5386 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 71, NO.

6, JUNE 2024

Two-Stage Three-Phase Transformerless Hybrid

Multilevel Inverter for Solar PV Application
Neha Tak , Student Member, IEEE, and Sumit K. Chattopadhyay , Member, IEEE

Abstract—The proposed inverter topology is emerged using clamping diodes, active switches are used in the active
from the multiple level-doubling-network (LDN) based neutral-point-clamped (ANPC) [6] topology. To prevent switch
topology for grid-connected solar photovoltaic (PV) sys- overvoltage, Zhuang et al. [7] proposed an ANPC zero-crossing
tem, where dc buses of three phases could not be merged
without electrical isolation. Three-phase T-neutral point switching method.
clamped (TNPC) is used to merge all the three phases A nonisolated converter is used along with single-phase full-
without transformer. Due to LDN operation, balancing the bridge inverter in [8] to effectively eliminate the leakage current
capacitors of TNPC (main bridge) is a major challenge. A [9]. However, this is achieved at the cost of one diode and two
dc–dc converter is customized in this work, which is nor- additional switches. This configuration is applicable for single
mally required for two-stage operation of solar PV inverter.
This converter topology reduces the size and losses of filter phase and may not be cost-effective solution for three-phase
inductors due to high-resolution inverter output voltage in system. Moreover, a greater number of switches are coming
addition to restricting the leakage current well below the rel- in the conduction path as compared to similar transformerless
evant standards. To validate the concept, experiments are topologies for PV application. Control of two-stage inverter
carried out using hardware prototype and those matched is presented in [10] and mainly addresses low-voltage ride-
well with simulation result.
through (LVRT) condition. LVRT in grid-connected PV inverter
Index Terms—Asymmetric multilevel converter topology, is achieved in most of the cases by deviating the PV array voltage
capacitor balancing, coupled inductor, leakage current, toward the open-circuit voltage during fault condition.
level doubling network (LDN), photovoltaic (PV), space vec-
tor modulation, T-neutral point clamped (TNPC).
A transformerless converter is developed in [11]. This con-
figuration provides low-cost converter with better efficiency. A
I. INTRODUCTION switched capacitor multilevel inverter is proposed in the litera-
ture, which improves power quality and reduces leakage current
RID-CONNECTED photovoltaic (PV) inverters have
G been developed and commercialized over the past few
years. Inverter topologies are currently being researched to
[12], [13]. In order to eliminate leakage current in five-level
cascaded H-bridge (CHB) multilevel inverter (MLI), a novel
PSPWM modulation technique is proposed in [14]. A two-stage
improve power quality, reliability, and efficiency while reducing transformerless T-neutral-point-clamped (TNPC) MLI, which
the number of components [1] to make cost-effective inverter consist of a three-level boost converter, is propsed to enhance
solution [2]. Due to insulation limit of PV module, solar PV array the efficiency and reduce the cost of inductor [15]. Neutral point
voltage must be restricted to 600 to 900 V, which means current converter is presented with some mechanism to limit leakage
has to be very high for multimegawatt converters. Chopping this current, common-mode voltage and to compensate neutral point
current at high frequency causes high switching losses mainly in [16].
two-level and three-level inverters. Tradeoff between switching Power frequency transformers are unavoidable in the topology
losses and power quality of a two-level inverter [3] makes it present in [17]. Voltage stress in all the switches will be equal
less desirable option for utility-scale PV applications. To meet to main dc-link voltage. In order to develop five levels in output
the high power handling requirement of large PV systems, voltage, 12 switches are used in [18]. Tradeoff is done between
multilevel converters are being considered as a viable solution. capacitor size and switching frequency in this work. A similar
One of the most widely used transformer-free topologies in topology is presented in the literature to generate 13 levels
solar PV systems is the neutral-point-clamped converter [4]. A in line-to-line voltage [19]. In this work, six-phase machine
neutral point controller is described in [5], which is expected is utilized to instantaneously remove the neutral-point voltage
to minimize the fluctuation of the neutral potential. Instead of fluctuations.
Multilevel inverters, such as CHB and modular multilevel
Manuscript received 3 February 2023; revised 15 April 2023 and 16
June 2023; accepted 5 July 2023. Date of publication 8 August 2023; converters, provide high power quality and efficiency [20],
date of current version 19 January 2024. (Corresponding author: Neha [21]. However, higher number of PV arrays are required to fed
Tak.) multiple H-bridges in CHBMLI [22]. When three phases are
The authors are with the Department of Energy Science and Engi-
neering, Indian Institute of Technology Delhi, New Delhi 110016, India fed by different PV array, it results in an unbalanced current
(e-mail: [email protected]; [email protected]). injection or power imbalance among the phases due to partial
Color versions of one or more figures in this article are available at shading. Various methods for resolving these issues are proposed
https://doi.org/10.1109/TIE.2023.3299027.
Digital Object Identifier 10.1109/TIE.2023.3299027 in [23]. Asymmetrical multilevel inverter [24] gives higher

0278-0046 © 2023 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See https://www.ieee.org/publications/rights/index.html for more information.

Authorized licensed use limited to: Corporacion Universitaria de la Costa. Downloaded on April 16,2024 at 00:05:31 UTC from IEEE Xplore. Restrictions apply.
TAK AND CHATTOPADHYAY: THREE-PHASE TRANSFORMERLESS HYBRID MULTILEVEL INVERTER FOR SOLAR PV APPLICATION 5387

power quality with same number of bridges as compared to

symmetrical CHBMLI. However, asymmetrical MLI bridges
[25] are fed by different PV array, which causes uneven power
distribution and interarray leakage current. More isolated dc
sources with different power ratings are needed for asymmetrical
multilevel inverter topologies and, moreover, these isolated dc
sources consume power in very different range, which causes
high cost of implementation. Maintaining power quality in input
side becomes difficult as power consumption is very different
in different bridges. Consequently, the voltage rating of the low
voltage cells and the need for bidirectional power sources are
some concerns.
Level doubling network (LDN) is proposed to reduce the
power rating of isolated auxiliary sources. In H-bridge, if half-
bridge is get added, then it helps to get double the number of out- Fig. 1. Circuit diagram of the prior art converter topology.
put voltage levels. Half-bridges are fed by capacitors and known
as LDN, and full-bridges are fed by dc sources. Capacitors are 4) A coupled inductor-based buck–boost converter is used as
self-balanced, and its voltage will be maintained at half of its first stage of the two-stage PV inverter topology in order
corresponding bridge voltage. This topology does not require to achieve NPC capacitor balancing without closed-loop
any closed-loop control to balance capacitor voltage. In the liter- control.
ature, it is found that 1:7 is the optimal ratio of asymmetry to get 5) In the proposed topology, the path of common mode
higher number of output voltage levels using concept of double current (reduced using a common mode choke) is dif-
LDN. In this work, auxiliary bridges are fed by bidirectional ferent from the PV array leakage current. Moreover,
isolated sources. Though power requirement by these sources both currents are maintained within standard safety
are just 7% of total rated power, however, it will contribute limits.
substantial cost to the converter. In this double LDN-based This article consists of six sections. Section I depicts motiva-
topology [26], dc buses of main bridges are merged with the help tion and challenges to use multilevel inverter in grid-connected
of three isolation transformers to reduce capacitor requirement PV systems. Section II gives brief idea of double LDN-based
and to prevent current imbalance in phases. However, the use topology with transformer. Section III explains different con-
of power transformers reduces system efficiency and increases verter configuration for hybrid multilevel inverter topology to
the cost, weight, and volume of the system. In [26], isolated develop efficient, cost-effective, and high-performance inverters
auxiliary sources are used, though the power requirement by that meet PV system requirements. Capacitor balancing method
these isolated sources is very less but it get reflected in total for TNPC is also explained in detail in this section. Section IV
system cost. These isolated sources have been removed in [27] discusses the simulation results. Section V validates the con-
with the help of low-gain PI controller. In this converter, only cept through hardware results. Section VI presents an in-depth
main bridge is fed by dc source. As observed in aforementioned analysis on performance of the converter based on analysis, sim-
literatures, in order to develop a practical inverter solution, main ulation, and experimental results. Finally, Section VII concludes
focus is given to remove transformer, mitigate leakage current, this article.
reduce part count, and semiconductor losses. The work reported
in this article addresses all the aforementioned aspects along
II. TOPOLOGY TO REMOVE TRANSFORMER AND ITS
with high power quality, reliability, and efficiency. The following
BACKGROUND
contributions are made in order to achieve the aforementioned
attributes in a single PV inverter. LDN-based topologies are useful to achieve high resolution
1) Double LDN-based high-resolution multilevel inverter in output voltage with limited switches. In prior art [26], main
is modified in this work for two-stage grid-connected bridge dc buses of three phases are merged in order to reduce
solar PV application. This architecture doubles the power capacitor requirement and to prevent unbalance current injection
handling capability of the inverter by incorporating a due to partial shading. To merge three phases, three isolation
dc–dc converter. transformers are required, as shown in Fig. 1, which makes
2) For central inverter applications, the dc-buses of three system bulky as well as expensive. Isolation transformers are
phases are merged without using isolation transformer to replaced in the presented work by using three-phase TNPC
decrease capacitor ripple and to reduce the probability topology along with a customized dc–dc converter for two-stage
of interarray leakage current. As three-isolation power PV application. This topology is not only useful to merge the
frequency transformers of the full-scale converter rating three PV arrays without isolation transformer, but also helpful to
are removed in this work, it is expected to reduce the limit the leakage current. Complete circuit diagram of the hybrid
implementation cost by 35%–50%. converter topology is illustrated in Fig. 2.
3) In this work, in order to remove transformers of full con- This hybrid power converter needs 12 switches per phase
verter rating, main H-bridges are replaced by three-phase and only one dc source in complete converter topology to
TNPC. reach optimal voltage resolution. The inverter section of the
Authorized licensed use limited to: Corporacion Universitaria de la Costa. Downloaded on April 16,2024 at 00:05:31 UTC from IEEE Xplore. Restrictions apply.
5388 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 71, NO. 6, JUNE 2024

using multiple redundancies. Hence, only one set of space vector

combination is used to balance the LDN capacitors. Moreover, it
brings the challenge to equalize the charging and discharging of
upper and lower capacitors “C1 ” and “C2 ,” respectively. Either
upper capacitor will supply power to LDN or lower capacitor will
supply power depending on the topological configuration. As
presented in [26], lower capacitor of TNPC will always supply
power to the LDN. Hence, it brings a limitation due to different
rate of discharging in both the capacitors. In absence of some
additional balancing circuit between C1 and C2 , the upper and
lower capacitors will substantially diverge. Unequal distribution
of voltage at upper and lower dc bus will not be able to maintain
the desired voltage across main and auxiliary LDN, which will
distort the output line-to-line voltage to an unacceptable level.
Hence, it is required to have some additional circuit with proper
Fig. 2. Proposed two-stage grid-connected PV inverter topology.
controller to equalize the charging and discharging of upper
and lower capacitors. For two-stage grid-connected PV system,
presented topology can generate high-resolution output voltage, dc–dc converter is modified and customized in such a way that
as discussed in [27]. The aim of this work is to merge dc buses no extra controller and converter will be needed to balance
of three phases of main bridges without isolation transformer. the capacitor voltages, as shown in Fig. 2. Operation of the
This converter is developed by hybridizing TNPC topology and customized dc–dc converter is presented in the next section.
LDN modules. It is relevant to note that TNPC topology is more
suitable for PV application (due to limited PV array voltage) III. GRID-CONNECTED TWO-STAGE OPERATION OF THE
compared to any other neutral-point-clamped topologies. PROPOSED HYBRID TOPOLOGY
In proposed topology, TNPC and main LDN are of value 20
Proposed work is specifically for two-stage grid-connected
and 5 V, respectively, and its combination is known as high-
PV system where dc–dc converter is important part of complete
voltage module. Low-voltage module that is auxiliary bridge
system, and it cannot be avoided in order to operate inverter
and auxiliary LDN is of value 2 and 1 V, respectively. Auxiliary
around unity modulation index and to have decoupled control
bridge needs closed-loop control to balance the capacitor volt-
of maximum power point tracking (MPPT) and grid current.
age, whereas LDNs are self-balanced, and they do not require
This dc–dc converter helps to double the inverter power handling
any closed-loop control. LDN voltage converges to half of
capacity for grid-connected solar PV application compared to a
its corresponding bridge voltage through self-balancing with-
single-stage PV inverter.
out closed-loop control. Main module work on fundamental
In presented work, the coupled inductor-based buck–boost
switching frequency and auxiliary module operates at relatively
converter itself is used to equalize the voltages of the upper
higher switching frequency. Main bridge and auxiliary bridge
and lower capacitors. Relation between voltage and flux and
maintain 1:5 ratio of asymmetry to remove auxiliary voltage
inductance of a single inductor coil is given in (1) and (2).
sources. To remove auxiliary sources, it is necessary to ensure
Hence, the voltage across single inductor coil can be calculated
that fundamental component of input reference voltage should
using (3). However, for coupled inductor, flux (Ф22 ) generated
match with fundamental component of output of main module
by secondary current (i2 ) known as mutual flux (Ф12 = Ф21 ) in
in the steady state. Hence, input of auxiliary module will not
the secondary winding is also linked to the first winding. The
have any fundamental component.
rate of change of current in both primary and secondary windings
TNPC is preferred over NPC as in curent path, the number of
will give voltage across primary (VL1 ) and secondary windings
semiconductor devices are lower in three-level TNPC as com-
(VL2 ), as given in (4) and (5)
pared to three-level NPC to achieve positive and negative volatge
levels. However, the voltage stress across TNPC switches is dΦ
V =N (1)
double as compared to NPC. Though conduction loss can be dt
reduced by selecting appropriate switches. TNPC is fed by two NΦ
floating capacitors connected to neutral point of the converter. L= (2)
i
Voltage across these two floating capacitors, i.e., C1 and C2 ,
di
should be maintained at half of the main dc-bus voltage. V =L (3)
Proposed hybrid topology brings the challenge of capacitor dt
balancing due to LDN operation. In the literature to balance di1 di2
VL1 = L11 + L12 (4)
the upper and lower capacitors of NPC-based topologies, space dt dt
vector redundancies are selected in such a way that charging N1 Φ11 N1 Φ12
where L11 = i1 and L12 = i2
and discharging will take place from both the capacitors to
maintain equal voltage across capacitors. However, in order to di1 di2
VL2 = L21 + L22 . (5)
satisfy LDN principle, it is not possible to form space vector dt dt

Authorized licensed use limited to: Corporacion Universitaria de la Costa. Downloaded on April 16,2024 at 00:05:31 UTC from IEEE Xplore. Restrictions apply.
TAK AND CHATTOPADHYAY: THREE-PHASE TRANSFORMERLESS HYBRID MULTILEVEL INVERTER FOR SOLAR PV APPLICATION 5389

Fig. 3. Circuit diagram of customized dc–dc converter with single-

phase TNPC and LDN.
Fig. 5. LDN operation. (a) LDN (half-bridge) charging through lower
capacitor of TNPC. (b) LDN discharging through upper capacitor of
TNPC.

Lb2 will be as high as Vpv or Vdc (neglecting loss due to leakage

and stray parameters). Turns ratio of coupled inductor also help
to determine the blocking voltage of diode Db1 and Db2 . Like
a buck–boost converter, the inductor currents ILb1 and ILb2 of
coupled inductor will vary with load
diLb1
VLb1 |Sb =1 = Vin = Lb1 (6)
dt
 
diLb1  ΔiLb1 ΔiLb1 ΔiLb1  Vin
= = or =
Fig. 4. Working of customized buck–boost converter circuit to equalize dt Sb =1 Δt DT DT Sb =1 Lb1
the voltage across upper and lower capacitors. (a) Schematic diagram.
(b) Charging of lower capacitor C2 when (Vc2 <Vc1 ) and charging of DT Vin
(ΔiLb1 )|Sb =1 = (7)
upper capacitor C1 when (Vc2 >Vc1 ). Lb1
VDb1 |Sb =1 = VLb2 + VC1 (8)
Customized dc–dc converter consists of one switch “Sb ,”
coupled inductors “Lb1 ” and “Lb2 ,” and diodes “Db1 ” and “Db2 ,” VDb2 |Sb =1 = VLb1 + VC2 (9)
which will help to boost the PV voltage in present case as well NS NS
as to balance the capacitor voltages. No additional capacitors VLb2 |Sb =0 = VLb1 = Vin . (10)
NP NP
are needed for this dc–dc converter because the capacitors used
in the main converter will be used in this dc–dc converter. When switch “Sb ” is ON, the inductor current will flow on
Circuit diagram of a customized coupled inductor-based dc–dc one side, and capacitors on the other side will be discharged
converter with single-phase TNPC and LDN is shown in Fig. 3. in response to the load. Lower capacitor C2 is having higher
Working of capacitor voltage balancing is explained through discharge rate owing to LDN operation as per the polarity of
Fig. 4. For capacitor voltage balancing, two cases are considered, LDN, as depicted in Fig. 5(a) and (b). In the present topo-
i.e., ideal and nonideal situations. For initial investigation, it logical configuration, LDN charges in the negative half cycle
is assumed that there is no parametric difference in terms of and discharges in a positive half cycle, as shown in Fig. 5(a)
stray resistance and inductance and only ideal components are and (b), which means in the negative half cycle, power goes
considered. from the lower capacitor of TNPC to charge LDN capacitor.
Current will flow through inductor Lb1 when switch “Sb ” is For this configuration, lower dc-bus capacitor will discharge
closed, as shown in Fig. 4(a). The polarity of “Lb1 ” will be more than the upper dc-bus capacitor. However, if LDN has
determined by the coupled inductor. When Sb is turned ON, given negative voltage with respect to the ground, this situation
current will flow through Lb1 irrespective of the three voltages, could have been reversed. Hence, any configuration can be
namely Vdc , VC1 , and VC2 , as given in (6) and (7). Furthermore, taken for this converter topology, and accordingly charging and
regardless of the voltage on the upper capacitor “C1 ,” the diode discharging rates of capacitors will be decided. In present work,
“Db1 ” will be reverse biased by the reflected voltage polarity on lower capacitor C2 is having higher discharge rate. The moment
Lb2 . The blocking voltage of diode Db1 will be equal to voltage switch Sb will be turned OFF and current will be developed in
appearing at Lb2 and the voltage of upper capacitor C1 , as given inductor as per polarity. In this situation, inductor voltage will be
in (8). Similarly, blocking voltage of Db2 will be decided on the clamped to the lower capacitor voltage, as depicted in Fig. 4(b).
basis of voltage across Lb1 and C2 , as given in (9). Voltage across For example, when VC2 is less than VC1 , then current will be
Lb2 will be decided based on turns ratio of coupled inductor, as able to flow in loop 2, i.e., through Lb1 , Db2 , and C2 . Relation
given in (10). When 1:1 ratio is considered, the voltage across between capacitor voltage VC2 and input voltage (Vin = Vdc )

Authorized licensed use limited to: Corporacion Universitaria de la Costa. Downloaded on April 16,2024 at 00:05:31 UTC from IEEE Xplore. Restrictions apply.
5390 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 71, NO. 6, JUNE 2024

is given in (11)–(14). Current can flow in a particular loop only TABLE I

SIMULATION PARAMETERS
when inductor voltage is clamped to capacitor voltage, as shown
in Fig. 4(b). Since Lb1 and Lb2 are mutually coupled, rate of
change of flux will be common for them (neglecting the effect
of stray resistance and leakage inductance). There can be other
situation when VC1 is less than VC2 , then current will be able
to flow in loop 1, i.e., through Lb2 , Db1 , and C1 , then change
in inductor current and VC2 can be calculated using (15)–(17).
However, this situation will not occur as LDN is charging from
lower capacitor C2 (resulting into higher discharge rate of C2 ).
At steady state, both the capacitor voltages will be equal and
accordingly number of turns will be decided, as given in (18)
and (19)
VLb1 |Sb =0 = Vc2 (11)

diLb1 
Lb1 = Vc2 (12)
dt Sb =0
Vc2
(ΔiLb1 )|Sb =0 = (1−D)T
Lb1 (13)
(ΔiLb1 )|Sb =1 + (ΔiLb1 )|Sb =0 = 0
D Fig. 6. Voltage waveforms without balancing circuit. (a) Upper and
Vc2 = −Vin (14) lower capacitors. (b) Main LDN and auxiliary LDN capacitors.
(1 − D)
VLb2 |Sb =0 = Vc1 (15)
 will be equalized. The moment voltage across both capacitors
diLb1 NS 
Lb1 = Vc1 (16) will equalize, the inductor current will be distributed equally
dt NP Sb =0 across both the capacitors. Then, the voltage in both capacitors
(1 − D) T Vc1 NP will begin to rise uniformly. At steady state, both the capacitor
(ΔiLb1 )|Sb =0 = (17) voltages given in (21) and (22) will get equalized, then it will
Lb1 NS
help to calculate the number of turns in primary as well as in
NS
Vc1 − Vc2 = VLb1 − VLb1 (18) secondary
NP
∫0DTs (Vin − r1 iLb1 (t))
NP = NS . (19) (ΔiLb1 )|Sb =1 = (20)
(Lb1 + Lleak1 )
Current distribution in both the capacitors will get affected by  2
diLb2 NS diLb2
nonidealities, such as stray resistance and leakage inductance. VC1 = Lb2 + Lleak2
In proposed circuit configuration, current in leakage inductance dt NP dt
will have path to flow through diodes Db1 and Db2 , as shown − r2 iLb2  (21)
in Fig. 4. Hence, snubber circuit is not required to handle the
leakage current in the proposed converter topology. The total diLb1 diLb1
VC2 = Lb1 + Lleak1 − r1 iLb1  . (22)
stray resistance (diode and inductor) and leakage inductance dt dt
in loops 1 and 2 is r1 and r2 , Lleak1 and Lleak2 , respectively.
However, for nonideal condition, when stray parameters and IV. SIMULATION RESULTS
leakage are considered, inductor voltage and voltage across The analysis of proposed transformerless hybrid asymmetri-
capacitor C2 will slightly differ in compared to ideal situation, cal multilevel inverter is done using MATLAB/Simulink. The
as given in (20)–(22). So, when VC1 > VLb1 , it will not allow main purpose of this simulation is to test a double LDN-based
current to flow in that particular loop. In this situation, lower inverter for two-stage grid-connected solar PV applications to
capacitor C2 will carry current through inductor Lb1 . However, merge three phases without an isolation transformer. The simu-
its coupled version is unable to supply any current, despite the lation parameters are presented in Table I.
voltage being impressed. This process will keep on going as In absence of some balancing mechanism to have a local
long as these two capacitor voltages are not getting equalized. power flow between C1 and C2 , the upper capacitor C1 and
While discharging takes place continuously in C2 due to LDN lower capacitor C2 will not be able to maintain its voltage and
operation, this capacitor will share majority of charging cur- voltage of both the capacitors will be diverge from its desired
rent from the inductor during the switch Sb is OFF. However, voltage, as shown in Fig. 6(a). Due to voltage imbalance in main
discharging can be through both the capacitors, and it varies dc-link capacitors, the main LDN and auxiliary LDN capacitors
at different instances. The lower capacitor will charge through will continue to discharge until they reach zero at steady state,
inductor Lb1 , and at some point, the voltages of both capacitors as shown in Fig. 6(b). Due to divergence of capacitor voltages

Authorized licensed use limited to: Corporacion Universitaria de la Costa. Downloaded on April 16,2024 at 00:05:31 UTC from IEEE Xplore. Restrictions apply.
TAK AND CHATTOPADHYAY: THREE-PHASE TRANSFORMERLESS HYBRID MULTILEVEL INVERTER FOR SOLAR PV APPLICATION 5391

Fig. 7. Voltage across upper and lower capacitors for grid-connected

system with customized coupled inductor-based dc–dc converter for
MPPT tracking. Fig. 10. Line voltage and corresponding phase voltage @ grid voltage
of 840 V.

Fig. 11. Grid voltage and current at unity power factor.

Fig. 8. Voltage across main LDN, main bridge, and auxiliary LDN
capacitors for the grid-connected system.

Fig. 9. Inverter output line-to-line voltage.

from their required value in inverter, output ac voltage and

Fig. 12. Impact of insolation on capacitor balancing, power, and grid
subsequently grid current will be distorted far beyond the accept- current.
able limit. With customized coupled inductor-based converter
for MPPT tracking, voltage across upper and lower capacitors
will also be equalized and maintained at desired voltage for to adjust direct component of grid current to stabilize the main
grid-connected system, as shown in Fig. 7. Due to equal voltage dc-link voltage. Fig. 12 represents the inverter operation with
at upper and lower capacitors, LDN capacitor voltage will be pure active power injection into the grid.
maintained at half of the corresponding bridge voltage. Fig. 8 Fig. 13 indicates the impact of load-side current and TNPC
depicts the balanced capacitor voltage across the main LDN, switching on the capacitor balancing. C1 and C2 are having
the auxiliary bridge, and the auxiliary LDN for a PV-fed grid- different currents, i.e., IC1 and IC2 , and current difference will
connected system without any addidtional balancing circuit. be determined by the switching combinations of three phases
High-resolution inverter ouput voltage is shown in Fig. 9. together. Impact of TNPC output voltage and load current on
Line voltage and corresponding phase voltage at a given grid IC1 , IC2 , VC1 , and VC2 for one cycle is shown in Fig. 13(a). Its
voltage (+), i.e., at 0.85 modulation index, are shown in Fig. 10. expanded view is illustrated in Fig. 13(b) to analyze the capacitor
Line voltage waveform is near to sinusoidal, and phase voltage voltage balancing. A particular instance is considered to analyze
contains triplen harmonics, which is getting cancelled out in different states. Combination of different phase output voltages
line voltage. Grid voltage and current waveform for a given can be categorized, as shown in Fig. 13(b), and its detailed study
modulation index at unity power factor operation are shown in is given in case 1 to case 4, which are as follows.
Fig. 11.
Fig. 12 depicts the impact of varying loads on capacitor Case 1: During T0 <T<T1 , R and B phases clamped to negative
balancing. It can be observed from Fig. 12 that after 3 s, solar dc bus and Y phase to positive dc bus. IC1 will be equal to IC2
insolation is reduced to half, which finally reduces the output at such condition as a result, VC1 and VC2 will converge to
power and grid current correspondingly. A PI controller is used Vdc /2. Same situation can be observed during T2 <T<T3 .

Authorized licensed use limited to: Corporacion Universitaria de la Costa. Downloaded on April 16,2024 at 00:05:31 UTC from IEEE Xplore. Restrictions apply.
5392 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 71, NO. 6, JUNE 2024

Fig. 13. Impact of TNPC voltage and load current on voltage and
current of upper and lower capacitors. (a) Overall view for one cycle.
(b) Expanded view. Fig. 14. Impact of nonidealities on coupled inductor currents Lb1 and
Lb2 , capacitor currents Ic1 and Ic2 , and voltage across capacitors Vc1
and Vc2 (a) with stray resistance and (b) with leakage inductance.

Case 2: During T1 <T<T2 , Y phase will be clamped to neutral,

and R and B phases to negative dc bus. At this instance, C1
will not be connected to load, whereas C2 will observe the regardless of whether the switch is turned ON or OFF. While in
load impedance. Lb2 , the current continues to flow when the switch is turned OFF.
Total current will flow through C2 , and capacitor voltage will This statement is true regardless of the kind of discharge in the
diverge till next switching, and this is the worst case. Diver- upper and lower capacitors. Fig. 14 depicts the impact of stray
gence of VC1 and VC2 is restricted by natural adjustment in parameters on current distribution in coupled inductor and both
current magnitude of IC1 and IC2 , respectively. the capacitors.
Case 3: During T3 <T<T4 , Y phase clamped to positive dc bus, Capacitor with higher voltage will have smaller charging
B phase to neutral, and R phase to negative dc bus. C1 and current and vice versa. Ideally, no charging current from dc–dc
C2 will observe unequal loading and consequently discharge converter should flow into C1 , as voltage across C1 is greater than
rates of C1 and C2 are different. Charging current from the C2 . However, both capacitors will carry charging current when
dc–dc converter IC1 and IC2 is naturally adjusted to limit the switch Sb is turned OFF. Distribution of charging current between
divergence of voltage at this stage. C1 and C2 depends on difference of those capacitor voltages,
respectively. The cause of voltage deviation, as explained in
The aforementioned analysis gives an idea about the effect Fig. 13, is time varying. So, entire window is difficult to illustrate
of different output voltage vectors on capacitor balancing and prominently. Hence, an arbitrary instance has been taken in
indicates that the capacitor voltages will remain balanced at all Fig. 14. In Fig. 14(a), it is clearly shown that at the instance
operating conditions. In dc–dc converter circuit, current ILb1 between T1 and T2 , C1 and C2 are having negative value of
and ILb2 will be different in both the inductors as per switching current. Say, at T2 instance, IC1 is zero and IC2 has some
action of Sb , as shown in Fig. 14. Inductor Lb1 is expected to value (IC2 is following load-side current). Since IC1 is zero and
have a higher average current because it is always conducting, IC2 has some value that means C2 is discharging more. Here,

Authorized licensed use limited to: Corporacion Universitaria de la Costa. Downloaded on April 16,2024 at 00:05:31 UTC from IEEE Xplore. Restrictions apply.
TAK AND CHATTOPADHYAY: THREE-PHASE TRANSFORMERLESS HYBRID MULTILEVEL INVERTER FOR SOLAR PV APPLICATION 5393

Fig. 15. Leakage current at full load in the solar PV array.

initially capacitor C2 is discharged higher so entire charging

current entered into capacitor C2 and then as time lapses, voltage
Fig. 16. Pictorial view of hardware setup for grid-connected system.
across C2 increases and both VC1 and VC2 increase. However, (a) Control desk. (b) Oscilloscope. (c) Additional buck–boost converter.
VC2 is increasing at higher rate. After certain instance at T3 , (d) Main module. (e) Auxiliary module. (f) NPC formed with discrete com-
they get equalized and then current IC2 is falling and IC1 is ponents. (g) Main LDN. (h) Synchronization switch. (i) Line inductors.
(j) Grid.
rising. When the difference between VC1 and VC2 , i.e., ΔV,
is larger, then it will affect the current distribution in IC1 and
IC2 at charging end, as shown in Fig. 14(a). Discharging is TABLE II
determined by the load. It can be observed that the current is EXPERIMENTAL SETUP PARAMETERS
not proportionally increasing in every cycle, as it depends upon
percentage of voltage difference in total capacitor voltage. At the
same time, leakage and other nonidealities are also constraints
for balancing the system. Appropriate balancing can be achieved
by adding few compensations turns in Lb1 winding. Capacitor
voltage balancing, inductor, and capacitor currents after con-
sidering leakage inductance case are shown in Fig. 14(b). In
proposed converter topology, leakage current is negligible since
the leakage capacitance (CL ) comes in parallel to the dc-link
capacitance (which is several hundred times bigger than the
leakage capacitance). Hence, no additional circuit is added to
restrict the leakage current, as shown in Fig. 2. Leakage current
in steady state is illustrated in Fig. 15. Peak magnitude of leakage
current is restricted to 0.15 A against rated ac-side peak current
of 85 A, i.e., leakage current is less than 0.2%.
The proposed topology can be directly connected to the dis-
tribution transformer, i.e., usually star grounded at low-voltage
side, which will cause common-mode current in three phases.
However, it may be found from Fig. 2 that path of common-
mode current will be different from the PV array leakage
current. Common-mode current can be effectively suppressed
Fig. 17. Voltage waveforms. (a) Inverter three-phase line-to-line volt-
within limit by common-mode choke or grid formed by delta- age. (b) Synchronization with polluted grid channel 1: Inverter voltage;
connected transformer. In the proposed topology, the magnitude channel 2: Grid voltage; math function: Difference between inverter
of common-mode current is less than 1% of the line current. voltage and grid voltage.
Hence, the effect of this common-mode current on output current
THD is negligible.
neutral-point-clamped configuration instead of TNPC, as shown
in Fig. 16. Parameters used in hardware are shown in Table II.
V. EXPERIMENTAL RESULTS
Fig. 17(a) illustrates balanced inverter output line-to-line volt-
Prototype of this inverter topology is developed in laboratory age at around 0.85 modulation index. The converter is designed
to verify the concept presented in earlier sections. dSPACE such that it will synchronize with the grid at its nominal voltage at
MicroLabBox is used for modulation and control of this inverter this modulation index. This is done to keep sufficient headroom
prototype. A three-phase NPC inverter is fed by a common dc for reactive power injection and to take care of the nonidealities.
source of 320 V and both the capacitors of NPC are maintained This hybrid converter is connected to the grid and its perfor-
at half of dc-link voltage, i.e., 160 V each. A pictorial view mance is verified with polluted grid, as shown in Fig. 17(b).
of the hardware setup for the grid-connected system and key Inverter line-to-line voltage and grid voltage are shown in
components of the hardware prototype is given in Fig. 16. In channels 1 and 2, respectively. It is important to note that even if
addition to this, six-pulse rectifier is used to supply the input the grid voltage is distorted, difference between inverter voltage
of the dc–dc converter. Experiments are conducted using a and grid voltage is not showing substantial distortion, as shown

Authorized licensed use limited to: Corporacion Universitaria de la Costa. Downloaded on April 16,2024 at 00:05:31 UTC from IEEE Xplore. Restrictions apply.
5394 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 71, NO. 6, JUNE 2024

Fig. 18. Inverter line voltage and corresponding phase voltages @ grid
voltage of 200 V for (a) three cycles. (b) Zoomed view.
Fig. 21. Capacitor voltage variation under influence of load change
channel 1: Main LDN voltage; channel 2: Auxiliary bridge voltage; chan-
nel 3: Auxiliary LDN voltage; and channel 4: Load current.

Fig. 19. Inverter output performance channel 1: Line voltage; channel

2: Load-side phase voltage; channel 4: Line current at modulation index.
(a) Unity. (b) 0.5.

Fig. 22. TNPC output voltage channel 1: R phase voltage; channel

2: Y phase voltage; channel 3: B phase voltage. (a) Complete cycle.
(b) Possible voltage combination of three phases.

the capacitor voltages of LDN. However, due to limited number

of channels in oscilloscope, it is not possible to illustrate voltage
across both the capacitors. Similarly, auxiliary bridge voltage
(32 V) will be maintained as per 1:5 ratio of asymmetry w.r.t.
voltage across TNPC capacitor (160 V), as shown in Fig. 20(a).
Capacitor balancing due to variation in modulation index, i.e.,
Fig. 20. Capacitor balancing channel 1: TNPC dc bus to neutral-point
voltage; channel 2: Main LDN voltage; channel 3: Auxiliary bridge volt- from unity to 0.9 modulation, is shown in Fig. 20(b). It is evi-
age; channel 4: Auxiliary LDN voltage. (a) Cold start. (b) At steady-state dent from Fig. 20(b) that voltage across capacitors is stabilized
variation from unity to 0.9 modulation index. without any significant deviation.
Impact of load variation on capacitor balancing is shown
in Fig. 21 to demonstrate the performance of inverter during
in the math menu of Fig. 17(b). Inverter voltage is developed in dynamic operation. Load is changed from 200 to 600 W at time
such a way that the effect of grid harmonics on output current is instance T1 and 600 to 1000 W at time instance T2 . This may be
nullified. observed that there is variation in capacitor voltage ripple w.r.t.
Inverter line voltage and its corresponding phase voltages are load variation. However, average values of capacitor voltage
shown in Fig. 18. It can be observed that the phase voltage remain unchanged.
consists of triplen harmonics (generated by space vector modula- In proposed converter topology, capacitors do not require
tion) to satisfy floating capacitor balancing operation, which gets any precharging circuit as all the floating capacitors are getting
cancelled out in line-to-line voltage. Fig. 18(a) and (b) depicts balanced through the load. However, it is important to note
inverter line and phase voltage for three cycles and its zoomed that LDNs are self-balanced in open loop and auxiliary bridges
view, respectively. Inverter output voltage and current for two require closed-loop control to reach desired capacitor voltage
different modulation index are shown in Fig. 19. Fig. 19(a) after cold start.
represents inverter output voltage at unity modulation index. Voltage across individual phases of TNPC is shown in Fig. 22.
Channel 1 represents 43 levels in output line-to-line voltage, An overall view of three TNPC phase voltages is illustrated in
channel 2 represents load-side phase voltage, and channel 4 Fig. 22(a), whereas Fig. 22(b) displays possible TNPC output
represents the line current. Capacitor balancing and voltage voltage combinations in an expanded view. This matches well
across all the floating capacitors are shown in Fig. 20. Start-up with the results shown in simulation.
transients for all floating capacitors are depicted in Fig. 20(a). Impact of load-side current and switching of main bridge
All floating capacitors converge toward their desired values. At on the balancing of capacitor is shown in Fig. 23. Voltage
instant T1 , dc-bus is charged and at instant T2 , inverter operation across upper and lower capacitors is shown in channels 1 and 2,
is started in Fig. 20(a). No closed-loop control is used to balance respectively. Both the capacitor voltages are equalized, and their

Authorized licensed use limited to: Corporacion Universitaria de la Costa. Downloaded on April 16,2024 at 00:05:31 UTC from IEEE Xplore. Restrictions apply.
TAK AND CHATTOPADHYAY: THREE-PHASE TRANSFORMERLESS HYBRID MULTILEVEL INVERTER FOR SOLAR PV APPLICATION 5395

TABLE III
COST COMPARISON

Fig. 23. Channel 1: Upper capacitor voltages; channel 2: Lower ca- count (no. of gate drivers, switches, and heat sinks) will remain
pacitor voltage; channel 3: ILb1 current; channel 4: ILb2 current; math same w.r.t. prior art shown in Fig. 1. Moreover, increase in VA
function: Sum of ILb1 and ILb2. rating of power semiconductor switches (increase in chip area)
does not necessarily result into proportional increase in cost due
to other materials and manufacturing cost (nearly fixed cost)
respective voltage is maintained at 160 V. Resultant inductor associated with a power semiconductor module. Moreover, it is
current (sum of current flowing in both the inductor windings) important to note that the cost of module is a smaller fraction
is shown in red color at a given instance and it is in continuous compared to total cost of the system, which includes cost related
conduction mode. It is found by summing up the two winding to controller, cabinet, GUI, heat sink, cabling, assembly, etc.
currents of the coupled inductor. Cost of these components will remain almost unaffected. In-
Inductor winding currents ILb1 and ILb2 are shown in channel verter cost is determined by number of components. Component
3 and channel 4, respectively. When switch “Sb ” of Fig. 2 is “ON,” cost keeps on changing with power rating, component availabil-
then the diode corresponding to upper capacitor (C1 ) comes in ity, inflation, and geographical location of the manufacturer and
reverse bias and ILb2 current immediately falls to zero. At the buyer. An indicative figure is provided in Table III on the basis of
same time, inductor current will be transferred from ILb2 to ILb1 . statistical data related to transformer and inverter cost for same
So, there is a step rise in the inductor current in Lb1 winding and power rating.
this winding is going to observe the PV array voltage. As long as In a typical transformer-isolated PV inverter, transformer cost
switch is ON, ILB1 will keep on increasing and ILB2 will remain is near about 40% of total system cost. Overall cost of inverter
zero. The moment switch is turned OFF, slope of rise and fall of owing to higher power rating of switches (only for main bridge
ILb1 and ILb2 will be determined by capacitor voltages, leakage and dc–dc converter) is expected to increase the total converter
inductance, and stray resistances. Due to leakage inductance of cost by less than 10%. In standard inverter, 65% is the cost of
the respective windings, current will not change immediately. power electronic circuitry, so in presented work, converter cost
It can be concluded that Lb2 on average is going to carry much may increase to 72% approximately. Hence, a tentative idea of
smaller fraction of current compared to Lb1 at all conditions. total system cost is highlighted in Table III. A power frequency
Inductor Lb1 is expected to have a higher average current because transformer also contributes a major components of system
it is always conducting, regardless of whether the switch is losses. Hence, removal of transformer, small size of filter (owing
turned ON or OFF. This statement is true regardless of the kind to high-resolution output voltage), and low switching losses
of discharge in the upper and lower capacitors. together can ensure better efficiency or at least no compromise
in efficiency other than its superior performance, as discussed
VI. DISCUSSION in earlier sections.
Usefulness of the presented two-stage high-resolution PV
inverter is analyzed in this section from component requirement, B. Leakage Current
leakage current, and efficiency perspective, respectively, as dis- PV inverters connected to the grid without a transformer
cussed in the following sections. present a significant challenge when it comes to restrict the PV
array leakage current, which flows through terminal to frame
A. Component Requirement and Efficiency capacitance known as leakage capacitor. Leakage current limits
are specified by the German VDE0126-1-1 standard. For low
There are number of factors that affect the cost of a power power rating, as per standards, common-mode leakage current
electronic converter. However, among them, two factors are should be below 300 mA. However, in presented work, even
considered, which affects the cost substantially, which are men- for high power (70 kWp), leakage current is restricted below
tioned as follows. 200 mA, as shown in Fig. 11.
1) Number of components.
2) Rating of components.
C. Comparison of Well-Known Multilevel Inverter
There is about 1.3 times increment in cumulative volt–ampere
(VA) rating of the power semiconductor switches (due to TNPC Topologies
converter) in the inverter part and nearly 2 times in the dc–dc In order to evaluate the merits and limitations of the proposed
converter part (due to high blocking voltage of switch used in hybrid converter topology, well-known converter topologies for
coupled inductor-based dc–dc converter). However, component two-stage grid-connected solar PV applications are compared

Authorized licensed use limited to: Corporacion Universitaria de la Costa. Downloaded on April 16,2024 at 00:05:31 UTC from IEEE Xplore. Restrictions apply.
5396 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 71, NO. 6, JUNE 2024

TABLE IV
COMPARISON OF VARIOUS TWO-STAGE CONVERTER TOPOLOGIES

to the number of output voltage levels. The topologies have cost. Removal of isolation transformer of full converter rating
been compared using a baseline of 12 switches (per phase). will reduce the overall implementation cost and increase the
In this comparison, it is considered that dc buses of three efficiency of PV plant with some tradeoff in CVA rating.
phases are merged in order to limit capacitor voltage ripple. The switch utilization factor (SUF) is analysed with respect
All the three phases are fed by a common PV array, which to a standard single-phase full-bridge inverter. For which the
will eliminate the interarray leakage current. However, three CVA rating of the switches is eight times higher than the VA
isolation transformers are required in order to merge three dc rating of the inverter. SUF is inversely proportional to the ratio
buses. Whereas, in present with this three-phase single-sourced of TCVAR of the H-bridge to the ratio of TCVAR of the given
double-stage transformerless converter topology in Table IV. converter topology. It is important to note that keeping the
One of the most important aspects of the proposed MLI topology number of switches unchanged with reduced utilization factor
is the elimination of isolated auxiliary dc power sources through is not expected to increase the implementation cost of the total
a fair tradeoff in converter topology, isolation transformers are system in linear proportion
removed, three-phase H-bridges are replaced by TNPC, and
CVA rating of total switches
one additional converter is added to balance the capacitors of TCVAR = (23)
double LDN-based converter topology. It is well known that for VA rating of inverter
two-stage converter topology, dc–dc converter is required and Hbridge
SUF = TCVAR /TCVAR . (24)
that dc–dc converter is customized in such a way that it will help
to balance capacitor voltages as well as to remove three isolation The cost of the gate driver circuits will not differ significantly
transformers. compared to the topologies mentioned in Table IV. Owing to
In proposed converter topology, as input capacitor of TNPC nearly same efficiency w.r.t. standard topologies, the heat sink
comes in parallel to stray capacitance, this ensures the leakage volume will not change significantly. Considering the require-
current within permissible limits. In order to remove isolation ment of limited computational resources, the controller cost is
transformer, TCVAR becomes high as compared to other con- also expected to be unaffected. On the contrary, this topology
verter topologies, as mentioned in Table IV. TCVAR is ratio of has advantages because it does not require an auxiliary dc
cumulative VA (CVA) rating of total switches to VA rating of supply, and three isolation power frequency transformers of the
inverter, as given in (23). Table IV indicates that the transformer full-scale converter rating are removed, which reduces the imple-
is removed in the presented topological configuration by trading- mentation cost by 35%–50%. In the proposed hybrid converter,
off with cumulative VA rating of switches. It is apparent from the switching loss is negligible since the TNPC operates close to
the comparison that the proposed topology has 50% higher CVA the fundamental switching frequency and auxiliary cells operate
rating. Though, the number of switches and corresponding gate at around 1 to 2 kHz. Proper selection of switches also helps
drivers will remain same w.r.t. prior arts indicated in Table IV and to limit the conduction losses. Due to high power quality of
increased CVA rating will not have proportional impact on the this converter topology, filter losses will be absent. In this way,

Authorized licensed use limited to: Corporacion Universitaria de la Costa. Downloaded on April 16,2024 at 00:05:31 UTC from IEEE Xplore. Restrictions apply.
TAK AND CHATTOPADHYAY: THREE-PHASE TRANSFORMERLESS HYBRID MULTILEVEL INVERTER FOR SOLAR PV APPLICATION 5397

TABLE V with limited switches. Since this loss is distributed in several

COMPARISON OF TRANSFORMERLESS TOPOLOGIES
devices, corresponding junction temperatures are restricted to
substantially lower value (40–50 °C lower than two-level case).
Lower junction temperature has two-fold effect on failure rate:
Peak junction temperature is reduced, and temperature variations
during thermal cycling is also reduced. As a consequence, failure
rate of individual devices can be lowered by a factor as high as
100 or even much higher. Considering this, overall reliability
is expected to be comparable or improved w.r.t. presently used
topologies.

VII. CONCLUSION
A grid-connected two-stage transformerless multilevel con-
verter for utility-scale solar PV application was presented in
this article. Hybridization of TNPC with double LDN-based
topology was done to remove isolation transformers. The dc–dc
converter for MPPT in two-stage solar PV inverter was cus-
tomized in this work to balance the TNPC capacitors and double
the inverter’s power handling capacity. No extra converter and
controller were needed to balance capacitor voltages. Hence,
complexity and cost related to controller were negligible in
presented work. The key benefit of the proposed work was the
use of a single PV source in a complete converter topology
that does not require any isolation transformer or any auxiliary
power supply. This converter topology also kept leakage current
this topology provides an efficient, reliable, and cost-effective within acceptable limits without adding any additional circuit
converter solution. elements. High-resolution output voltage of presented inverter
A few potential transformerless topologies are compared was expected to offer an efficient and reliable converter solution
w.r.t. major parameters in Table V. Other than NPC (or its for PV central-inverter application. This article presented a thor-
variants, such as T-NPC), most of the transformerless topologies ough simulation study, analysis, as well as experimental results.
presented in the literature are for single-phase and low-power The experimental results acquired from a hardware prototype
applications. In those topologies, also additional switch/circuit is resembled well with the simulation results.
added to remove the transformer. Moreover, for those topologies,
switching losses and filter losses are much higher than the pro-
REFERENCES
posed solution. Key merits and the limitations of the presented
work can be identified in this table. [1] N. Sandeep and U. R. Yaragatti, “Design and implementation of a sen-
sorless multilevel inverter with reduced part count,” IEEE Trans. Power
Electron., vol. 32, no. 9, pp. 6677–6683, Sep. 2017.
D. Reliability of Proposed Converter Topology [2] X. Zhang, T. Zhao, W. Mao, D. Tan, and L. Chang, “Multilevel inverters for
grid-connected photovoltaic applications: Examining emerging trends,”
Reliability analysis of the proposed topology can be broken IEEE Power Electron. Mag., vol. 5, no. 4, pp. 32–41, Dec. 2018.
[3] A. Aghazadeh, M. Davari, H. Nafisi, and F. Blaabjerg, “Grid integration of
into two parts: Reliability of the dc–dc converter (mostly de- a dual two-level voltage-source inverter considering grid impedance and
termined by switch Sb ) and reliability of the inverter consisting phase-locked loop,” IEEE J. Emerg. Sel. Topics Power Electron., vol. 9,
large number of switches. no. 1, pp. 401–422, Feb. 2021.
[4] K. Chen, W. Jiang, and P. Wang, “An extended DPWM strategy with
Reliability issues due to switch Sb can be addressed by having unconditional balanced neutral point voltage for neutral point clamped
multiple dc–dc converter units connected to same inverter. This three-level converter,” IEEE Trans. Ind. Electron., vol. 66, no. 11,
is a common practice in a large-scale two-stage PV inverters to pp. 8402–8413, Nov. 2019.
[5] Y. Tanaka et al., “ Three-level neutral point clamping PWM inverter and
have multiple dc–dc converters (i.e., multiple MPPT trackers) neutral point voltage controller,” U.S. Patent 6 795 323 B2, Sep. 21, 2004.
[31], [32]. In the proposed configuration, voltage and current [6] S. Xu, J. Zhang, G. Ma, Z. Sun, K. Liu, and C. Yao, “Operation of
stress in the coupled inductor-based dc–dc converter is also not a seven-level T-type active neutral-point-clamped converter with mod-
ified level-shifted PWM,” IEEE Trans. Ind. Electron., vol. 68, no. 11,
significantly high compared to the well-known dc–dc converters pp. 10970–10981, Nov. 2021.
used in two-stage applications. [7] J. Zhuang et al., “Active neutral point clamped inverter, zero-crossing
Reliability issue for the inverter part is achieved by ensuring switching method and zero-crossing switching apparatus thereof,” U.S.
Patent 2018 /0316 276 A1, Nov. 1, 2018.
lower junction temperature of the power semiconductor devices [8] F. Gao, “Non-isolated photovoltaic grid-connected inverter and control
due to significantly lower conduction loss per switch and neg- method therefor,” WO 2013/163777 AI, Nov. 7, 2013.
ligible overall switching loss. It is found that the overall effi- [9] C. A. Rojas, M. Aguirre, S. Kouro, T. Geyer, and E. Gutierrez, “Leakage
current mitigation in photovoltaic string inverter using predictive control
ciency of the converter (even without considering filter losses) with fixed average switching frequency,” IEEE Trans. Ind. Electron.,
is competitive (or even better) with the standard topologies vol. 64, no. 12, pp. 9344–9354, Dec. 2017.

Authorized licensed use limited to: Corporacion Universitaria de la Costa. Downloaded on April 16,2024 at 00:05:31 UTC from IEEE Xplore. Restrictions apply.
5398 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 71, NO. 6, JUNE 2024

[10] Y. Jiang, C. Qin, X. Xing, X. Li, and C. Zhang, “A hybrid passivity- [29] M. S. Manoharan, A. Ahmed, and J.-H. Park, “A PV power condi-
based control strategy for three-level T-type inverter in LVRT operation,” tioning system using nonregenerative single-sourced trinary asymmetric
IEEE J. Emerg. Sel. Topics Power Electron., vol. 8, no. 4, pp. 4009–4024, multilevel inverter with hybrid control scheme and reduced leakage cur-
Dec. 2020. rent,” IEEE Trans. Power Electron., vol. 32, no. 10, pp. 7602–7614, Oct.
[11] Y. Konishi et al., “Transformerless power conversion circuit for grid 2017.
connected power generation system,” U.S. Patent 7 440 300 B2, Oct. [30] S. K. Chattopadhyay and C. Chakraborty, “Performance of three-phase
21, 2008. asymmetric cascaded bridge (16 : 4 : 1) multilevel inverter,” IEEE Trans.
[12] R. Barzegarkhoo, S. S. Lee, S. A. Khan, Y. P. Siwakoti, and D. D.-C. Ind. Electron., vol. 62, no. 10, pp. 5983–5992, Oct. 2015.
Lu, “A novel generalized common-ground switched-capacitor multilevel [31] 2015. [Online]. Available: https://www.fimer.com/sites/default/files/
inverter suitable for transformerless grid-connected applications,” IEEE ULTRA-EN-Rev%20E.pdf
Trans. Power Electron., vol. 36, no. 9, pp. 10293–10306, Sep. 2021. [32] 2011. [Online]. Available: https://2.imimg.com/data2/QB/LY/MY-
[13] X. Zheng, J. Yao, Z. Xu, and A. Ioinovici, “Multioutput switched-capacitor 2452300/utility-scale-pv-inverter-system.pdf
multilevel inverter with intrinsic elimination of the photovoltaics induced [33] A. Srivastava and J. Seshadrinath, “A single-phase seven-
leakage current and high boost factor,” IEEE Trans. Ind. Electron., vol. 70, level triple boost inverter for grid-connected transformerless PV
no. 6, pp. 5850–5861, Jun. 2023. applications,” IEEE Trans. Ind. Electron., vol. 70, no. 9, pp. 9004–9015,
[14] M. Shahabadini and H. Iman-Eini, “Leakage current suppression in multi- Sep. 2023.
level cascaded H-bridge based photovoltaic inverters,” IEEE Trans. Power [34] S. Dhara and V. T. Somasekhar, “A nine-level transformerless boost
Electron., vol. 36, no. 12, pp. 13754–13762, Dec. 2021. inverter with leakage current reduction and fractional direct power transfer
[15] S. K. Kuncham, K. Annamalai, and N. Subrahmanyam, “A two-stage T- capability for PV applications,” IEEE J. Emerg. Sel. Topics Power Elec-
type hybrid five-level transformerless inverter for PV applications,” IEEE tron., vol. 10, no. 6, pp. 7938–7949, Dec. 2022.
Trans. Power Electron., vol. 35, no. 9, pp. 9510–9521, Sep. 2020. [35] S. Majumdar, K. C. Jana, P. K. Pal, A. Sangwongwanich, and F. Blaabjerg,
[16] M.-J. Tsai, Y.-Y. Shen, M.-J. Chao, and P.-T. Cheng, “An edge-aligned “Design and implementation of a single-source 17-level inverter for a
modulation technique to suppress leakage current in the back-to-back single-phase transformer-less grid-connected photovoltaic systems,” IEEE
neutral-point clamped converter,” IEEE Trans. Ind. Appl., vol. 57, no. 3, J. Emerg. Sel. Topics Power Electron., vol. 10, no. 4, pp. 4469–4485,
pp. 3179–3187, May/Jun. 2021. Aug. 2022.
[17] A. Salem, H. Van Khang, I. N. Jiya, and K. G. Robbersmyr, “Hybrid
three-phase transformer-based multilevel inverter with reduced component
count,” IEEE Access, vol. 10, pp. 47754–47763, 2022.
[18] A. Karthik and U. Loganathan, “A reduced component count five-level
inverter topology for high reliability electric drives,” IEEE Trans. Power
Electron., vol. 35, no. 1, pp. 725–732, Jan. 2020.
[19] A. K. Yadav, K. Gopakumar, K. R. R. L. Umanand, S. Bhattacharya, Neha Tak (Student Member, IEEE) received the
and W. Jarzyna, “A hybrid 7-level inverter using low-voltage devices and B.Tech. degree in electrical and electronics en-
operation with single DC-link,” IEEE Trans. Power Electron., vol. 34, gineering from Rajasthan Technical University,
no. 10, pp. 9844–9853, Oct. 2019. Udaipur, India, in 2012, and the M.Tech. degree
[20] S. Amamra, K. Meghriche, A. Cherifi, and B. Francois, “Multilevel in- in electric drives and power electronics from the
verter topology for renewable energy grid integration,” IEEE Trans. Ind. Indian Institute of Technology Roorkee, Roor-
Electron., vol. 64, no. 11, pp. 8855–8866, Nov. 2017. kee, India, in 2016. She is currently working to-
[21] A. Lewicki, C. Odeh, and M. Morawiec, “Space vector pulsewidth mod- ward the Ph.D. degree in power electronics spe-
ulation strategy for multilevel cascaded H-bridge inverter with DC-link cialization with the Department of Energy Sci-
voltage balancing ability,” IEEE Trans. Ind. Electron., vol. 70, no. 2, ence and Engineering, Indian Institute of Tech-
pp. 1161–1170, Feb. 2023. nology Delhi, New Delhi, India.
[22] E. Villanueva, P. Correa, J. Rodriguez, and M. Pacas, “Control of a She has 4.5 years’ experience in industry as well as in academics.
single-phase cascaded H-bridge multilevel inverter for grid-connected Her research interests include multilevel inverter and power electronic
photovoltaic systems,” IEEE Trans. Ind. Electron., vol. 56, no. 11, converter for solar PV application.
pp. 4399–4406, Nov. 2009.
[23] A. Lashab et al., “Cascaded multilevel PV inverter with improved harmonic
performance during power imbalance between power cells,” IEEE Trans.
Ind. Appl., vol. 56, no. 3, pp. 2788–2798, May/Jun. 2020.
[24] A. E. T. Maamar, M. Kermadi, M. Helaimi, R. Taleb, and S. Mekhilef,
“An improved single-phase asymmetrical multilevel inverter structure
with reduced number of switches and higher power quality,” IEEE Trans. Sumit K. Chattopadhyay (Member, IEEE) re-
Circuits Syst. II: Express Briefs, vol. 68, no. 6, pp. 2092–2096, Jun. 2021. ceived the B.E. degree in electrical engineer-
[25] A. Ahmed, M. S. Manoharan, and J.-H. Park, “An efficient single-sourced ing from Burdwan University, Burdwan, India, in
asymmetrical cascaded multilevel inverter with reduced leakage current 2004, the M.Tech. degree in industrial electrical
suitable for single-stage PV systems,” IEEE Trans. Energy Convers., systems from the National Institute of Technol-
vol. 34, no. 1, pp. 211–220, Mar. 2019. ogy Durgapur, Durgapur, India, in 2006, and
[26] S. K. Chattopadhyay and C. Chakraborty, “A new asymmetric multi- the Ph.D. degree in electrical engineering from
level inverter topology suitable for solar PV applications with varying the Department of Electrical Engineering, Indian
irradiance,” IEEE Trans. Sustain. Energy, vol. 8, no. 4, pp. 1496–1506, Institute of Technology Kharagpur, Kharagpur,
Oct. 2017. India, in 2016.
[27] N. Tak, S. K. Chattopadhyay, and C. Chakraborty, “Performance of He has about five years of industrial experi-
reduced-dc-source-based three-phase high-resolution multilevel inverter ence in power electronic system design. In 2018, he joined the Depart-
with optimal asymmetry,” IEEE Trans. Power Electron., vol. 37, no. 9, ment of Energy Science and Engineering, Indian Institute of Technology
pp. 10713–10726, Sep. 2022. Delhi, New Delhi, India, where he is currently an Assistant Professor. His
[28] E. E. Espinosa et al., “A new modulation method for a 13-level asymmetric research interests include development of reliable and efficient power
inverter toward minimum THD,” IEEE Trans. Ind. Appl., vol. 50, no. 3, electronic converters for renewable energy, electric vehicles, and other
pp. 1924–1933, May/Jun. 2014. industrial applications.

Authorized licensed use limited to: Corporacion Universitaria de la Costa. Downloaded on April 16,2024 at 00:05:31 UTC from IEEE Xplore. Restrictions apply.