Common EMC failures and what they usually mean

A failed EMC plot is not a diagnosis. Its frequency, shape, polarity, operating-mode dependence and response to controlled changes are what turn it into one.

Under schedule pressure, teams often start fitting ferrites, changing capacitors or adding shielding before identifying the coupling path. That can move a peak without fixing the mechanism, creating a product which passes one configuration and fails when a cable, peripheral or operating mode changes.

Reading the failure before changing the design

Common EMC failures become easier to investigate when the result is separated into source, coupling path and victim. For emissions, the source may be a switching edge, clock, converter or motor drive. The path could be a PCB return discontinuity, power lead, enclosure seam or external cable. For immunity, the same model runs in reverse: injected or radiated energy reaches a susceptible circuit through a coupling path and causes a defined performance degradation.

Start by preserving the failed configuration. Record cable positions, EUT mode, software build, loads, peripherals, supply arrangement, enclosure state and test equipment settings. Photograph it. If the arrangement cannot be reconstructed, later improvements cannot be distinguished confidently from test variation.

Next, classify the failure:

• narrowband or broadband;
• fundamental frequency, harmonic series or unrelated resonance;
• continuous, load-dependent or periodic;
• associated with a particular cable, port or enclosure orientation;
• repeatable at one immunity frequency or spread across a band;
• temporary degradation, latched malfunction, reset or physical damage.

The applicable product or product-family standard must then be checked. CISPR 11, CISPR 32 and other CISPR publications can contain emissions requirements for products within their scopes, while IEC 61000-4-x documents describe basic immunity test methods called up by product standards. They are not interchangeable, and the latest active edition, frequency ranges, limits, detectors, bandwidths, configurations and performance criteria need verification.

Harmonic combs usually point back to a periodic source

A series of narrow peaks with regular spacing normally indicates a periodic waveform. The spacing often corresponds to a clock, converter switching frequency or repeating data activity, while the visible lines are harmonics created by fast edge rates and non-sinusoidal current.

Do not assume that the strongest measured harmonic identifies the dominant circuit node. PCB structures, cables and enclosure apertures can resonate, making a higher harmonic radiate more efficiently than the fundamental. A 25 MHz digital clock, for example, might produce a much larger radiated feature at a later harmonic if an attached cable provides an efficient common-mode antenna at that frequency. This is an illustrative mechanism, not a universal cable-resonance calculation.

Near-field probing can narrow the source region, but probe position and orientation must be controlled. A magnetic-field probe is useful around current loops and switching paths; an electric-field probe responds more strongly to high-impedance voltage structures. Treat either as a comparative debugging tool unless the complete measurement method has been characterised.

Likely corrections include reducing unnecessary edge speed, improving local decoupling, shortening high di/dt loops, maintaining a continuous return beneath signal transitions and containing clock currents within the PCB. Randomly increasing decoupling capacitance can introduce anti-resonances. Check impedance over frequency rather than assuming that more capacitance always produces a quieter rail.

Broadband noise often exposes fast switching current

Broadband emissions are commonly associated with short rise times, diode recovery, MOSFET switching, relay or motor commutation and unstable control loops. A spectrum that rises or falls with converter load is particularly informative. Compare idle, nominal and maximum representative load while changing one variable at a time.

For conducted emissions on relevant power ports, a correctly configured LISN provides a defined impedance and a measurement port for disturbance voltage. Conducted emissions are commonly assessed from 150 kHz to 30 MHz in several standards, but that range is not universal. The applicable publication determines the frequency range, LISN arrangement, receiver bandwidth, detector sequence and limits.

Peak detection is useful for rapid scanning, while quasi-peak and average detectors may be required for final assessment under relevant CISPR methods. A spectrum analyser set to an arbitrary resolution bandwidth is not automatically equivalent to a compliant measuring receiver. Incorrect bandwidth or detector selection can change reported amplitudes and lead to either false confidence or unnecessary redesign.

If noise appears on both line and neutral relative to the reference plane, common-mode current deserves attention. If it appears primarily between conductors, differential-mode filtering may be more relevant. Clamp-on current probes and controlled component substitutions can help distinguish the two, but the probe transfer impedance and usable frequency range must cover the disturbance being investigated.

Cable movement implicates common-mode current

When a radiated peak changes substantially as an I/O cable is moved, shortened or disconnected, the cable is probably part of the radiating structure. The source may still be on the PCB. The cable is simply converting common-mode current into radiation.

Cable shielding only works as part of a complete current-return strategy. A shield terminated through a long pigtail has appreciable inductance at higher frequencies. It can leave common-mode current flowing over the enclosure and cable exterior, even though continuity appears satisfactory on a multimeter. A low-impedance circumferential termination at the enclosure entry is generally more effective where the connector and product construction permit it.

Be careful with blanket instructions to connect or disconnect shields at one end. The correct approach depends on frequency, safety, bonding architecture, cable function and the applicable product requirements. A low-frequency ground loop concern does not justify creating an uncontrolled RF termination. Hybrid arrangements may be appropriate, but parasitic capacitance and surge behaviour must be considered.

For debugging and robust test interconnections, correctly specified RF and microwave cables and connectors matter. Damaged connectors, unknown cable loss and poor screening can obscure small changes. EMC Hire can also advise on suitable RF coaxial cables for a defined frequency range and measurement setup.

Immunity weakness is often a demodulation problem

A narrow immunity failure at one or more frequencies can indicate resonance in a cable, PCB structure or enclosure, but the final malfunction may occur after RF is rectified by a semiconductor junction. Inputs, protection devices, amplifier stages and power-management circuits can demodulate RF into a low-frequency error, causing false measurements, communication faults or processor resets.

IEC 61000-4-6 addresses conducted RF immunity using coupling devices such as CDNs where the method and port permit. A CDN is not a conducted emissions LISN. BCI probes are used for current-injection immunity methods in applicable automotive, military, aerospace or product-specific procedures and are not interchangeable with a CDN simply because both couple RF onto cables.

During a controlled pre-compliance investigation, monitor injected level, forward power, relevant calibration data and EUT behaviour. Correlate the failure with individual ports and cable arrangements. EMC Hire provides information on conducted immunity equipment and complete conducted immunity system configurations for suitable methods.

Fixes may include common-mode filtering, feed-through capacitance, improved connector bonding, local RF filtering at susceptible inputs, better reset supervision or firmware recovery. Adding a capacitor deep inside the PCB may fail because RF has already travelled through the victim circuitry. Filtering is normally most effective at the boundary where the unwanted current enters.

Ground loops are not one single EMC fault

The term ground loop is frequently used for unrelated problems: low-frequency circulating current, shared impedance, shield current, reference-plane discontinuity or unintended chassis-to-circuit coupling. Each requires a different correction.

A shared return path allows load current to modulate a sensitive reference. At RF, an unnecessarily long bonding strap adds inductance and may cease to behave as an equipotential connection. Cutting protective-earth conductors is not an acceptable debugging technique. Instead, measure safely, identify current paths and distinguish protective bonding from functional and signal reference arrangements.

If a failure changes when a temporary short, wide bond is applied between two chassis points, that is evidence about RF impedance, not proof that the final product should contain an arbitrary wire. The permanent fix may require a revised enclosure joint, connector mounting, PCB-to-chassis bond or cable-entry treatment.

Typical scenario

Consider an illustrative networked controller which exceeds a radiated emissions limit at a harmonic of its processor clock and resets during conducted RF immunity on the Ethernet port. The emissions peak changes with cable position, while the immunity failure occurs only over a limited frequency band.

The likely investigation would preserve the formal setup, compare cable common-mode current against the radiated peak, probe the clock and interface regions, and assess shield termination at the enclosure boundary. For immunity, the team would confirm the method called up by the product standard, reproduce the affected frequencies using suitable coupling equipment and monitor supply rails, reset and communications.

Choosing the wrong test equipment can corrupt both diagnoses. A general-purpose analyser with unsuitable dynamic range may overload, while an uncharacterised injection arrangement may deliver an unknown stress. Testing only an idle software mode could miss the noisy clock activity or the firmware condition that triggers the reset.

Early pre-compliance work gives the team access to the PCB and firmware while changes remain practical. It can produce calibrated engineering data, improve confidence before formal testing and support the technical file and self-certification process where legally and technically appropriate. It does not by itself prove compliance.

EMC Hire can support the work through equipment hire, setup selection, accessible pre-compliance engineering, on-site testing, test-facility access and formal compliance testing where appropriate. Test equipment is used with calibration traceable through an appropriate ISO/IEC 17025 accredited calibration provider. Suitable traceability supports repeatability, comparison between development and formal measurements, and a more defensible evidence trail.

When to Hire EMC Equipment

Hiring is technically sensible when the investigation needs a receiver, LISN, current probe, CDN, RF amplifier, signal generator or monitoring equipment for a defined test window. It avoids capital expenditure on instruments that may not suit the frequency range, power level or standard used by the next programme.

Short-term hire also covers project peaks without transferring storage, servicing and calibration overheads to the development team. That matters when equipment is used irregularly. An instrument left unverified for long periods can create more uncertainty than a properly selected hired system with suitable calibration status.

Selection should begin with the applicable method, EUT ports, required frequency range, test level, modulation, cable type and monitoring needs. EMC Hire can help define a compatible setup rather than supplying isolated components that cannot achieve or measure the required condition.

Common EMC Testing Mistakes to Avoid

Changing several variables together

Adding ferrites, capacitors and shielding simultaneously may produce a pass, but it removes diagnostic value. The team cannot identify the effective control or judge its production tolerance.

Ignoring cable geometry

Cable routing, height, length and termination alter common-mode coupling and antenna behaviour. Poor documentation creates non-repeatable emissions plots and immunity thresholds.

Using the wrong coupling device

A LISN belongs to relevant conducted emissions measurements. A CDN belongs to applicable conducted RF immunity methods. Substituting one for the other invalidates the physical test setup.

Testing an unrepresentative operating mode

Inactive radios, static displays, reduced processor load or missing peripherals can suppress both noise sources and susceptible states. A clean result may describe only the test mode, not the marketed product.

Overlooking receiver settings and ambient signals

Wrong detectors or resolution bandwidths change emissions results. Ambient broadcast and radio signals can also resemble EUT emissions unless the source is checked by switching the EUT off, changing antenna orientation or using other controlled discrimination methods.

Failing to record the failure criterion

An immunity note stating only “failed” is weak evidence. Record frequency, applied level, modulation, dwell behaviour, affected function, recovery, port, cable configuration and the applicable performance criterion.

Frequently Asked Questions (FAQs)

Can harmonic spacing identify the noisy clock?

It can provide a strong lead. Compare the spacing with clocks, switching converters and periodic data activity, then verify by changing the suspected source frequency or mode. Resonances may make a higher harmonic dominant.

Does moving a cable prove that it needs better shielding?

No. Movement shows that the cable participates in the coupling path. The root cause could be common-mode current caused by PCB return discontinuities, poor connector bonding, inadequate filtering or shield termination.

Why does an immunity failure occur at only one narrow band?

A cable or structure may be resonant, a filter may lose effectiveness, or a semiconductor may demodulate RF particularly efficiently in that range. Repeat the test with controlled cable and port changes before selecting a fix.

Can pre-compliance data be used in a technical file?

Calibrated and well-documented engineering data can support risk assessment, mitigation evidence and technical documentation. The manufacturer remains responsible for confirming applicable legislation, standards, conformity route and whether further formal or accredited testing is required.

When might accredited final testing be required?

Defence, automotive and aerospace programmes may impose accredited laboratory or contractual testing requirements. Similar obligations can arise from customers or approval schemes. Check the current programme documents rather than assuming that general pre-compliance evidence is sufficient.

Move from the plot to the coupling path

The useful question is not simply why the product failed. It is which source generated the disturbance, which path transported it and which circuit or structure converted it into the recorded result.

For help selecting equipment, reproducing a failure or planning defensible pre-compliance and formal compliance work, speak with the EMC Hire engineering team. You can request an equipment hire quotation, arrange on-site testing or book space at the EMC Hire test facility by calling +44 (0)1462 817111 or emailing sales@emchire.co.uk.