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

Download

Media
831-UBC_1990_A1 L58.pdf [ 7.69MB ]
Metadata
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
831-1.0065498-fulltext.txt
Citation
831-1.0065498.ris

Full Text

A PHOTOVOLTAIC-POWERED PUMPING  SYSTEM  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 PARTIAL F U L F I L L M E N T O F T H E REQUIREMENTS FORT H E DEGREE OF  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  T H E 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 The University of British Columbia Vancouver, Canada Date  DE-6 (2/88)  QN6M££RMq )  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 reducing the maintenance requirements of the electrical subsystem. The reduction of maintenance 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 brushtype dc motor systems.  ii  Table of Contents  Abstract  ii  List of Figures  vi  List of Tables  ix  Acknowledgement  x  1  Introduction  1  1.1  3  1.2 2  An Overview of the Experimental Pumping System Objectives and Scope of the Thesis Research  4  Design Considerations  6  2.1  6  System Configuration 2.1.1  Existing Pumping Configurations  6  2.1.2  Selection of System Components  8  2.1.3 2.2  Brushless DC Motor Drive  2.4  12 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 2.3  The Selected PV Pumping Configuration  Position-sensorless Operation  The Converter-commutator Circuit Maximum Power Tracking Realization iii  25 26 30  2.5  z.o  2.4.1  The Need for Maximum Power Tracking  30  2.4.2  A Double-loop Maximum Power Tracking Scheme  31  System Losses and Efficiency  35  2.5.1  Maximum Efficiency from the Motor  . . .  2.5.2  Losses in the Converter-commutator Circuit and Motor . . . . . .  36  bummary  39  3 Hardware Design 3.1  35  41  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  3.6  PWM Signal Generation . . .  53  3.7  Microprocessor System Hardware  54  3.8  I/O Interface and Signal Conditioning  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  . . .  .  51  54  69  4.1  Functional Description  69  4.2  Description of Major Routines  72 iv  4.3 5  Main Routine  72  4.2.2  Modules in READY State  74  4.2.3  MOTOR.ST  4.2.4  VCONT, ADJUST and POWER  4.2.5  Fault Detecting and Handling Modules  ,  . . . .  Summary  75 79 83 84  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  5.4 6  4.2.1  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  Summary  95  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 P V 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 K  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  f  and Curve-fitted K  f  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 3.29 Voltage, emf and Position Signal Waveforms for Backward Rotation 3.30 Ring Counter for Motor Starting  . . .  63  . .  64  . .  3.31 State Diagram for the Ring Counter . . . . •  65 .  ,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  .  5.42 Recorded PV Array Operating Points in A Period of 6 Minutes 5.43 PV Array Operating Points During Sudden Change of Radiation  81 87  . . . .  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  Ill  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  L i s t 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  5.4  Combined Converter-motor Efficiency  90  5.5  Measured Efficiency for Sinusoidal and Trapezoidal Excitation  93  ix  . . .  26  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) pumping 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 existing PV pumping systems, have the advantages of high efficiency and simple control circuitry. However, these motors have an inherent drawback - the maintenance requirement 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 commutation 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 controlled rectifier (SCR) or bipolar junction transistor (BJT) version. With the use of three HEXSenses (current sensing power MOSFETs from International Rectifier Corp.) in the converter-commutator circuit, the PV array current is sensed with minimum power loss. Furthermore, by making use of the current sensing 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 brushtype 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 conventional 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. trapezoidal 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 suitable system configuration, the control of the brushless dc motor drive, the convertercommutator circuit, maximum power tracking realization and loss minimization.  2.1  System Configuration  In this section, the existing PV pumping configurations are examined first. Then major 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 Figure 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 output power of a given array area can be substantially boosted by using concentrator and sun-tracking mechanisms [4]. However, concentrator and sun tracking mechanisms require extra components such as heat sinks and servo systems and they also consume  6  3 to  sun battery  tracking  to i  <§  y  ? CL  PV ARRAY  converter or inverter  MOTOR  mechanical  buffer  transmission  tank  cn  water  concentrator  s  control  level  circuit  detector  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 induction 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 permanent 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 mechanical 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 developments 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 P M 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 information 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 efficiency. 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  Brushless PM Motor  SUN  Progressive Cavity Pump PV ARRAY  I A/D  I  A/D  MC6809 Microprocessor Development System  HEXFET COMVERTERCOMMUTATOR  I D/A  CONTROL ELECTRONICS  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 M o t o r 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 P M motor drive can be categorized 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 theory. 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  15  Considerations  emf should be trapezoidal. However, due to some manufacturing complications associated 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 generated 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 P M 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 2TT JO  E  m  sinut I sinwt d[wt] = ^ ^rnE m  (2.1)  m  2  For trapezoidal excitation, the total power output can be calculated by averaging the power output within a 60 electrical degree interval: Ptra  P  = -  f  2  [i e A  + i e  A  B  B  + i ec)d[wt)  (2.2)  C  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): Ptra  P  = ~ /  f  [I E^sinwt a  - I E sin(wt a  m  - ^ ) ] d[ut] =  Z  s  /  *  I  a  E  m  (2.3)  where J , E are the peak phase current and emf respectively; I is the average value of m  m  a  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: Isr . m  = %  (2.4)  Chapter 2. Design Considerations  Figure 2.4: Current and emf Waveforms for Trapezoidal Excitation  17  18  Chapter 2. Design Considerations  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 P.,... = V+Hf.  (K, + ^  T  T  + 0.5^)  (2.8)  where / , is the switching frequency in Hertz, K is the charge constant of the power q  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: (2.9)  h = h  V  E  a  &~-f  =  0.954V3 E £  m  (2.10)  where E is the equivalent dc motor emf can be related to the phase emf oi the brushless a  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:  P . .. T LO  where  PTIIOU  = l.«5^ /.(jr A  ^  f +  \2K I q  a +  (2.11)  0 J S ^ )  is the switching loss in trapezoidal excitation in micro-Watt.  For sinusoidal excitation: V f*2E de  I = \/2I L  Srmt  (2.12)  m  (2.13)  sin(ut)  J2/Zl  a  Substituting relations 2.12,2.13 and 2.14 into 2.8 yields:  Ps,io»»(wt)  2.2K I sin(u>t) = 2.2E - f, • I [K sin(u>t) + sin(ut) \ di/dt q  m  a  a  q  0.557 sin (u>t) ] di/dt 2  o  (2-15)  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):  20  Chapter 2. Design Considerations  Ps,io„ =  — / [K avn(wt) q  JO  7T  2.2K I sin{wt) +sin(wt) \\ di/dt q  .  i ^  w  +  M  T  l.ll sin (u>t) 2di/dt 2  a  ^ ^  +  a  J  M  M  lU,J  _ 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:  E  m  = 58V, / , = 6A5kHz, K = 0.7pC/A, di/dt = lOOA/pS q  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 excitation, only one leg is switching at frequency  in trapezoidal  Another factor is that, for a given  E , the required dc supply voltage V, for sinusoidal excitation is higher than that required m  for trapezoidal excitation. Friction, windage and iron core losses for the two excitation methods are considered 1  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 excitation 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 1  iion 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) Trapez. loss (Watt) Sinus, loss (Watt)  1 0.510 1.282  2 1.084 2.719  3 4 5 7 6 8 1.706 2.365 3.058 3.782 4.534 5.313 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 conventional (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 converter. Some parameters of a brushless dc motor can be defined as follows [24] (SI metric units are used): Voltage and torque constants: 0.954EMF  v  Ke =  — a T K =K t  peak  (2.17)  -  (2.18)  e  where EMFpeaj, is the peak value of motor line to line em/, Q is the steady state angular m  speed of the motor.  22  Chapter 2. Design Considerations  Right-hand side of A-A' is equivalent to a dc motor la SW1  SW3  SW5  Power supply  Brushless PM motor  B SW4  Electronic commutator  A'  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): K = 0.0007 + ° '  1 8 4 1  f  20 < <v < 200 rad./sec. m  (2.19)  where u> is the instant angular speed of the motor. The curve-fitted Kf and measured m  Kf are illustrated in Figure 2.6. Equivalent armature resistance and inductance: R = 2Rpha, + 2J?u( ) a  £  e  a  =2(L  p W  (m  -M)  (2.20) (2.21)  Chapter 2. Design Considerations  23  0.0100  0.0000H  0  1  1  1  1  50  100  150  200  Motor angular speed  250  ( 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 N is the number of poles of the motor. p  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 Symbol  Description Rated power Rated torque Voltage constant Torque constant Phase resistance Phase inductance Thermal resistance Moment of inertia Friction-windage const.  Prate T te K K Pphate Lphase r<1 e  t  RT  J Kf  Unit  Value  kilowatt 1.5 N •m 4.5 volt • sec./rad. 0.52 N • m/Ampere 0.52 Ohm 0.5 ± 5% mH 1.2 °C/Watt 0.67 kg • m 0.008 JV-m 0.001-0.008 (varies with w ) Rad./tec. 2  m  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 r t)Ra {  m  (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  25  Chapter 2. Design Considerations  0  5 0 0  1 0 0 0  Motor speed  1 5 0 0  2 0 0 0  ( RPM )  2 5 0 0  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 position 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 active phases rotor position  3,4 +B,-A 3  4,5 +C,-A 4  5,6 +C,-B 5  6,1 +A,-B 6  1,2 +A,-C 1  2,3 +B,-C 2  in one electrical cycle. However, the use of a position sensor requires extra wire connections 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  cor  B  0  cor  VC  0  cor  (a) Motor terminal voltages without PWM (all voltages are with reference to ground) VA  0  cor  VB  0  Ui KJ COf  vc  0  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) signal 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 armature 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 adjusted 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 E in Figure 2.9 is the equivalent average back emf oi the brushless motor; L and a  a  R are the equivalent armature inductance and resistance respectively. C is a capacitor a  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, dt  i,(v ) — Si C a  (2.24)  a  di _ Sv„ — Rglg — K u) ~dt ~ Lg dwm K i — KfU> — TL dt J a  e  t  a  (2.25)  m  (2.26)  m  where 8 is the duty ratio of the converter, u> is the motor angular speed, R~ and K m  e  t  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 8v and 6i . Besides Kf varies with mechanical speed w . t  a  m  Letting all the derivatives be zero leads to the following steady-state model of the brushless dc drive: (2.27)  V. = g{I ) t  / . = £>/„  DV, - R I - K n a  KJ  a  e  a  (2.28)  - KfSl - T m  = 0  m  L  (2.29)  = 0  (2.30)  where D,V ,I I and f i are the steady state values of 8,v ,i,,i and u> . t  t)  a  m  t  a  m  Equation 2.27 is the I-V curve of the PV array. D V, in Equation 2.29 can be considered as the armature voltage of the equivalent dc motor: V = DV, a  (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:  KH  -  (2 32)  is satisfied (assuming temperature and radiation are constant). P = V,J is the output t  t  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 2.4.1  Maximum Power Tracking Realization 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/m and the 2  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) = T  ambient  {°C) + 30 x Intensity(kW/m ) 2  (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 M a x i m u m 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  32  Chapter 2. Design Considerations  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 V  t  close to the set point voltage V and suppresses the effect of the disturbance in load or r  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(V ) is a nonlinear function with single maximum t  in the interval 0 < V < V^.; V^. is the open circuit voltage of the P V array, and e(t) 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)  - v (k)}  (2.35)  t  where D(k + 1) and D(k) are the new and old duty ratio values respectively, V (j) is r  the new reference voltage for the fast voltage loop, V (k) is currently measured array a  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 V . T  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  Q^  dP  should indicate the correct direction for V to move. When t  = 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  8  SUNLIGHT  •8  to Is(k)  A/D  Update timer  signal conditioning  IPV  Vs(k)  ARRAY  Vs(j) IsG)  A/D  u  Extremum Control Loop  Vs(k) Voltage Loop  Ps(j)=Vs(j)*Is(j); V )=VrO-l) AV*sign{J^- ^:- - -} Vs(j)-Vs(j-1)' )  r(j  1 )  +  Vr(j)  D(k+l)=D(k)-Ki* [Vr(j)-Vs(k)]  D/A  D(k+1)  signal  V  3*  S  conditioning signal  conditioning  CL  e-+-  fx.  Convertercommutator circuit I  § 2.  S  J5  Brushless PM motor  Fault diagnosis; Sensorless starting; Data recording; Displaying  MICROPROCESSOR SYSTEM  +T,a>m L  Progressive cavity pump Water output ^  Figure 2.10: Block diagram of maximum power tracking scheme co oo  34  Chapter 2. Design Considerations  pumping system: V (j) = V (j - 1) + AV • sign{ ™ r  r  ~_  ~ *)  }  (2  . 6) 3  where V (j) and V (j — 1) are the new and old reference values to the voltage loop, AV is r  T  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/m ). Therefore, it is safe to select an update period between 1 to 10 seconds. 2  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(V + AV)-f(V )\>\2e(t)\ t  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 requires a smaller AV, and vice versa. 1  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.  35  Chapter 2. Design Considerations  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 measurement 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 convertercommutator 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  6peed.  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 . The output power can be calculated by multiplying Equation 2.30 by the A  motor speed fi  m  (bearing in mind that power output equals TL • ft )« From Equations m  2.30, 2.29 and 2.31, and with input power being V • J , the efficiency of the brushless dc A  A  36  Chapter 2. Design Considerations  motor can be estimated by:  -^x ~ ^r - -^r  Vm=1  By calculating  2Ra+  )  I  (2>38)  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 V ff is the maximum-efficiency armature voltage, the values of K and R can be e  e  a  found in Table 2.2. For a given armature current J V ff can be estimated with Equation ai  e  2.39. For example, when /„ = 4.5A, V ff would be 85 volts according to Equation 2.39. e  Laboratory tests show that the maximum-efficiency voltage at the converter input for I  a  = 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 freewheeling 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 circuit diagrams for these two configurations are shown in Figure 2.11 (a) and (b) respectively. Thefirstconfiguration uses the internal diodes of the power MOSFETs (HEXFETs III from International Rectifier Corp.) for freewheeling. The advantage of this configuration 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  Q5  QI  Brushless PM motor  B Q4  (a) Power circuit using HEXFET III internal diodes  QI £ U i  D F 1  ? ! j j ^ D3  Q4  i-h  Q6 t-h  Bf  ^Sf  :  D F 3  B  Q5j  DF5  Brushless P M motor  D5  Q i-H 2  H5f  (b) Power circuit using external freewheeling diodes  Figure 2.11: Two Switching Configurations for the Converter-commutator  38  Chapter 2. Design Considerations  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: P a = SRolZw c  + (1 - 8)(^ll  {rmt)  + V*J.  (rm  .>)  (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  Summary  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  £ ^ w  o = using FET body diodes a = using external diodes 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 temperature. 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 10 ). 9  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 connected 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 drive  IRF740 Kgg i  0-  R  n^40,£n Cl  10  C2  Rg  from pin5ofIR2110A  To a motor terminal  (VS) from gate drive  IRF740  1 0  IRC740  10g  To current sensing circuit  r Rg  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: 1I  L  x  T  ch  (3.44)  where AV is the P V array voltage ripple, T h is the chopping period and Ii is the load P  c  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  3.2  46  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  47  Chapter 3. Hardware Design  The schematic diagram of the gate drive stage is shown in Figure 3.17. The key component in the gate drive stage is IR2110. 'LIN' and'HIN' are the input pins for controlling  +12V  RI  C3 lOOuF  To source of the top F E T •  o—H—vVW^-f—ill 10  CI luF *SW1  4001  VB  O  *SW4  VS  HJN 4001  O  jLI| L2  S  D  .  R3 47  LO VDP luF  -tr  To gates of the top FETs  IR2110A  LIN VSS  OVERCURRENT  R2 47  HO  11  To gates of the botton FETs  VCC  T  C4 lOOuF  +12V  Figure 3.17: HEXFET Gate Drive Schematic Diagram the bottom F E T and the top F E T respectively.  The 10 ohm resistor in the gate of  each F E T 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 F E T 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  48  Chapter 3. Hardware Design  IR2110. Shielded wires are used for the connection between the gate drive output to the gate resistor.  Over-current Protection  3.3  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.  VDD  o  current  signal  :  RI ioo  O-HVW^VDD  MC34072  X  s 4013 _  O—i  R2 5k  Ql l  D R  Q  2  OVERCURRENT •  (J) PWM  Figure 3.18: Over-current Protection Schematic Diagram  49  Chapter 3. Hardware Design  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  / O  Q sen4  9 sen2  sen6  9  *  t e m  P  Iovercurrent  oa Isen •47  10k  3 h  47: :  r  lOOuF  Rll.  R10^47 10k  2.k2  2k2  MC3407: +  IRFD120  2k2: I—1  -*  »  /VVvv100k  •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  3.5  51  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 P W M , 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  W i t h a duty ratio control ( P W M ) on the lower three F E T s and a R U N 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  •  SW2 = RUN • PWM  • (PAPC)  (PBPC)  The resulting converter-commutator logic circuit is shown in Figure 3.21.  Chapter 3. Hardware Design  53  Chapter 3. Hardware Design  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 ratio control  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:  2  .  5  RA  R5 + R6  i23 + J?4  R6  6  . ^ ! _ . ^ + B 6  J23 + RA  R6  =2 =  3  (3.42) (3.43)  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  3.7  54  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 interface 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 V I A l , 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 microprocessor 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 a n d S i g n a l C o n d i t i o n i n g  I/O interface and signal conditioning circuit transfers information between the microprocessor 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 P V array voltage and current are analogue signals and the duty ratio control also requires an analogue  Chapter 3. Hardware Design  55  is TEMP PUL  1  REV  D/A  OPER/ 4 f  A/D  A/D  MairuTame computer  START  PAO  PBO  PA1  PB1  VIAO  VIA1  (6522)  (6522)  RS232  TT 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-passfilterand 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  duty ratio control  bit 0, port A, VIA1 D/A AD588 bit 7, port A. VIA1  10 9 11 InF  T  — • — • bit 3, port A, VIAO +5V  'I CAl  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  58  Chapter 3. Hardware Design  current signal  +5V Vref  INTR in port B  0  Q  T  ^  INTR CS WR 3 -2 VREF RD VI+ DBO CLK 1 8  ^ bit 1 of port A of VIAO • bit 2 of port A of VIAO - bit Oof port B of VIAl  ADC0804 CLKR VIf-*jAGND  DB7 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.  59  Chapter 3. Hardware Design  7  6  5  4  OPER/START REV  3  2  _  1_ RD WR  1_ EN  1  0  • TEMP  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 convertercommutator 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. v vj, and ai  v in Figure 3.27 are the motor terminal voltages in reference to ground. C l and R3 e  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 v lags that of phase A em/30 degrees a  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  +5V  motor terminal A  68k  100 PA'  + v,  Ai i  2.2k:  motor terminal B ~, 68k  .115uF Cl  --vW-  Af  OluF  + BA  2.21 motor  .115uF  :  +5V  100  4.7k V  V  terminal  R3 4.7k  PB'  Bfj  .OluF  :  +5V  | C  +v  -  6  8  k  2.2i  100  4.7k  v  .115uF  I  cf|  .OluF  r  PC  47k 4  7  k  A»A/V  47k  Figure 3.27: Schematic of Rotor Position Signal Generator  62  Chapter 3. Hardware Design  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 voltages, 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 Position: 3  4.5 4  5,6 5  6,1 6  1,2 1  2,3 2  3,4 3  Figure 3.28: Voltage, emf and Position Signal Waveforms for Forward Rotation  Chapter 3. Hardware Design  SW ON: Position: v A  5,4 6  4.3 5  3,2 4  2,1 3  1,6 2  6,5 1  5,4 6  ii  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.  66  Chapter 3. Hardware Design  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 convertercommutator 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 3.0.4  68  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 theflowchartsof 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 algorithmic state machine (ASM) diagram as shown infigure4.34. 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 P V 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 V  . When the system is in READY state, the microprocessor and control electronics  mng  are connected to the P V 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. Normal 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 implemented 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 microprocessor 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 condition 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  72  Chapter 4. Software Development  rotor can be achieved by checking the following inequality: 6 v. < i.R 2  a  (4.47)  Theoretically, the left-hand side of Inequality 4.47 cannot be less than the righthand side. However the "less" condition is helpful for microprocessor implementation. 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  Figure 4.35: Flowchart for Main Routine  73  Chapter 4. Software Development  4.2.2  74  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 . Theflowchartsfor OUT JDRDY and OUTJD are shown in figure B.57 in Appendix B.  VIJSAM 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.  CURVE 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. Theflowchartfor 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  MOTOR_ST  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. Theflowchartof 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 1r  OPER/START set to low i  Dl<- DST  f OUT_D  r Reset pulse countr: PNUM Disable interrupt Initialize reverse counter  }  OPER/START Initialize zero  set to high  duty ratio counter 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  PERI<==(15/16)*PERI PCNR<=PERI  Clear interrupt flag  ^  EXTT^  Increment pulse counter: PNUM  Figure 4.38: Flowchart for PULSE  Disable timer interrupt  Chapter 4. Software Development  4.2.4  79  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 Chapter 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 memory location VR. VCONT calls a module ADJUST which implements the extremum control 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 D l . The flowchart for VCONT is shown in Figure 4.39. Theflowchartfor ADJUST is shown in Figure 4.40.  Figure 4.39 shows that a delay count DOUT is used to adjust the  ADJUST ENTRY  I  1  Decrement delay counter: ICN  Record V J  POWER  re-initialize delay counter: ICN  EXIT  Figure 4.40: Flowchart for ADJUST  Chapter 4. Software Development  80  VCONT  re-initialize DOUT  ADJUST  yes  no  1  r  DD=(Vs-Vr)*Ki  yes  DD=(Vr-Vs)*Ki  DD>LIMIT7  DD<=LIMIT  r  D1<-D1-DD  D1<-D1+DD  T ««  DEX=LIMTT  no  ^<T  Dl>255 ?  no Dl=255  ^  EXIT^  Figure 4 . 3 9 : Flowchart for VCONT  *  Chapter 4. Software Development  Figure 4.41: Flowchart for POWER  81  82  Chapter 4. Software Development  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. Theflowchartfor POWER is shown in Figure 4.41. In module POWER, the voltage reference VR is adjusted according to the sign of dP /dV . It can be seen from Figure 4.41 a  t  Chapter 4. Software Development  83  that the change in VR is afixedstep DV. DV is the microprocessor representation of A V in Equation 2.36 and should be selected as small as possible, but must be significantlylarger 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 radiation 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. Theflowchartof FAUJDT is shown in Figure B.60 in Appendix B. STLJDT is a module designed to detect a stalled rotor. Theflowchartof 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. Theflowchartof CHKRTR is shown in Figure B.62 in Appendix B.  84  Chapter 4. Software Development  When the inequality 4.47 is true, the motor is in a stalled condition. In microprocessor, v ,i, and 8 are represented by 3 separate bytes as shown in the following forms: t  v. = K • VOUT  (4.48)  i. = Ki • IOUT  (4.49)  v  S-g  .  (4.50)  where VOUT, IOUT and Dl are the three bytes representing v , i and 8. Knew constant t  t  is defined as:  =Tk  K  <- > 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 condition 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 development 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  M a x i m u m 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 lowfill-factorlargely 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. Immediately 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 n is defined u  as:  then the utilization factor of the PV array in the recorded 6-minute period can be approximated as: V u  ~ 128 x max{P (k)}  ( 5 , 5 4 )  t  where p(t) and p  mox  ( 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 P V 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  Figure 5.43: PV Array Operating Points During Sudden Change of Radiation  89  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 91.7 91.2 91.3 90.2 91.0  1.23 2.34 2.06 3.5 4.99 4.48  0.5 0.5 0.5 0.75 0.95 0.95  430 740 779 1155 1458 1550  67.3 74.6 76.9 79.6 81.2 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 5.2.1  Converter-motor Efficiency and Position-sensorless Operation 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  5.2.2  91  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 5.3.1  Motor and Converter-commutator Operation 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. Figure 5.46 shows the current and voltage waveforms for sinusoidal excitation. According to theoretical analysis, sinusoidal excitation results in lower copper loss and higher switching 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  EFFICIENCY OF THE BRUSHLESS MOTOR sensorless and conventional operation  Figure 5.44: Efficiencies with and without a position sensor  92  Chapter 5. Experimental Results  93  * ** -  <  Upper trace: motor current at mode change 5A/div.  f i t * " "  t  ... -y,. -  Lower trace: mode changing signal 2 V/div.  Time scale: lOOmS/div.  Figure 5.45: M o t o r Current D u r i n g M o t o r M o d e Switching efficiency increases w i t h duty ratio. T h e last line of the table shows that, i f the duty ratio and modulation depth are both 80%, trapezoidal excitation is 1 percent more efficient than sinusoidal excitation.  Table 5.5: Measured Efficiency for Sinusoidal and Trapezoidal E x c i t a t i o n Excitation method  D C voltage supply ( V )  D u t y ratio or modulation depth  efficiency  sinusoidal trapezoidal sinusoidal trapezoidal  80.7  0.8 0.67 0.8  74.03  0.67  75.84  0.8  76.9  trapezoidal  80.5 90.9 90.9 76.1  (%) ' 74.56 75.84  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. A n over-temperature condition is determined by the difference between these two outputs. Figure 5.49 shows the recorded voltage difference between body-resistor sensing and external resistor sensing at different M O S F E T case temperatures.  Chapter 5. Experimental Results  95  Upper trace: motor current (one phase) 5 A/div.  0'  Lower trace: three phase current sensing output (with opposite polarity to motor current) 2 V/div. Time scale: 2 mS/div.  0  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 efficiency 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 motormode 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  CO  MOSFET current in Amperes Figure 5.48: Filtered Current Signals With and Without an External Resistor scheme.  96  97  Chapter 5. Experimental Results  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 process. 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 regardless 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 convertercommutator 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 theflexibilityfor various functions and built-in "intelligence" to be implemented. The communication link between the microprocessor 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 IEE 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, Massachusetts, 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. PV Water Pumping with U  Reciprocating Volumetric (Jack) Pumps , Proceedings of 15th IEEE Photovoltaic n  Specialists Conference, Orlando, Florida, pp.1386-94, 1984. [7] Y. Roger Hsiao and Bruce A. Bleuins, Direct Coupling of Photovoltaic Power Source u  to Water Pumping System", Solar Energy, Vol. 32, No. 4, pp. 489-98, 1984.  101  102  References  [8] Ir. C. Franx, "A New Approach to Solar Pump Systems Using Submersible Motors" ,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 Conference, 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, The Quality of Load Matching in A Direct-Coupling Photovoltaic u  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?", Join 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  Industry Application,Vol. IA-21, No. 4, May/June 1985.  References  103  [17] J . Appelbaum, "Performance Characteristics of A Permanent Magnet D.C. Mot 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 Wat Supplies Using Brushles D.C. Motors", Solar Cells 13 (1984 - 1985), pp.231-51. [23] Giampiero Brentani, "Considerations on the Locked Rotor Torque Characteristics Brushless DC Motors", MOTOR-CON 1985 Proceedings, pp.107-14, April 1984.  [24] Jaroslav Tomasek, "Electro-Mechanical Performance Data for Sine- Wave Brushle 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.  104  References  [27] P. U. Calzolari, G. C. Cardinali and E. Faldella, " fiC-Controlled Direct Adaptive  Coupling of Photovoltaic Generators to Pumping Systems", MELECON '85/Volum 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 C rent Inverter ', Proceedings of the International Conference on Electrical Machines, 7  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 198 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 T  105  References  Maximum Power Point Tracking" , Proceedings of International Conference on Electrical 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 Motor", 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 Photovoltaic 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 Applications 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  107  References  the 1987 IEEE Industry Applications Society Annual Meeting, Part I, pp.816-20, October 1987.  [51] Marwan Mahmoud, "Substituting Diesel Engines with Photovoltaic Power System 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 (Preceedings 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 temperature 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/m , Vop varies from zero to 156 volts (array temperature being 25 °C). 2  t  However, in a pumping system, a minimum power is required to start the pump. When radiation is lower than 0.2 kW/rn , the PV array output is usually smaller than that 2  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/m . Figure A.52 shows a roughly linear relation between Vopt and array 2  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 , the power loss will be 168.1 Watts (11.7% of total power). t  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  Figure A.52: The Effect of Temperature on Maximum Power Voltage  Figure A.53: Power Loss vs. Operating Voltage  110  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:  T  Rad = A cos[ir(t - 12)/SSH]  (A.55)  = T + T cos[*(t-U.5)/12)  (A.56)  ami  i  m  Tarr = T  amb  (A.57)  + 25Rad  where Rad is the radiation on latitudely tilted, south-facing surface; T b is the ambient am  temperature; T  aTT  is the array temperature; A is the peak radiation of a day; SSH is  112 Appendix A. Simulation study on maximum power operation of a PV array  sunlit hours; Tj, is a base temperature ( temperature at 8:30 am); and T  m  is the peak  temperature deviation of a day. For a day in March the following parameters are assumed: A = 1.06 (kW/m ) 2  SSH = 12 (hours)  (A.58) (A.59)  T = 15 (°C)  (A.60)  = 8 (°C)  (A.61)  6  T  m  For a day in June the following parameters are assumed: A = 0.98 (kW/m )  (A.62)  SSH = 15 (hours)  (A.63)  T = 25 (°C)  (A.64)  = 10 (°C)  (A.65)  2  6  T  m  For a day in September the following parameters are assumed: A = 1.03 (kW/m )  (A.66)  SSH = 12 (hours)  (A.67)  2  T = 12 ("O  (A.68)  = 8 CC)  (A.69)  b  T  m  For a day in December the following parameters are assumed: A = 0.8 (kW/m )  (A.70)  SSH = 9 (hours)  (A.71)  2  Appendix A. Simulation study on maximum power operation of a PV array  113  o  r  W o  o  > o  >- 6 <»  June "Se'ptember" December  (X  <:  la  16  ao  HOUR OF THE DAY  Figure A.55: Variation of V^t with the Four Seasons T = 0 (°C)  (A.75)  T  (A.76)  b  m  = 6 (°C)  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 considerable 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  Someflowchartsdescribed in Chapter 4 are shown in this appendix:  ^ENTRY^  Setup stack pointer  set port A, port B to input ports  I  Set auxiliary cont. reg. in V I A 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  OUT D E N T R Y  OUTDRDY  E N T R Y  Set port A o f  I  1 D l  <=  VIA1  to o u t p u t p o r t  Drdy  PA1  <=  D l  I O U T  D  Enable D / A converter  Set port A EXIT  to  input port  EXIT  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  ^ENTRY^  I  Set starting adrr. of the mem. area OUT_D  T VI SAM  I  VI_SAM  D1<=0  131  maximum power MPW  Store V, I  Set delay TCNR1  I  Store V, I  <=0  • Decrement TCNR1  I  P <= I*V  no ^TCNl  f yes Decrement D l  yes ^ MPW<=P IMP <= IOUT VMP <= VOUT  ? Dl<DLOW  OUT_D  DMP <=D1  EXIT  Figure B.59: Flowchart for CURVE  Appendix B. Flowcharts  Figure B.60: Flowchart for FAUJDT  118  Appendix B. Flowcharts  STL DT  1 CHKRTR  no^' ? VABRA>IOUT\ Decrement counter ZCNR  Set STLFto'O' re-initialize ZCNR  SetSTLFto'l'  EXIT  Figure B.61: Flowchart for STLJDT  120  Appendix B. Flowcharts  CHKRTR ^ENTRY^  V A B R A < =  Dl^.m^your 256  256  Q EXIT ^  Figure B.62: Flowchart for CHKRTR  REVJ3T  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  122  Appendix B. Flowcharts  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.  sample 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25  Voltage  current  k  voltage  current  (V)  (A)  sample  (V)  (A)  69 . 50 70.00 69 . 50 72.00 69 . 50 69 . 50 71 . 50 72 .00 71 . 50 73 . 00 71 . 00 71 .00 70 .00 72.00 69 . 50 69.50 72 . 00 71 . 50 71 . 50 69 . 50 72 . 00 72 . 00 73 .00 72.50 74 . 50  2 . 27 2 . 27 2 . 27 2 . 19 2 . 27 2.25 2.21 2.21 2 . 21 2 .15 2 . 23 2 .23 2 . 25 2 .19 2 . 27 2.25 2 . 19 2 . 21 2 . 21 2.25 2.19 2 . 19 2.17 2 . 17 2 . 06  th  26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 123  73 . 50 73 . 00 74 . 50 73 . 50 73 . 00 74 . 50 73 .50 73.00 74.50 73 . 50 73 . 50 71.00 73 . 50 71.00 71 . 00 69.50 72.00 72.00 7 2.50 7 1 . 50 73.00 73.00 74 . 50 73 . 00 73 . 50  2 .15 2 . 15 2 .08 2 .15 2.15 2 .08 2 .15 2 . 15 2.08 2. 15 2 .15 2.21 2.13 2.21 2.21 2.25 2 .19 2.19 2.17 2.19 2 .17 2 . 15 2.08 2.15 2 .15  Appendix C. Recorded Operating Points of the PV Array  sample 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92  Voltage  current  k  voltage  current  (V)  (A)  sample  (V)  (A)  73 . 00 74 ,, 50 73 . 50 73 . 00 74 . 50 74 . 50 76 . 00 75 . 00 75 . 00 75 . 50 75 . 00 75 . 00 75 . 00 73 . 00 73 . 50 73 . 50 74 . 5 0 69 . 50 72 . 00 69 . 50 69 . 50 68 . 50 70 . 50 70 . 50 70 . 00 71 . 50 71 . 50 73 . 50 71 . 0 0 73 . 50 73 . 50 74 . 00 73 . 00 73 . 00 74 . 50 74 . 50 75 . 50 75 . 00 74 . 50 75 . 50 75 . 00 75 . 00  2 . 15 2 . 08 2 .15 2 ., 15 2 .11 2 .. 08 2 . 00 2 ..06 2 . 04 2 .. 02 2 . 04 2 ., 06 2 . 06 2 .. 15 2 . 15 2 .. 15 2 . 11 2 ., 27 2 . 19 2 .. 27 2 . 27 2 ,. 29 2 . 23 2 ., 25 2 . 25 2 ..21 2 .21 2 ., 15 2 .21 2 .15 , 2 . 13 2 ., 11 2 . 15 <L . , 13 2 . 11 2 ., 08 2 .00 2 .. 06 2 . 06 2 .02 , 2 . 06 2 ., 06  th  93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127  73 . 00 73 . 00 73 . 00 74 . 50 74 . 50 75 . 50 75 . 00 75 ..00 73 . 00 73 ., 00 74 . 00 73 . 50 73 . 50 71 . 00 71 . 00 73 ., 50 73 . 50 74 ., 00 73 . 00 73 . 00 71 . 00 71 . 00 69 . 50 72 . 00 72 . 00 73 . 00 71 . 50 73 . 50 71 . 00 71 . 00 70,.00 72 . 00 69 .. 50 70 . 00 69 ,. 50  2 . 15 2 ., 17 2 . 15 2 ., 11 2 . 08 2 ..00 2 . 06 2 .06 , 2 . 17 2 ., 15 : 2 . 11 2 .15 , : 2 .15 2 ., 23 2 .21 2 ., 15 2 . 15 2 ., 11 2 . 15 -> . 15 2 .21 2 ..21 2 .25 2 ..21 2 . 21 2 ., 17 2 .21 2 . 15 2 . 23 2 . 23 2 . 25 o 19 2 . 27 2 . 27 2 . 27  124  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

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"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0065498/manifest

Comment

Related Items