directly corrupt the current being measured the way it would a single-ended voltage signal. It is a different mechanism reaching a similar noise-immunity result.

Does twisting the pair matter, or is it enough to just run two wires side by side?

Twisting matters. Two untwisted parallel wires can still couple slightly differently to a nearby noise source depending on their exact geometry and orientation, especially over long runs. Twisting the pair keeps the average position of each conductor relative to the noise source essentially identical over any given length, which is what makes the common-mode noise on the two conductors nearly equal — the precondition for it to cancel well in a differential measurement.

What is CMRR and why does it matter for BAS wiring design?

Common-mode rejection ratio (CMRR) is a measurement of how well a differential input rejects a signal that is common to both its inputs, usually expressed in decibels. A CMRR of 80 dB means common-mode noise is attenuated by a factor of 10,000 relative to the differential signal. BAS designers use CMRR specs, along with expected common-mode noise levels near VFDs and motor wiring, to decide whether differential/twisted-pair wiring is required for a given sensor run.

If a single-ended sensor is far from any VFD or motor, does any of this still matter?

Less, but it does not disappear. Fluorescent and LED ballasts, nearby power wiring, ground loops between buildings, and even lightning-induced transients can all couple common-mode noise onto long cable runs. Differential wiring is cheap insurance relative to the cost of chasing an intermittent, noise-related BAS control problem after the fact, which is why it is the default recommendation for any long analog run in most BAS wiring standards.

🎓

Try our Smart Buildings Studio

More calculators, simulators, and guides for this discipline.

Related tools & guides

Drift, Offset & Span — Concept ExplainerBAS Design Reference — Interactive ReaderBACnet MS/TP Segment SizingDDC Panel Heat Dissipation