Industrial EMC case study: noisy motor control board

A motor control board can run perfectly on the bench yet fail emissions testing because the motor cable, enclosure and protective earth have become part of the RF circuit.

This industrial EMC case study is an illustrative engineering scenario rather than a report about a named customer. It follows the measurement sequence we would use to distinguish differential-mode switching noise, common-mode current and secondary radiation before changing filters or redesigning the PCB.

The board works, but the complete machine is noisy

Consider a mains-powered industrial controller driving a three-phase permanent-magnet motor through a pulse-width-modulated inverter. The control electronics, gate drivers and power stage have passed functional verification. Thermal performance is acceptable. Motor torque and speed regulation meet specification.

Initial conducted emissions measurements tell a different story. Disturbance voltage measured through the appropriate line impedance stabilisation network, or LISN, is elevated across parts of the applicable frequency range. Radiated measurements also show several broad peaks associated with inverter operation. The exact limits, frequency ranges, detector types and measurement bandwidths depend on the applicable product or product-family standard.

The temptation is to add a larger mains filter. That may reduce one result without addressing the dominant mechanism. It can also increase leakage current, create resonances with existing capacitance or shift RF current into the enclosure and protective-earth system. Filters are not noise absorbers. They alter impedances and return paths.

Start by defining the operating state

Motor drive noise depends heavily on operating mode. A measurement made with the motor unloaded at low torque may bear little resemblance to operation during acceleration, regenerative braking or maximum continuous load. PWM frequency, modulation index, DC-link voltage, gate resistance and motor cable length can all change the spectrum.

A defensible investigation records at least the software version, switching frequency, motor speed, torque or mechanical loading, supply condition, cable type, cable length, enclosure configuration and earthing arrangement. Any automatic spread-spectrum or variable-frequency function should also be documented. Otherwise, an apparent filter improvement may simply be the result of the controller entering a quieter operating state.

Photographs matter. Moving a motor cable by a few centimetres can alter its capacitance to the reference plane and change radiated coupling. Without a record of cable placement, the test cannot be reproduced reliably.

Separate differential-mode and common-mode motor drive noise

The inverter produces high rates of voltage and current change. Differential-mode noise flows between conductors in the intended power loop. Common-mode current flows from the switching nodes through parasitic capacitances into the heatsink, chassis, motor frame, cable screen, protective earth and surrounding structures.

Those mechanisms require different corrective action. A current probe placed around all conductors of a cable together largely cancels balanced differential current and reveals the residual common-mode component, subject to the probe's transfer impedance, bandwidth and loading. Measuring individual conductors can then help characterise conductor current, but the result must be interpreted carefully because it contains both wanted power current and RF disturbance.

If substantial RF current appears around the complete motor cable bundle, fitting only a differential inductor is unlikely to solve the dominant path. Conversely, a common-mode choke will not correct a poorly controlled high-current differential loop simply because it is labelled as an EMC component.

Probe selection matters. The current probe must cover the frequencies of interest without saturation from low-frequency motor current or excessive insertion impedance. Its calibration data and transfer impedance are needed if the reading is to be converted into current rather than used only for relative comparison.

Use the LISN result as evidence, not a diagnosis

A LISN provides a defined impedance at a relevant power port and couples conducted disturbance voltage to a measuring receiver. It does not identify where the noise originated. On many industrial drives, the measured mains disturbance is the final result of several coupling paths involving the DC link, parasitic capacitance across isolation barriers, heatsink bonding and the motor connection.

Conducted emissions are commonly investigated from 150 kHz to 30 MHz, but that range must not be assumed for every product or port. Receiver bandwidth, peak, quasi-peak and average detectors must follow the applicable standard and test plan. A fast peak scan is useful for debugging, yet it is not automatically equivalent to a final quasi-peak or average measurement.

Compare several controlled states:

• Controller energised with the inverter disabled.
• Inverter enabled with the motor stationary, where the design permits this safely.
• Motor running at defined speed and load points.
• Motor cable connected and disconnected only under a safe, documented procedure.
• Alternative cable-screen terminations or enclosure bonds fitted one at a time.

A sharp increase when PWM starts points towards the power stage or its coupling paths. Noise present with the inverter disabled may instead originate in the auxiliary switch-mode supply, communications interface or mains input circuitry.

Near-field probing finds regions, not compliance results

Electric- and magnetic-field probes are useful for locating energetic PCB regions. A magnetic loop can identify high di/dt current loops around MOSFETs, IGBTs, DC-link capacitors and gate-drive returns. An electric-field probe is more responsive to high dv/dt nodes and capacitive coupling.

Probe orientation, spacing and pressure must remain consistent. A near-field trace is a comparative engineering measurement, not a radiated emissions result in dBµV/m. Treating it as a compliance measurement creates false precision.

In this illustrative board, likely areas for investigation include the commutation loop between the DC-link capacitor and inverter bridge, the switching-node copper, heatsink capacitance and the connector transition into the motor cable. A long connection between the DC-link film capacitor and power devices adds inductance. The resulting voltage overshoot and ringing can generate harmonics extending far above the PWM fundamental.

Oscilloscope probing can confirm the ringing, but an unsuitable probe creates its own artefact. A long ground lead forms an inductive loop and can display ringing that is partly generated by the measurement setup. A correctly rated differential probe or appropriately configured high-voltage probing method, with adequate bandwidth and common-mode performance, is required for switching-node work. Manufacturer derating and safety instructions must be followed.

Correct the coupling path before choosing filters

Once measurements indicate the dominant path, modifications can be ranked by mechanism. Reducing the power-loop area, placing the DC-link capacitor closer to the switching devices and controlling gate transitions can reduce source energy. Slower edges may lower emissions but increase switching loss and device heating, so the electrical and thermal consequences need measurement rather than assumption.

For common-mode motor cable current, a low-inductance screen termination at the enclosure entry can outperform a long pigtail. At RF, a pigtail's inductance prevents the screen connection behaving as the intended low-impedance bond. The unwanted current then finds another route through control wiring, bearings or protective earth.

Common-mode chokes, feedthrough capacitors and dv/dt or sine-wave filters may be appropriate, depending on drive topology and product requirements. Component voltage rating, current rating, saturation behaviour, dielectric stress, leakage current, thermal rise and resonance all need checking. Filter manufacturer data should be reviewed under conditions representative of the circuit because catalogue insertion loss measured in a defined impedance system may not predict performance in a motor drive.

Typical scenario

An industrial development team receives an unfavourable pre-compliance plot shortly before a production release. The peaks move when the motor cable is repositioned, while the mains result changes when the enclosure door is removed. That behaviour strongly suggests a common-mode current path involving the cable screen and chassis rather than a single free-space radiator on the PCB.

A sensible test setup combines a suitable LISN and receiver for the power-port conducted emissions investigation, calibrated current probes for cable-current comparisons, near-field probes for source localisation and a representative radiated emissions setup. Each instrument answers a different question. Substituting an oscilloscope FFT for a compliant receiver may support diagnosis, but differences in dynamic range, detectors, bandwidth and overload behaviour limit direct comparison with formal results.

The team must decide whether to change gate resistance, PCB geometry, bonding, cable termination or filtering. Changing all five at once is quick but scientifically weak. If the result improves, nobody knows which measure worked. If it deteriorates, the interaction is even harder to unwind.

Early access to a controlled setup allows one change at a time and reduces the chance of committing to a new PCB before the coupling mechanism is understood. EMC Hire can support this work through equipment hire, pre-compliance engineering, on-site investigation, test-facility access and formal compliance testing where appropriate. Equipment used for relevant measurements is calibrated with traceability through an appropriate ISO/IEC 17025 accredited calibration provider, supporting repeatability and comparison between development and formal testing.

Choosing the applicable industrial EMC requirements

An adjustable-speed power drive system may fall within the scope of IEC 61800-3, depending on the product and installation. Generic industrial immunity or emissions standards should not be selected merely because the equipment will be installed in a factory. Product-specific and product-family standards take precedence where their scope applies.

Engineers evaluating other equipment categories can review EMC Hire's information on generic immunity standards and generic emissions standards. These pages are starting points, not substitutes for a standards assessment.

Always verify the latest active edition, product scope, port definitions, limits, test levels, frequency ranges, performance criteria, operating modes and documentation requirements. Customer contracts may add tests beyond regulatory requirements, particularly in defence, automotive and aerospace programmes. EMC Hire can provide pre-compliance support in these sectors, although final programme testing may require an appropriately accredited laboratory.

For CE or UKCA self-certification routes, testing forms only part of the manufacturer's assessment. The responsible manufacturer or economic operator must determine the applicable legislation, conformity route and technical file requirements. Further practical context is available on EMC Hire's commercial CE marking and EMC testing service page.

When to Hire EMC Equipment

A motor drive investigation often needs several specialised instruments for a short period. Purchasing a receiver, LISNs, current probes, near-field probes and suitable RF accessories for one project can tie up capital while leaving the business responsible for storage, servicing, calibration and future suitability.

Hiring is technically sensible when demand is irregular, a project peak exceeds internal capacity or the required frequency range differs from the equipment normally held. It also permits the setup to be matched to the defined test window rather than forcing an investigation around an unsuitable instrument already owned.

Selection should start with the applicable method, port type, frequency range, disturbance level and required detector functions. For current probes, cable aperture, saturation risk and transfer impedance are relevant. For LISNs, supply voltage, current, phase arrangement and network type must match the EUT and method.

Short-term hire can also reduce the risk of buying hardware that will not suit the next programme. Where the setup or interpretation is uncertain, combining hire with on-site support or a facility booking usually produces better evidence than assembling an unverified bench test.

Common EMC Testing Mistakes to Avoid

Treating every spectral line as a new source

PWM edges create harmonics, resonances and mixing products. Chasing each peak separately can lead to a collection of ineffective filters. Correlate frequencies with switching rate, ringing frequency and operating-state changes before modifying hardware.

Changing cable routing between measurements

The motor cable is often a dominant antenna and capacitive return path. Moving it changes coupling to the reference plane and enclosure, destroying the validity of an A-B comparison.

Using an unsuitable LISN configuration

An incorrect network, supply arrangement or bonding scheme changes the impedance presented to the EUT. The resulting disturbance-voltage measurement may be invalid or unrepresentative.

Ignoring receiver overload

Strong low-frequency switching components can drive an analyser or receiver into compression, creating false responses or hiding real ones. Input attenuation, preselection and preamplification must be managed without compromising sensitivity.

Applying filters without measuring mode current

A differential filter selected for a common-mode problem may offer little improvement. Worse, added capacitance can redirect current into the chassis and increase radiated emissions.

Recording only the final plot

A plot without cable photographs, EUT mode, detector, bandwidth, transducer details, correction factors and modification state is weak evidence. It cannot reliably support later diagnosis, technical documentation or comparison with formal testing.

Frequently Asked Questions (FAQs)

Can an oscilloscope FFT replace an EMI receiver?

It can help identify switching frequencies, harmonics and changes between modifications. It does not automatically reproduce CISPR detector behaviour, bandwidths, input protection, preselection or calibrated transducer correction. Use it as a diagnostic tool unless its suitability for the required measurement has been demonstrated.

How can we tell whether motor cable noise is common-mode?

Place a suitable calibrated RF current probe around all relevant cable conductors together, including the protective or screen arrangement only as required by the measurement plan. Balanced differential currents tend to cancel, leaving residual common-mode current. Probe bandwidth, saturation and placement must be controlled.

Will a shielded motor cable solve radiated emissions?

Only if the screen is suitable and terminated with low RF impedance at the required points. A good cable fitted with long pigtails can perform poorly because the termination inductance interrupts the high-frequency return path.

Should we reduce the PWM switching frequency?

Not automatically. Lowering it moves the fundamental and changes harmonic distribution, acoustic behaviour, current ripple and control performance. Edge rate and parasitic ringing often matter more than the nominal PWM frequency.

Does a pre-compliance pass prove conformity?

No. Pre-compliance testing improves confidence and can provide calibrated engineering data, but the setup, method and documentation must be assessed against the applicable formal requirements. It does not by itself complete a Declaration of Conformity or prove every obligation has been met.

When should formal testing begin?

Formal testing is most efficient once hardware, firmware, cables, enclosure, accessories and representative operating modes are stable. Entering too early risks testing a configuration that will not be manufactured. Leaving all testing until design freeze makes corrective changes expensive.

Turn the spectrum into an engineering decision

The useful outcome of an EMC investigation is not simply a lower trace. It is evidence showing which source produced the disturbance, how it coupled into the measured port or radiating structure, and why the selected mitigation changed that path.

That evidence can support the technical file, EMC risk assessment, mitigation records, Declaration of Conformity and stakeholder review where applicable. More examples of measurement-led fault finding are available in EMC Hire's EMC case studies.

To discuss a noisy motor control board, request an equipment hire quotation, arrange on-site testing, book space at the EMC Hire test facility or plan pre-compliance or formal compliance testing, contact the EMC Hire engineering team on +44 (0)1462 817111 or email sales@emchire.co.uk.