1. Introduction
The energy demand for industrial, commercial, and residential consumers is a fact in the 21st century. One way to accomplish this demand without harming the environment has been seen through the incorporation of distributed generation (DG) systems in distribution networks. Large and as well as small-scale photovoltaic (PV) arrays for home applications have become popular rather than other DG sources due to the improvement in manufacturing techniques and significant advances in power electronic interfaces. Electrical energy produced by a PV installation is basically a DC system. Therefore, it must be converted into AC power to make it available in power utility through a grid-tie inverter [
1,
2,
3]. Transformerless voltage source inverters (VSIs) are the standard choice in detriment of current source inverters (CSIs), because they show better conversion efficiency, smaller size, and lower manufacturing costs for the same power rating. Transformerless inverters generally have two classes of connection such as galvanic or non-galvanic categories. When a high-frequency low-size transformer on the DC side or low-frequency large-size transformer on the AC side are used, the electric connection is removed, which means electrical isolation between the two electrical systems [
4]. This electrical separation improves operation security and reliability. However, it deteriorates power conversation efficiency. If the transformer is removed (following protection guidelines [
5,
6]), renewable power generation investment cost is considerably lower. Many studies have been conducted so far on this issue to explain how common mode current (CMC) could be surpassed [
7]. CMC circulation is only possible, since there is no galvanic isolation which, combined with the stray capacitance that appears between PV installation and the DC side ground, allows a path for ground current to be injected into the neutral point at the AC side. The stray capacitance level may depend on:
Class of PV cells;
Climate type and residential localization;
The height between PVs and ground on the DC side;
Voltage level; and
Electromagnetic Interference (EMI).
To guarantee grid reliability, modern grid-connected PV standards propose leakage current protection device actuation according to CMC amplitude. This means that CMC values below 300 mA are permanently acceptable, while above this point, the protection must react and trigger a break by isolating the transformerless inverter from the utility grid. Furthermore, this current must not continue over 0.3 s during the operation scheduling period. If the CMC increases, the maximum permissible time to trip will be decreased, correspondingly.
Table 1 depicts the recommended tripping time as a function of CMC amplitude.
To eliminate the leakage current presence, several topologies have been studied and analyzed in the literature, which are shown in
Figure 1. The Karschny structure depicted in
Figure 1a [
8] is based on an asymmetrical output inductor. The topology works at low voltage on the DC side being configured to be operated in two different modes. In one half cycle, its operation is very similar to a buck converter. On the other hand, when S
1 and S
5 are switching in a complementarily way, the topology operates as a buck-boost converter. This allows generating a semi-pure sinusoidal current. In practical terms, it is not easy to implement because it requires a complex control technique. In addition, the three switches that contribute to the current path lead to an increase in the active power loss. Other forms of proposed topologies are based on dual-buck topologies [
9,
10], which can be seen in
Figure 1b,c. These structures are basically two buck converters that can be connected in series or in parallel being operated complementarily to perform DC decoupling of the current path. Switches S
L1 and S
L2 are commuted at a power grid frequency in order to prevent the reversal of the inductor current [
10]. The half-bridge-based transformerless inverters as shown in
Figure 1d may limit the CMC presence, but at the cost of operating restrictions such as the requirement of the DC input voltage being at least twice that of the utility grid voltage. In addition, from a design point of view, power switch components are subjected to higher voltage stress and require a larger output filter size [
11]. In
Figure 1e, the neutral point clamped (NPC) VSIs have been recently applied in grid-connected PV systems. Due to its structure, extended switching frequency is possible, allowing in turn smaller filter size calculations and reduced voltage stress on power switches. The NPC inverters can balance the voltage applied to off-biased switches. Despite their advantages, the active power loss is not negligible, taking into account the number of components used [
12].
The full bridge topology represented in
Figure 1f that uses symmetrical inductor filters cannot effectively decouple the CMC on the AC side, since high-amplitude CMC is generated with the sinusoidal pulse width modulation (SPWM) technique [
13].
A concept known as highly efficient and reliable inverter (HERIC) can be seen in
Figure 1g. It comprises a full bridge configuration where two additional high-frequency switches are arranged in opposite directions on the AC side. The freewheel path is also produced by those high-frequency switches in high-efficiency operation. The main disadvantage of HERIC VSI is that the exact regulation of them results in output voltage and current chopping. The H5 topology shown in
Figure 1h is obtained from a full bridge converter by incorporating a switch on the DC side to decouple it from the AC side, and therefore suppressing the CMC interference [
14,
15]. The three upper-side switches are controlled by the power grid frequency, while the rest are under high-frequency signal switching control. In the freewheel condition, only two semi-conductor switches are conducted, but in an active state, three of them are tuned on, which results in large active power loss consequently. Adding an extra switch to H5 topology will make those DC side switches work in pairs and will obtain the H6 topology represented in
Figure 1i. However, the modulation performance varies regarding H5 topology and the most important drawback of this structure is the active power loss achieved in conduction operation mode with three switches [
16].
This paper is organized as follows:
Section 2 presents the CMC conceptualization for a single-phase PV inverter. The mitigation method and CMC evaluations are discussed in
Section 3 and
Section 4, respectively. The design considerations and maximum power point tracking (MPPT) algorithm are presented in
Section 5.
Section 6 analyzes the active power loss into two categories: switching and conduction. The simulation results and experiments are conducted in
Section 7, and finally, the reactive power compensation is described in
Section 8. Briefly, in this paper, a high-efficiency grid-tie inverter structure for non-isolated PV systems is proposed. The main contributions of the presented VSI are briefly mentioned below:
An extended H6 topology by comprising seven switches, of which five are controlled in high frequency. The remaining switches operate in line frequency modulation to create the freewheel path.
The 7th switch is used to clamp the common mode voltage; if the voltages () are higher than half of the DC link voltage, freewheeling current flows through S7 and D1 to the midpoint of the dc link, which results in and being clamped at /2.
The maximum power point tracking (MPPT) algorithm based on perturb and observe (P&O) is considered to increase the total system efficiency.
The AC decoupling strategy has been chosen.
The unipolar SPWM pattern is applied to gate inputs.
A reduced filter size is hired to decrease the unsolicited harmonic generated by output current and restrict the total harmonic distortion (THD).
2. CMC Origin in Transformerless Inverter
In the first categories of PV inverter topologies without galvanic connections mentioned before, which may be known as switching inverters or transformerless inverters, many applications are verified. The main drawback of these configurations is the CMC generation, which has appeared between the AC mains and the DC side of the inverter. A general structure of a transformerless PV inverter highlighting the path of
is shown in
Figure 2. As it can be seen clearly, the common mode current is flowed in a loop that consists of a DC input voltage source or PV arrays voltage, inverter switches, output filter inductances, AC grid, grounding impedance, and parasitic stray capacitances of PV cells. The common mode voltage is the mean of voltages of node A and node B. Besides,
is the differential of the aforementioned nodes.
Figure 3 has been depicted to represent the simple single-line circuit of the discussing topology in order to calculate the total common mode voltage. Consequently, according to these explanations, it could be expressed with the following equations.
Since
appeared due to
flows in the abovementioned loop, it could be proved that the total common mode voltage is obtained in (4):
Clearly, it is observed that the criteria for the elimination of
is to make
remain constant. Thus, we should have:
In some configurations such as half-bridge family or Karschny inverters and some of those indicated in
Figure 1a, only one inductor is used in the output to filter the harmonics, so that the other inductor is equal to zero or does not need to be considered. Thus, by assuming that
, we find the
as (6).
If
,
will be extracted in (7).
In other PV transformerless inverter topologies, as shown in
Figure 1, both inductors
and
are existing and equal to each other. This equality results in
. Therefore, to eliminate the leakage current flow, Equation (8) is being obtained:
It is concluded that transformerless VSIs could be categorized into two groups. The first one is called “asymmetrical inductor based inverters”, which trust in (6) and (7). Similarly, the second group is named “symmetrical inductor based inverter”; Equation (8) is applied to them.
4. H7 Topology Evaluation
In order to evaluate the common mode leakage current impact on the proposed topology, we derived the equivalent circuit of the loop from
Figure 6 that imports the novel structure presented in
Figure 6. Since the input capacitances
are very small, they are not considered for calculation purposes. In addition,
has no influence in
derivation, being discarded, too. Manipulating Equations (1) and (3) can be reorganized as:
Based on that, the CM equivalent circuit of
Figure 6 is converted in
Figure 7a. Further simplification can be obtained in the form of a single loop arrangement (
Figure 7b).
Using the single loop circuit simplification, CM voltage and current values are computed as follows:
5. Prototype Design Consideration
This part of the paper is organized in three sections: filter inductance design, filter capacitance selection, and semi-conductor components sketching. In these topologies, typically the maximum ripple of the current is considered between 10% and 25% of the root meat square (RMS) current [
17]. Therefore, for a 36 W prototype fed a load with 30 V at 8 kHz switching frequency, the filter inductance is obtained as (15).
where the maximum current ripple is assumed to be 20% of the nominal current. Consequently, the capacitance of the output filter is generally designed based on the cut-off frequency. This frequency is normally between 10% and 20% of the switching frequency [
18]. Equation (16) is given to help the selection of filter capacitance.
Therefore, filter capacitance will achieved as approximately 220 µF. In order to select the semi-conductor components, it is noted to say that in high-frequency conditions, the switches S
1 to S
4 should work properly. On the other hand, the switches S
5 and S
6 are operated at line frequency. According to peak inverse diode (PIV), total standing voltage (TSV), and nominal parameters of load, the MOSFET IRFP150 was selected as a power switch that could withstand up to 100 V and 40 Amperes as the nominal current. For the diodes configuration, the BY3099 reference was chosen, while the MOSFET driver role is ensured by the ICL 7667. Finally, to guarantee galvanic isolation between the gate signals and power circuit, we used a 6N137 optocoupler [
19].
MPPT
The P&O approach is a well-known algorithm to perform the MPPT function. It works by applying a small disturbance in the system; then, the set-point of PV arrays are going to be changed due to tracking the maximum power point. The equation that describes the P&O algorithm operation is shown in (17) [
20].
where
. The MPPT flowchart of the P&O strategy is shown in
Figure 8.
7. Simulation and Experimental Results
In order to evaluate the H7 topology an inverter prototype was built. It can be seen in
Figure 9 the main elements of the inverter system used in bench tests.
Table 2 gather the main electrical parameters that characterize the electrical tests. The switching frequency is chosen as 8 kHz and the line frequency is 50 Hz.
Figure 10a,b represent the
and
signals obtained by simulation. Being signals with pulsating periodic waveform, this behavior makes
constant, as can be verified in
Figure 10c. Consequently, the common mode current elimination target is accomplished. For emulating as close as possible a real scenario, it was chosen a stray capacitance of 75 nF. The experimental waveforms are shown in
Figure 11, demonstrating the effectiveness of the novel photovoltaic transformerless inverter. The output voltage and load current amplitudes can be checked in
Figure 11a. Voltage THD measurement at inverter output is 1.51%, in accordance with the value estimated in simulation whose result is 1.07%. Measured waveforms of
and
have approximately square waveforms (
Figure 11b), confirming that
is constant (
Figure 11c). Finally, in
Figure 11d the CMC waveform is presented. Due to experimental device restrictions, DC input voltage of 30 volts conditioned
and
peak voltage around 30 V. The peak value of the output voltage and current are near 30 V and 1.2 A, respectively. From
Figure 11d data the current measured in common mode is approximately 1.14 mA. Some undesirable peaks shown in these pictures appear due to non-ideality in components.
Table 3 shows a comparative study concerning the proposed topology with published performance data regarding some well-known single phase inverter structures. The comparison highlights that the H7-type inverter with fewer components can not only reduce significantly the CMC circulation, as well as the power conversion efficiency is one of highest. On the other hand, these results denote that the CMC of the H6 topology is the lowest of all under comparison and the two others’ CMC is practically the same. However, the efficiency analysis shown in
Figure 12 relies on the best performance of the proposed topology. To illustrate this, it can obviously realized by comparing our topology with those evaluated in active and freewheel conditions. In the active state, the AC current flows through only two switches in the proposed topology, similarly to HERIC-type inverter, while H6 and H5 topologies require three power switches translating into higher active power loss. In the freewheel state, the H7-type design makes use of only one switch and one diode in line with the H6 structure. The same is not true for the H5 and HERIC converters whose operation depends on the two switches, generating consequently additional power losses.
The California efficiency that is used to calculate the diagrams shown in
Figure 12 is supposed to be what is written in (30) [
12,
13,
14,
15].
Figure 13 shows the total loss of active power through the energy transmission process. As it is understood, the suggested topology has the least loss in equal condition. When the same nominal output power is selected for all topologies, the power loss in terms of nominal load percentages will be changed due to the current drawn. According to
Figure 13, the proposed topology has the least power loss among all during load variations.
Figure 14 shows the percentage of each power loss shared during operation. As discussed before, the proposed topology has a conduction loss that is the same as the HERIC and the freewheel loss is the same as that of the H6 VSIs. The only difference between these converters appears in switching loss calculations. Whereas the rated power is loaded, the share of switching loss in the proposed structure obtains the biggest part among all quotas. This issue does not mean that lots of power transmitted through the VSI is wasted during switching transitions, but this justifies that the power loss could be decreased in terms of switching frequency. Consequently, the total loss will be mitigated in the proposed topology, and its exclusive capability does not appear in other VSIs.
Figure 15 compares the total loss obtained between IGBT and MOSFET utilizations. Since the switching frequency increases, the power loss growths, correspondingly, as expected. However, MOSFTE utilization can decrease the power loss about 40% rather than using IGBT in all switching scenarios. As a result, since the power loss increases as the switching frequency is enhanced, the efficiency will be decreased, subsequently.
Figure 16 represents the total efficiency of proposed topology in terms of load power variations, considering switching frequency effects. If the switching frequency is chosen as 30 kHz, the maximum efficiency is obtained as 97.42% in 2.75 kW of output power. However, the maximum efficiency is being calculated as 98.02% for the other switching frequency in
. The average efficiency is assessed nearly as 97.1% and 97.85% for
and
, respectively, in all loading conditions. As the output voltage and current waveforms of experimental results are achieved at 8 kHz switching frequency, due to our laboratory equipment, the output power has been varied between 20 W and 150 W step by step. This verification implies that the proposed prototype is working successfully with high efficiency and reliable performance, as shown in
Figure 17. The reactive power and its effect on efficiency is one of the most important concerns in single-phase photovoltaic inverters.
The H7-type VSI is equipped with a capability to control the active and reactive power instantly by the proposed switching pattern. The loading condition is considered as , and the load current is calculated as . Then, the average and the amount of current flowing through the diodes and switches are computed, and the loss calculation is obtained as 0.886 W at 8 kHz switching frequency. Then, the effective efficiency of the proposed topology will be figured out as shown in (31).
It should be noted that with a rise in load power, the current ripple and the sizing of passive components are minimized due to the increment in current amplitude. Furthermore, the proposed topology has the ability to accomplish the power factor correction (PFC) operation mode as an active filter application.