Design of a Modified-Bridge Circuit with a Master-Slave Input Supply Mechanism for Ozone-Driven System Applications

Abstract

This study proposes a design of a modified-bridge circuit with a master–slave input supply mechanism for ozone-driven system applications. Because the single-source supply design is becoming the mainstream choice in the existing ozone-driven systems, the input supply reliability of the ozone-driven system is crucial. Therefore, this proposed design involves a modified-bridge circuit combined with inductors and transistors, which can be augmented with the energy storage device as a backup source to improve the reliability of the input supply for the ozone-driven system. In addition, considering that the original source directly connected to DC BUS can re-charge the energy storage device, the energy recycling operation mode is designed in this proposed system to extend the duration of the energy storage device, which improves the supply reliability of the ozone-driven system further. To validate this proposed system, both model formulation and hardware realization are assessed through different test scenarios. Experimental outcomes of these tests confirm the practicality of the proposed design.

1. Introduction

Ozone technologies have seen wide application in various fields, such as agriculture, sterilization [1,2,3], water purification [4,5,6], portable air cleaners [7,8], and surface cleaning [9,10]. In ozone technology applications, signals with both high-voltage and high-frequency features must be generated to achieve proper electrical discharge in the ozone chip. Therefore, a suitable driving circuit or system circuit design is crucial. In some studies, advanced components with a high-frequency capacity have been used to build the ozone load driving resonant circuit [11,12,13,14]. In addition, resonant circuit designs such as LCC type or hybrid type have been adopted, due to the high voltage-gain property, to generate a high-voltage signal in a high-frequency range [15,16,17,18]. The existing ozone-driven systems are most often designed with a single input port. If a fault occurs in the power source during operation, the ozone-driven system must stop working and wait for power recovery, which significantly affects the ozone quality. Adopting designs using a single power source inevitably impairs the operational reliability of the ozone-driven system in the above application fields. To improve the reliability of the input source, connecting directly the energy storage devices to the ozone-driven system may seem to be a quick fix. However, there are issues that remain to be resolved to achieve the voltage balance between the energy storage device and the input power source.

To solve the voltage balance problem, energy storage devices are often integrated with the main input power for better power quality. Therefore, we believe that adopting a dual-input port structure in the construction of ozone-driven systems will provide a viable solution. In [19], Saleh et al. connected the multi-level converter to the energy storage devices to drive the DBD plasma load adopting multi-level switching architecture and using multiple DC converters, which provide multiple input ports for connecting energy storage devices to overcome the problem resulting from the single power source. In addition, some studies have proposed a multi-input converter with a special control strategy to extend the input amount when connecting to energy storage devices. For example, the integration of energy storage devices and supercapacitors provides the energy for reducing the numbers of DC converters through a novel multiple-input converter [20]. Furthermore, a triple-input converter is designed for three kinds of power sources, adopting an interleaved full-bridge converter that integrates each power source to improve the supply reliability [21]. The above studies have proposed novel architectures to integrate two or more power sources and achieve encouraging results. Next, because the energy storage device is adopted as another source, the re-charging function of the energy storage device is also essential. Therefore, some studies have discussed bi-directional power transmission. For instance, an isolated converter with a bi-directional power transmission is proposed for a two-power-source input and power control in [22]. In addition, these studies have discussed the interleaved buck/boost converter with zero-voltage switching control for some bi-directional converter designs [23,24]. Furthermore, to obtain a high voltage gain with high efficiency to drive the ozone chip, some resonant circuits have been proposed [15,16,17,18]. Such a circuit consists of magnetic inductors to reach the resonance while satisfying the requirement of soft switching. Yet, the characteristics of this resonant circuit were easily affected under the bi-directional power transmission since its architecture was asymmetrically connected. Therefore, some studies have adopted the symmetrical resonant circuit for the bi-directional power transmission to achieve the stabilization of the bi-directional power transmission [25,26,27]. The aforementioned methods reveal that encouraging results were observed, yet each of them came with some inherent demerits in the ozone-driven applications. Moreover, considering that the ozone-driven system is usually applied to a cleansing function, the reliability of the ozone-driven system will affect directly the cleansing quality. Ozone-driven systems, which have been applied widely, are costly. Therefore, substituting the novel ozone-driven equipment for the traditional one is a difficult mission. As mentioned above, considering the replacement issues, the design of the extra-installing circuit is an excellent choice, which lets the original ozone-driven system continuously provide a stable voltage by increasing the reliability of the power supply in order to maintain a stable ozone generator. To clearly understand the relationship between each aforementioned reference and the corresponding study subject, the features of the above references are as listed in

Table 1. References of feature analysis and the corresponding subjects.

No.	Features	Reference
[11]	An accurate analytical converter-level model is introduced. The model evaluates and optimizes the switching solution of a SiC MOSFET for the pulsed power converter.	SiC MOSFET application of the ozone-driven circuit design for increasing the component reliability
[12]	A high-power high-frequency ICP drive system that achieves high efficiency and durability is proposed. A 10 kW 4 MHz switching inverter and a novel control method are proposed for ICP with a high Q factor and highly variable impedance. The proposed phase lead sensing circuit can always guarantee ZVS operation of a high-power high-frequency switching inverter. Through the proposed system, ICP with a much higher plasma density can be safely driven.
[13]	An optimized version of a gate-boosting driver circuit for SiC MOSFETs for ultrafast switching is proposed. A MOSFET turn-on time of below 2 ns is achievable at a high operating voltage and moderate current and below 3 ns for a wider range of load-current and operating voltage conditions.
[14]	A high-voltage SiC MOSFETs module can serve as the main switch in a repetitive high-voltage nanosecond pulse generator. A series-connected MOSFETs module with only a single external gate driver requiring few components is suitable for a compact assembly. A 10 kV SiC MOSFETs module with a turn-on transition time ~10 ns is developed. A compact and high-voltage pulse generator composed of three 10 kV SiC MOSFETs module is tailored, with a typical rise time of ~40 ns and peak a voltage of ~30 kV.
[15]	A 2.0 kW high-voltage rep-rate charging power supply based on a non-isolated inductor–capacitor–capacitor (LCC)-type resonant converter operating in continuous conduction mode is proposed. It provides a constant charging current with high control accuracy/efficiency. The design of the proposed capacitor charging power supply (CCPS) is a low-cost, simpler, compact, and efficient driver for the miniature pulsed plasma sources.	Different resonant architecture analysis for acquiring a higher voltage gain
[16]	A hybrid LLC resonant converter design is proposed. There are three modes of operation: As a full-bridge converter, as a dual-phase half-bridge LLC converter, or as a single-phase half-bridge LLC converter. Using GaN devices with a power density of 65 W/in 3 has been designed.
[17]	A comparative half-bridge LLC resonant converter using the integrated LCLC resonant transformer is proposed. To reduce the demanded leakage inductance, a resonant capacitor can be connected in parallel to the magnetizing inductor through auxiliary winding as an LCLC resonant transformer. An integrated LCLC resonant transformer with lower leakage inductance for efficiency comparison is implemented.
[18]	The light-load efficiency of the LLC resonant converters is an improvement. The simplified optimal trajectory control (SOTC) for burst mode can solve this issue by using an optimal switching pattern for burst-on time. The digital implementation of SOTC for burst mode and its limitation in a high-frequency LLC converter are discussed.
[19]	Multistage and multilevel switch-mode converters are employed to construct the power supply. The proposed MSML-PEC power supply uses boost dc-dc PECs without high-frequency transformers. The removal of high-frequency transformers has improved the average efficiency and reduced the size of each dc-dc PEC. Modular, lightweight, and compact size power supply has been achieved.	Multiple input design for adding the input port to increase the supply reliability
[20]	A novel multi-input converter (MIC) is proposed in this article for interfacing a battery-SC combination. The SC converter is mostly rated for the full surge power. The battery is operated in current control mode under normal loading conditions. The proposed system balances the flow of power between the load and the HESS.
[21]	Modeling and two controller design techniques for a triple active bridge (TAB) three-port dc-dc converter comprising the fuel cell, the battery, and the load are proposed. A new decoupling matrix-based proportional-integral controller design method reduces the design complexity and improves the system dynamic performance in comparison to similar three-port converters reported in the paper.
[22]	A multiple-input configuration of an isolated bidirectional dual-active bridge dc-dc converter (MIBDC) is proposed for power flow control in a combinational battery storage. The proposed system can be operated in an independent source or a combinational source mode of operation to control the power transfer. It has the capability of bidirectional power flow and smoother transition. The proposed configuration can also be used with unequal voltage level sources by connecting them in series on the multi-input side.
[23]	A ZVS interleaved buck/boost bidirectional converter using the active-clamp technique is introduced in this paper. Applying a simple auxiliary circuit comprised of an auxiliary switch, an auxiliary inductor, and a clamp capacitor, ZVS condition is achieved by the converter operating duty cycle or load condition. The converter can operate with the conventional fixed-frequency pulse-width modulation (PWM) control.	Bidirectional converters and symmetrical resonant circuit discussion for delivering and re-charging of energy regulation
[24]	An interleaved bi-directional buck-boost is proposed for achieving zero-voltage switching. Applying a coupled inductor with a variable coupling coefficient achieves a wide input–output voltage operating range. The variable coupled inductor significantly improves the soft-switching range and reduces the circulating energy.The gallium nitride interleaved bi-directional buck-boost converter is applied in this paper.
[25]	An integrated half-bridge CLLC (IHBCLLC) resonant bidirectional dc-dc converter is proposed in various applications, including energy-storage systems. A half-bridge CLLC resonant circuit and a buck/boost circuit are integrated. The proposed converter obtains the high voltage gain by pulse-width modulation. The CLLC resonant circuit can help MOSFETs to achieve soft switching and high voltage gain.
[26]	The total power loss optimization and the magnetic design of the CLLC resonant converter based on the artificial intelligence (AI) algorithm are proposed in this paper. The parameters of the magnetic component and the resonant component are derived by the AI algorithm. The parameters of the transformer are optimized by the AI algorithm. A planar transformer is used for CLLC designing in this paper.
[27]	The CLLC bidirectional resonant converter is discussed in this paper. Application fields are battery chargers and dc microgrids. Asymmetric parameters methodology (APM) can design gain curves for charging and discharging modes separately.

.

In this paper, a modified-bridge circuit is proposed for the ozone-driven system, which modifies a bridge circuit with inductors isolated to provide dual-input ports. Here, the master source and the slave source are connected to different input ports to independently provide the energy while maintaining the source status of each port. In addition, a symmetrical resonant tank is adopted and designed for the bi-directional power transmission, letting the resonant property of the bi-directional power transmission be similar. Once the original source is connected directly to DC BUS, the energy can be re-charged to the slave source through the proposed system. Thus, delivering or re-charging the energy storage device becomes much easier by the proposed system and the discharging time of the energy-storage device is also improved. Moreover, through the voltage of each port being regulated by the controller, different operation modes can be selected by simple frequency modulation. The features of this proposed system are listed as follows:

(1)

The proposed modified-bridge circuit accomplished in this paper emphasizes the benefits of using two input power sources that are supplying coordinately to improve the supplied reliability and reduce the component cost for the ozone-driven system.

(2)

The proposed system can be extra-installed in the original ozone-driven system, which means that it is not necessary to replace the original one, keeping the ozone quality and reducing the replacement cost.

(3)

The resonant circuit design is of a symmetrical type, which achieves a similar resonant property under the bi-directional power transmission. Difficult control and a complicated algorithm are unnecessary to enhance the stability of the proposed system.

(4)

The voltage status of each port can be detected through the control system operation mode setting by judging whether the energy is being provided through DC BUS or re-charged to the slave source (energy storage device).

The rest of this article is organized as follows: Section 2 introduces operations of the proposed system, Section 3 discusses the parameter design and analysis of the symmetrical resonant circuit, Section 4 describes the control mechanism analysis, Section 5 reports the experimental results, and Section 6 presents the conclusion.

2. Operations of the Proposed System

Figure 1a shows the application illustration of the proposed system in this paper. As shown in this application illustration, the two parts are the original system and the extra-installed system. In the original system, the energy of the ozone-driven system is directly supplied by the original supply through DC BUS [28,29]. It is noted that the output voltage of the original supply fits the voltage of DC BUS. To avoid ozone-driven system malfunction when a fault occurs in the original supply, the proposed system is connected to DC BUS. Moreover, both sources are included in the proposed system, the low-voltage PFC source [30,31,32,33] and energy storage devices. As shown in Figure 1, when the main supply delivers the power to DC BUS, the main supply not only provides the energy for the ozone-driven system through DC BUS but also charges the energy storage device simultaneously. When the main supply fails, the energy of DC BUS will be supplied by the proposed system. The master and the slave sources can both provide the energy. If the master source is normal, it can provide the energy for DC BUS and re-charge the energy storage devices through the proposed system at the same time. If the master source provides insufficient energy and causes abnormality, the energy storage devices will provide independently the energy for DC BUS through the proposed system to regulate the energy of each source and maintain the input voltage for the ozone-driven system. Figure 1b shows the extra-installed system illustration of the proposed system. As shown in this block diagram, the proposed system feeds the energy to the original ozone-driven system through DC BUS. When the energy is provided by the master source or the slave source, the voltage signal of each port transfers to the feedback circuit, the controller, and the isolated gate driver to deliver suitable driving signals of the power transistors of the power stage and allow the output voltage of the proposed system to be stabilized. In Figure 1b, the voltage feedback of the ozone load transfers to the original ozone-driven system for stabilizing the ozone generation exciting and the proposed system does not intervene in the ozone control, which means that the proposed system can be applied to any existing ozone-driven system. The application compatibility of the proposed system is higher than any solutions.

2.1. System and Circuit Description

The architecture of the proposed system is shown in Figure 2. There are five parts: a master–slave input supply circuit, a symmetrical resonant tank, a full-bridge circuit, a control circuit, and an ozone generator. The proposed system uses a master–slave input supply circuit formed by six power transistors (Q₁–Q₆), two inductors (L₁ and L₂), and two diodes (D₁ and D₂). D₁ and D₂ provide current paths when L₁ and L₂ release the energy in order to avoid a higher voltage spike across the two inductors, which will occur when instantaneous current variation without current freewheeling path emerges. Furthermore, two input ports are designed to let two power sources be applied to this proposed system. The inductors are applied to isolate and store the energy. To smoothly release the energy of the inductors through any path and avoid the energy being delivered to the master source, Q₅ and Q₆ are located upside down and different as a conventional full-bridge circuit. The proposed system also includes a symmetrical resonant tank composed of two resonant inductors (L_r1 and L_r2), two resonant capacitors (C_r1 and C_r2), and a step-up transformer (T). In addition, the proposed system possesses a full-bridge circuit consisting of power transistors (Q₇–Q₁₀) and an output capacitor (C_D). The proposed system not only flexibly delivers the energy to DC BUS for the ozone generator but also opportunely recovers the energy from DC BUS. First, the master power supply (Master source) and the energy storage device (Slave source) are connected to the master–slave input supply circuit. Both the output of the full-bridge circuit and the ozone generator are connected to DC BUS. When the original source connects directly to DC BUS and provides the ozone generator and the proposed system with the energy, the frequency of Q₇–Q₁₀ will be controlled to convert the energy of DC BUS to the voltage at high frequency, which is transferred through the symmetrical resonant tank and the transformer, and the high-frequency voltage signal is rectified and stabilized by the body diodes of Q₁–Q₄ and capacitor C_s, respectively. Finally, the energy will be charged to the energy storage device (the slave source). When the original source fails, the energy of the ozone generator must be provided through the proposed system. The master source (V_m) delivers the energy through the master–slave input supply circuit with the high-frequency square AC voltage being converted by controlling the frequency of Q₁–Q₆. This high-frequency square AC voltage passes through the symmetrical resonant tank and the transformer and is rectified and stabilized by the body diodes of Q₇–Q₁₀ and capacitor C_D, respectively. Next, the energy of C_D is transferred to the ozone generator through DC BUS. In addition, to ensure that the energy storage device (the slave source) maintains high voltage levels for full charging, the master source provides the energy to the energy storage device by controlling the frequency of Q₁–Q₆. If a fault occurs in the master source, the energy will be provided by the slave source (V_s) by controlling the frequency of Q₁–Q₆ to convert the energy of the slave source to the high-frequency square AC voltage, which was transferred through the symmetrical resonant tank and the transformer and is rectified and stabilized by the body diodes of Q₇–Q₁₀ and the capacitor C_D, respectively. Subsequently, the energy of C_D is transferred to the ozone generator through DC BUS.

To improve the input supply reliability of the ozone generator, the master–slave input supply circuit is adopted. However, when two sources are applied to provide the energy for DC BUS, the voltage of each source must be considered. Therefore, the inductors of the master–slave input supply circuit (L₁ and L₂) are applied to regulate the energy of each source in this study. Here, to design these two inductors, the duty cycles of all transistors are 50% and the boundary between continuous and discontinuous inductor current I_Li-min must be 0. L₁ and L₂ are as follows:

$$I_{Li}=\frac{V_s{}^2{i}_s}{V_m\left(V_s/i_s\right)}=\frac{V_m{}^2}{{\left(1-D\right)}^2\left[V_m\left(V_s/i_s\right)\right]}=\frac{V_m{i}_s}{{\left(1-D\right)}^2{V}_s},i=1\mathrm{or}2$$

(1)

$$\Delta i_{Li}=\frac{V_m}{2L_i}{T}_s=\frac{\left(V_s-V_m\right)}{2L_i}{T}_s,i=1\mathrm{or}2$$

(2)

$$I_{\mathit{\text{Li-min}}}=I_{Li}-0.5\Delta i_{Li}=\frac{4V_m{i}_s}{V_s}-\frac{V_m{T}_s}{4L_i}=0,i=1\mathrm{or}2$$

(3)

where I_Li is the average current of the inductor; D is the duty cycle of the power transistors; which is set up at 0.5; and T_s is the time period. From Equation (3), the minimum inductor for continuous current can be written as below:

$$L_i=\frac{0.0625V_s}{f_s{i}_s},i=1\mathrm{or}2$$

(4)

From Equation (4), since the duty cycles of all transistors are about 50%, the voltage and the current of the storage device are only taken into consideration in the inductor design. If the current of the inductor is designed in the continuous mode for lower charging power of the storage device, the larger inductor must be designed; otherwise, the smaller inductor is enough in this proposed system.

Table 2. The parameters of the proposed system.

Specifications	Title 3
Master input source voltage, Vm	48 V
Slave input source voltage, Vs	40 V–54.6 V
Output voltage, Vo	400 V
Power transistors, Q1–Q10	IXFH44N50P
Storage inductors, L1 and L2	1.21 mH
Capacitors of the master source and the slave source, Cm and Cs	470 μF/100 V
Output capacitors, CD	470 μF/450 V
Maximum output current, Io	2 A
Maximum output power, Po	800 W

shows the circuit specification of the proposed system.

2.2. Operation Mode and Circuit Analysis

Since this proposed system can be applied in three operations, the circuit operations must be analyzed for each operation mode. Figure 3, Figure 4 and Figure 5 depict the circuit operation for three operations, and Figure 6 shows their corresponding waveforms. First, the circuit operations will be analyzed for the master source supply operation, as shown in Figure 3 and Figure 6a. The operation is described as follows:

(1)

Mode I (t₀–t₁): As shown in Figure 3a, Q₂ and Q₆ are turned off and Q₄ and Q₅ are turned on, achieving zero-voltage switching in this mode; the energy of the master source is stored in inductor L₂ by turning on Q₄ and releasing the energy of inductor L₁, which is stored in the energy storage device and C_s by turning on Q₅ and the diode of Q₁. Therefore, the currents of L₁ and L₂ (i_L1 and i_L2) linearly decrease and increase, respectively. At this time, since the resonant voltage v_r1 of the primary side is a positive square wave, the resonant current i_r1 of the primary side flowing through Q₁ and Q₄ is positive and the resonant current i_r2 of the secondary side flows through the body diode of Q₇ and Q₁₀ to transfer the energy to DC BUS with output capacitor C_D. It is noted that the current i_m of the master source has low ripples because i_L1 and i_L2 are interleaved to store and release energy, which can contribute to mitigating the burden of the master source supply. This operation enters the next mode until Q₄ and Q₅ are turned off.

(2)

Mode II (t₁–t₂): In this mode, all power transistors Q₁–Q₆ are turned off, as shown in Figure 3b. The energy of L₂ begins to be released and i_L2 linearly decreases; thus, the energy of L₂ is stored in the energy storage device and C_s through the body diode of Q₃ and i_r1 is freewheeling through the body diodes of Q₂ and Q₃ at the same time. When the voltage of the parasitic capacitor of Q₂ and Q₃ drops to zero, the body diodes of Q₂ and Q₃ provide a path for current freewheeling. Here, i_L1 still decreases and i_L2 begins to decrease. When Q₂, Q₃, and Q₆ are turned on, this operation enters the next mode.

(3)

Mode III (t₂–t₃): As shown in Figure 3c, Q₂ and Q₆ are turned on and achieve zero-voltage switching in this mode. The energy of inductor L₂ is charged to the energy storage device and C_s by turning on Q₆ and the body diodes of Q₃, and the energy of the master source is stored in inductor L₁ by turning on Q₂. Therefore, i_L1 and i_L2 linearly increase and decrease, respectively. Since the resonant voltage v_r1 of the primary side becomes a negative square wave in this time, the current i_r1 becomes negative and flows through Q₂ and Q₃. i_r2 also becomes negative and flows through the body diode of Q₈ and Q₉ and transfers the energy to DC BUS with output capacitor C_D. When Q₂ and Q₆ are turned off, this operation enters the next mode.

(4)

Mode IV (t₃–t₄): In this mode, all power transistors Q₁–Q₆ are turned off again, as shown in Figure 3d; the energy of L₁ begins to be released; and i_L1 linearly decreases. Thus, the energy of L₁ is stored in the energy storage device and C_s through the body diode of Q₁. Meanwhile, i_r1 is freewheeling through the body diodes of Q₁ and Q₄. When the voltage of the parasitic capacitor of Q₂ and Q₃ drops to zero, the body diodes of Q₁ and Q₄ provide a path for current freewheeling. Here, i_L1 begins to decrease and i_L2 continues to decrease. When Q₄ and Q₅ are turned on again and the circuit returns to Mode I, the aforementioned mode of operation continues.

When the master source cannot normally provide the energy, the slave source provides the energy for the proposed system. Next, the circuit operations will be analyzed for the slave source supply operation, as shown in Figure 4 and Figure 6b, with this operation described as follows:

(1)

Mode I (t₀–t₁): As shown in Figure 4a, Q₂ and Q₃ are turned off, Q₁ and Q₄ are turned on, and the zero-voltage switching of Q₁ and Q₄ is achieved in this mode. Because Q₅ and Q₆ are always turned off in this mode, if there are surplus energies in these inductors (L₁ and L₂), i_L1 and i_L2 must be freewheeling through D₁ and D₂ until both i_L1 and i_L2 linearly decrease to zero. Since v_r1 is a positive square wave, i_r1 is also positive and flows through Q₁ and Q₄. The current i_r2 flows through the body diode of Q₇ and Q₁₀, and the energy is transferred to DC BUS with output capacitor C_D. This operation enters the next mode when Q₁ and Q₄ are turned off.

(2)

Mode II (t₁–t₂): As shown in Figure 4b, all power transistors Q₁–Q₆ are turned off. When the voltage of the parasitic capacitor of Q₂ and Q₃ drops to zero, the body diodes of Q₂ and Q₃ provide a path for current freewheeling. i_r1 is still positive and freewheeling through the body diodes of Q₂ and Q₃. i_L1 and i_L2 are still freewheeling through D₁ and D₂ if there are surplus energies in L₁ and L₂. When Q₂ and Q₃ are turned on, this operation enters the next mode.

(3)

Mode III (t₂–t₃): As shown in Figure 4c, because the voltage of Q₂ and Q₃ drops to zero in the last mode, Q₂ and Q₃ are turned on in this mode and zero-voltage switching of Q₂ and Q₃ is achieved. Since v_r1 becomes a negative square wave at this time, i_r1 becomes negative and flows through Q₂ and Q₃; i_r2 also becomes negative and flows through the body diode of Q₈ and Q₉ and transfers the energy to DC BUS with output capacitor C_D. Because no load can consume the surplus energies of L₁ and L₂, both i_L1 and i_L2 linearly decrease slowly through D₁ and D₂. When Q₂ and Q₃ are turned off, this operation enters the next mode.

(4)

Mode IV (t₃–t₄): In this mode, all power transistors Q₁–Q₆ are turned off again, as shown in Figure 4d. When the voltage of parasitic capacitors of Q₁ and Q₄ drops to zero, the body diodes of Q₁ and Q₄ provide a path for current freewheeling. Therefore, i_r1 is freewheeling through body diodes of Q₁ and Q₄ at this time. In addition, because no load can consume the surplus energies of L₁ and L₂, both i_L1 and i_L2 linearly decrease slowly through D₁ and D₂. When Q₁ and Q₄ are turned on again and the circuit returns to Mode I, the aforementioned modes of operation continue.

When the original source directly connects and provides the energy for DC BUS, the energy will be recovered from DC BUS to the energy storage device through the proposed system. Finally, the circuit operations will be analyzed for the energy is transferred from DC BUS, as shown in Figure 5 and Figure 6c. Therefore, this operation is described as follows:

(1)

Mode I (t₀–t₁): As shown in Figure 5a, because the voltage of Q₇ and Q₁₀ drops to zero in the last mode, Q₇ and Q₁₀ are turned on and achieve the zero-voltage switching in this mode by turning off Q₈ and Q₉. The resonant voltage v_r2 is a positive square wave because this voltage is converted by the switching of Q₇ and Q₁₀. The resonant current i_r2 is positive and flowing through Q₇ and Q₁₀, and the resonant current i_r1 that is induced by transformer T is also positive and flows through the body diode of Q₁ and Q₄ to transfer the energy to the energy storage device with capacitor C_s. This operation enters the next mode until Q₇ and Q₁₀ are turned off.

(2)

Mode II (t₁–t₂): As shown in Figure 5b, all power transistors Q₇–Q₁₀ are turned off because the voltage of the parasitic capacitor of Q₈ and Q₉ drops to zero. i_r1 is freewheeling through the body diode of Q₈ and Q₉. When Q₈ and Q₉ are turned on, this operation enters the next mode.

(3)

Mode III (t₂–t₃): As shown in Figure 5c, because the voltage of Q₈ and Q₉ drops to zero in the previous mode, Q₈ and Q₉ are turned on in this mode and the zero-voltage switching of Q₈ and Q₉ is achieved. The resonant voltage v_r2 becomes a negative square wave because this voltage is converted by switching Q₈ and Q₉. The resonant current i_r2 is positive and flows through Q₈ and Q₉, and the resonant current i_r1 that is induced by transformer T is also positive and flows through the body diode of Q₂ and Q₃ to transfer the energy to the energy storage device with capacitor C_s. This operation enters the next mode until Q₈ and Q₉ are turned off.

(4)

Mode IV (t₃–t₄): In this mode, all power transistors Q₇–Q₁₀ are turned off, as shown in Figure 5d, because the voltage of the parasitic capacitor of Q₇ and Q₁₀ drops to zero. i_r1 is freewheeling through the body diode of Q₇ and Q₁₀. When Q₇ and Q₁₀ are turned on again, the circuit returns to Mode I for the aforementioned cycle to continue.

3. Symmetrical Resonant Circuit Analysis

To enhance the transmission efficiency, the resonant circuit and the transformer are usually applied to the power conversion. However, resonant characteristics are easily affected because the power transmission is not unidirectional. The symmetrical design of the resonant tank must be taken into consideration for reducing the variation in resonant characteristics in the different power transmission directions. The symmetrical resonant tank analysis and transformer design are described in detail in this section.

3.1. Operation Mode and Circuit Analysis

Figure 7 depicts the configuration of the symmetrical resonant tank, where L_r1 and L_r2 are the resonant inductors, C_r1 and C_r2 are the resonant capacitors, T is the transformer, and L_m is the magnetic inductor of the transformer. v_r1 is the resonant voltage of the primary side, which is generated by the front stage circuit. v_r2 is the resonant voltage of the secondary side, which is generated by the symmetrical resonant tank and input into the rear circuit. R_L is the equivalent load of DC BUS and V_o is the output voltage after rectification and stabilization by the rear stage circuit such as V_{DC BUS} of Figure 2. To select the resonant capacitor, the operation frequency f_s must be defined first, and the resonant inductor can be determined from Equation (5), as follows:

$$f_s=\frac{1}{2\pi \sqrt{L_{r1}{C}_{r1}}}=\frac{1}{2\pi \sqrt{L_{r2}{C}_{r2}}}$$

(5)

Due to the symmetrical structure of this resonant tank, the resonant parameters of both sides can be determined from Equation (5). To ensure the symmetrical configuration of the resonant tank, the resonant parameters of the primary side L_r1 and C_r1 are determined first and the resonant parameters of the secondary side L_r2 and C_r2 are calculated by using Equation (6), as follows:

$$C_{r2}=\frac{C_{r1}}{n^2},L_{r2}=L_{r1}{n}^2$$

(6)

where n is the turn ratio of the transformer. When all resonant parameters are selected, the resonant characteristics must be analyzed for the different operations. The process will be explored in the next section.

3.2. Analysis of Voltage Gain and Impedance Phase

3.2.1. The Energy Is Provided by the Master or the Slave Source

Next, the analysis of voltage gain and impedance phase is described when the energy is provided from the master source or the slave source, as shown in Figure 8. As shown in Figure 8b, the equivalent circuit of Figure 8a, to consider the variation of R_L, the impedance of the transformer is mapped to the primary side of the transformer, where L_r2/n², n²C_r2, and 8R_L/n²π² are the resonant parameters that mapped to the primary side from the secondary side, and these resonant parameters are in parallel connection with magnetic inductor L_m. This resonant equivalent impedance Z_{p_1} is calculated as

$$Z_{p\_1}\left(j\omega \right)=\frac{-\pi j{\omega }^3{L}_m{L}_{r2}{C}_{r2}-8{\omega }^2{L}_m{C}_{r2}{R}_L+n^4\pi j\omega L_m}{-n^2{\omega }^2\pi L_m{C}_{r2}-{\omega }^2\pi L_{r2}{C}_{r2}+8j\omega C_{r2}{R}_L+n^4\pi }$$

(7)

Consequently, the input impedance Z_{in_1} and the voltage gain G_v1 (jω) are derived below.

$$\begin{array}{cc}\hfill Z_{in\_1}\left(j\omega \right)& ={\left[n^4\pi j\omega C_{r1}\left(-{\omega }^2{L}_m{C}_{r2}-\frac{{\omega }^2{L}_{r2}{C}_{r2}}{n^4}+\frac{8j\omega C_{r2}{R}_L}{\pi n^4}+1\right)\right]}^{-1}\hfill \\ & \times \left[\pi {\omega }^2\left(-L_{r2}{C}_{r2}-n^2{L}_m{C}_{r2}-n^4{L}_m{C}_{r1}-n^4{L}_{r1}{C}_{r1}\right)\right.\hfill \\ & +8j\omega C_{r2}\left(-{\omega }^2{L}_m{C}_{r1}{R}_L-{\omega }^2{L}_{r1}{C}_{r1}{R}_L+\frac{n^4\pi }{8j\omega C_{r2}}+8R_L\right)\hfill \\ & \left.+\pi {\omega }^4{C}_{r1}{C}_{r2}\left(L_m{L}_{r2}+L_{r1}{L}_{r2}+n^2{L}_m{L}_{r1}\right)\right]\hfill \end{array}$$

(8)

$$\left|{\mathrm{G}}_{v1}(j\omega )\right|=\frac{\left|V_o\right|}{\left|v_{r1}(j\omega )\right|}=\frac{2j\omega L_m{R}_L{\left(j\omega L_m+\frac{8R_L}{n^2\pi }+\frac{n^2}{j\omega C_{r2}}+\frac{j\omega L_{r2}}{n^2}\right)}^{-1}}{n\left(j\omega L_{r1}+\frac{1}{j\omega C_{r1}}+\frac{\frac{8j\omega L_m{R}_L}{n^2\pi }+\frac{n^2{L}_m}{C_{r2}}-\frac{{\omega }^2{L}_{r2}{L}_m}{n^2}}{j\omega L_m+\frac{8R_L}{n^2\pi }+\frac{n^2}{j\omega C_{r2}}+\frac{j\omega L_{r2}}{n^2}}\right)}$$

(9)

Based on Equations (8) and (9), Figure 9 shows the curves of voltage gain and input impedance under different load conditions. When load variation occurs, the system frequency is adjusted to maintain the voltage gain, as shown in Figure 9a. Since the voltage gain is larger than unity, the operating frequency must be adjusted to meet the requirements of the system output. Then, as depicted in Figure 9b, where the phase of input impedance is larger than zero on the right-hand side of resonant frequency, it means that the condition of zero-voltage switching is achieved. Therefore, the operating frequency will be larger than the resonant frequency (f_n > 1).

3.2.2. The Energy Is Provided from DC BUS

In this section, the study discusses the scenario in which the energy is provided by DC BUS. As seen from Figure 10a, which presents the equivalent circuit of energy reversal, the residual load power of DC BUS is first converted to an AC squared signal and then delivered to the resonant circuit. As shown in Figure 10b, to consider the variation in R_s, the impedance of the transformer is mapped to the primary side of the transformer, where C_r1/n², n²L_r1, and 8R_s/n²π² are the resonant parameters that mapped to the secondary side from the primary side, and these resonant parameters are in parallel connection with magnetic inductor L_m. This resonant equivalent impedance Z_{p_2} is calculated as

$$Z_{p\_2}\left(j\omega \right)=\frac{\left(-n^4{\omega }^2{L}_{r1}{C}_{r1}+n^{4}8j\omega C_{r1}{R}_s+1\right)n^2\pi j\omega L_m}{-n^4{\omega }^2\pi L_{r1}{C}_{r1}-n^4{\omega }^2\pi L_m{C}_{r1}+8n^{4}j\omega C_{r1}{R}_s+\pi }$$

(10)

Given Equation (10), the transfer function of voltage gain G_v2(jω) and input impedance Z_{in_2}(jω) when the energy is provided from DC BUS are then derived as follows:

$$\begin{array}{cc}\hfill Z_{in\_2}\left(j\omega \right)& ={\left[\left(-n^4\pi j{\omega }^3{C}_{r1}{C}_{r2}\right)\left(L_{r1}+L_m\right)+\left(j\omega C_{r2}\right)\left(\pi +8n^{4}j\omega C_{r1}{R}_s\right)\right]}^{-1}\hfill \\ & \times \left\{\pi +\left({\omega }^4{n}^4\pi C_{r1}{C}_{r2}\right)\left(L_m{L}_{r2}+n^2{L}_m{L}_{r1}+L_{r1}{L}_{r2}\right)\right.\hfill \\ & -{\omega }^2\left[\pi C_{r2}\left(L_{r2}+L_m\right)+n^4\pi C_{r1}\left(L_{r1}+L_m\right)\right.\hfill \\ & \left.\left.+8j\omega C_{r1}{C}_{r2}{R}_s\left(n^6{L}_m+L_{r2}\right)\right]+8n^{4}j\omega C_{r1}{R}_s\right\}\hfill \end{array}$$

(11)

$$\left|{\mathrm{G}}_{v2}(j\omega )\right|=\frac{\left|V_s\right|}{\left|v_{r2}(j\omega )\right|}=\frac{2j\omega L_m{R}_s{\left(n^{2}j\omega L_m+\frac{8R_s}{n^2\pi }+\frac{n^{-2}}{j\omega C_{r1}}+n^{2}j\omega L_{r1}\right)}^{-1}}{j\omega L_{r2}+\frac{1}{j\omega C_{r2}}+\frac{\frac{8j\omega L_m{R}_s}{\pi }+\frac{L_m}{C_{r1}}-n^4{\omega }^2{L}_{r1}{L}_m}{n^{2}j\omega L_m+\frac{8R_s}{n^2\pi }+\frac{n^{-2}}{j\omega C_{r1}}+n^{2}j\omega L_{r1}}}$$

(12)

Next, Figure 11 shows the simulation results of transfer gain G_v2(jω) and input impedance Z_{in_2}(jω) under load variations. In Figure 11a, since the voltage gain is lower than 1, this means that the voltage steps down in this operation mode. Figure 11b shows the curve of the phase angle for input impedance. To achieve zero-voltage switching and considering that the intended resonant property requires inductance, the operating frequency must be larger than the resonant frequency (f_n > 1). Let the phase angle of impedance be higher than 0 degrees. When the parameters of the resonant tank are selected based on the above consideration, the parameters of the resonant tank of the proposed system are shown in

Table 3. Parameters of the symmetrical resonant circuit.

Items	Values
Resonant frequency	55 kHz
Resonant inductor of the primary side (Lr1)	124.8 uH
Resonant capacitor of the primary side (Cr1)	67.1 nF
Magnetic inductor (Lm)	2.75 mH
Resonant inductor of the secondary side (Lr2)	2.034 mH
Resonant capacitor of the secondary side (Cr2)	4.12 nF

.

4. Control Mechanism Analysis

Figure 12 shows the control flowchart of the proposed system, where three modes of power deliveries are studied. This flowchart starts with the initialization of frequency and each voltage reference setting. The controller begins with the acquisition of the voltage of DC BUS (V_{DC BUS}) from the feedback circuit. On comparing V_{DC BUS} with the setting voltage of V_{DC BUS} (V_{DC BUS-ref}), if V_{DC BUS} is lower than or equal to V_{DC BUS-ref}, then the process goes to voltage detection of the master source (V_m). If V_m is larger than or equal to V_m-min, it means that the energy of the master source is sufficient. Yet, if V_{DC BUS} is still lower than V_{DC BUS-ref}, it implies that the energy is insufficient for the ozone-driven system. Next, the controller decreases frequencies of Q₂, Q₄, Q₅, and Q₆ (f_Q2, f_Q4, f_Q5, and f_Q6) to raise the voltage gain and detects repeatedly V_{DC BUS} until V_{DC BUS} is larger than or equal to V_{DC BUS-ref}. Subsequently, the frequencies of Q₂, Q₄, Q₅, and Q₆ (f_Q2, f_Q4, f_Q5, and f_Q6) are increased to reduce the voltage gain. If V_{DC BUS} is lower than or equal to V_{DC BUS-ref}, the process is back to detecting V_m. Conversely, the operation mode is changed. This operation mode is called the master source supply mode in the flowchart in Figure 12.

Following the above operation, if V_m is smaller than V_m-min, it means that the energy of the master source is insufficient and the voltage of the slave source V_s must be detected at this time. If V_s is larger than or equal to V_s-min, it means that the energy of the slave source still needs to be replenished because V_{DC BUS} is still lower than V_{DC BUS-ref}. As a result, the controller decreases the frequencies of Q₁–Q₄ (f_Q1–f_Q4) to raise the voltage gain and then V_{DC BUS} is detected repeatedly until V_{DC BUS} is larger than or equal to V_{DC BUS-ref}. The controller then increases the frequencies of Q₁–Q₄ (f_Q1–f_Q4) to reduce the voltage gain and detects V_m again. If V_m is still lower than V_m-min, the same operation will continue. Conversely, the operation mode is back to the master source supply mode. However, if V_s is also smaller than V_s-min, it means that the master source and the slave source will both fail to provide the energy and then the process goes to stop all operations. This operation mode is named the slave supply mode, as shown in Figure 12.

Finally, when the controller begins with the acquisition of V_{DC BUS} from the feedback circuit and V_{DC BUS} is larger than V_{DC BUS-ref}, it implies that the energy of DC BUS is generated by the original source that connects directly to DC BUS. Then, the controller detects V_s. If V_s is less than V_s-max, this means that the energy can be charged to the slave source (energy storage device). Next, the controller decreases frequencies of Q₇–Q₁₀ (f_Q7–f_Q10) to raise the voltage gain and detects V_s repeatedly until V_s is larger than or equal to V_s-max. The controller increases the frequencies of Q₇–Q₁₀ (f_Q7–f_Q10) to reduce the voltage gain and detect V_{DC BUS} again. If V_{DC BUS} is lower than or equal to V_{DC BUS-ref}, the process is back to detecting V_m. Conversely, the operation mode processes at the same mode. This operation mode is called the DC BUS supply mode in the flowchart in Figure 12.

5. Experimental Results

This section validates the feasibility and performance of a modified-bridge circuit with a master–slave input supply mechanism. Figure 13 shows the photograph of the hardware circuit. This proposed system can be separated into two parts. The first is the main circuit, which consists of a master–slave input supply circuit, a symmetrical resonant tank, a full-bridge circuit, a controller (MCU), and two auxiliary powers. The other is the ozone generator, which includes an ozone load, a driven circuit, an auxiliary power, and a controller. It is noted that the terminal connectors for the DC BUS of the main circuit and the ozone generator are connected, allowing the energy to be exchanged between them.

To validate the capability of the modified-bridge circuit with a master–slave input supply mechanism, the voltages of the master power supply (V_m), the voltages of the energy storage device (V_s), the voltage and current of the resonant tank (v_r and i_r), and the output status (V_o and I_o) are discussed under different scenarios. Experimental results and corresponding waveforms are also analyzed below.

5.1. Input and Output Status

Here, the input and output validations start with different modes of testing, as shown in Figure 14. When the operation mode is the master source supply, it implies that the energy is provided solely from the master source. In Figure 14a, the voltage of the master source (V_m) is about 48 V and the voltage of the slave source (V_s) is about 30 V, which means that the voltage of the master source is ready to provide the energy and the voltage of the slave source is insufficient. In this operation mode, the voltage and the current of DC BUS (V_{DC BUS} and I_{DC BUS}) are about 400 V and 2 A, respectively. According to this waveform in Figure 14a, when the consumption of DC BUS reaches 2 A, the master source not only provides independently the energy for the ozone generator and stabilizes V_{DC BUS} but also charges the slave source through the proposed system. When the operation mode is the slave source supply, it means that the energy is provided from the slave source. In Figure 14b, V_m is about 30 V and V_s is about 50 V, which means that the master source cannot provide sufficient energy and the slave source must substitute for the master source. In this operation mode, when I_{DC BUS} is about 2 A and V_{DC BUS} is still maintained at about 400 V, it means that the slave source provides independently the energy for the ozone generator and it can stabilize V_{DC BUS} through the proposed system. Operating in the DC BUS supply mode implies that the original source directly connects to DC BUS to supply the energy. It is noted that the slave source is an energy storage device, so the energy from DC BUS charges the slave source. In Figure 14c, V_{DC BUS} and I_{DC BUS} are about 420 V and 2 A, respectively, and the waveform of I_{DC BUS} is negative, which means the energy is delivered from DC BUS because the original source directly connects to DC BUS. In addition, when V_{DC BUS} is higher than 400 V, the energy can be provided from DC BUS to the slave source through the proposed system. In this waveform of Figure 14c, V_s is about 40 V and V_m is 0 V, which means that the energy from DC BUS is only charged to the slave source and the master source cannot be charged, because the slave source is an energy storage device and the master source is a general power supply.

5.2. Resonant Property Verification

Here, the resonant property with the voltage and the current of the resonant circuit is verified under different operation modes, as shown in Figure 15. According to Figure 15a, the proposed system operates in the master source supply mode. The input voltage of the resonant circuit (v_r1) can be observed. v_r1 is seen to lead the input and the output current (i_r1 and i_r2) with a phase difference of θ_m, which means that the resonant property is exhibiting the inductive characteristics to help in the achievement of zero-voltage switching (ZVS). In Figure 15b, the proposed system is in the slave source supply mode and v_r1 still leads i_r1 and i_r2 with a phase difference of θ_s, which implies that the resonant property is exhibiting the inductive characteristics to help achieve ZVS. It is noted that though the front of the proposed system is an integrating circuit design, the power transistors of the front of the proposed system can still achieve the condition of ZVS. In Figure 15c, the DC BUS supply mode is operated in the proposed system and v_r2 also leads i_r1 and i_r2 with a phase difference of θ_D, which means that the inductive characteristic is exhibited in this resonant circuit even if the energy is provided from DC BUS for the output of the resonant circuit, which helps to achieve ZVS. Through the resonant property verification, the test results of the resonant property can meet the design of the resonant circuit, which can achieve the inductive property to help achieve ZVS irrespective of the supply mode under which the proposed system is operating.

5.3. Zero-Voltage Switching Verification

In this section, zero-voltage switching (ZVS) achievement is verified in the test results. Please refer to Figure 16 for an explanation. Here, the voltage and the current for the power transistors of the proposed system must be observed while power transistors turn on. In Figure 16a, when the operation mode is the master source supply, negative currents (i_ds2 and i_ds5) flow through the body diodes of Q₂ and Q₅ before the voltages (v_ds2 and v_ds5) become zero. When v_ds2 and v_ds5 become zero because Q₂ and Q₅ are turned on, i_ds2 and i_ds5 flow through Q₂ and Q₅, realizing zero-voltage switching. In Figure 16b, when the operation mode is the slave source supply, the currents (i_ds1 and i_ds2) are negative and flow through body diodes of Q₁ and Q₂ before Q₁ and Q₂ are turned on. Until the voltages (v_ds1 and v_ds2) become zero because Q₁ and Q₂ were turned on, i_ds1 and i_ds2 become positive and flow through Q₁ and Q₂ to achieve ZVS. In Figure 16c, the system is operating in the DC BUS supply mode and negative currents (i_ds7 and i_ds8) flow through body diodes of Q₇ and Q₈ before Q₇ and Q₈ are turned on. When the voltages (v_ds1 and v_ds2) become zero since Q₇ and Q₈ are turned on, i_ds7 and i_ds8 become positive and flow through Q₇ and Q₈ because of ZVS. These testing results show that the power transistors can be operated under ZVS regardless of the operation mode this proposed system is operating in, implying that the switching loss can be reduced effectively.

5.4. Operation Mode Change Verification

Since there are three operation modes in this proposed system and the operation mode changes according to the voltage of each port, the operation mode change must be verified in this section. In this experiment, DC power supplies are used in each port to simulate the master source, the slave source, and the original power source, which directly connects to DC BUS. The voltage of DC power is adjusted to simulate the voltage variety of each port, to observe the operation mode change situation of the proposed system as shown in Figure 17. Figure 17a,b shows the voltage variation in each port and the driving signal when the master source supply mode changes to the slave source supply mode. In Figure 17a, the voltage of the master source gradually drops to 40 V and the driving signal of Q₅ is turned off, which means that the energy of the master source is insufficient to provide energy and the driving signals of the corresponding power transistors (Q₅ and Q₆) must be stopped. After 20 ms, the driving signal of Q₁ is turned on and the voltage of the slave source drops a little bit, which means that the driving signals of the corresponding power transistors (Q₁ and Q₂) start to operate, letting the slave source substitute for the master source to provide the energy. In Figure 17b, there is a roughly 40 ms time difference between the voltage of the master source starting to gradually drop and the voltage of the slave source dropping a little bit. The voltage of DC BUS can be stabilized. Figure 17c,d shows the voltage variation in each port and the driving signal variation when the slave source supply mode switches to the master source supply mode. In Figure 17c, the voltage of the slave source drops to 36 V and the driving signal of Q₁ is turned off, which simulates the situation when the energy storage device (the slave source) is cut off and the driving signals of the corresponding power transistors (Q₁ and Q₂) are turned off. After 20 ms, the driving signal of Q₅ is turned on and the voltage of the master source drops a little bit, which means that the driving signals of corresponding power transistors (Q₅ and Q₆) start to operate and let the master source provide the energy. In Figure 17d, the voltage of DC BUS is still stable during operation mode change, which is about 60 ms. Finally, both the master source and the slave source cannot provide the energy and the energy is re-charged to the slave source (the energy storage device). As shown in Figure 17e, the voltage of DC BUS is stepped up to 420 V and the driving signal of Q₁ is turned off, which simulates that the original source directly connects to DC BUS and the energy will be provided through DC BUS from the original source. The driving signals of corresponding power transistors (Q₁ and Q₂) must be turned off. After 20 ms, the driving signal of Q₇ starts to operate and the voltage of the slave source is stepped up, which means that the driving signals of the corresponding power transistors (Q₇–Q₁₀) start to operate to provide the energy from DC BUS and re-charge the energy storage device (the slave source). Through these experimental results, it can be observed that the proposed system is functioning well regardless of the operation mode.

5.5. Comparison with the References

Finally, a comparison with the reference focuses on some indexes, such as input port amount, input voltage, used converter amount, supply coordination function, and output voltage in this section as listed in

Table 4. Comparison with the references of the ozone-driven circuit.

Paper No.	Input Port Amount	Used Circuit Amount	Supply Coordination Function	Input Voltage	Output Voltage
[11]	1	1	No	400 VDC	8.2 kV
[12]	1	1	No	208 VAC	NA
[13]	1	1	No	1200 VDC	1.6 kV
[14]	1	1	No	NA	10–30 kV
[15]	1	1	No	60 VDC	20 kV
[19]	5	6	No	30 VDC	1500 V
✓ The proposed system	2	1	Yes	Master: 48 VDC Slave: 40~54.6 VDC	400 V

. First, the proposed system has two input ports that connect the master source and the slave source (energy storage devices) by applying circuit integration design, which significantly improves the supply reliability. Other references have only one input port that connects a single source. If the source failure occurs, this kind of ozone-driven circuit will stop operating. It is noted that [15] has five input ports that connect five batteries to supply, yet six circuits are applied in this reference, which means complicated control and a higher circuit cost for this design. Furthermore, the supply coordination function must be discussed because the energy is provided by two sources in this paper and the proposed system has three operation modes to decide which one source delivers or re-charges, and other references have no supply coordination function because only a single source provides the energy. From the indexes of the input voltage and the output voltage, the proposed system can be extra-installed with the original ozone-driven system [11], which implies that the original system remains the same to maintain the ozone quality and minimize the system change to reduce the installation cost.

6. Conclusions

This paper presents the realization of a modified-bridge circuit with a master–slave input supply mechanism for ozone-driven system applications. Through the master–slave input supply circuit, the completed circuit exhibits better reliability during different operating conditions. In this study, a symmetrical resonant circuit that achieves zero-switching conditions under each operation mode is designed. The hardware prototype is realized in the laboratory, and experiments are executed under different scenarios. The experimental results validate the practicality of the proposed system, which should be most beneficial to improve the reliability of the ozone-driven system.