Dual Output and High Voltage Gain DC-DC Converter for PV and Fuel Cell Generators Connected to DC Bipolar Microgrids

This paper introduces a new topology for a DC-DC converter with bipolar output and high voltage gain. The topology was designed with the aim to use only one active power switch. Besides the bipolar multiport output and high voltage gain this converter has another important feature, namely, it has a continuous input current. Due to the self-balancing bipolar outputs, the proposed topology is suitable for bipolar DC microgrids. Indeed, the topology balancing capability can achieve the two symmetrical voltage poles of bipolar DC microgrids. Furthermore, it is possible to create a midpoint in the output of the converter that can be directly connected to the ground of the DC power supply, avoiding common-mode leakage currents in critical applications such as transformerless grid-connect PV systems. The operating principle of the proposed topology will be supported by mathematical analysis. To validate and verify the characteristics of the presented topology, several experimental results are shown.


I. INTRODUCTION
Energy conversion systems are extremely important for modern societies and are present in different forms in most domestic, commercial and industrial applications [1]. With the increasing challenges of CO 2 reduction and proliferation of solutions using clean and renewable resources (e.g. electrification of transport or photovoltaic systems), the importance and interest around power electronic converters has grown strongly over the last decades [2]. Numerous power converters have been proposed over the years to overcome different integration problems and, at the same time, improve different economic and operational aspects of such systems. Among all power converters, the DC-DC converter is currently one of the most important electronic circuit in the overall decarbonization process allowing to mitigate The associate editor coordinating the review of this manuscript and approving it for publication was Bidyadhar Subudhi . problems with integration of renewable energy sources (RES) [3], smart grids [4], integration and optimization of storage devices [5], electric vehicles [6], battery chargers [7] and many others. Some of the most desired characteristics of DC-DC converters are high-efficiency, cost effectiveness and the voltage regulation capability in order to allow a wide range of applicability. For certain applications, the step-up voltage of the classic DC-DC converters can be a simple and acceptable solution [8], [9]. However, for applications where a high step-up ratio is required other DC-DC converters have been reported in literature. In general DC-DC converters are classified into two main categories: isolated and non-isolated topologies [10]. Some isolated solutions are based on a single-switch and a high frequency transformer using either hard or soft-switching resonant or quasi-resonant circuits [10], [11]. Other isolated solutions using multiple switches, high frequency transformers, quasi-Z source topologies and soft switching techniques can be found in [12]- [14]. Other topologies are based on the boost full-bridge [15], [16], isolated push-pull boost [17] and isolated inductor-feed boost converter [18]. Using isolated converters, it is possible to achieve relatively high voltage regulation ratios and, in some cases, high efficiency by properly designing the transformer's turn ratio and the soft-switching circuits. The costs associated with the high number of switching devices and transformers, in addition to more complicated control schemes and more isolated sensors are some of the main drawbacks of these isolated topologies. Some existing isolated voltage-type converters also present pulsed input current, which negatively impacts the life of PV arrays. Several non-isolated topologies have also been proposed to extend the output to input voltage transfer ratio (voltage gain) of existing boost converters. Some advantages and applications of non-isolated topologies can be found in [9], [19]. Some topologies are based on the integration of different DC-DC converters and use a single switch [20]- [22]. Other high step-up voltage converters are based on coupled inductors [23], [24]. These solutions represent a simple, effective, and promising alternative to achieve a wide range of voltage conversion. Other high step-up voltage converters are based on switched inductors and switched capacitors cells [25]- [28]. Some solutions to achieve high step-up voltage ratios are based on voltage-multiplier or voltage-lift techniques [29]- [31]. Quadratic boost converters and quasi-Z source topologies also provide another effective method to obtain high-step-up voltage gains [32]- [36]. These converters have usually less volume, simpler circuits and are more suitable for extreme voltage applications. Despite several advantages of non-isolated boost converters, some require multiple switching devices to achieve high-step-up voltage gains, which increase the cost and complexity. Some converters require only a single switching device but present limited voltage gain. In practical terms, due to large parasitic capacitors at the ground of the PV panels, the leakage current in the non-isolated PV grid-connected system is an important issue and a key problem. A review of non-isolated high step-up DC-DC converters for PV applications regarding these issues can be found in [37].
Very important applications of DC-DC converters can be found in the context of the DC microgrids. Main architectures of DC microgrids include the unipolar and bipolar wiring configurations. The bipolar architecture microgrid has two poles presenting symmetrical voltages regarding a ground pole and is becoming attractive due to several advantages such as, higher resilience, increased efficiency and power supply with higher quality [38]. However, bipolar microgrids can suffer from an important disadvantage, namely the bipolar voltage unbalance. Unbalances are due to the connection of loads with different power or characteristics to the two poles whose voltage variation departs from the voltage symmetry. To eliminate or attenuate this disadvantage solutions like the use of voltage balancers were proposed [39], [40]. However, this approach requires an extra power converter, increasing the cost and losses of the system. Better solutions require that the DC-DC generators balance the bipolar DC microgrid voltages. To achieve this, the use of converters with dual outputs, a common connection and symmetrical voltage outputs are needed [41]- [46]. However, these solutions normally require extra switches and/or have limited voltage gain.
In this paper a new DC-DC converter indicated to be used in bipolar DC microgrids is proposed hereafter. To be able to balance bipolar microgrids voltages the proposed topology outputs dual symmetrical voltages. Besides, the converter has also been designed to offer a very high voltage gain and continuous input current using only a single active switch. Therefore, the proposed converter is extremely suited for RES such as PV and fuel cell generators. Also, the proposed design allows a relatively reduced voltage stress across all power semiconductors of the DC-DC converter. The topology is also suitable for use in transformerless grid-connect solar PV and fuel cell generators. The design and analysis of the converter will be addressed in detail in the next sections. Some experimental results from a laboratory prototype are presented to confirm the theoretical validity of proposed converter.

II. DESCRIPTION OF THE PROPOSED DC-DC CONVERTER
To mitigate voltage unbalances in bipolar DC microgrids, DC-DC converters of RES applications, like PV and fuel cell generators, must be able to inject the power into the two poles of the microgrid as a way to balance the bipolar voltages. Thus, the proposed converter has dual outputs V Co1 and V Co2 illustrated in Fig. 1. Besides the important advantage of self-balancing the bipolar DC microgrid, the proposed converter has high voltage gain and continuous input current. The proposed converter power circuit needs only a single active switch (S w ), two inductors (L 1 and L 2 ), three capacitors (C 1 , C 2 and C 3 ) and four diodes (D 1 to D 4 ). From circuit analysis it is verified that the active switch does not withstand all the load voltage, given by the sum of the voltages across the output capacitors C o1 and C o2 .
The proposed circuit is also suited for applications like grid connected solar PV and fuel cell generators. One of the problems associated to the use of DC-DC converters in PV and fuel cell applications with transformerless inverter is the common mode leakage current. So, similar to previous case, the ground of the PV panel (or fuel cell) is directly connected to the neutral of the AC grid (green line of Fig. 2).

III. ANALYSIS AND OPERATING PRINCIPLE OF THE PROPOSED DC-DC CONVERTER
As mentioned before, the circuit allows for continuous input current and high voltage gain. This can be verified through the analysis of the circuit in steady state behaviour and considering operation in the continuous conduction mode (CCM). In accordance with these assumptions, there are two possible operating modes, each one associated to one equivalent circuit (Fig. 4). The main waveforms of the voltages and currents in this circuit that are associated to the two operating modes are shown in Fig. 3.
From the analysis of the circuit the behaviour of the converter for each operating mode is described as follows: During this interval the turn ON voltage is applied in the gate of the transistor, resulting on the equivalent circuit of Fig. 4 a). Thus, inductors L 1 and L 2 have a positive voltage (input voltage source plus C 2 capacitor voltage and C 1 capacitor voltage) increasing their currents. The current in inductor L 1 will flow through capacitor C 1 (discharging it), transistor S W and voltage source V i . The current in the inductor L 2 will flow through transistor SW and capacitor C2 (discharging it). The discharge of the capacitor C 3 depends of the output buses. Thus, if the voltage in capacitor C o1 is lower than capacitor C o2 , since the voltage of capacitor C 3 is approximately equal to the capacitor C o1 then there will be no discharge of capacitor C 3 over the capacitor C o2 since diode D 4 is reverse biased. Otherwise, capacitor C 3 will discharge over C o2 flowing the current through those capacitors and transistor S W (see y in Fig. 4 a)).
From the analysis of this operating mode I, the following expressions are obtained by: In this time interval the transistor S W is turned OFF, originating the equivalent circuit presented in Fig. 4 b). The voltages across the inductors are now negative, forcing the currents in inductor L 1 and L 2 to decrease (discharge of the inductors). The current in the inductor L 1 will flow now through diode D 1 , capacitor C 2 (charging it) and voltage source V i . The current in the inductor L 2 will flow through capacitor C 1 (charging it) and diode D 1 . The charge of the capacitor C o1 depends on the output buses. Thus, if the voltage in capacitor C o1 is lower than capacitor C o2 then (since the voltage of capacitor C 3 is approximately equal to the capacitor C o2 ) there will be no charge of capacitor C 3 since diode D 2 is reverse biased. In this condition there will be a current flowing through this capacitor charging it (x in Fig. 4 b)). Otherwise, there will be a current flowing through capacitor C 3 charging it (y in Fig. 4 b)). In this operating mode the following expressions are obtained by: The voltage gain of this power converter topology can be determined from the analysis of the described two operating modes. Taking into consideration these modes and that the average value of voltages of the two inductors in steady-state are equal to zero, it is possible to obtain: Also considering that the capacitors C 1 and C 2 are sufficiently large, it can be assumed that the voltages at their terminals will change linearly, as well as, their average values, given by: Using these last two equations in equations (3), it will be derived the equations that are described by: From these equations and considering δ = (t 1 /T ) the duty cycle associated to the switch S w , the average voltages of V C1 and V C2 are obtained by: Using these equations with (1) and (2), the average value of the output capacitor voltages (or pole voltages) function of the input voltage will be determined by: From the voltage gain given by this last equation it is foreseen that the proposed topology extends the voltage gain to the double of the conventional quasi-Z DC-DC converter, or obtains two outputs with the same gain.

IV. DESIGN OF THE POWER CONVERTER COMPONENTS
To determine the power semiconductors ratings, it will be considered again the converter operating in CCM with no losses. Thus, in steady state operation the next relationship is given: where the current variables Ii and Io are the average value of the input voltage source and output load current. Considering the relationship of the last expression and the gain voltage of the converter given by (7), the average value of the input current source is a function of the duty cycle and output load current, being given by equation (9). This expression also gives the average value of the inductor currents of the power converter.
For the design of the inductors and capacitors, it will be considered that they depend of the rated voltages and currents and that they ensure small ripple for the current in the inductors and voltage across the capacitors. Thus, considering the differential equation that represents the evolution of the current in the inductors (10) and the voltage across the inductor almost constant, the linear solution (11) is obtained. From this last equation considering equation (10) and , the new relationship given by (12) is obtained.
To size the capacitors, a similar approach is used. Thus, considering the differential equation that mathematically represents the behavior of the capacitors voltage and the currents through the capacitors constant during the operation interval, the following relationship is obtained: VOLUME 9, 2021 From the previous relationship, the capacitors depend on average currents of the capacitors and the time interval (t 1 -T): The determination of the power semiconductors current stress can be done considering the average values of the currents in inductors and from the condition that the average values of the currents in each of the capacitors are zero in a switching cycle (T ), giving: These equations show that the current in the switch is higher than the load current, as expected since the switch voltage stress is lower than the load voltage.
To determine the voltages stress across the semiconductors, it is considered that the voltage ripple across the capacitors can be neglected. The voltages in the power semiconductors are dependent on the capacitors voltages as written in (17). These expressions show that the voltage stresses of the power semiconductors are only half of the converter output voltage.

V. EFFECTS OF NON-IDEALITIES OF THE INDUCTORS
The non-idealities of the circuit components will affect the voltage gain of the power converter. The elements that have more impact on that are the inductors. To verify that impact, the circuit will be analyzed considering that the losses in those inductors can be represented by an equivalent resistance. In this way, associated to those inductors will be considered that the resistances of r L1 and r L2 are in serial connection with the inductors L 1 and L 2 , respectively. For the analysis of the power circuit in non-ideal conditions with the previous assumptions, the load was considered as an equivalent resistance R 0 . Thus, with all these considerations it is obtained the static voltage gain of the converter, expressed by: To verify the impact of the inductors and load resistances, it is shown in Fig. 5 the voltage gain of the converter as a function of the duty cycle for different values of r L1 / R 0 = r L2 / R 0 . From Fig. 5 it is seen that the voltage gain is limited. This limitation is function of the resistors values. Considering for example r L1 / R 0 = r L2 / R 0 = 0.001, the maximum voltage gain will be around 11, while if it is considered r L1 / R 0 = r L2 / R 0 = 0.005, the maximum voltage gain is reduced to 5.

VI. COMPARISON WITH EXISTING SIMILAR CONVERTERS
To analyze the characteristics of the proposed converter, a comparison with other topologies was also performed. For this comparison it was considered equivalent topologies that could be connected to a bipolar microgrid (with the possibility to connect to a positive and negative pole). Table 1 presents a comparison between the characteristics of the proposed solution and similar topologies available on literature. The proposed converter (right column) presents the highest static voltage gain. It is also one of the topologies that require minimal number of switches. Regarding the passive components, it is the second topology that requires a smaller number of inductors (only topology [46] requires less). It also provides common ground between the neutral pole of the microgrid and the ground of the input voltage source. The proposed solution was developed to operate in hard switching which could contribute to increasing the power losses. However, the proposed topology can also be extended to a soft switching solution.

VII. SIMULATION AND EXPERIMENTAL TESTS
The proposed DC-DC converter based on a quasi-Z topology with extended voltage gain and dual output characteristic was  subject to simulation and experimental tests. For the experimental tests it was developed a prototype with demonstrating purposes regarding their operating principle and confirmation of their features. The simulation and experimental tests were made in accordance with Fig. 1, in which inductors L 1 and L 2 have 1,1 mH, and capacitors of 100 µF, 68 µF and 68 µF are used for C 1 , C 2 and C 3 respectively. For the output capacitors C 01 and C 02 have a capacitance of 470 µF. The load connected to the power converter consists in a 540 resistor. The power converter is supplied by a DC power source with 60 V and was operated with a switching frequency of 20 kHz. The duty cycle was adjusted to obtain an output voltage of 400 V.

A. SIMULATION TESTS
The obtained simulation results for a duty cycle of 0.35 are presented in Figs. 6 to 8. Fig. 6 presents the voltage waveforms of the input (V i ) and outputs (V Co1 , V Co2 and V o ) of the proposed power converter. In accordance with the defined duty cycle, it is expected a theoretical gain of 6.7. Analyzing this figure is possible to verify that a similar gain is obtained. Besides that, it is possible to confirm that the voltages of the output capacitors (C o1 and C o2 ) have practically the same average value equal to half of the total output voltage. The voltages across the power semiconductors can be seen in Fig. 7 where the waveforms that were obtained through this simulation test are presented. The analysis of these waveforms confirms that the voltage across semiconductors is lower than the output voltage. They are in accordance with the described in the theoretical analysis, being half of the total output voltage. The input current (also in the inductor L 1 ), inductor current L 2 and output current are presented in Fig. 8. From these waveforms obtained from the simulation test, it is possible to confirm the continuous input current of the power converter (CCM). Besides that, it can be seen that the current in inductor L 2 presents a similar behaviour of the one in inductor L 1 , ensuring also CCM. Regarding the output currents, i o1 and i o2 in Fig. 8, it is shown they are well balanced as they have the same average and ripple values.  Therefore, the proposed converter will contribute to balance the voltages of bipolar DC microgrids.
To test the capability of the proposed converter to inject the power in an asymmetric way into the two poles in an unbalanced situation, a new simulation with an unbalanced resistive load was also realized. For this simulation, it was initially considered the circuit with a balanced load (540 and 540 ). However, suddenly the resistor connected to the negative pole changed to 190 originating in this way an asymmetry in the circuit output. The obtained output voltages in this test can be seen in Fig. 9. This result shows that although the load became unbalanced, the output pole voltages remain balanced. Despite using only one switch, the output currents appear in this situation as unbalanced (Fig. 10). Moreover, the output current connected to the pole   with a higher load (i o2 ) has higher value contributing in this way to balance the microgrid voltages.

B. EXPERIMENTAL TESTS
The results obtained from the experimental tests realized through the developed prototype are presented in Figs. 11 to 13. As shown by Fig. 11 the output voltage was set for 400 V. This voltage was obtained for a duty cycle of 0.36. From this is possible to see that output/input voltage gain is near to the expected theoretical gain. This figure also shows the output capacitors voltages (V o1 and V o2 ) of the power converter, being possible to confirm that they are practically half of the total output voltage. The confirmation about the reduction of the voltages across the power devices when  compared with the output voltage of the converter can be seen in Fig. 12. From these voltages waveforms it is possible to see that the maximum voltage that they must hold-off is half of the total output voltage, confirming in this way the theoretical expectations.
The confirmation about the main currents in the circuit can be seen in Fig. 13. From the analysis of the waveforms of the experimental currents it is possible to see that the input current (also in the inductor L 1 ) presents a continuous conduction mode. The waveform of the experimental current in inductor L 2 also shows a similar behaviour with the one in inductor L 1 . This figure also shows that the output currents (i o1 and i o2 ) are balanced, like in the simulation. This confirms that in a situation of a balanced DC microgrid the proposed converter will not contribute to its unbalance.    To experimentally confirm the capability of the proposed converter to help in balancing the bipolar microgrid, a test with an unbalanced resistive load was also performed. For this test, an initially balanced load (with a resistor of 540 in each of the poles) was suddenly unbalanced by paralleling an extra load in the negative pole (to 190 ). Fig. 14 shows the input and output voltages of the converter. The waveforms of the positive and negative poles (V Co1 and V Co2 ) show that they maintain the same voltage value even when the load is unbalanced. This is ensured by the asymmetrical injection of the converter output currents (i o1 and i o2 ) as shown in Fig. 15. Indeed, the experimental waveforms of the converter output currents shows that when the load became unbalanced the currents also became unbalanced, since the output connected to the pole with higher load presents a higher value. This confirms that in an unbalanced situation the converter will inject in an asymmetric way the power into the two poles of the microgrid helping their balance, using only a single switch.
Despite quite similar, the dynamic response of the converter in the developed prototype regarding the output voltages and capacitor voltages is not so good as the simulation results. This is mainly due to parasitic elements introduced in the circuit design of the prototype.

VIII. CONCLUSION
This study proposed a novel DC-DC power converter topology that is especially suited for bipolar DC microgrids or transformerless grid-connect PV and fuel cell generators. The suitability to these applications is due to the fact that the topology is characterized by a double output, allowing connections to the positive, neutral and negative poles of bipolar microgrids. The proposed converter contributes to balance the voltages of the microgrid. The topology is also characterized by a high input-output voltage gain and continuous input current, which is very important for RES such as PV and fuel cell generators. The proposed DC-DC converter was developed with the purpose to maintain, as must as possible a reduced number of components, using only a single active switch. One of the features of the proposed DC-DC is the ability to create an output midpoint that is directly connected to the ground of the input DC power supply avoiding the common mode leakage current. The operating principles of the proposed DC-DC converter, as well, their component sizing were also presented. The assumptions, theoretical analysis and design were verified through the experimental results obtained from a laboratory setup.