What causes radiated emissions peaks at harmonics

A harmonic peak rarely identifies the radiating structure by itself. The source may be a clock, but the antenna is often a cable, PCB return-path discontinuity or enclosure aperture elsewhere in the product.

Effective debugging therefore requires two questions: what generates energy at that frequency, and what converts it into radiated electromagnetic fields?

Why digital circuits produce harmonic spectra

An ideal sinusoid contains one frequency. A digital clock contains a fundamental plus harmonics because its waveform includes rapid transitions rather than a smooth sinusoidal shape. The faster the edge, the further significant spectral energy extends above the clock frequency.

This distinction matters. A 25 MHz clock with sub-nanosecond edges can create measurable energy hundreds of megahertz above its fundamental. Replacing it with another nominally 25 MHz source will not necessarily preserve the emissions result if the replacement has different rise time, drive strength, duty cycle, overshoot or output impedance.

Odd harmonics tend to dominate a symmetrical square wave, but real hardware is not symmetrical. Duty-cycle error, unequal rise and fall times, ringing, modulation and nonlinear loading introduce even harmonics and alter harmonic amplitudes. A spectrum containing both odd and even clock-related peaks is therefore not automatically evidence of a second oscillator.

The commonly used relationship between rise time and spectral knee frequency can provide a debugging estimate, but it is not a compliance limit or a precise prediction of radiated level. Probe loading, package inductance, transmission-line behaviour and the actual switching waveform all affect the spectrum. Check the semiconductor manufacturer's signal-integrity documentation and IBIS models where available.

A source needs a coupling path and an antenna

Clock energy confined to a small loop with a continuous return plane may produce little far-field radiation. The same clock coupled onto a long external cable can fail a radiated emissions limit by a wide margin. The digital source has not necessarily become noisier. Its common-mode conversion and antenna efficiency have changed.

A useful working model separates the problem into three parts:

• The source, such as a clock, switch-mode converter edge, memory interface or processor activity.
• The coupling mechanism, including capacitive coupling, shared impedance, magnetic coupling or return-path discontinuity.
• The radiating structure, commonly a cable, PCB loop, heatsink, enclosure seam or aperture.

Changing any one of these can reduce a peak. Changing the wrong one may merely move it. For example, adding a ferrite to a cable can suppress common-mode current over one frequency region while leaving lower-frequency current largely unchanged. Ferrite impedance curves must be read under realistic bias and temperature conditions rather than treated as a single resistance value.

Clock harmonics and spectral correlation

Start by calculating candidate harmonic families from every known oscillator, bus clock and repetitive switching frequency. Do not stop at the crystal frequency. Phase-locked loops, clock multipliers, divided clocks, DDR interfaces, display timing, Ethernet references and DC-DC converter switching nodes can all produce related families.

Frequency coincidence is evidence, not proof. Two unrelated sources may share a harmonic, and instrument frequency accuracy or spread-spectrum modulation may make a family appear offset or broadened. Change one operating parameter at a time and observe whether the peak moves, disappears or changes amplitude.

Useful controlled changes include reducing processor clock frequency, disabling an interface, changing display resolution, pausing data traffic or selecting a different converter operating mode. A peak that tracks the altered clock strongly implicates that source family. If the peak remains fixed but its amplitude changes, the modified subsystem may instead be changing the coupling path or load current.

Spread-spectrum clocking can replace a narrow peak with a broader distribution. That may reduce a receiver reading under some detector and bandwidth conditions, but it does not remove the underlying energy. Confirm that spread-spectrum operation is permitted for the product and that the measurement settings match the applicable standard.

Cable resonance can select one harmonic from many

A source may generate dozens of harmonics while the scan shows only a few dominant peaks. Cable resonance often explains the selection.

A cable driven in common mode can behave as an unintended antenna. Resonant behaviour depends on electrical length, velocity factor, termination, proximity to the ground plane, attached equipment and the common-mode impedance at both ends. A rough quarter-wave estimate can help identify candidate structures, but a cable in a laboratory arrangement is not an isolated wire in free space.

This explains why moving a cable can change one peak by several decibels without changing the electronics. The cable's orientation, distance from conductive surfaces and current distribution have changed. If the harness routing during debugging differs from the prescribed test arrangement, the resulting improvement may not survive formal testing.

Measure common-mode cable current with a suitable RF current probe where possible. A current maximum at the same frequency as a radiated peak is a useful correlation. The probe's transfer impedance, usable frequency range, aperture and loading effect must be understood. Raw analyser amplitude is not automatically cable current unless the probe factor and measurement chain are correctly applied.

For more detail on antennas, receivers and measurement-chain requirements, see EMC Hire's radiated emissions test system information.

Return-path discontinuities create common-mode current

Differential current should return close to its outgoing path. A signal crossing a split reference plane, changing layer without an adjacent return via or passing through a poorly bonded connector forces the return current to take a larger route. Loop area increases, local field strength rises and part of the energy may be converted into common-mode current.

Connector pin allocation is a frequent cause. A high-speed signal may reach the connector beside one signal return, while the cable shield is bonded through a long PCB trace or pigtail. At higher frequencies that inductive connection prevents the enclosure and shield from acting as a low-impedance RF boundary. Harmonic current then flows on the outside of the cable.

Near-field magnetic and electric probes help localise energetic PCB regions, although they do not directly predict a compliant far-field result. Probe orientation, spacing and cable handling must remain consistent. Otherwise, apparent improvements may simply reflect a changed probe coupling coefficient.

Enclosure leakage and aperture resonance

A conductive enclosure is only as effective as its seams, apertures, cable penetrations and bonding arrangements. Long narrow slots can radiate efficiently when driven by internal fields. Display openings, ventilation patterns, removable covers and poorly compressed gaskets deserve attention when a harmonic peak changes sharply with lid position or applied pressure.

Maximum aperture dimension is often more informative than total open area. A row of small holes generally behaves differently from one continuous slot of the same area. Paint, anodising, contamination and inconsistent fastener torque can also raise joint impedance and undermine repeatability.

Use a near-field probe around seams to identify leakage regions, then confirm the result at the radiated measurement position. Covering an aperture with conductive foil can be a useful diagnostic experiment, provided electrical contact is controlled. It is not automatically a production-ready mitigation.

Distinguishing radiated and conducted origins

Radiated emissions harmonics can originate from noise first present on a power port. A switch-mode converter or clock-related current may appear in a conducted emissions measurement and subsequently excite a cable at higher frequencies. Conversely, a strong radiated peak does not prove that the mains port is the source.

Use a LISN only for conducted emissions measurements on relevant power ports and according to the applicable method. It is not a conducted immunity device. Comparing conducted and radiated signatures can reveal shared source frequencies, but the measured quantities and coupling paths are different. EMC Hire's conducted emissions system guidance explains the relevant measurement arrangement.

Typical scenario

Consider an illustrative industrial controller that shows narrow peaks at multiples of a processor reference clock during a pre-compliance scan. One harmonic is markedly higher than adjacent members of the family, and its amplitude changes when the Ethernet cable is repositioned.

The likely test setup includes a compliant measurement receiver or suitable analyser, antenna and characterised RF cable at the required test distance, with the equipment under test operating in a representative high-activity mode. A current probe and near-field probes may be added for diagnosis rather than used as substitutes for the radiated measurement.

The engineering team must determine whether the dominant mechanism is direct PCB radiation, common-mode current on the Ethernet cable or leakage through the enclosure around the connector. Reducing the processor edge rate could lower the source energy. Improving shield termination or connector bonding may instead interrupt the coupling path. Applying suppression without that distinction risks masking one configuration while leaving another operating mode exposed.

Early investigation in an accessible EMC pre-compliance environment allows cable positions, software modes and provisional mitigations to be explored before formal testing. Pre-compliance results do not prove conformity, but calibrated engineering data can reduce redesign risk and improve confidence when entering a formal programme.

EMC Hire can support equipment selection, test-facility access, on-site investigation, pre-compliance work and formal compliance testing where appropriate. Equipment used for relevant measurements has calibration traceable through an appropriate ISO/IEC 17025 accredited calibration provider. That traceability supports repeatability, comparison between development and formal measurements, and a stronger evidence trail for the technical file.

When to Hire EMC Equipment

Hiring is often more defensible than purchasing when a receiver, antenna, current probe or low-loss RF cable is required for a defined debugging window. It avoids tying up capital in equipment whose frequency coverage, detector capability or dynamic range may not suit the next programme.

Short-term access also helps during project peaks when several teams need to test simultaneously. Ownership carries storage, servicing, calibration and damage-control obligations, particularly for antennas, precision adapters and RF cables. Cable performance can deteriorate through repeated flexing or connector damage while remaining visually acceptable. Suitable phase-stable or low-loss assemblies should be selected for the frequency range and handling duty; see the available RF and microwave cables and connectors.

Before hiring, define the measurement objective, frequency range, expected signal level, detector requirements, transducers, connectors and software needs. This prevents a technically capable analyser being paired with an antenna, preamplifier or cable that limits the complete system.

Common EMC Testing Mistakes to Avoid

Chasing every peak independently

Treating each harmonic as a separate fault wastes time. Group peaks into frequency families and test whether they track a common clock or operating state. A mitigation at the source may reduce the whole family, while cable resonance may explain why only one member threatens the limit.

Changing cable routing without recording it

Cable position affects antenna efficiency and coupling to the reference plane. An undocumented routing change can create an apparent fix that cannot be reproduced. Photograph cable paths, support points, excess length, terminations and peripheral positions.

Using inappropriate receiver settings

Peak, quasi-peak and average detectors do not produce interchangeable results. Resolution bandwidth and detector selection must follow the applicable CISPR or product-standard method for the frequency region being measured. A fast peak scan is useful for diagnosis, but it may neither confirm a failure nor demonstrate compliance.

Ignoring the RF measurement chain

Uncharacterised cables, adapters, preamplifier compression or incorrect antenna factors distort amplitude data. Connector wear can introduce intermittent loss that looks like product variability. Record correction factors and verify the chain before drawing design conclusions.

Testing an easy operating mode

An idle processor, inactive port or static display may suppress the mechanism under investigation. Exercise representative worst-case modes, including data traffic and connected accessories. The product or product-family standard may define configurations and performance conditions that override engineering assumptions.

Confusing a diagnostic fix with a production solution

Foil over a seam, a clamp-on ferrite or an improvised bond can identify a coupling path. It does not establish durability, safety, manufacturability or performance across production tolerances. Retest the engineered implementation rather than relying on the diagnostic result.

Frequently Asked Questions (FAQs)

Why is one clock harmonic much higher than the others?

The harmonic may coincide with a cable, PCB or enclosure resonance, making the radiating structure more efficient at that frequency. Receiver response, antenna polarisation and multiple sources adding constructively can also contribute. Check common-mode current and alter physical geometry under controlled conditions.

Do evenly spaced peaks always come from a clock?

No. Repetitive converter switching, burst activity, digital frame timing and intermodulation can also create regularly spaced components. Correlate the spacing with known frequencies and deliberately alter one candidate source.

Can near-field probing replace a radiated emissions scan?

No. Near-field probes are strong diagnostic tools, but their readings depend heavily on probe position and local field structure. They do not reproduce the far-field geometry, antenna factor, test distance or prescribed measurement method.

Will slowing a clock edge always reduce emissions?

It often reduces high-order harmonic energy, but excessive slowing can affect timing margin, increase transition-region current or alter susceptibility. Use drive-strength and slew-rate controls within the component manufacturer's permitted operating conditions, then verify signal integrity and EMC performance.

How much scan repeatability should an engineer expect?

That depends on site performance, ambient signals, EUT stability, cable placement, antenna positioning and measurement-chain uncertainty. Repeatability improves when the setup, software mode, photographs, equipment identifiers, correction data and environmental conditions are recorded consistently.

Which standard should define the final measurement?

Use the current product-specific or product-family standard applicable to the equipment and intended market, not merely a familiar generic method. Confirm the latest active edition, limits, frequency range, detector, bandwidth, operating modes, test distance and documentation requirements. The manufacturer remains responsible for determining the applicable legislation and conformity route.

Turning harmonic patterns into defensible evidence

A good debug record links each observed peak to a probable source, coupling path, radiating structure and controlled experiment. That record supports design decisions, formal test planning, EMC risk assessment, mitigation evidence and, where applicable, self-certification documentation. Testing alone does not complete every CE, UKCA or FCC obligation, and the responsible manufacturer or economic operator must confirm the relevant requirements.

For help selecting equipment, arranging on-site testing, booking space at the EMC Hire test facility, or discussing pre-compliance and formal compliance testing, contact the EMC Hire engineering team on +44 (0)1462 817111 or email sales@emchire.co.uk. Bringing the scan, clock list, cable configuration and product-standard requirements to the discussion will make the first technical review more productive.