Electronics Guide

Switching Power Supply Emissions

Switching regulators and power supplies generate conducted emissions through several mechanisms. The fundamental switching action creates pulsating current at the switching frequency and its harmonics. A buck converter, for example, draws current from the input only during the on-time of the switching transistor, creating a rectangular current pulse whose harmonic content extends to frequencies determined by the edge rates. These pulsating currents, flowing through the finite impedance of input capacitors and interconnecting conductors, create differential-mode voltage noise.

Common-mode noise in switching power supplies arises primarily from rapid voltage transitions coupling through parasitic capacitances. In a flyback converter, the switching transistor's drain voltage swings rapidly between near zero and the input voltage plus the reflected output voltage. This rapid voltage change, occurring across the parasitic capacitance between the transistor and its heatsink or the circuit's grounding structure, injects common-mode current into the system. Similarly, transformer interwinding capacitance couples primary-side switching noise to the secondary, creating common-mode current on output cables.

The rectifier diodes in AC-DC converters contribute to conducted emissions through their reverse recovery behavior. When a diode switches from conducting to blocking, the reverse recovery current creates high-frequency ringing that appears as conducted emissions. Fast-recovery diodes reduce the duration of this event but may create sharper transitions with more high-frequency content. Soft-recovery diodes or silicon carbide devices offer improved EMC performance at additional cost.

Digital Circuit Contributions

High-speed digital circuits contribute to conducted emissions through their dynamic current demands and through coupling of digital switching noise onto power supply lines. Simultaneous switching of many logic gates creates transient current demands that cannot be fully supplied by local decoupling, causing noise to propagate back toward the power source. This mechanism creates conducted emissions at frequencies related to the digital clock and its harmonics.

Ground bounce in digital circuits creates noise between circuit ground and chassis or earth ground that can appear as common-mode conducted emissions. When switching currents flow through the inductance of ground connections, the resulting voltage appears as noise between the power supply return and external references. This mechanism depends on the inductance of ground paths and the magnitude and di/dt of switching currents.

Clock circuits and high-speed interfaces generate particularly significant noise due to their continuous, periodic operation. The spectral content of this noise concentrates at harmonics of the fundamental frequency, creating narrowband emissions that must comply with limits at each specific frequency. Digital noise coupling to power supply conductors adds to power supply switching noise, potentially creating emissions that exceed limits even when each source alone would comply.

Motor and Actuator Noise

Motors, particularly those with brush commutation, generate broadband conducted emissions from arcing at the brush-commutator interface. This arcing creates high-frequency noise across a wide spectrum, often appearing as an elevated noise floor rather than discrete harmonics. Filtering motor noise requires attention to the broadband spectrum rather than just specific frequencies. Brushless motors eliminate commutation noise but the drive electronics create switching-related emissions similar to power supplies.

Solenoids, relays, and other inductive actuators create transients when switched that can appear as conducted emissions. The rapid interruption of current through an inductor creates high voltage spikes that, while typically suppressed for voltage protection purposes, may leave residual high-frequency content. Snubber circuits designed for transient suppression may need optimization for EMC performance as well as voltage limiting.

Differential-Mode Noise

Differential-mode noise flows in opposite directions on the line and neutral conductors, representing the noise component of the equipment's input current. This noise results from the non-constant current drawn by rectifiers and switching circuits. The noise voltage appears between the line and neutral at the LISN measurement ports, with opposite polarity on the two conductors. Adding the two LISN outputs vectorially yields twice the differential-mode component, while subtracting yields twice the common-mode component.

Differential-mode noise typically dominates at lower frequencies, from the lowest regulated frequency up through several hundred kilohertz. The spectral content reflects the switching operation, with harmonics of the fundamental switching frequency extending across this range. The amplitude of differential-mode harmonics depends on the input current waveform shape, which in turn depends on the switching topology and operating conditions.

Filtering differential-mode noise uses series inductors on the line and neutral conductors, combined with capacitors across the line. The inductors present high impedance to differential-mode currents, while the X-class capacitors across the line shunt remaining noise around the source. The filter cutoff frequency and the number of filter sections determine the attenuation achieved at each frequency.

Common-Mode Noise

Common-mode noise flows in the same direction on both line and neutral conductors, returning through the ground connection or through parasitic capacitance to earth. This noise results from voltage transitions coupling through parasitic capacitances between switching nodes and grounded structures. Both LISN outputs show the same polarity for common-mode noise, allowing its identification through measurement comparison.

Common-mode noise often dominates at higher frequencies, typically above several hundred kilohertz. The parasitic capacitance coupling this noise acts as a decreasing impedance with increasing frequency, allowing more noise current at higher frequencies. The spectral content still relates to switching frequencies, but the amplitude versus frequency characteristic differs from differential-mode due to the capacitive coupling mechanism.

Common-mode filtering uses common-mode chokes, inductors wound so that differential-mode currents experience no net inductance while common-mode currents see high impedance. Y-class capacitors from each line to ground provide a low-impedance return path for common-mode currents, shunting them around the equipment. Safety regulations limit Y-capacitor values to control leakage current, constraining the filtering that can be achieved with capacitors alone.

Mode Separation Techniques

Separating differential-mode and common-mode components during measurement enables targeted troubleshooting. Commercial mode separation networks split the LISN outputs into their differential and common-mode components for simultaneous measurement. These networks use resistive combining networks or transformer-based splitters to perform the vector sum and difference operations that extract each mode.

When mode separation networks are not available, sequential measurements can provide the same information. Measuring each LISN output separately, then comparing the spectra, reveals the mode balance. If both measurements show similar amplitudes and spectral shapes, common-mode noise dominates. If the spectra differ, differential-mode noise contributes significantly. More quantitative analysis requires vector measurement of both outputs, accounting for both amplitude and phase.

Understanding which mode dominates at each frequency guides filter component selection. Frequencies dominated by differential-mode noise require differential filtering with series inductors and X-capacitors. Frequencies dominated by common-mode noise require common-mode chokes and Y-capacitors. Mixed situations may require combined filtering or multiple filter stages addressing each mode separately.

Filter Topology Selection

Single-stage filters using one inductor and one or more capacitors provide moderate attenuation with minimal components. The typical single-stage filter uses a series inductor followed by a shunt capacitor, creating a low-pass response with 40 dB per decade rolloff above the cutoff frequency. This basic topology suffices when the noise margin requirement is modest or when space and cost constraints limit complexity.

Multi-stage filters achieve higher attenuation by cascading multiple filter sections. Each additional LC section adds another 40 dB per decade to the rolloff, though component parasitics eventually limit the achievable attenuation. Two-stage or three-stage filters are common when significant noise reduction is required. The stages may be identical or may use different component values to spread the resonant frequencies and achieve smooth attenuation across the frequency range.

Combining differential and common-mode filtering addresses both noise types efficiently. The common-mode choke, wound with both line and neutral conductors on the same core, provides common-mode filtering while contributing minimal differential-mode inductance. X-capacitors across the line and Y-capacitors to ground complete the filter. This combined approach provides more complete filtering than separate structures would in the same volume.

Component Selection

Inductor selection for power line filters must consider both the inductance value and the current-carrying capability. Core material affects both the inductance stability with current and the high-frequency performance. Ferrite cores provide good high-frequency performance but may saturate at high current. Powdered iron cores handle higher current without saturation but may have lower permeability. The core material should be chosen for the specific application requirements.

Capacitor selection must address both the capacitance value and the voltage rating, along with safety certification requirements. X-class capacitors, rated for across-the-line connection, must safely fail without fire hazard. Y-class capacitors, rated for line-to-ground connection, must meet stringent safety requirements and are limited in value to control leakage current. Both types must have adequate voltage ratings for the application, including transient overvoltages.

Component parasitics limit filter performance at high frequencies. Inductor winding capacitance creates a parallel resonance that limits the frequency range over which the inductor provides effective impedance. Capacitor equivalent series inductance prevents the capacitor from providing low impedance above its self-resonant frequency. Understanding these limitations ensures that components are used within their effective frequency ranges.

Layout and Installation

Physical layout significantly affects filter performance. Input and output connections should be separated to prevent noise from bypassing the filter through stray coupling. Keeping filter components close together minimizes the loop areas that can pick up noise. Mounting the filter at the point where power enters the equipment ensures that noise is attenuated before it reaches the power conductors.

Ground connections for Y-capacitors require careful attention. These capacitors must connect to a stable ground reference with low impedance at the frequencies of concern. Long wire leads add inductance that limits high-frequency effectiveness. Using ground plane connections or short, wide traces to chassis ground maintains the capacitor's high-frequency performance.

Shielding between filter input and output prevents capacitive coupling that bypasses the filter. Metal partitions between input and output sections of filter housings provide this isolation. When filters are integrated on circuit boards, ground plane shields or physical separation between input and output traces prevents coupling. The effectiveness of the shielding determines the maximum attenuation the filter can achieve regardless of the component values.

Switching Circuit Optimization

Reducing the rate of change of voltage and current during switching transitions decreases the high-frequency content of switching noise. Snubber circuits across switching devices slow the transitions, trading switching loss for EMI improvement. Gate resistors in transistor drive circuits reduce switching speed with similar trade-offs. Soft-switching topologies achieve low-loss transitions by switching when voltage or current is naturally low, reducing both switching loss and noise.

Minimizing parasitic inductance in the power switching loop reduces the amplitude of switching transients. Tight layout of the switch, diode, and input capacitor keeps loop area small. Multi-layer printed circuit board construction enables placement of critical components directly over their return path. Integrated power stages that include switch and diode in a single package achieve the lowest possible loop inductance.

Input capacitor selection affects both differential and common-mode noise. Bulk capacitors provide energy storage but may have limited high-frequency effectiveness. Ceramic capacitors provide low impedance at high frequencies but limited bulk capacitance. Combining both types, with ceramics placed closest to the switching devices, addresses the full spectrum of noise frequencies.

Transformer Design

In isolated power supplies, the transformer can be a significant source of common-mode noise through interwinding capacitance. Minimizing this capacitance through interleaving of primary and secondary windings, adding shielding windings, or using sectioned bobbin construction reduces the noise coupled to the secondary. The trade-off involves increased manufacturing complexity and potentially reduced efficiency or increased leakage inductance.

Faraday shields between primary and secondary windings intercept the capacitively coupled noise and return it to the primary circuit rather than allowing it to flow to the secondary. The shield must be connected to the appropriate reference point, typically the primary circuit ground, with a low-inductance connection. Shield construction must ensure complete coverage without creating a shorted turn that would cause losses.

Core and winding resonances can amplify noise at specific frequencies. Careful design of winding structure and layer arrangement avoids resonances in the frequency range where noise is problematic. Adding damping through resistive elements or lossy magnetic materials can reduce resonant amplification when design changes alone cannot avoid the resonances.

Rectifier Improvements

Rectifier reverse recovery noise can be reduced through component selection and circuit techniques. Fast-recovery diodes reduce the duration of reverse recovery but may not reduce the peak current. Soft-recovery diodes control the rate of current change during recovery, reducing high-frequency content. Silicon carbide diodes offer nearly zero reverse recovery, eliminating this noise source at premium cost.

Snubber circuits across rectifier diodes damp the ringing that follows reverse recovery. RC snubbers provide simple, effective damping at the cost of some power dissipation. The resistor value should be chosen to critically damp the ringing; the capacitor value then sets the snubber impedance relative to the circuit impedance. Optimizing the snubber requires measurement of the ringing frequency and amplitude.

Rearranging the rectifier configuration can sometimes reduce noise. Using a bridge rectifier with controlled sequencing of diode conduction rather than allowing natural commutation can reduce transients. Adding small inductors in series with diodes limits di/dt during switching, reducing both the amplitude and frequency content of transients.

LISN Setup and Usage

The LISN must be properly installed to provide valid measurements. The LISN's ground terminal should connect to the reference ground plane with a low-impedance connection. The equipment under test should be positioned to minimize coupling to the LISN and ground plane other than through the intended power connection. When using multiple LISNs for multi-phase or neutral connections, each LISN should be properly grounded and positioned.

LISN maintenance affects measurement accuracy. The internal components can drift or degrade over time, changing the LISN's impedance characteristic. Periodic calibration verification ensures that the LISN presents the correct impedance to the equipment under test. Calibration certificates document the LISN's performance and establish traceability to measurement standards.

Auxiliary equipment grounding during conducted emissions tests requires attention. Equipment such as oscilloscopes or computers used to control the test should not introduce ground loops that affect measurements. Battery operation or isolation transformers for auxiliary equipment prevent ground loop currents from flowing through the measurement setup.

Diagnostic Measurements

Comparing emissions at different operating conditions isolates specific noise sources. Varying load, input voltage, or operating mode while monitoring conducted emissions reveals how each condition affects the noise spectrum. Sources that vary with switching frequency shift their harmonic pattern predictably. Sources related to load current change amplitude with load. This correlation guides identification of the dominant noise sources at each frequency.

Temporarily adding filtering at various points in the system reveals which filtering stages provide attenuation at each frequency. Clip-on ferrites on cables, temporary capacitors at key nodes, or provisional filter stages can be evaluated quickly. Measuring the noise reduction provided by each temporary modification guides permanent filter design and identifies the most effective filtering locations.

Internal measurements within the equipment complement LISN measurements at the power input. Current probes on internal power distribution reveal where noise currents flow. Voltage probes at filter component connections show how noise is distributed throughout the filtering chain. These internal measurements help optimize filter performance and identify component or layout issues that limit filter effectiveness.

Documentation and Trending

Recording conducted emissions results throughout the development cycle enables tracking of EMC performance. Early measurements establish baseline performance before filtering is implemented. Measurements after each filter modification track progress toward compliance. Final pre-compliance measurements before formal testing provide confidence in the outcome. This documentation supports troubleshooting if later issues arise.

Photographing the measurement setup ensures that conditions can be reproduced. Equipment positioning, cable routing, LISN connections, and any auxiliary equipment should be documented. Photographs supplement written test procedures by capturing details that might otherwise be overlooked. Consistent setups enable valid comparison between measurements.

Comparing results across similar products reveals common issues and successful solutions. When a new product uses similar circuitry to previous designs, the EMC experience with those designs guides expectation and troubleshooting for the new product. Building this institutional knowledge through documented results improves efficiency across the product portfolio.