Why EMC Is a First-Order Design Constraint in EVs

Electromagnetic compatibility in electric vehicles is not a compliance afterthought — it is a fundamental systems-engineering discipline that must be designed in from the earliest architecture decisions. The coexistence of high-voltage (HV) power electronics, including traction inverters, DC-DC converters, and on-board chargers (OBC), with low-voltage (LV) systems such as CAN/LIN communication buses, ADAS sensors, and infotainment units creates electromagnetic interference (EMI) coupling paths that, if unmanaged, can cause functional failures, data corruption, and regulatory non-compliance. According to IEEE research on power electronics, fast switching transients in wide-bandgap semiconductor devices — increasingly used in EV inverters for their efficiency advantages — produce voltage slew rates that generate broadband conducted and radiated emissions extending well into the frequency ranges occupied by LV electronics.

The transition from 400 V to 800 V HV bus architectures in premium EV platforms amplifies this challenge considerably. Higher bus voltages mean larger voltage swings during switching events, which directly increases the amplitude of common-mode and differential-mode noise injected into the vehicle’s electrical network. Simultaneously, the proliferation of LV electronics — modern EVs can carry more than 100 electronic control units (ECUs) — creates a dense electromagnetic environment where coupling paths are numerous and difficult to predict without systematic modelling.

Engineers who treat EMC as a late-stage validation exercise rather than a concurrent design discipline consistently encounter the most costly outcome: fundamental architectural changes required after prototype build. Rerouting HV harnesses, adding unplanned filter stages, or redesigning PCB layouts at prototype stage can add months to a programme and significant cost. The engineering community — as documented in SAE technical publications — has converged on a systems-level approach where EMC requirements cascade from vehicle-level targets down to subsystem and component specifications before a single component is selected.

Understanding the Noise Sources: Inverters, DC-DC Converters, and OBC Systems

The traction inverter is the dominant EMI source in an EV powertrain, generating both conducted emissions that travel along HV cables and radiated emissions that propagate through the vehicle structure. Pulse-width modulation (PWM) switching of the inverter’s power transistors — whether silicon IGBTs or silicon carbide MOSFETs — produces current and voltage waveforms rich in harmonics. The fundamental switching frequency and its harmonics up to and beyond 30 MHz fall within the frequency ranges covered by CISPR 25, requiring careful filter design on both the HV DC input side and the AC output side of the inverter.

The DC-DC converter — which steps the HV battery voltage down to 12 V or 48 V for LV loads — is the second major conducted-noise source. Its switching action injects noise directly onto the LV power rail, which is shared by ECUs, sensors, and communication transceivers. Because the LV rail is the power supply for virtually every sensitive electronic system in the vehicle, noise on this rail represents a direct coupling path to the most susceptible circuits. Engineers must specify EMC performance requirements for the DC-DC converter that account for both its own switching emissions and the noise it may couple from the HV side.

The on-board charger introduces a third noise source that is active during charging events rather than driving. OBC switching frequencies and the associated harmonic content must comply with both vehicle-level EMC standards and grid-side power quality regulations. In bidirectional OBC designs — increasingly common for vehicle-to-grid (V2G) applications — the EMC filter must function effectively in both power flow directions, which complicates the filter design considerably.

“The traction inverter, DC-DC converter, and on-board charger each introduce distinct switching noise profiles that overlap in frequency with the communication and sensing bands of low-voltage electronics — making isolation and filtering a multi-source, multi-path engineering problem.”

Grounding Architectures and Shielding Strategies

Grounding architecture is the single most consequential EMC design decision at the vehicle system level. A poorly designed ground network creates impedance differences between subsystem reference points, which drive ground currents that act as coupling paths for HV switching noise into LV circuits. The recommended approach — widely adopted across automotive OEM design guidelines and documented in standards published by ISO — is a star-point topology in which all subsystem grounds converge at a single low-impedance reference node, typically the vehicle chassis at a defined location near the HV battery.

In HV systems, the battery pack is intentionally isolated from chassis ground to provide a safety barrier against electric shock in the event of a single-point insulation fault. This isolation is maintained and continuously monitored by an insulation monitoring device (IMD), which detects degradation of the isolation resistance between the HV conductors and chassis. The IMD itself must be designed to operate without introducing additional EMI coupling, since it injects a measurement signal onto the HV network. The interaction between the IMD test signal, the HV bus, and the chassis ground structure requires careful impedance management to avoid false-positive fault detection triggered by EMI rather than genuine insulation degradation.

Shielding is applied at three distinct levels in a well-designed EV EMC architecture. At the cable level, HV traction cables use foil-and-braid composite shields that provide attenuation across the full CISPR 25 frequency range from 150 kHz to 30 MHz. At the module level, power electronics enclosures are constructed from conductive materials — typically aluminium die-cast housings — with gasket seals at all joints to prevent radiated emissions from escaping through apertures. At the vehicle level, the body structure itself acts as a partial Faraday cage, though the increasing use of composite and non-metallic body panels in EVs reduces this natural shielding effectiveness and places greater burden on component-level shielding.

PCB-level grounding within LV ECUs that must operate near HV components requires particular attention. Split ground planes — separating analogue, digital, and power sections — reduce the impedance of high-frequency return currents and prevent digital switching noise from coupling into analogue signal paths. The connection between split planes must be managed carefully: a single-point connection at the signal entry point prevents ground loops while maintaining a defined reference, whereas multiple connections create the loops that allow magnetically coupled noise to drive currents through the ground plane.

EMC Filtering and Spread-Spectrum Modulation Techniques

EMC filters for EV power electronics must simultaneously suppress common-mode and differential-mode conducted emissions across the 150 kHz to 30 MHz range defined by CISPR 25, while handling the full HV bus voltage and peak current of the traction system without saturation, thermal failure, or significant efficiency loss. This combination of electrical performance requirements makes EV EMC filter design substantially more demanding than equivalent filters in industrial power electronics, where space, weight, and operating temperature constraints are less severe.

A typical HV EMC filter assembly consists of a differential-mode inductor (or integrated common-mode choke with differential-mode leakage inductance), common-mode choke, X-capacitors across the HV bus for differential-mode suppression, and Y-capacitors from each HV conductor to chassis ground for common-mode suppression. The Y-capacitor values are constrained by the maximum allowable chassis leakage current — defined by safety standards including those published by IEC — which limits the total capacitance to chassis and therefore limits the attenuation achievable from capacitive filtering alone. This constraint drives the requirement for high-inductance common-mode chokes to achieve adequate high-frequency attenuation.

Spread-spectrum PWM modulation addresses the EMC problem from the source rather than the filter. By continuously varying the inverter switching frequency within a defined band — typically a few percent above and below the nominal frequency — the harmonic energy that would otherwise concentrate at discrete frequencies is distributed across a wider bandwidth. This reduces the peak spectral amplitude at any single frequency, which is the quantity measured by quasi-peak and average detectors in CISPR 25 testing. Spread-spectrum modulation is particularly effective at reducing the fundamental switching frequency harmonic and its lower-order multiples, where filter attenuation is limited by the size and weight constraints on inductors.

The interaction between spread-spectrum modulation and motor control performance must be carefully managed. Varying the switching frequency changes the current ripple characteristics in the motor windings and affects torque ripple, acoustic noise, and thermal distribution in the inverter. Motor control algorithms must be co-designed with the spread-spectrum strategy to ensure that the EMC benefit does not come at the cost of drivability or NVH (noise, vibration, and harshness) performance. This co-design requirement is one reason why EMC engineering in EV development is increasingly integrated into the powertrain systems team rather than treated as a separate specialisation.

For the LV power supply rail, the DC-DC converter’s output filter must attenuate switching noise to levels compatible with the most sensitive LV loads — typically CAN transceivers and analogue sensor circuits with noise immunity thresholds in the millivolt range. This often requires a two-stage filter: a first stage close to the DC-DC converter output to handle the bulk of the switching harmonic content, and a second stage at the point of load for particularly sensitive circuits. Ferrite bead inductors, which provide high impedance at high frequencies with minimal DC resistance, are widely used in the second stage of LV filtering.

Standards Landscape and Validation Methodology

CISPR 25 is the primary international standard governing conducted and radiated emissions from vehicle components and modules, establishing limit classes from 1 (least stringent) to 5 (most stringent) across frequency bands from 150 kHz to 2.5 GHz. Most automotive OEMs specify Class 5 limits for components in their EMC procurement specifications, reflecting the dense electromagnetic environment inside a modern vehicle. CISPR 25 is published by the IEC through its CISPR committee and is harmonised with regional regulations in Europe, North America, and Asia-Pacific markets.

ISO 11452 provides the complementary immunity test framework, specifying methods for exposing vehicle components to controlled electromagnetic fields and conducted disturbances to verify that they continue to function correctly in the presence of interference. The standard covers bulk current injection (BCI) for conducted immunity, strip-line and TEM cell methods for radiated immunity, and direct power injection (DPI) for component-level immunity at the IC pin level. Together, CISPR 25 and ISO 11452 define the full EMC test programme that a component must pass before it can be released for vehicle integration.

SAE J551 addresses vehicle-level EMC performance, defining test methods and limits for the complete vehicle rather than individual components. Vehicle-level testing captures system-integration effects — such as resonances in the vehicle wiring harness, coupling between subsystems through the chassis, and the cumulative emission contribution of multiple simultaneous noise sources — that cannot be predicted from component-level test results alone. OEMs typically conduct vehicle-level EMC testing at accredited facilities equipped with anechoic chambers and absorber-lined shielded rooms capable of measuring emissions down to the noise floor required by Class 5 limits.

The validation process follows a V-model development methodology. EMC requirements are defined at the top of the V alongside vehicle-level performance targets, then cascaded into subsystem and component specifications at the design phase. Simulation tools — including SPICE-based circuit simulation for conducted emissions and full-wave electromagnetic solvers such as FDTD and FEM for radiated emissions — are used throughout the design phase to predict compliance before hardware is built. Pre-compliance testing on early prototypes identifies issues while design changes are still low-cost, and formal compliance testing against CISPR 25 and ISO 11452 is conducted on production-representative hardware before homologation submission.

The increasing use of wide-bandgap semiconductors — silicon carbide (SiC) and gallium nitride (GaN) — in EV power electronics is driving a re-evaluation of existing EMC filter designs. These devices switch significantly faster than silicon IGBTs, with rise times below 20 nanoseconds in production designs, which shifts the dominant harmonic content to higher frequencies and increases the bandwidth over which filters must provide attenuation. The EMC engineering community, as reflected in standards evolution work at bodies including IEEE, is actively developing updated test methodologies and filter design guidance to address the specific emission characteristics of SiC and GaN-based converters in EV applications.