xsz sensor · industrial signal integrity guide
Common-Mode vs Differential-Mode Noise in Sensor Cables
Common-mode noise moves both sensor conductors together relative to the receiver reference. Differential-mode noise appears between the conductors and directly changes the measured signal. The practical job is to measure both modes, find the coupling and conversion path, and then choose wiring, shielding, grounding, filtering, isolation, or receiver changes that address the real mechanism.
Measure V+ and V− to the same receiver reference, then calculate Vdiff and Vcm. If both conductors move together, investigate common-mode coupling, grounding, and input range. If the voltage between them changes, investigate loop area, imbalance, source noise, and differential filtering.
Classify the measured voltage before choosing a cable fix
Noise classification describes where the unwanted voltage exists at the receiver. Coupling describes how the energy reached the circuit. One noise source can create both modes, and common-mode energy can become differential error before the receiver performs its subtraction.
Common-mode noise
Both input conductors move in the same direction relative to receiver reference. A balanced differential input can reject part of this voltage while its pins remain in range.
Differential-mode noise
The unwanted voltage appears across the pair. The receiver treats it like part of the wanted signal, so common-mode rejection cannot remove it.
Mode conversion
Unequal source, cable, connector, filter, leakage, or input impedance converts shared current into unequal input voltage.
| Decision area | Common mode | Differential mode |
|---|---|---|
| Measured between | Each conductor and the receiver reference; then averaged | V+ and V− directly |
| Typical first causes | Capacitive/RF coupling, ground-potential difference, conducted supply noise | Magnetic loop pickup, source noise, unequal coupling, conversion through imbalance |
| Receiver response | Rejected according to practical CMRR, balance, frequency, and input range | Added to the wanted differential signal |
| First checks | Common-mode range, ground/chassis path, shield current, cable-to-noise-source coupling | Loop area, pair assignment, source impedance, connector balance, differential bandwidth |
| Common mistake | Assuming high CMRR makes common-mode voltage harmless | Adding a shield without reducing loop area or finding the differential source |
The terms describe two simultaneous components of the same two input voltages; they are not two mutually exclusive cable types.
Every differential sensor input sees Vdiff and Vcm at the same time
Let V+ and V− be the conductor voltages measured relative to the receiver's local reference. Vdiff contains the wanted sensor signal plus differential interference. Vcm describes how far the pair moves together relative to the receiver.
Vdiff = V+ − V−Voltage between the conductorsVcm = (V+ + V−) / 2Average voltage relative to receiver referenceCommon-mode movement
Both conductors move in the same direction. A differential receiver subtracts the two inputs and attempts to reject the shared component.
Differential-mode movement
The conductors move in opposite directions, changing the voltage across the pair. The input cannot distinguish this error from the desired signal.
An ideal differential receiver has no common-mode gain. A real receiver has finite common-mode rejection ratio, or CMRR. CMRR normally changes with gain, common-mode level, source impedance, temperature, and frequency. It also cannot rescue an input that has exceeded its permitted common-mode range or caused protection clamps to conduct.
Signal and CMRR screening aid
Use documented voltages and a CMRR value specified at the frequency of interest. This is a planning calculation, not a product limit or safety test.
1. Separate differential and common mode
2. Estimate common-mode error from CMRR
Technical references: NI field wiring and noise considerations and Texas Instruments SBAA634.
Noise classification tells you what arrived; coupling tells you how it got there
Four coupling paths deserve separate checks
Capacitive coupling comes from changing voltage through stray capacitance. Long parallel runs beside drive or motor conductors increase exposure.
Magnetic coupling comes from changing current linking the signal loop. Twisting the signal with its own return reduces effective loop area.
Conducted noise enters through the supply, common return, reference, or grounding network. Physical cable separation cannot remove noise already carried into the circuit.
Radiated RF can enter through cable, connector, enclosure, or shield discontinuity. The cable and chassis must be evaluated as one high-frequency structure.
Common-mode current becomes differential error through impedance imbalance
The receiver rejects equal voltages, not approximately equal current paths. Any asymmetry before the subtraction point can turn shared current into unequal input voltage.
- STEP 01A noise source couples into the cableA drive, motor, contactor, heater controller, radio, or ground-potential difference creates a disturbance.
- STEP 02Common-mode current finds a return pathCurrent flows through capacitance, shield, chassis, signal reference, or another intended or unintended route.
- STEP 03The two conductor impedances differUnequal source impedance, contact resistance, cable capacitance, filter tolerance, leakage, or protection changes the paths.
- STEP 04The receiver sees differential errorThe input voltages are no longer equal, so subtraction cannot reject the complete disturbance.
Vdm,error ≈ Icm × ΔZWith 100 µA of common-mode current and 10 Ω of impedance mismatch, the simplified result is 1 mV of differential error. That may be small for a 10 V transducer and unacceptable for a millivolt bridge or thermocouple.
Common imbalance sources include connector corrosion, a conductor routed through a different terminal path, mismatched RC components, unequal cable-to-shield capacitance, protection-device leakage, or one input clamp conducting before the other. Replacing only the receiver may not solve an imbalance created upstream.
Twisted pair, shield, grounding, and CMRR perform different jobs
| Control | Main job | What it does not guarantee | What to verify |
|---|---|---|---|
| Twisted pair | Preserves similar exposure and reduces magnetic loop area | Does not replace shielding, isolation, surge protection, or correct pair assignment | Signal and its own return share one pair; untwisted length stays short |
| Cable shield | Intercepts electric-field and RF current and provides a deliberate path | Does not automatically remove differential pickup or serve as signal return | Coverage, continuity, connector termination, pigtail length, chassis path |
| Ground/chassis plan | Defines where safety and interference current returns | Does not make protective earth, functional earth, shield, and signal common identical | Ground-potential difference, bonding impedance, safety requirements |
| Differential receiver | Rejects shared voltage inside its operating range | Does not reject differential noise or remain linear outside common-mode range | Frequency-dependent CMRR, input range, gain, protection, source impedance |
For a deeper installation comparison, read Sensor Cable Shielding for AC and DC Sensors Explained and Sensor Cable Length: What to Check Before Ordering.
The signal architecture changes the correct cable and receiver decision
A solution suitable for a 4–20 mA loop may be unsuitable for a millivolt bridge, RTD, coaxial sensor, or high-speed digital pair. Start with the electrical interface, not the generic word “sensor.”
| Signal architecture | Priority checks | Frequent wrong assumption |
|---|---|---|
| Thermocouple or bridge | Small differential signal, source balance, thermal junctions, input common-mode range, low drift | High input impedance alone guarantees accurate millivolt measurement |
| RTD | 2-, 3-, or 4-wire method, lead matching, excitation noise, settling and filtering | Every conductor can be extended differently without creating error |
| 0–10 V sensor | Signal-reference drop, shared return current, load impedance, supply noise, common-mode range | Signal common remains at the same voltage everywhere in the machine |
| 4–20 mA loop | Compliance voltage, burden voltage, receiver reference/isolation, cable and surge exposure | A current loop is immune to grounding and common-mode problems |
| Coaxial/IEPE sensor | Specified coax, shield/return function, connector, constant-current source, ground topology | Shielded twisted-pair rules can be copied directly to coax |
| Differential digital pair | Pair impedance, skew, termination, common-mode range, connector discontinuity | Logic thresholds make cable geometry unimportant |
For analog proximity applications, compare 0–10 V and 4–20 mA sensor outputs. For switching signals, review PLC sensor input terminology and NPN versus PNP outputs.
Filter the unwanted mode without destroying pair balance or response time
Filtering should be chosen after the mode, source impedance, receiver, and required signal bandwidth are known. A component placed in the wrong path can increase imbalance or hide the real machine event.
Differential capacitor
Shunts high-frequency differential energy but also loads the wanted differential signal as frequency rises.
Common-mode RC paths
Shunt common-mode energy when both paths are closely matched. Tolerance and parasitics can convert remaining current into differential error.
Common-mode choke
Impedes high-frequency current moving in the same direction. It cannot solve low-frequency ground potential or insufficient common-mode range.
A low-pass filter can improve noise but also slows the signal. Define the fastest valid event, acceptable delay, settling requirement, PLC sample behavior, and alarm logic before selecting a cutoff. RF entering the amplifier may be rectified into a DC offset, so filtering only after the amplifier can be too late.
Analog Devices describes balanced common-mode and differential input filters in MT-070: In-Amp Input RFI Protection.
Diagnose sensor cable noise without changing several variables at once
A normal grounded oscilloscope probe can create a hazardous short, alter the reference path, or make the noise appear better or worse. Use an isolated or differential measurement method appropriate to the circuit and follow site electrical-safety procedures.
- Record the symptom and machine state. Note the error, target, drive speed, load transition, temperature, and repetition rate.
- Create a known baseline. Use a stable target or documented simulator when appropriate; confirm the receiver channel first.
- Measure V+, V−, Vdiff, and Vcm. Use the same receiver reference and a safe measurement method.
- Compare timing with likely sources. Trigger on a drive edge, contactor, solenoid, heater controller, or radio event.
- Change one coupling variable. Test separation, orientation, a documented termination, or another receiver channel without changing everything else.
- Inspect balance and continuity. Check conductor resistance, connectors, shield continuity, unintended contact, and damaged pair geometry.
- Retest the worst operating condition. Confirm the fix at realistic cable exposure, machine speed, load switching, temperature, and supply limits.
Three illustrative cases show why one universal fix fails
Illustrative scenario · not a reported customer result
A 0–10 V sensor becomes unstable only when the drive accelerates
Both signal and reference move relative to the PLC reference, but the voltage between them changes only after one terminal block. The useful investigation is not “shielded or unshielded?” It is whether drive-related common-mode current encounters unequal impedance at the connector, return, filter, or input.
Evidence that changes the decision: simultaneous V+, V−, Vdiff, and Vcm traces at both sides of the terminal block, plus conductor resistance and shield/ground continuity.
Illustrative scenario · not a reported customer result
A millivolt bridge looks clean to ground but noisy across the input
The error is already differential. A shield change may have little effect if the main cause is magnetic loop area, incorrect pair assignment, source noise, or a mismatched input filter.
Evidence that changes the decision: verify that each signal travels with its own return, reduce loop area, compare the source directly, and check the differential bandwidth.
Illustrative scenario · not a reported customer result
A 4–20 mA loop works on the bench but saturates in the machine
The loop current may remain correct while the receiver input or burden resistor is driven outside its common-mode range by the installed grounding arrangement. Calling the loop “noise immune” hides the actual voltage limit.
Evidence that changes the decision: measure loop current, burden voltage, receiver input voltage to its reference, supply compliance, and ground-potential difference under load.
A useful supplier review needs the complete sensor-to-receiver path
“The sensor is noisy” is not a reproducible requirement. Give the supplier the exact operating and installation information needed to review the cable, output, connector, power, PLC input, and environment together.
Include these items in the RFQ
Continue to the next wiring decision
Common-mode and differential-mode noise FAQ
Is common-mode noise harmless because a differential input rejects it?
No. Rejection is finite, frequency-dependent, and sensitive to balance. The input can also saturate or clamp if its common-mode range is exceeded.
Will a cable shield remove differential-mode noise?
Not automatically. A shield can reduce electric-field and RF coupling, but differential noise can remain because of magnetic pickup, source noise, conducted noise, or common-to-differential conversion.
Should a sensor cable shield be grounded at one end or both ends?
There is no universal answer. Frequency, ground potential, chassis design, cable construction, connector, safety requirements, receiver, and manufacturer EMC instructions determine the correct termination.
Does twisted pair still help when the cable is shielded?
Yes. Twisting reduces loop area and helps both conductors experience similar coupling. The shield and pair geometry perform different jobs.
Why is mains-frequency noise present with a high-CMRR input?
Possible causes include cable or source imbalance, ground-potential difference, common-mode range limits, filter mismatch, incorrect measurement setup, or differential pickup that CMRR cannot reject.
Is a 4–20 mA sensor loop immune to common-mode noise?
No. Current loops are often robust, but compliance voltage, burden placement, receiver reference, isolation, cable coupling, grounding, surge exposure, and common-mode limits still matter.
Should the noise filter be installed at the sensor or receiver?
Place filtering according to where the noise enters, source impedance, cable, receiver, and required bandwidth. Receiver-side filtering is common, but the complete circuit and response time must be checked.
How can I tell whether cable noise is common mode or differential mode?
Measure both conductors to the same receiver reference with a suitable isolated or differential method, and measure directly between the conductors. Compare V+, V−, Vdiff, and Vcm with the machine event.
Technical references
- NI, Field Wiring and Noise Considerations for Analog Signals—common-mode voltage, CMRR, source balance, and common-mode range.
- NI, Diagnosing and Reducing Measurement Noise—measurement and troubleshooting considerations.
- Analog Devices, MT-070: In-Amp Input RFI Protection—balanced common-mode and differential filtering.
- Texas Instruments, SBAA634—input differential/common-mode definitions and practical rejection limits.
This guide supports application planning. Exact wiring, input limits, filter values, grounding, safety procedures, and EMC requirements must come from the selected sensor, cable, connector, receiver, and machine documentation.