Vienna Rectifier with Gallium Nitride (GaN) Devices

Transcription

1 Vienna Rectifier with Gallim Nitride (GaN) Devices By Ytong Zh A thesis sbmitted in partial flfillment of reqirements for the degree of Master of Science (Electrical and Compter Engineering) at the University of Wisconsin Madison 206

2 Vienna Rectifier with Gallim Nitride (GaN) Devices By Ytong Zh Under the spervision of Professor Yehi Han at University of Wisconsin - Madison Approved by Yehi Han Date

3

4 Abstract As the technology on wide bandgap materials sch as gallim-nitride (GaN) has advanced rapidly, commercial GaN power devices with satisfying performance are available now. It is widely-known that GaN-based switching devices have several advantages over traditional Si-based switching devices, sch as lower ON-resistance, faster switching speed, better thermal condctivity, and smaller size. However, researchers have not yet flly explored and applied GaN devices in many important power conversion applications sch as power rectifiers. The three-phase three-level three-switch Vienna rectifier has advantages of low inpt crrent harmonics, low blocking voltage stress on the power semicondctor devices, high power density, high efficiency, and high reliability, and is widely sed in many power applications. It is a good candidate topology to demonstrate GaN applications. In this thesis, a three-phase three-level three-switch Vienna rectifier is designed tilizing GaN FETs. The advantages and challenges of tilizing GaN FETs in Vienna rectifiers are discssed. The topology and operation principles of the Vienna rectifier are carried ot. The control of the Vienna rectifier is introdced based on two types of the crrent control strategies, the instantaneos crrent control and the direct power control. A simlation model is established and rn in MATLAB/Simlink to verify theoretical analysis. To provide a comparative analysis of GaN FET and Si MOSFET based Vienna rectifiers, two prototypes are bilt with each type of the power devices on a similar scale. Experiment reslts and design experience of GaN FET based and Si MOSFET based Vienna rectifier systems are presented. Advantages and benefits of applying GaN FET devices in the Vienna rectifier are conclded based on simlation and experimental reslts. It proves the promising potentials of GaN power devices in Vienna rectifier applications. -I-

5 Acknowledgement I wold like to thank my advisor Professor Yehi Han, for his gidance and spport to my gradate stdy and research at Wisconsin Electric Machines and Power Electronics Consortim (WEMPEC) in University of Wisconsin-Madison. His enthsiasm in power electronics and patience with his stdents always inspire me to prse the best and face the challenges with corage. It is great pleasre for me to spend my gradate life with the WEMPEC family. The WEMPEC faclties are giving the most comprehensive and the best arranged lectres in power electronics, controls, electric machines and power grids, and I have learnt a lot from them. The WEMPEC stdents establish harmonic research atmosphere and I have enjoyed the great collaboration with them. I wold love to thank my parents, Jinling Zh and Xiaoyn Feng, and my hsband, Haichan Tang, for their nderstanding and spport dring my gradate life. I wold like to extend my sincere love and best wishes to them. Great thanks to all my friends in USA and China, for the pleasant and sad time we spent together. -II-

6 Table of Contents Abstract... I Acknowledgement... II Table of Contents... III List of Figres... VII List of Tables... XII Nomenclatre... XIII Introdction... XV Research Overview... XV Smmary of Chapters... XV Chapter... State-of-the-Art Review.... Unidirectional Rectifier Topologies..... Passive Systems Hybrid Systems Active Systems Wide Bandgap Semicondctor Power Devices Material Properties Existing Wide Bandgap Devices Research Opportnities and Challenges Chapter Vienna Rectifier Topology & Operation General Topology of Three-Level Rectifiers Single-phase Three-level Rectifying Circit Topology Three-phase Three-level Rectifying Circit Topology Vienna Rectifier Topology and Operation Bipolar Bidirectional Switch Topology of Vienna Rectifier Operation of Vienna Rectifier Basic Operation of Vienna Rectifier III-

7 Voltage Space Vector of Vienna Rectifier Modlation of Vienna Rectifier Space Vector PWM (SVPWM) Based Modlation PWM Carrier Based Modlation Chapter Control of the Vienna Rectifier Mathematical Model of Vienna Rectifier System Model in Three-Phase Stationary (ABC) Reference Frame Model in Two-Phase Stationary (αβ) Reference Frame Model in Two-Phase Rotating (dq) Reference Frame Controller Strctre Crrent Controller DC Otpt Voltage Controller Voltage Balance Controller Direct Power Controller (DPC) Chapter Simlation of the Vienna Rectifier Simlation Model of the Vienna Rectifier Power Circit Simlation Model Bild Power Circit Model in MATLAB / Simlink Parameter Design of the Power Circit Control Circit Simlation Model Instantaneos Crrent Controller DPC Based Crrent Controller Voltage Controller Voltage Balancing Controller Modlator Simlation Model Smmary of the Vienna Rectifier Simlation Model Simlation Reslts of the Vienna Rectifier Simlation Reslts of the Vienna Rectifier Based on Instantaneos Crrent Control (ICC) IV-

8 4.2.2 Simlation Reslts of the Vienna Rectifier Based on Direct Power Control (DPC) Simlation Reslts of the Vienna Rectifier with Load Change Smmary of Simlation Reslts of Vienna Rectifier Chapter Experiments of the Vienna Rectifier The Hardware Design of the Vienna Rectifier The Power Circit Hardware Design The Driver circit Hardware Design The Driver Circit Hardware Design of the Si MOSFET The Driver circit Hardware Design of the GaN FET The Sensing Circit Hardware Design The AC Voltage Sensing Circit Hardware Design The DC Voltage Sensing Circit Hardware Design The AC Crrent Sensing Circit Hardware Design The DSP Control Circit Hardware Design The Power Spplier Hardware Design The Prototypes of the Vienna Rectifier Systems The Software Design of the Vienna Rectifier Experiment Reslts of the Vienna Rectifier Si MOSFET Based Vienna Rectifier Prototype with ICC control Si MOSFET Based Vienna Rectifier Prototype with DPC control GaN FET Based Vienna Rectifier Prototype with ICC Control GaN FET Based Vienna Rectifier Prototype with DPC Control Smmary of Experimental Reslts Comparison between the GaN FET Based and Si MOSFET Based Vienna Rectifiers Comparison on Power Densities and Sizes of the GaN FET and Si MOSFET Devices and Drivers Comparison on Power Losses on the GaN FET and Si MOSFET Devices7 -V-

9 5.4.3 Comparison on Crrent Qalities of the GaN FET and Si MOSFET Based Vienna Rectifier Systems... 8 Chapter Conclsions, Contribtions and Ftre Work Conclsions Contribtions Ftre Work Appendix Bibliography VI-

10 List of Figres Fig.. Rectifiers classification based on power flow direction and converter types... 2 Fig..2 Classification of nidirectional three-phase rectifier topologies into passive, hybrid, and active systems [9]... 3 Fig..3 (a) Topology of diode rectifier; (b) Topology of thyristor rectifier... 5 Fig..4 Minnesota rectifier sing third harmonic injection into all three phases... 7 Fig..5 Korea rectifier sing third harmonic injection always into one phase.. 7 Fig..6 (a) Y-connection phase-modlar rectifier; (b) Δ-connection phasemodlar rectifier... 9 Fig..7 Single-switch three-phase rectifier operating in DCM... 0 Fig..8 Three-switch Vienna rectifier... Fig..9 Six-switch Vienna rectifier... 2 Fig..0 Two bipolar bidirectional switch strctres... 3 Fig.. NPC-converter-based nidirectional rectifier... 4 Fig..2 Concentration of intrinsic carriers in semicondctor vs. temperatre... 6 Fig..3 The specific on-resistance vs. breakdown voltage... 7 Fig..4 The strctre of EPC GaN FET Fig..5 Illstration of cascode soltion Fig. 2. (a) Symbolic topology of single-phase three-level rectifying circit; (b) Eqivalent ideal switch topology of single-phase three-level rectifying circit Fig. 2.2 Symbolic topology of three-phase three-level rectifying circit Fig. 2.3 Eqivalent ideal switch topology of three-phase three-level rectifying circit... 3 Fig. 2.4 Switch implementation with one power electronics device and for diodes and its operation modes VII-

11 Fig. 2.5 Topology of Vienna rectifier Fig. 2 6 Voltage space vector of Vienna rectifier Fig. 2.7 Control flowchart of the Vienna Rectifier SVPWM Fig. 2.8 (a)architectre of the modlator of the Vienna rectifier system; (b) Plse-width modlation sing two 80 -phase-shifted trianglar carrier signals at N 0 in ( ia 0, ib 0, ic 0 ) section Fig. 3. Basic controller strctre of the Vienna Rectifier system. (Signal paths being eqal to all three phases are shown in doble lines) Fig. 3.2 State block diagram of the crrent controller Fig. 3.3 Model of the DC side of the Vienne Rectifier with a resistive load Fig. 3.4 State block diagram of the DC otpt voltage controller Fig. 3.5 Plse-width modlation and switching seqence illstrating the self stability of an nbalanced otpt voltage Fig. 3.6 (a) Vienna rectifier connection and netral point crrent direction for the switching state ( Sa, Sb, Sc) (0,, ) for 0; (b) Vienna rectifier connection and netral point crrent direction for the switching state ( Sa, Sb, Sc) (,0,0) for Fig. 3.7 Inflence of a DC component N -VIII- N m added to all three modlation indexes on the switching seqence of the rectifier system for N Fig. 3.8 DPC controller strctre of the Vienna rectifier system. (Signal paths being eqal to all three phases are shown in doble lines) Fig. 3.9 State block diagram of decopled power control loops in the DPC system... 7 Fig. 3.0 State block diagram of simplified active power control loop in the DPC system... 7 Fig. 4. Strctre of the simlation model of the Vienna rectifier system Fig. 4.2 (a) Simlation model of the power circit of the Vienna rectifier system; (b) Detailed simlation model of the rectifying sbsystem in the power circit Fig. 4.3 Strctre of the simlation model of the control circit... 77

12 Fig. 4.4 (a) State block diagram of the instantaneos crrent controller; (b) Fnction of the instantaneos crrent controller Fig. 4.5 State block diagram of the DPC based crrent controller Fig. 4.6 Fnction of the modlation index generating section in the DPC based crrent controller Fig. 4.7 State block diagram of the DC voltage controller... 8 Fig. 4.8 State block diagram of the voltage balancing controller... 8 Fig. 4.9 (a) State block diagram of the modlator based on the carrier based PWM modlation; (b) State block diagram of the PWM generator in the modlator Fig. 4.0 (a) State block diagram of the Vienna rectifier system simlation model with the instantaneos crrent control strategy; (b) State block diagram of the Vienna rectifier system simlation model with the direct power control strategy Fig. 4. Simlation reslts of the ICC based Vienna rectifier system in the rated operating condition with fs 00 khz, and RL 0 Ω Fig. 4.2 FFT analysis reslts of the ICC based Vienna rectifier in the rated operating condition with fs 00 khz, and RL 0 Ω Fig. 4.3 Simlation reslts of the ICC based Vienna rectifier system in the low freqency operating condition with f _ 0 khz Fig. 4.4 Simlation reslts of the ICC based Vienna rectifier system in the s high freqency operating condition with f _ 500 khz Fig. 4.5 Simlation reslts of the DPC based Vienna rectifier system in the rated operating condition with fs 00 khz, and RL 0 Ω Fig. 4.6 FFT analysis reslts of the DPC based Vienna rectifier in the rated operating condition with fs 00 khz, and RL 0 Ω Fig. 4.7 Simlation reslts of the ICC based Vienna rectifier system when the resistive load low s high R L changes from 0 Ω to 80 Ω at 0. s IX-

13 Fig. 4.8 Simlation reslts of the DPC based Vienna rectifier system when the resistive load R L changes from 0 Ω to 80 Ω at 0. s... 9 Fig. 5. Strctre of the hardware design of the Vienna rectifier system Fig. 5.2 Schematic of the driver circit in the Si MOSFET based Vienna rectifier system Fig. 5.3 Schematic of the driver circit with GaN FET in the GaN FET based Vienna rectifier system Fig. 5.4 Photo of a real driver circit for the GaN FET Fig. 5.5 Schematic of the AC voltage sensing circit in the Vienna rectifier system Fig. 5.6 Schematic of the DC voltage sensing circit in the Vienna rectifier system... 0 Fig. 5.7 Schematic of the AC crrent sensing circit in the Vienna rectifier system... 0 Fig. 5.8 Photo of the DSP development board Fig. 5.9 Photo of the prototype of the Si MOSFET based Vienna rectifier system Fig. 5.0 Photo of the prototype of the GaN FET based Vienna rectifier system Fig. 5. DSP Control Program Flow Chart Fig. 5.2 Experiment reslts of the Si MOSFET based Vienna rectifier system with ICC control... 0 Fig. 5.3 FFT analysis on AC mains crrent in the Si MOSFET based Vienna rectifier system with ICC control... 0 Fig. 5.4 Experiment reslts of the Si MOSFET based Vienna rectifier system with DPC control... Fig. 5.5 FFT analysis on AC mains crrent in the Si MOSFET based Vienna rectifier system with DPC control... 2 Fig. 5.6 Experiment reslts of the GaN FET based Vienna rectifier system with ICC control X-

14 Fig. 5.7 FFT analysis on AC mains crrent in the GaN FET based Vienna rectifier system with ICC control... 4 Fig. 5.8 Experiment reslts of the GaN FET based Vienna rectifier system with DPC control... 5 Fig. 5.9 FFT analysis on AC mains crrent in the GaN FET based Vienna rectifier system with DPC control XI-

15 List of Tables Table. Material properties I... 5 Table.2 Material properties II... 8 Table.3 Comparison between 200 V Si and SiC devices... 2 Table.4 Comparison between 200 V Si and GaN devices Table 2. Section division based on phase voltage polarity Table 2.2 Switching states in Section 2 ( a 0, b 0, c 0 ) Table 2.3 Voltage space vectors of Vienna rectifier Table 4. Parameters of the power circit of the Vienna rectifier system Table 5. Parameters of the selected Si MOSFET and GaN FET devices Table 5.2 Power spplying chips and their characteristics XII-

16 Nomenclatre Symbol C, C 2 i f N f s i a, i b, i c i, i i d, i q i o L * m i R L p q S i DC side capacitors dty cycle grid freqency switching freqency AC phase crrents in the ABC reference frame AC phase crrents in the αβ reference frame AC phase crrents in the dq reference frame netral point crrent AC side boost indctors modlation index resistive load active power reactive power switching state (fnction) S ij, TABC SW i switch ON/OFF stats Clark transforming matrix T dq Park transforming matrix DC, DC 2 DC side voltages on otpt capacitors DC DC side otpt voltage ao, bo, co an, bn, cn rectifier inpt voltages referring to netral point rectifier inpt voltages referring to mains grond in ABC reference frame -XIII-

17 r, d, q ω θ voltage space vector rectifier inpt voltages in αβ reference frame rectifier inpt voltages in dq reference frame the speed of dq reference frame the angle of dq reference frame Abbreviations 2DEG BFoM DPC FET GaN MOSFET PCB PWM SiC SVPWM THD WBG 2 dimensional electron gas Baliga s figre of merit direct power control field effect transistor gallim nitride metal oxide semicondctor field effect transistor printed circit board plse width modlation silicon carbide space vector plse width modlation total harmonic distortion wide bandgap -XIV-

18 Introdction Research Overview The objective of this research is to design a Vienna rectifier sing Gallim Nitride (GaN) FET devices to maximize advantages of the Vienna rectifier topology and GaN power devices. The promising potential of applying the GaN power devices in power electronics applications is conclded. In this research, theoretical analysis, simlation stdy and experiments based on prototypes are carried ot on the Vienna rectifier system with GaN FET devices. All the concepts and techniqes to realize sch a system is covered in this research. Smmary of Chapters Chapter presents the state-of-the-art review of rectifier topologies and wide bandgap semicondctor power devices. Based on the reviews, research opportnities are identified. Chapter 2 introdces the Vienna rectifier topology. Technical details, sch as operation and modlation, of the Vienna rectifier are presented. Chapter 3 presents control of the Vienna rectifier system. Since control strategy largely determines the performance of the Vienna rectifier, two control strategies based on instantaneos crrent control and direct power control are introdced and compared. Chapter 4 develops the simlation model of the Vienna rectifier in MATLAB/Simlink and provides simlation reslts. Simlation reslts are presented to validate the theoretical analysis reslts. Chapter 5 presents the hardware design of the Vienna rectifier prototypes based on similar scaling GaN FET devices and Si MOSFET devices respectively. Experiences of hardware design are shared in this chapter. Experimental reslts are provided to compare the GaN FET and the Si MOSFET devices. Chapter 6 concldes the thesis and smmarizes the contribtions of this research. Ftre work is also proposed in this chapter. -XV-

19 Chapter State-of-the-Art Review This chapter presents the state-of-the-art review of existing nidirectional rectifier topologies. These topologies are classified into three categories inclding passive rectifier systems, hybrid rectifier systems, and active rectifier systems. The advantages and challenges of the Vienna rectifier system are also discssed. Wide bandgap (WBG) semicondctors are briefly reviewed in the second part of this chapter. Several existing WBG prodcts are introdced. At the end of this chapter, based on the review of rectifier topologies, and WBG semicondctor, research opportnities are identified.. Unidirectional Rectifier Topologies AC-DC converters, also known as rectifiers, are the most commonly sed power electronics circits in wide applications inclding adjstable-speeds drive (ASDs), ninterrptible power spplies (UPSs), HVDC systems, and tility interfaces with renewable energy sorces sch as solar photovoltaic systems (PVs), wind power, etc []-[7]. In general, rectifiers play the role of a front end in power chains by switching accessible three-phase AC power into specific DC power, which spplies following power systems. Since rectifiers directly connect the inpt AC mains to the DC bs, they may have problems of poor power qality in terms of injected crrent harmonics, resltant voltage distortion and poor power factor at inpt AC mains and slowly varying rippled DC otpt at load end, low efficiency, and large size of AC and DC filters [8]. In order to avoid these problems, researches focsing on improved power qality and advanced performance rectifiers have been carried ot for a long time, and important achievements have been made. The classification of rectifiers can be done in varios ways in terms of different aspects. Based on the power flow direction, rectifiers can be divided into nidirectional ones and bidirectional ones. Based on converter types, the rectifiers can be classified as boost, bck, bck-boost, mltilevel and mltiplse AC-DC converters [8]. Fig.. shows the classification according to power flow direction and sb-classification based on converter

20 Chapter. State of the Art Review types. Other than converter types, rectifiers can also be classified by the switching components applied in the systems. Traditionally, rectifiers are developed sing diodes and thyristors. Nowadays, MOSFETs, GTOs, and IGBTs are widely sed in rectifier systems as well. Fig.. Rectifiers classification based on power flow direction and converter types [8] In addition to above classification methods, the nidirectional rectifiers can be divided into passive systems, hybrid systems and active systems as well [9]. Passive systems sally se diode or thyristor bridge withot any active crrent control. This reslts in low-freqency harmonics in the mains crrents and an ncontrolled otpt voltage. Hybrid systems partially integrate a passive rectifier and an active circit part implemented with power semicondctors that can be actively switched OFF. As a reslt, hybrid systems exhibit either partly controlled mains crrents or otpt voltage. Active systems employ flly controlled power semicondctors and achieve desired characteristics sch as controlled otpt voltage and controlled sinsoidal mains crrents. A more comprehensive comparison is presented in Fig..2 [9]. -2-

21 Chapter. State of the Art Review Unidirectional Three-Phase Rectifier Systems Passive Systems Hybrid Systems Active PFC Systems Single Diode Bridge Rectifier Systems DC Side Indctor AC Side Indctors Passive 3 rd Harmonic Injection Mlti-Plse Rectifier Systems Partial Transf. Isol. Or Atotransf. Based AC or DC Side Interphase Transformer Passive Plse Mltiplication Electronic Reactance Based Rectifier Systems Combination of Diode Rectifier and DC/DC Converter Systems 3 rd Harmonic Injection Systems Direct Three-Phase Systems Phase Modlar Systems Single Diode Bridge & DC Side Electron. Ind. Single Diode Bridge & AC Side Electron. Ind. Or Cap. Mlti-Plse Rectifier System Employing Electron Interphase Transf. Passive/Hybr. Or Active 3 rd Harm. Inject. Network Boost- or Bck-Type or Uncontrolled Otpt Diode Bridge Or Mlti-Plse System with 3 rd Harmonic Inj. (Plse Mltipl.) Y-Rectifier Or Δ-Rectifier - Y-Arrangement with Converter Artificial Star-Point Connection 3/2-Phase Scott- Transf. -Based Boost-Type Single Diode Bridge & DC-DC Otpt Stage Half-Controlled Diode Bridge Mlti-Plse Rectifier System (Transf. or Atotransf. -Based) with DC-DC Otpt Stage Empl. DC or AC side Ind. Bck-Type Single Diode Bridge & DC-DC Otpt Stage Half-Controlled Diode Bridge Impressed Inpt Crrent (Boost-Type) Impressed Inpt Voltage (Bck-Type) DCM CCM DVM CVM Single-Switch Rectifier Two-Switch Rectifier Two-level Rectifier - Y- or Δ-Switch Rectifier Three-Level Rectifier (VIENNA Rectifier) Six-Switch Converter Single-Switch Rectifier Three-Switch Rectifier Six-Switch Rectifier Fig..2 Classification of nidirectional three-phase rectifier topologies into passive, hybrid, and active systems [9] -3-

22 Chapter. State of the Art Review In the following discssion, nidirectional rectifier topologies with passive systems, hybrid systems and active systems will be introdced and discssed respectively... Passive Systems For prely passive systems, sally sing three-phase diode/thyristor bridge shown in Fig..3, the rectifier characteristics inclde: ) Containing no trn-off semicondctors; 2) Working prely in mains-commtated mode; 3) Employing low freqency passive components (indctors and capacitors) for otpt voltage smoothing and mains crrent shaping; 4) Large low-freqency harmonics in mains crrents, poor power factor, high THD; 5) Poor or even no otpt voltage reglation capability. AC L N AC C RL AC (a) -4-

23 Chapter. State of the Art Review AC L N AC C RL AC (b) Fig..3 (a) Topology of diode rectifier; (b) Topology of thyristor rectifier In general, passive rectifier systems exhibit high mains crrent peaks, which lead to poor inpt crrent qality and power factor. However, this problem can be redced if filter indctors are inserted in AC or DC side of the rectifier bridge. Adding filter indctors considerably improves system performance, bt system THD is still above 30% and a power factor below exists. In addition, the concept of passive third harmonic injection can be sed to improve the inpt crrent qality according to [0] and a THD of 5% at fll load can be achieved sing this method. Moreover, the inpt crrent qality can also be enhanced with mlti-plse rectifier strctre, which consists of two or more phase-shifted rectifier bridges connecting in parallel. Transformers are sed for phase-shifting or isolation in mltiplse rectifiers. Even thogh performance of passive systems can be improved with additional components or circit design, the system size is big and its efficiency is low. Considering the strong limitations of ncontrolled switches, i.e. diodes, or halfcontrolled switches, i.e. thyristors, passive systems are not sitable for applications where high inpt crrent qality and high power factor is reqired. -5-

24 Chapter. State of the Art Review..2 Hybrid Systems Hybrid systems are developed based on passive systems by partially integrating fllcontrolled semicondctor devices, which can be trned off, into the systems. The hybrid systems fndamentally allow a reglation of the otpt voltage and a sinsoidal control of the mains crrent. However, there are limitations to otpt voltage reglation and mains crrent shaping. Frthermore, low freqency filter components of passive rectifier systems may be replaced/emlated by high-freqency PWM converters of relatively low rated power (electronic indctor [], [2]), e.g., in the sense of an increase of the power density [9]. Third harmonic injection concepts form a major grop of hybrid rectifier circits. Here, crrent is injected by a passive or active injection network into either one phase or all three phases reslting in avoiding the zero mains crrent period in each phase. Therefore, nearly sinsoidal crrent flows in all phases. The most famos topology of third harmonic injection was proposed by Prof. N. Mohan in 995 and is known as Minnesota rectifier [3], [4], as shown in Fig..4. It ses a third harmonic crrent injection transformer to inject third harmonic crrents into all three phases to achieve sinsoidal mains crrents. The rectifier system shows a controlled otpt voltage and prely sinsoidal mains crrents. The main drawback of the topology is the blky, low-freqency crrent injection transformer, which has a high weight. The blky and heavy third harmonic injection transformer can be omitted if the crrent is always injected into only one phase. An interesting approach was proposed in [5], [6] and the basic strctre of the rectifier system is given in Fig..5. The system ses only a single indctor and three bidirectional, bipolar switches for injection of the third harmonic crrent into one phase. The crrent in the indctor L is modlated by the transistors S+ and S. The three bidirectional switches always connect the phase with smallest (absolte) voltage vale to the indctor. -6-

25 Chapter. State of the Art Review Fig..4 Minnesota rectifier sing third harmonic injection into all three phases Fig..5 Korea rectifier sing third harmonic injection always into one phase In conclsion, the rectifier fnction of the hybrid systems with third harmonic injection is implemented by a diode bridge on the inpt side. The active network for crrent shaping, injection, and voltage reglation, is arranged on the dc side, ths it may be considered essentially as a dc dc converter working on a time-varying (six-plse) dc inpt voltage [9]. Hence, the circits are relatively simple and exhibit relatively low complexity control comparing with active three-phase converter systems. The characteristics of hybrid rectifiers are smmarized as follows: -7-

26 Chapter. State of the Art Review ) The diode rectifier circit is commtated by mains while the active network is nder forced commtation since it is implemented with power semicondctors that can be actively switched OFF; 2) Utilize low freqency and/or switching freqency passive components; 3) Limited otpt voltage reglation and/or sinsoidal mains crrent shaping can be achieved by trn-off power semicondctors...3 Active Systems Rectifier systems with active power factor correction (PFC) systems have otstanding advantages inclding high inpt crrent qality, low inpt crrent harmonics, high power factor, controllable otpt voltage reglation, high power density, etc. As a reslt, plenty of researches have been done in developing varios active rectifier topologies. According to Fig..2, the active rectifier topologies have two main branches, phase-modlar systems and direct three-phase rectifier systems. First, phase-modlar systems se a single-phase rectifier stage for each phase to carry ot active rectifier characteristics. The individal rectifier systems can be connected in star (Y-Rectifier [7], [8], as shown in Fig..6(a)) or between phases (Δ-Rectifier [9], as shown in Fig..6(b)). As the individal phase nits provide independent otpt voltages (Vo Vo2 Vo3), in order to have common DC voltage (Vo) on the DC bs, isolated DC/DC converters are essential in each phase. Large capacitors are reqired in the DC bs of phasemodlar rectifier as power flow plsating, which is typical in single phase rectifier systems. Besides, balancing isses of three independent rectifier nits need to be addressed. Overall, the phase-modlar systems exhibit good active rectifier performance, bt a DC/DC converter is typically reqired for each phase in the systems. Therefore, the phase-modlar systems are not the top choice of active rectifiers in applications with specification of high power density. -8-

27 Chapter. State of the Art Review (a) Fig..6 (a) Y-connection phase-modlar rectifier; (b) Δ-connection phase-modlar (b) rectifier -9-

28 Chapter. State of the Art Review Second, direct three-phase rectifiers perform a direct energy conversion from the threephase AC mains to the specific DC bs. Direct three-phase rectifier systems can be generally classified into boost-type and bck-type rectifier systems based on the voltage ratio. Typical bck-type rectifier systems, sch as Three-Switch Bck Rectifier System [20] or Six-Switch Bck Rectifier System [2], operate in discontinos condction mode (DCM), ths extra inpt filtering capacitors are reqired de to discontinos inpt crrents and high crrent peak vales. The high demand for filters limits the applications of bck-type rectifiers in high power density applications. In general, boost-type rectifier systems are more poplar and have more topologies. A very simple boost-type direct three-phase rectifier topology is shown in Fig..7, named as single-switch three-phase boost rectifier [22], [23]. This system operates in discontinos condction mode (DCM) and the switch is modlated with a constant dty ratio. No PWM nor crrent measrement is reqired for this rectifier. However, since the system has discontinos inpt crrent with high crrent peak vales, there is a large demand in EMI filter as well. Fig..7 Single-switch three-phase rectifier operating in DCM Frthermore, direct three-phase rectifier systems can be divided into two-level and three-level topologies tilizing two and three voltage levels for PWM voltage formation. Compared with two-level rectifiers, three-level rectifiers otstand by smaller crrent ripples -0-

29 Chapter. State of the Art Review and lower voltage stress on semicondctor switches. Since the crrent ripples are small, the size of boost indctors on the AC side is small. Moreover, de to low voltage stress on semicondctor switches, the switching loss is decreased, the filtering effort is redced, and high freqency operation is enabled. As a reslt, capabilities of operating with high switching freqency and high power density make three-level rectifier sitable in corresponding applications. On the other hand, three-level rectifiers have some drawbacks. Usally, the complexity of three-level rectifier is high and an additional controller needs to be employed to balance the two otpt voltages. The advantages, however, overrle the drawbacks for applications where high power density is reqired. Fig..8 Three-switch Vienna rectifier The most famos representative of a nidirectional three-phase three-level rectifier topology is the Vienna rectifier, which is shown in Fig..8 [24], [25]. This is the "original" Vienna rectifier implemented with only three switches, and all semicondctors (diodes and --

30 Chapter. State of the Art Review active switches) are only stressed with half of the otpt DC voltage. The reliability of Vienna rectifier is pretty high becase the DC otpt voltage can never be shorten. In each phase, two diodes are always condcting the crrent, which cases high condction losses. The condction loss problem of the original Three-Switch Vienna rectifier can be declined by slightly changing the topology in to Six-Switch Vienna rectifier [26], as shown in Fig..9, to redce nmber of condcting diodes. Fig..9 Six-switch Vienna rectifier The main difference between the Three-Switch Vienna rectifier and the Six-Switch Vienna rectifier is the implementation of the key element, the bipolar (voltage), bidirectional (crrent) switch, in the three-level rectifiers. Two poplar strctres, having the same fnction, are shown in Fig..0 (a) and (b). The Three-Switch Vienna rectifier ses the for -2-

31 Chapter. State of the Art Review diodes bridge and one switch strctre given in Fig..0 (a), while the Six-Switch Vienna rectifier ses the two switches strctre given in Fig..0 (b). (a) (b) Fig..0 Two bipolar bidirectional switch strctres In report [27], a derivation of nidiractional three-phase rectifier circits based on conventional bidirectional rectifier topologies, sch as the Netral Point Clamped (NPC), the Flying Capacitor and the converter employing symmetric Cascaded H-Bridges is discssed. A proposed nidirectional rectifier topology based on the NPC converter is shown in Fig... In this topology, the condction losses can be frther redced and high efficiency can be expected. In conclsion, the three-phase Vienna rectifier topology is an ideal candidate for applicarions reqiring high power density, high performance regarding power factor and inpt crrent qality. In [28], the Vienna rectifier topology is compared with a 2-plse passive rectifier system and a conventional six-switch rectifier circit, and the Vienna rectifier scores well and show several advantages. -3-

32 Chapter. State of the Art Review Fig.. NPC-converter-based nidirectional rectifier.2 Wide Bandgap Semicondctor Power Devices Silicon (Si) material has been widely sed in semicondctor power devices for decades. The technology of silicon power devices has been matrely developed as well. For siliconbased power devices, the limits in terms of power density, operating temperatre and switching freqency have almost been reached [29]. However, power converters with higher power density, operating temperatre, and switching freqency are very demanding in several application areas, sch as mining, military, transportation, renewable energy and etc. A new technological breakthrogh based on wide bandgap semicondctor materials, shows advantages to overcome the limits of silicon semicondctor materials. The bandgap characterization of a material pertains to the energy reqired for an electron to jmp from the top of the valence band to the bottom of the condction band within the semicondctor. The -4-

33 Chapter. State of the Art Review term "wide bandgap materials" refers to semicondctor materials that typically reqire energy larger than two or three electron-volts (ev) [30]. Nowadays, two very important wide bandgap materials showing great promise for ftre are Gallim Nitride (GaN) and Silicon Carbide (SiC). Based on development of wide bandgap materials, breakthrogh in semicondctor technology is nderway. Wide bandgap semicondctors show several advantages over traditional semicondctors and are gradally changing the market..2. Material Properties Table. [3] shows several material properties of silicon (Si), silicon carbide (SiC) and Gallim Nitride (GaN), inclding bandgap, critical field, electron mobility, and electron satration velocity. These material properties have major inflence on fndamental performance characteristics of the corresponding devices. Material properties Si SiC GaN Bandgap [ev] Critical field [MV/cm] Electron mobility [cm 2 /Vsec] Electron satration velocity [0 6 cm/sec] Table. Material properties I [3] The bandgaps of SiC and GaN are 3.2 ev and 3.4 ev respectively, which are abot three times higher comparing to that of Si (. ev). The bandgap property has major inflence on the concentration of intrinsic carriers n i in a semicondctor device in the form of an exponential fnction, represented by (.) [32]: -5-

34 Chapter. State of the Art Review 2 E G n i NcNv exp (.) k BT, where E G is the bandgap energy, T is the temperatre, N c is related to condction band density of states, N v is associated with valence band density of states, and k B is the Boltzmann constant. The relationship between intrinsic carriers and temperatre is also plotted in Fig..2, where characteristics of several materials in different crystal strctres are shown. Fig..2 Concentration of intrinsic carriers in semicondctor vs. temperatre [33] Since the p-n jnction leakage crrent is proportional to ni in a qadratic fnction, it is also determined by the bandgap energy E G of the material and the operating temperatre T. When the operating temperatre is high, the p-n jnction leakage crrent increases, ths the device losses increase. The main reason that the operating temperatre of normal Si devices is limited is that the leakage crrent is significant at high temperatre. However, for wide -6-

35 Chapter. State of the Art Review bandgap materials, sch as SiC and GaN, the p-n jnction leakage crrent in these materials can remain relatively low at high temperatre [34]. It is becase they have higher bandgap energy compared with Si, so that the inflence on leakage crrent de to temperatre increase can be redced. As a reslt, the wide bandgap characteristics allow SiC and GaN power devices to operate nder mch higher temperatre conditions than Si power devices. It is re orted that a -Si F T can o erate at e tremely high ambient tem eratre to C [35]. Fig..3 The specific on-resistance vs. breakdown voltage [37] Besides, SiC and GaN have higher critical field than Si,.7 and 0 times higher respectively, as shown in Table.. The higher critical field allows a higher breakdown voltage for identical epitaxial thickness, which means for a certain breakdown voltage, thinner epitaxial layer is needed for materials with higher critical field. As a reslt, very low on-resistances and very high breakdown voltages can be achieved by SiC and GaN devices -7-

36 Chapter. State of the Art Review [36]. Fig..3 shows the specific on-resistance vs. breakdown voltage for Si, SiC and GaN devices. This low resistance and high breakdown voltage characteristic is especially significant for high-power and high-efficient applications. Moreover, the electron mobility and electron satration velocity are important material parameters as well. They determine the switching performance of power devices by directly affecting the transcondctance and otpt gain [38]. As can be seen from Table. that GaN has the highest electron mobility and electron satration velocity, it has great potential to be the best material to make semicondctor devices for high freqency operation [3]. Table.2 shows the material properties associated with thermal performance. Higher thermal condctivity means the material is sperior in condcting heat more efficiently. Since the thermal condctivity of SiC is higher than Si and GaN, the heat can be easily dissipated in this material and high power density can be achieved. As a reslt, SiC material has advantages in high power applications. Material properties Si 4H-SiC GaN Thermal condctivity [W/cmK] [3] / 4.* BFoM [4] *The theoretical vale of GaN thermal condctivity [39] Table.2 Material properties II Crrently, the thermal condctivity of GaN is only.3 W/cmK, which makes GaN material less competitive in high power application. However, the theoretical vale of GaN thermal condctivity is estimated as 4. W/cmK [39], which is comparable to 4H-SiC. The key limitation for GaN thermal condctivity to reach its theoretical vale is the technical difficlty to form high qality defect-free GaN layers [40]. The crrently available defect GaN crystal reslts in lower thermal condctivity. Recently, development and improvement -8-

37 Chapter. State of the Art Review of GaN process was made to increase GaN thermal condctivity [40], and 2.53 W/cmK thermal condctivity was already achieved [4]. Frthermore, other than thermal condctivity, Baliga's figre of merit (BoFM) is another parameter inflencing device thermal performance. Baliga's figre of metir (BoFM) is a measre of the on-resistance of a nipolar device, which dominates the resistive condction loss of the device. A larger vale indicates smaller on-resistance and lower condction losses. In Table.2, the vales are calibrated to.0 relative to Si. The condction loss of a SiC and GaN device decreases by a factor of 500 and 2400 respectively when compared with a Si conterpart with identical device size and thermal dissipation area [42]. In conclsion, both silicon carbide (SiC) and gallim nitride (GaN) exhibit great advantages over silicon (Si) semicondctors in terms of power density, operating temperatre and switching freqency. By properly applying these wide bandgap materials to power devices and flly taking their advantages, power electronics will be able to operate in extreme high freqency, high temperatre and high efficiency conditions. SiC and GaN power devices are promising to dominate the ftre market of semicondctor power devices..2.2 Existing Wide Bandgap Devices Wide banggap semicondctors, sch as SiC and GaN, have been researched for a long time in laboratories. SiC and GaN light-emitting diodes (LEDs) have been well developed and commercially available for a long time. However, SiC and GaN semicondctor power devices jst began to present in the market in recent years. In this section, several representative SiC and GaN power devices are introdced. SiC schottky diode In general, SiC schottky diode is the most matre wide bandgap semicondctor power device in commercial market. Compared with traditional Si schottky diode, the biggest -9-

38 Chapter. State of the Art Review advantage of the SiC schottky diode is its extremely high rating voltage of 700 V [43]. For conventional Si schottky diode, typical rating voltage is 50 V. Althogh some 200 V Si schottky diodes are available, bt they have high reverse leakage crrent which is sensitive to operating temperatre, and the thermal stability is not good. Considering the shortcomings of Si schottky diode in terms of low rating voltage and bad thermal stability, SiC schottky diodes obviosly otstand with high rating voltage and good thermal stability. Since 20, 700 V SiC schottky diode has become available and its operating temperatre can be as high as 200 C. Since the rating voltage level and operating temperatre level exceeds the reach of Si schottky diode, SiC schottky diode has already taken the market to a large degree. SiC FET SiC FETs with the voltage rating in the range of 600 V to 700 V has become available in the market. Compared with Si FETs, SiC FETs otstand in performance obviosly, especially in high rating voltage range. For devices rating at V, there are a few Si FET prodcts sing advanced semicondctor technologies that are comparable to SiC FETs, sch as MDmesh by STMicroelectronics and CoolMOS by Infineon. However, for devices rating at V, SiC FETs are dominating the market. At this voltage level, SiC FETs have relatively small on-resistance and allow large rating crrents, while Si FETs have very large on-resistance and the crrent ratings are limited to few amps. Considering prodcts at this voltage level, a variety of choices are available for SiC devices, bt the se of Si devices is qite limited. On the other hand, SiC FETs have mch higher price compared with Si ones, which limits the wide penetration of SiC devices. To compare SiC FETs with Si FETs, a SiC FET and a Si FET with identical rating voltage of 200 V and similar on-resistance of 0.35 Ω are chosen as an example. Detailed comparison of parameters are shown in Table.3. Since the two FETs have the same onresistance, they will have similar condction loss dring operation. However, the performance of the SiC device is - 2 orders of magnitde better then the Si device, indicated by the inpt capacitance and otpt capacitance. Considering the body diode reverse recover time and charge, SiC FET has mch better performance than Si FET in high -20-

39 Chapter. State of the Art Review switching freqency operations as well. Moreover, SiC FETs with mch lower on-resistance are available in the market, bt no sch availability for Si FETs. Si SiC Manfactre IXYS Rohm Manfactre # IXFB30N20P SCT2280KE Rds(on)@25 C [Ω] Inpt cap. [pf] Otpt cap. [pf] Body diode reverse recovery time [ns] Body diode reverse recovery charge [nc] Table.3 Comparison between 200 V Si and SiC devices GaN Diode Researches directed towards GaN diodes have been taken place in niversities and laboratories for a while. However, commercial GaN diodes are not available in the market nowadays. In 205, Panasonic annonced the development of a GaN diode with high operating crrent (7.6 ka/cm²) and low trn-on voltage (0.8 V) based on a hybrid strctre of a low-voltage nit and a high-crrent-capable nit [5]. Since the GaN diode has achieved simltaneos high-crrent operations and low threshold voltage, it can handle high crrent with a small chip area. The capacitance of the chip is also redced, and lower switching losses and high freqency operation can be achieved. GaN diodes have promising ftre in the power device market considering its advantages, bt technology breakthrogh is still needed to bring commercial GaN diodes to the market. -2-

40 Chapter. State of the Art Review GaN FET Depletion mode (normally-on) GaN FET devices first started to appear in 2004 by Edyna Corporation in Japan [44]. In recent years, a large nmber of researches focsing on high performance GaN transistors have been done, bt the progress has never been easy. De to the thermal mismatch between different materials, there exists a thermal stress between GaN and the sbstrate when the temperatre changes. As a reslt, GaN is very difficlt to grow on the defect-free sbstrate [45]. Tremendos work has been down to improve the growth of GaN layer on Si sbstrate [46], sapphire sbstrate [47] or SiC sbstrate [48]. GaN has different advantages on different sbstrate materials. In general, Si material is the most common and commercialized sbstrate de to its cheap price. Bt it is difficlt to grow GaN on a Si sbstrate becase of the thermal mismatch. In contrary, it is easier to grow GaN on a SiC sbstrate, since GaN and SiC has a smaller thermal mismatch. Besides, the high thermal condctivity of SiC improves the thermal performance, which is significant in high power density applications. The drawback of SiC as the sbstrate of GaN is the high cost of the material. The depletion mode (normally-on) is an inherent behavior of GaN devices with conventional strctre. This behavior is de to the strong bilt-in polarization electric field, which cases two dimensional electron gas (2DEG) in the channel [49]. Since 2DEG cannot be easily depleted at zero gate voltage, GaN devices exhibit normally-on behavior. Generally speaking, normally-on behavior is not desirable for semicondctor devices, especially in voltage sorce power converter applications. The key reason is that the control nit loses its fnctionality in several conditions, inclding start-p, nder-voltage transient and etc., which leads to short-circit problems, device damages, control nit damages, or even inpt power sorce damages. In conclsion, normally-on devices reqire more components and more complicated circit design to garantee reliability. Therefore, normally-off devices are more preferred in power electronics applications. -22-

41 Chapter. State of the Art Review Nowadays, commercialized normally-off GaN devices are available, and the mainstream technologies transferring normally-on behavior to normally-off behavior divided into two major branches. The first is the invention of enhancement mode GaN devices, which have normally-off behavior. This technology breakthrogh was made by Efficient Power Conversion Corporation (EPC) that they introdced the first enhancement mode GaN FET on Si sbstrate. For enhancement mode GaN devices, the 2DEG problem is solved so that the devices exhibit normally-off behavior. Fig..4 shows the lateral strctre of an EPC GaN FET. EPC is one of the leading corporations providing advanced commercial enhancement mode GaN FET devices. A wide range of prodcts are available with the voltage ratings from 30 V to 450 V. These devices can be applied p to several kilowatts power converter applications. Fig..4 The strctre of EPC GaN FET [44] -23-

42 Chapter. State of the Art Review Fig..5 Illstration of cascode soltion [50] Another major soltion converts a normally-on GaN device into an eqivalent normallyoff GaN device by combining an enhancement mode Si device and a depletion mode GaN FET into a cascade switch. Hence, it is referred as "cascode soltion". As shown in Fig..5, the gate of the GaN FET is tied to the sorce of the bottom Si switch, ths the gate voltage of the pper GaN switch is the sorce to drain voltage of the bottom Si switch, which is a negative vale and secrely trns off the pper GaN device. As a reslt, the cascode switch behaves as a normally-off device. In 204, GaNSystems started to provide 00 V and 650 V cascode GaN devices. To compare GaN FETs with Si FETs, a GaN FET and a Si FET with identical rating voltage of 200 V and similar rating crrent of abot 20 A are chosen as an example. More comparison of parameters are shown in Table.4. It can be seen clearly that the on-resistance of GaN FET is nearly nine times smaller than the on-resistance of Si FET, which means the condction loss of the GaN device is mch smaller than that of the Si device. Compared with the Si device, the GaN device has better parameters in total gate charge, inpt capacitance, otpt capacitance and reverse transfer capacitance. Therefore, the switching loss of the GaN device is mch smaller than that of the Si device as well. Better performance in high switching freqency operation applications can be expected with the GaN device. -24-

43 Chapter. State of the Art Review Manfactre Si International Rectifier GaN EPC Manfactrer # IRF640N EPC200C VDS [V] ID [A] 8 22 Rds(on)@25 C [Ω] Total gate Charge [nc] Inpt cap. [pf] Otpt cap. [pf] Reverse Transfer cap. [pf] 53.8 Table.4 Comparison between 200 V Si and GaN devices To smmarize, commercial GaN FETs are available in the market. GaN FETs have many advantages over Si FETs, bt the device price is crrently very high compared to its Si conterparts..3 Research Opportnities and Challenges Theoretical limits of the silicon (Si) based power devices have almost been reached dring years of technology improvements, which increases the difficlty in frther improving device and system level performance [29]. Wide bandgap semicondctor material sch as gallim nitride (GaN) has many advantages inclding wide band gap, high electron satration velocity, high critical breakdown electric field, high thermal condctivity, etc. The GaN power devices have low ON-resistance, fast switching speed, nice thermal condctivity, small size and high reliability. Ths GaN power devices can be more sitable for high -25-

44 Chapter. State of the Art Review freqency, high power density, and high temperatre applications [55, 56]. Until recently, the research and applications of GaN power devices are still limited. The enormos benefits and possible applications of GaN devices have not been flly explored yet. The three-phase three-level three-switch Vienna rectifier is widely sed in electric aircraft systems, wind trbine systems and power factor correction systems. It has advantages of low inpt crrent harmonics, low blocking voltage stress on the power semicondctor devices, high power density, high efficiency, and high reliability [52]-[54]. However, all Vienna rectifiers were stdied based on Si power devices. The Vienna rectifier system can be frther improved when the Si power devices are replaced with GaN power devices. Since the Vienna rectifier topology only ses three switches, one in each phase, there is no shoot-throgh possibility and dead time is ths not necessary in a switching period. As a reslt, the switching freqency of the switches can be pshed even higher. With GaN FETs, the new system has a lot of potential improvements in terms of the power loss of switching devices, the performance of the rectifier like crrent harmonics, the size and power density of the rectifier, etc. This is the first work that the GaN power devices have ever been tilized in the Vienna rectifier. This thesis discsses the advantages and challenges in implementing the Vienna rectifier with GaN FET devices. Based on topology and control discssions, a model of the Vienna rectifier is bilt and verified by simlation. One Vienna rectifier is designed with GaN FETs and the other one is designed with Si MOSFETs on a similar scale in order to carry ot a fair comparison of the two types of devices. Comparative analyses on the power loss redctions of power devices, the performance improvements and size redctions of the Vienna rectifier will be stdied based on the two systems. There are many advantages to establish the Vienna rectifier with GaN FET devices. The power loss of power devices can be redced when GaN FET devices are applied. As the On-resistance of GaN FET is small, it has low condction loss. The switching loss is also low de to short voltage rise and fall time of the GaN FET. -26-

45 Chapter. State of the Art Review The size of the Vienna rectifier system can be miniatrized and the power density can be improved. As GaN devices can operate at high switching freqency, the vale and size of passive components, like indctors and capacitors, in the system can be redced. De to low power loss and good thermal condctivity of the devices, heat sinks can be eliminated. As a reslt, high miniatrization and integration can be realized by sing GaN devices in Vienna rectifier, which has promising tilization potentials in integrated modlar motor drive systems and other applications with limited space [57]. However, implementing the Vienna rectifier with GaN FET has some challenges as well. For the Vienna rectifier, topology characteristic of single switch per phase makes the control of the rectifier complex and difficlt. For the GaN FET, since the commercial devices are newly developed, the device tilization and corresponding driver design are challenging. -27-

46 Chapter 2 Vienna Rectifier Topology & Operation This chapter will introdce basic concepts of Vienna rectifier. A general topology of three-level rectifiers will be discssed first. Based on it, the typical Vienna rectifier topology and its operation will be demonstrated. Frthermore, the modlation of the Vienna rectifier based on carrier and SVPWM will be illstrated. 2. General Topology of Three-Level Rectifiers Typically, a general rectifier system consists of an AC side crrent filter, a DC side voltage filter, and a rectifying circit. Respectively, the AC side crrent filter is sally composed of passive indctive and capacitive components; the DC side voltage filter is often implemented by parallel capacitors; the rectifying circit is bilt with flly controlled power electronics switches in varios topologies regarding to different types of rectifier systems. Under the assmption that the switches are ideal and driving signals directly control switches' ON/OFF stats, the rectifier system can be seen as the DC otpt is connected to the AC inpt in a certain form. 2.. Single-phase Three-level Rectifying Circit Topology The basic classification of the rectifying circit is based on the voltage level. The term "level" implies the nmber of the voltage levels that can be achieved dring operation on the AC side of the rectifying circit. For instance, "two-level" demonstrates that there are two possible voltage vales on the AC side of the rectifying circit, while "three-level" indicates three vales. The topology of the typical three-level rectifying circit is shown below, where both the symbolic rectifying circit (Fig. 2. (a)) and its eqivalent circit consisting ideal switches (Fig. 2. (b)) can be seen clearly. In the ideal-switch eqivalent circit, the switches are represented by S ij. For Sij, the switch is ON, and for Sij 0, the switch is OFF. Since the otpt of the rectifying system is DC voltage, the otpt terminals 'p', 'o', and 'n' cannot

47 Chapter 2. Vienna Rectifier Topology & Operation connect to each other. Therefore, only one of the three switches shown in Fig. 2. (b), S i, S i 2, and i3 S, can trn ON at a time. Hence, the three switches, S i, S i2, and S i3, make a onepole three-throw switch. This is the operation principle for three-level rectifying circits. i DC p i DC p i ii Rectifying Circit i o DC o DC i ii S i S i3 i o DC o DC DC 2 S i2 DC 2 n n (a) (b) Fig. 2. (a) Symbolic topology of single-phase three-level rectifying circit; (b) Eqivalent ideal switch topology of single-phase three-level rectifying circit Take point 'o' on the DC side as the voltage reference of the system, and the connection stats of the rectifying circit can be analyzed based on switches' connecting stats as follows. () When Si, Si2 0, and Si3 0, which means S i is ON, and S i2 and S i3 are OFF, define Si. Now io DC, i DC i i, and io 0. (2) When Si 0, Si2, and Si3 0, which means S i2 is ON, and S i and S i3 are OFF, define Si. Now io DC 2, idc 0, and io 0. (3) When Si 0, Si2 0, and Si3, which means S i3 is ON, and S i and S i2 define Si 0. Now io 0, idc 0, and i o i i. As a reslt, there are three possible voltages vales ( DC, 0, DC are OFF, ) on the AC side of the rectifying circit ( io ). The AC side voltage io is a fnction of the switch stats and DC

48 Chapter 2. Vienna Rectifier Topology & Operation side voltages (2.). And the DC side crrent i DC, and the netral point crrent i o are fnctions of both the switch stats and AC inpt crrent i i (2.2)-(2.3). f ( S,, ) S S (2.) io v ij DC DC 2 i DC i2 DC 2 i f ( S, i ) S i (2.2) DC id ij i i i i f ( S, i ) S i (2.3) o io ij i i3 i 2..2 Three-phase Three-level Rectifying Circit Topology Based on the topology of the single-phase three-level rectifying circit, three-phase three-level rectifying circit topology can be developed. The symbolic topology of threephase three-level rectifying circit is shown in Fig Combining Fig. 2. (b) and Fig. 2.2, the eqivalent ideal switch topology of three-phase three-level rectifying circit can be derived, as shown in Fig Corresponding eqations of three-phase AC side voltages, the DC side crrent and the netral point crrent can also be written as (2.4)-(2.7). a i a i DC p b ib Rectifying Circit i o DC o DC DC 2 c i c n Fig. 2.2 Symbolic topology of three-phase three-level rectifying circit -30-

49 Chapter 2. Vienna Rectifier Topology & Operation p S a S b S c DC a b c S a3 S b3 i o o DC S c3 DC 2 S a2 S b2 S c2 n Fig. 2.3 Eqivalent ideal switch topology of three-phase three-level rectifying circit Inpt AC phase voltages S S S S S S ao a DC a2 DC 2 bo b DC b2 DC 2 co c DC c2 DC 2 (2.4) Inpt AC line voltages ( S S ) ( S S ) ( S S ) ( S S ) ( S S ) ( S S ) ab ao bo a b DC a2 b2 DC 2 bc bo co b c DC b2 c2 DC 2 ca co ao c a DC c2 a2 DC 2 (2.5) i S i S i S i (2.6) DC a a b b c c i S i S i S i (2.7) o a3 a b3 b c3 c -3-

50 Chapter 2. Vienna Rectifier Topology & Operation 2.2 Vienna Rectifier Topology and Operation The Vienna rectifier topology was proposed by Professor Johann W.Kolar in 994. This topology has many otstanding advantages, sch as high efficiency, high power density, low voltage stress, sinsoidal inpt crrent waveform, small volme of indctive components, and capability to operate nder nbalancing voltage conditions. In this section, the implementation of the one-pole three-throw switch with one active power electronics device and for diodes is introdced. Apart from the key switch of the rectifier, the Vienna rectifier power circit and its operation is discssed as well Bipolar Bidirectional Switch The main component in the Vienna rectifier power circit is the one-pole three-throw switch. This is a bipolar (voltage) bidirectional (crrent) switch. The position of the throw of this switch determines the connection between AC side and DC side of the rectifier. As shown in Fig. 2.4 (a), a typical implementation of the switch with one power electronics device and for diodes is given. Both the ON/OFF stats of the power electronics device and the direction of the AC side crrent determine the connection stats of the switch. Detailed switch connection modes are shown as well. When the power electronics device is OFF and the crrent flows from the AC side to the switch, terminal and terminal 3 are connected (Fig. 2.4 (b)). When the power electronics device is OFF and the crrent flows from the switch to the AC side, terminal and terminal 4 are connected (Fig. 2.4 (c)). When the power electronics device is ON and the crrent flows from the AC side to the switch, terminal inpts the crrent and terminal 2 otpts it (Fig. 2.4 (d)). When the power electronics device is ON and the crrent flows from the switch to the AC side, terminal 2 inpts the crrent and terminal otpts it (Fid. 2.4 (e)). The one power electronics device and for diodes topology of the one-pole three-throw switch has several advantages. Only one power electronics device is needed in this switch -32-

51 Chapter 2. Vienna Rectifier Topology & Operation topology. The control of the switch is easy. There is no shoot-throgh problem and dead time design is not necessary dring operation. However, this topology has some shortcomings: There are always two diodes connecting in each switching mode, so the forward voltage drop is high and the condction loss on the diodes is high. In this research, GaN device will be tilized in the Vienna rectifier. In order to simplify the control and drive system of the GaN device, the one power electronics device and for diodes topology is selected S 2 S 2 S (a) (b) (c) 3 3 S 2 S 2 (d) 4 4 (e) Fig. 2.4 Switch implementation with one power electronics device and for diodes and its operation modes -33-

52 Chapter 2. Vienna Rectifier Topology & Operation Topology of Vienna Rectifier The topology of Vienna rectifier based on the switch shown in Fig. 2.4 (a) is shown in Fig The Vienna rectifier system consists of three AC side filtering indctors (L and R), three bipolar bidirectional switches, a diode bridge, two DC side filtering capacitors ( C and C 2 ), and load ( R L ). The three bidirectional switches are implemented sing a single nidirectional switch SWi in combination with a diode bridge, and they are advantageosly integrated in the diode bridge ( Da, Da, Db, Db, Dc, Dc ) of the rectifier system. The otpt capacitors are split and the bidirectional switches are connected to the otpt voltage midpoint o. p Da Db Dc i p i C i L C DC ia SWa i b SWb i c SWc i o o R L DC C 2 DC 2 Da Db Dc i C2 i n n R R R L L L a b c N Fig. 2.5 Topology of Vienna rectifier -34-

53 Chapter 2. Vienna Rectifier Topology & Operation Operation of Vienna Rectifier Basic Operation of Vienna Rectifier Based on the Vienna rectifier topology given in Fig. 2.5, the basic operation of the Vienna rectifier can be analyzed. The inpt of the rectifier system, which is connected to the boost indctor L, can be switched to 0 V (o point) by closing switch Si, or depending on the crrent direction, switched to diodes Di and U DC (p point) or 2 U DC (n point) by the free-wheeling 2 Di. Ths, Vienna rectifier shows the three-level rectifier behavior. To smmarize, the operation stats of the Vienna rectifier is determined by the inpt crrent direction of each phase and the switching states of each switch. Based on the polarity of three phase inpt voltages, one power freqency (50Hz or 60Hz) period can be divided into six sections, and each section is 60. In each section, two phase voltages have the same polarity. An example of section classification is given in Table 2.. In each section, the directions of three-phase inpt crrents are also determined. Section Section 2 Section 3 Section 4 Section 5 Section 6 a b c a b c a b c a b c a b c a b c Table 2. Section division based on phase voltage polarity Define the switching fnction of the Vienna rectifier as Si: The switching fnction Si is a fnction of switch ON/OFF stats ( SWi, switch is ON; SW 0, switch is OFF) and inpt crrent direction, as expressed by (2.8). i -35-

54 Chapter 2. Vienna Rectifier Topology & Operation S [ SW ] sign( i ) (2.8) i i i In different sections, the same switch stats SW i leads to different switching fnction S i since the crrent direction is different. Take section 2 as an instant, the relationship between switching fnction S i and switch stats SW i is given in Table 2.2. Besides, the corresponding netral point crrent i o is given nder each switch stats in section 2 as well, in Table 2.2. The existence of i o cases voltage nbalance problem between C and C 2 that. DC DC 2 SW a SW b SW c S a ic i b i a i a i b i c Table 2.2 Switching states in Section 2 ( a 0, b 0, c 0 ) The operation of Vienna rectifier in other sections is very similar to that in Section 2. By controlling the trn-on and trn-off of switches ( S b S c ), the inpt crrent can be reglated to follow crrent demand, and the otpt voltage can be reglated steady as well. i o -36-

55 Chapter 2. Vienna Rectifier Topology & Operation Voltage Space Vector of Vienna Rectifier The Vienna rectifier has three phases and each phase has three switching fnction vales (referring to switching fnction definition), ths there are 27 theoretical switching states for the rectifier system. However, based on the specific topology of Vienna rectifier, switching fnction ( S a, S b, S c ) does not have (+, +, +) and (-, -, -). As a reslt, there are only 25 switching stateses of the system. Voltage space vectors of the Vienna rectifier can be fond based on the 25 switching stateses of the system. Under stable operation conditions, the voltages on two separate DC DC capacitors can be assmed to be the same ( DC DC 2 ). Then the rectifier inpt 2 voltages between the inpts of the rectifier and the netral point o are (2.9) 2 DC S S S ao bo co a b c The phase voltages between the inpts of the rectifier and the mains netral point N are an ao No bn bo No cn co No (2.0) Since the sm of the three-phase mains voltages is zero c in 0 (2.) ia The voltage between netral point o and mains netral point N can be calclated No c DC Si (2.2) 6 ia -37-

56 Chapter 2. Vienna Rectifier Topology & Operation Therefore, the voltage space vectors of the Vienna rectifier can be calclated by r ( an bn cn ) ( ao bo co ) DC ( Sa Sb Sc) (2.3) j e 3 where. Based on Eqations (2.8)-(2.3) and the 25 switching states of the Vienna rectifier, 9 different voltage space vectors, inclding 8 non-zero vectors and zero vector, can be obtained. Among the 8 non-zero vectors, there are 6 long vectors with the norm of 2, 6 medim vectors with the norm of, and 6 short vectors with the norm of. Table 2.3 shows the list of the voltage space vectors of the Vienna rectifier, and Fig. 2.6 shows the space vector diagram. Table 2.3 Voltage space vectors of Vienna rectifier i a Crrent direction i b i c SW a Switch ON/OFF SW b SW c [ Sa Switching states Sb S c ] [, -, -] [, -, 0] [, 0, -] [, 0, 0] Phase rectifier inpt voltage an bn cn 2 2 DC 0 2 DC 2 DC 0 6 DC 2 DC 6 DC Space vector norm 2-38-

57 Chapter 2. Vienna Rectifier Topology & Operation [0, -, -] [0, -, 0] [0, 0, -] 6 DC 6 DC 6 DC 6 DC 6 DC 6 DC [0, 0, 0] [,, -] [,, 0] [, 0, -] [, 0, 0] 6 DC 6 DC 2 DC [0,, -] [0,, 0] [0, 0, -] 6 DC 6 DC 6 DC 2 DC 6 DC 2 2 DC 6 DC 2 DC 6 DC [0, 0, 0] [-,, -] [-,, 0] [-, 0, -] [-, 0, 0] 2 DC 6 DC DC 6 DC 6 DC 6 DC

58 Chapter 2. Vienna Rectifier Topology & Operation [0,, -] [0,, 0] [0, 0, -] 6 DC 6 DC 2 DC 6 DC 2 DC 6 DC [0, 0, 0] [-,, ] [-,, 0] [-, 0, ] [-, 0, 0] [0,, ] [0,, 0] [0, 0, ] 2 2 DC 0 2 DC 6 DC 6 DC 2 DC 0 6 DC 6 DC 6 DC 2 DC 6 DC 6 DC 6 DC [0, 0, 0] [-, -, ] [-, -, 0] [-, 0, ] [-, 0, 0] 6 DC 2 6 DC 0 2 DC 6 DC 2 DC 6 DC

59 Chapter 2. Vienna Rectifier Topology & Operation [0, -, ] [0, -, 0] [0, 0, ] 6 DC 6 DC 2 DC 6 DC 2 DC 6 DC [0, 0, 0] [, -, ] [, -, 0] [, 0, ] [, 0, 0] 2 DC 6 DC [0, -, ] [0, -, 0] [0, 0, ] 6 DC 6 DC DC 6 DC 2 DC 6 DC 6 DC 6 DC 2 DC 6 DC [0, 0, 0]

60 Chapter 2. Vienna Rectifier Topology & Operation ( + ) (0 + ) (+ + ) ( + 0) (0 + 0) ( 0 ) (+ + 0) (0 0 ) (+ 0 ) ( + +) (0 + +) ( 0 0) (0 0 0) (+ 0 0) (0 ) (+ ) (0 0 +) (+ 0 +) ( 0 +) ( 0) (0 0) (+ 0) ( +) (0 +) (+ +) Fig. 2 6 Voltage space vector of Vienna rectifier By selecting proper voltage space vectors and creating optimal switching seqences, space vector modlation can be achieved to modlate the Vienna rectifier. Detailed modlation analysis is discssed in the following section. 2.3 Modlation of Vienna Rectifier A modlator is essential in the rectifier system to generate proper PWM signals (optimal switching seqence) to drive power electronics switches based on the specific modlation index m i given by the system controller. With proper modlation, the inpt crrents and otpt voltage can follow corresponding demands which are represented by the modlation index m i. Two typical modlation methods of the Vienna rectifier is introdced as follows: The first is space vector PWM (SVPWM) modlation, and the other is carrier based PWM modlation. -42-

61 Chapter 2. Vienna Rectifier Topology & Operation 2.3. Space Vector PWM (SVPWM) Based Modlation Since the Vienna rectifier has plenty of voltage space vectors, as shown in Fig. 2.6, space vector PWM (SVPWM) can be sed to modlate the rectifier. However, considering the inherent limitations of the inpt crrent direction on the voltage space vectors, only 7 vectors in a certain section of the space vector diagram are achievable at a time, e.g. the dark area shown in Fig This featre distingishes the Vienna rectifier SVPWM modlation from the general SVPWM modlation. A typical control flowchart of the Vienna rectifier SVPWM is shown in Fig Voltage & Crrent Sampling Reslts Inpt crrent vector calclation Demand voltage space vector calclation Determine the inpt crrent vector section Obtain the space vector section Determine the section of the demand voltage space vector Select proper voltage space vectors to bild p the demand vector Calclate the working time of each sefl voltage space vector and obtain the switching seqence Generate PWM plses Fig. 2.7 Control flowchart of the Vienna Rectifier SVPWM -43-

62 Chapter 2. Vienna Rectifier Topology & Operation It can be conclded from Fig. 2.7 that the first step of SVPWM in the Vienna rectifier is determining the inpt crrent space vector and its position. Based on the crrent section, the section where the voltage demand space vector belongs to need to be calclated. Since there are only 7 available voltage space vectors at a time de to the inherent limitations of the Vienna rectifier, the demand voltage space vector locating can only be done within one crrent section (Table 2.). Several times of calclation and jdgments are necessary to obtain correct section of the demand voltage space vector and determine effective voltage space vectors in the three-level space vector diagram. Another SVPWM method introdced in reference [58] transfers the three-level space vector diagram into two-level space vector diagram by coordinate transformation, ths the demand voltage space vector can be bilt in the two-level space vector diagram. Then the modlation reslts are transferred back to three-level space vector diagram with inverse coordinate transformation. In this method, fewer calclation and jdgments are needed to determine the section of the demand voltage space vector and corresponding effective voltage space vectors at the cost of coordinate transformation PWM Carrier Based Modlation Thogh a SVPWM modlation can be applied to implement the optimal switching seqence, it reqires high efforts of comptation. Therefore, its relatively long comptation time makes SVPWM not sitable for systems operating nder very high switching freqencies. A plse width modlation sing a well-designed carrier signal intrinsically achieves the optimal switching seqence is invented in [59]. This allows the redction of the processing demand. In practical operation, the modlator generates proper PWM signals to drive power electronic devices in three phases based on modlation index modlation index -44- * m i. The magnitde of the * m i is determined by the rectifier system controller, while the polarity of

63 Chapter 2. Vienna Rectifier Topology & Operation the modlation index expressed by (2.4). * m i corresponds to the direction of the phase inpt crrent, as m m sign( i ) (2.4) * i i i where m i is the magnitde of the modlation index, which is always a vale between 0 and and is the otpt of the rectifier system controller. The dty cycle can be expressed as a fnction of the modlation index as (2.5) (2.5) where the dty ratio is vale between 0 and. The proposed three-phase modlator sing two nipolar, 80 -phase shifted trianglar carrier signals is shown in Fig. 2.8(a). Dependent on the inpt crrent direction either the switch signal SW i or SW i has to be operated. According to Fig. 2.8(a) the modlation signal generated by the crrent controller is directly sed for plse-width modlation of SW i and the inverted modlation signal is sed for is the logic AND of SW i and SW i. SW i. And the final switch signal The modlator otpt of each phase is high if the trianglar carrier signal exceeds the modlation index signal * m i or SW i * m i (Fig. 2.8(b)). As the carrier signals for the PWM corresponding to different crrent directions are 80 ot of phase, inverse plse-patterns are generated inherently. In Fig. 2.8 (b), a specific modlation example in the crrent section ( ia 0, ib 0, ic 0 ) is shown. The optimal switching seqence (0)-(00)-(000)-(00)- (000)-(00)-(0) is in consistency with the reslt obtained from the SVPWM modlation method. -45-

64 Chapter 2. Vienna Rectifier Topology & Operation * mi SW i + - SW i And SW i Carrier Signals + (a) M Carrier Signals * m a 0 T / f s s Ts 2T s 3T s * m b * m c t SW a 0 t SW b 0 t SW c 0 ( Sa, Sb, Sc ) (,0,0) (,0,0) (, 0, ) (,0, ) (,, ) (,, ) (0,, ) t (b) -46-

65 Chapter 2. Vienna Rectifier Topology & Operation Fig. 2.8 (a)architectre of the modlator of the Vienna rectifier system; (b) Plse-width modlation sing two 80 -phase-shifted trianglar carrier signals at N 0 in ( ia 0, ib 0, ic 0 ) section In conclsion, both the SVPWM modlation method and the proper carrier based modlation method have the same capability to generate the optimal switching seqence. However, the SVPWM is more complex and processing time consming becase several calclation and jdgments are essential. On the other hand, the proper carrier based modlation is simple and fast. Therefore, in this thesis, the proper carrier based modlation method is selected. -47-

66 Chapter 3 Control of the Vienna Rectifier To design a controller for the active three-phase rectifier, an appropriate model of the rectifier system is reqired. In this chapter, a model for control of the three-phase three-level rectifier system sing state space averaging is introdced, and the model transformation among different reference frames is described. Based on this model, a basic controller strctre inclding the inner-loop crrent controller, the oter-loop DC voltage controller, and the voltage balance controller is demonstrated [6]. Detailed design of each specific controller is indicated respectively as well. Other than the basic controller's strctre based on the instantaneos crrent control, a Direct Power Control (DPC) strctre is also introdced at the end of this chapter. 3. Mathematical Model of Vienna Rectifier System In general, the mathematical model of the Vienna rectifier system can be established sing the state-space averaging techniqe in continos crrent mode. The averaging process is applied on two time intervals: the switching period for average crrent evalation, and the mains period for average voltage comptation. Neglecting the possible inpt EMI filter, the Vienna rectifier system exhibits five energy storage elements (three boost indctors and two otpt capacitors), ths a fifth order model is expected. 3.. Model in Three-Phase Stationary (ABC) Reference Frame Based on the Vienna rectifier topology shown in Fig. 2.5 and analysis of its operation principles given in Chapter 2, the mathematical model of the Vienna rectifier nder the steady-state and balanced load operating conditions in the basic three-phase stationary (ABC) reference frame can be derived (3.).

67 Chapter 3. Control of the Vienna Rectifier di 2 2 a L R 0 0 Sa Sb Sc Sa2 Sb2 Sc2 dt di 2 2 b L 0 R 0 Sa Sb Sc Sa2 Sb2 S c2 ia an dt i b di 2 2 bn c L = 0 0 R Sa Sb Sc Sa2 Sb2 S c2 ic + cn dt d DC 0 DC C S a Sb Sc dt RL R L DC 2 0 d DC 2 C2 Sa2 Sb2 Sc2 dt RL RL (3.) where the S i and S i2 were defined as the ideal switches connecting to positive otpt potential point 'p' and negative otpt potential point 'n' respectively in Section The relationship between S i and S i2 and the switching fnction S i was given in Section Model in Two-Phase Stationary (αβ) Reference Frame Frame transformation from the three-phase stationary ABC reference frame to the twophase stationary αβ reference frame can be achieved by Clark transforming matrix TABC. TABC (3.2) Define -49-

68 Chapter 3. Control of the Vienna Rectifier an T ABC bn cn i i a T ABC i b i i c S S a T ABC S b S S c (3.3) S S a2 2 T ABC S b2 S 2 S c2 and the mathematical model of the Vienna rectifier nder the steady-state and balanced load operating conditions in the two-phase stationary (αβ) reference frame can be obtained (3.4). L di dt R 0 S S 2 di 0 R S S i 2 L dt i S S d R L RL DC 0 DC C dt DC 2 0 S2 S 2 d DC C RL RL dt (3.4) 3..3 Model in Two-Phase Rotating (dq) Reference Frame Frthermore, based on the model in the two-phase stationary (αβ) reference frame, the model in the two-phase rotating (dq) reference frame can be formed with Park transformation matrixes. In the rotating reference frame, all AC terms, sch as rectifier inpt -50-

69 Chapter 3. Control of the Vienna Rectifier crrents and voltages, are transformed to DC terms, which largely simplifies the analysis process of the AC system. The Park transformation matrixes are given in (3.5) and the terms in the dq frame are defined in (3.7). The model of the Vienna rectifier nder the steady-state and balanced load operating conditions in the two-phase rotating (dq) reference frame is shown in (3.7), where θ is the position of the rotating reference frame t, and ω is the anglar speed of the reference frame, which eqals to the anglar freqency of the power grid 2 fgrid. cos sin T dq sin cos (3.5) d T dq q i i d T dq i q i Sd S T dq S q S (3.6) Sd 2 S 2 T dq S q2 S 2 L did dt R L S S d d2 di L R S S i q q q2 d d L dt i q q Sd Sq d R L RL DC 0 DC C dt DC 2 0 Sd2 S q2 d DC C RL RL dt (3.7) -5-

70 Chapter 3. Control of the Vienna Rectifier 3.2 Controller Strctre As the general model of the Vienna rectifier for control is obtained, the controller of the Vienna rectifier can be designed based on the model. In [6], a controller strctre was proposed after the analysis of the basic operation of the rectifier system. The model of the AC-side of the rectifier system and the model of the DC-side of the rectifier system can be seen as decopled to a certain degree. The decopling of the otpt voltage controller from the inpt crrent controller is jstifiable by the different dynamic behaviors of both sides. The dynamic behavior of the inpt crrent control loop is related to the switching freqency f s (relatively high freqency, sally kilo-hertz to mega-hertz) while that of the otpt voltage control loop is associated with the mains freqency f (mch lower than N f s, sally 50 Hz or 60 Hz). Typically, the large otpt capacitors are the decopling elements of the two control loops. As a reslt, a cascade control strctre can be sed in the Vienna rectifier system. A basic cascade control strctre is shown in Fig. 3. where a phase-oriented modlation strategy is assmed in combination with sperimposed otpt voltage and voltage balance controllers. A three-phase crrent controller K () s is reqired to ensre sinsoidal inpt crrents which are in phase with the mains phase voltages or leading/lagging the mains voltages by a limited amont. An otpt voltage controller K () s has to be employed to reglate the otpt voltage constantly. Since the otpt capacitor is split into two parts, a voltage balance controller K () s is essential to balance the corresponding otpt voltages DC and DC 2 S I. Otherwise, the system is not stable and cannot achieve designed performance. The design of the corresponding controllers is smmarized in the following sections. V -52-

71 Chapter 3. Control of the Vienna Rectifier in i i in * G e * i i i i 2 3U in * P o KI () s KV () s KS in () s m m * i * U DC 0 U DC PWM Modlator U U DC U DC 2 SWi AC p DC o U DC U DC 2 n Fig. 3. Basic controller strctre of the Vienna Rectifier system. (Signal paths being eqal to all three phases are shown in doble lines) 3.3 Crrent Controller In the following a very simple linear model for crrent controller design is derived which describes the main behavior of the rectifier system. As already mentioned average mode control is sed, which means that all signals are averaged over one switching period. The inpt of the crrent controller K () s is the difference between two crrents. One is the I reference phase crrent, which has sinsoidal waveform and is in phase with the phase -53-

72 Chapter 3. Control of the Vienna Rectifier voltage. Another is the measred phase crrent. The deviation between the measred phase crrent and the reference phase crrent generates corresponding PWM plse driving signal, which can be expressed by modlation index or dty cycle. Therefore, the otpt of the crrent controller is the PWM plse signal that makes the practical phase crrent following the reference phase crrent. Under the assmptions that the otpt voltages on otpt capacitors are balanced and the common mode voltage on between DC side netral point and mains netral point can be neglected, the operation of the phases is decopled. Therefore, (3.8) can be written sing the dty cycle () t. i dii ( t) DC ( t) in ( t) Li ( i( t)) (3.8) dt 2 This eqation is nonlinear since the otpt voltage is mltiplied by the dty cycle and time-varying as the sinsoidal mains voltage in () t is inclded. The otpt voltage DC () t is controlled to a constant vale by the otpt voltage controller and as the dynamic of the voltage control loop is mch slower than the dynamic of the crrent control loop, a constant otpt voltage ( t)= U can be assmed. This assmption eliminates the nonlinearity in DC DC (3.8). To eliminate the time variance of (3.8), a proper feed-forward signal, () t is applied, as shown in (3.9). Note that an ideal inpt voltage feed-forward signal is assmed which generates the dty cycle according the sinsoidal mains voltages () t and that the crrent controller only has to deal with the deviations from the reference crrent. in ff i ( t) ( t) ( t) res, i ff, i i ff, i ( t) U in DC () t /2 (3.9) Since the nonlinearity and time variance is eliminated, after the Laplace transform, a very simple linear model of the rectifier system is obtained (3.0). In this model, some details, -54-

73 Chapter 3. Control of the Vienna Rectifier sch as the impedance of the mains, the characteristics of the EMI-filter or the delay times of the switches are not considered for the sake of simplicity. ii () s U DC Gs () ( s) 2L s (3.0) i A state block diagram of the simplified crrent controller is shown in Fig In the state block diagram, Gs () is the derived simple model of the Vienna rectifier, K () s is the crrent controller, and M () s is the transfer fnction of the crrent measrement considering the I bandwidth limitation of the crrent sensor. i I i * i () s ii, meas () s * PWM m * i i () s KI () s kpwm MI () s z () s Gs () ii () s KI () s. Fig. 3.2 State block diagram of the crrent controller A simple proportional controller (P controller) can be sed to implement the fnction of K () s K (3.) I P, i Besides, a P + lag controller is also a good choice to KI () s. K () s K I P, i st st D (3.) In this work, the P controller is selected for the sake of simplicity. -55-

74 Chapter 3. Control of the Vienna Rectifier 3.4 DC Otpt Voltage Controller The DC otpt voltage controller K () s of the Vienna rectifier is sed to reglate the V DC otpt voltage of the rectifier system to a specified vale. The inpt of the DC otpt voltage controller K () s is the difference between the reference otpt voltage and the V measred otpt voltage. The final otpt of the voltage controller K () s is the reference crrents, which are sed as demand in the crrent controller K () s. As a reslt, the DC otpt voltage controller KV () s is the oter-loop while the crrent controller KI () s is the inner-loop, which matches the controller strctre shown in Fig. 3.. Previosly, only the AC side of the rectifier system has been modeled to design the crrent controller to control the AC crrents. In order to control the DC otpt voltage DC of the rectifier, an appropriate model of the DC side of the rectifier system has been derived as shown in Fig. 3.3, and the reslts are smmarized as follows. I V i DC i L C i C id id2 i D 3 C R L DC Fig. 3.3 Model of the DC side of the Vienne Rectifier with a resistive load The rectifier system has to obey the rle of power conversation, so that the otpt power p o eqals to the inpt power p in mins the rectifier system power losses p loss (3.2). -56-

75 Chapter 3. Control of the Vienna Rectifier po pin ploss (3.2) The otpt power and the inpt power of the rectifier system can be expressed respectively by (3.3) and (3.4). p i i (3.3) o DC, avg DC, avg DC DC where i DC is the DC side crrent and DC is the otpt voltage. p 3U I cos( ) (3.4) in in i i where U in is the RMS vale of the phase voltage and I i is the RMS vale of the phase crrent on the AC side. The efficiency of the rectifier can be defined by (3.5). po idcdc (3.5) p 3U I cos( ) in in i i Under the assmption that the rectifier system has nity power factor cos( i ) and negligence of the energy stored in the boost indctors, the power conversation rle of the system can be expressed as (3.6). i U I (3.6) DC, avg DC 3 in i and This nonlinear eqation can be linearized arond the operating point I i 0 sing U DC 0, U in 0, I DC 0-57-

76 Chapter 3. Control of the Vienna Rectifier U DC DC0 DC U in in0 in i I i DC, avg DC0 DC i I i i i0 i (3.7) The variation of the DC crrent i DC is inflenced by the variations of the AC crrents i i, the AC voltages in, and the DC voltage DC (3.8). 3U 3I I i i (3.8) in0 i0 DC0 DC i in DC U DC U 0 DC U 0 DC0 Hence, the small signal model between the AC inpt crrents and the DC otpt crrent is fond (3.9). This relationship proves that the DC otpt crrents are proportional to the AC inpt crrents. k p i 3 U DC in0 (3.9) i i U DC0 power The otpt of the DC voltage controller K () s is eqivalent to the demanded otpt * P o. Using the RMS vales of the mains voltages, a condctance V G * e P can be 3U calclated. And the instantaneos reference crrents can be generated based on the condctance and the instantaneos phase voltages (3.20). * * * o i e in 2 in 3U in * o 2 in P i G (3.20) The small signal model can be bilt by linearizing eqation (3.20) arond an operating point P, * o0 in 0 *, and I sing i0-58-

77 Chapter 3. Control of the Vienna Rectifier p P p * * * o o0 o U in in0 in i I i * * * i i0 i (3.2) which reslts in i p P 2I * * * * o0 i0 i o 2 in 3U in 3U 0 in U 0 in0 (3.8) Ths, the relationship between the demanded otpt power and the reference inpt crrents is obtained. k i 3U (3.23) * i p2 * po in0 In the case of a constant resistive load, the load side of the rectifier system can be modeled by H load RL () s (3.24) sr C L Based small signal models (3.9), (3.23) and (3.24), the state block diagram of the DC otpt voltage controller can be drawn, as shown in Fig In the state block diagram, K () s represents the DC otpt voltage controller, T () s represents the closed loop transfer V fnction of the crrent controller, and M () s represents the transfer fnction of the voltage measrement circit considering the bandwidth limitations of the voltage sensors. V I -59-

78 Chapter 3. Control of the Vienna Rectifier p L N * () DC s * p o () Kv s 2 U 3 N * G e * ii TI () s i i 3 U N DC, avg 0 U DC0 i RL DC () s sr C L, () s DC meas M () v s Fig. 3.4 State block diagram of the DC otpt voltage controller In order to prevent steady state DC otpt voltage control errors, a PI-type controller is sally applied. K () s K V K Iv, P, v (3.25) In conclsion, the DC otpt voltage control is achieved in the following process. Firstly, the DC otpt voltage demand determines the reqired otpt power (nder the condition of constant resistive load). Secondly, the reqired otpt power generates the reference AC inpt crrents. Thirdly, the crrent controller garantees that the AC inpt crrents follow corresponding reference crrents well. Then, the DC otpt crrent is proportional to the AC inpt crrents. Finally, the DC otpt voltage is inflenced by the DC otpt crrent becase of the constant resistive load. s -60-

79 Chapter 3. Control of the Vienna Rectifier 3.5 Voltage Balance Controller As discssed in [62] an nbalanced otpt voltage U U ( DC DC 2 ) (3.26) 2 which cold reslt from neqal leakage crrents of the otpt capacitors or from an asymmetrical load of the two rectifier otpts, reslts in an asymmetrical distribtion of the switching actions which finally yields to increased inpt crrent distortions. The average netral point crrent io, avg can be formed by averaging the netral point crrent i o over a switching period i i i i (3.27) o, avg a a b b c c This netral point crrent cases nbalanced otpt voltages according to du io 2C (3.28) dt where two otpt capacitors have the same vale of C. De to the modlation of the rectifier, a third harmonic netral point crrent is generated. As a reslt, the third harmonic netral point crrent leads to a third harmonic otpt voltage nbalance. It can be minimized by injecting a proper third harmonic signal, bt is still present. According to [63], if average mode crrent control in conjnction with a plse-width modlation is applied, the otpt voltage nbalance by Fig. 3.5 where the corresponding modlation indexes The voltage nbalance can be presented by (3.29) -6- U always exists stably. It is verified m are plotted for 0 in the ( ia 0, ib 0, ic 0 ) crrent section when the otpt voltage nbalance U is considered. i N

80 Chapter 3. Control of the Vienna Rectifier U U DC DC 2 U 2 U 2 DC DC U U (3.29) M Carrier Signals m a * m a 0 T / f s s Ts 2T s m b m c 3T s * m b * m c t SW a 0 t SW b 0 t SW c 0 ( Sa, Sb, Sc ) (,0,0) (,0,0) (, 0, ) (,0, ) (,, ) (,, ) (0,, ) t Fig. 3.5 Plse-width modlation and switching seqence illstrating the self stability of an nbalanced otpt voltage According to (3.8) and (2.5), the vales of the otpt voltages U DC and U DC 2 directly inflence the vales of corresponding dty cycle i and modlation index -62- m i. To be more specific, the effects of the otpt voltage nbalance in the condition of N 0, as shown in Fig. 3.5, are analyzed in details as follows. The increased otpt voltage U DC cases the decrease inpt crrent in phase a. Accordingly, the crrent controller increases the dty cycle of phase a, which is condcted by decreasing the modlation index of phase a by ma. On

81 Chapter 3. Control of the Vienna Rectifier the other hand, phase b and phase c generate a relatively low otpt voltage U DC 2. Hence, dty cycles of phase b and phase c are redced by increasing the corresponding modlation index by mb and mc respectively. The reslting modlation indexes, as altered by the crrent controllers, are plotted in Fig. 3.5 by dashed lines and the changes in optimal switching seqence and operating time are shown as well. The optimized switching seqence is nchanged, bt compared to a balanced otpt voltage condition, the dration of switching state ( Sa, Sb, Sc) (0,, ) is enlarged, the dration of switching state ( Sa, Sb, Sc) (,0,0) is redced, and the dration of other switching states is kept nchanged. The two switching states ( Sa, Sb, Sc) (0,, ) and ( Sa, Sb, Sc) (,0,0) are redndant concerning the differential mode inpt crrents, bt they lead to netral point crrents with opposite directions as shown in Fig. 3.6 (a) and (b). As shown in Fig. 3.6 (a), switching state ( Sa, Sb, Sc) (,0,0) reslts in a negative netral point crrent which charges C and discharges C 2. For switching state ( Sa, Sb, Sc) (0,, ), as shown in Fig. 3.6 (b), it yields a positive netral point crrent which discharges C and charges C 2. The asymmetric distribtion of the two redndant switching states cased by the crrent controllers conteract the voltage nbalance U by discharging C and charging C 2. Therefore, it atomatically stabilizes the otpt voltage nbalance. However, thogh the otpt voltage nbalance is stable, it cannot be redced to zero withot active control. A B C L L L C C 2 A B C L L L C C 2 (a) (b) -63-

82 Chapter 3. Control of the Vienna Rectifier Fig. 3.6 (a) Vienna rectifier connection and netral point crrent direction for the switching state ( Sa, Sb, Sc) (0,, ) for 0; (b) Vienna rectifier connection and netral point crrent direction for the switching state ( Sa, Sb, Sc) (,0,0) for 0 N Therefore, an otpt voltage balance controller is reqired to actively balance the two otpt voltages. The redndant switching states with different directions of the netral point crrent i o are sed for active balancing. Active balancing can be achieved by adding a variable DC component m to all three modlation index m a, m b and m c. N m m m a b c a, old m m m b, old m m m c, old (3.30) The reslting modlation signals, switching states and switching seqence are illstrated in Fig Since the DC component m inversely changes the relative on time of the redndant switching states, it can be sed for active otpt voltage balancing. In this specific case, if the otpt voltage nbalance is UDC UDC 2, C needs discharging and C 2 needs charging. Setting m with a proper negative vale wold decrease the nbalanced voltage U. In contrary, if the otpt voltage nbalance is UDC UDC 2, C wold be charged and C 2 wold be discharged when a positive m is selected. -64-

83 Chapter 3. Control of the Vienna Rectifier M 0 Carrier Signals T / f s s Ts 2T s m m m 3T s * m a * m b * m c t SW a 0 t SW b 0 t SW c 0 ( Sa, Sb, Sc ) (,0,0) (,0,0) (, 0, ) (,0, ) (,, ) (,, ) (0,, ) t Fig. 3.7 Inflence of a DC component on the switching seqence of the rectifier system for N 0 m added to all three modlation indexes As a reslt, the inpt of the voltage balance controller K () s is the nbalanced otpt voltage, while the otpt of KS () s is a variable DC component m which can redce the nbalanced otpt voltage by affecting all three phase modlation indexes. The implementation of the voltage balance controller K () s can be a PI controller. K () s K S S K S Is, P, s (3.3) s -65-

84 Chapter 3. Control of the Vienna Rectifier 3.6 Direct Power Controller (DPC) From the perspective of the power of a rectifier system, when the AC mains voltage is constant, if the instantaneos power (active power and reactive power) of the rectifier system can be controlled with in a specific range, then the instantaneos crrent (active crrent and reactive crrent) of the rectifier system can be conseqently controlled. This control strategy is named as Direct Power Control (DPC). In general, a DPC control system consists of a DC voltage control oter-loop and a instantaneos power control inner-loop. By selecting corresponding switching states in the switch table to generate reqired rectifier inpt voltages based on AC voltages and instantaneos power of the rectifier system, high qality rectifying can be achieved. Comparing with traditional crrent controllers, DPC has advantages sch as high power factor, low THD, high efficiency, easy algorithm, simple system strctre, and good dynamics corresponding to load changes. As a reslt, DPC has promising potential to be applied in the three-phase/level/switch Vienna rectifier system. Since the DPC directly controls the instantaneos active power and the instantaneos reactive power of the rectifier system, the key point of this control strategy is the accrate observation of the instantaneos active/reactive power. Instantaneos power theory is the theoretical base of the DPC. In 980's, the instantaneos power theory was proposed by H. Akagi [64], and in 990's, a PWM converter closed-loop control evolved the instantaneos power was proposed and frther developed by T. Ohnishi, T. Nogchi, etc. [65]. The instantaneos active power is defined as the scalar prodct of the instantaneos voltage vector and the instantaneos crrent vector i. The instantaneos reactive power is defined as the vector prodct of the instantaneos voltage vector and the instantaneos crrent vector i. In the three-phase stationary ABC reference frame, the definition of the instantaneos active power and reactive power are given as (3.32). -66-

85 Chapter 3. Control of the Vienna Rectifier p i i i i i cos i a a b b c c p q i i i i i sink i k * * * a a b b c c q (3.32) where p is the instantaneos active power of the AC mains, q is the instantaneos reactive power of the AC mains, i is the AC mains voltages, i i is the AC mains crrents, * a, and * a are srface of and i. * a c b cb * b a c ac * 3 3, k is a nit vector perpendiclar to the c b a ba Eqations in (3.32) can also be written in the matrix form, as shown in (3.33). * a, i a p a b c i * * * b q a b c i c (3.33) In the two-phase stationary αβ reference frame, the instantaneos active power p and the instantaneos reactive power q can be calclated by (3.34). p i q i (3.34) In the two-phase rotating dq reference frame, the instantaneos active power p and the instantaneos reactive power q can be calclated by (3.35). p i d q d q q d iq (3.35) Align the d-axis with the AC mains voltage vector, then q =0. Ths, sbstitte q =0 into (3.35), (3.36) can be gained. -67-

86 Chapter 3. Control of the Vienna Rectifier p d 0 id q 0 i d q (3.36) Based on the model of the Vienna rectifier in the two-phase rotating dq reference frame, given by (3.37), which is another form of (3.7), di dt di d L ed d Rid Liq q L eq q Riq Lid dt (3.37) where e d and e q are mains voltages, d and q are rectifier inpt voltages, i d and i q are mains crrents. Mltiply e d to (3.37) to get (3.38) dided L ed ( ed d Rid Liq ) dt diqed L ed ( eq q Riq Lid ) dt (3.38) Define L L / e and R R / e, and sbstitte (3.36) into (3.38), hence (3.39) can be obtained. new d new d dp Lnew ed d Rnew p Lnewq dt dq Lnew eq Rnewq Lnew p dt (3.39) Eqation (3.39) presents the essence of the DPC controller. The state block diagram of DPC controller is shown in Fig It can be seen clearly that there are two control loops in the DCP system. The first loop is the oter DC voltage control loop, where the sampled DC otpt voltage U DC is compared with the demanded DC reference voltage -68- * U DC, and the voltage error is sed to generate the d-axis crrent command i * d throgh a PI controller. The

87 Chapter 3. Control of the Vienna Rectifier instantaneos active power command * p is the prodct of the d-axis crrent command i * d and the DC otpt voltage nit power factor. U DC. The reactive power command * q is zero in the condition of p q Active Power Reactive Power Rotating Position Calclation in i i r * U DC U DC PI * i d * q 0 * p PI PI q d dq αβ U U DC SVM U DC U DC 2 SW a SW b SW c AC p DC o n U DC U DC 2 Fig. 3.8 DPC controller strctre of the Vienna rectifier system. (Signal paths being eqal to all three phases are shown in doble lines) The second loop is the inner instantaneos power control loop, where the instantaneos active power p and reactive power q, which are calclated based on sampled AC mains voltages and crrents, are compared with active power command * p and reactive power command * q respectively. The power errors are sed to generate desired rectifier inpt voltages ( d, q ) via PI controllers. Once the rectifier inpt voltages are selected, -69-

88 Chapter 3. Control of the Vienna Rectifier corresponding switching states can be chosen by the switch table in the Space Vector Modlator (SVM). Therefore, the rectifier is controlled to perform with specific DC otpt voltage and power (AC crrents) reqirements. Design of the oter DC voltage loop in the DPC system is similar to the otpt voltage controller design discssed in section 3.4, so that it wold not be repeated again. Design of the inner power loop in the DPC system will be illstrated in the following. As can be seen in (3.39), the d-axis and q-axis in the DPC system are cross-copled, which increases difficlty in designing the controller. In order to simplify the controller design and improve control performance, feed-forward decople control strategy is employed in the DPC system. When PI controllers are sed in the inner power control loop of DPC system, the control fnction can be expressed as (3.40) K K p p L q e s Ki * q K p q q Lnew p s i * d p new d (3.40) where controller. K p is the gain of the proportional controller and K i is the gain of the integrating Sbstitte (3.40) into (3.39) to get (3.4). Therefore, the active power p and the reactive power q are decopled and can be controlled independently. dp K K dt s s i * i Lnew K p p Rnew K p p dq K K dt s s i * i Lnew K p q Rnew K p q (3.4) The state block diagram of the decopled power control loops is shown in Fig

89 Chapter 3. Control of the Vienna Rectifier * p PI e d d Ls R e d p L new * q PI q L new Ls R e d q Fig. 3.9 State block diagram of decopled power control loops in the DPC system The state block diagram of the simplified active power control loop is shown in Fig. 3.0, where T K p i and s Ki T is related to sampling period. * p p K p Ts i Ts i e d kpwm R Ts s ( L ) s R p Fig. 3.0 State block diagram of simplified active power control loop in the DPC system By adjsting gains of the PI controllers, the DPC control of the Vienna rectifier can be achieved easily based on the feed-forward decople. -7-

90 Chapter 4 Simlation of the Vienna Rectifier In this chapter, the simlation of the Vienna rectifier system is discssed. In the first part, how to bild the simlation model of the Vienna rectifier system is bilt in MATLAB / Simlink is introdced. The simlation model mainly consists of the power circit, the modlator and the control circit. In the second part, the simlation reslts of the Vienna rectifier system are shown and analyzed. 4. Simlation Model of the Vienna Rectifier In this work, simlation of the Vienna rectifier system was done in MATLAB / Simlink. The simlation model of the Vienna rectifier system can be divided into three main sbsystems, the power circit sbsystem, the modlator sbsystem, and the control circit sbsystem. The general strctre of the simlation model of the Vienna rectifier system is shown in Fig. 4.. It can be conclded from Fig. 4. that at a specific operating point, the voltages and crrents signals can be sampled in the power circit and be sed as inpts to the control circit. The control circit processes the sampled signals and provide appropriate modlation index following the commands as the otpt. Based on the modlation index, the modlator generates proper plses to drive the switches in the power circit so that the power circit can follow the specific commands given by the controller. In the following sections, detailed information abot bilding the power circit simlation sbsystem, the modlator simlation sbsystem, and the control circit simlation sbsystem will be demonstrated conseqently.

91 Chapter 4. Simlation of the Vienna Rectifier Switch Driving Plses Power Circit (AC mains, Boost indctors, DC capacitors, Switches, Diodes, and Resistive load are embedded.) Sampled Signals: AC mains crrents (*3) AC mains voltages (*3) DC otpt voltages (*2) Modlator (PWM carriers, Plse generators are embedded.) Modlation Index Control Circit (Crrent Controller, DC Otpt Voltage Controller, Voltage Balance Controller are embedded. ) Fig. 4. Strctre of the simlation model of the Vienna rectifier system 4.. Power Circit Simlation Model 4... Bild Power Circit Model in MATLAB / Simlink The power circit model of the Vienna rectifier system is mainly composed of the AC mains, AC side boost indctors, switches and diode bridges, DC side capacitors, resistive load, crrent sensors and voltage sensors. The power circit simlation model has exactly the same topology with the Vienna rectifier topology given in Fig Fig. 4.2 (a) shows the power circit model of the Vienna rectifier system in MATLAB / Simlink. From left to right, the AC mains, AC side boost indctors, AC line-to-line voltage sensors, AC phase crrent sensors, rectifying sbsystem consists of one switch and six diodes of each phase which connecting AC side and DC side, DC side capacitors, DC voltage sensors and resistive load can be seen in conseqence in Fig. 4.2 (a). The rectifying circit of each phase leg is exactly the same for all -73-

92 Chapter 4. Simlation of the Vienna Rectifier three phases, and it can be expressed by a sbsystem with two inpt terminals and three otpt terminals, as shown in Fig. 4.2 (b). The first inpt of the rectifying sbsystem is the AC rectifier inpt voltage of the phase leg, and the second inpt is the plse to drive the switch in the phase leg and control rectifying performance. The three otpt terminals represent the positive point, netral point, and negative point of the DC voltage. (a) (b) -74-

93 Chapter 4. Simlation of the Vienna Rectifier Fig. 4.2 (a) Simlation model of the power circit of the Vienna rectifier system; (b) Detailed simlation model of the rectifying sbsystem in the power circit Parameter Design of the Power Circit The parameters of components in the power circit are specified in Table 4.. The simlation and the hardware prototype of the Vienna rectifier system have the same parameters. AC mains voltage (phase peak voltage) U 20 V ph pk DC command voltage U DC 50 V Resistive load RL 0 Ω Boost indctor L 2 mh DC capacitor C 500 μf AC mains freqency fn 50 Hz Switching freqency fs 00 khz Simlation type Discrete 8 Sample time 0 s Table 4. Parameters of the power circit of the Vienna rectifier system -75-

94 Chapter 4. Simlation of the Vienna Rectifier 4..2 Control Circit Simlation Model The control circit model of the Vienna rectifier system mainly incldes an inner-loop crrent controller, an oter-loop DC voltage controller, and an independent voltage balancing controller. As mentioned previosly, the control circit processes the sampled signals provided by the power circit, like comparing sampled signals with commands, generating control signals based on errors via controllers, etc. In conseqence, proper commands and modlation index are generated by the control circit of the Vienna rectifier. The general strctre of the control circit model of the Vienna rectifier is shown in Fig There are three crrent controllers (one for each phase), one DC voltage controller, and one voltage balancing controller. The inpts of the DC voltage controller are the sampled DC otpt voltage and the DC voltage command (reference), and the otpt of the DC voltage controller is the magnitde reference of the AC crrents. The inpts of the voltage balancing controller are the DC voltages on two DC capacitors, and the otpt of the voltage balancing controller is a variable DC vale added to the modlation index to redce the nbalanced voltage of the rectifier system to zero. The inpts of the crrent controller are the AC crrent reference generated by the DC voltage controller, the modlation index adjstment generated by the voltage balancing controller, the sampled AC mains voltage and crrent of the phase. The otpt of the crrent controller is the modlation index, which can be sed to provide corresponding plses to control switches in the modlator. -76-

95 Chapter 4. Simlation of the Vienna Rectifier * U DC U DCsample U DC sample Power Circit U DC 2 sample i sample i isample DC Voltage Controller Voltage Balancing Controller I ref m i Crrent Controller * m i Fig. 4.3 Strctre of the simlation model of the control circit Note that only line-to-line AC mains voltages can be measred in three-phase systems. Therefore, the sensed line-to-line AC mains voltages need line-to-phase transforming into phase AC mains voltages. In this work, two forms of the crrent control are simlated. One is the crrent controller based on the instantaneos crrent control theory. Another is the crrent controller based on the direct power control theory Instantaneos Crrent Controller The fnction of the crrent controller based on the instantaneos crrent control theory (introdced in section 3.3) is carried ot by programming in MATLAB / Simlink. The state block diagram of the crrent controller sbsystem is given in Fig. 4.4 (a), while the MATLAB fnction of the crrent controller is given in Fig. 4.4 (b). -77-

96 Chapter 4. Simlation of the Vienna Rectifier (a) (b) Fig. 4.4 (a) State block diagram of the instantaneos crrent controller; (b) Fnction of the instantaneos crrent controller DPC Based Crrent Controller The crrent controller based on the direct power control theory (introdced in section 3.6) is carried ot by bilding state block diagram models and programming in MATLAB / Simlink. The state block diagram of the crrent controller sbsystem is given in Fig

97 Chapter 4. Simlation of the Vienna Rectifier Fig. 4.5 State block diagram of the DPC based crrent controller As shown in Fig. 4.5, the DPC based crrent controller consists of three main fnctional sections, shown from left to right respectively: the power calclation section, the DPC PI control with feed-forward section, and the modlation index generating section. In the power calclation section, instantaneos active power P and reactive power Q of the rectifier are calclated based on sampled AC mains voltages and crrents. In the DPC PI control with feedforward section, DPC with feed-forward decople is condcted. Based on the comparison between active power command P _ set (determined by the AC crrent reference gained from the DC voltage controller) and calclated active power P as well as reactive power command Q_ set 0 and calclated reactive power Q, the desired d-axis and q-axis rectifier inpt voltages ( d and q ) can be obtained via the DPC PI control with feed-forward section ((3.39)-(3.4)). The modlation index generating section generates reqired modlation index with the inpts of the d-axis rectifier inpt voltage d, the q-axis rectifier inpt voltage q, the modlation index adjstment provided by the voltage balancing controller. The fnction of the modlation index generating section is shown in Fig. 4.6, which has similar strctre with the instantaneos crrent controller. -79-

98 Chapter 4. Simlation of the Vienna Rectifier Fig. 4.6 Fnction of the modlation index generating section in the DPC based crrent controller Voltage Controller The DC voltage controller is implemented by a PI controller based on theoretical analysis given in section 3.4. Since the resistive load in the rectifier system is fixed, the DC crrent is proportional to the DC voltage. As proved in section 3.4, the AC crrents have proportional relationship with the DC crrent, which reslts in the fact that AC crrents are proportional to DC voltage. To smmarize, the DC voltage controller provides the AC crrent reference based on DC voltage command. In Fig. 4.7, the state block diagram of the DC voltage controller is shown. The DC voltage controller is simple that the error between the DC voltage command and the sampled DC voltage determines the AC crrent reference to control the DC voltage to follow its command. The PI controller is expressed in (3.25). -80-

99 Chapter 4. Simlation of the Vienna Rectifier Fig. 4.7 State block diagram of the DC voltage controller Voltage Balancing Controller The voltage balancing controller is carried ot based on a PI controller, which can be written as (3.3). As indicated in section 3.5, the voltage balancing controller aims to redce the nbalanced otpt voltages on the two capacitors by adding a variable DC component to the modlation index to change the charging/discharging time of the two capacitors. The state block diagram of the voltage balancing controller can be seen in Fig Fig. 4.8 State block diagram of the voltage balancing controller -8-

100 Chapter 4. Simlation of the Vienna Rectifier 4..3 Modlator Simlation Model In this work, the modlator of the Vienna rectifier system is implemented with the carrier based PWM modlation method demonstrated in section Fig. 2.8 shows the modlation strctre and concepts. The state block diagram of the modlator simlation model is shown in Fig. 4.9 (a). The state block diagram of the PWM generator sbsystem in the modlator model is shown in Fig. 4.9 (b). (a) (b) Fig. 4.9 (a) State block diagram of the modlator based on the carrier based PWM modlation; (b) State block diagram of the PWM generator in the modlator It can be conclded from Fig. 4.9 (a) and (b) that when the modlation index is injected into the modlator, its polarity will be determined and corresponding effective modlation fnction will be obtained. And then, proper plse signal, which will be sed to drive the switch in each phase, can be generated accordingly. To smmarize, the simlation model of the modlator is in consistency with the theoretical analysis of the modlation discssed in section

101 Chapter 4. Simlation of the Vienna Rectifier 4..4 Smmary of the Vienna Rectifier Simlation Model Based on models of several sbsystems, the complete simlation model of the Vienna rectifier system can be established by combining and connecting the sbsystems correctly. Fig. 4.0 (a) shows the model of the Vienna rectifier system tilizing the instantaneos crrent control method, while Fig. 4.0 (b) illstrates the model of the Vienna rectifier systm employing the direct power control method. In conclsion, the simlation models of the Vienna rectifier system bilt in MATLAB / Simlink correlate to the strctre given in Fig. 4. very well. Accrate simlation reslts can be expected from the simlation models. -83-

102 Chapter 4. Simlation of the Vienna Rectifier Fig. 4.0 (a) State block diagram of the Vienna rectifier system simlation model with the instantaneos crrent control strategy; (b) State block diagram of the Vienna rectifier system simlation model with the direct power control strategy 4.2 Simlation Reslts of the Vienna Rectifier Based on the established simlation model, the Vienna rectifier system is simlated in MATLAB / Simlink nder different circmstances. Analysis and discssion are also given based on the simlation reslts Simlation Reslts of the Vienna Rectifier Based on Instantaneos Crrent Control (ICC) For the Vienna rectifier system tilizing the instantaneos crrent control (ICC), varios simlations are carried ot in different operating conditions. To start with, the simlation is -84-

103 Chapter 4. Simlation of the Vienna Rectifier condcted in rated operating condition, which means the parameters of the power circit of the Vienna rectifier are the same as given in Table 4.. To be more specific, the switching freqency of the system is 00 khz, and the resistive load is of Ω. After adjsting gain of the P-type crrent controller ( KPi, 0 ), gains of the PI-type DC voltage controller ( KPv, 3, K ), and gains of the PI-type voltage balancing controller ( K, 2, K, 0.5), the Iv, 0.2 simlation reslts of the ICC based Vienna rectifier system are acqired. Ps Is Fig. 4. Simlation reslts of the ICC based Vienna rectifier system in the rated operating condition with fs 00 khz, and RL 0 Ω Fig. 4. shows the for voltage/crrent waveforms of the ICC based Vienna rectifier in steady state in the rated operating condition. The first waveform is the DC otpt voltage U DC, the second waveform is the line-to-line rectifier inpt voltage rab, the third waveform is the phase voltage of phase a a, and the forth waveform is the phase crrent of phase a i a. All the simlation reslts in this work will be presented in the same formla with all for waveforms. -85-

104 Chapter 4. Simlation of the Vienna Rectifier It can be conclded from Fig. 4. that the crrent in phase a i a is in phase with the voltage of phase a a, which means the rectifier system has high power factor. Moreover, the crrent qality is high since it incldes little harmonics and the shape of the crrent is nearly sinsoidal. FFT analysis is done on the crrent waveform and the THD of the crrent is 6.92%. More details of the FFT analysis reslts can be seen in Fig The DC otpt voltage is reglated to the command vale with small low-freqency ripples (at freqency of 6 f N ) de to topology inherent featres. In conclsion, the simlation model for the ICC based Vienna rectifier system is bilt sccessflly, and the simlation reslts of the Vienna rectifier are in consistency with previos theoretical analysis. Fig. 4.2 FFT analysis reslts of the ICC based Vienna rectifier in the rated operating condition with fs 00 khz, and RL 0 Ω Frthermore, in order to stdy the inflence of the switching freqency -86- f s on the performance of the Vienna rectifier system, more simlations are carried ot nder varios

105 Chapter 4. Simlation of the Vienna Rectifier circmstances with different switching freqencies. Compared with the rated switching freqency f _ 00 khz, simlations are also done with low freqency at f _ 0 khz s rated and high freqency f _ 500 khz. The simlation reslts nder low/high freqency s high operating conditions are shown in Fig. 4.3 and Fig. 4.4 respectively. Note that the power circit parameters and controller parameters in the low freqency condition are set exactly the same as the rated freqency condition. However, in the high freqency condition, the crrent controller gain K Pi, is changed to 20, while the rest parameters are kept nchanged. s low Fig. 4.3 Simlation reslts of the ICC based Vienna rectifier system in the low freqency operating condition with f _ 0 khz s low -87-

106 Chapter 4. Simlation of the Vienna Rectifier Fig. 4.4 Simlation reslts of the ICC based Vienna rectifier system in the high freqency operating condition with f _ 500 khz s high It can be seen clearly by comparing Fig. 4., Fig. 4.3 and Fig. 4.4 that increasing the switching freqency improves the performance of the Vienna rectifier system. Firstly, the amplitde of the crrent ripples at switching freqency is closely related to the switching freqency of the system. When the switching freqency is low, the crrent ripples are high and the crrent qality is low. In contrary, when the switching freqency is high, the crrent ripples are low and the crrent qality is high. This conclsion is also proved by the FFT analysis reslts. For the crrent in low freqency operating condition, the THD is 3.44%. However, for the crrent in high freqency operating condition, the THD is only 5.74%. Secondly, with higher switching freqency, the amplitde of the DC voltage ripples is redced. As a reslt, by increasing the switching freqency, the performance of the Vienna rectifier system is better in terms of the AC crrent and DC voltage waveforms. Since high freqency operation is preferred in the Vienna rectifier system, GaN devices are the top choice becase they have tremendos advantages over traditional Si devices in high freqency applications. -88-

107 Chapter 4. Simlation of the Vienna Rectifier Simlation Reslts of the Vienna Rectifier Based on Direct Power Control (DPC) For the Vienna rectifier system sing the direct power control (DPC), simlation in the rated operating condition is condcted. For DPC, it shares the same DC voltage controller and voltage balancing controller with ICC. Corresponding controlling gains are set as KPv, 3, KIv, 0.2, KPs, 2, KIs, 0.5. However, for DPC, there is no direct crrent control loop K Pi,, instead the power control loop dominated by K PP,, K IP,, K PQ,, and K IQ, replaces the crrent control loop. Here, the parameters of the active/reactive controller are selected as KPP, 25, K, K, 25, K, 45. IP, 45 PQ IQ Fig. 4.5 Simlation reslts of the DPC based Vienna rectifier system in the rated operating condition with fs 00 khz, and RL 0 Ω Fig. 4.5 shows the simlation reslts of the DPC based Vienna rectifier system in the rated operating condition. Besides, FFT analysis on the crrent is shown in Fig The THD in the crrent is only 4.29%. -89-

108 Chapter 4. Simlation of the Vienna Rectifier Fig. 4.6 FFT analysis reslts of the DPC based Vienna rectifier in the rated operating condition with fs 00 khz, and RL 0 Ω Since the reslts illstrated in Fig. 4.2 and Fig. 4.5 are carried ot nder the same operating condition, the difference in the steady state performance of the ICC based and DPC based Vienna rectifier system can be indicated. In general, the DPC based Vienna rectifier system has advantages over the ICC based Vienna rectifier system in the steady state. The DC otpt voltage of the DPC system is reglated better with smaller ripples. The AC crrent qality of the DPC system is higher with relatively low THD vale Simlation Reslts of the Vienna Rectifier with Load Change In addition to steady state performance, dynamic performance of the ICC and DPC based Vienna rectifier systems also needs stdy. -90-

109 Chapter 4. Simlation of the Vienna Rectifier Fig. 4.7 Simlation reslts of the ICC based Vienna rectifier system when the resistive load R L changes from 0 Ω to 80 Ω at 0. s Fig. 4.8 Simlation reslts of the DPC based Vienna rectifier system when the resistive load R L changes from 0 Ω to 80 Ω at 0. s Specific simlations for ICC and DPC based Vienna rectifier systems are designed nder a load change condition, where the resistive load changes from 0 Ω to 8 Ω. The simlation -9-

110 Chapter 4. Simlation of the Vienna Rectifier reslts of the ICC based and DPC based Vienna rectifier systems are shown in Fig. 4.7 and Fig. 4.8 respectively. Based on comparison of Fig. 4.7 and Fig. 4.8, it can be conclded that the DPC based Vienna rectifier system still has advantages over the ICC based Vienna rectifier system in dynamic performance. When the load changes, the time of the transient stats in the DPC based system is abot s, which is mch shorter than the s of the transient stats in the ICC based system. Therefore, the DPC based Vienna rectifier system has better capability to rapidly respond to distrbance and command changes and it has better dynamic performance Smmary of Simlation Reslts of Vienna Rectifier The simlation reslts of the Vienna rectifier verifies that the theoretical analysis of the Vienna rectifier system, inclding the power circit, modlation method, control strategies, is correct, since all simlation reslts correlate with theoretical analysis reslts well. The ICC based Vienna rectifier system can sccessflly achieve desired performance with reglated DC otpt voltage and sinsoidal AC crrents. The power factor of the system is good and the THD of the crrent is acceptable. The switching freqency is a key element inflencing the performance of the Vienna rectifier system. The higher the switching freqency, the better DC otpt voltage reglation and smaller AC crrent distortion. The DPC based Vienna rectifier system has otstanding performance in both the steady state operation and transient state operation. Direct power control strategy has promising potential in applications in the Vienna rectifier system. -92-

111 Chapter 5 Experiments of the Vienna Rectifier Practical experiments are always essential to prove the theoretical analysis reslts and simlation reslts of a power electronics system. In this chapter, the experiments are carried ot on two prototypes of the GaN FET and Si MOSFET based Vienna rectifier systems with each type of the power devices on a similar scale. Therefore, comparative analysis of the Vienna rectifier based on different devices can be condcted. The selected GaN FET and Si MOSFET devices have the same rating voltage (200 V) and similar rating crrent (arond 20 A), and more parameters are compared in Table 5. Manfactre Si International Rectifier GaN EPC Manfactre # IRF640N EPC200C VDS [V] ID [A] 8 22 Rds(on)@25 C [Ω] Total gate Charge [nc] Inpt cap. [pf] Otpt cap. [pf] Reverse Transfer cap. [pf] 53.8 Table 5. Parameters of the selected Si MOSFET and GaN FET devices In this chapter, the hardware design and software design of prototypes of the GaN FET and Si MOSFET based Vienna rectifier systems are demonstrated. Experiment reslts of the two prototypes are shown and analyzed as well.

112 Chapter 5. Experiments of the Vienna Rectifier 5. The Hardware Design of the Vienna Rectifier For the two prototypes of the Vienna rectifier systems based on GaN FET and Si MOSFET respectively, the most parts of the systems, other than the switching devices and corresponding driver circit, are the same. In other words, the two prototypes share the power circit, sensing circit, DSP control circit and power spplier, bt the switching devices and the driver circit are different. To start with, the general strctre of the hardware design of the Vienna rectifier system can be illstrated by Fig. 5.. In the figre, the yellow sections are high power level sections and doble lines represent high-power electrical connections and energy flows. On the other hand, the other colorfl sections are low power level sections and single lines indicate low-power digital connections and signal flows. AC Mains 3 AC voltages 3 AC crrents Power Circit Driving plses Load 2 DC voltages (on capacitors) Drive Circit Sensing Circit DSP Control Cirtit Plse signals Power Spplier Fig. 5. Strctre of the hardware design of the Vienna rectifier system Based on the given strctre, the hardware design of the two prototypes of the Vienna rectifier can divided into several fnctional sections, sch as the power circit, the driver -94-

113 Chapter 5. Experiments of the Vienna Rectifier circit, the sensing circit, the DSP control circit, and the power spplier, and will be introdced in detail respectively. 5.. The Power Circit Hardware Design In the power circit hardware design, a key step is to select proper passive components, diodes and switching devices. Based on the power circit specification given in Table 4., components selection can be determined. The AC boost indctor has the vale of 2 mh with the satration crrent of abot 0 A (DC) to garantee that the indctor works in continos condcting mode and is not satrated dring operation. The DC capacitor has the vale of abot 500 μf, which consists one 470 μf electrolytic capacitor and several 0 μf ceramic capacitors in parallel. The rating voltage of these capacitors is 00 V (DC). For diodes, SS 50 Schottky diode from HY Electronic Corp. are selected, with 00 V reversed voltage and 5 A forward crrent. For switching devices, GaN FET EPC200C is employed in the GaN FET based system while Si MOSFET IRF640N is tilized in the Si MOSFET based system. The resistive load is a power load of Ω, 500 W. Considering the DC capacitors need time to charge and reach designed voltage level, a three-phase transformer is sed as the AC mains to provide adjstable voltages The Driver circit Hardware Design The Driver Circit Hardware Design of the Si MOSFET The otpt plse signals of the DSP controller have the voltage level of 3 V with low energy, however, the reqired gate voltage signals to drive the Si MOSFET have the voltage level of 5 V with high energy. As a reslt, a driver circit, which has the fnction as a power amplifier, is needed to amplify low-power digital plse signals to high-power electrical plse signals to drive the Si MOSFET. For the conventional Si -95-

114 Chapter 5. Experiments of the Vienna Rectifier MOSFET, the semicondctor switch and its driver circit are not always necessarily integrated together. Ths, the Si MOSFETs are connected in the power circit and corresponding driver circits provide gate signals to the switches. The schematic of the driver circit PCB board for Si MOSFET is shown in Fig There are two main chips in each driver circit. One is the MOSFET driver chip IR 20s, which isolates the low-power digital side and high-power electrical side and transfers the 0~3 V low-power plse signals on the first side to the 0~5 V high-power plse signals on the second side. Another is the isolated power spply chip MORNSUN QA04, which offers the isolated +5 V voltage to the second side of the driver chip IR 20s. Fig. 5.2 Schematic of the driver circit in the Si MOSFET based Vienna rectifier system The Driver circit Hardware Design of the GaN FET For the GaN FET, the appropriate gate voltage to drive the device is arond 5 V. Therefore, a driver circit is still essential to transfer low-power plse signals generated -96-

115 Chapter 5. Experiments of the Vienna Rectifier by the DSP controller to high-power plse signals, which can drive the GaN FET. Take the extremely small size of the GaN device and the sensitivity of the distortion of driving signals de to parasitic indctors and parasitic capacitors in practical circit into accont, the driver circit of GaN FET and the GaN FET device itself shold be integrated and careflly designed. Hence, the GaN FET and its driver circit are designed on one PCB board which is placed in the power circit directly. Fig. 5.3 shows the schematic of the driver circit PCB board for GaN FET. Fig. 5.3 Schematic of the driver circit with GaN FET in the GaN FET based Vienna rectifier system The driver circit for the GaN FET consists of one isolated power spply, one PWM signal isolator, one drive chip and one GaN FET device. This circit is ready to be sed in one phase of the power circit as the switch. The isolated power spply chip is B5050s- 2W from MORNSUN, which spplies isolated +5 V. The PWM signal isolator chip is Si840. The main fnction of the PWM signal isolator is isolating the GaN FET drive chip from the DSP controller, which redces distrbances and noises for the GaN FET drive. The GaN FET drive chip is LM54 from TI with a single driver. Since there is only one semicondctor switch in each phase in the Vienna rectifier topology, the single driver chip is sitable. A photo of a real driver circit for the GaN FET is shown in Fig The ble device on the pper left corner of the PCB board is a GaN FET. The size of the GaN FET device is very small, and considering the crrent it can condct and the -97-

116 Chapter 5. Experiments of the Vienna Rectifier voltage it can block, the power density of the GaN FET device is really high. At the same time, the total volme of the complete driver circit of the GaN FET is relatively small compared with that of the Si MOSFET. Fig. 5.4 Photo of a real driver circit for the GaN FET 5..3 The Sensing Circit Hardware Design According to Fig. 5., the Vienna rectifier system reqires eight sensors. Among them, three are AC voltage sensors, two are DC voltage sensors, and three are AC crrent sensors. These sensors measre crcial instantaneos information of the power circit and provide it to the controller. The sensing circit is an important part of the Vienna rectifier system. The accracy of the sensing reslts largely inflences the effectiveness of the control and the performance of the whole system. Detailed design information of the three types of sensing circits are demonstrated as follows. -98-

117 Chapter 5. Experiments of the Vienna Rectifier The AC Voltage Sensing Circit Hardware Design Voltage transdcer LV25-P from LEM is selected to measre the AC line-to-line voltages. This voltage transdcer can measre DC voltage, AC voltage and plse voltage, with a galvanic isolation between the primary circit (high-voltage) and secondary circit (electronic circit). The nominal RMS crrent of the primary side is 0 ma and the nominal RMS crrent of the secondary side is 25 ma. By selecting proper external resistance in the primary circit ( R 4 and R 4 _ 2 as shown in Fig. 5.5), the voltage transdcer can measre voltages with amplitde in the range of 0 V to 500 V, becase in order to measre the voltage, a crrent proportional to the voltage mst pass the primary circit. By setting vale of the external resistors in the secondary circit ( R 20 and R 2 as shown in Fig. 5.5), the ratio between the otpt and the inpt of the voltage transdcer can be determined. As the otpt signal of the sensing circit is the inpt signal of the DSP controller, the voltage range of the signal shold be limited within 0 V to 3 V, otherwise it will damage the DSP controller. Besides, there is no DC offset of the otpt of the voltage transdcer, which means when the inpt voltage is 0 V, the otpt voltage is 0 V. However, since the voltage transdcer is sed to measre AC voltage in this application, the DC offset of the sensor otpt need to be added so that the bipolar voltage can be represented in the range of 0 V to 3 V. In order to add DC offset to the otpt signal of the voltage transdcer, operational amplifiers are needed. The schematic of the AC voltage sensing circit in the Vienna rectifier system is given in Fig An amplifier chip LM642, with two operational amplifiers embedded, is employed to add DC offset to the otpt of the voltage transdcer to make the sensing circit generate accrate and effective signal within the range limitation. -99-

118 Chapter 5. Experiments of the Vienna Rectifier Fig. 5.5 Schematic of the AC voltage sensing circit in the Vienna rectifier system The DC Voltage Sensing Circit Hardware Design In the DC voltage sensing circit, the same voltage transdcer LV25-P, as sed in the AC voltage sensing circit, is tilized. Since in this application, the voltage transdcer only needs to measre nipolar voltage, the DC offset and adder in the AC voltage sensing circit are no longer necessary. By choosing vale of the external resistor in the primary circit ( R and R _ as shown in Fig. 5.6), the measring range of the voltage transdcer is determined. By selecting vale of the external resistor in the secondary circit ( R 5 and R 6 as shown in Fig. 5.6), the voltage ratio between the inpt and otpt of the voltage transdcer is set. As the schematic of the DC sensing circit shown in Fig. 5.6, an amplifier is still employed and playing the role of a voltage follower. It helps redce noise in the signal and provide isolation between the sensing circit and DSP controller. -00-

119 Chapter 5. Experiments of the Vienna Rectifier Fig. 5.6 Schematic of the DC voltage sensing circit in the Vienna rectifier system The AC Crrent Sensing Circit Hardware Design Hall-effect based linear crrent sensor ACS72-05B from Allergo Microsystem Inc. is selected in this work to measre AC crrents. The optimized range of the objective crrent is ±5 A, and the typical otpt sensitivity is 85 mv/a. The original DC offset of the crrent sensor is 2.5V. Therefore, DC offset adjstment is necessary to make the otpt signal of the crrent sensing circit meet the reqirement of the DSP controller. Fig. 5.7 shows the schematic of the AC crrent sensing circit in the Vienna rectifier. Fig. 5.7 Schematic of the AC crrent sensing circit in the Vienna rectifier system -0-

120 Chapter 5. Experiments of the Vienna Rectifier 5..4 The DSP Control Circit Hardware Design DSP TMS320F28335 from TI is selected as the microcontroller in the Vienna rectifier system. A DSP development board embedding a DSP TMS320F28335 chip and fndamental fnction sections is employed in the Vienna rectifier system. The DSP development board makes the applications of the DSP in the system easy. A photo of the DSP development board is shown in Fig Fig. 5.8 Photo of the DSP development board 5..5 The Power Spplier Hardware Design The power spplying system is a significant part of every power electronics system that it provides the electronic power and enables devices to work. In the Vienna rectifier -02-

121 Chapter 5. Experiments of the Vienna Rectifier system, several power sppliers are spporting the sensing circit, the DSP controlling circit, and the driver circit. Table 5.2 shows the power spply chips applied in the Vienna rectifier system and their characteristics. System Chip Inpt Otpt Isolation Object DSP control circit WRB0505-6W 4.5 ~ 9 VDC 5VDC/200mA Yes Si Voltage sensing circit MOSFET WRB0505-6W 4.5 ~ 9 VDC 5VDC/200mA Yes Crrent sensing circit Based WRA052-6W 4.5 ~ 9 VDC ±2VDC/250mA Yes Voltage sensing circit System External DC Power Spply 2VDC 2VDC/2A Yes Driver circit WRB0505-6W 4.5 ~ 9 VDC 5VDC/200mA Yes DSP control circit Driver circit GaN FET WRA052-6W 4.5 ~ 9 VDC ±2VDC/250m Yes Voltage sensing circit Based System L7805CV 7 ~ 35 VDC + 5 VDC/.5A Yes Voltage sensing circit Crrent sensing circit D/A circit L7905CV -7 ~ -35 VDC 5 VDC/.5A Yes D/A circit Table 5.2 Power spplying chips and their characteristics 5..6 The Prototypes of the Vienna Rectifier Systems With design introdction given in previos sections, the two prototypes of the Vienna rectifier system based on the GaN FET devices and Si MOSFET devices can be bilt. Fig. 5.9 and Fig. 5.0 show the photos of the final prototypes of the Si MOSFET based and GaN FET based Vienna rectifier systems respectively. -03-

122 Chapter 5. Experiments of the Vienna Rectifier Fig. 5.9 Photo of the prototype of the Si MOSFET based Vienna rectifier system -04-

123 Chapter 5. Experiments of the Vienna Rectifier Fig. 5.0 Photo of the prototype of the GaN FET based Vienna rectifier system -05-

124 Chapter 5. Experiments of the Vienna Rectifier 5.2 The Software Design of the Vienna Rectifier The control system of the prototype of the Vienna rectifier is based on the control circit, which contains the DSP TMS320F28335 as the key element. The DPS program is written in C langage and the compiling and debgging of the program are done in the CCS (Code Composer Stdio) environment. The control program of the Vienna rectifier has a main program and an interrpt program. In the main program, the DSP system is initialized. System initializations inclde parameters initializations and modle initializations. For the Vienna rectifier application, GPIO modle, epwm modle, ADC modle, etc. are tilized and need initializations in the main program. After the initializations of the DSP system, the main program keeps rnning. The interrpt program rns periodically to implement the control fnctions of the DSP control system. Signal processing, digit calclation, and plse generation are carried ot in the interrpt program. The flow chart of the thoght of the software design in the form of the DSP control program is shown in Fig. 5.. These fnctions are written and achieved in the interrpt program of the DSP. Therefore, desired control of the Vienna rectifier system can be carried ot. In general, the control method shown in Fig. 5. correlates with the control strategy of the Vienna rectifier introdced in chapter 3 well. They have the same strctre, which confirms the correctness of the DSP controller. In the following, detailed descriptions of each fnction given in Fig. 5. will be indicated in seqence. Since the inpts of the DSP control circit are eight analog signals generated by the sensing circit and representing sampled AC voltages, AC crrents and DC voltages, the first thing the DSP controller needs to do is to transfer the analog signals into digital signals. This analog-to-digital transformation is carried ot by the Analog Digital Converter (ADC) modle in the DSP. Calibration process is essential to transfer each analog signal to a corresponding digital signal. The relation between the analog signal -06-

125 Chapter 5. Experiments of the Vienna Rectifier and the corresponding digital signal is linear mapping. In conclsion, the "A/D Calibration" fnction transfers eight analog signals carrying sample information into eight digital signals withot losing any information. Sensing Circit DSP Controller 8 analog signals A/D Calibration PPL (Phase-Locked Loop) DC Otpt Voltage Control Instantaneos Crrent Control OR Direct Power Control Voltage Balancing Control Modlation Index Calclation Modlation PWM Plses Otpt Driver Circit 3 PWM plse signals Fig. 5. DSP Control Program Flow Chart -07-

126 Chapter 5. Experiments of the Vienna Rectifier Based on the digital signals provided by the "A/D Calibration", "Phase-Locked Loop" (PLL) can acqired the actal phase and freqency of the AC voltages. The phase and freqency information will be sed in the "Instantaneos Crrent Control" fnction. As it is named, "DC Otpt Voltage Control" fnction plays the role of the DC otpt voltage controller introdced in section 3.4. Given the DC otpt voltage command, corresponding AC crrent command (the reference crrent magnitde) is calclated in this fnction. Either the instantaneos crrent control (ICC) or the direct power control (DPC) can be sed as the inner control loop in the whole control system. For the ICC, a modlation index command is generated by comparing the sampled crrent with the crrent reference and process the difference. For the DPC, the active power command is calclated based on the crrent reference and the reactive power command is set to zero. Based on feedback control on the active power and the reactive power, a modlation index command can be calclated to make the active power and reactive power to follow their commands. The "Voltage Balancing Control" deals with the nbalance voltages on the DC capacitors by generating a compensating component in the form of the modlation index. Since the modlation index inflences the condcting dty cycle of the switch in each phase, a change in the modlation index cases changes in charging/discharging time on the DC capacitors and leads to balanced voltages. By adding the modlation index components generated by the ICC/DPC control fnction and the voltage balancing control fnction, the final modlation index commands can be calclated. The modlation fnction generates proper PWM plse signals based on corresponding modlation index commands. And the PWM plse otpt fnction exports the PWM plses as the otpt of the DSP control circit. -08-

127 Chapter 5. Experiments of the Vienna Rectifier 5.3 Experiment Reslts of the Vienna Rectifier Experiments are carried ot on both prototypes of the GaN FET based and Si MOSFET based Vienna rectifier systems. And for each system, the instantaneos crrent control strategy (ICC) and the direct power control strategy (DPC) are both applied as well. As a reslt, for types of experiments are condcted based on the classification of switching device type as well as control strategy. The experimental reslts of these for conditions are shown as follows Si MOSFET Based Vienna Rectifier Prototype with ICC control Fig. 5.2 shows the experiment reslts of the Si MOSFET based Vienna rectifier prototype with ICC control operating in steady state nder rated conditions ( f s =00 khz). For waveforms in the figre are respectively the DC otpt voltage U DC, the line-to-line rectifier inpt voltage rab, the phase voltage of phase a a, and the phase crrent of phase a i a. All experiment reslts will be given in the same form in this work. The experiment reslts shown in Fig. 5.2 are in good correlation with the simlation reslts shown in Fig The DC otpt voltage is nicely reglated at the command vale (50V) and the AC mains crrent (phase a) has nearly sinsoidal shape and is in phase with AC mains voltage in the same phase. FFT analysis of the AC mains crrent is done and the reslts is shown in Fig The THD is 8.29% in this case. The theoretical analysis reslts, the simlation reslts and the experiment reslts match each other very well. Therefore, the Vienna rectifier system employing Si MOSFET is sccessflly designed and instantaneos crrent control strategy is fnctional in the Vienna rectifier system. -09-

128 Chapter 5. Experiments of the Vienna Rectifier Fig. 5.2 Experiment reslts of the Si MOSFET based Vienna rectifier system with ICC control Fig. 5.3 FFT analysis on AC mains crrent in the Si MOSFET based Vienna rectifier system with ICC control -0-

129 Chapter 5. Experiments of the Vienna Rectifier Si MOSFET Based Vienna Rectifier Prototype with DPC control Fig. 5.4 Experiment reslts of the Si MOSFET based Vienna rectifier system with DPC control Fig. 5.4 shows the experiment reslts of the Si MOSFET based Vienna rectifier prototype with DPC control operating in steady state nder rated conditions ( f s =00 khz). The experiment reslts shown in Fig. 5.4 correlates well with the simlation reslts shown in Fig The DC otpt voltage is nicely reglated at the command vale (50V) and the AC mains crrent (phase a) has nearly sinsoidal shape and is in phase with AC mains voltage in the same phase. FFT analysis of the AC mains crrent is done and the reslts is shown in Fig The THD is 7.76% in this case, which is slightly smaller than the THD for the same prototype bt sing ICC control instead. The --

130 Chapter 5. Experiments of the Vienna Rectifier THD comparison proves that better THD is one of the advantages of the DPC over the ICC. The direct power control strategy performs well in the Vienna rectifier system. Fig. 5.5 FFT analysis on AC mains crrent in the Si MOSFET based Vienna rectifier system with DPC control GaN FET Based Vienna Rectifier Prototype with ICC Control Fig. 5.6 shows the experiment reslts of the GaN FET based Vienna rectifier prototype with ICC control operating in steady state nder conditions with f s =50 khz. The original designed switching freqency for the GaN FET based Vienna rectifier system is 00 khz, however, de to limited time and technical difficlty on the GaN FET drivers, the switching freqency of the system is only pshed to 50 khz. In near ftre, obstacles will be overcome and the switching freqency will be pshed to 00 khz soon. The experiment reslts shown in Fig. 5.6 correlates are in great consistency with the simlation reslts shown in Fig The DC otpt voltage is nicely reglated at the command vale (50V) and the AC mains crrent (phase a) has nearly sinsoidal shape and is in phase with AC mains voltage in the same phase. FFT analysis of the AC mains -2-

131 Chapter 5. Experiments of the Vienna Rectifier crrent is done and the reslts is shown in Fig The THD is.3% in this case. The THD vale is relatively high compared with that of the Si MOSFET based system. Possible reasons may inclde the difference in system switching freqency and the immatrity of the newly designed GaN FET drivers. Besides, controller parameters, like PI gains can be frther optimized to improve the performance of the system. In general, the theoretical analysis reslts, the simlation reslts and the experiment reslts match each other very well. Therefore, the Vienna rectifier system employing GaN FET is sccessflly designed and implemented. This is the first research that sccessflly applies the GaN devices in a Vienna rectifier system. It proves that GaN power devices have the ability and potential to be applied in rectifying systems and related applications. Fig. 5.6 Experiment reslts of the GaN FET based Vienna rectifier system with ICC control -3-

132 Chapter 5. Experiments of the Vienna Rectifier Fig. 5.7 FFT analysis on AC mains crrent in the GaN FET based Vienna rectifier system with ICC control GaN FET Based Vienna Rectifier Prototype with DPC Control The experiment reslts of the GaN FET based Vienna rectifier prototype with DPC control operating in steady state nder conditions with f s =50 khz is shown in Fig The experiment reslts match the simlation reslts shown in Fig The DC otpt voltage is nicely reglated at the command vale (50V) and the AC mains crrent (phase a) has nearly sinsoidal shape and is in phase with AC mains voltage in the same phase. FFT analysis of the AC mains crrent is done and the reslts is shown in Fig The THD is 4.8% in this case. De to limited time, the DPC control in the GaN FET based Vienna rectifier prototype is not flly developed and optimized. Bt the experiment reslts still reveal that the DPC control achieves fndamental control fnctions in the Vienna rectifier and has promising ftre. -4-

133 Chapter 5. Experiments of the Vienna Rectifier Fig. 5.8 Experiment reslts of the GaN FET based Vienna rectifier system with DPC control Fig. 5.9 FFT analysis on AC mains crrent in the GaN FET based Vienna rectifier system with DPC control -5-