The rapid advancement of power electronic technology has led to the widespread adoption of Silicon Carbide (SiC) devices in electric drive systems due to their superior switching speeds and efficiency. However, the high du/dt and di/dt associated with these fast switches exacerbate Electromagnetic Interference (EMI) problems. The conducted EMI generated can severely impact the safe operation of the system itself and surrounding equipment, making effective suppression crucial for ensuring electromagnetic compatibility (EMC).
Traditional EMI suppression methods primarily involve passive EMI filters (PEFs) and spread spectrum modulation strategies. PEFs, constructed from inductors and capacitors, are effective but often lead to increased volume and weight, particularly when targeting low-frequency noise, thereby limiting power density. Spread spectrum modulation strategies, such as Random Switching Frequency PWM (RSF-PWM), work by dispersing concentrated harmonic energy over a wider frequency band, reducing peak noise levels. However, they merely redistribute rather than eliminate interference energy, struggle with residual noise outside harmonic bands, and can degrade the control stability of the electric drive system. Therefore, a synergistic approach that combines the broadband attenuation capability of filters with the peak-shaving effect of modulation is necessary for comprehensive EMI management in modern high-performance electric drive systems.
Analysis of Conducted Interference Suppression and Control Performance Impact of Random Spread Spectrum Modulation Strategies
Generation Mechanism of Conducted Interference in Electric Drive Systems
In a typical permanent magnet synchronous motor (PMSM) electric drive system, conducted EMI manifests as differential-mode (DM) and common-mode (CM) noise. CM currents primarily flow through the parasitic capacitances between the inverter bridge midpoints and ground, forming a loop with the Line Impedance Stabilization Network (LISN). DM currents circulate between the phase lines. The fast switching transitions of SiC MOSFETs are the primary sources of these interference currents.
Suppression Effect of Random Spread Spectrum Modulation on Conducted Interference
When a fixed carrier frequency Space Vector Pulse Width Modulation (SVPWM) strategy is used, significant harmonic energy concentrates at the switching frequency and its multiples, creating distinct peaks in the spectrum. The RSF-SVPWM strategy mitigates this by randomly varying the switching frequency, thereby spreading the harmonic energy over a defined spread spectrum range \(\Delta f\). The switching period for the \(k\)-th cycle is given by:
$$
T_{sk} = T_{s0} + R_k \Delta T
$$
where \(T_{s0}\) is the central switching period, \(\Delta T\) is the variation range (\( \Delta f = 1/\Delta T \)), and \(R_k\) is a random number within \([-1, 1]\).
Analyzing the Fourier transform of the output line voltage \(v_{AB}\) reveals that the amplitude of high-order harmonics is influenced by the DC bus voltage \(V_{DC}\), the average switching frequency \(f_{s0} (=1/T_{s0})\), the spread spectrum range \(\Delta f\), and the random sequence \(R_k\). The expression for the Fourier transform \(S_{v_{AB}}(f)\) is complex but demonstrates that a larger \(\Delta f\) leads to a reduction in the peak amplitude of harmonics at the switching frequency multiples. However, when \(\Delta f\) exceeds approximately 50%, residual noise in the sidebands begins to accumulate, potentially elevating the interference level across the spread band and limiting the overall suppression benefit.
Impact of Random Spread Spectrum Modulation on Control Performance
While increasing \(\Delta f\) suppresses EMI peaks, it adversely affects the control performance of the electric drive system. In fixed-frequency SVPWM, the sampling period aligns with the constant PWM period, resulting in a fixed control delay. Under RSF-SVPWM with a fixed sampling rate, the varying switching period introduces a variable delay, leading to accumulated sampling errors.
This manifests as a non-uniform rotation increment \(\theta_{sk} = 2\pi f_m / f_{sk}\) for the voltage vector in the complex plane, compared to the constant \(\theta_{s0}\) in fixed-frequency modulation. Larger \(\Delta f\) causes \(\theta_{sk}\) to switch between wider extremes, increasing waveform distortion and current ripple, thereby degrading the control performance of the electric drive system.
Proposed Collaborative EMI Noise Suppression Method Combining Quarter-Frequency Spread Spectrum Modulation and Passive Filter
Quarter-Average Random Switching Frequency SVPWM (QARSF-SVPWM) Strategy
To balance EMI suppression and control stability, a Quarter-Average Random Switching Frequency SVPWM (QARSF-SVPWM) strategy is proposed. The core idea is to divide the total spread spectrum range \(\Delta f\) into four sub-ranges. The system operates in each sub-range for a duration \(T_b/4\), where \(T_b\) is the base period, typically set to \(1/f_m\) (the fundamental period) for harmonic distribution uniformity. The central frequencies for the four phases are defined as:
$$
\begin{aligned}
f_{s1} &= f_{s0} – \frac{3}{4}\Delta f \\
f_{s2} &= f_{s0} – \frac{1}{4}\Delta f \\
f_{s3} &= f_{s0} + \frac{1}{4}\Delta f \\
f_{s4} &= f_{s0} + \frac{3}{4}\Delta f
\end{aligned}
$$
Within each phase, the switching frequency varies randomly within \(\pm \Delta f/8\) around its central frequency \(f_{si}\). This approach reduces the average relative change \(\delta_k\) in switching frequency between consecutive cycles compared to standard RSF-PWM. A smaller \(\delta_k\), as derived from the harmonic amplitude equation, leads to lower high-frequency harmonic content. More importantly, by limiting the instantaneous frequency jump, the variation in the voltage vector rotation increment \(\theta_{sk}\) is minimized, thereby preserving the control performance of the electric drive system close to that of fixed-frequency SVPWM while maintaining effective EMI dispersion.
Optimized EMI Filter Parameter Design Method
Traditional EMI filter design based on Insertion Loss (IL) does not fully account for the dynamic noise spectrum reshaped by spread spectrum modulation. QARSF-SVPWM reduces peaks at \(nf_s\) but elevates the noise floor across the spread band \(\Delta f\). A collaborative design flow integrating the modulation parameters (\(f_{s0}, \Delta f\)) with filter synthesis is proposed.
- Modulation Strategy Application: Implement QARSF-SVPWM with a chosen \(f_{s0}\) and initial \(\Delta f\) (e.g., ±40%).
- Noise Measurement & IL Requirement: Measure the DM/CM noise spectrum without a filter. Determine the required IL profile from the EMC standard limit line.
- Initial Corner Frequency Selection: Find the intersection point (freq_min, IL_req) between the measured noise and the required IL. Identify the nearest switching harmonic order \(n\). The initial filter corner frequency \(f_c\) is set conservatively to:
$$f_c = n f_{s0} – \Delta f / 4$$
This accounts for the frequency spreading, ensuring the filter provides sufficient attenuation at the lower edge of the spread harmonic band. - Filter Parameter Calculation: Select an appropriate filter topology (e.g., π-type with separate CM and DM sections). For a simple single-stage LC filter, the corner frequency is:
$$f_c = \frac{1}{2\pi\sqrt{LC}}$$
Component values (L, C) are calculated based on the desired \(f_c\) and impedance matching principles. - Iterative Validation and Adjustment:
- Check if the suppressed noise in the 10 kHz–200 kHz band meets the standard with the designed filter. If not, increase \(\Delta f\) up to a maximum (e.g., 50%). If still non-compliant, reduce the corner frequency by recalculating with \(n = n-1\) (i.e., \(f_c = (n-1)f_{s0} – \Delta f / 4\)) and repeat.
- Check the >200 kHz band. If non-compliant, also trigger a reduction in the corner frequency calculation.
This iterative process ensures the filter is optimally sized to handle the smeared noise spectrum from the QARSF-SVPWM strategy, avoiding overdesign while achieving full-band compliance.
Experimental Verification
A 5 kW SiC-based PMSM electric drive system experimental platform was built to validate the proposed method. The key parameters are listed in Table 1.
| Parameter | Value |
|---|---|
| Output Power, \(P_o\) | 5 kW |
| DC Input Voltage, \(V_{in}\) | 270 V |
| Average Switching Frequency, \(f_{s0}\) | 20 kHz |
| Spread Spectrum Range, \(\Delta f\) | ±40% (for QARSF-SVPWM) |
| Motor Speed (Test Condition) | 6,000 rpm |
| Motor Torque (Test Condition) | 2.5 N·m |
Table 1: Design parameters of the experimental electric drive system platform.
The control performance under QARSF-SVPWM was first verified. The results showed that the d-axis and q-axis current ripples, as well as the phase current waveform, were nearly identical to those under fixed-frequency SVPWM, confirming the strategy’s minimal impact on control performance.
Following the proposed design flow, the EMI filter parameters were determined as: Common-Mode Inductance \(L_{CM} = 1 \text{ mH}\), Common-Mode Capacitance \(C_Y = 220 \text{ nF}\), Differential-Mode Inductance \(L_{DM} = 10 \mu\text{H}\), and Differential-Mode Capacitance \(C_X = 2.2 \mu\text{F}\).
The conducted EMI measurements compared four scenarios: 1) Fixed SVPWM, 2) Fixed SVPWM with PEF, 3) QARSF-SVPWM alone, and 4) QARSF-SVPWM with PEF. The results demonstrated that the combined approach (Scenario 4) yielded the best overall suppression. Crucially, in the critical 10 kHz–200 kHz band where passive filters are bulky, the synergistic method provided an additional 5–10 dBμV attenuation for DM noise and up to 5 dBμV for CM noise compared to the traditional method of fixed SVPWM with PEF (Scenario 2). The system successfully complied with the relevant EMC standard limits across the full frequency range.
Conclusion
This paper addresses the challenging conducted EMI issues in SiC-based electric drive systems. The analysis confirms that while random spread spectrum modulation reduces peak noise, it can increase sideband noise and impair control stability. The proposed Quarter-Average Random Switching Frequency SVPWM (QARSF-SVPWM) strategy mitigates these drawbacks by dividing the frequency spread into four segments, thereby preserving control performance.
The core contribution is a collaborative EMI suppression methodology that synergistically combines the QARSF-SVPWM strategy with an optimized passive EMI filter. The filter design explicitly accounts for the spread spectrum parameters (average frequency \(f_{s0}\) and range \(\Delta f\)), using an adjusted corner frequency \(f_c = n f_{s0} – \Delta f / 4\) to effectively attenuate the smeared interference spectrum. Experimental results on a 5 kW SiC motor drive validate the method’s effectiveness, showing significant additional noise reduction in the low-to-mid frequency range compared to conventional techniques, without compromising the dynamic performance of the electric drive system. This approach provides a practical solution for achieving high power density and robust EMC in advanced electric drive applications.