UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

A photovoltaic-powered pumping system Liu, Guang 1989

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-UBC_1990_A1 L58.pdf [ 7.69MB ]
JSON: 831-1.0065498.json
JSON-LD: 831-1.0065498-ld.json
RDF/XML (Pretty): 831-1.0065498-rdf.xml
RDF/JSON: 831-1.0065498-rdf.json
Turtle: 831-1.0065498-turtle.txt
N-Triples: 831-1.0065498-rdf-ntriples.txt
Original Record: 831-1.0065498-source.json
Full Text

Full Text

A P H O T O V O L T A I C - P O W E R E D P U M P I N G S Y S T E M Guang Liu B. Eng., Guangxi University, 1982 M. A. Sc., University of British Columbia, 1985 A THESIS S U B M I T T E D IN P A R T I A L F U L F I L L M E N T O F T H E R E Q U I R E M E N T S F O R T H E D E G R E E O F DOCTOR OF PHILOSOPHY IN T H E F A C U L T Y O F G R A D U A T E STUDIES DEPARTMENT OF ELECTRICAL ENGINEERING We accept this thesis as conforrjaing to the required standard THE UNIVERSITY OF BRITISH COLUMBIA December 1989 © Guang Liu, 1989 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of QN6M££RMq The University of British Columbia ) Vancouver, Canada Date DE-6 (2/88) Abstract This thesis studies the optimal design for a photovoltaic-powered medium-head (30 meters) water pumping system, with the emphasis on improving the efficiency and reduc-ing the maintenance requirements of the electrical subsystem. The reduction of main-tenance requirements is realized by replacing the conventional brush-type permanent magnet dc motor with a brushless dc (BLDC) motor. Different BLDC motor control techniques such as position-sensorless operation, sinusoidal and trapezoidal excitations are investigated. The improvement in efficiency is achieved by maximizing the output power from the photovoltaic array and by minimizing the losses in various parts of the electrical sub-system. A microprocessor-based double-loop maximum power tracking scheme is developed for maximization of the photovoltaic array output power. Over 99% utilization factor is achieved for a typical clear day regardless of the season of the year. The system losses are minimized mainly by performing loss analysis and selecting most suitable switching topologies and switching components. Experimental results show that the combined converter-motor efficiency is comparable to those of high-efficiency brush-type dc motor systems. ii Table of Contents Abstract ii List of Figures vi List of Tables ix Acknowledgement x 1 Introduction 1 1.1 An Overview of the Experimental Pumping System 3 1.2 Objectives and Scope of the Thesis Research 4 2 Design Considerations 6 2.1 System Configuration 6 2.1.1 Existing Pumping Configurations 6 2.1.2 Selection of System Components 8 2.1.3 The Selected PV Pumping Configuration 12 2.2 Brushless DC Motor Drive 14 2.2.1 Terminology 14 2.2.2 Trapezoidal Excitation and Sinusoidal Excitation 15 2.2.3 BLDC Motor Parameters and Losses 21 2.2.4 Position-sensorless Operation 25 2.3 The Converter-commutator Circuit 26 2.4 Maximum Power Tracking Realization 30 iii 2.4.1 The Need for Maximum Power Tracking 30 2.4.2 A Double-loop Maximum Power Tracking Scheme 31 2.5 System Losses and Efficiency 35 2.5.1 Maximum Efficiency from the Motor . . . 35 2.5.2 Losses in the Converter-commutator Circuit and Motor . . . . . . 36 z.o bummary 39 3 Hardware Design 41 3.1 Power Circuit 41 3.1.1 Power MOSFETs and Heat Sinks 41 3.1.2 Capacitors and Layout 44 3.2 MOSFET Gate Drives 46 3.3 Over-current Protection 48 3.4 Current Sensing and Over-temperature Detection 49 3.5 The Converter-commutator Logic Circuit . 51 3.6 PWM Signal Generation . . . 53 3.7 Microprocessor System Hardware 54 3.8 I/O Interface and Signal Conditioning . . . 54 3.9 Position-sensorless Operation 59 3.9.1 Rotor Position Signal Generator 60 3.9.2 Motor Starting Pulse Generation 62 3.9.3 Motor Mode Control 66 3.9.4 Reverse Rotation Detection 68 4 Software Development 69 4.1 Functional Description 69 4.2 Description of Major Routines 72 iv 4.2.1 Main Routine 72 4.2.2 Modules in READY State 74 4.2.3 MOTOR.ST , . 75 4.2.4 VCONT, ADJUST and POWER . . . 79 4.2.5 Fault Detecting and Handling Modules 83 4.3 Summary 84 5 Experimental Results 86 5.1 Maximum Power Tracking 86 5.2 Converter-motor Efficiency and Position-sensorless Operation 90 5.2.1 Efficiency Evaluation 90 5.2.2 Efficiencies With and Without a Position Sensor 91 5.2.3 Current Transient During Motor Mode Switching 91 5.3 Motor and Converter-commutator Operation 91 5.3.1 Sinusoidal and Trapezoidal Excitation 91 5.3.2 HEXSense Current Sensing and Over-temperature Detection . . . 94 5.3.3 MOSFET Waveforms 95 5.4 Summary 95 6 Conclusions 98 References 101 A Simulation study on maximum power operation of a P V array 108 B Flowcharts 114 C Recorded Operating Points of the PV Array 123 v List of Figures 1.1 I-V characteristics of a PV array . 2 2.2 Configurations of Most PV Pumping Systems 7 2.3 The configuration of the experimental pumping system 13 2.4 Current and emf Waveforms for Trapezoidal Excitation . 17 2.5 Simplified Electronic Commutator Circuit with the Brushless Motor . . . 22 2.6 A Plot of Measured Kf and Curve-fitted Kf 23 2.7 Friction and Windage Losses of the Motor 25 2.8 Brushless DC Motor Terminal Voltage Waveforms 27 2.9 Equivalent Circuit for the BLDC Motor with the Converter-commutator Circuit 28 2.10 Block Diagram of the Maximum Power Tracking Scheme 33 2.11 Two Switching Configurations for the Converter-commutator 37 2.12 Switching Loss for the Two Switching Configurations 40 3.13 Typical High Temperature Reverse Bias (HTRB) Failure Rate 42 3.14 One Leg of the Power Circuit 43 3.15 A Picture of the Converter-commutator Power Circuit . . 44 3.16 Voltage and Current Waveforms of a Bottom HEXFET 45 3.17 HEXFET Gate Drive Schematic Diagram 47 3.18 Over-current Protection Schematic Diagram 48 3.19 Current Sensing and Over-temperature Detection Circuit 49 3.20 Current Sensing Waveform 50 vi 3.21 Converter-commutator Logic Circuit 52 3.22 Circuit Diagram of the PWM Signal Generator 53 3.23 Block Diagram of the Microprocessor Development System and Interface 55 3.24 I/O Interface at Port A of VIA1 57 3.25 Interface at Port B of VIA1 58 3.26 Signals at Port A of VIAO 59 3.27 Schematic of Rotor Position Signal Generator 61 3.28 Voltage, emf and Position Signal Waveforms for Forward Rotation . . . 63 3.29 Voltage, emf and Position Signal Waveforms for Backward Rotation . . 64 3.30 Ring Counter for Motor Starting . . 65 3.31 State Diagram for the Ring Counter . . . . • . ,66 3.32 Motor Mode Control Logic 67 3.33 Reverse Rotation Detection Circuit 68 4.34 Top Level Control in Algorithmic State Machine Diagram . . . . . . . . . 70 4.35 Flowchart for Main Routine 73 4.36 Flowchart for MOTORJST 76 4.37 Flowchart for SYN 77 4.38 Flowchart for PULSE 78 4.40 Flowchart for ADJUST 79 4.39 Flowchart for VCONT 80 4.41 Flowchart for POWER . 81 5.42 Recorded PV Array Operating Points in A Period of 6 Minutes 87 5.43 PV Array Operating Points During Sudden Change of Radiation . . . . 89 5.44 Efficiencies with and without a position sensor 92 5.45 Motor Current During Motor Mode Switching 93 vii 5.46 Motor Current and Voltage in Sinusoidal Excitation 94 5.47 Motor Line Current Waveform and Three-phase Current Sensing Output Signal . . 95 5.48 Filtered Current Signals With and Without an External Resistor 96 5.49 Current Sensing Difference vs. MOSFET Case Temperature 97 5.50 Bottom MOSFET Current and Voltage Waveforms 97 A.51 The Effect of Radiation on Maximum Power voltage 109 A.52 The Effect of Temperature on Maximum Power Voltage 110 A.53 Power Loss vs. Operating Voltage 110 A.54 A Sample of the Radiation and Temperature in Vancouver I l l A. 55 Variation of Vopt with the Four Seasons 113 B. 56 Flowchart for INITIALIZE 114 B.57 Flowcharts for OUTJDRDY and OUT_D 115 B.58 Flowchart for VIJSAM 116 B.59 Flowchart for CURVE . 117 B.60 Flowchart for FAU_DT . 118 B.61 Flowchart for STLJDT 119 B.62 Flowchart for CHKRTR 120 B.63 Flowchart for REV_DT , 120 B.64 Flowchart for TEMPJDT 121 B.65 Flowchart for FAULTS 122 viii List of Tables 2.1 Switching Loss For Sinusoidal and Trapezoidal Excitation 21 2.2 BLDC Motor Parameters 24 2.3 Relation Between On-switches and Rotor Position . . . 26 5.4 Combined Converter-motor Efficiency 90 5.5 Measured Efficiency for Sinusoidal and Trapezoidal Excitation 93 ix Acknowledgement I would like to express appreciation to Dr. William G. Dunford, for his guidance and supervision, and to Dr. Malcome Wvong for assuming the role of acting supervisor when Dr. Dunford was on Sabbatical leave. Appreciation is also expressed to the staff of the Department of Electrical Engineering, UBC, for their help and assistance, particularly to Alan Prince and Lloyd Welder. I would like to thank my colleague Yin Yanan for his helpful assistance and discussion, and to Mr. David Carson of B. C. Hydro, for providing the operating record of the photovoltaic array installed in Surrey. Thanks are also expressed to National Research Council for providing the photovoltaic panels used in the thesis research. The work on this research has received financial support from the Science Council of British Columbia. x Chapter 1 Introduction This thesis studies the optimization of a microprocessor-based photovoltaic (PV) pump-ing system that uses a brushless dc (BLDC) motor drive. The rapid development in solid state technology has led to increasing installations of PV pumping systems, especially in areas remote from power utilities. As a developing technology, PV pumping has already shown its potential economic advantages [1, 2, 51]. However, more research is needed for improved overall performance of PV pumping systems. System efficiency and reliability are the main concerns in the development of a PV pumping system. As PV arrays are still the most expensive components in PV pumping systems, higher efficiency will lead to smaller array size and thus to lower expense for a given amount of water output. Since the PV pumping system must be able to operate for at least five to ten years to be economically competitive and since maintenance in remote areas is expensive, reliability is an important index in system performance. Conventional brush-type permanent magnet (PM) dc motors, extensively used in ex-isting PV pumping systems, have the advantages of high efficiency and simple control circuitry. However, these motors have an inherent drawback - the maintenance require-ment entailed by their brush-type commutation. The use of a BLDC motor eliminates brush-type commutation, so that longer motor life and reduced maintenance result. The efficiency of a BLDC motor is comparable to that of a conventional PM dc motor and is considerably higher than that of an induction 1 Chapter 1. Introduction 2 motor (IM), especially when the power rating is below 1 kilowatt. There are two aspects in maximizing overall system efficiency: one is to minimize the system losses for the full operation range; the other is to extract maximum power available from the PV array. Minimization of system losses can be obtained by carefully selecting system configuration, circuit components and parameters, motor excitation method, and by conducting loss analysis on different stages of the circuit. For maximum power to be extracted from a PV array, special control techniques are required. These techniques are often called maximum power tracking (MPT). The current-volt age (I-V) characteristics of a typical PV array is plotted in Figure 1.1, in which the I-V curves are nonlinear and vary with insolation levels and temperature. Figure 1.1: I-V characteristics of a PV array At the high voltage section of each I-V curve, the voltage is relatively constant while at Chapter 1. Introduction 3 the high current section the current is fairly constant. There is one point on each I-V curve, the maximum power point (MPP), at which maximum power can be extracted. For the array capacity to be fully utilized, the PV array should operate, under all weather conditions, as close to the MPP as possible. A simple PV pumping system consists of a PV array, a dc motor and a pump. The PV array converts solar energy to electrical energy. The motor fed by the PV array turns the pump which pumps water from underground. However, it is difficult to utilize the PV array capacity fully with this simple configuration, especially when the motor load has a constant torque characteristic. This problem is mainly due to the mismatch of the power supply (PV array) and the load (motor and pump). Better utilization of the available power from a PV array can be achieved by using suitable MPT techniques. An overview of the experimental pumping system is presented in the following section. 1.1 An Overview of the Experimental Pumping System The experimental pumping system is designed for medium water head (between 25 to 45 meters) applications. Major components of the system include: a PV array; a power MOSFET converter-commutator circuit; a BLDC motor; a progressive cavity pump, and a Motorola 6809 microprocessor development system. Since a progressive cavity pump is roughly a constant torque load on the system (for a given water head), some kind of power conditioning is desirable. A switching converter can be used as an efficient power conditioning circuit. An electronic switching circuit is also required to provide the commutation process normally obtained from a conventional mechanical commutator. In the experimental pumping system, a switching converter function and a commu-tation function are accomplished with a power MOSFET converter-commutator circuit. Chapter 1. Introduction 4 As will be explained in Chapter 2, the converter-commutator circuit which eliminates the use of a series power MOSFET that is required in some commercial BLDC motor drives, reduces the circuit losses. Because of the fast switching characteristics of power MOSFETs, the switching loss of the converter-commutator circuit is greatly reduced in comparison with a silicon con-trolled rectifier (SCR) or bipolar junction transistor (BJT) version. With the use of three HEXSenses (current sensing power MOSFETs from Interna-tional Rectifier Corp.) in the converter-commutator circuit, the PV array current is sensed with minimum power loss. Furthermore, by making use of the current sens-ing characteristics of the HEXSenses, over-temperature protection can be implemented without a temperature sensor. A double-loop structure, an inner voltage loop and an outer extremum control loop, is implemented with a microprocessor for MPT function. The voltage loop keeps the PV array voltage following a voltage set point by controlling the duty ratio of the converter. The extremum control loop modifies the voltage set point periodically. This structure accomplishes MPT function robustly and effectively. The use of a microprocessor development system accommodates the development of the MPT algorithm, position-sensorless operation, etc. The development system also facilitates some simple diagnoses which improve system reliability. The following section describes the objectives and scope of this thesis work. 1.2 Objectives and Scope of the Thesis Research The main objectives of the thesis research are to maximize the efficiency and reduce the maintenance requirements of the electrical sub-system. The more specific aims of the research work can be listed as follows: Chapter 1. Introduction 5 • to reduce maintenance and improve reliability by replacing a conventional brush-type PM dc motor with a BLDC motor; • to compare sinusoidal excitation and trapezoidal excitation of the motor in terms of efficiency; • to maximize the circuit efficiency over the full range of operation; • to eliminate the motor shaft position sensor thus simplifying the motor terminal wiring; • to develop a robust maximum power tracking algorithm which keeps the PV array operating near the maximum power point (MPP) and which can tolerate sudden radiation changes; • to evaluate the performance of the BLDC motor drive and compare it with con-ventional permanent magnet brush-type dc motor drives. The scope of this thesis work focuses on the maximization of system efficiency, and includes selection of a suitable system configuration, motor control methods (e.g. trape-zoidal excitation and position-sensorless operation), loss analysis and MPT realization. Improvements to the PV array installation (for example, sun tracking) and to the pump are not in the scope of this thesis work. Hardware and software developments for the PV pumping system are described. The effectiveness of the optimization is verified by experiment. Chapter 2 Design Considerations In this chapter, some important design considerations in the optimization of the PV pumping system are presented. These considerations include the selection of a suit-able system configuration, the control of the brushless dc motor drive, the converter-commutator circuit, maximum power tracking realization and loss minimization. 2.1 System Configuration In this section, the existing PV pumping configurations are examined first. Then ma-jor components of the experimental PV pumping system are selected. Finally a new PV pumping configuration is presented. Some distinctive features of this configuration include a combined converter-commutator structure, position-sensorless operation of a BLDC motor and a robust MPT realization. 2.1.1 Existing Pumping Configurations A functional block diagram representing most PV pumping systems is shown in Fig-ure 2.2. The blocks with broken lines represent the optional functions that exist in some systems and are absent in other systems. As the output power of a PV array rises in proportion to the radiation, the out-put power of a given array area can be substantially boosted by using concentrator and sun-tracking mechanisms [4]. However, concentrator and sun tracking mechanisms re-quire extra components such as heat sinks and servo systems and they also consume 6 sun tracking PV ARRAY concentrator battery i y converter or inverter MOTOR mechanical transmission buffer tank control circuit water level detector 3 to to <§ ? CL s cn Figure 2.2: Configurations of Most PV Pumping Systems Chapter 2. Design Considerations 8 power. They are more suitable for high power rating PV systems than for small pumping applications. Storage batteries can be used to match the load with the PV array. But due to the need for maintenance and the loss in batteries and also due to the development of modern high speed, low loss semiconductor switching devices (SCR, GTO, power BJT, MOSFET and GIBT), the use of batteries should be avoided [4] [5]. Power conditioning required in a PV pumping system is mainly continuous impedance matching between the PV array and the load. This matching can be achieved by using switching converters. The converter, by varying the load impedance appearing at the PV array terminals, can vary the operating point of the PV array. However, a converter causes extra losses (switching loss and conduction loss of the electronic switches) and some control circuitry is needed besides the converter in order to accommodate MPT. Generally speaking, using a converter and MPT is more desirable when a positive displacement pump is used because the load characteristic of a positive displacement pump does not directly match the PV array [9]. Use of mechanical transmission should be avoided whenever possible because it causes energy loss and may reduce reliability. 2.1.2 Selection of System Components Motor Various motors have been used in PV pumping systems: three-phase squirrel cage in-duction motors [8, 19], series field dc motors [5], shunt dc motors [2], brushless dc (also called commutatorless or electronically commutated) motors [20]. Induction motors have the advantages of low cost and robustness and require no maintenance. However, their efficiency at small power rating (below 1 kW) is noticeably lower than that of permanent Chapter 2. Design Considerations 9 magnet dc motors. Besides, the control of an induction motor is more complicated than that of a PM dc motor. A combined motor-converter efficiency of 43% was reported in [8]. Dc motors are most often used in PV pumping systems because a PV array is a dc power source. The most common dc motor used in these applications is of the perma-nent magnet type. Among 64 PV pumping systems tendered for the UNDP/World Bank project, 54 used permanent magnet dc motors [1]. A permanent magnet dc motor has high efficiency due to the absence of field current. Another advantage of such a motor is the simplicity of control. The disadvantages of this type of motor are that the me-chanical commutator requires regular maintenance and that it is more expensive than an induction motor. With MPT, a combined motor-converter efficiency of 80 to 85 percent was achieved [9, 50]. Brushless dc motors are a relatively new addition to the motor family. Recent de-velopments in solid state devices and in rare earth magnetic materials have resulted in the availability and widespread use of this kind of motor. A BLDC motor, an inverted version of a conventional PM dc motor, maintains the advantages of a dc motor, i.e., high efficiency and ease of control. Moreover, the brush contact is eliminated so that it has a longer life and requires no maintenance. The price paid for this advantage is that an electronic commutation circuit must be added. Although nine PV pumping systems using BLDC motors were tendered for the UNDP/World Bank project [1], little infor-mation about BLDC motor applications in PV pumping systems is available except in [22, 20]. Due to their robustness, low maintenance and high efficiency, BLDC motors are promising in PV pumping applications. Since these advantages are among the original aims of this thesis research, a BLDC motor is selected for the experimental pumping system. Chapter 2. Design Considerations 10 Pump Centrifugal pumps are often used in PV pumping systems, especially when the water heads are below 15 meters. The best inferred peak efficiency of single stage centrifugal pumps tested was 65 % [1]. However, as the pumping head increases, the efficiency decreases. Progressive cavity pumps are suitable for medium and high water head applications. Impbed pump efficiencies of over 70 % at 49 meter head have been recorded in field tests in Egypt [1]. Other pumps suitable for high head applications are jack pumps and piston pumps. These pumps have pulsating torque characteristics [6]. The disadvantage of a progressive cavity pump is that, since it requires relatively high starting torque, a converter is usually needed, since this type of pump has been used in many diesel pumping systems, a PV-powered electrical subsystems accommodating a progressive cavity pumps may find application in replacing those diesel engines. Because of its high efficiency and wide application prospect, a Mono progressive cavity pump is chosen for the experimental pumping system described in this thesis. Power conditioning For maximum power from the PV array, power conditioning is needed for most situations. The power conditioning in a PV pumping system is basically impedance matching which can be carried out by a switching converter. For instance, the effective load emf and impedance appearing at the PV array terminals can be varied by varying the duty ratio of a buck converter which functions as a dc transformer in steady state. With a switching converter incorporating a proper control scheme, maximum power can be extracted from the PV array constantly. Different methods of obtaining MPT of the PV array have been reported. Most of Chapter 2. Design Considerations 11 them can be roughly categorized into two groups: one uses voltage feedback (often called "voltage tracker") [8], the other uses power feedback (often called a" power tracker") [2, 50]. In a voltage tracker, the PV array voltage is fed back and compared with a set point voltage. The control function keeps the PV array voltage following the set point voltage. One drawback of this scheme is that, since the optimal set point voltage varies with radiation and temperature, the PV array cannot always operate on its maximum power point. In a power tracker, the steering direction of the duty ratio depends on the change in PV array output power. A small change in duty ratio is applied after which the power change of the array is measured. If the power has increased, the duty ratio keeps changing in the same direction. If the power has decreased, the duty ratio changes in an opposite direction. Thus the PV array keeps operating around the MPP. One problem with this scheme is that, when thick clouds suddenly cover the sun, the radiation can drop quickly. In this situation, the output power of the PV array keeps decreasing regardless of the small change in duty ratio. When the radiation settles down at a lower level, the duty ratio is still stepping back and forth around the old value (for high radiation). The PV array and the load are severely mismatched and the output power from the PV array is very small. The system may even be forced to a stall. Another method to match a PV array and its load is to switch the combination of the series and parallel panel numbers either manually or electronically. This method is not effective when the total number of PV panels is small. In the past few years, microprocessors have been increasingly used in PV pumping systems. The use of a microprocessor enables a designer to build some " intelligence" into a PV pumping system. One example is a hybrid control scheme developed at UBC [41] which is basically a voltage tracker with a maximum power tracking function turned on and off under program control. Chapter 2. Design Considerations 12 A double-loop structure is developed for the experimental pumping system to realize maximum power operation. It is based on the similar idea as that of the hybrid control method — to exploit the advantages of both voltage trackers and power trackers. Details of this MPT realization are described later in this chapter. 2.1.3 The Selected P V Pumping Configuration The configuration of the experimental pumping system is shown in Figure 2.3. In order to reduce maintenance requirement and maintain high efficiency, a brushless PM motor is selected. The converter-commutator in Figure 2.3 performs two functions: a buck converter function and a commutator function. The buck converter function is designed for the maximum power tracking of the PV array. The commutator function is required by the operation of the brushless dc motor. The purpose of combining the two functions into one circuit is to reduce the number of serial switching elements (MOSFETs) in the power circuit and to increase the circuit efficiency. A microprocessor is used as the main control unit for the PV pumping system. It is used to implement maximum power control and other functions such as diagnosis of faulty conditions. The PV array is formed of a number of Solinex NSL5925 panels (rated 16 volts, 32 watts each). Because the power rating of the array is less than one horsepower, a sun-tracking mechanism is not economic and hence not considered. Batteries are not used due to their maintenance requirement and relatively low effi-ciency. A Mono progressive cavity pump is used in the experimental pumping system, partly due to its good efficiency at a wide range of pumping heads and partly due to its widespread use in existing pumping systems using other energy sources. The water SUN PV ARRAY A/D I I A/D MC6809 Microprocessor Development System HEXFET COMVERTER-COMMUTATOR I D/A Brushless PM Motor CONTROL ELECTRONICS Progressive Cavity Pump Figure 2.3: Selected PV Pumping Configuration Chapter 2. Design Considerations 14 output will be stored in a buffer tank. Compared to other PV pumping system configurations, this new configuration has the advantage of reduced maintenance due to the use of a brushless PM motor and the absence of storage batteries. The microprocessor-based MPT realization is effective and robust. The efficiency is comparable to that using a brush type PM dc motor. 2.2 Brushless D C Motor Drive This section presents the considerations of motor control and efficiency maximization of the brushless dc motor drive. Terminology of brushless PM motors is defined; sinusoidal excitation and trapezoidal excitation are compared; losses in the motor and commutator are analyzed; and position-sensorless operation is presented. 2.2.1 Terminology The terminology for brushless PM motors used by various authors differs. The term brushless PM motor used in this thesis refers to a motor with a permanent magnet rotor and a polyphase wound stator. The term brushless PM motor drive used in this thesis refers to a brushless PM motor with an inverter or electronic commutator and some other control circuitry. According to its supply voltage waveform, a brushless PM motor drive can be cat-egorized as a trapezoidal type or a sinusoidal type. The motor used in a sinusoidal brushless PM drive requires a sinusoidal voltage and current supply and the motor back em/is sinusoidal. This type of drive is usually analyzed with synchronous machine the-ory. A trapezoidal type of brushless PM motor drive requires commutation (in the most common three-phase wye-connected system) every 60 electrical degrees. This results in a trapezoidal line to line voltage and a rectangular line current. Ideally the motor Chapter 2. Design Considerations 15 emf should be trapezoidal. However, due to some manufacturing complications associ-ated with the production of a trapezoidal emf, most brushless PM motors designed for a trapezoidal control scheme exhibit em/waveforms that are much closer to sinusoids than trapezoids [24]. A trapezoidal brushless PM drive is analyzed with dc machine theory. A sinusoidal brushless PM drive provides smooth torque control and accurate position control. However, the sinusoidal control scheme is considerably more complicated than the trapezoidal control scheme. In this thesis, the term brushless dc (BLDC) motor refers to a brushless PM motor with an electronic commutator (functionally equivalent to a brush-type PM dc motor). The term brushless dc motor drive refers to a BLDC motor with some kind of control function, such as speed control using a buck converter. 2.2.2 Trapezoidal Excitation and Sinusoidal Excitation A brushless PM motor can be powered with a sinusoidal voltage source or a trapezoidal voltage source. These two excitation methods are investigated in terms of efficiency. For sinusoidal excitation, the inverter is modulated by a PWM control signal gener-ated by a 68HC11 microcontroller. A table look-up method is used in the generation of the PWM signal. In the experiment, the motor voltage is adjusted to be in phase with the motor current so that best efficiency for sinusoidal excitation is measured. For trapezoidal excitation, the bottom three MOSFETs of the inverter are modulated by a PWM signal generated by hardware. At any moment, only two of the three phase coils are connected to the power supply. The emf of the brushless PM motor used in the experiment has sinusoidal emf waveforms. In order to compare the two different excitation methods, the following assumptions are made: Chapter 2. Design Considerations 16 1. For both methods, the speed and output power are the same. 2. For sinusoidal excitation, the current and voltage are in phase (unity power factor). 3. For trapezoidal excitation, current waveforms are approximated as rectangles. 4. For sinusoidal excitation, the modulation depth is 1. 5. For trapezoidal excitation, the duty ratio is very close to 100%. For sinusoidal excitation, the total power output is three times that of the per-phase output and can be calculated from the following equation: Ptin = 7r~ I Em sinut Im sinwt d[wt] = ^ ^ rnEm (2.1) 2TT JO 2 For trapezoidal excitation, the total power output can be calculated by averaging the power output within a 60 electrical degree interval: PtraP = - f2 [iAeA + i B e B + iCec)d[wt) (2.2) The motor current and em/waveforms shown in Figure 2.4 indicates that at any moment only two phases are carrying current and the current in the other phase is zero. The total power output is the sum of the output from the two conducting phases and can be estimated from the following equation (see Figure 2.4): PtraP = ~ /f [Ia E^sinwt - Ia Emsin(wt - ^ ) ] d[ut] = Z s / * I a E m (2.3) where J m , Em are the peak phase current and emf respectively; Ia is the average value of the equivalent dc motor armature current; u> is the electrical angular frequency depending on motor speed. The rms currents in sinusoidal and trapezoidal excitations are: Isrm. = % (2.4) Chapter 2. Design Considerations 17 Figure 2.4: Current and emf Waveforms for Trapezoidal Excitation Chapter 2. Design Considerations 18 iTrms = y | / a (2.5) Based on equations 2.1, 2.3, 2.4, 2.5, the ratio of the two rms currents for a given output power is: ^ l i = 1 = 1.047 (2.6) Equation 2.6 indicates that, for a given output power and speed, sinusoidal excitation would have an rms current about 5% lower than trapezoidal excitation. Since copper loss is proportional to the square of the rms current of the conductors, the copper loss ratio for the two methods can be shown as: = copper loss trapezoidal = ^ = ^ copper loss sinusoidal It can be seen that, for a given power output, the copper loss in trapezoidal excitation is about 10% higher than that in sinusoidal excitation. Nonetheless, the switching loss in trapezoidal excitation is much lower than that in sinusoidal excitation. For the two cases, switching losses are estimated according to MOSFET manufacturer's handbook (see [21] pp. 1-135-49; a basic formula is represented in form 2.8). For a load current of Ir, and a supply voltage of V^., the corresponding switching loss is estimated by: IK T T P.,... = V+Hf. (K, + ^ + 0 . 5 ^ ) (2.8) where / , is the switching frequency in Hertz, Kq is the charge constant of the power MOSFET (/iC/Amp) and is provided by the manufacturer; di/dt is the current rise rate in ampere per microsecond and can be estimated through an oscillogram of the MOSFET current. Chapter 2. Design Considerations 19 For trapezoidal excitation, the following relations stand: h = h (2.9) Ea 0.954V3 Em  V&~-f = £ (2.10) where Ea is the equivalent dc motor emf can be related to the phase emf oi the brushless PM motor as shown in relation 2.10 [24]; 8 is the duty ratio of the converter function and is assumed to be very close to unity (assumption number 5). From formula 2.8 and relations 2.9,2.10, switching loss in trapezoidal excitation can be estimated as: \2KqIa PT.LO.. = l . « 5 ^ A / . ( j r f + ^ + 0 J S ^ ) (2.11) where PTIIOU is the switching loss in trapezoidal excitation in micro-Watt. For sinusoidal excitation: Vdef*2Em (2.12) IL = \/2ISrmt sin(ut) (2.13) J2/Zla Substituting relations 2.12,2.13 and 2.14 into 2.8 yields: Ps,io»»(wt) = 2.2Em- f, • Ia[Kqsin(u>t) + sin(ut) 0.557o sin2(u>t) 2.2Kq Iasin(u>t) \ di/dt ] (2-15) di/dt where psilon is the instant switching loss for sinusoidal excitation. The average switching loss for sinusoidal excitation can be obtained by Equation 2.15 averaged over 180 electrical degrees and the result multiplied by 3 (phases): Chapter 2. Design Considerations 20 Ps,io„ = — / [Kqavn(wt) 7T JO +sin(wt) 2.2KqIasin{wt) l.llasin2(u>t) \\ di/dt 2di/dt J lU,J . i ^ w + M T ^ ^ + M M _ A _ , ( 2 . 1 6 ) Based on Equations 2.16 and 2.11, the switching losses for the two excitations are calculated using some assumed parameters: Em = 58V, / , = 6A5kHz, Kq = 0.7pC/A, di/dt = lOOA/pS The result is listed in Table 2.1 for comparison. Table 2.1 shows that the switching loss for sinusoidal excitation is more than two times that in trapezoidal excitation. Two factors contribute to the higher switching loss for sinusoidal excitation. One factor is that at any moment all of the three legs are switching at frequency in trapezoidal excitation, only one leg is switching at frequency Another factor is that, for a given Em, the required dc supply voltage V, for sinusoidal excitation is higher than that required for trapezoidal excitation. Friction, windage and iron core 1 losses for the two excitation methods are considered to be the same because of the assumptions of equal speed and output power. Based on the above loss analysis, the efficiency of the two excitation methods can be compared. For instance, 10% copper loss is 3.9 Watts when motor current is 5 Amperes (equivalent armature resistance: 1.55 Ohm), and the switching loss for trapezoidal exci-tation is 4.5 Watts lower than that of sinusoidal excitation (see Table 2.1). As will be shown in Chapter 5, laboratory tests show that trapezoidal excitation is slightly more 1iion core loss is lumped into "friction and windage loss" in the rest of the thesis. Chapter 2. Design Considerations 21 Table 2.1: Switching Loss For Sinusoidal and Trapezoidal Excitation RMS current (A) 1 2 3 4 5 6 7 8 Trapez. loss (Watt) 0.510 1.084 1.706 2.365 3.058 3.782 4.534 5.313 Sinus, loss (Watt) 1.282 2.719 4.262 5.894 7.603 9.384 11.23 13.14 efficient than sinusoidal excitation. Since sinusoidal excitation gives no advantage in efficiency and is more complicated, trapezoidal excitation is selected. 2.2.3 B L D C Motor Parameters and Losses A two horse-power, three-phase, twelve-pole, surface magnet brushless PM motor is used in the experimental P V pumping system. Since a trapezoidal control scheme is chosen, an electronic commutator is required to run the motor. A simplified circuit diagram of an electronic commutator is shown in Figure 2.5. In Figure 2.5, the right-hand side of the dashed line A-A' is equivalent to a conven-tional (brush type) PM dc motor. Like a conventional PM dc motor, the speed of a brushless dc motor can be controlled by varying the supply voltage or by a dc-dc con-verter. Some parameters of a brushless dc motor can be defined as follows [24] (SI metric units are used): Voltage and torque constants: v 0.954EMFpeak K e = — a T -Kt = Ke (2.17) (2.18) where EMFpeaj, is the peak value of motor line to line em/, Q m is the steady state angular speed of the motor. Chapter 2. Design Considerations 22 Right-hand side of A-A' is equivalent to a dc motor la Power supply A' SW1 SW3 SW4 SW5 B Brushless PM motor Electronic commutator Figure 2.5: Simplified Electronic Commutator Circuit with the Brushless Motor The friction and windage torque coefficient is determined experimentally and is curve fitted to the following form (using minimum square error criterion): Kf = 0.0007 + ° ' 1 8 4 1 20 < <vm < 200 rad./sec. (2.19) where u>m is the instant angular speed of the motor. The curve-fitted Kf and measured Kf are illustrated in Figure 2.6. Equivalent armature resistance and inductance: Ra = 2Rpha,e + 2J?u((m) (2.20) £ a = 2 ( L p W - M ) (2.21) Chapter 2. Design Considerations 23 0.0100 0.0000H 1 1 1 1 0 50 100 150 200 250 Motor angular speed ( Rad./sec. ) Figure 2.6: A Plot of Measured Kf and Curve-fitted Kf where RphaieiLphase are the phase resistance and self inductance of the motor winding, M is the mutual inductance between any two phase coils, RD(on) is the on resistance of the power MOSFET(s) for one switching element. Torque ripple frequency: where Np is the number of poles of the motor. The parameters of the brushless PM motor used in this thesis project are shown in Table 2.2. Chapter 2. Design Considerations 24 Table 2.2: BLDC Motor Parameters Description Symbol Value Unit Rated power Prate 1.5 kilowatt Rated torque Tr<1te 4.5 N • m Voltage constant Ke 0.52 volt • sec./rad. Torque constant Kt 0.52 N • m/Ampere Phase resistance Pphate 0.5 ± 5% Ohm Phase inductance Lphase 1.2 mH Thermal resistance RT 0.67 °C/Watt Moment of inertia J 0.008 kg • m 2 Friction-windage const. Kf 0.001-0.008 (varies with wm ) JV-m Rad./tec. The losses in a brushless dc motor are similar to the losses in a brush-type PM dc motor except that the former has two switching elements in series with the armature. The series switching elements (MOSFETs) result in the copper loss of a brushless dc motor being slightly larger than that of its brush type equivalent. The copper loss of the motor can be calculated by: Plc = ll{rmt)Ra (2.23) Friction and windage losses of the brushless dc motor are experimentally determined and are shown in Figure 2.7. While the friction and windage losses of a BLDC motor are inherent, the copper loss can be minimized by reducing the current ripple and RD(<m)--Ri?(on)can be reduced by MOSFET paralleling at the cost of extra MOSFETs. The ripple in the motor current depends on such factors as motor emf (trapezoidal or sinusoidal), load current and A motor with trapezoidal emf has less ripple than that with sinusoidal emf. The higher the load current, the lower the ripple. It is not desirable to put filtering inductors in series with the armature because this will add to the armature inductance and shift the phase angle of the armature current, resulting in reduced output Chapter 2. Design Considerations 25 0 5 0 0 1 0 0 0 1 5 0 0 2 0 0 0 2 5 0 0 Motor speed ( RPM ) Figure 2.7: Friction and Windage Losses of the Motor torque at high speed. As will be described in Section 2.3, the switching frequency of a buck converter incorporated into the BLDC motor also affects current ripple. Current ripple and thus copper loss decrease when the switching frequency increases. However, since a higher frequency leads to a higher switching loss, the switching frequency should be selected to minimize the total losses. 2.2.4 Position-sensorless Operation The commutation of a brushless dc motor is based on the detection of the instantaneous rotor position. Conventionally, the rotor position is detected by some kind of rotor po-sition sensor. For example, in a three-phase brushless motor, three Hall sensors and a magnetic interrupter can produce a 3-bit position signal representing 6 different intervals Chapter 2. Design Considerations 26 Table 2.3: Relation Between On-switches and Rotor Position on-switches 3,4 4,5 5,6 6,1 1,2 2,3 active phases +B,-A +C,-A +C,-B +A,-B +A,-C +B,-C rotor position 3 4 5 6 1 2 in one electrical cycle. However, the use of a position sensor requires extra wire con-nections between the motor and the control unit. This requirement is a drawback when the motor is not close to the control unit, as when a submersible motor is placed at the bottom of a well and the control unit is installed on top of the well. Nonetheless the rotor position signal of a brushless dc motor can be obtained by filtering the motor terminal voltages and comparing the filtered voltages to a neutral point. Compared with other methods of deriving a rotor position signal, such as third harmonics method and numerical sampling method [15], the selected method is simple and reliable. Since the power supply (PV array) is a nonlinear, varying source, a minimum starting power must be satisfied prior to motor starting process. The use of position-sensorless operation with a PV array has not been found in the literature. 2.S The Converter-commutator Circuit The operation of the electronic commutator (see Figure 2.5) can be outlined as follows: At any moment one top and one bottom MOSFETs are on, connecting two of the phase windings in series to the power supply. The switches that should be on at a given moment depend on the instantaneous rotor position. The relationship between rotor position and on-switches is shown in Table 2.3. The motor terminal voltage waveforms are illustrated in Figure 2.8(a). For a buck converter er 2. Design Considerations \ k ^ . and V c are motor terminal voltages with reference to ground 0 V B 0 VC 0 cor cor cor (a) Motor terminal voltages without PWM (all voltages are with reference to ground) V A 0 V B 0 vc 0 cor Ui KJ COf COf (b) Motor terminal voltages with PWM (all voltages are with reference to ground) Figure 2.8: Brushless DC Motor Terminal Voltage Waveforms Chapter 2. Design Considerations 28 function to be incorporated into the commutator, a pulse width modulation (PWM) sig-nal is ANDed with the gating signals for the lower three MOSFETs. As a result, the motor terminal waveforms are "chopped", and the average value of the equivalent arma-ture voltage is reduced by a factor equal to the duty ratio of the PWM signal. The motor terminal voltage waveforms with pulse width modulation are shown in Figure 2.8(b). The equivalent impedance appearing at the left hand side of A-A' (in Figure 2.5) can be ad-justed by varying the duty ratio of the PWM signal. This is equivalent to a conventional PM dc motor fed by a buck converter shown in Figure 2.9. • Figure 2.9: Equivalent Circuit for the BLDC Motor with the Converter-commutator Circuit Ea in Figure 2.9 is the equivalent average back emf oi the brushless motor; La and Ra are the equivalent armature inductance and resistance respectively. C is a capacitor Chapter 2. Design Considerations 29 used to accommodate the buck converter function. Based on the equivalent circuit in Figure 2.9, the averaged state-space equations for the brushless dc motor drive can be obtained by using the state averaging technique [13]: dv, i,(va) — Sia dt C dia _ Sv„ — Rglg — Keu)m ~dt ~ Lg dwm Ktia — KfU>m — TL (2.24) (2.25) (2.26) dt J where 8 is the duty ratio of the converter, u>m is the motor angular speed, R~e and Kt are voltage and torque constants respectively; Kf is a friction and windage constant; J is the moment of inertia, and TL is the load torque which is assumed to be constant in this application. The system is nonlinear because of the nonlinear power supply and the cross product terms 8vt and 6ia . Besides Kf varies with mechanical speed wm. Letting all the derivatives be zero leads to the following steady-state model of the brushless dc drive: V. = g{It) (2.27) / . = £>/„ (2.28) DV, - RaIa - Kenm = 0 (2.29) KJa - KfSlm - T L = 0 (2.30) where D,Vt,It)Ia and f i m are the steady state values of 8,vt,i,,ia and u>m. Equation 2.27 is the I-V curve of the PV array. D V, in Equation 2.29 can be consid-ered as the armature voltage of the equivalent dc motor: Va = DV, (2.31) Chapter 2. Design Considerations 30 Equations 2.28 and 2.31 show that the converter acts like a dc transformer in steady state. For a given armature current /„ , the PV array current J, can be varied by varying the duty ratio D. As I, varies, V, also varies according to the I-V curve of the PV array. The maximum power point is reached if: K H (2-32) is satisfied (assuming temperature and radiation are constant). Pt = V,Jt is the output power from the array. In real-time operation, maximum power voltage can be searched by examination of the partial derivative | ^ (discussed in Section 2.4). 2.4 Maximum Power Tracking Realization 2.4.1 The Need for Maximum Power Tracking The I-V characteristics of a typical PV array is displayed in Figure 1.1. Figure 1.1 shows the dependency of the I-V characteristic upon radiation and temperature. During the operating period in one day, solar radiation may vary from zero to 1 kW/m2 and the temperature may also drift in a range of 10 to 20 degrees Celsius or more. For a fixed temperature, the maximum power voltage increases logarithmically with the increases of the radiation. For fixed radiation, the maximum power voltage decreases linearly with the increase of the array temperature. The array temperature is governed by an approximate equation as follows [55]: Tarray(°C) = Tambient{°C) + 30 x Intensity(kW/m2) (2.33) During the day both the radiation and array temperature vary. The total effect on the maximum power voltage of the array over a sunny day was simulated with a computer Chapter 2. Design Considerations 31 program (see Appendix A). The results show that, on a sunny day with 20 °C temperature deviation, for a single crystal silicon array the maximum power voltage remains relatively constant during the operating hours. The radiation and temperature data recorded on a sunny autumn day in Vancouver were used to estimate the maximum power voltage deviation in one day. The results show that the maximum power voltage varies ±2.5% of the array open circuit voltage. If a fixed voltage is maintained, the output power is only 0.5 % less than that with the ideal maximum power tracking. Therefore the voltage feedback scheme is satisfactory for single-sunny-day operation. For different seasons of the year, the average temperature in a location varies considerably. The seasonal weather changes affect the day-average value of the maximum power voltage (this can be seen from a simulation detailed in Appendix A). According to the simulation, the maximum power voltage on a winter day can be 11% higher than that on a summer day. If the array voltage is controlled at a proper value for the whole year, the power wasted due to non-maximum power operation can be 2.7% on a winter day. For the areas where the seasonal temperature varies in a wide range, the power loss due to non-maximum power operation cannot be overlooked, and a maximum power tracking mechanism is worthwhile. 2.4.2 A Double-loop Maximum Power Tracking Scheme Maximum power tracking in a PV powered system can be considered as an extremum control problem [14]. The power feedback scheme reviewed in Chapter 1 is a possible solution. However, this scheme may suffer from a stability problem when the solar radiation changes suddenly, as may happen on a cloudy day. Incorporating advantages of both voltage feedback and power feedback schemes, a double-loop maximum power tracking scheme is developed for the experimental pumping system. The block diagram of the double-loop maximum power tracking scheme is shown in Chapter 2. Design Considerations 32 Figure 2.10. From Figure 2.10 it can be seen that there is an inner voltage control loop and an outer extremism control loop. The voltage loop keeps the PV array voltage Vt close to the set point voltage Vr and suppresses the effect of the disturbance in load or in the power supply, e.g., when radiation changes quickly in a cloudy day. The voltage set point of the voltage loop is periodically adjusted by the outer loop. Suppose the output power of the PV array (seen by the microprocessor) is given as: P. = f(V.).+ e{t) (2.34) P, is the output power from the array; f(Vt) is a nonlinear function with single maximum in the interval 0 < Vt < V .^; V .^ is the open circuit voltage of the PV array, and e(t) represents the noise due to current ripple, measurement error and the changes in radiation and temperature. As described in Section 2.3, the PV array voltage V, can be varied by varying the duty ratio D. Therefore a voltage control loop can be implemented with the microprocessor. Integral control was used in the voltage loop as shown below: D(k + 1) = D(k) - K^VrU) - vt(k)} (2.35) where D(k + 1) and D(k) are the new and old duty ratio values respectively, Vr(j) is the new reference voltage for the fast voltage loop, Va(k) is currently measured array voltage and Kj is an integral constant. With the voltage loop implemented the steady state voltage of the PV array can be varied by changing the set point voltage VT. Since the goal is to maximize the average output power, the extremum control is based on steady state values. If the noise term e(t) is ignored, the first order partial dPQ^ should indicate the correct direction for Vt to move. When = 0 , the system is at the optimal operating condition. An extremum control algorithm based on the sign of | r f was chosen for the PV Is(k) Vs(j) uIsG) Update timer A/D Vs(k) Extremum Control Loop Ps(j)=Vs(j)*Is(j); V r ( j)=VrO-l) + AV*sign{J^-)^:-1-)-} Vs(j)-Vs(j-1)' Vs(k) A/D Vr(j) Voltage Loop D(k+l)=D(k)-Ki* [Vr(j)-Vs(k)] D(k+1) D/A Fault diagnosis; Sensorless starting; Data recording; Displaying MICROPROCESSOR SYSTEM signal conditioning signal conditioning VS signal conditioning SUNLIGHT PV ARRAY 3 * Converter-commutator circuit I J5 Brushless PM motor +TL,a>m Progressive cavity pump Water output ^ 8 •8 to I-§ 2. CL e-+-f x . S Figure 2.10: Block diagram of maximum power tracking scheme co oo Chapter 2. Design Considerations 34 pumping system: Vr(j) = Vr(j - 1) + AV • sign{ ™ ~_ ~ *) } ( 2 . 36) where Vr(j) and VT(j — 1) are the new and old reference values to the voltage loop, AV is the step of voltage change. The direction of the voltage change (increasing or decreasing) is determined by gradient |^-. The resulting operation of the extremum loop resembles the activity of "hill-climbing". It is very important to select correct values for the update rate and voltage step AV in Equation 2.36. The update rate should be slow enough so that the system has reached steady state when update occurs. On the other hand, the update rate should be fast enough so that the changes in radiation and temperature are insignificant during one update period. It takes less than a second for the voltage loop to reach steady state. On a clear day, it takes about 10 seconds for solar radiation to vary 0.1% of its maximum value (lkW/m2). Therefore, it is safe to select an update period between 1 to 10 seconds. To ensure the correct operation of the extremum control algorithm (Equation 2.36), the voltage step AV should be large enough to cause a power change greater than |2e(i)|, or: \f(Vt + AV)-f(Vt)\>\2e(t)\ (2.37) For a given amount of power change, the required voltage step A V varies with different arrays. A PV array with a larger fill factor 1 requires a smaller AV, and vice versa. In order to achieve optimal performance, the noise in Equation 2.34 should be made as small as possible. The noise is caused by a number of factors, such as the switching of the power MOSFETs, the current ripple caused by the motor, digitization error and changes in radiation and temperature. defined in [54] page 322. Chapter 2. Design Considerations 35 In practice, the noise term e(t) can sometimes be the major factor that causes the change of the PV array power. This phenomenon occurs when the solar radiation changes suddenly on a cloudy day. Since the extremum control algorithm is based on the measure-ment of , it cannot find the maximum power point under such a condition. However, the voltage control loop is still functional and will maintain the P V array voltage at about the old value. Since changes in radiation often cause insignificant changes in maximum power voltage, the PV array is likely to continue to operate near the maximum power point. 2.5 System Losses and Efficiency In order to achieve maximum efficiency, it is necessary to analyze the losses in different parts of the system and to minimize the losses. The analysis is focused on the converter-commutator circuit and the brushless PM motor. The losses in the pump and the PV array are largely determined by their manufacturers. 2.5.1 Maximum Efficiency from the Motor Since the motor operates with a roughly constant torque load (progressive cavity pump), the motor current is basically constant for a given water head. Therefore the copper loss is constant regardless of the motor 6 p e e d . As the motor speed varies with the radiation during the day, the friction and windage losses also vary. As a result, the efficiency of the brushless dc motor is a function of operating speed or a function of equivalent armature voltage V A . The output power can be calculated by multiplying Equation 2.30 by the motor speed fim (bearing in mind that power output equals TL • ftm)« From Equations 2.30, 2.29 and 2.31, and with input power being VA • J A , the efficiency of the brushless dc Chapter 2. Design Considerations 36 motor can be estimated by: Vm = 1-^x ~2Ra+^r)-I-^r (2>38) By calculating and letting it equal zero, the maximum-efficiency armature voltage can be found (note that Kf is a function of armature voltage): V.„ = + 2 * ° ° - 0 0 0 7 / ^ 2 ) ** ~ °-^^} (2.39) where Veff is the maximum-efficiency armature voltage, the values of Ke and Ra can be found in Table 2.2. For a given armature current Jai Veff can be estimated with Equation 2.39. For example, when /„ = 4.5A, Veff would be 85 volts according to Equation 2.39. Laboratory tests show that the maximum-efficiency voltage at the converter input for Ia = 4.5.4. is 90 volts. 85-volt is a close estimate since there is voltage drop on the equivalent converter. 2.5.2 Losses in the Converter-commutator Circuit and Motor The losses in the converter-commutator circuit mainly include switching loss, diode free-wheeling loss and conduction loss. Gate drive loss is very small compared to switching, freewheeling and conduction losses, and is therefore ignored. Two switching configurations for the converter-commutator are considered. The cir-cuit diagrams for these two configurations are shown in Figure 2.11 (a) and (b) respec-tively. The first configuration uses the internal diodes of the power MOSFETs (HEXFETs III from International Rectifier Corp.) for freewheeling. The advantage of this config-uration is that no external freewheeling diodes are required. However, the presence of cross-conduction (an inherent phenomenon of this type of converter and load) between a top and a bottom HEXFET requires fast recovery diodes. Although the reverse recovery time of the internal diode of a HEXFET III has been improved, it is still considerably Chapter 2. Design Considerations 37 QI Q4 Q5 B Brushless PM motor (a) Power circuit using HEXFET III internal diodes QI Q4 £ U i D F 1 ? ! j j ^ : D F 3 Q5j D3 D5 B DF5 Brushless PM motor i-h Q6 t-h Q2 i-H Bf ^ S f H 5 f (b) Power circuit using external freewheeling diodes Figure 2.11: Two Switching Configurations for the Converter-commutator Chapter 2. Design Considerations 38 longer than some fast recovery diodes. The slow recovery time of the internal diodes results in higher switching loss. This drawback can be overcome in the second configuration at the penalty of two extra diodes, one of which is in series with the motor windings, and causes extra loss. The switching loss of the converter-commutator circuit is a complicated function of supply voltage, load current, stored charge in the freewheeling diode and chopping frequency. It would be very complicated to compare the two switching configurations by theoretical analysis. In order to select a configuration from the two discussed earlier, the switching losses for the two configurations were measured and plotted in Figure 2.12. The measurement was done on one leg of the three phase converter with an R-L load. The supply voltage was 100 volts and the load current was 6 amperes for both cases. Figure 2.12 shows that the switching loss for the second configuration (using external fast-recovering diodes) is about 3 watts lower than that of the first configuration when the switching frequency is 20 kHz. However, the extra loss incurred by the series diode in the second configuration is about 4 watts. This experimental result indicates that the second configuration does not give a clear advantage in efficiency and it requires 6 extra diodes. The first switching configuration, therefore, is selected. The conduction loss of the MOSFETs and the motor plus internal diode freewheeling loss can be estimated with: Pca = SRolZw + (1 - 8)(^ll{rmt) + V * J . ( r m . > ) (2.40) Formula 2.40 shows that the conduction loss can be reduced if the current ripple and the equivalent armature resistance Ra are reduced. In order to reduce the current ripple, the chopping frequency should be increased. However, a higher frequency would lead to higher switching loss since switching loss is directly proportional to switching frequency. There is a frequency at which maximum Chapter 2. Design Considerations 39 efficiency can be obtained. With the HEXFET III switching configuration, the optimum frequency is around 6 kHz. In practice, a higher frequency can be used in order to eliminate audible noise. 2.6 S u m m a r y A new PV pumping configuration is selected for the experimental PV pumping system. This configuration accommodates efficient energy conversion and robust operation with a reduced maintenance requirement. Comparison of trapezoidal excitation with sinusoidal excitation shows that sinusoidal excitation has no advantage over trapezoidal excitation in terms of efficiency. The parameters of the BLDC motor are described and the losses discussed. The concept of position-sensorless operation is introduced. The ideas of converter-commutator combination and double loop MPT realization are presented. A loss analysis on each part of the electrical system is conducted and some guidelines on minimizing overall system losses are derived. Chapter 2. Design Considerations £ o = using FET body diodes ^ a = using external diodes w 8-1 Figure 2.12: Switching Loss for the Two Switching Configurations Chapter 3 Hardware Design The hardware design is discussed in this chapter. The guideline for hardware design is to maximize efficiency and reliability. The design can be roughly divided into two parts. The first part includes the MOSFET power circuit, its drive circuits and protection circuits. The second part includes motor control circuits, such as converter-commutator logic, microprocessor I/O interface circuit and control circuits required by position-sensorless operation. Some practical problems in implementation of the hardware design are also discussed in this chapter. 3.1 Power Circuit The configuration of the power circuit is shown in Figure 2.11(a). The detail of the power circuit is described in the following subsections. 3.1.1 Power M O S F E T s and Heat Sinks There are six power MOSFETs in Figure 2.11 (a), which represents the configuration of the power circuit. In the actual power circuit design, the current rating of the power MOSFETs are over-designed for two reasons: one is to limit the device temperature rise; the other is to reduce on-resistance (RD(<m)) of the switching elements. The reliability of a power MOSFET decreases with the increase of its junction tem-perature. As an example, Figure 3.13 illustrates the typical high temperature reverse bias (HTRB) failure rate of a HEXFET IRF740 from International Rectifier Corporation [21]. 41 Chapter 3. Hardware Design 42 Figure 3.13: Typical High Temperature Reverse Bias (HTRB) Failure Rate Figure 3.13 shows that, if the junction temperature of an IRF740 is lower than 55 ° C, the HTRB random failure rate is lower than 1 FIT (failure in 109). A higher rating power MOSFET has a lower on-resistance. A smaller on-resistance results in lower conduction loss. Based on these considerations, two IRF740s are con-nected in parallel to form one switching element in the actual design. The specification of IRF740 can be found in reference [21] (pp. C293-300). Since it is necessary to sample the PV array current for MPT function, 3 HEXSenses are used in the three bottom switching elements. IRC740s from International Rectifier Corporation are selected. The specification of IRC740 can also be found in [21] (pp. E25-26). As a result, the power circuit contains a combination of IRF740s and IRC740s. One leg of the actual power circuit is shown in Figure 3.14. Three 6" x 10" x 1" heat sinks are used for the three legs of the power circuit. Each Chapter 3. Hardware Design , + Vs from gate IRF7400-drive R g ri e Kg i n ^ 4 0 , £ n 10 from pin5ofIR2110A (VS) from gate drive IRF740 Rg IRC740 Cl C2 1 0 1 0 g r Rg To a motor terminal To current sensing circuit Figure 3.14: One Leg of the Power Circuit Chapter 3. Hardware Design 44 Figure 3.15: A Picture of the Converter-commutator Power Circuit leg can be easily tested or replaced. A l l H E X F E T s are electrically isolated from the heat sinks by thermal pads. The temperature of the H E X F E T cases is in the range of 26 to 30 °C when the power circuit is handling 300 watts of power. Figure 3.15 shows the power circuit of the experimental pumping system. 3.1.2 Capacitors and Layout C l in Figure 3.14 is used as one third of the storage capacitor in a buck converter (C in Figure 2.11(a) ). The capacitance of C l affects the voltage ripple due to chopping. The selection of C l can be based on the following formula: 1IL x Tch (3.44) where AVP is the P V array voltage ripple, Tch is the chopping period and Ii is the load current (assumed continuous and smooth). C2 is used to absorb any voltage spikes due to stray inductance in circuit layout. Chapter 3. Hardware Design 45 Power circuit layout is important because power MOSFETs are fast switching devices. The top and bottom HEXFETs should be mounted close together. Stray inductance can be minimized by keeping conduction paths as short as possible, by minimizing the area of current loops, and by using twisted pairs of leads. Power wires should be as far away from the gate drive stage as possible. The drain-source voltage and drain current waveforms of a bottom HEXFET are shown in Figure 3.16. The current spike at turn-on is caused by internal diode recovery. No significant spike is found in the voltage waveform. Upper trace: drain current 5A/div. Lower trace: drain-source voltage 50 V/div. Time scale: 5uS/div. Figure 3.16: Voltage and Current Waveforms of a Bottom HEXFET Chapter 3. Hardware Design 46 3.2 M O S F E T Gate Drives The performance of the gate drive of a power MOSFET is important to the behaviour of that MOSFET. One misdesigned gate drive stage may lead to the destruction of the corresponding MOSFET and even some other MOSFETs in the power circuit. The switching speed of a power MOSFET is determined mainly by its gate drive stage. Fast switching of a MOSFET results in lower switching loss. However, since fast switching causes high cross conduction current (internal diode reverse recovery current), trade-off between reliability and efficiency has to be made. The maximum transient current for IRF740 is 40 amperes. Since the PV pumping system is designed for a long lifetime, the actual design allows only 20 amperes peak of cross conduction current. The switching speed of the MOSFET can be varied by varying the gate resistance (between gate drive output and MOSFET gate). Since the source potential of a top HEXFET changes between zero and the supply voltage, its gate drive must be isolated from ground. Several methods can provide gating for the top FETs. For instance, the use of opto-couplers and of pulse transformers. After some design and experimental effort, a drive IC chip IR2110 from International Rectifier Corporation was chosen for the gate-drive stage. Some important features of IR2110 are: • 500V rated floating supply offset voltage • lOV/ns dv/dt immunity • 2A peak output current capability per channel • 100 ns propagation delay time • 10 to 20V output drive operating voltage range Chapter 3. Hardware Design 47 The schematic diagram of the gate drive stage is shown in Figure 3.17. The key compo-nent in the gate drive stage is IR2110. 'LIN' and'HIN' are the input pins for controlling +12V RI C3 lOOuF o—H—vVW^-f—ill 10 To source of the top FET • 4001 *SW1 O *SW4 O 4001 OVERCURRENT CI luF VB HJN VS HO jLI| S D IR2110A L2 LIN LO VSS VDP VCC luF 1 1 T R2 47 . R3 47 -tr C4 lOOuF +12V To gates of the top FETs To gates of the botton FETs Figure 3.17: HEXFET Gate Drive Schematic Diagram the bottom FET and the top FET respectively. The 10 ohm resistor in the gate of each FET prevents oscillation between the two parallel FETs. With the gate resistance shown in Figure 3.17, the cross-conduction current for the two paralleled FET6 are less than 40 amperes (half of the rated maximum). The current rise time is about 300 nS. 'OVERCURRENT' is a signal from over-current detection circuit. Whenever over current occurs, the lower FET is turned off so that the motor current is limited. C l , a tantalum capacitor with low stray inductance, and C2 should be placed as close as possible to Chapter 3. Hardware Design 48 IR2110. Shielded wires are used for the connection between the gate drive output to the gate resistor. 3.3 Over-current Protection Over-current protection is done with hardware because of the response speed required by the protection. The schematic diagram of the protection circuit is shown in Figure 3.18. The current limit level can be adjusted with a potentiometer in the circuit. The circuit works on a cycle by cycle basis. The three bottom HEXFETs are turned off for the rest of the cycle whenever the current exceeds preset limit; they are turned on again when the next cycle starts. Since the HEXFETs switch at high speed, the op-amp used in the over-current protection circuit should have a high slew rate. current signal : RI ioo O-HVW^ -VDD O—i R2 5k MC34072 VDD o X D s 4013 _ R Q (J) PWM Ql l 2 OVERCURRENT • Figure 3.18: Over-current Protection Schematic Diagram Chapter 3. Hardware Design 49 3.4 Current Sensing and Over-temperature Detection With HEXSenses, the PV array current can be sensed with virtually no power loss. Because of the temperature dependency of the on-resistance of the power MOSFET, over-temperature can be detected without a temperature sensor. The current sensing and over-temperature detection circuit is shown in Figure 3.19. There are two ways to achieve current sensing with HEXSenses: one is virtual earth To current sense pins of the three HEXSenses / Q sen4 O Iovercurrent o-sen6 9 sen2 9 * t e m P a Isen •47 47: 10k 3 h : -* » R l l . R10^47 10k 2.k2 IRFD120 2k2 lOOuF MC3407: + 2k2: r /VVvv-I—1 100k •ArM 1|> 50K Figure 3.19: Current Sensing and Over-temperature Detection Circuit sensing, the other is resistor sensing. Earth sensing is more accurate than resistor sensing and is less affected by junction temperature. However, it requires positive and negative power supply which is inconvenient in PV pumping applications. Chapter 3. Hardware Design 50 The resistor sensing method is used in the current sensing and over-temperature detection circuit. Although the temperature dependency of the resistor-sensing method is a drawback, it makes detection of an excessive temperature situation possible. In Figure 3.19, resistor RIO is either connected to the HEXSense or disconnected depending on the signal "*temp". In normal current sensing process, *temp is high, RIO is connected to the current sense pin and kelvin source pin of the HEXSense. When the microprocessor needs to check the temperature, it outputs a low level at *temp and disconnects RIO from the HEXSense. The current-sensing output will be based on the on-resistance of the HEXSense. As the on-resistance increases with the increase of HEXSense temperature, the current sensing output will also increase with the increase of HEXSense temperature. The difference between this current sensing output and the normal current sensing output indicates the HEXSense temperature. "sen2" and "sen4" are current sensing signals from the other two phases. These current sensing signals are added with an op-amp. The current sensing waveform is shown in Figure 3.20. Upper trace: motor current 5A/div. Lower trace: current sensing waveform 20m V/div. Time scale: 100 uS/div. Figure 3.20: Current Sensing Waveform Chapter 3. Hardware Design 51 3 .5 The Converter-commutator Logic Circuit The converter-commutator logic circuit generates the gating signals for the gate drives. The inputs to the logic circuit are the position signals PA, PB, PC, the duty ratio enable signal PWM, and a control signal RUN. The outputs are the six gating signals SW1 through SW6. For forward rotation defined by the manufacturer of the brushless motor, the relation between the rotor position signals and active switches is shown in Table 2.3. Based on Table 2.3, the following logic functions can be written: ONE = RAPB TWO = P~APC THREE =TBPC FOUR = PAP~B FIVE = PA-TC SIX = PBPC With a duty ratio control (PWM) on the lower three FETs and a RUN input to enable and disable all the switches, the outputs of the logic circuit can be written: SW1 = RUN • (FJPB) SW3 = RUN • (PBPC) SWh = RUN • (PAPC) SWA = RUN • PWM • (PAPB) SW6 = RUN • PWM • (PBPC) SW2 = RUN • PWM • (PAPC) The resulting converter-commutator logic circuit is shown in Figure 3.21. Chapter 3. Hardware Design Chapter 3. Hardware Design 53 3.6 P W M Signal Generation The PWM signal generator is implemented with a special purpose IC NE5561. The circuit is shown in Figure 3.22. The chopping frequency can be adjusted by varying RI 9 +5V D u t y r a t i o c o n t r o l PWM Figure 3.22: Circuit Diagram of the PWM Signal Generator and CI. For a PWM signal of 5 - 98% duty ratio at pin 7 of the NE5561 to be outputted, the voltage at point "P" should vary from 2 to 5 volts. As the D/A converter output is from 0 to 2.56 volts, an op-amp is needed for the required voltage. R3 through R6 can be calculated based on the following relation: RA R5 + R6 i23 + J?4 R6 = 2 2 . 5 6 . ^ ! _ . ^ + B 6 = 3 (3.42) (3.43) J23 + RA R6 If R3 and R6 are 10k Ohms, from equation 3.42 and 3.43 R4 and R5 can be determined: £ 4 = 3.4fcfi; i?5 = 5.7M2 Chapter 3. Hardware Design 54 3.7 Microprocessor System Hardware A single board microprocessor development system is used in the experimental pumping system. The block diagram of the microprocessor development system and its I/O inter-face is shown in Figure 3.23. The connection between the microprocessor development system to a mainframe computer is necessary for efficient data storage and analysis. There are two Versatile Interface Adaptors (VIAs) on the development system, VIAO and VIAl, each of which accommodates two 8-bit I/O ports, two timers and some other functions. Figure 3.23 shows that port A of VIAl is used as an input and output port. Seven bits of the I/O port A of VIAO are used as control input/output signals. Some of them are programmed as input bits, others as output bits. The input signals to the microprocessor system are PV array voltage and current signals. The output from the mi-croprocessor system is a control voltage which determines the duty ratio of the converter. Other I/O signals include an input logic signal indicating the direction of rotor rotation, an output logic signal to provide an increasing frequency clock for a ring counter, and a logic output signal to switch the motor operation between starting mode and operation mode. 3.8 I/O Interface and Signal Conditioning I/O interface and signal conditioning circuit transfers information between the micropro-cessor and the external world. In the experimental pumping system, the PV array voltage and current signals are needed by the microprocessor. Also, the microprocessor outputs a control voltage to control the duty ratio of the converter. Since the PV array voltage and current are analogue signals and the duty ratio control also requires an analogue Chapter 3. Hardware Design 55 is TEMP PUL OPER/ 4 f START REV 1 D/A A/D PAO PBO VIAO (6522) A/D PA1 PB1 VIA1 (6522) TT MairuTame computer RS232 DATA BUS Figure 3.23: Block Diagram of the Microprocessor Development System and Interface Chapter 3. Hardware Design 56 signal, analogue-to-digital (A/D) and digital-to-analogue (D/A) converters are needed. Some signal conditioning such as filtering and scaling are necessary to reduce the effect of noise and to match different devices. Figure 3.24 shows the I/O interface at port A of VIAl. The input interface at port A of VIAl includes an voltage divider to obtain the PV array voltage signal, an op-amp used as a voltage follower, RC low-pass filters and an A/D converter (ADC0804). ADC0804 converts an 0 - 5 volt input analogue voltage to a digital quantity (a byte). CS, WR, RD and INTR pins on ADC0804 are used for conversion control and handshaking. When WR goes low, the A/D converter starts to convert the analogue signal on pin 6 to digital form. Upon completion of conversion, INTR will become low. If RD is set to low at this moment, the digital data will be latched at the output of the A/D converter so that the microprocessor can read the PV array voltage. RD is set to high after the data are read. When RD is high, the output of the A/D converter is on third state, that is, disconnected from the data pins of Port A and the microprocessor can output a control voltage through the same I/O port. CS is a chip select pin which is active low. A handshaking pin C A l sets an interrupt flag upon completion of A/D conversion. When INTR of both A /D converters are low, C A l is pulled low, setting an interrupt flag bit in an interrupt flag register. The program polls the interrupt flag. When interrupt flag has been set, the program reads the data from Port A data register and the interrupt flag is automatically cleared. The output interface includes a D/A converter (AD558), a low-pass filter and a voltage follower. The input to AD558 is an 8-bit digital signal and the output is a voltage ranging from 0 to 2.56 volts. The low-pass filter at the output pin of the D/A converter smoothes the voltage waveform. The voltage follower is necessary for the control voltage to be coupled to the PWM signal generator. EN on AD558 is an active low enable pin. When EN is B e t to low, the output is set to a new value depending on the input digital data. When EN is high, the output remains unchanged regardless of the change on the Chapter 3. Hardware Design bit 0, port A, VIA1 bit 7, port A. VIA1 D/A AD588 10 9 11 duty ratio control InF 'I T — • — • bit 3, port A, VIAO +5V C A l INTR in port B bit 1 of port A of VIAO • bit 2 of port A of VIAO 1§ ^. bit Oof port A of VIA1 bit 7 of port A of VIA1 Figure 3.24: I/O Interface at Port A of VIA1 Chapter 3. Hardware Design 58 current signal +5V Vref 0 Q INTR in port B T ^ INTR CS WR RD DBO VREF VI+ CLK ADC0804 CLKR VI-f-*jAGND DB7 3 ^ bit 1 of port A of VIAO -2 • bit 2 of port A of VIAO 1 8 - bit Oof port B of VIAl 11 bit7ofportBofVIAl Figure 3.25: Interface at Port B of VIAl data pins of the I/O port. When a D/A conversion has been completed, EN should be set to high so that the output of the D/A converter is not affected by the input process at the same I/O port. Figure 3.25 shows the interface circuit at I/O Port B of VIAl. This circuit is similar to the input interface at Port A, but there is no output interface at Port B. Figure 3.26 shows the I/O signals at Port A of VIAO. Seven bits of this port are used for the I/O interface. Bit 0 is used in the detection of an over-temperature condition; bit 1 and 2 are used for A /D converter control; bit 3 is D /A converter enable signal; D4 and D5 are output signals used for position-sensorless operation; bit 6 is a reverse rotation signal. Chapter 3. Hardware Design 59 7 6 5 4 3 2 1 0 OPER/START REV _ 1_ • TEMP 1_ EN RD WR PUL Figure 3.26: Signals at Port A of VIAO 3.0 Position-sensorless Operation As mentioned earlier, the converter-commutator logic requires a rotor position signal as its input. Conventionally, a rotor position signal can be obtained with three Hall effect sensors and a magnetic interrupter. This method requires extra wire connection between the motor and the control unit. The wiring can be simplified and the Hall effect sensors eliminated, if position-sensorless operation is used. In position-sensorless operation, a three-bit position signal is obtained from a rotor position signal generator which makes use of the motor terminal voltages. There is a position sensor on the motor which is used for calibration and comparison purposes. Since there is no terminal voltage when the motor is in standstill condition, some method must be used to start the motor. One method is to ramp up the motor like a stepper motor. In this method, the motor operation is divided into two modes, starting mode and operation mode. In starting mode, the position signal is replaced by a pseudo position signal which simulates a ramping rotor. When the motor is brought up to a certain speed, the motor will enter the operation mode. In the operation mode, the rotor position signal derived from the motor terminals is used to drive the converter-commutator logic. Unlike the rotor position signal generated by a Hall effect position sensor, the position Chapter 3. Hardware Design 60 signal derived from motor terminal voltages depends on the direction of rotation. If the direction of rotation reverses due to a transient load, the resulting position signal will have a new relationship with the rotor position so that the rotor may rotate in reverse direction continuously. Because undesired reverse rotation may unscrew the progressive cavity pump, it should be avoided. 3.9.1 Rotor Position Signal Generator Two major components in the motor terminal voltages are: one, the trapezoidal waveform with a frequency proportional to the motor speed; two, the PWM content with a fixed chopping frequency. The zero crossing of the phase back em/, a good indication of rotor position, occurs when terminal voltage is half of the source voltage (unity duty ratio assumed). For position signal generation, a circuit similar to that described in [16] is used. However, a 30 degree lag angle in motor terminal voltages is used instead of the 90 degree lag angle used in [16] because of the noise appearing in the filtered signal at a large lag angle. This noise may be explained by the motor used in the motor used in the experiment having more poles and working at higher frequency than the one used in [16]. The schematic of the rotor position signal generator is shown in Figure 3.27. vai vj, and ve in Figure 3.27 are the motor terminal voltages in reference to ground. C l and R3 form a low-pass filter to filter out the high frequency content. The values of C l and R3 should be chosen such that the zero crossing of va lags that of phase A em/30 degrees at center operating speed. The references of the three comparators are connected to a synthesized neutral point. The resulting PA (phase A position signal) will be 30 electrical degrees behind phase A emf. Similarly, PB and PC will be 30 electrical degrees behind phase B and C emf respectively. The ideal motor terminal voltages, back em/s and position signal for forward rotation are shown in Figure 3.28 (unity duty ratio assumed Chapter 3. Hardware Design motor terminal A 68k + v, A i i R3 4.7k 2.2k: .115uF Cl motor terminal B ~, 68k + V B A 2.21 4.7k .115uF motor | terminal C - 6 8 k + v 4.7k 2.2i .115uF Af 100 --vW-OluF : 100 V B f j .OluF : 100 vcf| I r .OluF 47k 4 7 k A»A /V 47k +5V PA' +5V PB' +5V P C Figure 3.27: Schematic of Rotor Position Signal Generator Chapter 3. Hardware Design 62 for clarity). If the required direction of rotation is opposite to the forward direction defined by the motor manufacturer, the schematic in Figure 3.27 must be modified. When the motor rotates in the reverse direction, the corresponding waveforms of the motor terminal volt-ages, emfs and position signal generator outputs are shown in figure 3.29. The position signals PA, PB and PC in Figure 3.29 are generated from the circuit in Figure 3.27 with the motor rotating backwards. If these position signals are used directly for the required backward rotation, the converter-commutator logic circuit will be completely different from that for forward rotation. However, if the position signal for backward rotation is defined as follows, the forward converter-commutator logic can be used for reverse rotation: RPA = PU (3.44) RPB = PA (3.45) RPC = P~B (3.46) where RPA, RPB and RPC are the position signals for phases A, B and C respectively. In the implementation of the rotor position signal generator circuit of Figure 3.27, care must be taken to ensure balanced current waveforms for the three phases. The experiment showed that the three low-pass filters should be as identical as possible otherwise the resulting position signal would have unequal on and off periods which, in turn, cause unbalanced motor phase currents. 3.0.2 Motor Starting Pulse Generation When the motor is idle, no position signal is generated from the position signal generator. For the motor to be started from standstill, a ring counter circuit provides the pseudo Chapter 3. Hardware Design SWON: 3,4 4.5 5,6 6,1 1,2 2,3 3,4 Position: 3 4 5 6 1 2 3 Figure 3.28: Voltage, emf and Position Signal Waveforms for Forward Rotation Chapter 3. Hardware Design SW ON: 5,4 4.3 3,2 2,1 1,6 6,5 5,4 Position: 6 5 4 3 2 1 6 v A i i Figure 3.29: Voltage, emf and Position Signal Waveforms for Backward Rotation Chapter 3. Hardware Design 65 position signal. The counter schematic is shown in Figure 3.30. SA, SB and SC form a Figure 3.30: Ring Counter for Motor Starting 3-bit pseudo position signal; PUL is a clock signal generated by the microprocessor. The state diagram for the counter is shown in Figure 3.31. There are six states in the counter. Each state provides the position signal for 60 electrical degree interval. There are two unused states, "000" and "111". For these states to be prevented, C l , R l and C2, R2 are added so that, when the power is up, the initial value of SC is always low and SB is always high. The clock signal PUL has an increasing frequency. Since the counter outputs SA, SB and SC drive the converter-commutator logic in starting mode, the motor speed will be ramped up. If the required direction of rotation is backwards, the circuit of Figure 3.30 need not be changed except that QI should be used as SA and Q2 as SB. Chapter 3. Hardware Design 66 For forward rotation SC.SB.SA SC.SB.SA Figure 3.31: State Diagram for the Ring Counter 3.0.3 Motor Mode Control When the motor is starting, it is running in starting mode. In starting mode, the motor is ramped up to a certain speed at which a reliable rotor position signal can be derived from the motor terminal voltages. The ring counter will be disabled and the converter-commutator logic will be driven by the position signal generator. The motor then operates in operation mode. The switching of motor modes is performed with the logic circuit shown in Figure 3.32. The exact moment of mode change is determined by software and will be discussed in the next chapter. In the starting mode, OPER/START is set to low. At the moment of mode switching, OPER/START is set to high, so that the ring counter is disabled and the rotor position signal generator is enabled. The motor then switches from starting mode to operation mode. Chapter 3. Hardware Design Chapter 3. Hardware Design 68 3.0.4 Reverse Rotation Detection Unlike Hall effect position sensors, the position signal generator of Figure 3.27 may, depending on the direction of rotation, output two possible waveforms for a given rotor position. At low speed in position-sensorless operation, the motor may be forced to unwanted reverse rotation due to a transient load. This problem can be prevented with a simple circuit shown in Figure 3.33. In Figure 3.33, PA and PC are the outputs from REV • Figure 3.33: Reverse Rotation Detection Circuit the position signal generator for phase A and phase C respectively. When the motor rotates in the forward direction, PA lags PC 120 degrees, and REV is high. When the motor rotates in the reverse direction, PA leads PC 120 degrees, and REV becomes low indicating reverse rotation. Chapter 4 Software Development Software development was done on a Motorola 6809 microprocessor development system. A modular design method was used in the entire development process. In this chapter, the functions to be implemented with software are described and the flowcharts of different modules are illustrated and explained. The source codes are written in 6809 assembly language. A Motorola 6809 microprocessor is used because it was available at the time this research started. The resulting assembly program can be easily modified to be run on a more compact single-chip microcomputer such as Motorola 68HC11. 4.1 Functional Description For the microprocessor based pumping system to operate efficiently and reliably, some control functions must be implemented. The top level control flow of the PV pumping system can be represented in an algo-rithmic state machine (ASM) diagram as shown in figure 4.34. The circles indicate states and arrowed lines indicate transition between states. The conditions for a transition to happen are listed beside the arrowed lines. The system is in IDLE state when there is not sufficient power from the PV array. In this state the microprocessor and the control electronics are disconnected from the PV array. The PV array voltage is close to open circuit voltage. To leave IDLE state, the PV array voltage must be higher than a value Vmng. When the system is in READY state, the microprocessor and control electronics are connected to the PV array and the array voltage drops to a lower level, but it must 69 Chapter 4. Software Development 70 Figure 4.34: Top Level Control in Algorithmic State Machine Diagram be higher than a voltage limit VTdy in order for the system to remain in READY state, or the system will move back to IDLE state. To enter the START state, the maximum PV array power must be higher than a minimum value requirement PST • START is a temporary state which, upon completion, will always lead to OPERATION state. Nor-mal operation of the system stays in OPERATION state, unless faulty conditions such as reverse rotation, over-temperature or stalled rotor are detected. Based on the top level control of the PV pumping system, many functions are im-plemented to ensure good performance of the pumping system. Some major functions are: 1. In READY state, the microprocessor executes an initialization program to set up Chapter 4. Software Development 71 the I/O ports and initial values for some output signals. Then the microproces-sor scans the I-V curve of the PV array by slowly varying the duty ratio of the converter. The maximum available power is calculated. When it is higher than a threshold determined by the water head, the microprocessor executes a motor starting routine. PV array voltage is sampled in READY state in order to detect the condition for state transition. Fault flags are tested in READY state. Different fault conditions are handled differently. ST and TEMP should be high and there should be no output at PUL . 2. In START state, a clock pulse with increasing frequency is output on pin PUL . ST is held low so that the motor speed is ramped up. When motor speed reaches 100 rpm, ST is switched to high and the START state is completed. 3. In OPERATION state, the double-loop control algorithm should be executed for maximum power tracking. The algorithm of this double-loop structure is given in Chapter 2. The implementation of this algorithm will be described later in this chapter. 4. Functions to detect various faulty conditions, e.g., reverse rotation, over-temperature and stalled rotor, are executed in OPERATION state. The reverse rotation condi-tion is tested by hardware. The software only needs to poll the signal REV. For over-temperature detection, the microprocessor reads the HEXSense current with body resistance and external resistance after which it compares the two measured values. If the difference is larger than a preset value, an over-temperature condition is determined. As the motor stops when power from the PV array is insufficient, the microprocessor must detect the stalled rotor in order to prepare for sensorless starting when the array power becomes sufficient again. The detection of a stalled Chapter 4. Software Development 72 rotor can be achieved by checking the following inequality: 62v. < i.Ra (4.47) Theoretically, the left-hand side of Inequality 4.47 cannot be less than the right-hand side. However the "less" condition is helpful for microprocessor implementa-tion. 5. Auxiliary functions that are needed in the development phase include: an I-V curve recording function, data display function, etc. 4.2 Description of Major Routines The entire program is written in 6809 assembly language in a modular fashion. There is one main routine which consists of a number of modules. Each of these modules may contain some other modules. The following subsections describe the main routine and some major modules. 4.2.1 Main Routine The flowchart for the main routine is shown in Figure 4.35. The routine is entered when the system enters READY state. After initialization, an initial duty ratio is output to the PWM control circuit and fault flags are checked. If no faulty condition has occurred, the program checks the conditions for state transition. MOTORJST is a module for motor starting. VCONT is a module for maximum power control. FAU-DT checks reverse rotation, stalled rotor and over-temperature, and sets the corresponding fault flags. FAULTS checks the fault flags and waits for a certain length of time depending on the type of fault, and then returns to normal operation. Chapter 4. Software Development 73 Figure 4.35: Flowchart for Main Routine Chapter 4. Software Development 74 4.2.2 Modules in R E A D Y State INITIALIZE The flowchart for INITIALIZE is shown in Figure B.56 in Appendix B. INITIALIZE sets up the two VIAs, and some system parameters and flags. No parameters are passed to this module. O U T J D R D Y and O U T _ D OUT_DRDY is a module that outputs an initial value of duty ratio for READY state. One module (OUTJD) is called. The initial value DRDY is passed to OUT_D through memory location D l . The flowcharts for OUT JDRDY and OUTJD are shown in figure B.57 in Appendix B. V I J S A M VIJSAM reads the PV array voltage and current simultaneously. Port A and B must be set to input ports before VIJSAM is entered. The sampled voltage and current signals are stored in memory locations VOUT and IOUT. The flowchart of VIJSAM is shown in Figure B.58 in Appendix B. C U R V E Module CURVE is designed for the testing of the maximum power operation of the PV array. It outputs a zero duty ratio at the beginning after which it increments the duty ratio with an adjustable rate until the upper limit DHIG is reached. For each increment step, the PV array voltage and current are stored in a predefined memory area. The PV array output power is calculated and the maximum power and the corresponding voltage and current are recorded. The program will then decrement the duty ratio until a lower Chapter 4. Software Development 75 limit DLOW is reached. PV array current and voltage values are still stored for each decrement step, but the maximum power is not calculated. The flowchart for CURVE is shown in Figure B.59 in Appendix B. VIJSAM and OURJD are called in CURVE. The results are stored in memory locations MPW, IMP, VMP and DMP. 4.2.3 M O T O R _ S T MOTOR-ST facilitates the motor starting process and: 1. outputs duty ratio value for starting; 2. outputs a series of pulses with increasing frequency; 3. keeps track of the number of pulses that have been sent and enters the OPERATION state when the number reaches a preset value. The flowchart of this module is shown in Figure 4.36. This module calls OUTJD and SYN. SYN initiates timer 1 of VIAl. The flowchart for SYN is shown in Figure 4.37. In Figure 4.37, PERIOD determines the timer period. For example, if PERIOD is 1000, the timer will count 1000 clock pulses (1 ms) before timer interrupt occurs. PERI determines the initial frequency of the starting pulses. If PERI is 100, the initial period of the starting pulses will be 100 ms (PERIOD = 1000 assumed), and the initial frequency is 10 Hz. At the end of starting state, the starting pulse frequency will reach 62.5 Hz, corresponding to approximately 100 rpm. Upon timer 1 time out, an interrupt service routine PULSE is entered. A pulse will be sent to pin TUT if PULSE has been entered PERI times. The flowchart of PULSE is shown in Figure 4.38. Chapter 4. Software Development 76 ENTRY 1 r OPER/START set to low i Dl<- DST f OUT_D r Reset pulse countr: PNUM Initialize reverse counter Initialize zero duty ratio counter Disable interrupt } OPER/START set to high EXIT Figure 4.36: Flowchart for MOTORJ5T Chapter 4. Software Development SYN Set PERI Enable timer interrupt Set PERIOD m Start timer ^ EXIT ^ Figure 4.37: Flowchart for SYN Chapter 4. Software Development PULSE (Interrupt service routine) ^ENTRY^ Decrement period counter: PCNR Increment pulse counter: PNUM PERI<==(15/16)*PERI PCNR<=PERI Clear interrupt flag ^ E X T T ^ Disable timer interrupt Figure 4.38: Flowchart for PULSE Chapter 4. Software Development 79 4.2.4 V C O N T , A D J U S T and P O W E R VCONT is a module to implement the double-loop control algorithm described in Chap-ter 2. The inner voltage loop has an integral controller. The PV array voltage is controlled to follow the voltage reference of the voltage loop. The reference voltage is stored in mem-ory location VR. VCONT calls a module ADJUST which implements the extremum con-trol algorithm (outer loop of the double-loop structure). In ADJUST, a module POWER adjusts the reference voltage according to the extremum control algorithm described in Chapter 2. The output from VCONT is a duty ratio value stored in memory location Dl . The flowchart for VCONT is shown in Figure 4.39. The flowchart for ADJUST is shown in Figure 4.40. Figure 4.39 shows that a delay count DOUT is used to adjust the ADJUST ENTRY I Decrement delay counter: ICN re-initialize delay counter: ICN 1 Record V J POWER EXIT Figure 4.40: Flowchart for ADJUST Chapter 4. Software Development 80 V C O N T re-initialize D O U T yes 1 DD=(Vs-Vr)*Ki yes DD<=LIMIT DD>LIMIT7 D1<-D1+DD T «« ^ < T Dl>255 ? Dl=255 no ADJUST no r DD=(Vr-Vs)*Ki no D E X = L I M T T r * D1<-D1-DD ^ E X I T ^ Figure 4 . 3 9 : Flowchart for VCONT Chapter 4. Software Development 81 Figure 4.41: Flowchart for POWER Chapter 4. Software Development 82 actual sampling rate for the voltage loop. In OPERATION state, the PV array voltage and current are sampled each time the program runs through a cycle. Adjusting DOUT can vary the execution time of one cycle so that the sampling rate of the voltage loop can be adjusted. Figure 4.39 also shows that the change of duty ratio in each cycle is limited to a preset value. The duty ratio is not allowed to jump to a large value because a sudden increase of duty ratio can cause a very high transient current through the power MOSFETs. There is also a delay count in ADJUST. The reference voltage for the voltage loop is updated once when the main body of VCONT is entered a certain number of times. The number is the delay count stored in memory location ICN. The value in ICN should be chosen carefully. If it is too small, the updating of the extremum control loop will be too fast and the PV array voltage and current will then be sampled before they reach steady state. The operation of the extremum control loop would deteriorate. However, if the value in ICN is too large, the updating of the extremum control loop will be too slow. The I-V curve of the PV array may have been changed noticeably within one update period. Fortunately, the settling time for a motor drive is usually short enough to ensure the assumption of an unchanged I-V curve on a sunny day. Several update period values were tested and an update period of 3 seconds was selected. A "RECORD I,V" box in module ADJUST in Figure 4.40 records the PV array current and voltage and saves them in the memory of the development system. The values recorded represent the operating points of the PV array. These operating points are compared with the recorded I-V curve to see how close the operating points are to the maximum power point of the PV array. The flowchart for POWER is shown in Figure 4.41. In module POWER, the voltage reference VR is adjusted according to the sign of dPa/dVt. It can be seen from Figure 4.41 Chapter 4. Software Development 83 that the change in VR is a fixed step DV. DV is the microprocessor representation of A V in Equation 2.36 and should be selected as small as possible, but must be significantly-larger than the noise and ripple in PV array voltage. A large DV will cause unnecessary oscillation around the maximum power point. If DV is top small, the noise and ripple in the PV array voltage will reduce the effectiveness of the algorithm. As described in Chapter 2, AV (DV) should be selected according to the system noise and the I-V characteristics of the specific PV array. In experiment, DV was set to 4 which corresponds to about 2% of the open-circuit voltage of the PV array. POWER searches for the maximum power point effectively when the change in radi-ation is slow. When the radiation changes suddenly, as may happen on a cloudy day, VR remains basically unchanged. The system remains in normal operation until the power from the PV array becomes insufficient to run the pump. 4.2.5 Fault Detecting and Handling Modules FAUJDT is designed for fault detection. It calls three modules: STL_DT, REV JOT and TEMPJDT. The flowchart of FAUJDT is shown in Figure B.60 in Appendix B. STLJDT is a module designed to detect a stalled rotor. The flowchart of the module is shown in Figure B.61 in Appendix B. The detection of a stalled rotor is done by checking the condition of inequality 4.47. With voltage feedback control and if the duty ratio is forced to zero at steady state, the rotor must have stalled. However, in a transient situation the duty ratio may drop to zero even when the motor is not stalled. A zero count ZCNR is used to avoid false detection. Only when zero duty ratio is recorded for 256 times will the microprocessor set up a stalled-rotor flag. STLJDT calls a module CHKRTR to check the rotor condition. The flowchart of CHKRTR is shown in Figure B.62 in Appendix B. Chapter 4. Software Development 84 When the inequality 4.47 is true, the motor is in a stalled condition. In microproces-sor, vt,i, and 8 are represented by 3 separate bytes as shown in the following forms: v. = Kv • VOUT (4.48) i. = Ki • IOUT (4.49) S-g . (4.50) where VOUT, IOUT and Dl are the three bytes representing vt, it and 8. Knew constant is defined as: K=Tk <4-51> the stalled rotor detection inequality can be written in the following form: This inequality shows that the calculation of the left hand side of 4.52 is reduced to three 8-bit by 8-bit multiplications in the microprocessor. REVJDT is designed to poll the reverse rotation signal generated by hardware. The flowchart for REVJDT is shown in Figure B.63 in Appendix B. TEMPJDT is designed to detect over-temperature condition. The flowchart of TEMPJDT is shown in Figure B.64 in Appendix B. FAULTS, a fault handling module, is entered whenever a faulty condition is detected. This module delays a certain period of time according to the type of the faulty condi-tion and then sets the system to ready state. The flowchart for FAULTS is shown in Figure B.65 in Appendix B. 4.3 Summary This chapter describes the software aspect of the PV pumping system. The top level control flow of the system is illustrated with an ASM diagram. The entire software is Chapter 4. Software Development 85 written in 6809 assembly language in a modular fashion. Major modules are described in detail with flowcharts. In practice, the software developed in the research work should be loaded onto a dedicated 6809 microprocessor system because the microprocessor devel-opment system is relatively expensive. Given the modular nature of the software design, the source codes can be relatively easily converted to fit a single chip microcomputer (or microcontroller), such as 68HC11. The result would be a simpler and more compact system. Chapter 5 Experimental Results This chapter describes the experimental results of the research work. An experimental PV pumping system was implemented according to the analysis and design described in Chapters 2, 3 and 4. Operating data of the experimental pumping system were obtained in the laboratory. Among the major experimental results are data recorded on the maximum power tracking operation, data on converter-motor efficiency and data on position-sensorless operation of the BLDC motor. 5.1 Maximum Power Tracking The maximum power tracking function described in the previous chapters was tested experimentally in the laboratory. An array of ten PV panels were used as the power supply. Although these PV panels have relatively low fill-factor largely due to their age, they are sufficient to verify the effectiveness of the maximum power tracking scheme proposed in this thesis. The operating points were recorded for both sunny and cloudy conditions. Immedi-ately after the operating points were recorded, the I-V curve of the PV array was scanned and saved for reference. Figure 5.42 shows a plot of the recorded operating points for about 6 minutes on a clear day with a sampling rate of about 0.33 Hertz. The operating points are scattered around the knee point of the I-V curve. The maximum-power voltage is 71.5 volts. The corresponding current is 2.21 amperes. The mean value of the recorded voltages is 72.5 volts, and the standard deviation is 1.8 (volt). The recorded operating 86 Chapter 5. Experimental Results OPERATING POINTS IN A SUNNY DAY RECORDING PERIOD = 6 MIN. 100 Array voltage (V) Figure 5.42: Recorded PV Array Operating Points in A Period of 6 Minutes Chapter 5. Experimental Results 88 points are also listed in Appendix C. If utilization factor of the PV array nu is defined as: then the utilization factor of the PV array in the recorded 6-minute period can be ap-proximated as: V u ~ 128 x max{Pt(k)} ( 5 , 5 4 ) where p(t) and p m o x (i) are instant array output power and maximum available power respectively. T l and T2 form a time period for which utilization factor is estimated. P,(fc)'s are sampled array output power calculated from the recorded current and voltage samples. The result from approximation 5.54 is 99.26%. Since the maximum power tracking function is working continuously during normal operation, this utilization factor is maintained for all seasons. This consistent efficiency is the main advantage of the proposed maximum power tracking scheme over constant voltage schemes. As is described in Appendix A, the loss due to a fixed "optimum" voltage for different seasons can be 2.7% in an area where the day-average temperature varies 25 °C during the year. The degradation of utilization factor due to PV array aging is much smaller for the proposed maximum power tracking scheme than that for a voltage tracking scheme. The approximation of expression 5.54 assumes that I-V curve of the PV array is unchanged. This assumption can be justified by examination of the operating points recorded in Figure 5.42. Figure 5.43 shows the PV array operating points in a period of 6 minutes when radiation fluctuates. The PV array current varied between 0.75 to 1.5 amperes. The array voltage remained relatively constant while the current varied with the radiation. The I-V curve in Figure 5.43 was scanned and recorded immediately after the last operating point had been recorded. The scanning of the I-V curve took 40 seconds and the scanning result Chapter 5. Experimental Results 89 Figure 5.43: PV Array Operating Points During Sudden Change of Radiation Chapter 5. Experimental Results 90 Table 5.4: Combined Converter-motor Efficiency Input voltage (V) Input current (A) duty ratio Speed (rpm) v(%) 94.3 1.23 0.5 430 67.3 91.7 2.34 0.5 740 74.6 91.2 2.06 0.5 779 76.9 91.3 3.5 0.75 1155 79.6 90.2 4.99 0.95 1458 81.2 91.0 4.48 0.95 1550 82.4 shows significant variation in I-V characteristic during the 40-second period. Figure 5.43 shows that stable operation is maintained when radiation varies quickly. The observed stable operation is attributed to the voltage control loop. Without the voltage loop, the PV array voltage may drop to a very low level and the motor may even stop. 5.2 Converter-motor Efficiency and Position-sensorless Operation 5.2.1 Efficiency Evaluation The converter-motor efficiency was measured in the laboratory and is listed in Table 5.4 which shows that, when the speed varies from about 500 to 1500 rpm, the efficiency varies from 67 to 82%. In order to compare with brush-type permanent-magnet dc motors, converter-motor efficiencies for two other systems are examined [49, 50]. In [49] the converter-motor efficiency varies from 61 to 80% when speed varies from 500 to 1500 rpm. In [50] the converter-motor efficiency varies from 64 to 84% when the speed varies from 500 to 1800 rpm. This comparison shows that the brushless dc motor used in the experimental pumping system is slightly less efficient than the brush-type dc motors used in [50]. The slightly lower efficiency is the price paid for the reduced maintenance requirement in brushless dc motor. Chapter 5. Experimental Results 91 5.2.2 Efficiencies With and Without a Position Sensor The combined motor and converter efficiency vs. speed curves are plotted for two different situations: conventional operation and position-sensorless operation. The plot is shown in Figure 5.44 which shows that the efficiency for position-sensorless operation is close to that of conventional operation, especially when the speed is above 500 rpm. Since the motor operates mainly at speeds higher than 500 rpm, the reduction in overall efficiency due to position-sensorless operation is marginal. 5.2.3 Current Transient During Motor Mode Switching When the motor is in the starting process, the motor transient current must be limited to an allowable level. For the transient current to be minimized, some parameters should be chosen carefully. These parameters include motor accelerating rate, duty ratio and the speed at which the motor switches from starting mode to operating mode. Figure 5.45 which shows the motor current during motor mode switching indicates that the transient current is about 50% higher than the steady state current. 5.3 Motor and Converter-commutator Operation 5.3.1 Sinusoidal and Trapezoidal Excitation Both trapezoidal excitation and sinusoidal excitation are tested. The motivation for the experiments with sinusoidal excitation is to verify the use of trapezoidal excitation. Fig-ure 5.46 shows the current and voltage waveforms for sinusoidal excitation. According to theoretical analysis, sinusoidal excitation results in lower copper loss and higher switch-ing loss (refer to Equation 2.7 and Table 2.1). Experimental results show that sinusoidal excitation does not result in higher efficiency than trapezoidal excitation. Table 5.5 shows the efficiency measurement for the two excitation methods. Table 5.5 shows that Chapter 5. Experimental Results 92 EFFICIENCY OF THE BRUSHLESS MOTOR sensorless and conventional operation Figure 5.44: Efficiencies with and without a position sensor Chapter 5. Experimental Results 93 * * * < -t .. -y,. -f i t * " " Upper trace: motor current at mode change 5A/div. Lower trace: mode changing signal 2 V/div. Time scale: lOOmS/div. Figure 5.45: Motor Current During Motor Mode Switching efficiency increases with duty ratio. The last line of the table shows that, i f the duty ratio and modulat ion depth are both 80%, trapezoidal excitation is 1 percent more efficient than sinusoidal excitation. Table 5.5: Measured Efficiency for Sinusoidal and Trapezoidal Excitat ion Excitat ion D C voltage Duty ratio or efficiency method supply (V) modulation depth (%) ' sinusoidal 80.7 0.8 74.03 trapezoidal 80.5 0.67 74.56 sinusoidal 90.9 0.8 75.84 trapezoidal 90.9 0.67 75.84 trapezoidal 76.1 0.8 76.9 Chapter 5. Experimental Results 94 Upper trace: motor current 5A/div. Lower trace: motor terminal voltage in reference to ground 20 V/div. Time scale: 10 mS/div. Figure 5.46: Motor Current and Voltage in Sinusoidal Excitation 5.3.2 HEXSense Current Sensing and Over-temperature Detection As described in chapter 3, HEXSense devices are used to obtain current signal and to detect an over-temperature condition. Figure 5.47 shows motor current waveform and current sensing output waveform. The polarity of the current sensing output is opposite to that of the motor current in Figure 5.47. The current sensing output follows the motor current closely. Figure 5.48 shows the filtered current sensing outputs with M O S F E T body-resistor sensing and with external resistor sensing. An over-temperature condition is determined by the difference between these two outputs. Figure 5.49 shows the recorded voltage dif-ference between body-resistor sensing and external resistor sensing at different M O S F E T case temperatures. Chapter 5. Experimental Results 95 0 0 Upper trace: motor current (one phase) ' 5 A/div. Lower trace: three phase current sensing output (with opposite polarity to motor current) 2 V/div. Time scale: 2 mS/div. Figure 5.47: Motor Line Current Waveform and Three-phase Current Sensing Output Signal 5.3.3 M O S F E T Waveforms The power MOSFET drain to source voltage waveform and drain current waveform are shown in Figure 5.50. 5.4 Summary Experimental results are presented in this chapter. The double-loop maximum power tracking algorithm is proved effective and robust. The combined converter-motor effi-ciency is reasonably good compared with that of conventional brush-type systems. The PV array powered position-sensorless operation is feasible. The reduction in efficiency is marginal when the speed is above 500 rpm. The transient current during motor-mode changing is limited within an allowable level. The converter-commutator circuit with trapezoidal excitation is simpler and more efficient than the sinusoidal excitation Chapter 5. Experimental Results 96 CO MOSFET current in Amperes Figure 5.48: Filtered Current Signals With and Without an External Resistor scheme. Chapter 5. Experimental Results 97 Figure 5.49: Current Sensing Difference vs. MOSFET Case Temperature Upper trace: bottom FET current 5A/div. Lower trace: drain-source voltage of a bottom FET 50 V/div. Time scale: 10 uS/div. Figure 5.50: Bottom MOSFET Current and Voltage Waveforms Chapter 6 Conclusions A microprocessor based PV pumping system using a BLDC motor was optimized for efficiency and the reliability of the system was considered throughout the design pro-cess. The maximization of efficiency was done in two aspects: one, to extract maximum available power from the PV array at all radiation and temperature conditions; two, to search for the most suitable scheme for motor excitation, power circuit structure, and the entire P V pumping configuration and to rninimize the loss in every part of the electrical subsystem. Some conclusions are: 1. The double-loop structure for maximum power tracking proved to be effective and reliable. The inner voltage loop ensures the stable operation of the system regard-less of sudden changes in radiation. The outer extremum control loop searches the maximum power point of the PV array continuously so that the PV array operates around the maximum power point at all seasons of the year. The inner voltage loop must have a much faster sampling (updating) rate than that of the outer extremum control loop so that the operation of the outer (extremum) loop is not affected by the transient of the inner (voltage) loop. An update period of 1 to 10 is considered suitable. The change rate of duty ratio should be limited so that power MOSFET damage is avoided. The outer extremum control loop is based on the sampling of the steady-state value of PV array operating points. One assumption for this algorithm is that radiation and temperature do not change suddenly. Step A V in extremum control should be as small as possible, but must be 98 Chapter 6. Conclusions 99 significantly larger than the noise and ripple on the PV array voltage and current. The value of AV is mainly affected by system noise and the fill factor of the PV array. 2. For a given motor output power, the copper loss for sinusoidal excitation is about 10% lower than that for trapezoidal excitation. However, for the same condition, the switching loss for sinusoidal excitation is over 2 times that for trapezoidal excitation. Experimental results show that trapezoidal excitation is slightly more efficient than sinusoidal excitation. In practice, since the control hardware and software for sinusoidal excitation are more complicated than that for trapezoidal excitation, trapezoidal excitation is preferable to sinusoidal excitation. 3. A converter-commutator circuit was designed and implemented to accommodate trapezoidal excitation of the brushless PM motor. The combination of the converter-commutator circuit and the brushless PM motor can be modeled as a buck converter driving a PM dc motor. Compared with the structure of a separate buck converter driving a BLDC motor, the converter-commutator scheme requires fewer power switches and results in lower power loss. This advantage is realized by a converter function incorporated into the electronic commutator which is inherent of a BLDC motor. 4. Comparison of two ways of implementing the power circuit of the converter-commutator finds that the way using third generation HEXFETs and their internal diodes are more efficient and less expensive. 5. The position sensorless operation of a BLDC motor studied in this thesis is found suitable for application where the high-to-low speed ratio is within 2 to 3. For wider range of operating speed, the position signal generator must be modified for Chapter 6. Conclusions 100 efficiency. For balanced motor currents, the three voltage filters in the position signal generator must be made as identical as possible. The elimination of a rotor position sensor simplifies the wiring. The added cost for the position-sensorless operation is less than that of a rotor position sensor. 6. The use of HEXSense devices to sense current signal is efficient and economic. It is much cheaper than using Hall effect sensors and the power consumption is less than using a series resistor. 7. The use of a microprocessor provides the flexibility for various functions and built-in "intelligence" to be implemented. The communication link between the micropro-cessor system and a remote computer enables convenient display and analysis of real-time system performance. For further work on improvement of the efficiency of a PM brushless motor, advancing commutation angle with microprocessor control is suggested. The use of a microprocessor permits adjustment of the commutation angle according to the motor speed and facilitates position-sensorless operation. The potential advantages are improved efficiency at high speed and increased range of operating speed for position sensorless operation. References [1] M.B. Aylward and B. McNelis, " Small-Scale Solar Pumping Systems - present status and future prospects ", International Journal of Ambient Energy, Vol. 5, No. 3. pp. 147-58, July 1984. [2] J . P. Requier, M. Barlaud and P. Rouan, " Optimization of Photovoltaic Solar Pump by Means of Static Converters Driven by Microprocessor, Fifteenth IEEE Photovoltaic Specialists Conference Record, pp. 1395-8, May 1981. [3] H. Durand and G. J. Naaijer, "Water Pumps Driven by Photovoltaic Modules", Proceedings of the International Conference on Heliotechnique and Development, Saudi Arabia, pp. 315-30, 1975. [4] J.A Merrigan, Sunlight to Electricity, p.113, the MIT Press, Cambridge, Mas-sachusetts, 1982. [5] A. Braunstein and A. Kornfeld, "Analysis of Solar Powered Electric Pumps", Solar Energy, Vol. 27, No. 3, pp. 253-40, 1981. [6] R. W. Matlin, F. W. Sarles, A. Rangarajan, S. Reyes, Jr. UPV Water Pumping with Reciprocating Volumetric (Jack) Pumpsn, Proceedings of 15th IEEE Photovoltaic Specialists Conference, Orlando, Florida, pp.1386-94, 1984. [7] Y. Roger Hsiao and Bruce A. Bleuins, uDirect Coupling of Photovoltaic Power Source to Water Pumping System", Solar Energy, Vol. 32, No. 4, pp. 489-98, 1984. 101 References 102 [8] Ir. C. Franx, "A New Approach to Solar Pump Systems Using Submersible Mo-tors" ,2nd E. C. Photovoltaic Solar Energy Conference, pp. 1038-45. April 1979. [9] Peter R. B. Ward, William G. Dunford and D. L. Pulfrey, "Performance of Small Pumps with Solar Power ", Canadian Society for Civil Engineering Annual Confer-ence, pp. 461-77, 1985. [10] J. Appelbaum, " The Operation of A DC Permanent Magnet Motor Powered By A Solar Cell Generator", MELECON '85, Vol. IV: Solar Energy. [11] J. Appelbaum, "The Degree of Load Matching in Photovoltaic Systems ", The Pro-ceedings of the seventh European Community Photovoltaic Solar Energy Conference, pp.172-6, 1986. [12] J. Appelbaum, uThe Quality of Load Matching in A Direct-Coupling Photovoltaic System", IEEE Transactions on Energy Conversion, Vol. EC-2, No. 4, pp.534-41, December 1987. [13] R. D. Middlebrook and Slobodan Cuk, " A General Unified Approach to Modelling Switch-converter Power Stages ", IEEE PESC Record 1976, pp.18-31. [14] Jan Sternby, "Extremum Control Systems - An Area for Adaptive Control?", Joint Automatic Control Conference, Vol. 1, WA2-A, August 13 - 15, San Francisco, 1981. [15] M. Jufer, "Self-commutation of Brushless DC Motors without Encoders ", Power Electronics and Applications Proceedings, Vol. 2, pp. 3/275-9, Brussels, Belgium, 16 - 18 Oct. 1985. [16] Kenichi Iizuka, Hideo Uzuhashi, Minoru Kano, Tsunehiro Endo and Katsuo Mohri, 11 Microcomputer Control for Sensorless Brushless Motor*', IEEE Transactions on Industry Application,Vol. IA-21, No. 4, May/June 1985. References 103 [17] J. Appelbaum, "Performance Characteristics of A Permanent Magnet D.C. Motor Powered By Solar Cells ", Solar Cells, Vol. 17, pp.343-62, 1986. [18] Peter D. Lawrence and Konrad Mauch, Real-Time Microcomputer System Design: An Introduction, McGraw-Hill Inc., 1987. [19] J. M. Gichuni and N. S. Walkade, "PV-Solar Water Pump for Kenya at Kiserian", African Electrical Technology Conference Proceedings, pp.El.4/1-3, Dec. 1983. [20] A. Hori, K. Kanematsu, T. Abe and Y. Hamakawa, "An Optimum Design of Pho-tovoltaic Direct Coupled Water Pumping System", Eighteenth IEEE Photovoltaic Specialists Conference, pp.1626-31. Oct. 1985. [21] INTERNATIONAL RECTIFIER, H E X F E T Designer's Manual Fouth Edition, International Rectifier, 233 Kansas St., El Segundo, California, Sept. 1987. [22] Paul Longrigg, "Use of Solar Photovoltaics to Transport and Desalt Ground Water Supplies Using Brushles D.C. Motors", Solar Cells 13 (1984 - 1985), pp.231-51. [23] Giampiero Brentani, "Considerations on the Locked Rotor Torque Characteristics of Brushless DC Motors", MOTOR-CON 1985 Proceedings, pp.107-14, April 1984. [24] Jaroslav Tomasek, "Electro-Mechanical Performance Data for Sine- Wave Brushless Servo Systems", MOTOR-CON 1986 Proceedings, pp.16-28, October 1986. [25] Jaroslav Tomasek, " Velocity and Position Feedback in Brushless DC Servo Systems", MOTOR-CON 1985 Proceedings, pp.61-74, April 1985. [26] Jonathan Klein, " Brushless DC Motor Commutation", ELECTRO/86 and MINI/MICRO Northeast 1986 Conference Record, Vol. 11, pp.16/3/1-5, 1986. References 104 [27] P. U. Calzolari, G. C. Cardinali and E. Faldella, " fiC-Controlled Direct Adaptive Coupling of Photovoltaic Generators to Pumping Systems", MELECON '85/Volume IV: Solar Energy, pp.141-5, 1985. [28] V. G. R.au, A Saha and G Banerjee, " Analysis of A Solar Photovoltaic Converter Fed Electric Motor Drive", Proceedings of the Fifth International Conference on Energy Options, pp.99-102, April 1987. [29] Mahesh J. Shah, Norman L. Kopp and Jayant G. Vaidya, "Stall Torque Analysis of Brushless DC PM Motors ", MOTOR-CON 1986 Proceedings, pp.30-40, October 1986. [30] J. M. Rolland, S. Astier and C. Masselot, " A Photovoltaic Generator Powering An Under Water Centrifugal Pump Coupled To A Synchronous Motor Using A Cur-rent Inverter7', Proceedings of the International Conference on Electrical Machines, pp.1445-8, September 1986. [31] Robert H. Comstock," Trends in Brushless PM Drive And Motor Technology", MOTOR-CON 1986 Proceedings, pp.1-14, October 1986. [32] A. C. Stone and M. G, Buckley, "Novel Design and Control of A Trapezoidal Back EMF Motor - The smooth transition from brush to brushless", MOTOR-CON 1985 Proceedings, pp.86-95, April 1985. [33] S. Singer and A. Braunstein, " A General Model of Maximum Power Point Trackers", MELECON '85/Volume IV: Solar Energy, pp.147-51, 1985. [34] E. Olivier, M. Mahmoud and J. Perard, " Optimizing A System 'Photovoltaic Generator-Power eidth modulation inverter-Asynchronous Machine' Thanks To A References 105 Maximum Power Point Tracking" , Proceedings of International Conference on Elec-trical Machines, pp.859-62, 1984. [35] Andreas F. Boehringer, "Self-Adapting DC Converter for Solar Spacecraft Power Supply", IEEE Transactions on AEROSPACE AND ELECTRONICS SYSTEMS, Vol. AES-4, NO. 1, pp.102-11, January 1968. [36] S. Merlina, G. Peluso, C. Rossati and A. Sorokin, "Pumping Station for Fresh Water Supply", International Journal of Solar Energy, 1985, Vol. 3, pp. 164-72, 1985. [37] J . Garcia, L. Castaner, A. Poveda and I. Guruceaga, "Improvements in Switch-ing DC-DC Converters for Photovoltaic Applications", MELECON '85/Volume IV: Solar Energy, pp. 133-36, 1985. [38] R. Hill and N. M. Pearsall, " Photovoltaics - Present status and Future Prospects", Proceedings of the Fifth International Conference on Energy Options, pp.83-7, April 1987. [39] D. L. Pulfrey, P. R. B. Ward and W. G. Dunford, "A Photovoltaic-Powered System for Medium-Head Pumping>\ Solar Energy, Vol. 38, NO. 4, pp.255-65, 1987. [40] Dean De Galan, "Optimizing Commutation Efficiency in A DC Brushless Mo-tor", ELECTRO/86 and MINI/MICRO Northeast 1986 Conference Record, Vol. 11, pp.16/1/1-5, 1986. [41] William A. Passmore, " A DC-DC Converter Suitable For Controlling A Photo-voltaic Powered Pumping System", Master of Applied Science thesis, Department of Electrical Engineering, the University of British Columbia, March 1989. [42] G. Liu and W. G. Dunford, "The Investigation Into Position-Sensorless Operation of A Brushless DC Motor*', Proceedings of Canadian Conference on Electrical and References 106 Computer Engineering, pp.326-29, November 1988. [43] W. G. Dunford and P. R. B. Ward, "The Efficient Use of a DG Motor in a Low Voltage Pumping Application", Conference Record of the 1987 IEEE Industry Ap-plications Society Annual Meeting, Part I, pp.477-80, October 1987. [44] R. Hanitsch, " Calculations and Experimental Data of Water Pumping Systems with Respect to Photovoltaic Converter Arrays", 2nd E. C. Photovoltaic Solar Energy Conference, pp.1046-53, April 1979. [45] S. K. Dey, J. Kott and J. H. Simpson, "Design Features of A Photovoltaic Water Pumping System for Western Canada", Proceedings of ENERGEX '82, Solar Energy Society of Canada, Vol. I/II, August 23-29, 1982. [46] E. P. Evans, P. L. Fraenkel, E. M. Mitwally and M. W. Duffy, "The Development of A Practical Mathematical Simulation Model for the Evaluation of Small Scale Solar Photovoltaic Pumping Systems", Solar World Forum, Vol.2 pp.1140-45, 1981. [47] A. Derrick, P. L. Fraenkel, B. McNelis and M. R. Starr "Small Scale Solar Water Pumps - the State of the Art", Solar World Forum, Vol.2 pp.1146-50, 1981. [48] A. Derrick, R. J . Hacker, A. P. Napier and A. R. O'Hea, "Performance Monitoring of Small-scale Solar-Powered Pumping Systems", Solar World Forum, Vol.2 pp.1187-92, 1981. [49] W. G. Dunford and P. R. B. Ward, "Final Report to Science Council of British Columbia on DC-DC converter", 11 pages, 1985. [50] W. Passmore and W. G. Dunford, " The Design of A Maximum Power Point Tracking Circuit for A Photovoltaic Powered Pumping Application", Conference Record of References 107 the 1987 IEEE Industry Applications Society Annual Meeting, Part I, pp.816-20, October 1987. [51] Marwan Mahmoud, "Substituting Diesel Engines with Photovoltaic Power Systems in Water Pumping from Jordan Desert Wells", Advances in Solar Energy Technology (Preceedings of the Biennual Congress of the International Solar Energy Society), vol. 3, pp.2476-9, 1987. [52] Damitha P. Liyanage and G. R. Whitfield, " Photovoltaic Water Pumping System Simulation, Design and Testting ", Advances in Solar Energy Technology (Preceed-ings of the Biennual Congress of the International Solar Energy Society), vol. 3, pp.2467-70, 1987. [53] Bernard McNelis, " Solar Water Puming—the current status", Advances in Solar Energy Technology (Preceedings of the Biennual Congress of the International Solar Energy Society), vol. 3, pp.2355-6, 1987. [54] Matthew Buresch, Photovoltaic Energy Systems - Design and Installation, McGraw-Hill Book Company, 1983. [55] M. A. Green, Solar Cells: operating principles, technology, and system application, Prentice-Hall, 1982. [56] S. A. Nasar, Handbook of Electric Machines, McGraw-Hill Book Company, 1987. Appendix A Simulation study on maximum power operation of a P V array The power extracted from a PV array is a function of resident radiation, array tempera-ture and the operating point (current and voltage) on the PV array. For given radiation and temperature, different operating voltage (operating point) results in different output power. There is one and only one operating voltage V^t at which maximum available power from the PV array can be extracted. The value of Vopt varies with temperature and radiation. Appendix A describes the simulation results of the effect of temperature and radiation on Vopt. The PV array model developed in MIT [54] was used in the simulation. The effect of radiation on Vopt is shown in Figure A.51. Figure A.52 shows influence of temperature on Vopt. It can be seen from Figure A.51 that as radiation varies from zero to lkW/m2, Vopt varies from zero to 156 volts (array temperature being 25 °C). However, in a pumping system, a minimum power is required to start the pump. When radiation is lower than 0.2 kW/rn2, the PV array output is usually smaller than that minimum requirement. Therefore interest is mainly in the Vopt changes when radiation is above 0.2 kW/m?. Figure A.51 changes from 132 to 156 volts when radiation changes from 0.2 to 1 kW/m2. Figure A.52 shows a roughly linear relation between Vopt and array temperature. When the temperature increases from -30 to 70 °C so that Vopt decreases from 193 to 126 volts, the array temperature affects Vopt in the opposite direction to the radiation. The array temperature increases with the increase of radiation during the day. The increase of the array temperature tends to reduce Vopt while the increase of radiation 108 Appendix A. Simulation study on maximum power operation of a PV array 109 *» o RADIATION kxu/sq.m. Figure A.51: The Effect of Radiation on Maximum Power voltage tends to increase V^*. This complementary effect results in a relatively small variation of Vgpt during a day. When a PV array operates at a voltage other than its maximum power voltage, the power output is smaller than the maximum power available from the array. Let Pd denote the difference between the maximum available power and actual power from the PV array. Figure A.53 which shows the power loss vs. operating voltage, indicates that: when array voltage is 5 volts (2.5% of open circuit voltage) higher than Vopt, Pd is 5 Watts (0.5% of total output); when the operating voltage is 10 volts higher than V ^ , the power loss is 34.1 Watts (2.4 % of total output); if the array operating voltage is 20 volts (10% of open circuit voltage) above Kp t , the power loss will be 168.1 Watts (11.7% of total power). Figure A.52 shows that when the temperature decreases 15 °C, Voptwill increase by over 10 volts and nonnegligible power loss may result. It would thus be helpful to investigate the typical daily operation of the system. For the simulation of daily operation of a PV pumping system, clear days are assumed. Appendix A. Simulation study on maximum power operation of a PV array 110 Figure A.52: The Effect of Temperature on Maximum Power Voltage Figure A.53: Power Loss vs. Operating Voltage Appendix A. Simulation study on maximum power operation of a PV array 111 HOUR OF THE DAY Figure A.54: A Sample of the Radiation and Temperature in Vancouver The radiation and temperature recorded on an autumn day in Vancouver is plotted in Figure A.54. Based on this weather pattern, simulation results show that, during the time the pump is running, varies between 131 to 141 volts. If the voltage is fixed at 136 volts, the daily energy output will be about 0.5% less than that with optimum operation. In other words, the power loss is negligible in this case. However, for the four seasons of the year, the temperature changes significantly. If the radiation and temperature of the four seasons are approximated as follows: Rad = A cos[ir(t - 12)/SSH] (A.55) Tami = Ti + Tmcos[*(t-U.5)/12) (A.56) Tarr = Tamb + 25Rad (A.57) where Rad is the radiation on latitudely tilted, south-facing surface; Tamb is the ambient temperature; TaTT is the array temperature; A is the peak radiation of a day; SSH is Appendix A. Simulation study on maximum power operation of a PV array 112 sunlit hours; Tj, is a base temperature ( temperature at 8:30 am); and Tm is the peak temperature deviation of a day. For a day in March the following parameters are assumed: A = 1.06 (kW/m2) (A.58) SSH = 12 (hours) (A.59) T 6 = 15 (°C) (A.60) Tm = 8 (°C) (A.61) For a day in June the following parameters are assumed: A = 0.98 (kW/m2) (A.62) SSH = 15 (hours) (A.63) T 6 = 25 (°C) (A.64) Tm = 10 (°C) (A.65) For a day in September the following parameters are assumed: A = 1.03 (kW/m2) (A.66) SSH = 12 (hours) (A.67) Tb = 12 ("O (A.68) Tm = 8 CC) (A.69) For a day in December the following parameters are assumed: A = 0.8 (kW/m2) (A.70) SSH = 9 (hours) (A.71) Appendix A. Simulation study on maximum power operation of a PV array 113 o W o r o > o >- 6 < » (X <: June "Se'ptember" December la 16 ao HOUR OF THE DAY Figure A.55: Variation of V^t with the Four Seasons Tb = 0 (°C) (A.75) T m = 6 (°C) (A.76) Figure A.55 shows the simulated variation of Vopt in the four days in the four different months. The assumed seasonal temperature variation is moderate (25 °C variation in base temperature). It can be seen that Vopt varies from 132 volts in the summer to 155 volts n the winter. If the array voltage is controlled at a fixed voltage of 142 volts for the whole year, the loss due to non-maximum power operation is 0.2% for the simulated March and September days; 2.1% for the simulated day in June, and 2.7% for the day in December. These simulation results show that temperature is the major factor causing a consid-erable shift of Vopt in different seasons of the year. In order to achieve maximum power operation throughout the year, a maximum power tracking function is necessary. Appendix B Flowcharts Some flowcharts described in Chapter 4 are shown in this appendix: ^ E N T R Y ^ Setup stack pointer set port A, port B to input ports I Set auxiliary cont. reg. in V IA l Set interrupt vector 1 Set port A of VIAO 1 Initialize system variables EXIT Figure B.56: Flowchart for INITIALIZE 114 Appendix B. Flowcharts OUTDRDY E N T R Y 1 D l < = D r d y O U T D OUT D E N T R Y S e t p o r t A o f VIA1 to o u t p u t p o r t I PA1 <= D l I E n a b l e D / A c o n v e r t e r E X I T S e t p o r t A to i n p u t p o r t E X I T Figure B.57: Flowcharts for OUT JDRDY and OUT JD Appendix B. Flowcharts V_SAM Figure B.58: Flowchart for VUSAM Appendix B. Flowcharts 117 OUT_D T VI SAM Store V, I I P <= I*V MPW<=P IMP <= IOUT VMP <= VOUT DMP <=D1 ^ENTRY^ I Set starting adrr. of the mem. area I D1<=0 1 3 1 maximum power MPW <=0 I Set delay TCNR1 VI_SAM Store V, I • Decrement TCNR1 no ^T C N l f yes Decrement Dl yes ^ ? Dl<DLOW OUT_D EXIT Figure B.59: Flowchart for CURVE Appendix B. Flowcharts 118 Figure B.60: Flowchart for FAUJDT Appendix B. Flowcharts STL DT 1 CHKRTR Decrement counter ZCNR n o ^ ' ? VABRA>IOUT\ re-initialize ZCNR Set STLFto'O' SetSTLFto'l' EXIT Figure B.61: Flowchart for STLJDT Appendix B. Flowcharts 120 CHKRTR ^ENTRY^ V A B R A < = D l ^ . m ^ y o u r 256 256 Q EXIT ^ Figure B.62: Flowchart for CHKRTR R E V J 3 T Figure B.63: Flowchart for REVJ3T Appendix B. Flowcharts f ENTRY ) 1 Decrement counter TEDLY — OTF <= 1 Re-initialize OTCNR Figure B.64: Flowchart for TEMPJDT Appendix B. Flowcharts 122 FAULTS Figure B.65: Flowchart for FAULTS Appendix C Recorded Operating Points of the P V Array The following is a list of the operating points recorded on July 27th, 1988. The sampling period was approximately 0.33 Hz. These operating points are also shown in Figure 5.42 along with the recorded I-V curve of the PV array. Voltage current kth voltage current sample (V) (A) sample (V) (A) 1 69 . 50 2 . 27 26 73 . 50 2 .15 2 70.00 2 . 27 27 73 . 00 2 . 15 3 69 . 50 2 . 27 28 74 . 50 2 .08 4 72.00 2 . 19 29 73 . 50 2 .15 5 69 . 50 2 . 27 30 73 . 00 2.15 6 69 . 50 2.25 31 74 . 50 2 .08 7 71 . 50 2.21 32 73 .50 2 .15 8 72 .00 2.21 33 73.00 2 . 15 9 71 . 50 2 . 21 34 74.50 2.08 10 73 . 00 2 .15 35 73 . 50 2. 15 11 71 . 00 2 . 23 36 73 . 50 2 .15 12 71 .00 2 .23 37 71.00 2.21 13 70 .00 2 . 25 38 73 . 50 2.13 14 72.00 2 .19 39 71.00 2.21 15 69 . 50 2 . 27 40 71 . 00 2.21 16 69.50 2.25 41 69.50 2.25 17 72 . 00 2 . 19 42 72.00 2 .19 18 71 . 50 2 . 21 43 72.00 2.19 19 71 . 50 2 . 21 44 7 2.50 2.17 20 69 . 50 2.25 45 71. 50 2.19 21 72 . 00 2.19 46 73.00 2 .17 22 72 . 00 2 . 19 47 73.00 2 . 15 23 73 .00 2.17 48 74 . 50 2.08 24 72.50 2 . 17 49 73 . 00 2.15 25 74 . 50 2 . 06 50 73 . 50 2 .15 123 Appendix C. Recorded Operating Points of the PV Array 124 Voltage current kth voltage current sample (V) (A) sample (V) (A) 51 73 . 00 2 . 15 93 73 . 00 2 . 15 52 74 , , 50 2 . 08 94 73 . 00 2 , . 17 53 73 . 50 2 .15 95 73 . 00 2 . 15 54 73 . 00 2 . , 15 96 74 . 50 2 , . 11 55 74 . 50 2 .11 97 74 . 50 2 . 08 56 74 . 50 2 . . 08 98 75 . 50 2 . .00 57 76 . 00 2 . 00 99 75 . 00 2 . 06 58 75 . 00 2 . .06 100 75 . .00 2 , .06 59 75 . 00 2 . 04 101 73 . 00 2 . 17 60 75 . 50 2 . . 02 102 73 . , 00 2 , . 15 : 61 75 . 00 2 . 04 103 74 . 00 2 . 11 62 75 . 00 2 . , 06 104 73 . 50 2 , .15 : 63 75 . 00 2 . 06 105 73 . 50 2 .15 64 73 . 00 2 . . 15 106 71 . 00 2 , . 23 65 73 . 50 2 . 15 107 71 . 00 2 .21 66 73 . 50 2 . . 15 108 73 . , 50 2 , . 15 67 74 . 50 2 . 11 109 73 . 50 2 . 15 68 69 . 50 2 , . 27 110 74 . , 00 2 , . 11 69 72 . 00 2 . 19 111 73 . 00 2 . 15 70 69 . 50 2 . . 27 112 73 . 00 -> . 15 71 69 . 50 2 . 27 113 71 . 00 2 .21 72 68 . 50 2 , . 29 114 71 . 00 2 . .21 73 70 . 50 2 . 23 115 69 . 50 2 .25 74 70 . 50 2 , . 25 116 72 . 00 2 . .21 75 70 . 00 2 . 25 117 72 . 00 2 . 21 76 71 . 50 2 . .21 118 73 . 00 2 . , 17 77 71 . 50 2 .21 119 71 . 50 2 .21 78 73 . 50 2 , . 15 120 73 . 50 2 . 15 79 71 . 00 2 .21 121 71 . 00 2 . 23 80 73 . 50 2 , .15 122 71 . 00 2 . 23 81 73 . 50 2 . 13 123 70, .00 2 . 25 82 74 . 00 2 , . 11 124 72 . 00 o 19 83 73 . 00 2 . 15 125 69 . . 50 2 . 27 84 73 . 00 <L , . 13 126 70 . 00 2 . 27 85 74 . 50 2 . 11 127 69 , . 50 2 . 27 86 74 . 50 2 , . 08 87 75 . 50 2 .00 88 75 . 00 2 .  06 89 74 . 50 2 . 06 90 75 . 50 2 , .02 91 75 . 00 2 . 06 92 75 . 00 2 , . 06 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items