Electronics Guide

Clock Harmonic Radiation

The spectral content of a digital clock depends primarily on its edge rate rather than its fundamental frequency. A 10 MHz clock with 1 nanosecond rise time contains significant energy at harmonics up to several hundred megahertz. The amplitude of harmonics follows a predictable envelope, rolling off at approximately 20 dB per decade above the corner frequency determined by the rise time. Understanding this relationship helps predict which harmonics pose the greatest risk and which can be safely ignored.

Clock distribution networks that route signals across circuit boards create opportunities for radiation. Long traces act as antennas, particularly at frequencies where the trace length approaches a quarter wavelength or odd multiples thereof. Traces crossing gaps in ground planes encounter impedance discontinuities that cause reflection and radiation. Return currents that cannot follow directly beneath the signal trace must take longer paths, creating loop areas that radiate magnetic fields.

The physical layout of clock circuits significantly affects emissions. Keeping clock traces short and directly over continuous ground planes minimizes radiation. Avoiding routing clocks near board edges or connector areas reduces coupling to cables and enclosure apertures. Using clock buffers to create local copies of clocks rather than distributing a single clock across long distances limits the extent of radiating structures. These layout practices address clock radiation at its source.

Crystal Oscillator Problems

Crystal oscillators themselves can be EMI sources when their output signals are not properly managed. The oscillator output often swings rail-to-rail with fast edges, creating a rich harmonic spectrum. Decoupling the oscillator power supply prevents noise from coupling to other circuits and reduces conducted emissions. Keeping the oscillator close to the circuits it drives minimizes trace lengths and associated radiation.

Some oscillators generate spurious outputs at frequencies other than the fundamental and its harmonics. These spurs may result from internal mixing products, parasitic resonances, or mode jumping. When unexplained emissions appear at frequencies not harmonically related to any obvious clock, the oscillator itself may be generating spurious signals. Replacing the oscillator with one from a different manufacturer sometimes reveals whether the oscillator is the source.

Spread spectrum clocking offers a mitigation technique for clock-related emissions. By deliberately modulating the clock frequency over a small range, the energy at each harmonic spreads across a wider frequency band, reducing the peak amplitude measured with standard EMI receivers. The reduction, typically 6-12 dB, can transform marginal compliance into comfortable margin. However, spread spectrum clocking is not suitable for all applications, particularly those requiring precise timing or communicating with external systems expecting fixed clock frequencies.

Clock Jitter and Phase Noise

Clock quality issues manifest differently than simple harmonic radiation. Phase noise creates skirts around the carrier and harmonics that can violate narrowband limits even when the peak harmonic amplitude complies. Jitter causes timing uncertainties that may create emissions during digital state changes that do not correlate cleanly with the nominal clock period. These subtle clock quality problems require different diagnostic approaches than straightforward harmonic radiation.

Power supply noise modulating the oscillator frequency is a common source of phase noise. When switching regulator ripple couples into the oscillator, it modulates the output frequency, creating sidebands offset from each harmonic by the switching frequency. Improved oscillator decoupling, using linear regulators for oscillator power, or selecting oscillators with better power supply rejection ratio addresses this mechanism.

Conducted Emissions from Switchers

Conducted emissions on the AC input of switching power supplies typically dominate the low-frequency portion of emissions tests. Differential-mode noise results from the pulsating current drawn by the power supply, which contains the switching fundamental and harmonics. Common-mode noise results from high-frequency voltage transitions coupling through parasitic capacitances to the input lines. Both noise modes must be filtered to achieve conducted emissions compliance.

The input filter design must address both differential and common-mode noise. Differential-mode filtering uses series inductors and capacitors across the lines. Common-mode filtering uses common-mode chokes with capacitors to ground. The filter topology and component values depend on the noise spectrum and the required attenuation at each frequency. Proper filter placement, with components close to the noise source and adequate separation from input and output connections, maintains filter effectiveness.

Input rectifier recovery behavior affects conducted emissions in AC-DC converters. When diodes switch from conducting to blocking, reverse recovery current creates high-frequency ringing that appears as conducted and radiated emissions. Fast-recovery or soft-recovery diodes reduce this mechanism. Snubber circuits across rectifier diodes damp ringing and reduce the high-frequency content. Silicon carbide diodes offer extremely fast recovery with minimal reverse current, eliminating this noise source in applications where their cost is justified.

Output Noise and Ripple

Switching regulator output contains ripple at the switching frequency and high-frequency noise from switching transitions. While this noise is primarily a power integrity concern for sensitive loads, it can also couple into circuits and contribute to emissions. Output filter design must balance ripple reduction, transient response, and stability requirements.

High-frequency noise from switching events often requires different filtering than low-frequency ripple. Bulk capacitors provide ripple reduction but may have limited high-frequency effectiveness due to equivalent series inductance. Small ceramic capacitors provide low impedance at high frequencies but limited bulk capacitance. Combining both types, with ferrite beads or small inductors where needed, addresses the full spectrum of output noise.

Layout of the output filter significantly affects its performance. Keeping filter components close to the switching stage and connecting them with short, wide traces minimizes parasitic inductance that degrades high-frequency attenuation. Proper placement of capacitors, with decoupling capacitors at the switching stage and bulk capacitors following, ensures each component operates in its effective frequency range.

Radiated Emissions from Power Supplies

Power supply radiation originates from several mechanisms. The switching transistor and its drive circuit generate fields directly. Current loops formed by switching components and their decoupling capacitors create magnetic field radiation. High dV/dt nodes, particularly the switching node in buck converters, capacitively couple to nearby conductors and can excite secondary radiation from cables or enclosure structures.

Minimizing power supply radiation requires attention to loop areas. The primary switching loop, from input capacitors through the switches to return, should be as small as possible. Using integrated power stages that include switches and driver in a single package minimizes this critical loop. When discrete components are necessary, placing them to minimize loop area and using multi-layer boards to provide close return planes reduces radiation.

Shielding may be necessary when layout optimization alone cannot achieve adequate emissions reduction. Local shields around the power supply section prevent coupling to other circuits and reduce direct radiation. The shield must be properly grounded, with continuous connection to the ground plane rather than pigtail connections that introduce inductance. In severe cases, completely shielded power supply modules eliminate radiation from the power conversion circuitry.

Ground Loops

Ground loops form when multiple ground connections create closed loops through which external magnetic fields can induce current. In systems with connections to earth ground at multiple points, power line frequency magnetic fields induce circulating currents that appear as power line hum in sensitive circuits. Higher frequency fields from nearby equipment or the system's own switching power supplies induce correspondingly higher frequency noise.

Breaking ground loops requires either eliminating multiple ground connections or preventing current flow through signal returns. Single-point grounding concentrates all ground connections at one location, eliminating the loop. When multiple ground connections are necessary, isolating the signal ground from chassis or earth ground prevents external currents from flowing in signal returns. Isolation transformers, optocouplers, or balanced signaling with differential receivers provide signal coupling without ground current flow.

In complex systems with many interconnections, complete elimination of ground loops may be impractical. In such cases, reducing loop area, using shielded cables with properly terminated shields, and ensuring that ground connections have low impedance relative to the coupled interference frequencies minimizes the loop's effectiveness at picking up interference.

Split Ground Planes

Intentional ground plane splits, often implemented to separate analog and digital grounds, frequently cause more problems than they solve. When signals cross the split, their return currents cannot follow directly beneath the trace and must take long detours around the split. The resulting large current loops radiate strongly and create impedance discontinuities that cause reflections on high-speed signals.

A better approach uses a continuous ground plane with careful component placement to separate sensitive circuits from noisy ones. Return currents automatically follow directly beneath their associated signal traces when the ground plane is continuous. Sensitive analog circuits placed away from noisy digital circuits experience minimal coupling. When absolutely necessary, splits should be small and positioned where no signals need to cross them.

If signals must cross a ground plane gap, providing a high-frequency return path across the gap is essential. A small bridge connection or a decoupling capacitor across the gap provides a path for high-frequency return currents. The bridge or capacitor should be located directly beneath the crossing signal to minimize loop area. Multiple bridges may be necessary if many signals cross the gap.

Ground Bounce and Simultaneous Switching Noise

Ground bounce occurs when the simultaneous switching of multiple outputs causes current transients through the ground connection inductance, momentarily raising the internal ground potential relative to the external ground. This voltage bounce appears as noise on all signals referenced to that ground and can cause logic errors in circuits sharing the ground connection. Ground bounce represents a power integrity issue that also manifests as EMI.

Reducing ground bounce requires minimizing the inductance of ground connections and limiting the rate of change of ground current. Using multiple ground pins in parallel reduces effective inductance. Slowing output edge rates, where timing permits, reduces di/dt and the resulting voltage drop across ground inductance. Separating outputs with high drive current from sensitive inputs prevents the bounce from affecting critical signals.

Simultaneous switching output noise extends the ground bounce concept to the power supply connection. Large numbers of outputs switching together create current transients that cannot be supplied fast enough by the power distribution network, causing voltage droops that affect circuit operation and create emissions. Adequate decoupling, with capacitors providing the instantaneous current demand, addresses this mechanism at its source.

Common-Mode Current Generation

Common-mode currents flow in the same direction on all conductors of a cable, returning through displacement current to the environment or through ground connections. Unlike differential-mode currents that represent intentional signals, common-mode currents result from imbalances and unintended coupling. Several mechanisms generate common-mode currents that drive cable radiation.

Voltage differences between signal ground at the cable connection point and the signal reference at the receiving end create common-mode currents. These differences arise from ground bounce, power supply noise, or different ground potentials at the two ends of the cable. The cable acts as an antenna driven by this common-mode voltage. Ensuring clean, stable ground at cable connection points minimizes this drive mechanism.

Imbalanced signal coupling also generates common-mode currents. When a single-ended signal couples more strongly to one conductor of a cable pair than to the other, the imbalance appears as common-mode. Using balanced signaling with differential drivers and receivers inherently minimizes common-mode generation. When single-ended signals must drive cables, careful attention to symmetry in the connection design reduces imbalance.

Cable Shielding Effectiveness

Shielded cables can provide effective radiation control when the shield is properly implemented. The shield must provide a low-impedance path for common-mode currents to return without flowing on the signal conductors. This requires proper termination of the shield at both ends, with low-impedance connections to the enclosure or ground plane.

Shield termination technique dramatically affects shielding effectiveness. A 360-degree termination using a backshell or connector shell that contacts the shield continuously around its circumference provides the best high-frequency performance. Pigtail connections, where the shield braid is gathered and connected through a single wire, introduce inductance that degrades shielding effectiveness above a few megahertz. The longer the pigtail, the lower the frequency at which shielding fails.

Shield grounding at one end versus both ends presents trade-offs. Single-end grounding prevents low-frequency ground loop currents but allows the shield to act as an antenna at high frequencies. Double-end grounding provides high-frequency shielding but creates ground loops. In many cases, the solution involves double-end grounding for high-frequency shielding combined with circuit design that prevents ground loop currents from flowing in signal circuits.

Cable Routing and Dressing

Physical cable routing affects both emissions and immunity. Cables routed near noise sources pick up interference that appears at the connected circuits. Cables routed parallel to each other over long distances experience crosstalk. Cables forming loops with large enclosed areas create more efficient antennas than straight runs. Attention to cable routing during system integration can make the difference between compliance and failure.

Separating power cables from signal cables reduces coupling between the typically noisy power distribution and sensitive signals. When cables must cross, perpendicular crossing minimizes coupling. Bundling cables tightly keeps them from forming large loops, though bundling unrelated cables may increase crosstalk between them. Finding the optimal cable arrangement often requires experimentation and measurement.

Cable length affects radiation efficiency at specific frequencies. Cables with lengths near quarter wavelengths of emission frequencies radiate most efficiently. When emissions problems occur at specific frequencies, checking whether cable lengths correspond to resonant dimensions can identify the radiating structure. Changing cable length, if system constraints permit, can reduce radiation at critical frequencies.

Aperture Radiation

Any opening in a shielded enclosure allows electromagnetic energy to pass. The shielding effectiveness of an aperture depends on its size relative to the wavelength of the electromagnetic energy. Openings smaller than one-twentieth of a wavelength provide minimal leakage, while openings approaching half a wavelength can completely compromise the shielding. At 1 GHz, where the wavelength is 30 cm, even a 1.5 cm opening provides only about 20 dB of shielding.

Slot-shaped apertures are more problematic than round holes of the same area because the long dimension determines shielding effectiveness. A long, narrow slot for ventilation or cable entry radiates as effectively as if the slot were its full length wide. Many small holes provide better shielding than one large hole of the same total area, because each small hole individually provides high attenuation.

Display windows, ventilation openings, and access covers frequently create the dominant leakage paths in shielded enclosures. Conductive mesh over ventilation openings maintains shielding while allowing airflow. Conductive gaskets around cover edges ensure continuous shielding across the seam. Shielded display windows using conductive coatings or embedded mesh provide visibility while blocking electromagnetic radiation. Each aperture requires specific treatment appropriate to its function.

Seam Leakage

The seams where enclosure pieces join represent potential shielding compromises. Even when the enclosure material provides excellent shielding, a poorly designed seam can leak significantly. The seam must provide low-impedance electrical contact along its entire length to maintain shielding continuity. Gaps, particularly long slots formed by warped panels or loose fasteners, create leakage apertures.

Fastener spacing affects seam shielding effectiveness. The gap between fasteners forms a slot antenna, with the shielding effectiveness determined by the longest slot length. More closely spaced fasteners provide better high-frequency shielding. As a guideline, fastener spacing should not exceed one-twentieth of the wavelength at the highest frequency of concern for effective shielding.

Conductive gaskets improve seam shielding by providing continuous contact between panel edges. Various gasket materials and forms address different requirements: knitted wire mesh for high compression and repeated cycling, conductive elastomers for environmental sealing combined with shielding, and finger stock for lower compression requirements. The gasket must make contact with clean, conductive surfaces on both sides, requiring attention to surface finish and any non-conductive coatings.

Cable Penetration Shielding

Cables passing through shielded enclosures compromise the shield unless properly filtered or shielded at the penetration point. An unfiltered cable penetration creates an aperture that allows both conducted and radiated energy to pass. For shielded cables, proper bonding of the cable shield to the enclosure at the entry point maintains shielding continuity. For unshielded cables, filtering each conductor at the enclosure penetration provides the necessary isolation.

Connector selection significantly affects penetration shielding. Connectors designed for EMC applications provide 360-degree shield bonding and filtering options. Standard commercial connectors may provide adequate mechanical function but poor EMI performance. When using connectors not specifically designed for EMC, additional measures such as backshell adaptors, filtered connector inserts, or external filtering may be necessary.

Multiple cable penetrations grouped together in one area of the enclosure simplify shielding implementation and reduce the total aperture area compared to penetrations distributed around the enclosure. Designing the enclosure with a dedicated cable entry area, often at the rear panel, enables concentrated attention to penetration shielding and simplifies assembly.

Differential Signaling Imbalance

High-speed interfaces typically use differential signaling to achieve high data rates with low emissions and good immunity. Differential signaling ideally creates no common-mode, as the equal and opposite signals on the two conductors cancel. In practice, perfect balance is never achieved, and the imbalance generates common-mode that can radiate from cables or couple to other conductors.

Sources of imbalance include asymmetric trace routing, different loading on the two signal lines, and driver or receiver imbalance in the interface devices. Even small differences in trace length or coupling to ground create measurable common-mode at the high data rates of modern interfaces. Layout practices that maintain signal pair symmetry throughout the path from driver to receiver minimize imbalance.

Common-mode chokes on differential pairs filter common-mode while passing differential signals with minimal degradation. Properly designed chokes present high impedance to common-mode signals, attenuating the noise that would otherwise radiate from cables. The choke must be rated for the signal's data rate and must maintain adequate balance across its operating frequency range to avoid converting common-mode to differential or vice versa.

Return Path Discontinuities

High-speed signals require continuous, low-impedance return paths to maintain signal integrity and minimize radiation. Any discontinuity in the return path creates impedance changes that cause reflection and radiation. Layer transitions in multi-layer boards, gaps in reference planes, and connector transitions all potentially introduce return path discontinuities.

When high-speed signals change layers, they change reference planes, and the return current must also transition. Providing decoupling capacitors near via locations enables return current to transition between planes. The capacitors should be placed as close as possible to the signal vias, with minimized loop area in their connections to both planes. Multiple small capacitors distributed near transitions provide better high-frequency performance than single larger capacitors.

Connectors present particular challenges for return path continuity. The signal pin assignments and ground pin placements in standard connectors may not provide optimal return paths. Using connectors with abundant ground pins interleaved with signals improves return path quality. When connector selection is constrained by interface standards, careful attention to the PCB layout approaching the connector minimizes the impact of connector-related discontinuities.

Spread Spectrum and Clock Recovery

Many high-speed interfaces use spread spectrum clocking to reduce peak emissions while maintaining the same total energy. The embedded clock in serial interfaces recovers from the data transitions, so spreading the data rate spreads both the data and the clock-related emissions. This approach can provide several decibels of emissions reduction at harmonic frequencies.

Interface standards specify limits on the amount of spread spectrum that the receiver can tolerate while maintaining reliable clock recovery. Exceeding these limits causes bit errors even as it reduces emissions. When spread spectrum is employed, ensuring that the implementation complies with the interface specification prevents creating reliability problems while solving EMI problems.

Spread spectrum is not appropriate for all high-speed interfaces. Some applications require precise, stable frequencies that preclude spreading. Systems that must interoperate with equipment expecting fixed frequencies cannot use spread spectrum. When spread spectrum cannot be used, alternative emissions reduction techniques such as improved shielding, filtering, or layout optimization must provide the needed margin.

Spectral Signature Analysis

The frequency spectrum of emissions often reveals the source type. Narrowband emissions at harmonics of a known frequency point to that source: clock harmonics appear at exact multiples of the clock frequency, switching supply harmonics at multiples of the switching frequency. The harmonic rolloff rate indicates edge speed, with faster edges maintaining higher harmonics. Spectral signatures thus provide immediate clues about likely sources.

Broadband emissions suggest different sources than narrowband harmonics. Broadband noise may indicate arcing, corona discharge, or wideband noise sources such as switched-mode controller noise. The noise floor in a spectrum may rise due to broadband sources even when no distinct spectral peaks appear. Distinguishing broadband from narrowband emissions guides the search toward appropriate source types.

Modulation on carriers provides additional diagnostic information. Power line frequency modulation (50 or 60 Hz sidebands) indicates coupling from mains power. Modulation at the switching supply frequency suggests power supply noise coupling. Data modulation may appear on clock harmonics in digital systems. Examining the fine structure of spectral peaks reveals these modulation effects and their underlying causes.

Correlation with System Activity

Correlating emissions with system operating states identifies the circuits responsible. Emissions that change when specific functions are enabled point to those functions as sources. Testing with minimal system activity establishes a baseline, then progressively enabling functions identifies which contribute to emissions. This systematic approach efficiently narrows the search in complex systems with many potential sources.

Loading effects reveal coupling mechanisms. Emissions that change when cables are connected or disconnected indicate cable radiation. Changes with different loads on power supplies may reveal common impedance coupling. Variations with enclosure panels in place versus removed show the contribution of shielding and any resonances created by enclosure dimensions. These loading tests provide information about coupling paths in addition to sources.

Environmental factors can affect emissions in ways that provide diagnostic information. Temperature changes may shift resonant frequencies or affect component values. Vibration may create intermittent contacts in shields or grounds. Humidity changes affect surface resistance and electrostatic effects. While typically undesirable, these sensitivities can provide clues when emissions vary unexpectedly.