State-of-the-Art EMC Designs to Consider Before Your Next Module Project

When helping clients in the electric vehicle (EV) industry with their module design, I often find that engineers tend to follow an out-of-date list of do’s and don’ts in the form of EMC design rules without understanding the basics. These kinds of design rules are often borrowed from other industries, and they are not up to date with the latest technology involved in the fast-paced EV industry.

In this article, using a powertrain module as an example, I will first introduce the high voltage EMC regulations with which a powertrain module needs to comply. I will then highlight the risks and challenges when designing such modules. The main part of this article will share design techniques that engineers can apply on various parts of a powertrain module, including ground, front-end filter, inverter design, and so on. Examples are given to demonstrate some of the key design techniques.

It should be noted that the introduced design techniques in this article share the same principles as any other EMC engineering. Therefore, engineers from other industries can also benefit from these techniques.

Background

The past decade has witnessed the fast-paced adoption of electrification in the automotive industry, with an increasing number of hybrid and full electric vehicles coming to the market. A recent study has shown that going half-way (as plug-in hybrid vehicles do) might not be sufficient to bring carbon emissions in line with new regulations, which will require more full EVs on the road [1].

Powertrain modules are one of the key differentiators in the EV industry. Both vehicle manufacturers and Tier-1 suppliers have been spending considerable resources researching and developing state-of-the-art technologies for EVs. The current trend is to achieve a more compact module design with higher power density and system efficiency. For instance, the Nissan LEAF has achieved a very compact e-powertrain module design by integrating an on-board charger (OBC), a DC-DC converter, and a junction box with the electric drive unit (EDU) [2].

This article presents the EMC design of a powertrain module, which consists of an electric motor, an inverter, and a mechanical gearbox. The electric system diagram of an EV powertrain module is illustrated in Figure 1.

The importance of design engineers taking a system-level overview of an EV powertrain module was presented in my previous work, “Demystifying EMC in an Electric Vehicle’s Drive Unit” [3]. It is critical to factor in EMC design considerations at an early stage so as to achieve the overall system design goal. The high voltage (HV) EMC regulations and requirements present a daunting task for not only new entrants but also for well-established companies in the automotive industry. Therefore, we’ll first review in this article the HV standards and regulations that apply to electric powertrain modules. Then, we’ll highlight EMC challenges in the powertrain module design and demonstrate design techniques to address potential EMC issues. Engineers will then have a better understanding of how to design a module that will pass the EMC requirements in the EMC test chamber.

HV EMC Standards and Regulations for EV Powertrain Modules

CISPR 25:2016 [4] serves as a general EMC guideline for automotive developers, although vehicle manufacturers often have their own proprietary EMC specifications [5]. Annex I of CISPR 25 defines test methods for shielded power supply systems for high voltages in electric and hybrid vehicles. CISPR 36:2020 [6] was released recently, and it defines the test methods for electromagnetic field emission on a vehicle level. A component-level electromagnetic field emission test to reflect this standard is expected to be available soon from vehicle manufacturers.

(Note that OBCs require a different set of test methods that are related to charging and are not covered in this article. Also not included here is a discussion of low voltage (LV)-associated EMC tests, electrical tests, or electrostatic discharge (ESD) tests.)

HV Component Emission, Immunity, and Transients

Annex I of CISPR 25 defines both conducted and radiated emission limits for shielded HV systems, and unshielded systems shall comply with the same limits as shielded systems. Annex I of CISPR 25 also introduces the HV/LV coupling attenuation test. This test is performed while the equipment under test (EUT) is unpowered. Essentially the test result is the -S₂₁ plot of the EUT by an impedance analyzer. For good system decoupling behavior, the Class A1 or A2 in requirements for minimum coupling attenuation given in [4] need to be achieved.

For RF-immunity, the ISO 11452 series is relevant for vehicle component tests. The newest revisions of subparts under ISO 11452, such as ISO 11452-4:2020, include HV component test setups and high voltage artificial networks (HV-ANs). Other subparts are expected to adopt these HV component requirements accordingly.

ISO/TS 7637-4:2020 deals with transient emissions and transient immunity on HV lines.

Electric and Magnetic Field Strength

The standards SAE J551-5 [7] and GB/T 18387-2017 [8] define limits and test methods in the U.S. and China, respectively, for the magnetic and electric field emissions from electric vehicles. CISPR 36:2020 [6] deals with electromagnetic field emissions. A recent comparison study between GB/T 18387-2017 and CISPR 36:2020 on the magnetic field radiated disturbance test requirements has found that GB/T 18387-2017 is more stringent [9].

Magnetic field exposure will be the most critical aspect for electrical vehicles because of the high currents associated with their electric drives. To cover this aspect also on the component level, test methods have to be defined.

Human exposure to magnetic fields is tested in accordance with a Guideline issued by the International Commission on Non-Ionizing Radiation Protection (ICNIRP). IEC TS 62764-1 defines the measurement procedures for magnetic field levels generated by electronic and electrical equipment in the automotive environment with respect to human exposure.

A summary of the HV EMC test is listed in Table 1.

HV EMC Test	Reference
RF Conducted & Radiated Emission	Annex I, CISPR 25:2016 Edition 4
Electric and Magnetic Field Strength	Component test plan shall reflect vehicle test requirements GB/T 18387/2017 or CISPR 36:2020
HV Transient Emission and Immunity	ISO/TS 7637-4:2020
Human exposure to Magnetic Field	ICNIRP Guideline
RF Conducted & Radiated Immunity	ISO 11452 Series
HV LV coupling	Annex I, CISPR 25:2016 Edition 4

List only consists of HV EMC tests, LV, electrical and ESD tests are not included

Table 1: List of HV EMC tests for powertrain module [1]

EMC Challenges

A more compact, larger power rating, and higher efficiency powertrain module generally means more EMC challenges.

High Voltage

Currently, the most common HV rating adopted by modern automotive manufacturers is 400V (Audi e-tron, Tesla, Nissan Leaf, etc.). The Porsche Taycan is the first EV in the market to have adopted an 800V power system [10]. For the same power rating, increasing the voltage reduces the current in the system, resulting in reduced copper loss (I²R loss). Therefore, higher voltages correlate with an increased system efficiency. However, the HV rating of a powertrain module is limited by factors such as the voltage ratings of commercially available power electronics devices, the insulation breakdown of HV cables, worse EMC performance, and so on.

The roadmap for the next generation of power electronics devices for powertrain modules has indicated a breakdown voltage level beyond 1000V. It won’t be long before most of the automotive design houses move to 800V (and above) systems for the benefit of achieving higher efficiency. This predictable trend also poses great challenges for EMC design. As voltage level doubles (from 400V to 800V, for instance), and assuming the same parasitic characteristics of a design, noise levels associated with electric field will increase because of the high dV/dt characteristics.

Another great challenge associated with high voltage systems is safety. EMC and safety cannot be discussed separately in an HV system. Global Technical Regulation on Electrical Vehicle Safety (EVS) [11] defines the maximum capacitor energy that may be stored in the Y-capacitors to be 0.2J. This hard limit has a profound impact of front-end input filter design of all HV modules because Y-capacitors are very effective filters for broad band noise attenuation, particularly in the lower frequency range (starting from 300kHz).

This means that when voltage level doubles, the available Y-capacitance value drops by 75% according to Equation 1:

E = 0.5 · C_YV² Eq. 1

where E is the total energy stored in the Y-capacitor, C_Y is the available Y-class capacitance, and V is the upper band of HV system nominal voltage.

The relationship between noise level (as represented by common mode current) and available Y-capacitance is illustrated in Figure 2.

High Power/Current

Since the powertrain module is the performance unit of a vehicle, its power rating directly determines the acceleration rate, the horsepower, and the torque of a vehicle. Given a defined voltage rating of a powertrain module, higher power means higher current. As current is directly related to the magnetic field, higher power also means an increased magnetic field for a vehicle.

As the electric motor and inductors in the inverter are both inductive, higher current also means higher transient behavior caused by sudden state change. The back electromotive force (EMF) or kickback voltage caused by L·di/dt can stress or destroy components if not contained and send huge voltage spikes propagating on the HV bus line.

Fast Switching Power Electronics Devices

Insulated-gate bipolar transistors (IGBTs) were adopted in the early days of powertrain modules (such as the one in an earlier version of the Tesla Model S). The switching frequency of an IGBT-based power system is theoretically limited to 20kHz. The thermal concern of an IGBT often limits its switching frequency below its theoretical value. Wide-band-gap devices such as SiC MOSFETs have recently started replacing IGBTs as the device of choice thanks to their fast switching speeds (hence low switching losses) and better thermal characteristics. Switching frequencies of 20 kHz and above could be comfortably achieved with SiC MOSFET-based powertrain module.

The downside of adopting wide-band-gap devices is the increase in electromagnetic interference (EMI)-related issues caused by their faster switching events. The rise time of a SiC MOSFET can be as small as a few nanoseconds, leading to a slew rate of 50-200V/ns [12]. To enable such fast speed characteristics, gate drivers are equipped with short high peak pulse current features, which could also pose EMI issues.

More SMPS and ICs

To provide power to microprocessors, gate drivers, and analog and digital integrated circuits (ICs), multiple power suppliers are often integrated into the design of the control unit board, with switched-mode power supply (SMPS) units the most common.

SMPS units such as buck and boost converters in the automotive application often have a switching frequency range between 150kHz and 500 kHz. The rise and fall time of the switches can be as short as a few nanoseconds. The noise spectrum shows less energy compared with power switching devices but covers a much wider frequency range. Figure 3 demonstrates the switching noise profile between an LV buck converter and an HV inverter.

Bearing Current

Bearing currents are mainly caused by electrostatic discharges, magnetic asymmetries (caused by unbalanced three-phase windings), and common-mode voltages paired with high switching rates [13].

Bearings in an electric motor have moving metal balls or rollers in fixed metal shells. Very thin layers of lubricant sit between the two parts which therefore have a high capacitance and can carry high displacement currents. Because the lubricant is so thin, and because the bearings are not perfect, there can be an occasional electrical breakdown and even direct touching of the two metal parts. Therefore, the bearing current is partly capacitive, which gives a pulse of current during every switching transition and partly random high current spike [3]. This random breakdown can cause very high random peak currents that can give high quasi-peak noise in the EMI scan.

Bearing current can cause an electric motor’s bearings to deteriorate, hence reducing the lifetime of the powertrain. Bearing currents that circulate in the powertrain also cause conducted and radiated emissions, as shown in Figure 4.

EMC Design Techniques in Electric Powertrain Module

Grounding Design

It is not uncommon to see many ground symbols in one module design, even though the point of clarifying the use of the term ground has been stressed by many EMC experts [14] [15]. One recent project that I’ve reviewed had more than five ground symbols in one schematic. It is very confusing to see all these symbols in the first place, not to mention how the grounds are connected.

In Figure 1, the term reference is used rather than the term ground. It should be noted that circuit grounds are not necessarily the same as EMC grounds. To keep it simple and clear, there can only be one EMC ground, which is the metal chassis of the unit, or what we call the RF reference. The metal enclosure of a module has contact points to the vehicle chassis (either through direct bonding or mechanical fixtures); therefore, we treat the metal enclosure as a chassis reference.

The HV-line is the HV design reference, and it should be isolated from the vehicle chassis reference (either by dielectrics or by Y-capacitors). LV designs should have 0V as the reference points, and that should be the only design reference point. The idea of splitting analog and digital ground points is based on misconceptions and is not a good design approach [16].

The connection between the 0V reference (either on a PCB or a connector pin) and the chassis reference will introduce inductance [16]. The RF currents will inevitably flow in those inductances, leading to noise voltages that will help drive emissions. This is shown in Figure 5. As a result, efforts to minimize the inductance of the connection should be made in the module design. Among the schemes that reduce inductances between two reference points, the most effective way is to increase direct contact areas between the 0V Ref and Chassis Ref. This is demonstrated in Figure 6. Most of the time, multiple contact points along the edges and around corners of a PCB create a quasi-360-degree contact.

Connector Design

HV cables of a powertrain module are usually shielded, and shielding terminations have closely-spaced contacts (360 degrees) to the module enclosure [17]. The reason for having a low impedance bond is the same as what we explained in the previous section. However, the mechanical quality of the contactor design needs to be considered very carefully because high temperatures, aging, vibration, and chemical ingress can damage the bonding configuration over time. An example of a failed shielding connector is shown in Figure 7.

Front-end Filter Design

The front-end filter design is crucial for electric powertrain modules as it helps to block the noise from the inverter power switches. It also suppresses the noise traveling from outside of the module enclosure via the HV DC wirings.

There are many types of front-end filters, including the two-stage filter shown in Figure 1. In Figure 1, the L₁ and C₂ configuration forms the first-stage low-pass filter. The second-stage filter consists of a CMC and Y-capacitors. Notice that, together with the DC link capacitor C₁, the first-stage filter effectively acts as a p (C-L-C) filter.

Nanocrystalline Core

Due to its high voltage and high current characteristics, the saturation of the magnetic core needs to be accounted for when designing HV inductive components. Nanocrystalline materials enjoy a very high saturation magnetization. Because windings that carry such high currents will inevitably increase the size and the weight of a module, toroid or oval shape nanocrystalline cores are typically used in common-mode suppression chokes in automotive applications. They are effective in the frequency region between 150kHz and 120MHz.

For powertrain applications, cores can be either used on a single power line (HV+) as an inductor or on both power lines as a CMC. It should be noted here that the designers might find CMCs such as nanocrystalline cores are not needed due to other good EMC practices in place. However, it is best to allow for Murphy’s law and to design properly from the start so that the cores can be added later if necessary [18].

Y-capacitors

Compared with the use of CMCs, the benefits of using Y-capacitors include great high frequency conducted emission attenuation (generally effective starting from 5 MHz), smaller sizes, lighter weight, and no saturation concerns. The connection of Y-capacitors to the chassis reference should also be designed to achieve very low impedances. The internal equivalent series resistance (ESR) and equivalent series inductance (ESL) are the main factors that affect the effectiveness of Y-capacitors. The imbalance of the impedance of two Y-capacitors also affects the common-mode filtering performance. These drawbacks can be compensated by layout or by using alternative parts such as X2Y capacitors.

X2Y capacitors

X2Y capacitors(see Figure 8) [19] provide great EMI filtering and can be used to replace CMCs (Figure 8(a))in applications where size, weight, and cost are design constraints. Two capacitors are balanced shunt so as to create a cancellation of the mutual inductance. The X2Y capacitors also provide a shielding effect. Alternatively, as Figure 8 (b) shows, the X2Y capacitors can be configured as decoupling capacitors with ultra-low inductance. Currently, 500V X2Y parts are not automotive qualified, but it is worth paying attention to this component as manufacturers will probably upgrade the high voltage parts so they are AEC-Q200 qualified.

DC Link Design

The DC link includes the HV DC bus bars and DC link capacitors. The DC bus bars should be designed as short as possible and in close proximity to each other to reduce the loop area between them. DC link capacitors should be designed to cope with high voltage, high-frequency switching ripples, and high temperatures. The capacitance value should be large enough for the full power operation of a powertrain module. Low ESR and ESL film capacitors and electrolytic capacitors are often used in powertrain modules.

Film Capacitors

Film capacitors are widely seen in powertrain module design due to their high performance and reliability. The brick size of film capacitors limits the design freedom; therefore, it is important for design engineers to engage film capacitor suppliers early in the design stage. Parameters such as ESL and ESR are crucial for EMC, as self-inductance of the capacitor is caused by geometry of the component (such as capacitor wrapping, upper conductor rail, and lower conductor rail).

Electrolytic Capacitors

In some cases, manufacturers use a high-performance electrolytic capacitor bank as the DC link. The current flowing in the self-inductance of a capacitor creates a magnetic field. The smaller the self-inductance, the smaller the magnetic field for a given current, and vice-versa. Connecting a number of capacitors in parallel connects their self-inductances in parallel. But, if they are too close together, their magnetic fields will interact since the overall inductance is not reduced when the magnetic fields are all in the same direction. As a result, the overall inductance is not reduced to 1/N (as we might expect from circuit theory or SPICE simulations).

But if we arrange N-paralleled capacitors closely together, and so that they are alternately reversed or so that the self-inductance of the capacitors are in perpendicular position to each other, their magnetic fields will tend to oppose each other, canceling them out to some extent (as now mutual inductance is kept at a minimum). Since weaker fields mean lower inductances, we may be able to achieve greater than a 1/N reduction in overall self-inductance [20].

Figure 9 demonstrates the magnetic field coupling due to mutual inductance between capacitors. Examples are given to show how layouts of arrays of capacitors can achieve lower overall inductances by their magnetic fields and cancel each other out, to some extent.

Inverter Power Electronics Devices Design

The EMC design consideration of using wide-band-gap devices such as SiC MOSFETs was introduced in [3]. The subject itself could easily inspire a few dedicated articles. Therefore, we only summarize here some of the best design practices.

Because the common-mode current I_CM can be calculated by Equation 2:

I_CM = C_stray · dV/dt Eq. 2

where dV/dt is the slew rate of the switching device, C_stray is the stray (parasitic) capacitance and can be calculated by Equation 3:

C_stray = C_FET + C_D + C_L Eq. 3

where C_FET is the SiC MOSFET parasitic capacitance (predominantly drain to source capacitance C_DS), C_D is the free-wheeling diode capacitance, and C_L is the parasitic capacitance caused by layout (for instance, the capacitance between the device and a heatsink).

Reducing the slew rate helps to mitigate spikes or ringing, but at the cost of increased switching loss. Reducing C_stray can be achieved by selecting optimized packaging and applying good layout practice. The ringing of the switching is caused by the L–C circuit resonance, and good layout practice to achieve lower stray inductance (e.g., device connections to the bus bar) helps reduce the ringing.

To share the large current, N SiC MOSFETs are placed in parallel. This configuration results in 1/N R_DS(ON), allowing very low conduction loss. The total ESL of the devices might not be as low as 1/N for the same reason we explained when we talked about multi capacitors in parallel. However, ways of shortening the connections, such as connections between the devices and bus bars and the connections between the devices and motor windings, can minimize the inductance.

Figure 10 demonstrates the SiC MOSFETs layout in the powertrain module of a Tesla Model 3. Four MOSFETs are put in parallel to form one switching block. Altogether, there are 24 switching devices in a very tight package space, with short connections to minimize the parasitic inductance. Direct sintering of the SiC MOSFET to the bottom of the heat sink helps remove the heat efficiently.

Techniques such as using SiC Schottky diodes in parallel to SiC MOSFETs to eliminate the reverse recovery charge effect were presented in [3]. But more manufacturers are integrating very fast and robust intrinsic body diode into the device package; hence separate antiparallel diodes are not required. Generally, locating decoupling capacitor arrays close to the switching devices is also crucial to reduce the ringing effect of the switching events.

Inverter Control Unit Design

Compared with the power stage, the low voltage control unit often has various high-frequency noise sources. The noise spectrum of a control unit covers a much wider range, as listed in Table 2.

LV Electronics	Frequency Range
Switched Mode Power Supply (SMPS)	150 kHz – 2 MHz
Communication Lines (LIN, CAN, FlexRay, etc)	10 kHz – 2.5 MHz
Microcontroller/DSP	200 MHz – 600 MHz
Digital Isolator	200 kHz – 10s of MHz

Table 2: List of LV electronics frequency range in powertrain module

The EMC design follows guidelines similar to those we previously discussed, which is to apply good layout practice, design front-end and output filters on both power and signal lines, and apply sufficient global and local decoupling capacitors.

CMCs are often seen in the LV power system design, and the X2Y balanced capacitor was introduced previously in this article. Although the HV (above 400V) X2Y part is not automotive qualified, there are plenty of AEC-Q200 qualified parts [19] that can be used in the control unit of a powertrain module. [21] introduced X2Y capacitor in an SMPS design.

Bearing Current Mitigation

At the design stage, there are two approaches one can apply to mitigate the bearing current, as described in the following paragraphs.

Hardware Mitigation Schemes

Ceramic bearings or hybrid bearings (a combination of steel rings matched with ceramic) are good candidates for an electric motor because of their mechanical characteristics. Because ceramic is an electric insulator, it can reduce the bearing currents and mitigate the electrical arcing [22]. Alternatively, the rotor of an electric motor can be directly grounded with a small thrust bearing that can easily be changed [13] [23]. This grounding also helps prevent the motor shaft from radiating due to the stray RF currents induced into it.

Other methods of mitigating the common-mode noise, which contributes to the shaft voltage and bearing current, are to add the common-mode filter along the motor windings. Shielded cables also help [13]. But for a compact module design, these methods are generally not considered.

Software Mitigation Schemes

Apart from using the hardware approach, certain switching schemes can be adopted to reduce the common-mode voltage of an electric motor, therefore reducing the bearing current. The method proposed in [24] proved to achieve both high performance and low common-mode voltage and current but unfortunately does not include a spectrum analyzer evaluation of the results as compared with normal PWM schemes. The implementation of the proposed method also requires an in-depth understanding of the motor drive system.

Conclusion

In this article, we reviewed the HV-related EMC regulations for powertrain modules in EV applications. We then discussed the design challenges they present and demonstrated the EMC design techniques that can be implemented at the design stage.

Most of the techniques introduced in this article follow EMC design principles, such as reducing parasitic parameters, 360-degree shielding, and bonding. New passive components (such as nanocrystalline core and X2Y capacitors) can also be considered, as well as software schemes to mitigate the common-mode noise. By adopting these design techniques, engineers can be more confident that the powertrain module they design will pass the EMC tests, potentially even the first time around!

References

1. J. Miller, “Plug-in hybrid vehicles face attack despite gains in range,” Financial Times Special Report – Energy Efficiency, 2020.
2. Unknown, “Lightweight, compact and high-efficiency powertrain for electric vehicle EVs,” Nissan Motor Corporation.
3. M. Zhang, “Demystifying EMC in an Electric Vehicle’s Drive Unit,” Interference Technology, 2020.
4. CISPR 25:2016, 4th edition, “Vehicles, boats and internal combustion engines – Radio disturbance characteristics – Limits and methods of measurement for the protection of on-board receivers,” 2016.
5. H. W. K. R. (Hrsg.), Handbuch Kraftfahrzeugelektronik, ATZ/MTZ-Fachbuch, 2006.
6. CISPR 36:2020, “Electric and hybrid-electric road vehicles – Radio disturbance characteristics – Limits and methods of measurement for the protection of off-board receivers below 30 MHz,” 2020.
7. J551/5, “Performance Levels and Methods of Measurement of Magnetic and Electric Field Strength from Electric Vehicles, 150 KHz to 30 MHz.”
8. GB/T 18387, “Limits and Test Method of Magnetic and Electric Field Strength from Electric Vehicles,” 2017.
9. B. Z. X. L. Xuemei Huang, “Analysis and Verification of Magnetic Field Radiation Disturbance Standard for Electric Vehicle Below 30 MHz,” Safety and Compliance, no. 5, pp. 37‑39,55, 2020.
10. Porsche, “The battery: Sophisticated thermal management, 800-volt system voltage.”
11. “Global Technical Regulation on Electric Vehicle Safety (EVS),” United Nations, 2018.
12. Infineon, “CoolSIC 1200V SiC MOSFET Application Note.”
13. T. Hadden et al., “A Review of Shaft Voltages and Bearing Currents in EV and HEV Motors,” IECON 2016 – 42nd Annual Conference of the IEEE Industrial Electronics Society, Florence, 2016, pp. 1578-1583.
14. K. Armstrong, “EMC and Safety for Installations: Part 1,” In Compliance Magazine, October 2020. 
15. B. R. Archambeault, PCB Design for Real-World EMI Control, Norwell, Massachusetts: Kluwer Academic Publishers, 2002.
16. K. Armstrong, “Advanced PCB design and layout for EMC Part 4 – Reference planes for 0V and power.”
17. Patrick G. André and Kenneth Wyatt, EMI Troubleshooting Cookbook for Product Designers, SciTech Publishing, 2014.
18. K. Armstrong, “Design Techniques for EMC Part 5 – Printed Circuit Board (PCB) Design and Layout.”
19. Johanson Dielectrics, “EMI Filter & Decoupling Capacitors.”
20. K. Armstrong, EMC for Printed Circuit Boards, Nutwood UK Ltd, 2010, ISBN:978-0-9555118-5-1.
21. Keong Kam, David Pommerenke, Federico Centola, Cheung-wei Lam, Robert Steinfeld, “EMC Guideline for Synchronous Buck Converter Design,” in 2009 IEEE International Symposium on Electromagnetic Compatibility, Austin, TX, 2009.
22. MRC, “MRC Engineering Handbook.”
23. D. W. C. Terry L. Bossaller, “Kit and Method for Attaching a Grounding Ring to an Electrical Motor,” United States of America Patent US2010/0001602A1, 7 Jan 2010.
24. A. M. Hava and E. Ün, “A High-Performance PWM Algorithm for Common-Mode Voltage Reduction in Three-Phase Voltage Source Inverters,” IEEE Transactions on Power Electronics, vol. 26, no. 7, pp. 1998-20089, July 2011.