Review of DC-AC converters for photovoltaic conversion chains

Received Jul 6, 2020 Revised Jan 11, 2021 Accepted Feb 22, 2021 This paper is devoted to the state of the art in photovoltaic (PV) conversion chains and their architectures. Two major characteristics are considered to classify these chains. These are the galvanic isolation and the number of stages; characteristics generally localized around the DC-AC converter (inverter) at the end of the PV conversion chain. Therefore, this paper deals with a comprehensive review of the different inverter topologies that can be integrated into PV conversion chains, distinguishing between the transformer based and the transformer-less conversion chains. The paper demonstrates that to this date, transformer-based inverters are very common and widely used, have a long record of accomplishment as a component of solar energy systems, in particular for residential applications because of their greater efficiency, smaller size, and lower costs. Nevertheless, transformer-less chains are touted with some problems and shortcomings. Moreover, solar energy storage devices, wireless charging systems in stations and along the highways require the re-examination of the existing solar PV conversion chains, their architectures and possibly new conversion chains suitable for all distributed generation including electric cars and storage devices.

f. DC to AC inverter is as important as the solar panels and they at the heart of domestic solar power systems, converting the DC to AC. Inverters have been experiencing continued development since late nineteenth century. In 2000 Sandia Laboratories invented the modern inverter paving the way to residential solar systems deployment g. Apart from the above developments, the widespread smart grids have been dictating additional requirements on solar PV inverters such as autonomy, adaptivity, cooperation, plug-and-play functions, communication, and self-awareness [4]. Such requirements are expected to affect the inverter topology and physical properties. However, these additional requirements can be addressed in a separate study. h. However, the global production of photovoltaic energy has so far remained low compared to the potential of this resource. This is mainly due to the high production cost of this energy, which is still dominated by that of fossil energy. As a result, the main challenge is finding low-cost and high-efficiency solutions for developing new photovoltaic power systems. i. Improving the efficiency of the photovoltaic generator by integrating new cell technologies such as organic cells or light-concentration cells. j. Improving the efficiency and cost of the entire system by setting up new architectures of photovoltaic conversion chains. Given the fact that the global market for solar PV grew about 44% in 2019 [5], the fact that the global total of 627 GW including on-and off-grid systems [5] and the fact that the solar solar efficiency has been witnessing marginal improving, other alternatives should be investigated to maximize the power from solar photovoltaic systems. In particular and it is without doubt that the architecture of the DC-AC conversion chains shall be evaluated, and it is in this context that comes the work developed in this paper.

PHOTOVOLTAIC CONVERSION CHAIN ARCHITECTURE
A solar photovoltaic conversion chain consists of several subsystems [6] that provide various functionalities as illustrated in Figure 1. Such functions include production, conversion, and interface. These functions are common in all photovoltaic conversion chains. Auxiliary functions include control, protection, and storage and differ from one chain to another. Grid connected conversion chains do not require storage batteries as the conversion chain is fed directly from the solar panels and produce alternating voltage and supports the customer load. Solar PV chains can be classified by different families according to the number of power processing stages, the use and of a transformer and otherwise the elimination a transformer. a. The galvanic isolation between the source and the grid consists of an isolation transformer designed at high frequencies (HF) on the source side via a high frequency of 1 kHz or higher. Alternatively, a transformer is placed on the grid side and operates at a low frequency (LF) of 50 to 60Hz. b. The number of stages in the chain. In one hand, the conversion of the energy is achieved straight from direct current (DC) the alternating current (AC). Otherwise, the conversion is done through a buck-boost DC-DC stage to regulate the DC voltage before the DC to AC conversion. c. Additionally, solar photovoltaic chains architecture is characterized by constraints such as efficiency improvement, the input power fluctuations reduction, production optimization during irradiation intermittency and the reliability improvement and extension of the lifetime of key components such as the electrolytic capacitors, inductors, etc. As a result, various topologies of photovoltaic conversion chains have been developed and reported in the literature. In the following, a special focus will be given to the topologies that have caught the attention of most researchers in the field of photovoltaic systems. This study focuses on the essential function of any chain which is the "conversion" function as well as the auxiliary functions such as protection, connection with the grid and others.

Galvanic conversion lines of high frequency type
Galvanic isolation is ensured by means of a compact HF transformer. However, the transformer exhibits extra losses penalizing therefore the efficiency of the conversion chain [7]- [10]. The diagram in Figure 2 shows the galvanic isolation principle. This type of isolation is based on the flyback approach that uses high value capacitors at the input stage. This approach tends to significantly reduce the entire conversion chain lifetime.
The input voltage, generally of the order of 12V, is converted into very high AC voltage (HF pulses of the order of 400V). This voltage is rectified and filtered to provide a high DC voltage that is applied to the input of the inverter to produce "consumable" AC voltages. The use of HF galvanic isolation applies to systems having few PV modules in series producing few hundred Watts. The current waveforms are quasisinusoidal due to the adopted control strategies.

Conversion chains with low-frequency galvanic isolation
The schematic diagram of such chains is shown in Figure 3. In this case, the DC input voltage is converted into AC voltage through the inverter. The LF transformer amplifies this voltage to standard values (110/220V, 50-60Hz). This type of isolation is the least used because of their relatively high prices, weight and size compared to conversion with HF transformers [7]- [10].

Non-isolated conversion chains
In this galvanic isolation chains, the non-isolated dc-dc converters are designed to step-up or stepdown the voltage. As a result, the size, the weight and the volume of the chains.are compact and the conversion efficiency is enhanced.

Non-isolated mono-stage conversion chains
In this case, the conversion is done in a single step without a voltage amplification stage. As shown in Figure 4, many photovoltaic modules are required to generate a voltage at the output of the inverter that is sufficiently high, close to that of the grid. The additional functions (MPPT and protection) are all integrated in a single stage of the conversion chain.

Non-isolated two-stage conversion chains
In this case of photovoltaic systems, the conversion chain consists of two cascaded stages. The first stage is a DC-DC converter to boost the DC voltage generated by the photovoltaic panel while the second stage is DC-AC type, allowing the generation of an AC voltage to interface the photovoltaic system with the grid as shown in Figure 5. In this case, the additional functions could be distributed between the DC-DC converter and the DC-AC converter.

Photovoltaic conversion chains based on string inverters
Nowadays, as per the "system technology for photovoltaics" [16], the string inverter configuration the most used system configuration. The schematic diagram of this configuration is given in Figure 7 [17], [18]. It involves connecting an inverter to each row of photovoltaic modules and paralleling the outputs of the different inverters.
The inverters need certain synchronization to avoid any exchange of power. Each inverter has its own MPPT which allows better control of the power. In addition, in case of failure of an inverter, only the row associated with it is isolated which allows continuity of service. Finally, such an arrangement offers a modularity that allows an easy chain extension.

Multistring system configuration
The efficiency and reliability of the conversion chain could be improved by connecting the DC-DC converter to each row or string [19], [20]. The converters' outputs are connected in parallel to the input of a central inverter as shown in Figure 8. This configuration combines the advantages of the central inverter configuration with those of the string inverter configuration. Indeed, the maximum power point trackers (MPPT) are integrated at the converter level, which makes it possible to overcome the losses due to the phenomenon of mismatch. This configuration is certainly more economical than the string configuration but cannot ensure the continuity of the service offered by the latter in case of failure of the central inverter.

Integrated system configuration
This category includes the following three configurations, namely [21]- [23], i) parallel converter configuration, ii) series converter configuration, and iii) microinverter configuration. The three configurations listed above are respectively shown in Figure 9, Figure 10 and Figure 11. As shown in Figure 9, each of the parallel or the parallel converter configuration shown in Figure 9, each converter is connected to a photovoltaic module having its own MPPT which improves the power generation management. In addition, this configuration allows a more precise control system and a quick response in the event of a fault.
However, this configuration requires high amplification gain of the voltage to match the required voltage at the input of the inverter, which significantly increases the losses and thus penalizes the efficiency of the entire conversion chain.
To reduce this gain, series converter configuration shown in Figure 10 is recommended. Indeed, this configuration allows access to the parallel converter configuration options except that the gain of amplification is reduced as the number of converters increases. The drawback of this configuration lies in connecting several converters which results in the dependence of the MPPT of different converters.
Another configuration considered in the literature is the microinverter system design shown in Figure 11. This configuration consists in connecting the inverter directly to the photovoltaic module via a voltage amplification stage. This design would reduce losses due to the mismatch phenomenon and would allow flexible extension through the "plug and play" technique.

INVERTER TOPOLOGIES OF PHOTOVOLTAIC CONVERSION CHAINS 4.1. Categories of inverters
The inverter is the major element of any photovoltaic conversion chain. Inverters could be classified into two categories, namely: a. "Line-switched" inverters which are usually of high powers. They are made of thyristors whose switching frequency is controlled from the line current. However, the thyristor requires forced switching circuits to bring the current to zero to turn them OFF. In addition, inverters made up of thyristors require complicated filtering circuits. For these reasons, this type of inverter is rarely used as reported in the [24]. b. "Self-commutated" inverters that are based on power transistors such as bipolar transistors, field effect transistors (MOSFETs) and insulated gate bipolar transistors (IGBTs). Those inverters are controlled using simple switching techniques implemented in analog or digital integrated circuits [25]. They could be voltage controlled called "voltage source inverter" (VSI), or current controlled called "current source inverter" (CSI). c. VSI inverters are the most used because a high-power factor could be achieved by a simple pulse width modulation (PWM). d. The use of CSI inverters is generally limited to medium and high-power applications. CSIs have a series of inductors at the DC bus to maintain constant current. These inverters have the following advantages [26], i) in the event of a short-circuit, the current remains limited within the inverter which, thanks to the inductances of the DC bus, cannot exceed the short-circuit current of the PV panel, and ii) the use of sensors is not necessary because the connection of the inverter to the grid does not use any control loop.

Inverter specifications
For all inverters categories, there are technical specifications that must be taken into consideration when developing a PV conversion system. Among these specifications, the total harmonic distortion (THD) and power factor (PF) are retained [27].
The "929-1988-IEEE Recommended Practice for Utility Interface of Residential and Intermediate Photovoltaic (PV) Systems (1987)" [28] has stated that the total harmonic distortion is defined as the ratio of the square root of the sum of the squares of the rms values of the analyzed signal to the rms value of its fundamental term. This value is always given in percent as indicated in (1).
where Hn is the rms value of the nth harmonic and H1 is the rms value of the fundamental term. According to US IEEE P929 [28], the THD of the currents must always be less than 5%. Table I shows the THD limits adopted by most standards [29]. The second specification is the power factor which is of dominant importance. A unit power factor is often required where it is associated with low values of current. This results in a reduction of losses within the chain and consequently a better profitability of the photovoltaic conversion.
The effective value of the current is expressed as (2).
Where In is the amplitude of the harmonic of order n. Similarly, the rms value of current could be expressed as a function of THD as (3).
In addition, the power factor PF is described in the "1547.1-2005 -IEEE Standard Conformance Test Procedures for Equipment Interconnecting Distributed Resources with Electric Power Systems (2011)" [30], [31] as (4).
with  is the back-phase shift of the current fundamental term with respect to the fundamental term of the voltage. Rewriting (4) while considering (3), makes it possible to express the power factor as a function of the THD as shown in (5). In addition to the two specifications described above, any photovoltaic system is characterized by further specifications related to security and coupling to the grid. These specifications are dictated by the following standards, a) IEC 60364-7-2005: Electrical installations of buildings section 712: photovoltaic power systems (2016) [30], b) IEEE 1547-1-2005: Standard IEEE conformance test procedures for equipment connecting distributed sources to electrical grids (2011) [29], c) IEEE 929-2000: Practical interfacing of photovoltaic systems (2000) [32], d)IEC 61727: Photovoltaic systems: characteristics of the interface (2004) [33], and e) DS / EN 61000-3-2: EMC, limits for harmonic emissions (Input current of equipment that can wait 16A per phase) (2019) [34]. Table 2 shows a comparison of forth mentioned standards in terms of current injected into the grid. The VDE 0126-1-1 standard is the only one that imposes a disconnection time from the grid in the order of 200 ms if a direct current greater than 1A is injected into the grid [35]- [37]. This standard also requires the disconnection of the conversion system if a leakage current occurs, as shown in Table 3 [38]. Considering the technical specifications of photovoltaic systems, several inverter topologies that more closely meet these specifications have been studied and are presented within this paper that mainly focuses on, a) inverters with galvanic isolation, and b) inverters without galvanic isolation.

Flyback mono-transistor inverter
The schematic diagram of this topology is shown in Figure 12. This is a galvanically isolated topology using a high frequency (HF) transformer. It is a low power inverter (around 100W) based on a single-transistor flyback converter whose outputs feed a mid-point transformer [39]- [42]. Both outputs of the transformer are connected to the grid through two diodes, two transistors and a filter. This configuration allows the Flyback converter to provide two currents of opposite signs. The two output transistors synchronously switch with the single transistor of the flyback converter in the high frequency regime. The Flyback converter operates in a discontinuous conduction mode (DCM) which implies that the current through the transformer reaches zero before the start of each switching cycle.
The major disadvantage of this topology lies in the fact that the decoupling of the power in parallel with the photovoltaic module is generally carried out using electrolytic capacitors of high values. Moreover, the two stages of this topology must be sized for a power equal to twice the nominal power. Finally, another drawback of this topology is the low power factor due to the existence of a zero-crossing distortion.

Flyback inverter with high power decoupling
This topology consists of the flyback inverter with a buck-boost converter [43]. The circuit of this topology is given in Figure 13. The topology is a modified version of the one presented in Figure 12 in order to avoid the use of high value electrolytic capacitors. Indeed, the decoupling capacitor is eliminated using a buck-boost converter. This is a topology where the transistor Spv, the diode Dpv, the primary winding of the transformer, the freewheeling diode of the transistor SDC and the capacitor CDC form the buck-boost converter operating in the discontinuous conduction mode.
Each cycle begins with the conduction of the transistor SPV which results in a linear increase of the magnetization current. When this current reaches a so-called reference value, the transistor SPV is blocked and the energy stored in the magnetization inductance is transferred to the capacitor CDC. The transistor SDC must be active when the current is being discharged to the capacitor which implies the state of zero voltage switching.
In addition, one of the two output transistors is simultaneously activated with the transistor SDC. Thus, the transistors on the secondary side of the inverter switch at high frequency regime unlike the topology of the Flyback inverter. The magnetization current continues to decrease until reaching a second reference. At this time, the transistor SDC will turn OFF and the energy stored in the inductor is transferred to the secondary side of the transformer and then to the grid. The diode DPV is used to eliminate the reverse voltage across the capacitor CPV when the energy is being transferred to the secondary side.

The modified "SHIMIZU" inverter
The previous topology shows voltage spikes at the leakage inductance included in the transformer [44]. Such voltage spikes exist between the terminals of the transistor SDC during its transition to the OFF state. Also, this topology has been improved by replacing the single-transistor converter with a bi-transistor converter as shown in Figure 14. Initially, the input energy of the inverter is converted using the boost converter and is saved in the intermediate capacitor. This energy is then converted using the Flyback converter. Finally, the energy stored in the magnetization inductance is returned to the grid through the transformer.

The isolated inverter with parallel-parallel configuration
The inverter made in the parallel-parallel configuration is shown in Figure 15. It delivers a power of the order of 200W [45]. It is an inverter that is designed to operate in the discontinuous conduction mode, DCM. This results in a discontinuous current in one of the inductances Lpv or L. The advantage of this topology lies in the fact that the discontinuous conduction mode is generally associated with: a) a simple control, b) a reduced size of the inductance LPV, and c) a current injected into the grid free of HF ripples/fluctuations.
At the output of the transformer, there are two conversion stages whose outputs are connected in quasi-parallel. One stage is responsible for producing the positive alternation of the current, the second stage is responsible for the negative alternation.

The bi-transistor flyback inverter
This inverter topology is shown in Figure 16. Its output power is around 160W. The operating principle is detailed in [46]. At the beginning of a period of the grid current, the transistors S1 and S4 (S2 and S3 for the negative half-period) are activated. This will generate a current and the energy produced will be saved in the magnetization inductance. When these two transistors are deactivated, the energy will be

Isolated flyback inverter in parallel-series connection
The topology is shown in Figure 17. It is built around two independent flyback converters [47]. In this case, the grid is connected in series with the outputs of the two converters while the photovoltaic module is connected in parallel with the inputs of the converters. Producing a power of 160W, the inverter operates in continuous conduction mode, CCM, which results in a more complicated control. It produces an output voltage of AC type by modulating two sinusoidal voltages shifted by 180 ° across the capacitors C1 and C2.

Low frequency inverter associated with a flyback converter
This inverter topology is shown in Figure 18. It is built around a Flyback converter connected to a thyristor-based inverter. It is dedicated to produce a power of about 150W. In some cases, this power is limited to 100W. The use of thyristors instead of power transistors can cause some switching problems as the thyristors are current controlled [48].

Inverter with series resonance converter
The topology of this inverter associated to a converter is shown in Figure 19. It is built around a series resonance converter connected to a high frequency galvanically isolated inverter. It is designed to produce a power of about 110W [49] and can reach 250W in some cases [50]. Moreover, the transformer leakage inductance and the capacitor connected in series form a resonance circuit that reduces the switching losses of the inverter.
The DC-DC converter switches at 100KHz with a duty cycle slightly less than 50% and it operates with a fixed conversion ratio, which makes it possible to overcome the power decoupling between the photovoltaic module and the grid. The losses within the converter are quite small while those of the transformer are considerable. The inverter uses two types of switching frequency: high frequency and low frequency. Indeed, the left arm of the inverter operates at frequencies between 20 and 80KHz while the right arm switches to 100Hz. In this case, the switching losses are halved compared to those of the inverters whose two arms switches at high frequencies. Figure 19. Inverter with series resonance converter

H-bridge inverter
This inverter is whose topology is shown in Figure 20. Made up of two transistor arms and having no transformer, this topology is satisfactory in variable speed drives for AC motors and in UPS. About control strategy, different PWM techniques could be implemented [51].
The H-bridge inverter is characterized by a low efficiency compared to other transformerless topologies due to the use of high frequency PWM control signals. The output voltage of the inverter exhibits significant fluctuations which are at the origin of a relatively large leakage current depending only on the value of the parasitic capacitances CPV-G that exist between the photovoltaic module and the ground. One solution for reducing the leakage current, in the case of bipolar PWM, is to use an LCL filter whose two inductors are located on either side of the load as shown in Figure 21. Such a filter makes it possible to solve the problem of leakage current for the bipolar PWM H-Bridge inverter provided that the two inductors, on either side of the load, are perfectly symmetrical. Although the problem of leakage current is solved, the energy efficiency remains low compared to the transformerless topologies.
The unipolar PWM strategy could also be applied to H-Bridge inverter topology. In this case, the voltage of the inverter takes three different values, namely: + VIN, 0 and -VIN. Also, the filter becomes simpler and the fluctuations of the output voltage are reduced. This strategy is characterized by a high efficiency. However, the leakage current is quite high so that the H-Bridge topology cannot be used for transformerless topologies.

The HERIC inverter topology
The HERIC topology that was firstly introduced by Sunways is shown in Figure 22 [52]. It is a transformerless inverter topology designed to combine the advantages of bipolar modulation (low leakage current) with those of unipolar modulation (high energy efficiency). The inverter is based on the conventional full bridge topology with two extras transistors interposed at the AC side, as shown in Figure 22.
These transistors are used during the freewheeling time. Indeed, during this phase, one of these transistors is activated while the four transistors of the H-bridge are all OFF and the grid is completely disconnected from the photovoltaic source. the grid current is short-circuited and saved in the load through the transistor S5 during the positive alternation and through S6 during its negative half-cycle. The fluctuations of the output current are eliminated, and the efficiency of the inverter is significantly improved.

The H5 inverter
SMA has also developed a transformerless inverter topology as shown in Figure 23. The idea is to disconnect the photovoltaic source from the grid during the freewheeling time of the four main transistors of the full bridge topology [53]. In this case, the modulation used is hybrid. Indeed, the transistors S1 and S3 switch at the frequency of the grid while S2 and S4 are switching at high frequencies (S1 and S4 for the positive half period and S2 and S3 for the negative half one).

Single phase inverter with dc decoupling
By still adopting the same idea of decoupling the inverter from the grid, another topology has been developed in [54] and is shown in Figure 24. The disconnection is carried out using the two extras transistors, S5 and S6, and the two extra diodes, D1 and D2, connected at the DC bus to the input of the inverter.
The output of the inverter is of the three-level type as in the case of the HERIC and H5 inverters. In addition, the level of the leakage current remains within the international standards. This cumulative potential makes this topology a good candidate for structures without transformers.
For this transformerless inverter topology, the modulation is hybrid. Indeed, the two transistors S5 and S6 switch at high frequency while the other transistors switch at the grid frequency (S1 and S4 for the positive half periods and S2 and S3 for the negative half periods).

Half bridge inverter
The inverter based on this topology is built using only two transistors as shown in Figure 25 [55]. For this topology, the two transistors switch at high frequency. As a result, the output of the inverter is of two-level type, which requires larger filtering elements. This topology has low leakage current, which makes it suitable for structures without transformers. However, the major disadvantage is that it requires a DC input voltage equal to twice the input voltages of the other topologies.
Indeed, for a single The NPC inverter topology is very often used for inverters without transformers. Its structure is shown in Figure 26 [56]- [58]. The phase and the neutral are clamped through two diodes at the mid-point of the two capacitors of the DC bus. The control strategy is hybrid while the transistor S2 being active during the positive half-cycle, and the transistor S1 is in the switching state. The current flows from the source to the load through the two switches S1, S2 and the diode D1. During the negative half cycle, the transistor S3 is kept in the active state while the transistor S4 is in the switched state.
The inverter based on this topology has some advantages such as the low switching losses rate, a reduced current ripple, an output voltage having three levels which reduces the size of the filter. In addition, the inverter operates with a power factor close to the unit which improves the efficiency of the complete system. However, this topology has certain limitations, particularly in the case of single-phase applications. Among these limitations, it is necessary to distinguish the high level of the DC bus voltage which should be doubled compared to that adopted for the other inverters. Moreover, this topology generates high transient voltages at the transistors S2 and S3 because, unlike the two transistors S1 and S4, these two transistors are not connected to the coupling capacitors.

Inverter with floatig inductor "Karshny"
The floating inductance inverter, "Karschny", belongs to the "Sitop Solar" family of Siemens [59]-[61]. The topology is presented in Figure 27. One can highlight a direct connection between the neutral of the output and the negative terminal of the photovoltaic module. such a connection helps to, a) eliminate the oscillations of the voltage, and b) use thin-film photovoltaic modules. The basic structure of this topology is built around a buck-boost stage with additional transistors to define the polarity of the output quantities. The major disadvantage of this topology lies in the fact that the current passes through several semiconductor components in its path which results in a reduction in efficiency and an increased inverter cost. Table 4 summarizes the differences among between transformerbased inverters and transformer-less inverters discussed in this paper.

CONCLUSION
Within this paper, a review of photovoltaic conversion chains architecture has been conducted. Two major characteristics have been retained to classify these chains, namely: the galvanic isolation and the number of conversion system stages; characteristics usually localized around the DC-AC inverter. For this reason, a special focus is granted to the different inverter topologies that can be integrated into photovoltaic conversion chains.
Apart from the number of stages, photovoltaic inverters are classified into two groups -inverters with transformers and inverters without transformers. In the first group, the transformer is of high frequency (HF) type and it is located at the DC side, thus making it possible to increase the voltage of the photovoltaic module. A low frequency type transformer is placed at the output side of the inverter to amplify the output voltage to the local standard voltage values of 110 V or 220 V. In both cases, inverters with HF or LF transformers have been reported to have additional losses of at least 2%. Moreover, transformers are bulky and big in size, especially the LF ones penalizing the compactness of the conversion chain.
The second group includes the inverter topologies without transformer. Indeed, these topologies offer advantages in terms of size, mass and efficiency. However, they suffer from certain shortcomings related to the absence of galvanic isolation. These disadvantages are essentially the existence of a resonance circuit made up of the filter, the photovoltaic module(s) and the parasitic capacitances that appear between the photovoltaic module and the ground. This resonance circuit generates a leakage current that can sometimes exceed certain standards set by international institutions.
As a result, various transformerless inverter topologies with acceptable leakage current level have been developed and presented in the literature. The most famous ones have been presented within this paper. Although these topologies present an acceptable level of performance, some improvements are required. Nonetheless, the transition from centralized structures to other distributed architecture has paved the way to develop new structures where the DC-DC power converter includes multi-string structures or DC optimizers used significantly in residential photovoltaic installations.