Variable Dead-Time Based Novel Soft-Start Method for Dual Active Bridge Converters

This paper proposes a novel soft-start method for Dual Active Bridge converters that gradually reduces variable dead-time to achieve a smooth voltage buildup, effectively minimizing inrush current and voltage overshoot while ensuring safe and reliable startup across various operating conditions.

The Gist

Imagine you have a very powerful, high-speed train (the Dual Active Bridge Converter) that needs to move energy from a battery to a device, like charging an electric car.

The problem is: How do you get this train moving without derailing it?

The Problem: The "Hard Start" Crash

In traditional systems, when you turn the power on, it's like slamming the train's throttle from "Off" to "100%" instantly.

• The Result: The train lurches forward violently. The wheels (the electrical components) skid, sparks fly (voltage spikes), and the tracks shake (inrush current).
• The Danger: This sudden jolt can break the engine (burn out transistors), shatter the windows (blow capacitors), or trigger the emergency brakes (safety systems shutting down the whole system).

Engineers have tried to fix this by using complex "soft-start" methods, like slowly pressing the gas pedal or using special gearboxes. But many of these methods are complicated, expensive, or still cause a little bit of a jerk at the beginning.

The Solution: The "Variable Dead-Time" Trick

This paper proposes a clever, simple new way to start the train: The Variable Dead-Time Method.

Think of "Dead Time" as a safety pause between two people passing a heavy box back and forth.

• Normal Operation: They pass the box quickly, with just a tiny split-second pause to make sure they don't bump hands.
• The Startup Problem: If they try to start passing the box immediately, they might collide, dropping it or hurting themselves.

The New Strategy:
Instead of starting with a tiny pause, the system starts with a huge pause (almost the entire time the box is supposed to be moving).

1. The "Long Pause" Start: Imagine the two people stand still for a long time, holding the box. Nothing moves. No energy flows. The system is safe.
2. The "Slow Release": Instead of suddenly starting, the system gradually shortens the pause.
   • First second: They pause for 90% of the time. Only a tiny bit of energy gets through.
   • Next second: They pause for 80%. A little more energy flows.
   • Next second: 70%... 60%... and so on.
3. The Smooth Cruise: By the time the pause is reduced to the normal, tiny safety gap, the train is already moving smoothly at full speed.

Why is this better?

The authors compared their method to the old ways using a 15,000-watt power system (roughly the power of a fast EV charger).

• Old Methods: Like trying to pour a bucket of water into a glass by tipping the bucket over instantly. You get a splash (voltage overshoot) and water spills everywhere (inrush current).
• Their Method: Like using a hose with a nozzle that you slowly open. The water (energy) flows in gently, filling the glass perfectly without spilling a drop.

The Real-World Test

They didn't just simulate this on a computer; they built a real machine in a lab.

• They tested it at low power (200V), medium power (400V), and high power (650V).
• The Result: Every time, the system started smoothly. No sparks, no loud bangs, no damage. The voltage rose like a gentle hill rather than a cliff.

The Bottom Line

This new method is like a smart, automatic cruise control for starting up.

• Simple: It doesn't need extra wires or expensive parts. It's just a software trick that tells the computer to "pause a little longer, then a little less, then a little less."
• Safe: It protects the expensive electronics from the shock of sudden power.
• Versatile: It works for small chargers and massive industrial power systems alike.

In short, this paper teaches us that sometimes, the best way to go fast is to start slow and steady, using a simple pause to ensure a smooth ride.

Technical Summary

1. Problem Statement

The Dual Active Bridge (DAB) converter is a critical topology for high-power applications (e.g., EV charging, DC microgrids) due to its bidirectional power flow and galvanic isolation. However, conventional startup methods face significant challenges:

• Uncontrolled Inrush Current: Abrupt application of a fixed 50% duty cycle or standard phase-shift control at startup causes a step-change in transformer flux. This leads to magnetic core saturation and massive current spikes.
• Voltage Overshoot: The rapid charging of the output capacitor results in severe voltage overshoot, potentially damaging power semiconductors and passive components.
• Limitations of Existing Solutions:
  • Phase-shift modulation (SPS, DPS, EPS, TPS): Often complex to implement during startup and may still result in current spikes.
  • Single-cycle control: Insufficient to manage the long-term charging dynamics of the output capacitor.
  • Fixed dead-time methods: While they mitigate initial spikes, they often fail to prevent overshoot during the transition to steady-state or require complex auxiliary pre-charge circuits.
  • Hardware dependency: Many existing methods require additional hardware (pre-charge resistors/relays), increasing system cost and reducing power density.

2. Proposed Methodology

The authors propose a Variable Dead-Time Based Soft-Start Strategy that utilizes the dead-time parameter as the primary control variable to regulate the startup trajectory, rather than relying solely on phase-shift adjustments.

Core Mechanism:

• Initialization: The startup sequence begins with a dead-time ($t_{d\_start}$) set to a value approaching the full switching period ($T_{sw}$). This effectively reduces the applied duty cycle to near zero, keeping the secondary-side voltage at zero and minimizing energy transfer.
• Linear Ramp: Over a predefined time window (e.g., 150 ms), the dead-time is linearly decreased from the initial large value to the nominal hardware-defined minimum ($t_{d\_final}$).
• Control Logic:
  • The microcontroller adjusts the timing between complementary PWM signals.
  • The programmed dead-time value is set to half of the desired total dead-time to ensure equal delay on rising and falling edges.
  • As the dead-time decreases, the effective duty ratio increases, allowing energy to transfer incrementally to the output capacitor.
• Theoretical Basis: The method controls the rate of change of the output voltage ($dV_{dc}/dt$) by modulating the effective power transfer. By limiting the initial power transfer, the method prevents transformer saturation and ensures a monotonic, controlled rise in the DC-link voltage.

3. Key Contributions

• Novel Control Variable: Shifts the focus from phase-shift modulation to dead-time modulation for startup, offering a simpler and more direct way to limit initial energy transfer.
• Hardware-Friendly Implementation: The method requires no additional hardware components (like pre-charge circuits). It is easily implementable on standard microcontrollers (e.g., TI C2000 series) by simply adjusting PWM timing registers.
• Scalability: The approach is applicable to $n$-th order DAB architectures and is robust across a wide range of input voltages and load conditions.
• Elimination of Overshoot: Unlike other methods that may only mitigate initial spikes, this method ensures a smooth, monotonic voltage buildup, completely eliminating voltage overshoot and high-frequency ringing.

4. Results and Validation

The proposed method was validated through PLECS simulations and experimental testing on a 15 kW SiC-based DAB prototype (650 V DC-link, 32 kHz switching frequency).

Simulation Results:

• Hard-Start (Conventional): Showed immediate voltage spikes, severe inrush current, and transformer saturation.
• Proposed Method: Demonstrated a gradual, linear rise in secondary-side voltage. The leakage inductor current rose in a controlled manner without spikes, and the output capacitor current was significantly smoothed.
• Comparison: Compared against existing methods (DPS-based, single-cycle, fixed large dead-time), the proposed method showed:
  • Zero voltage overshoot.
  • Well-controlled inductor peak current (no abrupt spikes).
  • Low complexity and high compatibility with high-power (15 kW) models.

Experimental Results:

• Tests were conducted at 200 V, 400 V, and 650 V DC-link voltages.
• The experimental waveforms confirmed the simulation predictions:
  • Monotonic Voltage Rise: The DC-link voltage increased smoothly without transients.
  • Current Suppression: Leakage inductor current remained within safe limits throughout the startup.
  • Robustness: The method performed consistently across different voltage levels and startup ramp rates.
• Variable Ramp Rate: The study showed that adjusting the rate of dead-time reduction allows tuning between a rapid startup (higher rate) and a smoother transition (lower rate), offering flexibility for different application needs.

5. Significance and Impact

• Enhanced Reliability: By eliminating inrush currents and voltage overshoot, the method significantly reduces stress on power semiconductors (SiC MOSFETs) and magnetic components, extending system lifespan.
• Cost and Density Reduction: Removing the need for auxiliary pre-charge circuits increases power density and reduces system cost.
• Simplicity: The algorithm is computationally efficient and easy to integrate into existing DAB control loops without complex modulation schemes.
• Versatility: The solution is suitable for diverse high-power applications, including integrated EV chargers, high-voltage traction inverters, and modular charging infrastructures where transient stability is critical.

In conclusion, this paper presents a robust, software-defined solution to a critical hardware challenge in DAB converters, offering a superior alternative to traditional hard-start and complex modulation-based soft-start techniques.