A Comparative Review on Single Phase Transformerless Inverter Topologies for Grid-Connected Photovoltaic Systems

Abstract

The uses of grid-connected photovoltaic (PV) inverters are increasing day by day due to the scarcity of fossil fuels such as coal and gas. On the other hand, due to their superior efficiency, lower cost, smaller size, and lighter weight when compared to inverters with transformers, transformerless inverters for low-voltage single-phase grid-tied photovoltaic (PV) systems have recently attracted more interest. However, there are some specific challenges, especially leakage current issues, which must be adequately addressed to ensure the safety standards of the grid codes. A variety of topologies has been presented in the literature to eliminate the leakage current using the decoupling or clamping technique. However, choosing an appropriate topology after comparison is challenging because each topology has a unique set of parameters. In this paper, the authors have selected a common set of parameters and simulated all the selected eighteen well-known topologies in MATLAB/Simulink to fairly analyze and compare their common mode characteristics and other output parameters. In addition, the power loss distribution of every switch of each of the eighteen topologies was calculated and presented to better understand the insight of the topologies.

1. Introduction

Photovoltaic (PV) solar panels are now the most promising source of renewable energy. This source is now extensively used around all over the world. Solar energy installation is expected to cross the terawatts level globally by 2022 [1]. Better performance, reduced expenses, and lower weight are just a few of the benefits of a solar PV system [2]. According to the 2021 annual report of the International Energy Agency-Photovoltaic Power Systems Program (IEA-PVPS) [3], China and the USA contributed 45.6% of the total installed PV capacity of 767.2 GW globally up to 2020. Solar PV system output power, on the other hand, is intermittent and heavily reliant on the irradiance and temperature of solar systems. In solar PV systems, a cascaded inverter can provide maximum power point tracking (MPPT) and control the dc-link voltage. A power inverter is used in a solar system to convert the dc power taken from PV modules into equivalent ac power for an ac load. The grid-tied inverter links PV modules to the electricity grid, and the magnitude of the inverter’s output voltage and frequency varies depending on grid conditions [4,5,6,7,8].

In comparison to grid-connected PV systems, the percentage of standalone PV power is relatively low [9]. The PV power can be integrated into the grid in two ways: using a transformer or without a transformer. When the transformer is used with an inverter, galvanic isolation is developed between the PV and the grid. However, the system is bulky due to the use of a transformer. The other technique is to use an inverter without a transformer, which is called a transformerless inverter. There are two ways to connect this transformerless inverter to the grid: one is a single-stage structure, and the other is a two-stage structure. The single-stage structure and two-stage structure are shown in Figure 1. Transformer-less grid-tied inverters have emerged as a sensible future industrial choice for solar power generating systems in the scale of low to medium due to a number of advantages such as reduced inverter cost, higher conversion efficiency, and smaller size/weight [10,11,12,13,14].

There are some challenges to it despite its many benefits. One of these is the leakage current that passes through the electrical grid and the PV panels’ parasitic capacitor [4,5,6,7,8,9], which results in grid current harmonics, electromagnetic interference, and device losses, all of which increase grid current ripple and can cause safety problems such as electric shock [15,16,17,18,19,20,21,22].The total harmonic distortion (THD), a steady state disturbance that affects the power quality in low voltage networks [23], is the other concern. THD has a detrimental impact on various grid-connected equipment, including the operation of motors and electronic devices [24]. Additionally, harmonic distortion lowers the efficiency and reliability of power networks [25].

There are some grid codes and standards provided by different concerned authorities that must be followed by any inverter connected with the grid [26,27,28,29,30,31,32,33,34,35]. As the leakage current issue is crucial due to the absence of transformer galvanic isolation, to reduce the leakage current, numerous topologies have been proposed. Two main techniques have been employed to address this issue. They are decoupling and clamping. The common mode behavior of transformerless inverters has been improved using DC or AC decoupling techniques in [36,37,38,39,40,41,42,43,44,45]. The mid-point clamping technique has been employed to stabilize the total common mode voltage in [46,47,48,49,50,51]. Using the solidly clamped topology, transformerless inverters’ common mode performance was enhanced in [52,53,54,55,56,57,58,59,60]. Keeping the value as constant as possible or lowering the total common mode voltage value are the two basic ways to reduce leakage current [61,62,63,64,65,66]. The majority of solidly clamped topologies [52,53,54,55,56,57,58,59,60] may keep the total common mode voltage at zero or, at the very least, constant values, which reduces leakage current. However, they require special arrangements for a solid connection between the negative terminals of the grid and the PV. The total common mode voltage can be clamped to a nearly constant value using the topologies based on mid-point clamping [46,47,48,49,50,51], but they cannot completely prevent the flow of leakage current on the zero voltage state. Zero state decoupled topologies [36,37,38,39,40,41,42,43,44,45] can prevent leakage current during the zero state, but they cannot maintain a constant total common mode voltage.

The overall analysis and comparison were made in some review papers [67,68,69,70,71] in different ways. In this paper, a critical review of all the topologies based on common mode behavior was performed. Eighteen well-known topologies based on the same parameters for fair comparison were simulated by the authors after being divided into three primary categories, and they were then compared and their results examined. In addition, the loss distribution across different switches, and the maximum efficiency of all the selected topologies were calculated for better understanding. The paper’s novel contribution is as follows:

• Comparing performance indicators based on simulation results using a common set of simulation parameters.
• Highlighting the switching loss and conduction loss of each switch in each topology using a common set of switches that depicts the advantages and disadvantages of each topology in terms of loss distribution.

The rest of the paper is organized as follows: Common mode behavior, grid requirements, and loss analysis related to transformerless inverters are discussed in Section 2. Section 3 includes the classification, analysis, and simulation results of common mode behavior of all the selected well-known topologies. Comparative analysis, discussion, and future trends are presented in Section 4. Finally, Section 5 concludes the paper.

2. Common Mode Behavior, Grid Requirements, and Loss Analysis

2.1. Common Mode Behavior of TransformerlessInverter

Figure 2 shows the parts of a single phase grid-connected PV system with a transformerless inverter. When a transformerless inverter is connected to the PV and the grid, the leakage current is its primary concern. The leakage current flowing path is identified in Figure 3. The output terminals, A and B, of the transformerless inverter generate the pulsed voltage source V_AN and V_BN, respectively. They are the main sources of leakage current that flows through the grid to the PV that is shown in Figure 4. The effect of filter inductance remains active, but the effect of the filter capacitor and grid may be neglected [72,73]. Then the final common mode model is shown in Figure 5. The total common mode voltage (V_tCM) is the final combined source of leakage current (I_LC) that flows through the filter inductance and parasitic capacitance (C_PV) of the PV.

The common mode voltage (V_CM) and differential mode voltage (V_DM) can be defined as [17,72,73,74,75,76,77,78,79,80,81]

$${\mathrm{V}}_{\mathrm{CM}}=\frac{{\mathrm{V}}_{\mathrm{AN}}+{\mathrm{V}}_{\mathrm{BN}}}{2}$$

(1)

V_DM = V_AN − V_BN

(2)

Rearranging (1) and (2), V_AN and V_BN can be expressed, respectively, in terms of V_CM and V_DM:

$${\mathrm{V}}_{\mathrm{AN}}={\mathrm{V}}_{\mathrm{CM}}+\frac{{\mathrm{V}}_{\mathrm{DM}}}{2}$$

(3)

$${\mathrm{V}}_{\mathrm{BN}}={\mathrm{V}}_{\mathrm{CM}}-\frac{{\mathrm{V}}_{\mathrm{DM}}}{2}$$

(4)

Now, if the two sources are converted into a single source of V_tCM, then the total common mode voltage is

$${\mathrm{V}}_{\mathrm{tCM}}={\mathrm{V}}_{\mathrm{CM}}+\frac{{\mathrm{V}}_{\mathrm{DM}}}{2}\frac{{\mathrm{L}}_{\mathrm{B}}-{\mathrm{L}}_{\mathrm{A}}}{{\mathrm{L}}_{\mathrm{A}}+{\mathrm{L}}_{\mathrm{B}}}$$

(5)

where, L_A and L_B are the filter inductors connected to the A and B terminals of the inverter, respectively.

If L_A = L_B, then Equation (5) becomes [49]

$${\mathrm{V}}_{\mathrm{tCM}}={\mathrm{V}}_{\mathrm{CM}}=\frac{{\mathrm{V}}_{\mathrm{AN}}+{\mathrm{V}}_{\mathrm{BN}}}{2}$$

(6)

However, if L_B = 0, then Equation (5) can be written as

$${\mathrm{V}}_{\mathrm{tCM}}={\mathrm{V}}_{\mathrm{CM}}-\frac{{\mathrm{V}}_{\mathrm{DM}}}{2}$$

(7)

In this paper, the total common mode voltage (V_tCM) was generated in the MATLAB/Simulink environment using Equations (6) and (7) considering the design of the filter.

2.2. Grid Requirements

Different types of grid codes and standards are recognized globally. In the case of transformerless inverters, the leakage current is the main concern. The standard called VDE 0126-1-1 [26] for the leakage current issue is shown in

Table 1. Leakage current values and their corresponding disconnection times listed in the VDE 0126-1-1 standard [26].

Leakage Current Value (mA)	Disconnection Time (s)
30	0.3
60	0.15
100	0.04

. From this standard, the PV system with the transformerless inverter must discontinue its service if the leakage current value of 100 mA can persist up to 0.04 s.

The other issues related to the grid-connected system, i.e., the upper limit of THD, the DC current injection limit, the grid frequency limit, and the power factor requirement covered by different standards originated by different countries, are shown in

Table 2. Grid codes and grid-connected PV system standards [19,27,28,30,31,32,33,34,35,82,83,84].

Standard Name and Origin	(THD Limit; Grid Frequency Limit in Hz; Power Factor Limit; DC Current Injection Limit)
IEEE 1547 [27], USA (IEEE)	(<5%; 57 to 60.5; 0.9–0.97; <0.5% of rated value of output current)
IEEE 929-2000 [30], USA (IEEE)	(<5%; 59.3 to 60.5;>0.85; <0.5% of rated value of output current)
IEC 61727 [82], Swiss (IEC)	(<5%; 49 to 51;>0.90; <1% of rated value of output current)
AS4777 [83,84,85], Australia	(<5%; 48 to 52; 0.8–0.95; <0.5% of rated output current/phase)
EN 61000-3-2 [32], England	(<5%; 47.5 to 50.2; NA; <0.22 A corresponding to a 50 W half wave rectifier)
EREC G83 [31], England	(<5%; 49 to 51; 0.95; 0.25% of AC current rating/phase)
VDE 4105 [28], Germany	(<5%; 47.5 to 51.5; 0.89–0.95; <1 A; maximum trip time 0.2 s)
BDEW [33], Germany	(<5%; 47.5 to 51.5(−5% to +3%); 0.95; NA)
GB/T 19964-2012 [34], China	(<5%; 48 to 50.5; 0.95; <1% of rated output current)
JEAC 9701-2012 [35], Japan	(<5%; 47.5 to 51.5 (Eastern Japan)/57 to 61.8 (Western Japan); 0.9–0.95; NA)

[19,27,28,30,31,32,33,34,35,82,83,84]. The THD limit of all standards is below 5%, but the DC current injection limit and the power factor requirement are not identical. The minimum DC current injection limit is either 0.25% of the rated output current or 0.22 ampere, corresponding to a 50 W half-wave rectifier. The power factor must be greater than or equal to 0.8 if the PV system can maintain at least a minimum standard.

2.3. Loss Analysis

In PV systems, switching loss and conduction loss are the two main types of losses. In practical cases, when an IGBT operates, a certain voltage called saturation voltage ($V_{CE\left(SAT\right)}$) drops across this switch. Hence the conduction loss across the IGBT can be calculated by [49,76,79,80,81]

$$P_{CONIGBT}=V_{CE\left(SAT\right)}{I}_{C}D$$

(8)

where, $I_C$ = on state current, and $D$ = duty cycle.

When a diode operates, a certain voltage called saturation voltage $V_F$ drops across this switch. Hence the conduction loss across the diode can be calculated by [49,76,79,80,81]

$$P_{CONDIODE}=V_F{I}_{F}D$$

(9)

where, $I_F$ = on state current, and $D$ = duty cycle.

In the case of unipolar SPWM, when the switch is operated on a switching frequency, and the reference signal is greater than the carrier signal, then the duty cycle is

$$D=M\sin \omega t$$

(10)

When the carrier signal is greater than the reference signal, then the duty cycle is

$$D=1-M\sin \omega t$$

(11)

where, $M$ = modulation index. Then total conduction loss is

$$P_{CONTOT}=P_{CONIGBT}+P_{CONDIODE}$$

(12)

There are two types of switching losses in IGBT, i.e., turn on switching loss and turn off switching loss. Switching losses can be calculated as follows [86]:

Turn-on switching loss:

$$S{W}_{ON}=E_{ON}f{V}_{DC}/V_{DCDATASHEET}$$

(13)

Turn-off switching loss:

$$S{W}_{OFF}=E_{OFF}f{V}_{DC}/V_{DCDATASHEET}$$

(14)

where, $E_{ON}$ and $E_{OFF}$ are the turn on and turn off energy losses, respectively, given in the datasheet, $V_{DC}$ is the actual DC voltage, and $V_{DCDATASHEET}$ is the voltage at which $E_{ON}$ and $E_{OFF}$ are calculated on the datasheet.

Now, the total switching loss of IGBT is

$$S{W}_{TOT}=S{W}_{ON}+S{W}_{OFF}$$

(15)

Finally, total loss ($TL$) in PV systems is

$$TL=S{W}_{TOT}+P_{CONTOT}$$

(16)

In this paper, for calculating the loss across each switch and efficiency, the IGBT called IKW30N65ET7 and the diode called IDC08D120T6M were used.

3. Classification and Analysis of Single Phase Transformerless Inverters

In the literature, numerous single-phase transformerless inverter topologies have been proposed. However, there are some topologies from the beginning to now that can be considered the major topologies in this area.

In this paper, eighteen major topologies were selected and classified into three categories. The classification of the topologies is shown in Figure 6. Then all the selected eighteen topologies were simulated on MATLAB/Simulink software based on the simulation parameters provided in

Table 3. Parameters for simulations and comparisons.

Parameters	Value
AC output voltage	230 V
Line frequency	50 Hz
AC output current	4.34 A
DC input voltage (except Conergy NPC topology)	400 V
DC input voltage (for Conergy NPC topology)	800 V
Output load	53 Ω
Rated power	998.2 W
Switching frequency	24 kHz
DC bus capacitor (Cdc = Cdc1/2 = Cdc2/2)	940 µF
Flying capacitor (CF = C1 = C2)	1880 µF
Filter capacitor (Co)	0.68 µF
Filter inductor (L1 = L2)	2 mH
Parasitic capacitor (CPV = CPV1 = CPV2)	75 nF
Modulation technique	Unipolar SPWM

for fair comparison and analysis. In this section, circuit structures for each topology are shown along with their benefits and drawbacks based on simulations of total common mode voltage (V_tCM) and leakage current (I_LC). The analysis also takes into account the authors’ calculated efficiency for each of the topologies they selected. For consistent comparison, IGBTs are treated as switches for all topologies.

3.1. Zero State Decoupled Topologies

In this category, the best performed six topologies were selected for analysis and comparison purposes. They applied the dc or ac decoupling technique at the zero state to improve the common mode behavior. They are as follows:

3.1.1. H5 Topology

The H5 topology is shown in Figure 7a. This topology was first proposed in [36]. Using advanced modulation techniques in [87] and the bidirectional clamping approach in [88], the reactive power control of this topology was further enhanced. The switching loss of the H5 topology was reduced using a soft switching strategy in [89] and new modulation techniques in [90]. In terms of filter settings and switching frequency, the H5 topology design was optimized in [91]. Moreover, the full bridge topology was modified to make the structure of this topology. One extra switch, S₅, was added for disconnecting the grid from the PV side in the zero state of voltage for the purpose of reducing leakage current. The upper two switches of the full bridge were operated on the line frequency, and the lower two switches on the switching frequency. Figure 7b shows the simulation results of this topology. The advantages of the topology were realized, namely, its higher efficiency, a lower rms value of leakage current, and improved THD in the simulation results. However, the simulation results showed that the peak value of the leakage current was around 200 mA, where the rms value was only 13.23 mA, and the total common mode voltage (V_tCM) oscillated from 0 V to 400 V.

3.1.2. ImprovedH6 Topology

An improved H6 topology was proposed in [37]. In Figure 8a, the topology is shown, and Figure 8b shows the simulation results of this topology. The H5 topology was modified by adding an extra switch in the neutral side of the PV panel to lower the leakage current compared to that of the H5 topology. Here, the left bridge side was operated on the line frequency and right bridge side on the switching frequency. The disconnecting switches were kept on for a half cycle in an alternate manner to avoid more switching loss. The merits of the topology are lower THD and lower leakage current than that of the H5 obtained by simulation. However, the total common mode voltage swung between 0 V and 400 V with a different nature from H5, the peak leakage current was around 150 mA, and the computed efficiency was lower than that of H5 due to the increased number of switches.

3.1.3. Modified H6 Topology

A modified H6 topology was proposed in [38]. The topology is shown in Figure 9a, and the simulation results of this topology are depicted in Figure 9b. This topology disconnects the PV side from the grid on the free-wheeling mode. The left side of the bridge is operated on the line frequency, and the other four switches are operated on the switching frequency. The nature of the total common mode voltage is evident from the minimal value of leakage current (rms = 12.05 mA, peak = 130 mA), which was somewhat better than that of H5 and the improved H6 topology. The calculated efficiency was higher than the improved H6 but the same as H5. However, in the simulation, the THD was greater than the H5 and the improved H6 topology.

3.1.4. HERIC Topology

The HERIC topology was first proposed in [39]. The topology is drawn in Figure 10a. By utilizing the bidirectional clamping approach in [88] and the advanced modulation technique in [87], this topology’s reactive power capacity was further increased. By using soft switching techniques in [92,93], and enhanced modulation techniques in [94], this topology’s efficiency was raised. Figure 10b displays the simulation results of this topology. Throughout the entire cycle, a portion of the grid current circled through the S5 and S6, progressively reducing the leakage current. By functioning on the line frequency, the S5 and S6 switches serve as ac decoupling switches in the zero and non-zero states. The other four switches are run on the switching frequency. The excellent operation of S5 and S6 makes the topology the most efficient of all decoupling topologies. The simulation results showed the rms leakage current as 12.09 mA, but the peak value was around 125 mA, and the total common mode voltage was also oscillating. The THD of voltage was higher than that of H5 and the improved H6.

3.1.5. H6-I Topology

The H6-I topology was primarily proposed in [76]. Figure 11a depicts the topology. Using an advanced hybrid modulation technique, this topology’s capacity to inject reactive power was further improved in [95,96]. By using a new control strategy, the common mode voltage behavior was enhanced in [97]. Figure 11b shows the simulation results of this topology. In this topology, S5 and S6 switches were operated on both active and free-wheeling modes. The diodes were used to flow the free-wheeling current. The other switches were operated on switching frequency. The efficiency of this topology was higher than that of the improved H6 topology due to the lesser switching loss of switches S5 and S6. Its rms leakage current was recorded as 14.94 mA, and the peak was about 150 mA. The THD performance was worse than that of other selected zero state decoupled topologies except for H6-II. This topology’s total common mode voltage oscillated as well.

3.1.6. H6-II Topology

The H6-II topology was firstly proposed in [79]. The topology is sketched in Figure 12a. The reactive power capability was further improved using advanced modulation techniques in [88,98]. Figure 12b shows the simulation results of this topology. S5 and S6 switches were operated in this topology in both active and free-wheeling modes, similar to the H6-I topology. The diodes were used to flow the free-wheeling current. The other switches were operated on the switching frequency. The efficiency of this topology was the same as the H6-I topology due to the same number of switching elements and similar operation principles. The main difference between H6-II and H6-I could be identified from the location of the output terminal of the inverter. Its THD performance and common mode behavior were worse than those of other chosen zero state decoupled topologies.

3.2. Zero State Mid-Point Clamped Topologies

The top six topologies in this category were selected for analysis and comparison. To enhance common mode behavior, they used the mid-point clamping approach at the zero state. They are as follows:

3.2.1. OH5 Topology

The OH5 topology was proposed in [46]. The topology is shown in Figure 13a, and Figure 13b demonstrates the simulation results of this topology. The upper two switches of the bridge were selected for free-wheeling mode current flowing. The total common mode voltage was now roughly constant at 200 V due to the effective use of switch S6 in the zero state or free-wheeling mode. However, because both the inductor and the capacitor impact leakage current, when the first half of the cycle was ignored, simulation results showed a continuous leakage current of 65.05 mA without a peak. However, this topology’s efficiency was the lowest among those in this category due to the running of more switches at the switching frequency.

3.2.2. PN-NPC Topology

The PN-NPC topology was proposed in [47]. Figure 14a depicts a topology drawing, and Figure 14b displays simulation results of this topology. Eight switches were utilized in this topology, four of which were operated on line frequency and the remaining four on the switching frequency. Switches S7 and S8 were selected for the clamping function on the zero state as well as on the active state. When compared to the other zero state mid-point clamping topologies, its THD performance was ranked second. The standard efficiency was maintained by running the four switches on the line frequency. Although only 13.29 mA rms value of leakage current was present, simulation results showed a 100 mA spike at an incorrect clamping point.

3.2.3. H5-D Topology

The H5-D topology was proposed in [48]. The topology is shown in Figure 15a. Figure 15b shows the simulation results of this topology. In this topology, diode D1 was used for clamping purposes. Switches S2 and S4 were turned on during the free-wheeling mode. Despite not obtaining a single stable common mode voltage, the swing was restricted to 0 V to 200 V. The rms leakage current was 13.88 mA, although simulations showed 100 mA peaks due to incorrect clamping. However, the THD performance was the best among the chosen zero state mid-point clamping topologies. The estimated efficiency was higher than OH5 at 97.3%.

3.2.4. HERIC Active-I Topology

The HERIC Active-I topology was proposed in [69]. Figure 16a sketches out the topology, and Figure 16b shows the simulation results of this topology. Switch S7 solely clamped the midpoint voltage in the positive half cycle’s zero state. S5 and S6 were operated on the line frequency to create an extra free-wheeling path using the ac decoupling technique. The efficiency of this topology was the highest among the selected topologies under this category due to lower conduction losses. The rms leakage current was 41.41 mA, and the peak value was about 120 mA. Because only the positive half cycle was clamped, there was a relatively higher leakage current.

3.2.5. HERIC Active-II Topology

The HERIC Active-II topology was proposed in [69]. The topology is shown in Figure 17a, and the simulation results of this topology are depicted in Figure 17b. Switch S7 was used for clamping purposes, and it was operated on the switching frequency on the zero voltage state. For the purpose of ac decoupling, switches S5 and S6 were operated at the line frequency. The efficiency of this topology was higher than that of H5-D and PN-NPC topologies. The leakage current’s peak value was restricted to 100 mA, despite the fact that the leakage current rms value was somewhat greater than that of other chosen topologies in this category.

3.2.6. HBZVR-D Topology

The HBZVR-D topology was proposed in [49]. Figure 18a shows a drawing of the topology, and Figure 18b shows the simulation results of this topology. Two diodes, D5 and D6, were employed for clamping the mid-point voltage. Switch S5 was used for creating the free-wheeling path during the zero state. The free-wheeling path was made possible by the anti-parallel diodes D1–D4. The rms leakage current was 22.75 mA, and the total common mode voltage was about a constant 200 V. This topology’s computed efficiency was 97.6%, which was higher than that of the OH5 and H5-D topologies. However, the leakage current’s peak value was about 110 mA.

3.3. Solidly Clamped Topologies

In this category, the most optimum six topologies were selected for analysis and comparison purposes. They used a solid connection from the grid negative terminal to either the PV neutral or the mid-point of two series dc link capacitors to improve the common mode behavior. They are as follows:

3.3.1. Virtual DC Bus Topology

The Virtual DC Bus topology was proposed in [52]. The topology is shown in Figure 19a, and the simulation results of this topology are sketched in Figure 19b. The capacitor C₂ was charged during the positive half cycle to create a virtual DC bus for the negative half cycle. The S3 and S5 switches were activated during the positive half cycle free-wheeling mode, while the S2 and S4 switches were activated during the negative half cycle free-wheeling mode. This topology’s total common mode voltage was about 0 V because the grid’s negative terminal was directly connected to the PV side’s neutral. The rms leakage current was only 0.75 mA, but the leakage current peak value was roughly 7.5 mA. The efficiency was only 96.8% due to high conduction losses.

3.3.2. Flying Capacitor Type-I topology

The Flying Capacitor Type-I topology was proposed in [53]. Figure 20a shows a sketch of the topology, and Figure 20b shows the simulation results of this topology. The flying capacitor C_F was charged during the whole positive half cycle, and it acted as the source for the negative half cycle through discharging. Although all switches functioned at the switching frequency, because the length was only half a cycle, switching loss was cut in half. Although capacitor charging caused additional conduction loss, the efficiency was still quite high at 97.2%. Since it is evident from the simulation that the total common mode voltage was set at zero volts, the rms leakage current was only 0.58 mA. However, starting the leakage current’s peak value at 8.5 mA, it decreased over time. Furthermore, the output voltage’s THD was 4.2%.

3.3.3. Flying Capacitor Type-II Topology

The Flying Capacitor Type-II topology was proposed in [54]. The topology is drawn in Figure 21a, and the simulation results of this topology are shown in Figure 21b. Throughout the entirety of the positive half cycle, the flying capacitor C_F was charged through switch S1 and diode D. The grid was powered by the charged capacitor C_F during the negative half cycle. To lower the switching loss, switches S1 and S4 were activated during the free-wheeling mode for a half-cycle period on the line frequency and another half-cycle on the switching frequency. The total common mode voltage was confirmed by simulation results to be 0 V, and as a result, the rms leakage current was just 0.58 mA. Although the leakage current’s peak value starting from 8 mA diminishes with time. The obtained efficiency of this topology was the same as that of the Flying Capacitor Type-I topology at 97.2%, and the voltage THD was 4.2%.

3.3.4. Siwakoti-H Topology

The Siwakoti-H topology was first proposed in [55]. The topology is represented in Figure 22a. The nonlinear behavior of this topology was properly addressed using a discrete-time state feedback control scheme in [99], and using a direct model predictive control scheme in [100]. Figure 22b shows the simulation results of this topology. Two switches called S1 and S4 were reverse blocking IGBT (RB-IGBT), and they were operated on the zero state to charge the flying capacitor C_F. This fully charged capacitor was used as the source of the negative active half cycle. The positive active half cycle was controlled by switch S3. The total common mode voltage was observed as 0 V from the simulation, and hence the rms leakage current value was only 0.47 mA. However, the 40 mA leakage current peak was seen in the first cycle. The computed efficiency was second in this category at 98.1%, and the simulation showed a voltage THD of 4.3%.

3.3.5. Conergy NPC Topology

The Conergy NPC topology was primarily proposed in [56]. The topology is redrawn in Figure 23a. According to switching frequency and filter parameters, this topology with SiC switches was designed optimally in [91]. Figure 23b shows the simulation results of this topology. The negative terminal of the grid was directly attached to the middle point of two equal-series dc link capacitors. To flow the current in the free-wheeling mode, switches S3 and S4 were operated in the zero state in the opposite way. Although the total common mode voltage was about 400 V, nevertheless, the leakage current rms value was limited to 0.14 mA with no noticeable peaks due to its solid connection and constant V_tCM. Due to its extremely low conduction loss, this topology had the highest efficiency among those in this category. The voltage THD was also 1%. However, the input voltage requirement of this topology was twice than that of other topologies.

3.3.6. S4 Topology

The S4 topology was proposed in [57]. Figure 24a depicts the topology, and the simulation results of this topology are depicted in Figure 24b. In the positive half cycle active mode, C₁ charged through diode D₂, and then C₂ charged from the discharge of C₁ through D₁ in the positive zero state. In the negative half cycle, C₂ provided power to the grid. The total common mode voltage was 0 V, and hence the rms leakage current was only 0.56 mA. Because switching frequency was used the majority of the time to run all switches, efficiency was impacted, and the computed efficiency was only 96.2%. Furthermore, compared to other chosen topologies, the voltage THD was worse.

4. Comparative Analysis, Discussion, and Future Trends

Transformerless inverter topologies are categorized into three groups in this paper. MATLAB/Simulink was used for analysis and comparison purposes to simulate a total of 18 topologies, 6from each category, using the parameters shown in Table 3. Using the equations provided in Section 2.3, the loss distribution across all switches, and the maximum efficiency of all topologies were computed. The simulation results of the common mode behavior of each topology were discussed thoroughly in Section 3. In order to compare the topologies fairly,

Table 4. Comparison of the selected topologies based on simulation results and physical parameters.

Category	Topology Name	No. of Input Capacitor	Input DC Voltage	Modulation Index	No. of IGBTs	No. of Diodes	No. of Filter Inductors	No. of Filter Capacitors	Simulation Results			Calculated Maximum Efficiency@ 998.2 W
									Total Common Mode Voltage (VtCM)	RMS Leakage Current (ILC)	THD of Voltage
Zero state decoupled topologies	H5 [36,87,88,89,90,91]	1	400 V	0.82	5	0	2	1	(0–400) V, Floating	13.23 mA	0.68%	97.8%
	Improved H6 [37]	1	400 V	0.82	6	0	2	1	(0–400) V, Floating	12.77 mA	0.60%	97.1%
	Modified H6 [38]	1	400 V	0.82	6	0	2	1	(0–400) V, Floating	12.05 mA	1%	97.8%
	HERIC [39,87,88,92,93,94]	1	400 V	0.82	6	0	2	1	(0–400) V, Floating	12.09 mA	1%	98%
	H6-I [76,95,96,97]	1	400 V	0.82	6	2	2	1	(0–400) V, Floating	14.94 mA	4.9%	97.8%
	H6-II [79,88,98]	1	400 V	0.82	6	2	2	1	(0–400) V, Floating	15.19 mA	5%	97.8%
Zero state mid-point clampedtopologies	OH5 [46]	2	400 V	0.79	6	0	2	1	200 V, Constant	65.05 mA	3.3%	97.2%
	PN-NPC [47]	2	400 V	0.82	8	0	2	1	(100–230) V, Semi-floating	13.29 mA	1.7%	97.4%
	H5-D [48]	2	400 V	0.795	5	1	2	1	(0–200) V, Floating	13.88 mA	1.3%	97.3%
	HERIC Active-I [69]	2	400 V	0.795	7	2	2	1	200 V, Constant (half cycle); (0–200) V Floating (half cycle)	41.41 mA	3.5%	97.8%
	HERIC Active-II [69]	2	400 V	0.79	7	0	2	1	(100–200) V, Semi-floating	71.48 mA	3.7%	97.6%
	HBZVR-D [49]	2	400 V	0.78	5	6	2	1	200 V, Constant	22.75 mA	4%	97.6%
Solidly clamped topologies	Virtual DC Bus [52]	2	400 V	0.775	5	0	1	1	0 V, Constant	0.75 mA	5.1%	96.8%
	Flying Capacitor Type-I [53]	2	400 V	0.782	4	1	1	1	0 V, Constant	0.58 mA	4.2%	97.2%
	Flying Capacitor Type-II [54]	2	400 V	0.782	4	1	1	1	0 V, Constant	0.58 mA	4.2%	97.2%
	Siwakoti-H [55,99,100]	2	400 V	0.80	4	0	1	1	0 V, Constant	0.47 mA	4.3%	98.1%
	Conergy NPC [56,91]	2	800 V	0.815	4	0	1	1	400 V, Constant	0.14 mA	1%	98.2%
	S4 [57]	2	400 V	0.795	4	2	1	1	0 V, Constant	0.56 mA	9.2%	96.2%

was created based on the simulation results, computed efficiency, and physical parameters of the topologies. The loss distribution over each switch of the 18 topologies is shown in Figure 25.

In Figure 26, the output power versus efficiency curve for three chosen topologies—one from each category—is displayed.

In the case of the zero state decoupled topologies category, the circuit structure is comparatively simple due to absence of a clamping branch. The total common mode voltages for all six topologies fluctuate between 0 V and 400 V. Under this category, the HERIC topology displays the highest efficiency. Even though this group’s rms leakage current ranges from 12 to 15 mA, the concerns associated with their peak leakage current are detailed in Section 3. In comparison to other topologies in this group, the loss distribution across various switches is best in modified H6. HERIC can be selected as this group’s best topology after considering all the factors, even though H5 performs well.

In the case of zero state mid-point clamped topologies, the leakage current’s peak value is comparatively smaller than that of zero state decoupled topology, but the topological arrangement is more complex. The constant total common mode voltages are guaranteed by the OH5 and HBZVR-D topologies. Other topologies in this group have semi-floating or partially total constant common mode voltages. Although this group’s average rms leakage current is higher than that of the zero state decoupled topologies group, the average peak value of leakage current of this category is lower due to the mid-point clamping technique. HERIC Active-I is the most efficient topology in this group in terms of efficiency, and PN-NPC performs best in terms of leakage current.

HERIC Active-II is best in terms of loss distribution among switches. However, PN-NPC can be regarded as this group’s best topology when all the factors are taken into account.

All solidly clamped topologies guarantee the rms leakage current less than 1 mA, but the topological construction is the most complicated, since a solid connection between the electrical grid and the PV is necessary. Due to a strong solid connection between the grid and the PV negative terminal, the total common mode voltage of this category, with the exception of Conergy NPC topology, is zero. In the Conergy NPC topology, the negative terminal of the grid side is connected to the halfway point of two series dc link input capacitors; therefore, the topology’s total common mode voltage is not zero volts. However, because of the strong connection between the PV side and the grid side and the constant V_tCM, this topology only has a leakage current of 0.14 mA. Siwakoti-H operates most effectively in this group for 400 V input. Conergy NPC topology, however, can be regarded as the best topology in this category when all factors are taken into account.

However, choosing the best topology among the three groupings is challenging. The solidly clamped topologies perform the best regarding the leakage current problem, but the circuit configuration and control system are more complicated than those of other groups. Each group has unique benefits and drawbacks. Thus, the optimum topology may be selected based on requirements and consideration of all parameters.

Future research on various control strategies that can be used with the major topologies discussed in the paper to enhance reactive power control has significant opportunities. Different advanced modulation schemes can be suggested for improving the efficiency of the selected topologies. Using novel or improved modulation techniques will lower the leakage current and total harmonic distortion. Additionally, to further improve the reliability of the inverter system, fault-tolerant techniques can be incorporated into the already-existing topologies.

5. Conclusions

Transformerless inverters are the choice of the future due to their many benefits, such as minimal weight, low cost, and outstanding efficiency. However, there are certain obstacles to overcome. Leakage current and complexity in topology and control systems are two. In this paper, every topology is extensively simulated using a common set of parameters to evaluate and fairly examine its common mode characteristics and other output characteristics.

In Figure 25, the loss distribution across each switch is also displayed to highlight the topologies’ strengths and weaknesses in terms of loss spread. Table 4 offers a comparative analysis of all selected topologies of single-phase transformerless inverters, especially in regard to their key properties based on simulation and theoretical calculation, as a summary of this evaluation. Additionally, based on simulation and theoretical research, the top three topologies, one from each category, were chosen. The authors hope that this review will be useful for researchers, engineers, manufacturers, and users related to the academy and industry as a resource on transformerless grid-connected PV inverters. In the future, each selected topology of this paper can be verified experimentally based on the same parameters for best understanding of the insights of the topologies.