SPLIT WINDING SWITCHED RELUCTANCE MACHINE DRIVES FOR WIDE SPEED RANGE OPERATIONS

Transcription

1 SPLIT WINDING SWITCHED RELUCTANCE MACHINE DRIVES FOR WIDE SPEED RANGE OPERATIONS A Thesis Presented to The Graduate Faculty of the University of Akron In Partial Fulfillment of the Requirement for the Degree Master of Science Oguzhan Kilic August 2018

2 SPLIT WINDING SWITCHED RELUCTANCE MACHINE DRIVES FOR WIDE SPEED RANGE OPERATIONS Oguzhan Kilic Thesis Approved: Accepted: Adviser Dr. Yilmaz Sozer Interim Department Chair Dr. Joan Carletta Committee Member Dr. Malik Elbuluk Dean of the College Dr. Donald P. Visco Committee Member Dr. Jose A. De Abreu-Garcia Dean of the Graduate School Dr. Chand Midha Date ii

3 ABSTRACT In this thesis, a switched reluctance machine (SRM) with a new winding configuration and a novel power converter design is presented. The new winding topology along with the power converter are developed to improve the machine efficiency by exciting a portion of the phase windings at high speeds. The new topology helps to overcome the effects of the high back EMF voltage (Electromotive Force) and improve the torque characteristics of SRMs. The phase windings are split into two interconnected windings, each with a different number of turns. The two windings act as a transformer due to the magnetic coupling between them. To drive the split winding machine with a different number of turns at different speed levels, the H-bridge converter was improved by adding one extra switch on each leg. The extra switches are only used at high speed operation to reduce the number of turns. At high speed operations, the proposed set-up using only one portion of the winding, pushes higher phase currents. The new winding configuration and extra switch converter topology drive the proposed machine efficiently over a variable speed range. In addition, the proposed work improves the torque characteristic of the machine at both medium and high-speed levels by reducing the number of turns which, as mentioned above, enables the application of higher currents to the phase. The proposed design is investigated using Finite Element Analysis (FEA) circuit simulations on a case study SRM. The new power converter and its associated control algorithm have been tested experimentally. The experimental and simulation results for both full and split winding, SRMs are compared. iii

4 DEDICATION I dedicate my thesis work to my beloved parents and my wife to be. iv

5 ACKNOWLEDGMENTS First and foremost, I would like to express my sincere gratitude to my advisor Dr. Yilmaz Sozer for his tremendous academic support, his patience and for giving me so many wonderful opportunities. I wish to convey my gratitude and appreciation to my committee members Dr. Jose A. De Abreu-Garcia and Dr. Malik E. Elbuluk for their support and follow up. Special thanks go to the my colleagues at the Alternative Energy Laboratory for their generous help and support through the friendly atmosphere. I am also thankful to Turkish Petroleum Company for their financial support and to the Turkish Educational Attaché for helping in my social life during my master education abroad. v

6 TABLE OF CONTENTS Page LIST OF TABLES.vii LIST OF FIGURES.ix CHAPTER Ⅰ. INTRODUCTION Switched Reluctance Machines Switched Reluctance Machine Configurations Advantages and Disadvantages of SRMs Motivation for Research Thesis Outline... 4 Ⅱ. LITERATURE REVIEW Introduction Principle of Operation Design of SRMs Converter Topologies SRMs Control Average Torque Improvement and Torque Ripple Minimization Design Based Torque Ripple Minimization Control Based Torque Ripple Minimization Ⅲ. SPLIT WINDING STRUCTURE Introduction Proposed Work vi

7 3.3 Phase Winding Scheme Proposed Converter Topology Magnetization Mode Freewheeling Mode Demagnetization Mode Proposed Control Algorithm Ⅳ. SYSTEM MODELING AND SIMULATION RESULTS Introduction Split Winding Configuration Machine Modelling Finite Element Analysis Model Matlab Simulink Model Power Converter Model Control Algorithm Simulation Results Conclusion Ⅴ. EXPERIMENTAL SET-UP AND TEST RESULTS Introduction The Split Winding Machine Power Converter and Control Circuit Printed Circuit Board (PCB) Design Buffer for Gate Signals Current Sensor and Conditioning Circuit Gate Driver vii

8 5.5 Experimental Set-up Control Algorithm Experimental Results Conclusion Ⅵ. CONCLUSION AND FUTURE WORK Conclusion Future Work viii

9 LIST OF FIGURES FIGURES PAGE 1-1 Structure of 6/4 SRM Phase inductance profile Electromagnetic profile of SRMs Energy conversion of SRMs Torque-speed characteristic of SRMs Typical SRM power converter Magnetization mode Freewheeling mode Demagnetization mode Split-capacitor Converter Miller Converter Energy Efficient C-dump Converter Control block diagram of SRM Basic motoring operation Single pulse operation Torque sharing Control block diagram of SRMs Transformer winding configuration in the stator pole The proposed converter design Low speed magnetization High speed magnetization ix

10 3-5 Low speed free-wheeling High speed free-wheeling Low and high-speed /8 SRM finite element analysis model Torque characteristic of full winding topology Flux characteristic of full winding topology Phase torque characteristic of split winding topology Phase flux characteristic of split winding topology Flux and torque Look-up Tables Phase equation Matlab schematic Conventional power converter Proposed power converter Phase currents at 3000 rpm Phase currents at 4000 rpm Flux linkages at 2000 rpm Flux linkages at 3000 rpm Flux linkages at 4000 rpm Split and full winding torque comparison at 2000 rpm Split and full winding torque comparison at 3000 rpm Split and full winding torque comparison at 4000rpm Split winding torque at 4500rpm Torque comparison split winding at 5000rpm, full winding at 4000rpm At 3000 rpm torque ripple comparison x

11 4-21 At 4000 rpm torque ripple comparison Stator of the experimental SRM Winding configuration of the split winding machine Split winding machine winding configuration for phase B Extra switch power converter Block diagram of the control and power circuit Power and control circuit schematics Buffer circuit schematic Current conditioning circuit schematic Current sensor schematic Gate driver circuit schematic Power and control circuit layout Experimental set-up configuration Incremental encoder disk A, B and Z signal of encoder respectively SRM experimental setup Control algorithm Full and split winding topologies phase currents at 3000 rpm Full and split winding topologies phase currents at 4000 rpm Power comparison Torque comparison Model verification curves xi

12 LIST OF TABLES Table Page 1 Specification of SRM Full winding topology power and torque output Split winding topology power and torque output xii

13 CHAPTER Ⅰ INTRODUCTION 1.1 Switched Reluctance Machines Switched reluctance machines (SRMs) are electric machines that were initially developed in the 1980s [1]. SRMs have been extensively used in many applications such as in the aerospace, automotive and renewable energy industries, [2] [3], due to their outstanding inherent characteristics including simple mechanical structure, reliability, wide speed operation and robustness [4]. 1.2 Switched Reluctance Machine Configurations SRMs are doubly salient electric machines with independent stator windings that produce a torque by variable reluctance. The rotor has no windings or permanent magnets and consists simply of steel laminations. This minimizes losses on the rotor and SRMs can function properly without overheating. Depending on the application, various SRM configurations may be used in terms of the number of stator and rotor poles and the number of phases [5]. The number of phases can be increased by adding legs to the power converter. The different number of stator and rotor poles and its unsymmetrical configuration is the essential characteristic of the machine. Some common configurations include 6/4, 8/6, 10/8 12/8, the number of stator and rotor poles, respectively. The three phase SRM shown in Fig. 1.1 has six stator and four rotor poles, commonly referred to as a three phase 6/4 SRM [6]. 1

14 Figure 1-1 Structure of 6/4 SRM Stator windings can be connected by either a parallel or series structure. If connected by a parallel structure, the stator windings share the bus voltage; otherwise, they share the current. In low voltage applications, a parallel configuration structure is preferred. Torque is produced by the tendency of the rotor to move, in order to minimize the reluctance of the excited winding [6]. Since each phase is separately independent from the other phases, SRMs are highly reliable machines. The SRM control algorithm has a sequential pattern in order to maintain positive torque production from each phase [7]. SRMs are good candidates for variable speed applications because they have higher efficiency, higher starting torque and competent high speed characteristics. However, at high speed operations, due to the short excitation time, slow demagnetization and high back EMF voltage, the phase current is prevented from reaching the desired value [8]. In addition, SRMs also encounter high torque ripple and high acoustic noise problems. 1.3 Advantages and Disadvantages of SRMs The main advantages of SRMs are as follows: 2

15 Low cost due to simple construction Good fault tolerance and robustness due to independent phases Independent current direction can be applied to reduce complexity and losses on the power inverter The main losses occur on the stator due to the lack of rotor windings and permanent magnets; thus, cooling can be easier. High starting torque without excessive current Can operate over a wide speed range High torque ratio High efficiency at high speeds The disadvantages of SRMs are as follows: The independent phases and nonlinearity characteristics require complex control Torque ripple can be high Torque ripple and other radial forces cause high acoustic noise 1.4 Motivation for Research The proposed winding configuration, along with the new converter design, are developed to address the problems associated with high speed operations in SRMs. The typical driver used with SRMs is the asymmetric converter which is composed of two switches and two diodes in each phase [9]. The two switches are used to control the 3

16 excitation current developed in each phase. The diodes provide the discharge path for the current in the freewheeling and the demagnetization modes [10]- [11]. One of the factors that affects the performance of an SRM is the number of turns (Ns) in the stator of the machine, which means that with a high number of turns, high torque can be achieved at low speed operations. Conversely, when the number of turns decreases, a higher torque at high speeds can be achieved [12]. In this thesis, the objective is to use split-winding per phase such that at low speeds, a higher number of turns is used, while at high speeds, a smaller number of turns is used. In order to implement the control strategy, a new asymmetric converter design with an extra switch in each phase and its associated control algorithm are presented. 1.5 Thesis Outline This thesis is divided into six chapters, the first chapter provides a brief overview of SRMs, and explains the motivation and objective of this thesis. The second chapter reviews the structure of the SRMs, their control algorithms and power converter topologies. In the third chapter, the proposed machine design and modeling process are explained. The power converter design and implementation of the control algorithm are described. Chapter 4 presents the experimental setup of SRM. Chapter 5 discusses the SRM and its power converter test results, including a summary of the thesis and discussion of future work. 4

17 CHAPTER Ⅱ LITERATURE REVIEW 2.1 Introduction There are various methods found in the literature that improve the SRM torquespeed characteristics and minimize the torque ripple at medium and high-speed applications. SRMs principle of operation, design and their converter topologies are examined in next three sections, proceeded by torque improvements and torque ripple minimization. 2.2 Principle of Operation The SRMs main principle of operation relies on the reluctance difference in the air gap. The rotor has minimum reluctance when the stator and rotor poles are on the fully aligned position. The rotor phase reluctance reaches a maximum, and the inductance reaches a minimum at the unaligned position, where the rotor and stator poles are fully unaligned [13]. The torque direction is independent from the direction of the current, and depends on the rate of change of the inductance with respect to the rotor position. SRMs provide a positive torque in the motoring region where the slope of the inductance is positive. Correspondingly, they provide negative torque in the generating or breaking region where the rate of change of the inductance with respect to rotor position is negative. Fig. 2.1 presents the typical inductance profile of the SRM. By synchronizing each phase 5

18 excitation to the corresponding inductance profile, SRMs can provide continuous torque [9], [14]. Inductance Motoring Generating Rotor Unaligned Aligned Figure 2-1 Phase inductance profile The nonlinear characteristic of SRMs depends on the instantaneous torque-currentposition and flux-current-position. Flux linkage with respect to rotor positions for different current levels are shown in Fig The energy conversion of SRMs can be explained using the stored magnetic energy (W) and co-energy (W ) as shown in Fig Energy provided by the electrical supply is equal to the stored energy in the field (W) and the mechanical output energy neglecting the losses. 6

19 Figure 2-2 Electromagnetic profile of SRMs Figure 2-3 Energy conversion of SRMs The energy conversion ratio (ER) in an SRM is equal to the stored energy over the input energy as; EEEE = ww ww+ww (1) The phase voltage equation of SRMs can be written as; VV pph 7 = iiii +

20 (2) 8

21 where, VV pph is the DC bus voltage, i is the phase current and λλ is the flux linkage. SRMs have nonlinear flux linkage characteristics dependent on the phase current and the rotor position. VV = iiii + + (3) pph The derivative of the flux linkage with respect to current is defined as the instantaneous inductance LL(θθ) while the derivative of the position with respect to time is the angular speed ωω. When magnetic saturation is neglected, the phase voltage expression can be rewritten as; VV = iiii + LL(θθ) + ii (θθ) ωω (4) pph The second term on the right side of Eqn. (4) represents the back electromechanical force (back-emf), which is directly related to angular speed. The power of SRMs can be obtained by multiplying the voltage equation times the phase current, shown as; PP = VV ii = ii 2 RR + iiii(θθ) 1 2 (θθ) ωω (5) pph + ii 2 The first term on the right side of Eqn. (5) represents losses, the second term represents the stored energy and the third term represents the mechanical power. SRMs operate under magnetic saturation because under magnetic saturation, the input energy and the energy ratio increase. Power can also be represented in terms of the energy and co-energy with respect to time as: 9

22 PP = + (6) tt 10

23 The term can be defined as the instantaneous mechanical power. Neglecting the ohmic losses in Eqn. (2) power can be represented as: PP = ii (7) Substituting of Eqn. (7) into Eqn. (6), yields: ii = + WW (8) To simplify to the general torque equation, the differential co-energy can be defined as; (θθ. ii) = + (9) Assuming that the current is constant, from Eqn. 9, the torque can be written as; TT = (10) WW = ii2 LL(θθ) (11) 2 Substituting Eqn. (11) into Eqn. (10), the simplified general torque equation can be written as; TT = 1 2 (θθ) ii 2 (12) Similar to other machines, the maximum torque of the SRM is limited by current and the maximum speed is limited by the bus voltage. Fig. 2.4 shows the torque speed characteristics divided into three regions. Under the base speed, also known as the low- 11

24 speed region, the machine has a constant torque. In this region, the current reference and firing angle are the control parameters of the system. The SRM has constant power in the 12

25 control angle region. Increasing the speed of the machine increases the back-emf, thus the stator current cannot be built up to the maximum allowable levels. Therefore, the magnitude of the current cannot be controlled. This region is also known as the single pulse operation, which means current is not regulated. The only control parameters are the turnon and turn-off angles in the single pulse operations. To build a desired high current value, the turn-on angle is decreased when the machine speed is increased. To avoid the breaking region, the turn-off angle should be decreased because there is not ample time for demagnetization at high speeds [15]. Figure 2-4 Torque-speed characteristic of SRMs 2.3 Design of SRMs The number of phases, stator and rotor poles and the pole arcs provide the SRMs fundamental design parameters. Minimization of the core losses, good starting torque and ripple minimization are the objectives of a design [16]. The SRMs fundamental operating frequency is given by; ff = ωω NN (13) ss 60 rr 13

26 where, ff ss is the fundamental switching frequency, ωω is the motor speed rev/m and NN rr is the number of rotor poles. Several for number of stator and rotor poles combinations can be applied. The relation between the number of stator and rotor poles can be defined as; NNrr = NNss ± kkkk (14) where, k is an integer number, m is the pole number per phase. Further design considerations include available coil area, saliency ratio (the ratio between the maximum and minimum inductance), split ratio (the ratio between the rotor radius and outside radius), and reluctance variation. 2.4 Converter Topologies Unipolar current and electrically isolated stator poles allow several power converter topologies. The asymmetric/novel converter is the most common power converter, shown in Fig It has two switches and two diodes for each phase. Independent control of each phase is the primary advantage of the novel converter. The two switches are used to control the excitation current developed in each phase. The diodes provide the discharge path for the current in the freewheeling and demagnetization modes [10]. There are some approaches which require a new converter topology to reduce the torque ripple. Including even applying a high demagnetization voltage [17] to reduce the demagnetization time. Thus, the average torque improves due to the longer excitation time. 14

27 S1 VDC A+ A- S3 S5 B+ B- C+ C- S2 S4 S6 Figure 2-5 Typical SRM power converter. In the magnetization mode, a phase is excited when both switches are on, as shown in Fig In order to regulate the current, the upper switch turns on and off while the lower switch is kept on. In this case, to demagnetize the current, zero voltage is applied to the phase as shown in Fig This regulation is also called soft chopping because the current decreases slowly. To regulate the current, both switches can be turned on and off. In this case, the current decreases with the bus voltage applied to phase winding, as shown in Fig. 2.8 This method is also referred to as hard chopping because the phase current decreases faster [6]. 15

28 Figure 2-6 Magnetization mode Figure 2-7 Freewheeling mode 16

29 Figure 2-8 Demagnetization mode The split capacitor converter, shown in Fig. 2.9, has one switch per phase with a split DC supply. The phases are energized by upper or lower phases switches, and demagnetized through diodes. The main disadvantages are that only half of the DC bus voltage is applied, and the machine requires an even number of poles. D4 VDC Figure 2-9 Split-capacitor Converter In low speed applications, the novel converter simplified by Miller is shown in Fig It requires one switch per phase and an additional switch for magnetization. The main advantages are reduced switching losses and a simplified control algorithm. 17

30 D3 VDC Figure 2-10 Miller Converter The C-dump converter has a lower number of elements per phase. In addition to the simplified control strategy of the C-dump, hysteresis and PWM control can also be applied [18]. The voltage on the dump capacitor must be at least 2Vdc to implement -Vdc in the demagnetization region. The dump switch frequency is higher and it is hard to prevent overcharging of the capacitor. The converter requires an inductive load to limit the current during dump voltage regulation. VDC Dd Dc Figure 2-11 Energy Efficient C-dump Converter The energy efficient of the c-dump converter, shown in Fig. 2.11, converter has a diode block to prevent the overcharge current instead of the inductance. The dump capacitor applied the phase during demagnetization, that reduces the dump capacitor voltage level from 2Vdc to Vdc [19]. 18

31 2.5 SRMs Control SRMs are controlled by sequential energizing patterns due to the independent phases. The commutation angle, and currents profile determine the machine performance. A typical SRM closed loop speed control block diagram is shown in Fig To control the SRMs, current wave shape and magnitude must be controlled. For the speed control algorithm, illustrated in Fig. 2.12, the speed error, the reference current and the firing angle are the main control parameters. The energizing pattern is directly correlated to the speed of the machine. Gate signals are the output of the controller [23]. Current fedback Converter Actual Speed Commutator Figure 2-12 Control block diagram of SRM Feedback For motoring operation, the phases should be energized while the phase inductance increases as shown in Fig For generating operation, the phases should be energized while the phase inductance decreases. 19

32 Figure 2-13 Basic motoring operation Commutation angles, referred to as firing angles, should be selected to maximize the output torque and minimize losses. To increase torque, the dwell angle tends to increase; however, maximizing the dwell angle also maximizes losses. Efficiency can roughly be defined as the total output torque per rated current. There can be multiple turnon and turn-off angle configurations which provide the same output power. To optimize the dwell angle, the pair which provide the same power for lower rated current should be chosen [20]. The phase excitation controlled by the firing angle is required to be optimized, considering the speed of the machine and desired torque. There are several approaches in the literature to optimize the firing angle [21]. The methods to optimize the firing angle can be mainly classified as analytical [22] [23], self-tuning [24], model based, and artificial intelligent [25] methods. In [22] the turn-on angle is calculated analytically considering the desired position for peak current, and the angle is controlled with a feedback in order to maximize the 20

33 motoring efficiency. In addition, the authors control the turn-off angle with data gathered from simulation and experimental test results. For generating operation, optimizing the firing angle is more complicated due to the essential profile of the phase current. The low-speed and high-speed control algorithms of the SRMs have different characteristics. At low-speed operations, the phase currents can be profiled to achieve a desired torque. Regulating the current is more difficult when the machine speed increases. Eventually, the current cannot be regulated at high speed operation, also referred to as single pulse mode of operation. The turn-on and turn-off angles are the only control parameters at single pulse operation, as shown in Fig Figure 2-14 Single pulse operation When the machine operates at high speeds, the firing/dwell angle increases to obtain the desired torque. Optimizing the turn-on angle allows the current build, where the phase inductance is more reluctant to the increased current. In addition, the optimum turn- 21

34 off angle is decreased when the machine speed increases to prevent breaking torque during demagnetization. 2.6 Average Torque Improvement and Torque Ripple Minimization SRMs have an inherit torque ripple production mechanism because of their doubly salient structure. The main torque ripple, also called torque dip, occurs, when the exciting phase commutates from one to the other [20]. At that point, these two phases equally share the torque as shown in Fig There are several torque ripple minimization methods which can be broadly classified into two categories; machine design based and control based. Design based torque ripple minimization methods are reviewed in Section 2.5.1, and control based torque ripple minimization methods are reviewed in Section To reduce torque ripple in the SRMs, several converter topologies have also been developed. In Section 2.5.3, some examples of these converter topologies are discussed. Figure 2-15 Torque sharing Design Based Torque Ripple Minimization Increasing the number of phases reduces the torque ripple because the main torque ripple (torque dips) occurs when the excitation phase is changing. However, increasing the 22

35 number of phases also increase cost and complexity due to the need for new switches and diodes on the power inverter. Torque dips can also be reduced by increasing the number of strokes per revolution with a larger number of rotor poles. The saliency ratio, which is the ratio between maximum and minimum unsaturated inductance levels, decreases due to an increase in the number of strokes. Unfortunately, decreasing the saliency ratio reduces the average torque and increases the control volt-ampere requirement. In addition, higher frequency switching is required for an increased number of strokes, which increases core and switching losses [6] Control Based Torque Ripple Minimization The common control algorithm for SRMs provides a current reference optimal value which is determined by the rotor position and torque requirement. To decide the optimum current value, look-up s gathered from finite element analysis can be used [24]. The block diagram of general torque ripple minimization control system is shown in Fig Current fedback Converter Actual Speed Commutator Figure 2-16 Control block diagram of SRMs Feedback 23

36 One of the most commonly used control approach to reduce torque ripple is torque sharing function techniques (TSFs). The total torque is governed by the individual phase torque, which references to a torque value that uses a control parameter instead of current. However, it requires a high-bandwidth current regulator and define individual phase torque. Because of the nonlinear characteristics of SRMs, it is difficult to define the phase torque with respect to rotor position [25]. Current profiling is another common approach used to reduce torque ripple. To achieve the desired instantaneous torque, the reference current value can be obtained with a detailed machine model [26], which is used to obtain the T-i- θθ characteristics of the machine. The current profiling is updated with the mutual torque, after selecting current profiling via the machine model to obtain the desired torque (with minimum ripple). Adaptive fuzzy logic control is another approach which is applied to reduce torque ripple. In principle, fuzzy logic approximates the relationship between control parameters for weighting the parameters. Adaptive fuzzy logic control can also be adjusted to changes in the control parameters [27]. The PWM current control method is applied for profiling the phase current. The reference current profile calculated from encounter constant torque reference. This approach also considers the effect of magnetic saturation. The main disadvantage of this approach is that it requires a more accurate model of the machine [13]. 24

37 CHAPTER Ⅲ SPLIT WINDING STRUCTURE 3.1 Introduction This thesis proposes a SRM split winding configuration driven with an extra switch novel converter design to improve the machine torque-speed characteristic at medium and high-speed operations. High back-emf voltage and short demagnetization time limit the current at high speeds. The multi winding configuration allows a portion of the winding to be used at high speeds to reduce the back-emf voltage and phase inductance. Due to the lower back-emf voltage and phase inductance, the phase current can then reach higher values in the high-speed regions than that of a conventional set-up. The associated converter requires an extra switch per phase to use a portion of the winding, but the rated currents of these switches are lower because they are used only at high speeds, where the rated stator current is lower. After defining the split winding configuration in Section 3.2, the proposed converter topology is investigated in Section Proposed Work The principle is to allow more room for the current to reach higher values at high speeds, thereby providing higher torque density and better torque profile. The advantage 25

38 of this method is the reduced flux linkage due to the smaller number of turns, according to Eqn.3.1 [21]: λλaa = NN (3.1) where λλaa, N and are the flux linkage, number of turns and total flux associated with phase A respectively. Since λλ = λλ (θθ, ii), the change of flux linkage and torque in phase A can be given by Eqns [22]- [6]: dddd = dddd dddd + dddd dddd = ll dddd + ωω dddd (3.2) dddd dddd dddd ddθθ dd ddtt dddd ee TT = 1 ddddaa(θθ) ii (3.3) 2 dddd where, ll is the incremental inductance and θ, ω and TTee are the rotor position, motor speed and the average torque produced by phase A, respectively. From Eqns. 3.1 and 3.2; reducing the number of turns, the total flux linkage for phase A will decrease which will, in turn, drop the back-emf voltage as defined by the second term in the right-hand side of Eqn.3.2. Accordingly, the instantaneous current will rise, which should increase the torque production. However, from Eqn. 3.3; the torque will also decrease by decreasing the number of turns. Clearly, the turns ratio requires further analysis in order to account for the gain in torque by boosting the current and the reduction in torque by sacrificing some of the windings turns. The torque ripple is defined by Eqn. 3.4 [13]: aa Torque Ripple = TT iiiiiiii(mmmmmm) TT iiiiiiii (mmmmmm) 100% (3.4) TT aaaaaa The average torque will increase along with the instantaneous maximum and minimum 26

39 torque which would minimize the torque ripple. 27

40 3.3 Phase Winding Scheme The proposed SRM requires the proposed split winding configuration to energize only a portion of the winding in the high-speed region. The split winding scheme is obtained by interconnecting the primary and secondary windings as shown in Fig The two windings act as a transformer due to the magnetic coupling between them. In other words, the leakage inductance between them is close to zero. The machine physical and geometrical specifications remain the same as that of a full winding configuration, the split winding configuration is the only difference. Figure 3-1 Transformer winding configuration in the stator pole The total number of turns is divided into two windings where N1 is the primary and N2 is the secondary winding. The split ratio can be optimized by considering operation speeds and applications. The optimization of the number of turns for split windings is essential for torque improvement. At low speed demagnetization operations, the windings act like a transformer and so it should also be considered to optimize the number of turns. 28

41 3.4 Proposed Converter Topology The proposed power converter design along with the winding topology is presented in Fig The converter is mainly an asymmetric convert with an extra switch per phase to achieve a different number of turns for different speed ranges. The rated current of the extra switch is much lower than that of the other switches as it carries current only during high speed operations, which is much lower than the rated current of the inverter. The converter modes of operations are defined below for slow and high-speed operations. S1 S7 S3 S8 S5 S9 A+ B+ C+ VDC A N2 B N2 C N2 N1 N1 N1 A- B- C- S2 S4 S Magnetization Mode Figure 3-2 The proposed converter design. For low speed high torque operation, to energize the whole winding, switches S1 and S2 are used as shown in Fig In this mode, bus voltage is applied to the whole winding and the machine operates the same as that of the full winding topology. At high speeds, switches S7 and S2 should be used to energize only the secondary winding as shown in Fig

42 S1 A+ S7 V DC A N2 A- N1 S2 Figure 3-3 Low speed magnetization V DC ia S1 A+ A N2 S7 A- N1 S Freewheeling Mode Figure 3-4 High speed magnetization For the freewheeling mode at low speed operations, to apply zero voltage on both winding, S2 is switched ON and OFF to regulate the current while S1 is on for the full conduction time, as shown in Fig For high speed operations, to apply zero voltage on to the primary winding, S7 is used for the same purpose while S2 is on for the full conduction time, as shown in Fig

43 VDC S1 A+ ia N2 A S7 A- N1 S2 Figure 3-5 Low speed free-wheeling Demagnetization Mode Figure 3-6 High speed free-wheeling For the demagnetization mode; the portion of the windings defined by N1 is used to discharge the current at both high and low speeds, as shown in Fig The demagnetization capability at low speeds is improved as the phase current is discharged through a smaller number of turns, hence a high voltage/turns ratio will be applied during the demagnetization. 31

44 V DC S1 A+ ia N2 A S7 A- N1 S2 Figure 3-7 Low and high-speed At low speed operations, the split winding topology allows reduction the of demagnetization time by boosting the demagnetization voltage. During excitation, the windings share the bus voltage (VVdddd) as follows; VVdddd = VV1 + VV2 (3.5) VV 1 = NN1 (3.6) VV 2 NN 2 1 VV = NN 1 VV (3.7) NN 1+NN 2 dddd During the demagnetization time, the voltage can be represented as follows; VV1 = VVdddd (3.8) VV = NN 2 VV (3.9) 2 NN 1 dddd 32

45 where as VV1 increases, a faster demagnetization is expected. The demagnetization current flows thorough the N1 winding. Since the two coils are wound around the same core, the 33

46 main coil flux will cause the current level in the N1 winding to increase very fast. 3.5 Proposed Control Algorithm The SRM with split winding and its associated converter have been developed to improve the torque-speed characteristics of the machine at medium and high-speed operations. Hysteresis current control will be applied to regulate the current [30]. Only soft chopping current control will be used for low speed operations because of the transformer effects explained in Section 3.2. The rated speed of the machine is 1000 rpm. Medium and high-speed regions are taken as 2000 rpm or higher, where only a portion of the winding is excited. At high speed operations, the full winding current cannot reach the desired values because of the high back-emf voltage. Exciting only a portion of the winding by using an extra switch at medium and high-speed operations boosts the current, as mentioned in Section Conclusion The proposed split winding configuration, converter topology and the control algorithm were presented in this chapter. In addition, the split winding structure and associated power converter topology were introduced and optimization of the control parameters was explained. The proposed configuration improves the torque magnitude and profile in the high-speed regions using a lower number of turns. The purposed work is supported analytically in this chapter. 34

47 CHAPTER Ⅳ SYSTEM MODELING AND SIMULATION RESULTS 4.1 Introduction In this chapter, split and full winding machines models are built with their associated power converters. Each configuration is tested at different speed levels, with soft chopping hysteresis current control. 4.2 Split Winding Configuration Machine Modelling Finite Element Analysis Model The models of the electrical machines are typically described using differential equations. However, a detailed model can be obtained through finite element analysis methods [23]. Both machines are modeled with geometric specifications which include airgap, rotor and stator dimensions, overlapping, tapping angle and operation conditions of the machine. 100W 3 phase 12/8 SRMs are modelled geometrically using FEA software. Due to the simple geometric structure of the machine, a 2D drawing and the axial length of the machine are enough to geometrically describe the machine. Steady state and dynamic behaviors of the machine have been obtained via FEA [24]. The method analyzes the magnetic distribution and magnetic performance of the machine. Torque-current-position (TT ii θθ) and flux-current-position (λλ ii θθ) characteristics can be obtained from 35

48 finite element analysis. Fig. 4.1 shows a 2D FEA model of the machine built in Flux 12.1 by CEDRAT. Figure /8 SRM finite element analysis model 4.1 Geometric specifications of the SRM Parameter Value Units Rotor outer radius mm Rotor internal radius 30.5 mm Machine outer radius mm Shaft radius 8.54 mm Number of stator poles 12 Number of rotor poles 8 Stack length mm Number of phases 3 phase N1 Number of turns 100 number of turns N2 Number of turns 50 number of turns Phase resistance 2 ohm 36

49 To obtain the phase flux and torque characteristics, the phase is excited with a constant current for at least one electrical cycle. Sweeping the current magnitude with a small enough step size and a smaller rotation angle and increasing the mesh resolution, more accurate characteristic can be obtained. However, the solver processor speed limits the step angle, mechanical degree step and the mesh resolution. The phase torque and flux characteristics of the full winding topology with respect to angular rotor position at 5 different current levels are obtained, as shown in Fig , respectively. The torque characteristic shows the maximum torque that can be produced by phase A, or around 1.4 Nm and 0.5 Nm for given 5A and 3A constant currents, respectively. Both the flux and torque characteristics show that the phase has an aligned position at 22.5 mechanical degree, where the flux reaches the maximum value and the torque production changes direction. Figure 4-2 Torque characteristic of full winding topology 37

50 Figure 4-3 Flux characteristic of full winding topology To describe the proposed machine, the torque and flux produced by portions of the winding characteristics are also gathered via FEA. In Fig. 4.4, the torque characteristic of the proposed machine on the N1 winding, is presented where current is swept from 0 to 6 Amps in 0.3 A steps, while rotational steps are taken as 0.1 mechanical degree. Fig. 4.5 shows the flux characteristic of the proposed machine, on the N1 winding, for the same current step and bandwidth. The torque characteristic shows the maximum torque that can be produced by the proposed machine, or around 0.7 Nm and 0.25 Nm given 5 A and 3 A constant current, respectively. The flux characteristics for the full winding and split winding topologies show the reduction of the flux on the phase, where the maximum was around 0.2 Wb to 0.8 Wb for the full winding and split winding, respectively. 38

51 Q) ::,,Ọ_'" Q) Cl) cu [: A - 2A - 3A - 4A - 5A Rotor Position (degree) Figure 4-4 Phase torque characteristic of split winding topology LL cu. a.. c A - 2A - 3A - 4A - 5A Rotor Position (degree) Figure 4-5 Phase flux characteristic of split winding topology To obtain a detailed characteristic of the machine both winding topologies are 39

52 characterized with small enough current steps of 0.1 A and mechanical degree steps of

53 mechanical degrees. The flux and torque characteristics defined as function of the current and position information, are as follows; λλ = λλ(ii, θθ) 3.1 TT = TT(ii, θθ) 3.2 These two functions represent the model of the machine with each angle and current values corresponding to only one flux and torque value Matlab Simulink Model The machine is also modeled using Matlab Simulink for faster simulation. The Matlab model of the machine is described by flux and torque data gathered from FEA. This makes for a more accurate model compared to the model obtained from differential equations. The data, uploaded into the Matlab simulation via look-up s, can provide multiple dimensional data sets. For unknown positions directly interpolation is used, as shown in Fig For a given current and position values, the instantaneous torque and predicted flux can be defined from these look-up s. This allows the use of a faster solver speed to simulate a more detailed model. Figure 4-6 Flux and torque Look-up s 41

54 The flux data gathered from the look-up is used for the phase voltage of Eqn.3. Integrating the voltage, after accounting for ohmic loses, gives the flux linkage value from Eqn.4. The Matlab schematic is shown in Fig An algebraic constraint is used to find the actual current on the phase. The algebraic constraint Matlab block solver finds the only current value which satisfies both the phase equation and flux look-up data. Figure 4-7 Phase equation Matlab schematic 4.3 Power Converter Model The control algorithms are applied to both the conventional and the proposed power converters. The Matlab simulation is used to tune the firing angle and the PI parameters. The final simulation results are gathered from the FEA. The conventional and proposed converter are built on the Flux FEA software, as shown in Fig respectively. The switches and diodes on the models have a resistance 0.1 mω for the ON state and 10 kω for the OFF state. The phase windings in the converters represent the machine winding placed at the slot on the stator in the FEA model. 42

55 Figure 4-8 Conventional power converter Figure 4-9 Proposed power converter 43

56 The control signals for the switches are generated based on the given turn-on and turnoff angles, rotor angular position and desired torque values. In addition, current sensors are placed at each the phase to control phase currents. Hysteresis current control is implemented to regulate the phase current within a 0.2 A bandwidth. The DC bus voltage is kept at 50 V for both converter topologies. The current reference at the rated speed is chosen as 5 A with the current density of 4,7-5,2 A/mm2 [6]. 4.4 Control Algorithm The hysteresis current control with soft chopping is applied to control the torque for a given speed. As mentioned before, soft chopping is applied with freewheeling to reduce the switching frequency. For simulation purposes, speed control was not applied, as the speed was directly taken from given speed. To apply speed control, the mass of the rotor and force should be known to calculate the inertia and PI parameters. The integration of the speed gives the angular rotor position, which is the control feedback. The phase angle is compared with the turn-on and turn-off angles. The excitation period of each phase deicide where the phase position is between these angles. The current reference indicates the torque component, as a result, to produce a more torque, the reference current should be increased, or vice versa. During the excitation period, the lower device on the phase is kept on the ON position, and the upper switch is turned ON and OFF to regulate the current. For simulation results, the feedback of the current is only used for regulating the current. The desired torque value is directly given as an input with the reference current signal. For different speed levels the maximum torque can be produced from the full and split winding machines are gathered. The maximum 44

57 current value is referenced at all speed levels, but at the medium and high-speed regions, the full winding topology cannot reach the desired current levels. The set-up was simulated at different speed levels. For each simulation, the tur-on and turn-off angles and the current reference values are specified. To optimize the turn-on and turn-off angles for different speed levels, the tuning method used. The same turn-on and turn-off angles, along with the same current reference are applied for both machines. 4.5 Simulation Results The split and full winding machines are operated at different speed levels to provide a comparison. The proposed split windings with the new power converter improve the machine performance at high speeds by boosting the excitation current to levels that cannot be reached at high speeds in the full winding topology. In the low speed regions, the machine operation is similar to that of the full winding machine, except during the demagnetization. The split winding system boosts the negative voltage applied to the secondary windings to discharge the current and provides a shorter time, as mentioned Section 3.2.C. The faster discharge of the split winding machine phase current is compared to the typical time elapsed with the full winding machine. After switching off, the proposed machine phase current on the secondary winding jumps to higher values as shown in Fig Faster demagnetization allows more time to excite the phase. The main torque ripple, or torque dip, can be reduced by faster demagnetization. However, this effect only occurs in the low-speed regions, where there is enough time to reduce the current. 45

58 Figure 4-10 Phase currents at 1000 rpm. In the high-speed regions, the power converter control excites just the primary winding which has the higher number of turns. The improvement in the current profile is shown in Fig and 12 respectively at 3000 rpm and 4000 rpm. At 3000 rpm with the full winding machine, the current cannot be controlled because the back-emf is too high. However, the desired current can still be developed with the split winding machine, as shown in Fig At 4000 rpm, which is the highest speed of the machine, the phase current can be built up to around 4.2A and 2.2A with the full and split winding system, respectively. The current can be built up to higher values on the split winding machine than the full winding machine in the medium and high speed regions. 46

59 Figure 4-10 Phase currents at 3000 rpm. Figure 4-11 Phase currents at 4000 rpm The comparison between the flux linkages are shown in Fig for different speeds. An advantage is that the new system can improve the SRM torque characteristics at high speeds using the same machine design by changing the winding configuration and using 47

60 one extra switch per phase. The split winding characteristics of the flux density was much lower than that of the full winding one for a given current level and rotor position. However, as seen from the flux figures, the flux values are quite close and at higher speed levels, they became closer. The reduction on the phase inductance reduces the flux value for same current levels, but the split winding topology improves the current, as a result, the flux value for dynamic systems are closer to the full winding topology. Figure 4-12 Flux linkages at 2000 rpm 48

61 Figure 4-13 Flux linkages at 3000 rpm. Figure 4-14 Flux linkages at 4000 rpm. 49

62 The torque comparisons for the case study, at different speeds, are presented in Fig The higher current values enhance the average torque production and enables the driver to control the current, hence minimizing the torque ripple and the acoustic noise typically associated with the SRM operation. At 2000 rpm, the first high speed level, both split and full winding topologies produce a maximum average torque of 0.53 Nm, as shown in Fig However, the split winding topology improves the torque profile with a significant torque ripple reduction due to the different current magnitude and profile that are applied. The torque ripples are 66% and 110% at maximum torque on the split and full winding respectively. Figure 4-15 Split and full winding torque comparison at 2000 rpm 50

63 At 3000 rpm, the proposed machine reaches a higher maximum torque with a better profile, as shown in Fig The split winding topology produces maximum of 0.34 Nm average torque with 74% ripple. The full winding topology produces a maximum 0.28 Nm average torque with 124% ripple. Figure 4-16 Split and full winding torque comparison at 3000 rpm At 4000 rpm, which is the highest speed level for the full winding topology, the split winding produces a recognizable higher torque than the full winding. The split winding topology produces a maximum 0.28 Nm average torque with 110% ripple. The full winding topology produces a maximum 0.17 Nm average torque with 123% ripple, as shown in Fig

64 Figure 4-17 Split and full winding torque comparison at 4000rpm At 4500 rpm, the split winding machine can produce 0.22 Nm torque with 110 % ripple, as shown in Fig At 5000 rpm, the split winding can produce a maximum 0.19 Nm torque which, is broadly higher than the full winding torque at 4000 rpm. Fig shows the output torque when the split winding is driven at 5000 rpm and the full winding is driven at 4000 rpm. 52

65 ze (]) 0.2 ::J O" L.. 0.1,, I L - I - Ill r,r- Ill, 0 -Split-winding Average Torque Time (seconds) Figure 4-18 Split winding torque at 4500rpm E - z - O" L Time (seconds) Figure 4-19 Torque comparison split winding at 5000rpm, full winding at 4000rpm 53

66 The split winding topology improves both the magnitude and profile of the torque at high speed. The torque profile can be optimized by the firing angle which is kept the same for the full and split winding topologies. Both the split and full winding topologies produced the same torque of 0.22 Nm at 3000 rpm and 0.10 Nm at 4000 rpm respectively. The split winding topology can produce the desired torque with 70 % ripple as compared to the full winding torque with 110 % ripple at 400 rpm, as shown in Fig Similarly, at 4000 rpm, the split winding topology can produce the desired torque with 20 % ripple while the full winding torque with 80 % ripple as shown in Fig Figure 4-20 At 3000 rpm torque ripple comparison 54

67 4.6 Conclusion Figure 4-21 At 4000 rpm torque ripple comparison. Both the full and split winding configurations are modeled via CEDRAT Flux, FEA software. The characteristics of the machines obtained from FEA are used to define the machines using MATLAB Simulink. The machines, and their associated converter topologies, are simulated in both Flux and MATLAB Simulink. The turn-on and turn-off angles are optimized for different speed levels. To optimize the firing angle, a dynamic model of the system is derived using MATLAB. Given that the simulation results support the objectives of the proposed work, the next step would be to validate them, experimentally. Thus, an experimental set-up is required. The proposed split winding topology improves both the torque-speed and power-speed characteristics of the machine. The main improvements occur at the high-speed region, as expected. The improvement in the magnitude of output torque is more obvious than the ripple minimization. However, the ripple minimization is also observed at the desired 55

68 torque for the given speed. The split and full winding topologies are driven with the same current refence and firing angle. The proposed machine configuration torque ripple still can be improved by optimizing the firing angle because the phase current can be controlled even at the highest speed region. 56

69 CHAPTER Ⅴ EXPERIMENTAL SET-UP AND TEST RESULTS 5.1 Introduction The simulation results verified that the split winding topology improves the torque characteristics of the machine at the medium and high-speed regions. This is due to the increased magnitude of the torque and the reduction of the torque ripple. To validate the simulation results, the proposed topology set-up is designed, built, and tested. The proposed extra switch converter can also be used as a conventional asymmetric converter to drive the full winding machine. The process of the hardware implementation and their results are discussed in this chapter. 5.2 The Split Winding Machine A case study for a 12-8, 100 W 1000 rpm motor is used to investigate the competency of the winding configuration along with the new power converter design for the SRM. The specifications of the machine are depicted in 5.1. The total slot area for the machine is 104 mm^2 and is divided as 69.3 mm^2 for the primary windings and 34.7 mm^2 for the secondary windings according to the number of turns distributed between them; for N1= 100 and N2=50 turns. The disassembled motor is shown in Fig

70 Figure 5-1 Stator of the experimental SRM 1: Specification of SRM Quantity Value Units Rated Power 100 Watts Based Speed 1000 rpm Maximum Speed 4000 rpm DC Voltage 50 Volts Number of Stator Poles 12 Number of Rotor Poles 8 Number of Phases 3 Each stator phase has 4 poles. Two of the poles are connected in series. Two of the series connected pairs are connected in parallel. The winding configuration is the only difference between the proposed and conventional machines, all of the other specifications 58

71 are the same. The proposed machine has two windings, (50 turn and 100 turns, respectively), on each coil conductor, as shown Fig Figure 5-2 Winding configuration of the split winding machine Figure 5-3 Split winding machine winding configuration for phase B 59

72 5.3 Power Converter and Control Circuit The proposed power converter is an asymmetric converter with an extra switch per phase, as shown in Fig The converter rated power is 200 W and the bus voltage is 50 V. An insulated gate bipolar transistor (IGBT) is used as a switch, which is a well sui soft chopping and machine control inverter with up to 15 A and 600 V. Nine IGBT and six diodes are the main component of the converter. To obtain a stiff bus voltage, capacitors are used to eliminate the ripple caused by high or low frequency switching [32]- [33]. The converter reference current was chosen to be 5 A with a current density limit of 3.4 A/mm^2, while the converter DC voltage was set to be 50 V. Figure 5-4 Extra switch power converter The control circuit, which consists of current sensors and conditioning circuit, [34]. The control circuit main supply voltage is 5 V. The current sensors are required to measure phase currents at each phase to implement a current control algorithm. The current conditioning circuit designed to read the measured current value accounts for the DSpace 60

73 input signal requirements. There are two buffers in the control circuit to boost the gate signal from the processer and encoder signals. The gate drivers are responsible for driving the gates with the gate signal produced by DSpace. The gate drivers boost the voltage of the gate signal to achieve the required current level. The block diagram of the power and control circuit is shown in Fig Power Circuit Control Circuit Q7 N2 N1 +Vcc GND Current Sensor D2 GND +Vcc Conditioning Driver Buffer +Vcc GND DSpace Connecter Switching Signals Current Signals Figure 5-5 Block diagram of the control and power circuit 5.4 Printed Circuit Board (PCB) Design After designing the circuit and choosing the components considering the rated current and voltage levels, switching frequency at maximum speed, application type and cost, the printed circuit board was designed taking into account the signal clearance, cooling, 61

74 component, dimensions and footprints. The main concern was to meet the driver requirements for each component with an optimized layout. The schematic of the power and control circuit, as shown in Fig. 5.6, consists of IGBTs, diodes, gate drivers, current sensors, and buffers. 62

75 Figure 5-6 Power and control circuit schematics 63

76 5.4.1 Buffer for Gate Signals The PWM signals produced by DSpace are boost via buffer, signal S1-S8 are produced by DSpace while the boosted signal S1B-S8B are the input of the gate driver as shown Fig The buffers simultaneously enhance the signal quality and isolate the DSpace and gate driver circuits. Figure 5-7 Buffer circuit schematic Current Sensor and Conditioning Circuit The current sensors convert the current value of the phase into an analog voltage ranging from 0 to 5 V with a 2.5 V DC offset. The current sensor and its conditioning circuit schematics are shown in Fig. 5.8 and 5.9. The current conditioning circuit compares the signal with two series operational amplifiers that enhance the signal quality and isolate the current sensor and DSpace processor. C7 is used to reduce the voltage ripple of the supply (5V) and obtain a stiff voltage. R1-2 and C1-C5 are used as a low pass filter. 64

77 Figure 5-8 Current conditioning circuit schematic Figure 5-9 Current sensor schematic Gate Driver The gate driver circuit consist of a DC-DC converter to boost voltage, driver IC and fault detection, as shown in Fig The output of the gate driver is ±15V required to drive the IGBT. This gate signal is applied to the gate to emitter of the IGBT. 65

78 Figure 5-10 Gate driver circuit schematic The PCB layout consists of both control and power circuits, as shown in Fig The board has 2 layers with dimensions of inches. 5.5 Experimental Set-up Figure 5-11 Power and control circuit layout The 12/8 100W SRM, which is used in a washing machine, with a split winding is experimentally tested via the extra switch novel converter. To apply the control algorithm, 66

79 C1 C2 D1 Q1 Q2 D2 Q3 D3 Q4 a Dspace DS1401 board and Controldesk interface is used [36]. The experimental set-up includes a dynamometer, an incremental encoder, the split winding machine, a converter and control board, and the Dspace board and processor, as shown in Fig Switched Reluctance Machine M Dynamometer M Speed/Position D 4 Power converter and Control Circuit Control I/O Computer Dspace Board Figure 5-12 Experimental set-up configuration The dynamometer is an induction motor used as a mechanical load. The dynamotor is controlled to achieve the desired torque and loading for given speed. Sudden load changes can lead to erroneous torque values, when the machine speed control is too slow or poorly designed. To avoid overloading (the processor cannot respond within the system period), it is important that the load be smoothly applied [37]. The incremental encoder configuration, as shown in Fig. 5.13, is used to measure the angular rotor position and speed of the machine. The incremental encoder has three output square waveforms which are usually called A, B and Z. The speed can be measured 67

80 from the A or B signals which are 90-degree phase shifted from each other and Z is the 68

81 reference signal given in Fig The mechanical position is sensed from the A or B signals of the encoder, where the count of rising or falling edge of the signal. The speed of the machine measured from the total edges number during each period. For high resolution both A and B signals raising and falling edge can count [38]. The direction of the machine movement is sensed from the leading angel signals (A or B). If the signal A is leading B the machine turns in the positive direction, and vice versa. Figure 5-13 Incremental encoder disk Figure 5-14 A, B and Z signal of encoder respectively The set-up is controlled with a DS1401 board, its connection configuration is shown in Fig It is preferable to use control-desk interface because MATLAB simulations 69

82 can be uploaded directly. The angular rotor position and phase currents feedbacks are the inputs of the DS1401, and the gate signals are the output. Figure 5-15 SRM experimental setup 5.6 Control Algorithm The PI speed control algorithm is used to obtain maximum produced torque for a given speed level. The PI parameters are chosen via tuning. The error between the actual and desired speed values indicates the refence current. The PI parameters represent the speed and reliability of the control system. When the machine is driving for a desired torque at a given speed, if the load increases, the machine slows down. When the machine is slower than the desired speed, a positive error occurs. As a result, the reference current value increases in order to maintain the desired speed, or vice versa. At some point, the machine cannot build enough current to maintain the desired speed, and slows down. Before slowing down, the maximum torque value is the maximum torque that can be produced at the given speed level. 70

83 ɷ + Current Power Phase Figure 5-16 Control algorithm The optimized firing angle for each speed level is tuned during the simulations. There is no feedback for the firing angle parameters, they are controlled with an open loop control. The phase current is regulated by the soft chopping to reduce the switching loses by using the freewheeling mode as mentioned in Chapter 5.3. The control algorithm has two feedback signals; namely, phase current and position information, as shown Fig The control algorithm also has a different protection and initializing methods which improve the reliability of the system. First of all, the current reference value is limited on the software to overcurrent protection. Secondly, there is software lock which is not allowed the machine start working before one manual spin for initialize the angle and also safety for user. In addition to the main system start protection, the speed control has extra control to change into open loop torque control to overcome unexpected spikes on the refence value. 71

84 5.7 Experimental Results Both the conventional and proposed machine configurations are tested at different speed levels. Fig show the full and split winding topologies, phase currents, at different speed levels respectively. At 3000 rpm, the full winding topology phase current cannot be controlled because the current never reaches the desired values, as shown in Fig a. On the other hand, the split winding topology phase current can still be controlled, as shown in Fig b. a) Full winding topology phase currents at 3000 rpm b) Split winding topology phase currents at 3000 rpm Figure 5-17 Full and split winding topologies phase currents at 3000 rpm 72

85 a) Full winding topology phase currents at 4000 rpm b) Split winding topology phase currents at 3000 rpm Figure 5-18 Full and split winding topologies phase currents at 4000 rpm At 4000 rpm, which is the maximum speed, the current can still be built to the reference current value (5A) by energizing only a portion of the winding, as shown in Fig b. On the other hand, the current is barely half that of the conventional method as shown in Fig a. The torque of the machine is measured at different speed levels. After the speed is controlled by the PI controller, each speed level dynamometer torque increased until the machine slows down. The maximum torque, before the machine slows down, is the output 73