Analysis of dual active bridge-based on-board battery charger for electric and hybrid vehicles

This paper presents an in-depth analysis of a Gallium Nitride (GaN) Dual Active Bridge (DAB)-based on-board battery charger for electric and hybrid vehicles. It focuses on the superior efficiency and reliability of GaN-based power electronics over traditional Silicon-based systems. The study explores the implementation of the DAB topology, recognized for its bidirectional power flow and high efficiency, in enhancing vehicle charging infrastructure. Efficiency analysis, switching behavior and losses are thoroughly examined. Simulation results in PSIM software validate the theoretical predictions, showing significant improvements in operational flexibility and energy efficiency


Introduction
In the rapidly evolving landscape of electric and hybrid vehicles (EVs and HVs), the efficiency and reliability of charging systems have become critical factors in advancing automotive technology and promoting wider adoption of these ecofriendly transportation modes.Among the myriad of technologies being explored, Gallium Nitride (GaN) based power electronics have emerged as a ideal innovation, offering superior performance characteristics compared to traditional Silicon-based systems [1][2][3][4][5].The on-board battery charger is a cornerstone component of EVs and HVs, dictating the vehicles' charging speed, energy efficiency, and operational flexibility.The Dual Active Bridge topology (DAB), renowned for its bidirectional power flow capability and high efficiency, provides an ideal framework for exploiting GaN's high switching frequency, low on-resistance, and exceptional thermal performance.This analysis aims to unravel the technical intricacies and potential benefits of integrating GaN technology into DAB chargers, focusing on aspects such as efficiency improvements, power density, thermal management, and the impact on the vehicle's overall charging infrastructure.
In addition, the conversion of electrical energy plays a key role in many aspects of modern society, from consumer electronics to high-power applications [6].In the latter case, efficiency, harmonic quality, power density demands, among other requirements, have triggered the development of advanced electronic adjustable-speed drives (ASDs) and sophisticated power semiconductors.More precisely, the use of medium-voltage (MV)-ASDs has been a key technology for high-power applications such as fans, pumps, conveyor belts, propellers, crushers, photovoltaic systems, wind turbines, static compensators, and so on.Further improvements in power semiconductor and converter technologies have allowed the replacement of load-commutated inverters (LCIs), which are still used for output powers in the tensof-megawatt range, with self-commutated converters.The increased demand for intermediate storage of electrical energy in battery systems, particularly due to the use of renewable energy, has resulted in the need for bidirectional DC/DC power converters with galvanic isolation [7][8].Uninterruptible Power Supplies (UPS), battery charging systems [9], photovoltaic systems [10][11][12][13][14], and auxiliary power supplies in traction applications are examples of some fields of application of this kind of converter.A Dual Active Bridge (DAB) bidirectional DC/DC converter is a topology with the advantages of fewer devices, soft-switching commutations [15][16], low cost, and high efficiency.The use of this topology is proposed for applications where power density, cost, weight, and reliability are critical factors.In the present paper, the steady-state analysis of the DAB converter has been carried out, providing some guidelines for design (considering soft switching and the amount of RMS current) and modulation techniques for DAB topology are discussed with providing simulations results in PSIM.

Dual-Active-Bridge Topology, and its Modulations
The two-level dual-active bridge is a bidirectional and controllable DC-DC converter with substantial power capabilities [17].Figure 1 shows the structure of DAB with eight semiconductor devices, a high-frequency transformer, an energy transfer inductor, and DC-link capacitors where the converter resembles a more common full bridge with a controllable rectifier.Its symmetrical design, featuring identical primary and secondary bridges, enables bidirectional power flow control.The interconnected full bridges, linked by a high-/medium-frequency transformer for galvanic isolation, facilitate power transfer through the transformer's leakage inductance.Controlled switching actions generate square voltages namely VT1 and VT2 on both sides, enhancing efficiency compared to hard-switched topologies.This design achieves higher switching frequencies without excessive losses, particularly with soft switching.

Figure 1
The structure of dual active bridge At elevated frequencies, the isolation transformer's magnetizing inductance diminishes, simplifying the model to its leakage inductance and reducing weight for easier transport.Figure 2 illustrates an equivalent system used to derive the power equation.Additionally, the dual-active bridge's bidirectional nature makes it suitable for energy storage applications, such as linking batteries in automotive use [18][19].

Figure 2 The equivalent circuit of dual active bridge in high frequencies
There are various modulation techniques for DAB converters including single-phase-shift (SPS) control, dual-phaseshift control, extended-phase-shift control, triple-phase-shift control, as well as trapezoidal, triangular, and optimized modulation methods.The single-phase-shift modulation and the trapezoidal modulation scheme are chosen for modeling, simulation, and testing in PSIM® software.In a DAB converter, three parameters affect power flow between primary and secondary sides: phase-shift between square voltages, duty cycle of square voltages, and switching frequency [20][21].The modulation techniques considered here involve changes in phase-shift and/or duty cycle to control power flow, while frequency switching methods are excluded from consideration in this study.The single-phaseshift modulation only uses a phase shift between the two transformer voltages to control the power flow, while the trapezoidal modulation uses a phase shift and additionally changes the duty ratio of the transformer voltages, introducing a zero-voltage period.The zero-voltage period is attained by introducing a phase shift between the two legs of each full bridge [22].

Single-Phase-Shift Modulation
The SPS control is the standard modulation scheme for the DAB and describes a classical method to implement a voltagecontrolled DC-to-DC converter.The method is easy to implement and shows excellent control performance, but the overall efficiency is not sufficient.The square voltages in a circuit that is modulated with this scheme will always have duty cycles of 50% of the switching period while the frequency stays constant [23].Two square voltages  1 and  2 are generated on the primary and secondary side of the transformer by giving respective switching signals to the switches Q1 to Q8.A phase-shift  is introduced between the switching signals for the primary side and the switching signals for the secondary side, leading to the same phase-shift  between the two voltages  1 and  2 .A voltage difference is induced and a current flow from the primary to the secondary side.This is shown in Figure 3.There are four commutation states of the power switches to achieve the SPS modulation method which are shown in Figure 4 and Table1.

Trapezoidal Modulation
Equal to the single-phase-shift modulation [24], the two transformer voltages will be phase-shifted in the trapezoidal modulation scheme.In addition to that, two inner phase shifts are introduced between the two legs of each full bridge.This causes the duty cycle of and to change and introduces a period during which and will be zero [25].These intervals are named  1 and  2 can be seen in Figure 5.It is notable that the on and off times of the switches continue to equal 50% of one switching period, and it is only the duty cycles of the transformer voltages that change.There are eight commutation states of the power switches to achieve the trapezoidal modulation method which are shown in Figure 6 and 7. Commutation sequence of switches for the Trapezoidal modulation is tabulated in Table 2.

Triangular Modulation
Figure 8 Primary and referred secondary transformer voltage and inductor current for the triangular modulation The triangular modulation is a special case of a trapezoidal modulation.The current ramped to achieve zero-current switching on one full bridge.This modulation is only possible, if the two input voltages   and   are different [26].But if one of the two voltages is equal to zero, this method cannot be used.The variables for controlling the power flow are the phase-shift angle between primary and secondary transformer voltage as well as a change in duty ratio of these voltages [27].In distinction from the trapezoidal modulation scheme, in the Triangular modulation features two time periods during which both square voltages  1 and  2 are zero.This results in two-time intervals with zero inductor current   as can be seen in Figure 8.There are eight commutation states of the power switches to achieve the trapezoidal modulation method which are presented in Figures 9 and 10 and the commutation sequence of switches are tabulated in Table 3.

Simulation results and discussion
The model of the DAB can be built in PSIM®.Parameters that are defined in the Simulation Parameter Initialization dialogue in PSIM® are given in Table 4.The definition of certain parameters will be explained in the following sections.
A schematic of the PSIM® model is presented in Figure 11 and 12.

Single-Phase-Shift Modulation
In Figures 33 and 31, the DAB is subjected to an output voltage ranging between 250 V and 400 V.All waveforms exhibit the theoretically expected behavior illustrated in Figure 3.It is evident that the current slope increases as the output voltage decreases, and the current value decreases accordingly due to the reduction in maximum output power.

Trapezoidal Modulation
The waveforms from the simulation of the trapezoidal modulation exhibit the expected behaviour outlined in Figures 15 and 16.Like SPS Modulation, the current value decreases as the load decreases.According to the equations in the Trapezoidal Modulation section, the zero-voltage widths ω1 and ω2 decrease with a decreasing phase-shift angle, while the duty cycles increase.Furthermore, the current slope alternates in the third time interval when (Vin−nVo) is applied, depending on the relationship between Vin and nVo, as described in the Trapezoidal Modulation section.

Switching Losses
Generally, it is evident that the percentage shares of switching losses relative to total losses increase with decreasing load.This is because the current, and consequently, the absolute value of conduction losses decrease.

Conclusion
The research confirms that integrating GaN technology into DAB-based on-board battery chargers significantly enhances the performance of charging systems in electric and hybrid vehicles.The use of GaN results in higher efficiency

Figure 3 Figure 4
Figure 3 Primary and referred secondary transformer voltage and inductor current for the single-phase-shift modulation

Figure 5 Figure 6 Figure 7
Figure 5 Primary and referred secondary transformer voltage and inductor current for the trapezoidal modulation

Figure 9 Figure 10
Figure 9 Switching signals for the gates Q1 to Q4

Figure 11 Figure 12
Figure 11 Schematic of the DAB model in PSIM® for the SPS Modulation

Figure 31 Figure 31
Figure 31 Primary and referred secondary transformer voltage   and   and respective inductor current   and the output power at   =   with SPS Modulation

Figure 15 Figure 16 4 . 1 .
Figure 15 Primary and referred secondary transformer voltage   and   and respective inductor current   and the output power at   =   with Trapezoidal Modulation

Figure 17
Figure 17 RMS inductor current at different output voltage in both SPS Modulation and Trapezoidal Modulation4.2.Total Losses and EfficiencyFigure18displays the total losses, while Figure19illustrates the corresponding efficiencies of the constructed model.These losses comprise switching losses and conduction losses of the semiconductor switches.It's evident that total losses decrease with a decrease in output voltage and consequently with a decrease in load.As anticipated from the RMS current trends, the total losses in Trapezoidal Modulation are slightly larger than those in Single-Phase Shift Modulation.Consequently, it's observable in Figure19that SPS Modulation yields better efficiency than Trapezoidal Modulation.For lower output voltage values, the figures indicate that despite having fewer total losses, efficiency is reduced.This outcome can be elucidated by examining the distribution of switching and conduction losses at various load levels, as presented in the subsequent section.

Figure 18
Figure 18 Total losses at different output voltages and MOSFET switchesOverall, the efficiency of the constructed DAB controlled via SPS Modulation ranges from 95.48% to 96.97%, depending on the load and voltage conditions.In contrast, if the control is executed through the Trapezoidal Modulation scheme, the efficiency ranges from 96.91% to 98.17%.

Figure 19 Figure 20 Figure 21
Figure19 Efficiency at different output voltages and MOSFET switches4.3.Switching BehaviourBased on the simulations depicted in Figures20 and 21, which illustrate the switching behaviors in SPS Modulation and Trapezoidal Modulation when Vo=400 V, we can observe the switching behavior consistent with our previous discussions in both modulation schemes.
Additionally, SPS Modulation and Trapezoidal Modulation exhibit different characteristics regarding soft switching, as discussed in previous sections.Based on these observations, Trapezoidal modulation is expected to demonstrate a lower proportion of switching losses compared to SPS modulation.This assumption is supported by Figure 22, where at Vo=250 V, the switching losses with Trapezoidal Modulation account for only around half of the switching losses in SPS Modulation.This behavior is reflected in the overall efficiency at all loads.Moreover, as shown in Figure 17, the RMS inductor current in Trapezoidal Modulation is consistently smaller than in SPS Modulation.This results in smaller absolute values of conduction losses in Trapezoidal Modulation compared to SPS Modulation.

Figure 22
Figure 22 Total Switching Losses in both SPS Modulation and Trapezoidal Modulation in full load when the output voltage varies

Table 1
Commutation sequence of switches for the SPS modulation

Table 2
Commutation sequence of switches for the Trapezoidal modulation

Table 3
Commutation sequence of switches for the Triangular modulation

Table 4
Simulation Parameter Initialization