The transition towards sustainable transportation has placed battery electric vehicles (battery EV cars) at the forefront of automotive innovation. A critical subsystem within any modern battery EV car is the on-board power supply, responsible for efficiently stepping down the high-voltage from the traction battery to stable lower voltage levels (e.g., 12V or 24V) required for auxiliary systems, control units, and low-voltage batteries. The efficiency, reliability, and power density of this DC-DC converter directly impact the overall range and performance of the battery EV car. Among various topologies, the Phase-Shifted Full-Bridge (PSFB) converter has been widely adopted due to its ability to achieve soft-switching, which minimizes switching losses and enables higher frequency operation, leading to more compact magnetic components.
This article delves into the research and design of a high-performance on-board power supply for a battery EV car based on an enhanced PSFB converter topology. We will systematically analyze the operational principles, identify the limitations of conventional designs, and propose a refined topology with optimized parameter design and control strategy. The goal is to present a converter that meets the stringent efficiency and reliability demands of automotive applications, thereby contributing to the enhanced viability of the battery EV car.
Fundamental Analysis and Limitations of the Conventional PSFB Converter
The canonical PSFB converter, as applied in the context of a battery EV car, consists of a primary-side full-bridge formed by switches (Q1-Q4), a high-frequency transformer for isolation and voltage scaling, a resonant inductor (Lr), and a secondary-side full-bridge or center-tapped rectifier followed by an LC output filter. The core principle of “phase-shifting” involves operating the switches within each bridge leg with a fixed 50% duty cycle but introducing a controllable phase shift between the switching of the two legs. This phase shift directly regulates the effective voltage applied to the transformer primary, thereby controlling the output voltage delivered to the low-voltage network of the battery EV car.
The soft-switching achievement—Zero-Voltage Switching (ZVS) for the primary switches—is facilitated by the resonant inductor (Lr) and the intrinsic capacitance of the MOSFETs. The operational cycle can be divided into several distinct modes. For analysis, we assume ideal components and focus on the positive half-cycle.
Key Operational Modes and Governing Equations
Mode 1 [t0-t1]: Power Transfer Phase. Switches Q1 and Q4 are conducting. The input voltage Vin is applied across the primary winding (leakage inductance Lr and magnetizing inductance). Energy is transferred to the secondary, with rectifier diode D5 conducting. The primary current ip is approximately equal to the reflected output inductor current.
$$ i_p(t) \approx I_1 = \frac{n \cdot I_{Lf}}{2} $$
where \( n \) is the transformer turns ratio (Np/Ns) and \( I_{Lf} \) is the output filter inductor current.
Mode 2 [t1-t2]: Leading Leg Commutation. At t1, Q1 is turned off. The parasitic capacitors of Q1 (C1) and Q3 (C3) begin to resonate with Lr. C1 charges from zero, and C3 discharges from Vin. The primary current is considered constant during this brief interval.
$$ v_{C1}(t) = \frac{I_1}{2C_{oss}} (t – t_1) $$
$$ v_{C3}(t) = V_{in} – \frac{I_1}{2C_{oss}} (t – t_1) $$
where \( C_{oss} \) is the effective output capacitance of each switch. At t2, \( v_{C3} \) reaches zero, and its body diode D3 begins to conduct, creating a ZVS condition for turning on Q3.
Mode 3 [t2-t3]: Freewheeling Phase. With D3 and Q4 conducting, the primary voltage VAB becomes zero. The secondary rectifiers D5 and D6 both conduct, effectively shorting the secondary winding. The primary current ip remains nearly constant, circulating through the primary loop. No power is transferred to the load.
Mode 4 [t3-t4]: Lagging Leg Commutation. At t3, Q4 is turned off. The capacitors of Q4 (C4) and Q2 (C2) resonate with Lr.
$$ i_p(t) = I_2 \cos(\omega_1 (t-t_3)) $$
$$ v_{C4}(t) = Z_1 I_2 \sin(\omega_1 (t-t_3)) $$
$$ v_{C2}(t) = V_{in} – Z_1 I_2 \sin(\omega_1 (t-t_3)) $$
where \( Z_1 = \sqrt{L_r / (2 C_{lag})} \), \( \omega_1 = 1 / \sqrt{2 L_r C_{lag}} \), and \( C_{lag} \) represents the capacitance of the lagging leg switches. Achieving ZVS here is more challenging as the energy is solely from Lr, not the large output filter inductor.
Mode 5 [t4-t5]: Resonant Reset Phase. After D2 starts conducting, the primary voltage reverses. The primary current decreases linearly, crosses zero, and increases in the negative direction.
$$ i_p(t) = I_p(t_4) – \frac{V_{in}}{L_r} (t – t_4) $$
Mode 6 [t5-t6]: Negative Power Transfer. The primary current is negative, and power is transferred to the load through rectifier diode D6. This half-cycle concludes when Q3 is turned off.
Inherent Challenges in Conventional PSFB for Battery EV Car Applications
Despite its advantages, the standard PSFB converter presents several drawbacks critical for a battery EV car’s power supply:
- Duty Cycle Loss (Dloss): The commutation intervals (Modes 2 & 4) where the primary voltage is zero cause a loss of effective duty cycle on the secondary side. This necessitates a larger transformer turns ratio to achieve the same output voltage, increasing component stress.
- Transformer Flux Imbalance and Saturation Risk: Asymmetric switching characteristics or slight component mismatches can cause a DC offset in the transformer magnetizing current, potentially driving the core into saturation over successive cycles, leading to increased losses and failure.
- Secondary-Side Voltage Oscillation and Ringing: The abrupt commutation of current between rectifier diodes interacts with the transformer leakage inductance and diode junction capacitance, causing high-frequency ringing across the rectifier outputs. This generates significant Electromagnetic Interference (EMI) and increases voltage stress on the diodes, a major concern in the sensitive electronic environment of a battery EV car.
- Limited ZVS Range for Lagging Leg: As noted, ZVS for the lagging leg switches is load-dependent. At light loads, the energy stored in Lr may be insufficient to complete the capacitor discharge, leading to hard switching and increased losses, reducing the overall efficiency of the battery EV car’s power system.
| Challenge | Cause | Impact on Battery EV Car System |
|---|---|---|
| Duty Cycle Loss | Commutation time of primary switches | Requires higher turns ratio, larger transformer, reduced efficiency |
| Transformer Saturation | DC flux imbalance from asymmetries | Increased core loss, potential thermal runaway, reliability risk |
| Secondary-Side Ringing | Leakage inductance & diode capacitance resonance | High EMI, diode voltage stress, potential failure |
| Limited Lagging-Leg ZVS | Insufficient energy in Lr at light load | Increased switching loss, lower light-load efficiency |
Proposed Enhanced PSFB Topology for Battery EV Car Power Supply
To address the aforementioned issues and create a more robust power supply for the battery EV car, we propose an enhanced PSFB topology incorporating three key modifications:
- Adjusted Transformer Turns Ratio: The transformer turns ratio is intentionally designed to be higher than theoretically required for the nominal input/output voltages. This pre-compensates for the expected duty cycle loss, ensuring the desired output voltage is achieved without over-stressing components under normal operation in the battery EV car.
- Series DC-Blocking Capacitor: A capacitor (Cb) is inserted in series with the primary winding of the transformer. This capacitor automatically blocks any DC component from developing across the transformer primary, fundamentally eliminating the risk of flux imbalance and core saturation. Its value is chosen to have a negligible impedance at the switching frequency.
- Diode-Clamp Circuit on Secondary Side: A clamp circuit, consisting of diodes and a capacitor, is connected across the secondary winding. This circuit provides a path for the leakage inductance energy during diode commutation, snubbing the voltage spike and dramatically damping the oscillation. This results in cleaner waveforms, reduced EMI, and lower voltage stress on the rectifier diodes, enhancing the longevity of the battery EV car’s auxiliary power system.
The introduction of these elements refines the converter’s operation. The series capacitor influences the resonant transitions, while the clamp circuit modifies the secondary-side commutation dynamics, leading to a more predictable and efficient performance essential for the demanding environment of a battery EV car.
Parameter Design and Optimization Methodology
A systematic design approach is crucial for the on-board power supply of a battery EV car. Key specifications are defined first: Input Voltage \( V_{in} = 380V \), Output Voltage \( V_o = 24V \), Rated Output Power \( P_o = 500W \), Switching Frequency \( f_s = 100 kHz \).
1. Transformer and Resonant Inductor Design
The transformer turns ratio \( n = N_p/N_s \) is chosen considering the duty cycle loss \( D_{loss} \). First, the maximum effective duty cycle \( D_{max} \) is estimated.
$$ D_{max} \approx \frac{V_o}{V_{in,min} \cdot n} $$
The duty cycle loss can be approximated by:
$$ D_{loss} \approx \frac{2 \cdot L_r \cdot I_{o,max} \cdot f_s}{n \cdot V_{in,min}} $$
where \( I_{o,max} = P_o / V_o \). Therefore, the designed ratio must satisfy:
$$ n > \frac{V_o}{V_{in,min} \cdot (D_{max} + D_{loss})} $$
We select \( n = 16 \).
The resonant inductor \( L_r \) is critical for ZVS. Its value is derived from the energy required to charge/discharge the MOSFET capacitances during the dead time \( t_d \). For the lagging leg (most critical):
$$ \frac{1}{2} L_r I_{p,min}^2 \geq \frac{4}{3} C_{oss} V_{in}^2 + C_{Tr} V_{in}^2 $$
Here, \( I_{p,min} \) is the minimum primary current at the commutation instant (near light load), \( C_{oss} \) is the MOSFET output capacitance, and \( C_{Tr} \) is the transformer’s equivalent parasitic capacitance. Solving for \( L_r \):
$$ L_r \geq \frac{2 (\frac{4}{3} C_{oss} V_{in}^2 + C_{Tr} V_{in}^2)}{I_{p,min}^2} $$
A practical value ensuring ZVS down to 20% load is selected as \( L_r = 12 \mu H \).
2. Output Filter Design
The output filter inductor \( L_f \) is designed based on the desired current ripple \( \Delta I_{Lf} \), typically 20-40% of the rated current.
$$ L_f = \frac{V_o \cdot (1 – D_{min})}{\Delta I_{Lf} \cdot f_s’} $$
where \( f_s’ = 2 f_s \) is the ripple frequency seen by the output filter due to full-wave rectification, and \( D_{min} \) is the minimum duty cycle at maximum input voltage. For a ripple of 30%:
$$ \Delta I_{Lf} = 0.3 \times \frac{500W}{24V} = 6.25A $$
$$ L_f = \frac{24V \cdot (1 – \frac{24}{380/16})}{6.25A \cdot 200kHz} \approx 2.2 \mu H $$
The output capacitor \( C_f \) is sized to handle the output voltage ripple \( \Delta V_o \).
$$ C_f \geq \frac{\Delta I_{Lf}}{8 \cdot f_s’ \cdot \Delta V_o} $$
For a voltage ripple \( \Delta V_o < 0.5\% V_o = 120mV \):
$$ C_f \geq \frac{6.25A}{8 \cdot 200kHz \cdot 0.12V} \approx 32.6 \mu F $$
Considering ESR and holdup time, a larger value like \( C_f = 1000 \mu F \) is used.
3. Clamp Circuit and DC-Block Capacitor Design
The clamp capacitor \( C_c \) must be large enough to maintain a relatively constant voltage during a switching cycle.
$$ C_c \gg \frac{n^2 \cdot L_{lk} \cdot I_{pri}^2}{\Delta V_c^2} $$
where \( L_{lk} \) is the transformer leakage inductance reflected to the primary, \( I_{pri} \) is the primary current at commutation, and \( \Delta V_c \) is the allowable voltage rise on the clamp capacitor. A value in the range of \( 1-10 \mu F \) is typical.
The DC-blocking capacitor \( C_b \) should have an impedance much lower than the magnetizing inductance at switching frequency.
$$ C_b \gg \frac{1}{(2 \pi f_s)^2 L_m} $$
A film capacitor in the range of \( 1-10 \mu F \) is suitable.
| Parameter | Symbol | Designed Value | Notes/Rationale |
|---|---|---|---|
| Input Voltage | \( V_{in} \) | 380 VDC | Typical battery EV car auxiliary bus voltage |
| Output Voltage/Power | \( V_o / P_o \) | 24 V / 500 W | Standard low-voltage bus for a battery EV car |
| Switching Frequency | \( f_s \) | 100 kHz | Balance between size and switching loss |
| Transformer Turns Ratio | \( n \) | 16 | Accounts for duty cycle loss |
| Resonant Inductor | \( L_r \) | 12 μH | Ensures ZVS down to ~20% load |
| Output Filter Inductor | \( L_f \) | 2.2 μH | Limits current ripple to ~30% |
| Output Filter Capacitor | \( C_f \) | 1000 μF | Limits voltage ripple < 0.5% |
| DC-Blocking Capacitor | \( C_b \) | 4.7 μF | Prevents transformer saturation |
| Clamp Capacitor | \( C_c \) | 2.2 μF | Suppresses secondary-side ringing |
4. ZVS Achievement Analysis
The condition for achieving ZVS can be summarized by the following inequality, which must hold for both leading and lagging legs, with the lagging leg being the limiting case:
$$ E_{L_r} \ge E_{C_{oss}} $$
$$ \frac{1}{2} L_r i_p(t_{comm})^2 \ge \frac{4}{3} C_{oss} V_{in}^2 + C_{par} V_{in}^2 $$
Here, \( i_p(t_{comm}) \) is the primary current at the commutation instant, and \( C_{par} \) includes transformer capacitance. For the lagging leg commutation in a battery EV car application, the current is approximately:
$$ i_p(t_{comm}) \approx \frac{I_o}{n} – \frac{\Delta I_{Lf}}{2n} $$
where \( I_o \) is the output current. This shows that ZVS is easier to maintain at higher loads, a crucial consideration for the variable load profile of a battery EV car’s auxiliary systems.
Control Strategy for the Battery EV Car Power Supply
To ensure precise and stable output voltage regulation for the sensitive electronics in a battery EV car, a voltage-mode closed-loop control strategy is employed. The control block diagram is implemented digitally but can be analyzed in the continuous s-domain for design.
The output voltage \( V_o \) is sensed and compared with a reference voltage \( V_{ref} \). The error signal \( e(s) = V_{ref}(s) – V_o(s) \) is processed by a Proportional-Integral (PI) compensator \( G_c(s) \).
$$ G_c(s) = K_p + \frac{K_i}{s} $$
The compensated error signal modulates the phase shift \( \phi \) between the bridge legs. The plant transfer function \( G_{vd}(s) \) relating phase shift to output voltage for a PSFB converter can be approximated as a double-pole system dominated by the output LC filter.
$$ G_{vd}(s) \approx \frac{V_{in} \cdot n}{2\pi} \cdot \frac{1}{s^2 L_f C_f + s \frac{L_f}{R_{load}} + 1} $$
The PI controller parameters \( K_p \) and \( K_i \) are tuned to achieve adequate phase margin (>45°) and bandwidth (typically 1/10 to 1/5 of the switching frequency) to ensure fast dynamic response to load transients, which are common in a battery EV car environment. Additionally, a feedforward term based on the input voltage \( V_{in} \) can be added to improve the line transient response.
Simulation Results and Performance Analysis
The proposed enhanced PSFB converter design for the battery EV car on-board power supply was simulated under the specified conditions (380V input, 24V/500W output, 100kHz). The results validate the theoretical analysis and improvements.
Steady-State Waveforms: The output voltage \( V_o \) stabilizes at precisely 24V with a negligible steady-state error due to the integral action of the controller. The output current \( I_o \) is approximately 20.8A for the 500W load. The output power \( P_o \) is constant, confirming stable energy delivery to the low-voltage systems of the simulated battery EV car.
Key Voltage and Current Characteristics: The primary-side voltage \( V_{AB} \) across the bridge exhibits clean square waves with the characteristic phase-shifted flat regions, confirming proper switching. The primary current \( i_p \) shows the expected trapezoidal shape with resonant transitions during commutation, indicating ZVS operation. Crucially, the secondary-side rectifier voltage \( V_{rect} \) shows a dramatic reduction in high-frequency ringing compared to a conventional topology, thanks to the diode-clamp circuit. This is a significant achievement for the EMI profile of the battery EV car’s power supply.
| Metric | Simulation Value | Specification/Note |
|---|---|---|
| Output Voltage (Vo) | 24.01 V | Steady-state, meets 24V target |
| Output Current (Io) | 20.83 A | Corresponds to 500W load |
| Output Voltage Ripple | < 100 mV pk-pk | Well within 0.5% (120mV) limit |
| Peak Efficiency (at full load) | ~94.2% | Estimated from simulated losses |
| ZVS Achievement | Full ZVS for all primary switches at load >15% | Validates Lr design |
| Secondary Ringing Amplitude | Reduced by >70% | Compared to non-clamped design |
Hardware Implementation and Experimental Verification
A laboratory prototype was built to validate the design under real conditions, emulating the electrical environment of a battery EV car. The prototype used 600V-rated MOSFETs for the primary bridge, a custom wound transformer with the designed turns ratio, and the calculated passive components. A digital signal controller was programmed to implement the phase-shift modulation and the voltage PI control loop.
Experimental Waveforms: The measured gate drive signals for the leading and lagging leg switches confirmed the proper phase-shift operation. The drain-to-source voltage of a lagging-leg switch was observed to fall to zero before the gate signal turns on, providing direct evidence of ZVS, which is vital for high-efficiency operation in the battery EV car power supply.
Input and Output Performance: The primary voltage and current waveforms matched the simulations closely, showing the resonant transitions. On the secondary side, the output voltage and current were clean and stable at 24V under various load conditions. The output voltage ripple was measured to be within the specified limits.
Efficiency Measurement: The converter’s efficiency was measured across the load range typical for a battery EV car’s auxiliary systems (from 50W to 500W). The efficiency peaked at approximately 93.5% near full load, slightly lower than the simulation due to practical losses (PCB trace resistance, component ESR, core losses) not fully captured in the model. This efficiency level is highly satisfactory for an on-board power supply in a battery EV car, contributing to extended range by minimizing auxiliary system losses.
| Load Condition (% of Full Load) | Output Voltage (V) | Measured Efficiency (%) | Notes |
|---|---|---|---|
| 10% (50W) | 23.98 | 88.1 | ZVS maintained, good light-load performance |
| 25% (125W) | 24.02 | 91.3 | — |
| 50% (250W) | 24.00 | 92.8 | — |
| 75% (375W) | 24.01 | 93.2 | — |
| 100% (500W) | 23.99 | 93.5 | Peak efficiency point |
Conclusion
This comprehensive study has presented the research, design, and validation of an advanced on-board power supply for a battery EV car, based on an enhanced Phase-Shifted Full-Bridge DC-DC converter topology. By methodically addressing the inherent limitations of the conventional PSFB circuit—namely duty cycle loss, transformer saturation risk, and secondary-side voltage oscillations—through an adjusted transformer design, the introduction of a series DC-blocking capacitor, and a diode-clamp snubber circuit, a robust and high-performance solution has been developed.
The detailed parameter design methodology, grounded in the operational principles of the converter, ensures Zero-Voltage Switching across a wide load range and stable output regulation. Simulation and experimental results from a hardware prototype confirm that the proposed design successfully meets the critical requirements for an automotive-grade power supply: high efficiency (>93% peak), stable output under varying loads, and excellent noise characteristics. The implementation of such a power supply contributes significantly to the overall energy efficiency and reliability of the battery EV car, supporting the broader adoption of electric vehicles by making their auxiliary power systems more efficient and dependable.