Performance Analysis of Half-Bridge Commutation Cells of a Modular Multilevel Voltage Source Converter ^†

Abstract

This study presents a performance analysis of a half-bridge commutation cell using MATLAB/Simulink simulations. The behavior of IGBTs as semiconductor switches is examined under diverse load conditions, and the power losses, including conduction and switching losses, of two IGBT models, IXGH25N120 and IRG7IA19UPbF, are evaluated. A simulation-based design methodology is proposed for selecting suitable IGBTs for the half-bridge commutation cell. The comparative analysis reveals that the IXGH25N120 IGBT is well-suited for high-performance applications, while the IRG7IA19UPbF IGBT is optimized for energy recovery circuits. Furthermore, it is observed that IGBT1 (IXGH25N120) exhibits higher power losses due to its significant conduction losses compared to IGBT2 (IRG7IA19UPbF). The IGBT2 exhibits low conduction losses but has dominant switching losses due to high switching frequency. The utilization of a simulation-based approach provides valuable insights into the influence of IGBT characteristics on the responses of the half-bridge commutation cell, enabling an enhanced understanding for the design and performance analysis of the cell within the context of MMC.

1. Introduction

In the past, thyristor-controlled line commutated converters (LCCs) have been the primary means of HVDC transmission, which has been a significant technique of power transmission. LCCs offer consistent DC current flow through a significant inductance, which injects grid-frequency and harmonic currents into the AC network, acting as a current source on the AC side, in contrast to diode-based HVDC converters, which lack controllability. The success of HVDC networks depends on the use of suitable converters that can convert AC power to DC and vice versa. The advent of insulated gate bipolar transistors (IGBTs) in the 1970s marked a significant breakthrough in the semiconductor switching device field. Combining the gate-drive characteristics of power MOSFETs with bipolar transistors’ high-current, low-saturation-voltage capability, IGBTs are now used widely in power applications.

The IGBT has paved the way for the development of voltage source converter (VSC) technology in HVDC systems. Through the series connection of IGBTs, the VSC enables the transmission of high DC voltages, currently up to 500 kV [1,2,3,4,5]. Compared to classic line-commutated thyristor HVDC converters, VSC HVDC technology offers a range of advantages, including independent reactive power control, black start capability, the use of extruded polymer cables, a smaller station footprint, and standard transformers [6]. The VSC transmission technology has advanced to its fourth generation and employs a modular converter design that eliminates the requirement for filters. Additionally, this latest iteration of the VSC has achieved a significantly reduced level of converter loss [7]. In [8], numerous HVDC configurations and emerging applications are reviewed.

In MMC, the building-block cell is a fundamental component that acts as a power converter, with the ability to be connected either in series or parallel in order to meet specific application needs [9]. The cells’ internal switches facilitate both bidirectional current flow and unidirectional voltage blocking, allowing them to function in two quadrants and to generate two voltage levels. A straightforward half-bridge commutation cell design involves connecting semiconductor switches with antiparallel diodes. The switches operate in a complementary manner to prevent a short circuit. A modular multilevel converter is made up of multiple sub-modules connected in series, and the converter’s number of sub-modules can vary based on the application’s needs [10]. The performance analysis of the half-bridge cell using IGBTs in MMC–VSC is important for understanding system behavior and identifying potential limitations. This information helps improve the design and performance of MMC–VSC systems, making them more efficient, reliable, and cost-effective.

This study employs MATLAB/Simulink simulations to analyze the performance of a half-bridge commutation cell. This study investigates the behavior of IGBTs, which serve as semiconductor switches under various load conditions. Additionally, the analysis calculates both conduction and switching losses and assesses the performance of two IGBT manufacturers: the IXYS Semiconductor GmbH—IXGH25N120 and Infineon International Rectifier—IRG7IA19UPbF.

2. Cell Linkage Multi-Level Approach

By expanding the building block matrices, various types of cell-linkage modular converters can be synthesized, resulting in a multilevel structure. The higher the number of cells, the higher the voltage-blocking ability and the higher the output voltage quality. However, conduction losses are a function of the number of cells inserted in the conduction path.

The cell-linkage structures maintain the same characteristics of their corresponding building block cells, such as the IGBT-based bipolar configuration of the cell linkage multilevel converter resulting in a bipolar staircase voltage. The series-connected cell or sub-modules constitute the phase arm, while the upper and lower phase arm together form the phase unit as shown in Figure 1. This half-bridge cell plays a role in controlling the operation of MMC. Its switching frequency is a critical factor for efficient operation. In the absence of a sorting algorithm, the half-bridge cell operates at a default switching frequency of 50 Hz. However, with the implementation of a sorting algorithm, the switching frequency dynamically changes to optimize the performance and enhance the efficiency of the MMC system.

3. Design Methodology

A comprehensive description of a simulation-based approach to designing the half-bridge cell is discussed as follows.

3.1. Selecting the IGBTs

This study analyzed the performance of two different IGBTs in a half-bridge MMC configuration. The first IGBT, IXGH 25 N120, is suitable for high-performance applications, such as motor drives and inverters, due to its MOS gate turn-on and second generation HDMOSTM process [11]. The second IGBT, IRG7IA19UPbF, is optimized for energy recovery circuits in plasma display panel (PDP) applications, with low Vce (on) and EPULSETM features improving panel efficiency and high repetitive peak current capability [12]. The IXGH25N120 IGBT operates at 500 V, with a switching frequency of 588 Hz and td (on) of 100 ns, while the IRG7IA19UPbF IGBT operates at 240 V, with a switching frequency of 5208 Hz and td (on) of 15 ns.

Table 1. IGBT input characteristics.

IGBT Model	Ices	V (Ices)	Vge (th)	Vce (sat)	Ic (Vce (sat))	Cies	Cres	Coes	Ron Ω
IXGH 25 N120	250 uA	960 V	6 V	3 V	25 A	2750 pF	50 pF	200 pF	0.06
IRG7IA19UPbF	1 uA	360 V	4.7 V	1.49 V	30 A	1100 pF	30 pF	57 pF	0.049

showcases the meticulously derived input characteristics of IGBTs from datasheets, serving as crucial input parameters for simulations.

3.2. Designing the Half-Bridge Cell

For designing the half-bridge cell, simulations were conducted using parameters obtained from the IGBT datasheet. To assess the influence of IGBTs certain values, such as the total forward transit time, emission coefficient (N = 1), forward early voltage (VAF = 200 V), collector-emitter resistance, and forward current transfer ratio (BF), were kept constant. All simulations and calculations were performed at a temperature of 25 degrees Celsius. The emission coefficient, N, controlled the shape of the current–voltage curves in the vicinity of the origin, while the collector and emitter resistance parameters affected the slope of the current–voltage curve at higher currents.

The simulation of a half-bridge cell configuration consisting of two IGBTs connected with internal diodes, with an active load connected between them, is illustrated in Figure 2. The simulation model incorporates two subsystems and pulse generators to drive the gate signals of IGBT1 and IGBT2 with square wave pulses. The simulation employs a 50% duty cycle and a backward Euler solver configuration with a sample time of 1.0 × 10⁻⁷ s, starting from a steady state established by initial parameter settings from the datasheet.

4. Analysis

IGBT losses are classified as conduction losses and switching losses. The total power loss in an IGBT can be calculated using (1). The derived equations were obtained from [13].

$$P_{total }=P_{{conduction}_{\left(IGBT\right)}}+P_{{switching}_{\left(IGBT\right)}}+P_{{conduction}_{\left(Diode\right)}}+P_{{switching}_{\left(Diode\right)}}$$

(1)

In the conduction phase, conduction losses arise as a result of current flow through the IGBT or freewheeling diode. The power dissipation during this phase was calculated by multiplying the on-state voltage and the on-state current. The average conduction losses for IGBTs and diodes were calculated using (2) and (4).

$$P_{{conduction}_{\left(IGBT\right)}}=1/T{{\displaystyle \int }}_0^T\left[Vce\left(t\right)\ast Ice\left(t\right)\right] dt$$

(2)

$$P_{{conduction}_{\left(Diode\right)}}=V_{Do}\ast I+R_{Do}\ast I^2$$

(3)

The switching losses contribute to overall device losses, and their impact is directly related to the switching frequencies. The switching losses for IGBTs and diodes were calculated using (4) and (5).

$$P_{{switching}_{\left(IGBT\right)}}=(E_{on}@I_{nom}+E_{off}@I_{nom})\ast F_{sw}\ast \frac{\sqrt{2}}{\pi }\ast \frac{I_{out}}{I_{nom}}\ast \frac{V_{dc\_link}}{V_{nom}}$$

(4)

$$P_{{switching}_{\left(Diode\right)}}=\left(E_{REC}@I_{nom}\right)\ast F_{sw}\ast \frac{\sqrt{2}}{\pi }\ast \frac{I_{out}}{I_{nom}}\ast \frac{V_{dc\_link}}{V_{nom}}$$

(5)

IGBT modules in operation suffer from conduction loss and switching loss. Conduction power loss depends on load current and duty cycle, while switching power loss is influenced by junction temperature, DC link voltages, and load current. Analytical calculations employ equations using parameters (Eon, Inom, Vce, Vnom, and Vdo) from the datasheet to assess these losses.

5. Results and Discussion

This section presents the results of the investigation on IGBT modules, analyzing their characteristics and performance under varied conditions and the findings are discussed in relation to the research objectives.

The simulation results of half-bridge cell pulsating output waveforms reveal that the magnitude of load current, load power, and load voltage varies in response to changes in load values, emphasizing the influence of load characteristics on these parameters, as shown in Figure 3.

After performing analytical calculations using equations, it was observed that IGBT2 had higher switching losses, indicating a higher switching frequency or more frequent switching events. In contrast, IGBT1 exhibited higher conduction losses, suggesting higher current flow during the conduction phase, as shown in Figure 4a,b. Increasing the load value resulted in minimized losses. The diode connected to IGBT1 experiences higher conduction loss but lower switching loss compared to the diode connected to IGBT2, as shown in Figure 4a,d.

The power losses, including conduction and switching losses for both IGBTs and diodes in each IGBT model, were thoroughly examined. In order to obtain power losses that cannot be directly obtained from simulations, a simulation-based plot was generated for two distinct IGBT models operating at varying loads. This was achieved by subtracting the load power from the input power. It had been observed that IGBT 1 incurs greater power loss than IGBT 2, as depicted in Figure 5, primarily due to its higher conduction loss resulting from the higher on-state voltage drop in the diode. The lower switching frequency of IGBT 1 compared to IGBT 2 leads to reduced switching losses. As the load increases, the losses gradually decrease due to the lower voltage and current stress on the devices, resulting in improved efficiency and reduced power dissipation.

6. Conclusions

In conclusion, the analysis of various factors concerning IGBT characteristics highlights the favorable performance of IGBT2 in terms of load output and its lower power loss during on-state conduction. Conversely, IGBT1 demonstrates significantly lower switching power losses across all loads and frequencies. Hence, based on these findings, it is recommended to utilize Infineon International Rectifier’s IGBT2 model, specifically the IRG7IA19UPbF, as it outperforms other IGBTs in most criteria relevant to MMC. By optimizing the operation of the IGBT modules at higher loads, losses are minimized as the devices approach their ideal working conditions, thus confirming the observed trend.