Battery Energy Storage System With Interleaving Structure of Dual-Active-Bridge Converter and Non-Isolated DC-to-DC Converter With Wide Input and Output Voltage

We propose a circuit topology suitable as a battery charge/discharge tester with a DAB converter and a non-isolated dc-dc converter as a module structure. The module structure can be configured to have a wide input and output voltage range because it is easy to expand. DAB converters are used for bidirectional power transfer and galvanic isolation. By controlling the phase difference of the voltages across the DAB to be almost identical, the generation of circulating current due to the phase difference is minimized, and the ZVS is guaranteed. In addition, by switching at a phase angle that flattens the DAB switch current, the peak current is reduced to diminish the switch’s conduction loss, lowering the switch current rating, and reducing the magnetic core size. The non-isolated dc-dc converter is used to control the battery output voltage. Through the interleaved structure, good dynamic characteristics and current ripple can be reduced, so the effect of improving the battery lifecycle can be expected. We present four series-parallel structures of the proposed converter and compare them with the case of generating an output voltage only with a DAB converter. Analyze the characteristics of automatic voltage balancing due to the difference between the input capacitors’ initial voltage and capacitance connected in series. We analyze the features of the proposed system through simulation and verify the technical feasibility and excellence of the proposed battery charge/discharge tester through charge/discharge experiments using supercapacitors instead of batteries.


I. INTRODUCTION
High-voltage DC transmission systems are widely used in rail transport systems, electric vehicle battery charging systems, and DC grids [1]. In these applications, DC supply voltages can reach up to several kilovolts, which is often difficult to construct using a single power semiconductor. To solve The associate editor coordinating the review of this manuscript and approving it for publication was Zhehan Yi . this problem, a modular approach is applied to improve output performance, such as increasing voltage and current levels handled by power converters, improving efficiency, reducing costs, and reducing output ripple [3], [4], [5], [6], [7], [8], [9]. Among them, input-series and output-parallel (ISOP) connected modular dc-to-dc converters enable the optimization of high input voltage and high output current converters [4], [6], [7], [8], [10], [11]. In general, DAB (dual active bridge) is more suitable for high-power applications VOLUME 10, 2022 This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/ with high efficiency, bidirectional operation, and fault isolation compared to other isolated dc-to-dc converters [12], [13], [14] and is used for applications such as high and medium power [14], [15], [16], [17], [18], [19], [20], [21], [22], [23], [24], [25], [26], [27]. The control of the modular structure must ensure proper voltage and current distribution between modules to achieve safe and suitable operation [4], [6], [9], [10]. For ISOP connections, each module's input voltage and output current must be kept within acceptable ranges, ideally the same value for all modules. The output voltage must also be adjusted [11]. The most straightforward power balance control strategy is to use an output voltage regulator (OVR) to regulate the transferring power of all modules [29], [30]. However, due to the inevitable mismatch of parameters between modules, transmit power balancing cannot be achieved with OVR alone. Many control methods of ISOP dc-to-dc converters, including distributed control strategies and master-slave control strategies, can accomplish the transmit power balancing of DAB ISOP dc-to-dc converters. Distributed control strategies such as input voltage droop control and output current reverse droop control based on positive output voltage slope have the advantage of no control interconnection between modules and a simple structure [31], [32], [33], [34]. Still, on the other hand, there are many limitations and poor dynamic and steady-state properties [35], [36]. In addition, the transfer current of the DAB dc-to-dc converter includes alternating current, so transfer current-based control is not suitable. Above all, droop-based control schemes generally have poor dynamic and steady-state performance [37]. Furthermore, since the PI controller that regulates the input voltages of both modules is used to determine the phase shift ratio, not the delivered power, it also affects the output voltage. These existing studies focus more on powersharing performance. They have not conducted studies on the disturbance of input voltage and load current by the fast dynamic response of the ISOP DAB dc-to-dc converter [12]. This paper, rather than solving the ISOP problem through the control method, solves the problem through the combined topology using DAB and the non-isolated converter. Through simple control of DAB and output power control of nonisolated converter, we propose a topology that is effective for fast dynamic characteristics and load fluctuations and is used for battery charge/discharge tester. First, the operation of the DAB converter and the interleaved non-isolated dc-to-dc converter constituting the proposed converter are analyzed. The proposed converter has a modular structure and can be applied to a wide voltage range through serial/parallel connection of input/output terminals. By minimizing the phase difference between the voltages on both sides of the DAB converter, the circulating current due to the phase difference is minimized. The current peak is reduced by setting the switching angle to flatten the DAB switch current. Therefore, the loss due to the generation of reactive power is slight, and the current peak is low, which has the advantage of lowering the switch rating and reducing the core size of the magnetic component. In addition, controlling the output voltage through an interleaved non-isolated dc-to-dc converter reduces current ripple and has fast response characteristics. To prove the technical feasibility and superiority of the proposed converter structure, automatic voltage balancing by the initial voltage difference and capacitance difference for the input capacitors connected in series is analyzed. The design procedure to obtain the critical design parameters of the proposed converter is presented, and its validity is verified through simulation and experimentation. We measure the efficiency when operating the four series-parallel structures that the proposed converter can implement at 50V, 100V, and 150V and compare it with the case of generating an output with only a DAB converter.

II. REVIEW OF HIGH-FREQUENCY ISOLATED BIDIRECTIONAL DC-DC CONVERTERS SUITABLE FOR BATTERY CHARGE/DISCHARGE TESTER
A converter capable of bidirectional power transfer and galvanic isolation between the grid and the battery is required for ESS (Energy Storage System) and battery charge/discharge tester applications. Generally, a transformer is used to insulate the grid and the battery. The use of a low-frequency transformer has the disadvantage of increasing the volume of the transformer. Recently, an isolated dc-to-dc converter using a highfrequency transformer has been mainly used. This chapter analyzes and summarizes the characteristics of a highfrequency isolated bidirectional dc-to-dc converter suitable for a battery charge/discharge tester [1]. Fig. 1 shows a representative bidirectional isolated dc-to-dc converter suitable for battery charge/discharge testers; Phase shifted full-bridge (PSFB) converter, resonant dual active bridge (DAB) converter, and DAB converter. These converters have an active switch-based bridge circuit and a high-frequency transformer in common at both [38], [39].

A. PHASE SHIFTED FULL-BRIDGE (PSFB) CONVERTER
The PSFB converter given in Fig. 1(a) has the same circuit structure as the most basic isolated full-bridge dc-to-dc converter. A phase shift method is selected to improve the characteristics of ZVS. The transformer's primary is connected through a capacitor to show the features of the voltage source, and the secondary shows the characteristics of the current source due to the dc inductor for the filter. A high-frequency transformer provides galvanic isolation and adjusts the voltage gain. V 2 /V 1 = ND if the transformer turns ratio is N when the duty ratio D determines the voltage gain. Therefore, the PSFB converter voltage gain is limited to a maximum of N . A dc inductor is additionally connected to reduce the output ripple, and a relatively larger inductance is required compared to other converters, so the price increase and power density decrease. To ensure the ZVS, the transformer must have a leakage inductance above a certain level. However, the usable duty ratio range is limited due to the influence of the inductance, thus limiting the voltage gain. In addition, since switching occurs between the leakage inductance of the transformer and the dc inductor of the output stage, there is a problem with generating a voltage spike in the secondary switch due to the current difference between the two inductors [40], [41], [42], [43]. Fig. 1(b) is the resonant bridge converter's most widely used LLC-type converter. The series resonant converter has a limited voltage gain. The parallel resonant converter has a disadvantage in that the conduction loss due to the circulating current is significant at a light load. The resonant dc-to-dc DAB converter was introduced to solve the shortcomings of the existing series resonant and parallel resonant dc-to-dc converters. Since the voltage gain-frequency characteristic is steep near the resonance point, there is an advantage that it can be controlled only with a relatively small frequency change in a wide voltage range. However, the asymmetric circuit structure differs between forward and reverse operation characteristics. In particular, in the case of reverse control, the voltage gain width is limited because it operates in the same way as the existing series resonant converter. Because it is controlled at a variable frequency, it is not easy to optimize the converter design. Additional components are required for resonance, and the inductance needed for resonance is designed to be larger than the inductance used in the DAB converter. It is also difficult to optimize the power density because a capacitor for resonance is also required [44], [45], [46].

C. DUAL ACTIVE BRIDGE CONVERTER
The advantage of DAB converters given in Fig. 1(c) is that they can be electrically isolated and transfer power in both directions with only a small number of devices. In a DAB converter, power transfer is only through an inductor connected to the transformer. The series inductance may use an additional inductor or a transformer's leakage inductance. In addition, since the inductor of the DAB converter operates as an AC inductor, the same power delivery is possible with a much smaller inductance than the PSFB converter. Therefore, when both ends of the transformer are configured as a bridge circuit, there is no need for additional passive components, which is very useful for increasing power density. The inductor current waveform is determined by the voltage applied across the DAB converter. Therefore, since the inductor current has a lagging phase compared to the primary voltage and a leading phase compared to the secondary voltage, ZVS operation of all switches is possible in a wide range. The most basic control method of the DAB converter is to control only the phase difference between the primary and the secondary voltage. However, as the voltage ratio changes and drops to a light load, the ZVS characteristic deteriorates, and the circulating current increases, so the conversion efficiency decreases rapidly. Also, since the input and output of the DAB converter are both configured as voltage sources, the input and output capacitors have to filter the AC current, so the capacitor has to filter the high RMS current [47], [48], [49], [50].

III. PROPOSED CONVERTER STRUCTURE SUITABLE FOR BATTERY CHARGE/DISCHARGE TESTER
It can be seen that the overall system performance is improved by minimizing the current stress of the converter switch [51], [52], [53], [54], [55], [56], [57], [58], [59], [60], [61]. Reducing the current stress decreases the switch current rating and the volume and loss of magnetic components. In addition, the conduction loss and switching loss of the converter can be reduced, which helps to improve the system's efficiency. Fig. 2 shows the proposed circuit configuration. DAB converter and non-isolated dc-to-dc converter are combined. Two DAB converters are coupled in parallel, and four non-isolated converters are connected in parallel. The input stage is connected in series to obtain a wide range of input voltage. The basic module of the proposed converter minimizes reactive power generation due to circulating current by maintaining almost no voltage phase difference between both sides of the DAB. It can overcome the limitations of the existing phaseshifted modulation (PSM) by using an interleaved structure that combines a non-isolated dc-to-dc converter for output control. In the case of the existing DAB converter, current VOLUME 10, 2022 stress reduction and ZVS were achieved through a complex control at light load or when the voltage ratio was not 1:1. On the other hand, in this paper, a battery charge/discharge tester with reduced current stress and fast response is realized by utilizing the interleaved structure with the parallel combination of two DAB units and four non-isolated converters.

A. STEADY-STATE ANALYSIS OF DAB CONVERTER CONTROLLED BY PHASE-SHIFT MODULATION
This chapter reviews the case where the DAB converter is controlled by the single-phase shift (SPS) modulation method under steady-state conditions. The most basic driving method of the DAB converter is the single-phase shift (SPS) modulation, which applies V P and V S as square waves and controls using the phase difference between the two voltages. Fig. 3 shows the primary voltage, secondary voltage, inductor current, primary current, and secondary current when the phase difference is 45 degrees during two cycles. The PSM method performs two operational modes during one cycle according to the phase state of the two voltages [47].
Since v pH and v sH are in a steady state, the average value of the current flowing through the inductor is 0. Therefore, we can obtain the following boundary conditions.
Substituting (1) and (2) into (3) to calculate the inductor current at t 2 is as follows.
The input current (i inH ) of the DAB converter is determined by the switching state of the primary and secondary and the condition of the inductor current. The average value of the input current is the same as (5).
Therefore, assuming that the converter is ideal, calculating the average power of the converter using the average input 127208 VOLUME 10, 2022 current becomes (6). Fig. 4 shows the power by (6) when the phase difference φ varies from −π to π. If the magnitude of the power is normalized to the standard power P 0 , it is equal to (7).
When the phase difference is greater than 0, since the primary voltage leads to the secondary voltage, power is transferred from the input to the output. On the other hand, when the phase is less than zero, the primary voltage lags behind the secondary voltage, and the power transfer direction is reversed. In Fig. 4, the phase value that can transmit an arbitrary amount of power P 1 is obtained from two operating points, φ 1 and φ 2 . However, if a certain φ is minor than −π/2 or larger than π/2, the phase difference between the full bridge voltage and the current is significant, causing a circulating current. The maximum power transfer condition of the DAB converter is when the input/output voltage and the inductor are constant when φ is ±π/2. As for the magnitude of the maximum power, when the magnitude of the DC voltage and the switching frequency of both ends are fixed in (6), the smaller the value of the series reactance of the transformer element, the greater the current flowing.

B. BIDIRECTIONAL NON-ISOLATED CONVERTER
This chapter describes the non-isolated dc-to-dc converter for battery charging and discharging on the output side. Fig. 5 shows only the non-isolated dc-to-dc converter among the overall system diagram of Fig. 2. It adopts an interleaved method that consists of two modules (4 in parallel) of a Buck-Boost converter. This modular method is a structure in which power capacity can be expanded through additional series-parallel coupling. In addition, since the interleaved method increases the switching frequency by the number of interleaved, there is an advantage in that the device rating and size of the inductor and capacitor are reduced, and the current ripple is reduced.

FIGURE 5.
Interleaving structure and operation of non-isolated dc-to-dc converter for battery charging and discharging, (a) circuit configuration of the interleaved non-isolated converters, (b) key waveforms at step-down operation.
In the interleaved structure, the number of inductors and switches is required as much as the number of phases (N ), and the battery charging current i B flows through each phase divided by 1/N times. In Fig. 5(b), each phase's inductor current (i Lx , x=1,2,3,4) has a phase difference of 2π/N , that is, 360/N degrees. The slope of the inductor current is as follows.
Here, i Lxon is the inductor current in the rising section, and i Lxoff is the inductor current in the falling area. Therefore, the VOLUME 10, 2022 current ripple for one phase is defined as (10).
The converter's input current is represented by the sum of the inductor currents in each phase.
Since the inductor currents of each phase cross and cancel each other, the battery charging current ripple is reduced by at least 1/N times compared to a single phase, and the period becomes T/N . The ripple frequency of the charging current increases by N times the inductor current. The ripple of the battery charging current can be expressed as (12).
Here, N on denotes the number of switches that maintain an ON state during the T period of the input current, and N off denotes the number of switches that maintain an OFF state. By substituting (8) and (9) into (12), the result is as follows.
Here, N is the number of phases, and d is the ratio of the rising section of the input current. The current ripple of the non-isolated converter of the proposed interleaved method is summarized according to the duty ratio as follows.
The reduction of capacitor voltage ripple due to the battery charging current ripple reduction can be obtained by using the average charge Q of the capacitor in Fig. 6.
Therefore, the capacitor voltage ripple is as follows. In the interleaved converter, the current sharing ratio is proportionally reduced by the number of phases, so the magnetic field energy accumulated in each phase inductor is expressed as (21).
According to (21), the stored energy of the inductor is reduced by 1/8 times as much as the current sharing ratio of each phase is reduced by the 4-phase interleaved method. This has the effect of lowering inductance and inductor volume. In the non-isolated bidirectional converter of Fig. 5(a), Buck mode and Boost mode operate interlock. In this paper, only the step-down mode is described for the convenience of analysis.
In step-down mode, the upper switches (T 1A , T 3A , T 1B , T 3B ) become the main switches, and the lower switches (T 2A , T 4A , T 2B , T 4B ) provide a freewheeling current path for the internal body diode. Fig. 5(b) shows the fundamental waveforms when the bidirectional dc-to-dc converter operates in stepdown mode. The current flowing through the inductor of each phase rises during the switch's turn-on and falls during the turn-off. At this time, each phase current has a phase difference of 360/N degrees. In the step-down mode, as the upper switch is turned on, energy is accumulated in the inductors L 1 , L 2 , L 3 , and L 4 and is discharged to the battery through the lower switches T 2A , T 4A , T 2B , and T 4B . At this time, the average voltage (v L ) across the inductor is the same as (22), and the relationship between the input/output voltage and the duty ratio is defined as (23).

1) STEADY-STATE ANALYSIS FOR THE CURRENT RISING SECTION OF THE INDUCTOR
Since the operation of all phases in the steady state is the same except for the phase difference of 360/N degrees, only one phase (L 1 ) is described for convenience of analysis. The relationship between the voltage across the inductor and the inductor current in the rising slope section in the step-down mode is shown in (24). Since the inductor voltage is equal to the difference between the battery voltage and the dc-link voltage, the slope of the current flowing through the inductor can be expressed as (25).
In the steady state, if the initial current value of the inductor is an arbitrary minimum value (i L1min ), the inductor current during the turn-on period of the switch rises to the maximum value (i L1max ) at t=DT. Therefore, the maximum inductor current is expressed by (26), and the inductor current ripple equals (27).

2) STEADY-STATE ANALYSIS FOR THE CURRENT FALLING SECTION OF THE INDUCTOR
The voltage across the inductor and the current slope in the section when the switch is turned off is as follows.
While the switch is turned off, the inductor current falls from its maximum value (i L1max ) to its minimum value (i L1min ) at the point where t=T . At this time, the minimum current value of the inductor is (30), and the current ripple is expressed as (31).

C. DESIGN CONSIDERATION
The inductor should be designed considering the buck and boost modes in the proposed circuit configuration. In bidirectional operation, when the same load current in continuous current conduction mode (CCM) flows, the inductance of the inductor is generally set based on the buck, which requires a higher value. Table 1 shows the design specifications for one module among the proposed battery charging/discharging system circuit structures. The switching frequency is 40kHz, and the dc-link is boosted using the battery voltage in boost mode.

1) INDUCTOR DESIGN
When one 8kW module (2-phase interleaved) operates in buck mode, it is designed to work as CCM at 10% load current of the rated power capacity. The battery current corresponding to 5% of the rated power capacity is 2.67A, and the current flowing through the inductor of each phase becomes 1.33A. When the dc-link voltage reaches the upper limit when operating in Buck mode, the duty ratio D is 0.83 by (23). And to operate the battery charging current corresponding to 5% of the rated power capacity in CCM mode, the minimum value of each phase inductor current is 1.33A. The minimum inductance to satisfy CCM operation under the buck mode condition is 235µH. In the experiment, a value of 330µH, which is sufficiently more significant than the boundary value, is used to ensure complete CCM.

2) CAPACITOR DESIGN
Since the dc-link voltage ripple increases when operating in boost mode, the capacitance of the capacitor is selected based on the boost mode in Fig. 9.
In Boost mode, the duty ratio considering the step-up ratio of the battery voltage and the dc-link voltage is 0.167. Also, when operating at rated power capacity in Boost mode, R is 4.05 when obtained with a rated current of 44.4A for a dc-link voltage of 180V.
In (33), the dc-link voltage ripple was set to 1.8V, 1% of the dc-link voltage. Regarding voltage magnitude and stress, capacitors are generally selected up to 10 times larger. Three 470µF electrolytic capacitors were connected in parallel in the experiment, and 1410µF was selected. Fig. 10 shows the battery charge/discharge control block diagram. It consists of a current control unit, a voltage control unit, and a compensator. It is composed of a PI controller with anti-windup added to the voltage controller to prevent saturation of the controller when changing from CC mode to CV mode. In the battery charge/discharge system, a compensator is added to the current control loop to ensure a fast response to load fluctuations. Voltage sensing accuracy is critical in battery control. Suppose the error between the actual battery voltage and the measured battery voltage is about 1V. In that case, the CC mode can be controlled by the integrator of the battery charge/discharge controller, and an overcharge may occur in the battery. Methods such as a lookup table and curve fitting are widely used to compensate for the accuracy of the battery sensor voltage. In this paper, the low and high-voltage sections are divided and designed to reduce the voltage sensing error through curve fitting.  A more accurate fitting value can be obtained in the low voltage section through a function higher than the quadratic function. Still, the low voltage section is 1.8V, which is not the charging/discharging region of the battery, so it is expressed as a quadratic function that can tolerate the error rate. Table 2 compares the error rate between the measured voltage error and the actual voltage error. The curve fitting shows that the control error rate from 10V to 150V is up to 0.067%, and a precisely accurate sensing voltage performs control.

IV. SIMULATION AND EXPERIMENTAL RESULTS
In this chapter, the feasibility and superiority of the proposed battery charge/discharge circuit topology are verified through simulations and experiments using prototypes. In the proposed battery charging/discharging system, DAB has a significant purpose of electrical insulation with a proper voltage ratio. The power delivery direction is determined by the dead time of the DAB switch and the voltage difference between the primary and secondary of the transformer. Fig. 12(a) shows that the DAB primary voltage phase precedes the secondary voltage phase, so power is transferred from the transformer primary to the secondary. In Fig. 12(b), since the voltage phase of the transformer's primary lags that of the secondary, power is transferred from the secondary to the primary. At this time, the magnitude of the inductor current is determined by the voltage across the transformer magnetizing inductance. Fig. 13 is the simulation result of analyzing the voltage balancing characteristics for the case where the split capacitor's difference in capacitance and voltage connected in series to the DAB input side occurs in the proposed circuit structure. In Fig. 13(a), it can be confirmed that the voltages of the split capacitors are balanced after about 5ms even when the voltage difference between them is from 10% to 90%. However, when the voltage difference is 90%, the peak voltage due to the initial overshoot is about 550V and the primary current increases from about 40A to 150A. Suppose the peak voltage and current generated by the difference in the initial voltage of the capacitor is greater than the rated value of the capacitor or switch. In that case, there is a possibility of damage. Therefore, if a specific voltage difference or more occurs, a protection operation or an additional balancing operation is necessary. Fig. 13(b) is the simulation result for the case where the capacitance of the split capacitor connected in series to the input of DAB differs from 10% to 90%. Within 5ms after the overshoot, the split capacitors are voltage balanced. When the capacitance difference between the split capacitors is about 90%, the initial overshoot voltage and current generate peaks of about 440V and 100A. However, unlike Fig. 13(a), the overshoot continuously occurs due to the difference in capacitance, which causes continuous stress in the system. Fig. 14 shows the simulation results of the split capacitor voltage and DAB current according to the difference in initial voltage and capacitance of the split capacitor. The maximum overshoot of the capacitor voltage generated by the difference in the initial voltage of the capacitor becomes higher than 500V, which is the working voltage of the capacitor,  after about 70%. The primary current of both DAB modules increases constantly, and the magnitude of the current after 70% becomes about 100A or more. After about 70% of the voltage balancing due to the difference in capacitance, the upper module is 370V, and the lower module is 430V, resulting in a voltage difference of about 50V. The current also increases in proportion to the voltage. Fig. 15 shows the simulation results of analyzing the characteristics of the converter output voltage change (changed to 25% from the rated load) according to the load variation. It is the simulation result of switching from 8kW to 2kW and vice versa. At this time, the output voltage fluctuation appears as good as 10V. Fig. 16 shows the prototype for the battery charge/ discharge tester proposed in this paper. Two DABs and four non-isolated converters have an interleaved structure. An initial charging relay is added to prevent inrush current, and SMPS is used as a power source for the controller and relay operation. The controller used TI's TMS320F28377D and separately designed the sensing unit (CPU2) and the control unit (CPU1). Fig. 17 shows the results of the operation experiment for the DAB used to secure a wide input voltage range and electrical insulation in the proposed approach. Since one module is designed as 8kW and the input voltage is 360V, in the case of a prototype in which two modules are combined, the 16kW input voltage can be operated at 720V. However, it was tested up to 5kW, considering the laboratory power supply conditions. Fig. 17(a) is the upper module's DAB primary and secondary voltage and current waveforms. The output voltage (input voltage of the non-isolated converter) is controlled to 150V. Fig. 17(a) is the experimental waveform before setting the switching angle so that the DAB switch current is flat. Therefore, the switch currents i pH and i sH are not flat current waveforms and the current peak increases as the load power capacity increases. Fig. 17(b) is an enlarged waveform at the turn-on moment of the DAB switch. From the primary and secondary voltages of the DAB, it can be seen that the phase of the primary voltage leads to the secondary voltage. Therefore, power is transferred from the primary to the secondary. Fig. 18 shows the general operation waveform of the DAB converter when the load power capacity is increased to 500W, 1kW, 3kW, and 5kW. Each waveform is a primaryside voltage, secondary-side voltage, primary-side current, and secondary-side current waveform from top to bottom. The current peak becomes more significant as the magnitude of the power increases, and in particular, the current peak of the secondary is more significant than that of the primary. Fig. 19 shows the operation waveform of the DAB converter after setting the switching angle so that the DAB switch current is flat. Each waveform is a primary-side voltage, VOLUME 10, 2022  secondary-side voltage, primary-side current, and secondaryside current waveform from top to bottom. Since the DAB switch current is controlled flat, it can be seen that the current peak does not increase significantly even when the load power capacity increases. Since the current peak is reduced, the conduction loss of the switch can be diminished, and the current rating of the switch can be reduced. In addition, since the phase difference of the voltage across the DAB is maintained not to be significant, it can be confirmed that the generation of reactive power due to the circulating current is small, so the loss due to the DAB is not significant.   20 is an experimental waveform that is soft-started by slowly decreasing the PWM dead time after charging the internal input capacitor through resistance and relay operation to prevent inrush current during the initial operation. Even with a soft-start function, the peak current is up to about 60A due to the large capacitance. It shows the stable operation in a steady state after the transient. Fig. 21 shows the experimental waveform of response to load change. Fig. 21(a) is the experimental waveform when the load is changed from 100% load to 50% load. It shows stable response characteristics without overshooting. Fig. 21(b) is the test result waveform for voltage fluctuation. Even if you change the reference voltage from 150V to 100V, back to 150V, and then to 50V, you can see that the output voltage follows quickly. Fig. 22 is an experimental waveform for analyzing the interleaved characteristics of the non-isolated converter. It can be seen that the 4-phase interleaved operation works with a 90-degree phase difference. Fig. 23 is an experimental waveform to check the battery's charge/discharge response characteristics. A battery was modeled using a supercapacitor, and a charge/discharge system was implemented using a voltage controller and a current controller in the CV-CC controller loop. The target voltage was increased by about 10V to operate in CC mode, and it was confirmed through an experiment that the battery voltage reached nearly 90% and then worked in CV mode. When self-discharge is established by giving the battery a rest period, the voltage drop due to self-discharge is tiny.     converter is around 1, it shows that ZVS is possible in almost all load regions. Fig. 26 shows four series-parallel structures for the proposed converter with wide input/output voltage ranges. The proposed converter circuit structure is modular and easy to expand. When input terminals are connected in series, the input voltage range can be increased, and when combined in parallel, input power capacity can be increased. Similarly, if the dc-link stages are connected in series, the range of the dc-link voltage can be increased, and when combined in parallel, the dc-link power capacity can be increased. What is unique is that the input and output voltages of individual DAB converters are all the same regardless of the series-parallel structure. The voltage range increases with the series combination, and the power capacity rises only with the parallel combination. The output stage is composed of an interleaved non-isolated dc-dc converter, which can reduce current ripple as the module coupling increases. Table 3 shows four series-parallel structures that can be created by combining two modules in the proposed converter. In Case 1, two input voltages are connected in series, and the dc-link stages are connected in parallel. In Case 2, two input voltages are connected in series, and the dc-link stages are also connected in series. In Case 3, two input voltages are connected in parallel; and the dc-link stages are also connected in parallel. In Case 4, two input voltages are connected in parallel, and the dc-link stages are connected in series. In all four cases, the actual input voltage applied to the individual DAB is 360V, and the dc-link voltage applied to the individual non-isolated dc-dc converter is 180V, which is the same. Therefore, the non-isolated dc-dc converter uses 180V as an input voltage to control the battery charging voltage in the range of 0 to 150V. Fig. 27 compares the efficiency of the four series-parallel structures that can be created using the combination of two modules in the proposed converter. In the average efficiency for each case in Fig. 27(d), Case 1 shows the highest efficiency of 96.48% at 2kW load power capacity and the lowest efficiency of 93.118% at 5kW. The best efficiency is displayed in Case 3, and the highest efficiency was measured to be 96.8% at 1kW load power capacity, and the lowest FIGURE 23. Battery charge/discharge test, (a) 50V charge, then discharge to 40V, (b) 60V charge, then discharge to 50V, (c) 70V charge, then discharge to 60V, (d) 80V charge, then discharge to 70V. efficiency was measured to be 93.9% at 5kW. It can be judged as a result of Case 3 having a parallel structure in both input and dc-link. Overall, the average efficiency for all cases shows a difference of 1%. These results indicate that although the proposed converter can easily extend the input and output range through the module's series-parallel combination, the module series-parallel combination of the module does not significantly affect the efficiency because it operates in the form of a unit module. Table 4 shows four circuit configurations that can be combined using two DAB converters. The voltage applied to each DAB is 360V regardless of the series or parallel combination of input voltages. If the outputs are parallel coupled, the output of each DAB must be equal to the battery charging voltage. When the output stage is series-coupled, the sum  of the outputs of each DAB becomes equal to the battery charging voltage, so each DAB is output at half the battery voltage, and the sum of the two output voltages appears as VOLUME 10, 2022 FIGURE 26. Series-parallel structure for the proposed converter to have a wide input/output voltage range, (a) the input terminal is connected in series, and the dc-link terminal is connected in parallel, (b) the input terminal is connected in series, and the dc-link terminal is also connected in series, (c) The input terminal is connected in parallel, and the dc-link terminal is also connected in parallel, (d) the input terminal is connected in parallel, and the dc-link terminal is connected in series.
the battery charging voltage. Therefore, all four cases that can charge and discharge the battery using two DAB converters use 360V as input and generate a battery voltage of 50∼150V. Therefore, it can be equivalent to the condition that one DAB converter generates an output voltage of 50∼150V  with an input voltage of 360V. The efficiency of generating output voltages of 50V, 100V, and 150V with one DAB converter module is measured and compared with the proposed converter.
As shown in Fig. 28(a), one DAB converter has the highest efficiency of 96.45% under the output voltage of 150V and 1kW generation and the lowest efficiency of 91.5% when   generating 50V and 5kW. This is because when the output voltage is high, the phase difference between both sides of the DAB is insignificant. Still, when the output voltage is low, the phase difference between both sides of the DAB is significant.
Hence, the circulating current dramatically increases, and the loss increases due to the increase in reactive power. Fig. 28(b) shows the efficiency comparison result between the average value of Fig. 28(a) and the four possible combinations of the proposed converter. One DAB converter shows relatively low efficiency by 1∼2% compared to the average efficiency of the four cases of the proposed method. In the case of the proposed method, the DAB can lower the peak of the operating current, the circulating current is small because the phase difference is not significant, and the switching loss is low due to soft switching. However, the switching loss increases as the load power capacity increases due to the hard switching of the non-isolated dc-dc converter. On the other hand, in the DAB converter, as the load power capacity increases, the DAB current peak increases proportionally to increase the conduction loss. When the output voltage is 50V, the phase difference of the DAB converter is more significant than when the output voltage is 150V, so the circulating current and reactive power increase accordingly, resulting in increased loss. In addition, since the output voltage control is also performed at the same time, a problem arises that the configuration of the controller becomes complicated. Table 5 compares the characteristics of the proposed method with a similar approach using a DAB converter. The proposed converter connects the DAB converter and the nonisolated converter in series-parallel module form to ensure ZVS and have a wide input/output voltage range and to reduce the output ripple current with the interleaving structure. Regarding efficiency, [54] and [65] are the highest. However, it can be said that the proposed approach is the most excellent structure that has good efficiency characteristics and can cope with a wide input voltage and output voltage range.

V. CONCLUSION
We propose a converter structure for a battery charge/ discharge tester that can respond to a wide input/output voltage range. The switching angle is controlled to minimize the circulating current and flatten the DAB current by keeping the minimum phase difference between the voltages at both ends of the DAB converter used for bidirectional power transfer and galvanic isolation. A non-isolated dc-to-dc converter is used to control the output voltage, and interleaved operation is used to reduce the output current ripple and improve the battery lifecycle.
The voltage balancing characteristics due to the difference between the initial voltage of the series-connected input capacitor and the capacitor capacity were analyzed, and design parameters were presented. The technical feasibility of the proposed system was verified through simulation and prototype experiments. As a result of the testing and measuring the efficiency of the battery charging and discharging of 50V∼150V under the condition that the proposed converter composes two modules in series and parallel, it showed good efficiency of 93.3∼96.6% and confirmed fast and stable response characteristics. Since the proposed circuit structure is easy to expand in a module, it will be suitable as a high-capacity battery charge/discharge tester with a wide input/output voltage range.