Society is more than ever compelled to efficiently utilize existing energy resources while also incorporating Renewable Energy Sources (RES) into the energy infrastructure. This move is aimed to reduce Greenhouse Gases (GHG) emissions, comb the imminent climate crisis, and meet the increasing demand for electricity around the world. This energy demand has promoted the development of alternative sources such as wind turbines, tidal energy generation, Photovoltaic (PV), and green hydrogen energy systems (Ashok Kumar et al., 2021). PV systems are becoming cost-competitive against conventional thermal power generation systems and will be a major source of electrical power in the future (Ullah et al., 2020).

Power inverters are a crucial component of photovoltaic systems. They are responsible for increasing the input voltage and converting the DC power to AC power that can be connected to the grid or used in standalone systems. Typically, inverters have multiple stages of conversion, where the first stage is a DC-DC converter that increases the input voltage and performs Maximum Power Point Tracking (MPPT). The final stage of the inverter chain is the DC-AC converter. In some cases, the photovoltaic system may require transformers for galvanic isolation.

Power quality is a crucial aspect of power inverters. Ideally, inverters should produce a pure sinusoidal waveform, but this is not always possible in real-life applications (Cherkaoui Jaouad et al., 2022). To achieve high power output power quality, 2-level inverters must be switched at high frequencies, resulting in decreased conversion efficiencies (Kabalcı, 2021). However, the introduction of multilevel inverters has led to cost reduction, lower Total Harmonic Distortion (THD), and improved efficiency (Vakacharla et al., 2020).

Improvements on the inverters can be attributed to two factors, the development of new topologies and control methods that incorporated innovations and modernization in semiconductor and computing technologies. With MLI, high voltage inverters that were previously unfeasible with 2-level inverters due to power rating constraints on semiconductor switches can now be implemented (Kala and Arora, 2017). MLIs offer several other benefits over 2-level inverters, such as reduced rate voltage change over the change in time (dV/dt), lower common-mode voltages, and elimination of the need for bulky output filters (Rodriguez et al., 2009; Dhanamjayulu et al., 2022a; Atar et al., 2023).

Multilevel inverters offer many advantages, but they also have some drawbacks. The cost of the many switching devices and passive components necessary for MLIs can be a limiting factor in their potential applications. This article discusses the research that is being done in academia and industry to develop new reduced switch count and hybrid multilevel inverter topologies. These topologies aim to achieve high output power quality and improved efficiency with reduced switch counts and low complexity. It is important that topologies proposed as reduced switch count types do not come at the expense of increased complexity or use of multiple voltage sources. To assess the features and benefits of new advanced topologies, several key parameters are taken into consideration: total harmonic distortion, component count, efficiency, and complexity (Bhaskar et al., 2020; Sedaghati et al., 2022; Stöttner et al., 2022; Wikipedia, 2024). Figure 1 shows the classification of multilevel inverters. They may be categorized into single source, multiple source, resonant, and reduced switch count inverters. Only reduced switch count types and their improvement types or hybrids will be reviewed in this article and are highlighted in blue in the figure.

Research in multilevel inverters is an ever-evolving endeavor, and newer topologies that bring better performance are always contributed to the body of knowledge. In (Mehta and Puri, 2022), a review of reduced component MLI topologies classified as symmetric, asymmetric, and hybrid configurations was presented (Bughneda et al., 2021), reviewed photovoltaic systems and highlighted promising use cases of MLI in such systems (Barros et al., 2022), looked at modular multilevel inverters, their submodules, and modulation techniques, and (Jana et al., 2017) reviews multilevel topologies with a focus on soft-switched topologies. A recent review publication (Vinay Kumar and GowriManohar, 2023), reviewed and classified RSCs in asymmetric and symmetric topological structures, and compared their performance merits. The authors observed superior characteristics in asymmetrical structures in terms of higher efficiency, better switch utilization, and low THD. This review article takes a different approach not only to review topologies and their operations, advantages, and disadvantages but also to identify research opportunities for further development.

The paper is structured as follows. Section 2 presents a topological review of classical MLIs, discusses their advantages, and disadvantages, and highlights the motivation to move towards reduced switch count and hybrid MLIs; Section 3 provides an overview of modulation techniques that can be applied to MLI, Section 4 presents an analysis of recently proposed reduced switch count inverters and hybrid multilevel inverters, Section 5 summarizes the advantages and disadvantages of the reviewed topologies, and suggests potential research opportunities for further development, and Section 6 of the paper summarizes the conclusions and observations that were made during the review process.

The neutral-point clamped (NPC), cascaded h-bridge (CHB), and flying-capacitor (FC) MLIs were the first invented MLI topologies and are referred to as classical topologies. Coverage of the fundamental topologies is done extensively in other publications, (Abd Halim et al., 2016; Choudhury et al., 2021; Barnawi et al., 2023), and a summary is presented in this article for brevity. The CHB requires the least number of components among the three, as shown in Figure 2; Table 1, the reason for its widespread adoption in industrial applications. The main drawback of CHB MLI is the requirement for independent voltage sources, the provision of which is often achieved with expensive transformers. Asymmetric configurations of the CHB have the advantage of generating many voltage levels with fewer power sources and components, although this comes at a loss of redundancy, fault tolerance, and modularity (Rotella et al., 2006; Barzegarkhoo et al., 2016; Kannan et al., 2017; Busarello et al., 2018; Kumar Gupta and Bhatnagar, 2018; Chitra and Valluvan, 2020; Odeh et al., 2021). The resulting implementations have increased cost and complexity. FC MLIs have capacitor voltage balancing problems and high inrush currents. The NPC has the highest number of components. The issue with NPCs is the voltage balancing of the DC link capacitors at the input as the number of levels increases, which can cause undesirable transients that can damage the switching devices if not properly managed (Ahmad et al., 2024; Kieferndorf et al., 2000; Es-Saadi et al., 2018; Ghani et al., 2023). When the FC is compared to the NPC, the FC requires half the number of flying capacitors and clamping diodes compared to the NPC, resulting in a significant reduction in component count and cost. Additionally, the FC offers redundancy in configurations to generate zero output voltage, which can be valuable for optimization purposes (Chang-xin et al., 2009; Bressan et al., 2019; Humayun et al., 2020; Laamiri et al., 2017; Vincotech, 2024). In summary, all three have the benefit of generating low THD waveforms but in general, require many components to generate at high levels. A quantitative summary of component count for the three topologies is shown in Table 1 on the basis of the number of sources (N_DC), number of switches (N_SW), number of diodes (N_D), and number of capacitors (N_C). A qualitative summary is presented in Table 2, summarizing the merits and demerits of each topology. Great effort is directed toward developing multilevel inverters to enhance the level count while using fewer components.

Multilevel inverters need specific switching patterns to achieve the desired output voltage. Various techniques have emerged over time to control the topologies, and modulation techniques are an advanced field on their own. These techniques can be categorized on the basis of switching frequency into high-frequency and low-frequency techniques. Although there are many modulation techniques available, here we present a summary of some of the commonly used ones found in the literature (Kumar Gupta and Bhatnagar, 2018).

In the work of Thakre et al. (Thakre et al., 2022), the authors developed a topology as shown in Figure 3, that can accommodate symmetric and asymmetric voltage sources to improve the level count. The basic module of the topology consists of two unidirectional switches, two free-wheeling diodes, and two isolated voltage sources and can generate 3 voltage levels. Its ability to generate enhanced level count is overshadowed by the many voltage sources, which make it a cost-sensitive topology. The polarity inversion full bridge driving the output limits application to low and medium-voltage applications.

Ahmed et al. (2018), proposed a reversed source half-bridge multilevel inverter, eliminating the need for a dedicated polarity generator. It is usually custom to employ a full-bridge front end for this specific purpose of polarity generation, however, the front end in this case requires switches of high voltage blocking capability vis-a-vis the level generation switches and will only be limited to low voltage applications. To eliminate the full bridge front-end, the authors developed the topology as shown in Figure 4, that inherits simple cascaded half-bridge modules. Cascaded half-bridges cannot generate negative voltages, and to solve this, the authors implemented another half-bridge module with a negative polarity voltage equal to all the other cascaded positive modules combined. Intermediate voltage levels are generated through subtractive operations. The control complexity is low, and the asymmetric nature allows for low frequency modulation to enhance efficiency. Nevertheless, the negative half bridge limits its application to low-voltage applications and the many power sources can increase the cost.

In their paper Islam et al. (2023) proposed a five-level voltage-boosting switched capacitor MLI as shown in Figure 5. This structure consists of seven switches, 3 diodes, and 2 capacitors that act as virtual voltage sources. The topology uses a full-bridge frontend for polarity inversion, which makes it highly modular and extensible for high-voltage applications, although that comes at the expense of many voltage sources. The capacitor voltage balancing is inherent and allows for simple control. The switch utilization is low for this topology.

A switched series-parallel source multilevel inverter is proposed in (51) with a single DC source and 2 switched capacitors. Using capacitors in a topology to replace voltage sources helps to reduce the overall component count in a system. The inverter as shown in Figure 6, generates a 7-level output waveform and inherently balanced capacitor voltages. The excellent merit of the topology is the boosting capability, which additionally eliminates the need for boost converters, especially in renewable energy systems. The topology is highly modular and can be extended either by adding more switched capacitor units in series or cascading through the outer full bridge. The switch S2 has the highest conduction loss compared to other switches. This unbalanced nature will lead to reliability issues. Another downside of the topology is that the number of carrier signals used is tied to the voltage gain. This requires increased computational capability when extending and is costly. Fundamental-frequency modulation may erode the capacitor balancing capability and therefore not appropriate. Hybrid modulation can be explored to try to reduce the quantity of carrier signals.

A 31-level topology is presented in (52) to enhance output level count and shown in Figure 7 with asymmetric input voltage sources. The level generation circuits are innovative; however, the structure requires isolated voltage sources, in this case, provided using a high-frequency link which makes it unattractive cost-wise. As with all asymmetric topologies, the level count is enhanced. Generally, the component count is reduced on the level generation side but taking into account the high-frequency link and its associated components the prices will be compounded. The topology is complex, considering all the design variables including the design and sizing of the high-frequency transformer. It makes up for the demerits with high efficiency associated with low-frequency modulated asymmetric structure.

In (53), the authors present a novel high-frequency linked series switched-source MLI shown in Figure 8. It distinguishes itself from the topology in (52) by using the high-frequency transformer to generate only the low-voltage sources in an asymmetric configuration. The main voltage input to the inverter is supplied directly from the voltage source, hence the lack of complete galvanic isolation. The center-tapped secondary winding output allows for an enhanced level generation with the switched rectifier circuits, however, the many rectifiers involved considerably reduce the efficiency to a reported 95%. The topology still retains many components and the output full-bridge limits it to low to medium-voltage applications. However, cascading allows the system to be built to higher voltages, albeit costly considering the transformers.

In (54), a novel 13-level switched capacitor MLI with a voltage gain of six is proposed and is depicted here in Figure 9. The topology combines a switched capacitor series-parallel unit and an L-type unit for extending to high levels and gain. The topology benefits from a high number of complementary switching pairs, reducing the cost of drivers. The high gain is an advantage, especially for photovoltaic applications where the requirements for a boost converter are eliminated. However, the carrier signal requirement is twice the gain, making the control complex. The topology has inherent capacitor voltage balancing, an additional feature for reducing complexity. The structure could benefit from decoupling the charging of C1 and C2 in parallel, charge them independently and benefit from enhanced voltage gain and still retain the extensibility offered by the L-type unit. Furthermore, all the capacitors are charged in series with the voltage source and the addition of a series inductor can help with limiting the initial inrush currents.

In (55), a series-connected switched-source configuration with an auxiliary module and polarity changing full bridge is proposed. The topology was designed with hybrid modulation in mind from the onset and is shown in Figure 10. The switched sources are dedicated to the level generation, switched at low frequency, and the auxiliary modules, are switched at carrier signal frequency to smoothen the output waveform. The zero level is generated by the polarity changing full-bridge. All voltage sources must be equal to avoid short-circuiting through the body diodes. For this reason, it cannot accommodate asymmetric configurations. The voltage sources can be connected in parallel to share the load equally. One major benefit is that if the auxiliary module source is chosen to be half of the series sources, the number of levels generated is doubled. The topology generates a high number of levels with very few switches but requires isolated voltage sources even for the minimal configuration. Because the polarity is generated through the full bridge, it is limited to low to medium voltage applications. Even though possible to cascade, it would prove expensive. The topology benefits from a high number of complementary switches, reducing the cost of drivers drastically.

Kumari et al. (2023), proposed a 5-level extendible topology as shown in Figure 11. Its simple structure and inherent capacitor voltage balancing reduce the complexity of control. The design has retained a full bridge end, making it highly modular and extensible for higher voltage applications. For single source applications, it will only be limited to low-voltage applications. Its low gain makes it unattractive for photovoltaic applications requiring high gains. The topology has inherent voltage balancing capability. A single discharging state is always followed by two charging states, giving it a good capacitor voltage balancing capability and making it suitable for high-current applications.

In reference (Khan et al., 2022), a new topology of switched-capacitor is presented as shown in Figure 12. The topology and switching patterns implemented for it are designed to achieve inherent self-balancing of capacitor voltages, eliminating the need for auxiliary circuits and reducing the control complexity. The authors solve the issue of high inrush currents by incorporating a series inductor and implementing a precharge soft-start modulation scheme. The quasi-resonant recharging operations reduce losses considerably, a big advantage as low-loss capacitor charging or resonant charging in multilevel inverters is a current topic. With the inclusion of the pre-charge soft start, the control complexity has increased.

Sheik Tanzim et al. (Meraj et al., 2021), proposed a novel hybrid T-type inverter, a hybridization of a T-type, and a cross-switched MLI shown in Figure 13. The multisource structure allows for asymmetric configurations for increased level generation. The structure is extensible although the requirement for many isolated voltage sources is a drawback, and complexity is greatly increased. The cross-connected voltage sources require higher voltage blocking devices, a variation that will increase costs. Notwithstanding, the topology uses low-frequency switching, resulting in good efficiency results. The main drawback of this topology is the many different voltage sources.

Dhanamjayulu et al. (2022b), have proposed a novel 35-level asymmetric inverter applicable to renewable energy and electric mobility systems, comprising of eight unidirectional switches, 2 bidirectional switches, and five voltage sources. The inverter is shown in Figure 14. All asymmetrical topologies have the advantage of producing many levels with fewer components, but this topology contains a high number of voltage sources. The topology has four bidirectional switches designated as S1, S2, S6 and S12. The topology lacks extensibility and is limited to low to medium-voltage applications. As each of the voltage sources is of a different size, power distribution, and handling is unbalanced and varied. It is suited to battery systems where isolated power sources are readily available. The reported efficiency of the inverter is 93.37%, a comparatively poor performance figure in general, owing to the many switches present.

Figure 15 illustrates the 17-level asymmetric polygon-type asymmetric 17-level RSC MLI proposein (Samadaei et al., 2019). The topology is constructed with 2 voltage sources and 2 switched capacitors. One source has voltage V_dc while the second has 3V_dc, with the same scaling for the capacitor sources. The topology employs the NLC modulation technique that results in low switching losses. The switching states and the corresponding output voltages are presented in Table 14. The 1Vdc source is used to charge the 1Vdc capacitor, and the same is true for the 3Vdc source and capacitor. The inherent capacitor charging is achieved through careful planning of the switching paths. Both capacitors are charged one after the other within one window at the 0Vdc output voltage and their charging paths are created as follows: to charge the 1Vdc capacitor, switches S₁, S₂, S₄, S_8, and S₁₁ are engaged; and to charge the 3Vdc capacitor, switches S₄, S₅, S₆, S₇ and S₁₁ are switched ON. The sizing of the capacitors is such that it charges with sufficient charge to last all switching states within one-half of a single fundamental cycle. With such a long discharging period and a single short window for charging capacitors, high voltage ripples will be present. Very large and expensive capacitors will be necessary to keep voltage ripples to a minimum.

Srivivasan et al. proposed switch-ladder MLI (SLMLI) shown in Figure 16 (Srinivasan et al., 2021). The structure generates an output voltage of 81 levels. The output voltage levels can be increased by increasing the number of DC sources within the cell or through cascading modules or stages with fixed DC sources. The latter will use fewer switching devices compared to the former when increasing MLI voltage levels. In the presented topology. V₁ = V_dc, V₂=(z+2) * V_dc, V₃=(4z) *V₁, V₄=(z+2) *V₃. Negative output values are generated by the polarity inversion full bridge. The nearest-level modulation technique was employed and achieved low THD values. The topology can generate a high number of levels, however the modules use four voltage sources and cascading the modules increases the cost and complexity even further.

In (62) an asymmetric H-6 topology is proposed and depicted in Figure 17. The structure uses an H-6 inverter with a string of isolated DC sources on either side of the H-6 cell. The DC sources are controlled by the outer switches for level generation and the center switches S_H1-S_H6 are used for polarity inversion. The level generation switches operate at high PWM frequencies while the H-bridge switches operate at fundamental frequency to reduce the switching loss. Asymmetric voltage sources can be chosen for the DC source values yielding a high number of output levels. This is a highly complex topology and requires isolated voltage sources. In practical terms, obtaining different voltage ratios with photovoltaic systems or fuel cells can make the implementation expensive. It is suitable for battery operated inverters.

Siddique et al. proposed in (63) a new design of 11-level active neutral-point clamped (ANPC) inverter with high gain for photovoltaic applications. Figure 18 shows the inverter topology, consisting of a single voltage source, two switched capacitors, and 12 switching devices. Of the 12 switches, S₁, S₂, S₅, S₈, S_9, and S₁₀ have voltage ratings of the input DC source, S₃ and S₄ have half the DC source rating, switched S₆, S₇, S₁₁ and S₁₂ have voltage ratings twice the DC source voltage. The variability in the switch rating makes the inverter design complicated. Since the topology has a voltage gain of 2.5, it is well suited for use in photovoltaic systems and eliminates the need for a separate boost converter. The switch utilization, however, is low.

In (64), a 9-level SC inverter with voltage boost and capacitor self-balancing capability is proposed and shown in Figure 19. It consists of two switched capacitors, C1 and C2. C1 is charged to the input voltage and C2 is charged to 2Vin, through a series connection of C1 and the input voltage source. The topology is capable of switching at fundamental frequency, leading to efficient implementation. Having a gain of four eliminates the need for boost converter for photovoltaic applications. The topology’s main drawback is that it retains many switches. It boasts inherent capacitor voltage balancing but suffers from high inrush currents at startup.

In (65), a modified 7-level packed U-cell (PUC) inverter with voltage gain is presented and shown in Figure 20. The traditional 7-level PUC has a single voltage source V_dc and a capacitor source clamped to V_dc/3. The downside of this topology is the requirement for external closed-loop control for capacitor voltage balancing. The complexity is high. To solve the issue of voltage balancing, a PUC5 was invented along with a novel modulation method with inherent voltage balancing in (Abarzadeh et al., 2019). Both the PUC7 and the PUC5 are attractive with their low switch counts, but they lack voltage gain and as such not suitable for where voltage boosting is required as so in photovoltaic systems. The generic PUC is extensible but requires many voltage sources because the level generation is done for the most part by subtraction of voltage sources. In the modified PUC presented here, the capacitor is replaced by a voltage source with reversed polarity, and this allows the generation of higher voltage amplitude through the additive series connection of V₁ and V₂. In the case of modified PUC, V₁ = 2V₂, and the output voltages that are generated are 0, ±V_dc, ±2V_dc, ±3V_dc. The requirement for more than one voltage source is a disadvantage for this topology. An opportunity presents itself to explore a PUC7 with inherent voltage balancing and will be mentioned in the following section.

Table 3 summarizes the features, and points out the merits, and demerits of the reviewed MLI topologies. Table 4 shows a comparative analysis of the different topologies on the basis of the number of levels generated (N_L), number of switches (N_SW), number of diodes (N_D), number of magnetic transformers or inductors (N_TR/IND), the ratio of levels to the total number of components (N_L/N_COMP), and lastly a statement of the whether the reported topology employes asymmetric sources or not.It is crucial to understand that the overall cost of MLIs increase with a rise in the number of components, voltage sources and the complexity involved to implement the modulation strategies. The review looked at recently proposed inverters categorized as reduced switch count topologies. While galvanic isolation can be mandatory, the trend is in favor of nonisolated topologies as they are cost-effective. The topologies reported in (Ahmed et al., 2018; Samadaei et al., 2019; Islam et al., 2020; Meraj et al., 2021; Srinivasan et al., 2021; Dhanamjayulu et al., 2022b; Thakre et al., 2022; Hatas et al., 2023; Radhakrishnan et al., 2024) benefit from the use of asymmetric configurations to achieve high switch utilization or generate a high number of levels. This comes at the expense of a loss of modularity, which is highly desired to extend topologies for high voltage applications. The aforementioned topologies, utilizing multiple active voltage sources, present with high complexities as it is not the case that isolated voltage source are readily available. The cost of implementing multiple voltage sources is likely to be prohibitive. Single voltage source topologies utilizing capacitors to achieve voltage gain and level generation are attracting more attention from industry because or their low cost and manageable complexities. Many switched capacitor topologies are proposed, as those reviewed in this article, (Khan et al., 2022; Deng et al., 2023; Islam et al., 2023; Jena et al., 2023; Kumari et al., 2023), and increased research in this domain will lead to more optimal designs. Single source switched-capacitor topologies are favourable. In developing many-level SC topologies, soft-charging of capacitors becomes ever more important to be incorporated in inverters beyond five levels or single capacitor.

The focus of the review work was to identify opportunities for further development in multilevel inverter research. Insightful observations are made from the work presented in the paper.

⁃ Asymmetry is used extensively in the development of new reduced switch count topologies and its superiority is documented in the mentioned review article (Vinay Kumar and GowriManohar, 2023). Although it helps to generate more levels with a small number of switches as in the asymmetric CHB, and as evidenced from Table 20, the technique still retains the requirement for isolated active sources. The asymmetry in voltage sources is used in SC topologies such as (Samadaei et al., 2019) and others of the same type. It will be noticed that such topologies only use the additive operations possible for asymmetric topologies, to the exclusion of subtractive operations as in [56]. Including the subtraction of asymmetric voltage sources has the potential to generate more levels, especially in switched capacitor topologies, eliminating the need for isolated DC sources, and development in this direction can yield positive outcomes.

The focus of this work was to review reduced switch count and hybridized improvement types and identify challenges and opportunities to conduct research and development to improve multilevel inverters. The merits and drawbacks of classical topologies have been reviewed and presented as the primary motivation for the advancement and development of new reduced switch count and hybrid multilevel inverters. It is observed that even in the recent topologies, there still exist some challenges and opportunities for further development of multilevel inverters. The use of asymmetry has been a crucial technique in reducing the number of switches, and topologies with the highest output level count to the number of switches are asymmetric. However, asymmetric topologies have increased cost and complexity as they require many isolated active voltage sources. Elimination of additional active sources prompts the use of switched capacitors that employ the full set of operations applicable to asymmetric MLIs, addition, and subtraction of voltage sources. This remains unexplored in SC topologies and has the potential to achieve a high switch utilization as it does in isolated voltage source topologies. The problem of high inrush currents in multi-capacitor SC inverters still needs to be resolved with the developmemt of new topologies involving many capacitors. Resonant charging of capacitors in multilevel converters remains a challenge because of the high-frequency variable voltage transfer ratio. Fundamental frequency-modulated topologies present an opportunity to explore the resonant charging of the switched capacitors. The authors intend to explore the opportunities presented in this review article for further research and development and hope it to be helpful to other researchers as well.