UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

An approach to improving battery life time in a PV application using high energy density double layer… Majaess, David 2006

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

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata


831-ubc_2006-0248.pdf [ 9.06MB ]
JSON: 831-1.0064954.json
JSON-LD: 831-1.0064954-ld.json
RDF/XML (Pretty): 831-1.0064954-rdf.xml
RDF/JSON: 831-1.0064954-rdf.json
Turtle: 831-1.0064954-turtle.txt
N-Triples: 831-1.0064954-rdf-ntriples.txt
Original Record: 831-1.0064954-source.json
Full Text

Full Text

A N APPROACH TO IMPROVING B A T T E R Y LIFE TIME IN A PV APPLICATION USING HIGH ENERGY DENSITY DOUBLE L A Y E R CAPACITORS by DAVID MAJAESS B. Eng., Dalhousie University, Halifax, NS, 2003 A THESIS SUBMITTED IN PARTIAL F U L F I L L M E N T OF THE REQUIREMENTS FOR THE DEGREE OF MASTER IN APPLIED SCIENCE in THE F A C U L T Y OF G R A D U A T E STUDIES (Electrical and Computer Engineering) THE UNIVERSITY OF BRITISH C O L U M B I A April 2006 © David Majaess, 2006 Abstract An application to monitor Steller sea lions using a sub dermal sensor requires a power supply to provide energy for wildlife monitoring base stations. The base stations are positioned near the habitats of the North Pacific Steller sea lions in isolated coastal areas of North British Columbia and South Alaska. The locations expose the base station to high winds, storms, ice, snow, debris, impacts from ocean waves, salt corrosion, and wide temperature swings. Furthermore, due to the remote distance, there is limited infrastructure; connection to the electrical grid is impossible and installation/maintenance is costly. The project requires a ruggedized, autonomous power supply requiring minimum maintenance and a long operating life. Therefore, the thesis proposes a unique power supply design incorporating high energy density double layer capacitors (ultracapacitors) that extends serviceability and improves immunity to cold climates. Finally, the researchers working on the Steller sea lion project have a limited economic budget and require a low cost system that is simple to transport and easy to install. The purpose of this research is to extend the battery's service life and improve the base station's immunity to cold climates. In this thesis, two methods are used to accomplish the objective. The first undertaking is to extensively research, design and implement high efficient components to minimize battery demand. As a result, the input source, its electronics and the load are well matched for the application. The next task is to incorporate a battery/capacitor bank to store energy. By integrating ultracapacitor technology to create a hybrid energy storage system, the battery cycling is minimized. 11 Table of Contents Abstract » Table of Contents i" List of Tables vii List of Figure viii Introduction 1 1.1 Research Objectives 2 1.2 Structure of the Thesis 2 Chapter 2 Application Design 4 2.1 The Base Station System 4 2.2 Power Supply Requirements and Restrictions 5 2.3 Location and Climate 5 2.4 Component Sizing 7 2.5 Idea Generation 7 2.5.1 Ultracapacitor Storage 8 2.5.2 The Power Supply 8 2.6 Base Station's Energy Budget 10 2.7 System Energy Portfolio 12 2.7.1 Transceiver 12 2.7.2 Power and Data Management 12 2.7.3 Satellite Phone 13 2.7.4 Climate Control 13 2.7.5 Efficiency Factor 14 2.7.6 Energy Consumption 15 2.8 Renewable Technology 15 Chapter 3 Photovoltaic Systems 17 3.1 Solar Cell 17 3.1.1 Cell Modeling 17 3.2 Solar Insolation 18 3.3 Performance Characteristic 18 3.4 Wattage 18 3.5 Temperature 18 3.6 Tilt Angles 20 3.7 Solar Declination Adjustment 20 3.7.1 Base Station Tilt Angle 23 3.7.2 Solar Panel Sizing and Selection 25 Chapter 4 Energy Storage Systems 26 4.1 Types of storage systems 26 4.2 Battery Technology 27 4.2.1 The Battery Cell 27 4.2.2 Capacity 28 4.2.3 Battery Energy Losses 29 4.3 Battery Types 30 4.3.1 Primary Batteries 30 iii 4.3.2 Secondary Batteries 30 Lead-Acid 31 Vented Nickel Cadmium 31 Sealed Nickel Cadmiums 32 Nickel-Metal Hydride 32 Lithium-Ion 32 Lithium-Polymer 32 4.3.3 Summary of Battery Technology 33 4.4 Battery Selection 35 4.5 Lead-Acid Battery Technology 35 4.5.1 Plate Limited Versus Electrolyte Limited 36 4.5.2 Flooded and Valve Regulated Lead Acid Batteries 36 4.5.3 Battery Life 37 4.5.4 Grid corrosion 37 4.5.5 Sulfation 38 4.5.6 Charging Lead-Acid Batteries 38 4.6 Lead-Acid Applications 39 4.6.1 Starting/Lighting/Ignition Battery 39 4.6.2 Deep-Cycle Battery 39 4.6.3 Float Service Battery 40 4.7 The Trojan T-105 and The Prevailer PV30H Battery 40 4.7.1 Cost 40 4.7.2 Discharge Rate 41 4.7.3 Temperature Performance 41 4.7.4 Battery Sizing 42 4.8 UltraCapacitor Technology 42 4.8.1 Long Life 43 4.8.2 Discharge/Charge Capability 43 4.8.3 Modeling Ultracapacitor's ESR and EPR 44 4.8.4 Temperature Performance 45 4.8.5 Environment 45 4.8.6 Disadvantage of Ultracapacitors 45 4.8.7 Voltage Balance 46 4.8.8 Ultracapacitor Selection 47 4.8.9 Energy Capacity 48 4.9 Energy Bank Topology 49 Chapter 5 Power Converters 50 5.1 Linear Regulators 50 5.2 Switchmode Power Converters 50 5.2.1 The Buck Converter 51 5.2.2 The Buck Topology 51 5.2.3 I R M S Calculations 54 Inductor RMS Current 54 High Side Switch (Switch 1) rms Current 54 Low Side Switch (switch 2) rms Current 55 5.2.4 Capacitor Bank dc/dc Converter 55 iv 5.2.5 Maximum Switching Frequency 57 5.3 Switch-Mode Power Inductors 58 5.3.1 Powder Cores 58 5.3.2 Ferrite Cores 58 5.3.3 Inductor Losses 59 Winding Losses 59 Core Loss 60 5.4 Inductor Design 61 5.4.1 Inductor Selection 63 5.4.2 Inductor Construction 65 Chapter 6 Semiconductor Switches 66 6.1 Bipolar Junction Transistors 68 6.2 Metal-Oxide Semiconducting Field Effect Transistors 68 6.3 Insulated Gate Bipolar Transistors 69 6.4 Switch Selection 70 6.5 The Characteristics of the Power MOSFET 70 6.5.1 The MOSFET Operating Regions 70 6.5.2 Parasitic Capacitances and Inductance 72 6.6 Power Losses 73 6.6.1 Conduction Losses 73 6.6.2 MOSFET Switching Losses 74 The Gate Charge Loss 74 Crossover Loss 76 Power Dissipation due to cross conduction 80 Total MOSFET Loss 81 6.7 High Side Losses (Switch 1) 81 6.8 Low Side Losses (Switch 2) 81 6.9 MOSFET Selection 81 Chapter 7 Circuit Implementation and Evaluation 83 7.1 Maximum Power Point Tracker 86 7.1.1 Implementation of the Hi l l Climbing MPPT 86 TheP-V andP-D Curve 87 Step Size 89 7.1.2 MPPT Logic 90 7.1.3 Input Capacitance 92 Input Capacitor's RMS Current 93 Input Capacitance Value 94 7.1.4 Settling Time 95 7.1.5 MPPT Discontinuity 96 7.1.6 MPPT Result 96 7.2 Implementation of the Capacitor Bank dc/dc Converter 97 7.2.1 Start Mode 103 7.2.2 Capacitor Converter's Efficiency 104 7.3 System Implementation 104 Chapter 8 Conclusion 108 Chapter 9 Future Work I l l v Bibliography 113 Appendix A Base Stations locations 117 Appendix B Schematic 118 Appendix C Software Code 120 Appendix D Local Manufacturer Contact Information 143 v i List of Tables Table 1 Climate data at Annette, Alaska 6 Table 2 The base station's energy consumption portfolio 15 Table 3 The day of month to day of year conversion table 22 Table 4 Solar panel size 25 Table 5 The BP350's properties 25 Table 6 Lead acid battery 33 Table 7 Vented NiCad battery 33 Table 8 Sealed NiCad battery 34 Table 9 N i M H battery 34 Table 10 Lithium ion battery 34 Table 11 Lithium polymer battery 34 Table 12 Full charge battery voltage at different temperatures 38 Table 13 Freezing point of DOD 41 Table 14 Battery sizing chart 42 Table 15 Magnetic's inductor properties 63 Table 16 Magnetic's inductor core dimensions 64 Table 17 Top 15 cores with the lowest losses 64 Table 18 Operating parameters at the PV's maximum power point 82 Table 19 MOSFET selection 82 Table 20 Nominal operating conditions 86 Table 21 Component pricing 107 Table 22 Base station locations 117 vii List of Figures Figure 2-1 A generalized diagram of the Steller sea lion base station 4 Figure 2-2 The geographic location of interest 6 Figure 2-3 The base station's power supply 9 Figure 3-1 The photovoltaic's circuit model 18 Figure 3-2 The photovoltaic's performance characteristics 19 Figure 3-3 The earth's solar orbit 21 Figure 3-4 Earth's orientation at the Solstice 21 Figure 3-5 The earth's declination angle 23 Figure 3-6 The tilt angle at different days of the year 24 Figure 4-1 The energy storage system 26 Figure 4-2 Inside view of a battery cell 28 Figure 4-3 The polarization and electrical resistance losses in batteries 29 Figure 4-4 The V R L A battery's cycle life 37 Figure 4-5 The temperature versus capacity performance 41 Figure 4-7 The ultracapacitor classical equivalent circuit model 44 Figure 4-8 Passive circuitry for voltage balancing 46 Figure 4-9 Active circuitry for voltage balancing 47 Figure 4-10 The percentage of energy versus capacitor voltage 48 Figure 4-11 The energy bank topology 49 Figure 5-1 Buck converter topology and switching action 52 Figure 5-2 The inductor's voltage and current signal 53 Figure 5-3 The capacitor bank's charging characteristics 56 Figure 5-4 The skin effect 60 Figure 5-5 The inductance (a) physical model and (b) electrical equivalent circuit 62 Figure 5-6 The inductor's saturation curve 65 Figure 6-1 Switching device selection at different frequencies and voltages 67 Figure 6-2 MOSFET and IGBT conduction loss 69 Figure 6-3 The MOSFET's operating regions 71 Figure 6-4 The MOSFET's parasitic elements 72 Figure 6-5 The MOSFET'S normalized R D S o n 74 Figure 6-6 The gate's charge plot 75 Figure 6-7 The gate's current path 78 Figure 6-8 MOSFET turn on behavior 79 Figure 7-1 Microcontroller state diagram 85 Figure 7-2 MPPT functional diagram and control loop 87 Figure 7-3 Photovoltaic P-D curve 89 Figure 7-4 MPPT logic flow 90 Figure 7-5 MPPT illustration 92 Figure 7-6 The input capacitor's current plot 93 Figure 7-7 The BP350's performance using the MPPT 97 Figure 7-8 UltraCapacitor converter's functional diagram 98 Figure 7-9 Capacitor current limiting topology and control loop 99 Figure 7-10 The converter's current sense and drive signal at 1 A 100 viii Figure 7-11 The capacitor's converter operating in buck mode 101 Figure 7-12 The capacitor's converter operating in boost mode 102 Figure 7-13 The discontinuous mode signal to prevent reverse current 103 Figure 7-14 The power supply's digital controller, MPPT tracker and capacitor converter 106 Figure 7-15 The power supply with its energy storage banks 106 Figure 7-16 UBC' s two BP350W photovoltaic panels 107 I X Introduction Researchers at the University of British Columbia in Vancouver, Canada are examining the possibility of using sub-dermal radio frequency identification tags for monitoring Steller sea lions. Dispersal and mortality rates of Steller sea lion pups are unknown and the implantable telemetry tag under development will be used to help determine these rates. A base station located nearby will monitor the tags and identify the distinct radio signals from nearby animals. Monitoring takes place in remote, marine regions located off the British Columbia's North West and Alaska's South coast. The stations will be located in isolated, harsh climate areas where they are required to survive high winds, ice, snow, salt water, abrasion, U V light, and wide temperature swings. Regardless of these environmental conditions, the power supply is expected to provide a reliable energy source for the Steller sea lion application. The monitoring locations are remote, making a connection to the electrical grid unfeasible. Thus, power is extracted from renewable energy sources such as wind or solar. Unfortunately, renewable energy sources are an intermittent source of power and do not always coincide with the load. Either the energy generated cannot be consumed or when necessary, the source cannot deliver enough energy. An energy storage solution significantly improves power availability because it regulates the energy flow. Batteries are the standard method used to store energy. However, batteries require maintenance and below freezing temperatures, their performance and capacity degrade dramatically. The base stations are located in remote, Arctic areas where battery performance and long service life become questionable. Therefore, this thesis proposes a unique power supply design utilizing double layer capacitors to improve performance. These capacitors, also known as ultracaps (or ultracapacitors), are a new technology that can be used in conjunction with or even replace batteries to store electrical energy. Ultracapacitors have three significant advantages: they can operate in very low temperatures, their full energy capacity can be utilized, and their recharging cycle life significantly exceeds battery technology. Power supplies implemented using this technology offers reliability in cold weather and less maintenance. The objective of this thesis is to design a power supply that will operate reliably using a storage system that incorporates ultracapacitors. 1 Introduction 2 1.1 Research Objectives The purpose of this research is to improve the limitations of battery energy storage systems by integrating capacitor technology to create a battery/capacitor hybrid system with reliable low temperature performance and extended operating life. The hybrid storage system will be used to develop a low power supply that can extract energy from renewable resources in an Arctic environment. The thesis will examine the reality of implementing such a project because economics and practicality plays a significant role in design. The sea lion group requires an affordable supply that is convenient to transport and simple to install. The following objectives have been identified: • identify the base station's power budget • research the wind velocity and solar insolation at the locations of interest • develop a power point tracker to operate a solar panel at its optimum efficiency • study the power characteristics and efficiency of electrical devices • optimize semiconductor selection for best performance • investigate the performance of battery technology • hybridize ultracapacitor and battery technology • implement a power management scheme 1.2 Structure of the Thesis This section describes the structure of the thesis. The chapters are summarized as follows: • Chapter 2 details the application. The base station's function, requirements, power budget and design restrictions are identified. Finally, an implementation solution is introduced. • Chapter 3 discusses photovoltaic systems. Their performance characteristics, operation, sizing, cost and selection are described. • Chapter 4 describes energy storage devices and their significance for this project. Specifically, the hybridization of battery and ultracapacitor technologies is studied in depth. • Chapter 5 explains dc/dc converters and their applications within the base station. The synchronous buck converter's theory is briefly summarized. In this chapter, the filter inductor is discussed in great detail. A theoretical model is developed to help optimize the efficiency during the core's design. Introduction 3 • Chapter 6 is an extension on dc/dc converters. However, this chapter focuses on the semiconductor switch. For best performance, different semiconductor technologies are explored. In particular, the switching behavior of MOSFETs is studied and equations modeling the losses are derived. • Chapter 7 discusses the circuit implementation. The base station's digital controller is constructed. The solar panel's maximum power point tracker is presented and its logic is explained. Based on experimental results, the tracker's performance and efficiency are analyzed. Finally, the ultracapacitor's charging system is described. • Chapter 8 concludes the thesis. The implementation and final performance is summarized. • Chapter 9 suggests the future work. Chapter 2 Application Design This chapter discusses the base stations requirements, identifies the restrictions, estimates the power budget and researches the wind velocity and solar insolation at the locations of interest. 2.1 The Base Station System The base station retrieves data from monitoring tags implemented in Steller sea lions. The data is stored in the base station and is periodically uploaded to a remote server via a satellite network. Figure 2-1 illustrates the base station. As shown, the power supply provides energy for several systems: • a radio transceiver to communicate and retrieve information from the sea lion tags • a data processor to gather the information • a satellite modem to upload data • a climate controller to regulate the operating temperature within the base station • a power management system to monitor and maintain the energy supply Seal Lion Tag Data Radio Link Satellite link Satellite Renewable Resources S e a L i o n B a s e S t a t i o n Transceiever Data Flow Remote Server Received Data • • • • » Output Energy Flow Input Energy Power Supply Power Supply Controller Power Management J o -Data Processor Uplink I Climate Controller Figure 2-1 A generalized diagram of the Steller sea lion base station 4 Chapter 2 Application Design 5 2.2 Power Supply Requirements and Restrictions The base station requires a heavy duty, robust, off the grid power plant designed for Arctic and marine conditions. The power supply design for this application must be extremely durable because it will operate in an environment with subzero climate, high winds, rain, salt air and humidity. It will also have to survive impacts from ocean waves. The system must operate with a temperature range between -30 °C to 45 °C and must survive a range between -45 °C to 65 °C. The base station wil l be placed in isolated areas with no infrastructure and limited access. To minimize the costly installation/service overhead, the power supply wil l be low maintenance, simple to transport and easy to install. An operating time of five years between servicing is desired. The sea lion group has a limited budget and the final design is a compromise between cost, efficiency and reliability. To lessen the disturbance on wildlife, the base station must merge into the natural surroundings as much as possible. Any generated noise or movement must be kept to a minimum. Even the station's color will be chosen for camouflage. For wildlife and human safety, sharp objects protruding from the station, large radiation and high voltage are not permitted. The input power is supplied from renewable sources such as wind and/or solar energy. A backup energy storage system utilizing ultracaps and batteries will be designed to provide power when the renewable resources are unavailable. 2.3 Location and Climate The base stations are commissioned at different locations throughout the coast of south Alaska and British Columbia. Figure 2-2 is a map indicating the area of interest. The Latitude and Longitude coordinates for each site are included in Appendix A . Daily solar radiation data is recorded at Annette's meteorological station (located in Alaska). This information is available from the United States Department of Energy's Office of Energy Efficiency and Renewable Energy. Environment Canada records temperature and precipitation statistics at several stations along British Columbia's coast but lacks information on solar radiation. Thus, the data used to predicate the amount of energy available from renewable resources was gathered at Annette, Alaska (Table 1, Climate data at Annette, Alaska [1].) The following map indicates the region of interest [2]. Chapter 2 Application Design 6 Gulf Of Alaska Yukon N> Territory Annette, Alaska Station Number 25308 Latitude 55.03°N Longitude 131.57° W Elevation 34 m Mean Pressure 1009 millibars Figure 2-2 The geographic location of interest Table 1 Climate data at Annette, Alaska Solar Radiation (kWh/m2/day) Uncertainty ±9 % Tilt D Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Year 0 Minimum 0.40 0.90 1.60 2.70 4.00 4.10 3.50 3.20 2.20 1.30 0.60 0.30 2.40 Maximum 0.90 1.60 3.00 4.40 5.40 6.40 5.90 4.80 3.30 1.80 1.10 0.60 2.80 Average 0.65 1.25 2.30 3.55 4.70 5.25 4.70 4.00 2.75 1.55 0.85 0.45 2.60 Lat- 15 Minimum 0.70 1.00 1.90 2.90 4.00 3.90 3.40 3.30 2.50 1.50 0.80 0.40 2.70 Maximum 2.30 3.70 4.60 5.40 5.80 6.40 6.10 5.50 4.50 2.90 2.60 1.50 3.40 Average 1.50 2.35 3.25 4.15 4.90 5.15 4.75 4.40 3.50 2.20 1.70 0.95 3.05 Lat Minimum 0.70 1.00 1.80 2.80 3.70 3.50 3.10 3.10 2.40 1.50 0.80 0.40 2.60 Maximum 2.70 4.10 4.80 5.30 5.40 5.90 5.70 5.20 4.50 3.00 3.00 1.70 3.30 Average 1.70 2.55 3.30 4.05 4.55 4.70 4.40 4.15 3.45 2.25 1.90 1.05 2.95 Lat + 15 Minimum 0.70 1.00 1.70 2.50 3.20 3.10 2.70 2.80 2.30 1.40 0.80 0.40 2.40 Maximum 2.80 4.30 4.70 4.80 4.70 5.10 4.90 4.70 4.30 3.00 3.10 1.90 3.10 Average 1.75 2.65 3.20 3.65 3.95 4.10 3.80 3.75 3.30 2.20 1.95 1.15 2.75 Chapter 2 Application Design 7 (Table 1 continued) Temperature (°C) Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Average 1.3 2.8 4.1 6.3 9.4 12.4 14.6 14.8 12.3 8.3 4.1 2.1 7.7 Daily Min -1.3 0.1 1.1 2.8 5.7 8.8 11.1 11.3 8.9 5.5 1.5 -0.3 4.6 Daily Max 3.8 5.5 7.1 9.7 13.2 16 18.1 18.3 15.7 11 6.6 4.4 10.8 Record Lo -17.2 -16.7 -17.2 -6.1 -0.6 2.8 4.4 4.4 0.6 -7.8 -19.4 -17.2 -19.4 . Record Hi 16.1 17.2 16.7 27.8 31.1 31.7 31.7 32.2 27.8 21.7 19.4 16.7 32.2 Wind Speed (m/s) Average 4.9 5.0 4.4 4.4 3.8 3.7 3.4 3.4 3.7 | 4-8 [ 5.0 | 4.8 | 4.2 2.4 Component Sizing To supply the base station's energy budget, proper selection of the renewable source and storage component is critical. Sizing too small, prevents continuous operation and drains the storage elements causing component exhaustion, wear out and failure. Over sizing is also restrictive because as components enlarge, cost increases and transportation/installation becomes a burden. Factors to consider for selecting the renewable source and energy storage are as follows: The insolation patterns, temperature and wind speeds in the geographical area of installation Load requirements The number of days that energy must be supplied from the energy storage Temperature effects on capacity and performance Aging effect Component efficiency Cost Weight and physical size 2.5 Idea Generation The idea is to design an autonomous system that supplies the power needs for a radio base station from a renewable source. A maintenance free, energy storage, subsystem will supply power when the input source is unavailable. To insure that the energy reserve is not completely depleted, the power consumption is monitored. If the stored energy becomes low and the renewable source remains unavailable, then the base station enters a "sleepmode" state. In this state, the radio communication with the Stellar sea lion is placed temporally offline until the Chapter 2 Application Design 8 energy is replenished. This offline condition is a worst case scenario that prevents the system from completely exhausting its energy reserves. To minimize the likelihood of entering the sleepmode state, the sizing of the source and its energy storage system is critical. 2.5.1 Ultracapacitor Storage An ultracapacitor storage system will be incorporated in the design. Using a capacitor as an energy supply has many advantages over rechargeable batteries. A capacitor charge/discharge life is large, greater than 500,000 cycles [3]. A battery's cycle life is typically 200 times less. Capacitors have no disposable parts. They can withstand large currents and may be discharged to any level without damage. Accurate charge and time monitoring are unnecessary. Capacitors can operate at a wider temperature range than batteries making them attractive for this base station application. Unfortunately, ultracapacitors have a breakdown voltage, approximately 2.5 V. This is effectively their over charge limit. This application utilizes a series connection to obtain a larger voltage range. However, stacked capacitors involve additional voltage equalization circuitry [4]. Finally, a capacitor's energy density is not as large as batteries. Section 4.8 discusses ultracapacitors in more detail. 2.5.2 The Power Supply The power supply provides a reliable source of energy for the base station by monitoring and regulating the energy from renewable resources. The operation of the power supply is categorized into several subsystems. They are as follows: input regulator and maximum power point tracker, power management, power supply dc/dc converters, energy reserve monitoring/charging, sleepmode controller and power circuitry. Figure 2-3 demonstrates how the subsystems collaborate with each other. The solid arrows indicate the energy flow direction throughout the power supply. The dashed arrows are an illustration of the communication signals between each subsystem. The systems are numbered and their functions are explained below the figure. Chapter 2 Application Design 9 INPUT E N E R G Y f Solar 1 I' / l \ I Wind 1 Input Regulator/ Max Power Point Tracker m I Energy 1 Flow J • Power Circuitry Power Management m Determine the energy needs for the station Surplus energy Charge Reserves Power Supply DC/DC Converters m (12 V, 5 V) | Stored • Energy Amount I Signal Sleep Mode (On/Off Signal) Energy Reserve Monitoring/ Charging E Energy Reserve Battery 7 ^ utiracap Sleep Mode Control ler m Excess Energy Bleed? Shut Input? Energy Flow For Subsystem Output Energy for Base Station Temp Signal Temperature Control DO Figure 2-3 The base station's power supply 1. Input Regulator and Maximum Power Point Tracker The input power regulator and maximum power point tracker determine whether wind and/or sun energy is available. If so, the energy is then fed to the power management system. For optimum efficiency, the power point tracker operates the input source at its peak operating point. 2. Power Management The power management is a digital based system responsible for controlling the energy flow. The power management directs the renewable energy to the base station and reroutes surplus energy to storage. If no input energy is available, then the base station's energy is retrieved from storage. 3. Power Supply dc/dc Converters The dc/dc converters are responsible for the voltage/current conversions. The power supply will provide the base station with two voltage outputs: 12 V and 5 V . The 12 V output provides the power for the transceiver and satellite communication. The 5 V is the logic voltage used for the base station's controlling circuitry. Chapter 2 Application Design 10 4. Energy Reserve Monitoring/Charging The energy reserve monitoring/charging regulates the charging and discharging process insuring that the energy reserves are not overcharged or depleted beyond their lower threshold. This system communicates the amount of stored energy with the power management system. 5. Sleepmode Controller The sleepmode controller prevents complete exhaustion of the energy reserves. The base station is placed offline and the power supply enters a minimum function mode allowing the power management system to operate in an ultra-low power consumption state while it waits for the renewable energy to return. The system's average sleepmode power consumption is a few mW, allowing the sleepmode state to operate for months. 6. Temperature Control The temperature controller monitors and regulates the operating temperature of the power supply. The power supply is designed to operate between -30 °C to 45 °C. However, i f the base station's internal temperature exceeds its normal operating range, the temperature controller signals the system to enter the sleepmode state. 7. Power Circuitry The power circuitry supplies the power for the base station's subsystems. The critical subsystems such as the power management, energy reserve monitoring, and temperature controller have a constant power supply regardless i f the base station is online or in sleepmode. The next section calculates the amount of energy the power supply and its load (the base station) requires. 2.6 Base Station's Energy Budget The amount of energy required determines the capacity of the energy storage banks and the size of the input power source. The power supply provides energy for the following system within the base station: a radio transceiver, data storage, and satellite link. Energy is also required for the power supply's circuitry and temperature monitoring. For this application, the banks are sized by averaging the power (over a weekly window) for each system. The weekly Chapter 2 Application Design 11 average is utilized because not all the systems operate continuously. For instance, the satellite phone transmits 10 minutes a week and remains offline for the remainder. By determining each system's average power and predicting its operating time, the energy consumption can be calculated. It is convenient to discuss the energy consumed per day because batteries are rated by their Amp-hour capacity and the insolation data is provided as a daily average. The base station's energy budget is the summation of the energy consumption for all its systems. The storage banks are sized by determining the difference between the energy budget and the amount of energy the renewable sources can provide. To offset the unpredictability in renewable energy resources, estimations are based on a worst case scenario. The energy storage banks must provide sufficient energy for at least ten days of continuous operation. For clarity, the terms run power (PR), weekly averaged power (Pw), averaged daily energy (E D), peak power (PP) and watt-hours (Wh) are defined as follows: Run Power Run-power is the instantaneous amount of power necessary for a device to operate, such as a 60W light bulb. Weekly Averaged Power Weekly average is defined as the run power averaged over a one week window. Weekly average is calculated using the following equation TT. ,. . run power x operating hours per week Weekly averaged power = (W). (1) total hours per week Watt-hour A common unit of electrical work, representing the energy delivered at a rate of one watt for one hour of time. It is the average power (W) multiplied by the run time (h). One watt-hour (Wh) corresponds to 3.6 kJ since a Joule (J) is defined as J = Watt x second. Averaged Daily Energy For this project, it is convenient to discuss the average energy consumed in one day. It is simply Averaged daily energy = the average power x 24 hours (Wh). Chapter 2 Application Design 12 Peak Power The maximum power a device will require. The storage banks can deliver a peak power up to 100 W. This enables the system to handle transients caused by charging caps or turning on cold filaments used for heating. However, the base station is optimized for high efficiency and is intended for low power use. In this application, the peak power is not as important as minimizing the average power consumption. 2.7 System Energy Portfolio The energy portfolio for each system is described as follows. 2.7.1 Transceiver The transceiver communicates with the radio tags at 915 MHz. To keep the base station's power consumption low, the sea lion group is implementing their communication protocols to achieve a rated power of 1.6 W. The transceiver monitors a colony of sea lions and operates continuously. The transceiver is designed to have a peak power no larger than 4 W. 2.7.2 Power and Data Management The power and data management is a digital system that controls the energy flow and processes information retrieved from the tags. The data is stored within the R A M of a microcontroller until it is uploaded, once a week, to a remote server through a satellite connection. To reduce power consumption, this system utilizes a low current controller, operating at a reduced clock frequency. A microcontroller's power consumption increases at a higher operating frequency and with a larger voltage supply. Consumption also varies over temperature and this relationship is nonlinear. A Microchip's 18Fxx series microcontroller, with no peripherals, operating at its maximum limits of 5.5 V , 20 MHz, at 85 °C will consume 44 mW. At half the clock speed the power drops by 40 % [5]. The power management microcontroller operates at 4 M H z and manages the energy flow using comparators, operational amplifiers, transistor drivers, and analog/digital converters. These components are all low power, integrated circuits. The power management system measured a rated power of 130 mW (section 7.3.) During sleepmode, the power management disables most of its peripherals and operates at approximately 20 mW. However, the sleepmode duration or its Chapter 2 Application Design 13 frequency is hard to predict. Thus, a worst case assumption is that the power management will operate continuously at 130 mW. 2.7.3 Satellite Phone The collected data is transmitted via satellite phone. Several satellite communication products are available and the GSP-1620 data modem on the Globalstar satellite network is considered. The duplex modem requires a run power of 5.4 W and has a data rate of 1.2 kBps [6] (assuming 1 Byte = 8 bits.) A simplex system is required. However, the available simplex products are slow, having a data speed approximately 148 Bytes in 6 to 8 minutes [7]. The wildlife monitoring system detects an animal's presence. A radio tag is mounted onto a Stellar sea lion and a maximum of 100 animals are tagged per site. Every 15 minutes, the base station attempts to communicate with the tags. A presence is recorded i f the animal is within radio range. When detection occurs, the station records a binary one [8]. Therefore, reporting a week's worth of data per animal takes 672 presence bits. The system is designed so that each animal has a unique 24 bit ID code. Thus, a total of 8700 Bytes of information per week is required. Using the GSP-1600 phone, the weekly data can be sent in 7.25 s. However, before transmission, the modem requires approximately 5 to 10 min to power up, track the satellite network and establish a system connection. Therefore, the total operating time is estimated at 10 min per week resulting in a weekly averaged power consumption of 5.3 mW. This amount of power is insignificant for the storage bank. Nonetheless, when stored energy becomes critically low, the satellite link remains offline. 2.7.4 Climate Control The base station's operating temperature depends on its components. For instance, batteries do not perform well in subzero conditions. Likewise, electronics do not function correctly at high temperatures. Therefore, the climate controller monitors the internal temperature within the base station. The power required for climate control is difficult to determine analytically because the transfer of thermal energy between the base station's enclosure and the outside environment is complex. The mechanism of convection, conduction and radiation are dependant on many variables. However, based on design, an approximation is devised. To house the components, the base station's enclosure has the following physical dimensions yielding a volume of 0.29 m 3 [9]: Chapter 2 Application Design 14 • width: 600 mm height: 800 mm • depth: 600 mm The base station is a low power system and does not generate more than 2 W of heat. The enclosure is constructed from sheet steel with a depth of 2 mm. With a temperature conductance of 16 W / m K , the steel povides an easy path for heat to transfer between the enclosure's internal compartment and the outside environment [10]. To reduce the power required for climate control, the base station's components are designed to accommodate an operating range that attempts to exceed the environment. Thus, based on the climate data from Table 1, the base station's operating range is between -30 °C and 45 °C. However, in the summer, the collaboration of direct sun and warmer weather may require air ventilation. A solar vent is suggested because it has a built in solar cell and dc fan that activates only when the sun shines. This device has its own power source and is not included in the power budget. When activated, the vent circulates the air at 0.9 m3/min. It is available from a local distributor called SPS Energy Solution. Refer to Appendix D for the distributor's contact information. The base station's internal temperature is monitored using the AD7414 digital temperature probe. If the external climate becomes too severe and the enclosure's internal temperature exceeds the operating range, the temperature probe signals the system to enter the sleepmode state. 2.7.5 Efficiency Factor The base station's circuitry is designed for efficiency. Switching components operate at low frequencies and are optimized to reduce loss. The dc/dc converter utilizes a synchronous topology to minimize the diode voltage drop. However, the power budget requires a scaling factor to account for the I 2R loss throughout copper wires, switching losses, component equivalent series resistance, battery efficiency and the effectiveness in converting renewable energy into electrical energy. For instance, the photovoltaic used in the base station (discussed in chapter 3) is tested on the roof of U B C s electrical engineering department (Figure 7-16 on page 107). In the figure, two photovoltaics are shown and each is loaded at their maximum power point. At noon, during bright, clear, sunny days, the largest output power from either panel was observed at 43 W. This is a 14 % discrepancy from their rated 50 W. The difference is from environmental characteristics explained in section 3.3. Therefore, a 15% factor is added to Chapter 2 Application Design 15 account for the additional inefficiencies. The correction is a conservative factor chosen as additional protection to prevent excessive battery discharge. 2.7.6 Energy Consumption The following table summarizes the required power and duration for each system. The total energy budget is determined to have an average daily energy consumption of approximately 48 Wh/day. The power system is rated at 12 V. The system wil l be designed to safely handle a peak power of 100 W. The average daily amp hours is calculated as: 48 Wh/day + 12 V = 4 Ah. Table 2 The base station's energy consumption portfolio System Run Power (W) Operating Time (h/day) Operating Time (Days/Week) Weekly Averaged Power (W) Averaged Daily Energy Consumption (Wh/day) Transceiver* 1.6 24 7 1.6 38.4 Power Management and 0.13 24 7 0.13 3.12 Data Storage Satellite Phone 5.4 0.167 1 0.005 0.12 Subtotal- 1.735 41.64 Total (15 % add-on): 2 47.9 =48 * Transceiver: The transmitter and 1.6 W. receiver are one integrated unit and their combined power consumption is 2.8 Renewable Technology Wind and solar technologies are considered as the power sources. Deciding upon wind turbines and/or solar panels are site and application specific. The determination depends on practicality, available resources and the required energy production. Compared to other components of the base station, a wind turbine is the largest moving mechanical system. It is directly exposed to high winds, ice, snow, salt water, abrasion, U V light, and large temperature swings. In these conditions, the turbine will be stressed, resulting in a high chance of breakdown. In general, heavier wind turbines have proven to be more rugged and dependable Chapter 2 Application Design 16 than light weight machines. Heavier, robust machines typically survive longer [11]. However, large machines make transportation and installation more difficult. Furthermore, larger turbines require higher towers to clear rotor diameter and avoid turbulence. A smaller, micro turbine is cheaper, easier to transport, but less likely to survive the winter. While there is minimal evidence that turbines frighten animals, there is a concern that the spinning motion and created noise pollution may scare the Steller sea lions. Wind energy is a good source of renewable energy. However, due to location and requirements of the base stations, turbines are not a practical solution for this project. Solar power technology is reliable because it has no moving parts. Wind turbines have a three year warranty whereas solar panels are guaranteed for 25 years. Solar panels will be used for the source because they are straightforward to install, reliable, do not require maintenance and can be conveniently disconnected from the load when they supply excessive energy. Chapter 3 Photovoltaic Systems The conversion of the sun's insolation to electrical current is a practical way to generate power for electrical devices operating away from the power grid. Harnessing energy from the sun is becoming more possible as photovoltaic technology matures. 3.1 Solar Cell Solar panels, also called photovoltaic (PV) panels or solar modules, are flat arrangements of series connected silicon solar cells. Silicon is the most common material used in solar cells. There are three types of silicon solar cells. In order of decreasing efficiency and manufacturing cost, they are the single crystal cell (mono-crystalline), the semi-crystalline cell (multi-crystalline), and the amorphous (non-crystalline) cell. Lower efficiency means that larger surface area is required to produce the same power. For a given power rating, semi-crystalline panels are 20 % to 30 % larger than single crystal types. Amorphous cell panels are about twice the size of the other types [12]. Solar panel packaging and mounting vary to suit a wide variety of conditions. Solar panels are used at locations from homes to space applications. The panels can be wired as series or parallel arrays to produce higher voltages and currents. 3.1.1 Cell Modeling A solar cell is modeled as a constant current source in series with a diode as shown in Figure 3-1. A solar panel is a stack of solar cells connected in series and the equivalent circuit is a single large current source driving a string of diodes. The current flowing from the output terminal I O U T is equal to the light-generated current I I minus the diode-current I D and the shunt-leakage current I S H - In practical cells, I S H is small and can be omitted [13]. The series resistance Rs represents the resistance to the current flow and depends on the panel's doped p-n junction depth, the impurities and the contact resistance. ThePV's efficiency is sensitive to variation in Rs. An increase in Rscan significantly decrease the PV's output. With no external load, all the current from the source is through the series diodes. The output voltage of the panel is the sum of 17 Chapter 3 Photovoltaic Systems 18 the voltages across the diodes. This is the solar panel open circuit voltage V 0 c - Typically, a solar panel consisting of 36 series connected cells will have an open circuit of about 18V. Rs A A A o + OUT V o c = n 0 . 7 V Figure 3-1 The photovoltaic's circuit model 3.2 Solar Insolation Solar insolation describes how much sunrays an object is receiving. PV manufacturers rate their products at a reference point called air mass 1.5 (AMI.5). AMI .5 is classified as the air found on a typical day with average humidity and pollution. The typical solar insolation on a bright day with AMI .5 is about 1 kW/m 2 [14]. 3.3 Performance Characteristic The performance characteristic of a PV panel is described by the voltage versus current voltage (V-I) curve shown in the plot Figure 3-2 (a). Figure 3-2 (b) is the power curves and is simply the product of the voltage and the current outputs. As the amount of insolation decreases, the V-I curve shifts down. 3.4 Wattage Silicon solar cells have simple power characteristics. Each module has a specific point on its power curve which yields the greatest power. This maximum power point occurs at the knee point of the I-V curve. At open voltage and short circuit a panel produces zero power. 3.5 Temperature As shown in Figure 3-2 (a) and (c) when the temperature of the module increases, two things happen: the voltage output of each cell decreases and the current of the cell increases with Chapter 3 Photovoltaic Systems 19 increasing temperature. However, the net effect is a shift in the power knee, resulting in more power at lower temperatures, typically, a 0.45 % decrease of power for each increase in degree centigrade. Solar radiation is never consistent. Many factors affect the amount solar radiation a body receives. Variables such as weather, altitude, and reflection make realistic standardization of insolation impossible. Manufacturers test their models with reproducible laboratory standards that are not at the mercy of solar insolation and weather. Testing is performed at temperatures between 25 °C and 28 °C. However, on a very sunny day, even at an air temperature of 25 °C, modules have been measured to have a temperature of 40 °C. To compensate for the power shift at different operating temperatures, PV manufacturers usually publish the power temperature coefficient on their product's datasheet. The plots in Figure 3-2 are taken from [13], [15]. (a) 1000W/m2 Curve's knee Max power region @ constant temperature (b) Voltage Voltage :.High Temperature — Low Temperature (c) b d lOOOW/m 2 Temperature ° C 80° P 5 10 V o l t a g e 60c 20° 40° 0° 5 20 25 Figure 3-2 The photovoltaic's performance characteristics Chapter 3 Photovoltaic Systems 20 3.6 Tilt Angles A solar panel's output power depends on the amount of sunlight available and on how much of this light the solar panel collects. Sunlight is the essential ingredient. If the panel is fixed, as the direction of the sun with respect to the panel changes throughout the day, the effective area or aperture of the panel will change [16]. At dawn, very little sunlight reaches the active surface of the panel and the aperture of the panel is near zero. As the sun goes higher in the sky, the panel intercepts more sunlight until maximum aperture is reached when the sun is at right angle to the plane of the panel's active surface. The aperture then decreases as the sun falls in the sky. When the angle is low with respect to the panel, the aperture is small and losses due to reflection from the surface of the panel are large. The largest possible aperture collects the most light and keeping the module perpendicular to the incoming sunlight means that the module intercepts the maximum amount of sunlight. As the sun moves throughout the day, more energy is collected i f the PV module is installed on a tracker and follows the sun like a sunflower. However, due to additionally cost and mechanical complexity, trackers are not considered for the sea lion project. Future developments may implement a sun tracking system as an improvement in design. While a tracker provides from 25 % to 35 % more power by changing the PV's East to West angle (Solar Azimuth), adjusting the sun's north to south angle (Solar Declination) will compensate for the changes as the height of the sun changes during the year [17], [18]. 3.7 Solar Declination Adjustment Solar Declination is an angular measurement of how many degrees North (positive) or South (negative) of the equator that the sun is when viewed from the centre of the earth. As shown in Figure 3-3, the earth rotates on its axis which is inclined 23.45° with respect to the plane of its orbit around the sun. The rotational angle remains constant, so throughout the year, as the earth orbits the sun, the declination varies ±23.45°. On June 21, the summer solstice, the earth's northern hemisphere is titled closer to the sun because the declination is at its maximum of+23.45°. Chapter 3 Photovoltaic Systems 21 Axis normal to the plane of the Earth's Orbit Autumnal Equinox (September 21) Earth's rotational axis pointed neither toward nor away from the sun Earth's Axis\ Winter S o l s t i c e M ^ ( D e c e m b e r 2 2 ) I \ Southern Hemisphere tilted closer to the Sun Summer Solstice (June 21) Northern Hemisphere tilted closer to the sun Vernal Equinox \ (March 21) Earth's rotational axis pointed neither toward nor away from the sun Figure 3-3 The earth's solar orbit Figure 3-4 illustrates the earth's orientation during the solstice. Half a year later, on December 22, the declination is at its minimum of-23.45°. This is called the winter solstice and the southern hemisphere points closer to the sun. At the autumnal (September 21) and the vernal (March 21) equinoxes, the declination is at 0° and the earth's rotational axis is pointed neither toward nor away from the sun. During the equinoxes, at noon, the sun appears directly overhead at the earth's equator. garth's Orbiy On December 22: • North Hemisphere furthest On June 21: • North Hemisphere furthest from the sun • Equator above parallel axis • Declination: -23.45° from the sun • Equator above parallel axis • Declination: +23.45° Figure 3-4 Earth's orientation at the Solstice Chapter 3 Photovoltaic Systems 22 In Figure 3-5, the angular orientation of the earth is modeled and the resulting geometric equation is derived Tilt angle = Latitude angle + 23.45cos(a). (2) The scale factor a is in degrees and converts days of the year to an angular quantity. a = (360 ) 365 ' (3) Where n = the n t h day of the year (refer to Table 3.) Equation 2 gives the solar panel's tilt angle to obtain maximum aperture on a specific day of the year. Table 3 converts the i t h day of the month into the n t h day of the year. Table 3 The day of month to day of year conversion table Month im day of the month January n = = 3 1 + i February n = = 5 9 + / March n = = 9 0 + i April n = 120+ / May n = 151+ / June n = 181+ / July n = 2 1 2 + / August n = 2 4 3 + / September n = 2 7 3 + / October n = 3 0 4 + / December n = 334+/ The tilt angle is a function of the solar panel's physical latitudinal position. During the earth's orbit, when the planet is at the equinox, the declination angle is the latitude angle. However, on June 21, the equator is offset by 23.45° and this difference must be subtracted from the latitude angle. The opposite takes place on December 22, where 23.45° is added to the latitude angle. Since the earth orbits the sun in a circular path, the declination angle throughout the year is a periodic function. December 22 (day 356) corresponds to 0° and June 21 is 180°. For the derivation of (2), the following assumptions are made: • The earth's orbit is circular • The panel is perpendicular to the sun at noon • The earth's tilt angle of 23.45° remains fixed • Leap years are ignored Chapter 3 Photovoltaic Systems 23 Axis Normal to the Figure 3-5 The earth's declination angle Figure 3-6 shows a graphical representation of (2) for different latitudinal positions at various days of the year [17]. There are fifteen curves, each for 5° degrees of latitude. To determine the correct tilt angle, on the right side of the chart, find the curve that nearly corresponds to the solar panel's latitude. Follow that curve until it intersects the current date on the x-axis (month/day). The corresponding angle read on the y-axis is the angle between the ground and panel's face. This angle will result in the PV module being perpendicular to the sun's rays at noon on that date. 3.7.1 Base Station Tilt Angle Due to the base stations remote locations, the tilt angle for the sea lion base station cannot be frequently adjusted and the optimum tilt angle is determined as follows. December has the least amount of solar radiation during the year. From Table 1, the sun hours in December are four times less than June. The storage capacity and solar panel are sized to compensate for this condition. Even though the year's average energy will be compromised; the summer days are long enough to offset the winter's angle. For December, the calculated tilt angle using (2) is 78°. This predicted angle agrees with the insolation data recorded at Annette. In December, the values in Table 1 have longer solar hours at a tilt angle of "Lat + 15." Therefore, based on the Chapter 3 Photovoltaic Systems 24 theoretical predictions and matched climate data, the panel tilt angle is adjusted compensated for December. 12/21 1/20 2/20 3/22 4/22 5/22 6/22 7/22 8/22 9/21 10/22 11/21 12/22 Day of the Year Figure 3-6 The tilt angle at different days of the year Chapter 3 Photovoltaic Systems 25 3.7.2 Solar Panel Sizing and Selection Based on the power budget and the climate data, the following table determines that 50W solar panel is capable of supplying the base station's energy requirements [19], [20]. Table 4 Solar panel size 1. Average daily energy consumption 48 Wh/day 2. Average daily sun hours (worst case insolation from Annette, Alaska, compensated for December) 1.15 h 3. Divide line 1 divide by line 2. This is the required solar panel size (PMAX) 42 W In the industry, solar panels with the following output power are commonly available: 10W, 20W, 50W, 60W and 65W. Due to material overhead, the pricing of panels is not linearly related to its output power. For example, a 50W panel cost less than purchasing two 20W panels. BP Solar offers a SOW (model: BP350) that satisfies the requirements for the base station. It is available from a local distributor called SPS Energy Solutions. Refer to Appendix D for their contact information. SPS quoted $331.00 (CDN) for each unit. The following table is the characteristics of the BP350. Table 5 The BP350 's properties Characteristics* Units Maximum power (PMAX) 50 W Voltage at P M A X ( V M p ) 17.5 V Current at P M A X (Imp) 2.9 A Warranted minimum PMAX 45 W Short-circuit current (I s c ) 3.17 A Open-circuit voltage (Voc) 21.8 V Temperature coefficient of I s c (0.065±0.015)%/ °C Temperature coefficient of Voc -(80±10) mV/°C Temperature coefficient of power -(0.5±0.05) % / °C Dimensions Length 839 mm Width 537 mm Depth 50 mm Weight 6 kg NOCT (Air 20 °C; Sun 0.8 kW/m2; wind 1 m/s) Chapter 4 Energy Storage Systems The photovoltaic is an intermittent source of power and does not always coincide with the load. Either the energy generated by the PV array cannot always be consumed or when required the array itself cannot deliver enough energy. An energy storage solution significantly improves power availability by regulating the energy flow. The functions of the storage system are [21]: 1. To store electrical energy when it is produced by the PV array and to supply energy to electrical load when needed 2. To supply power to the load at a stable voltage, by suppressing or smoothing out transient voltages that may occur in PV systems 3. To supply surge or high peak operating current Figure 4-1 The energy storage system 4.1 Types of storage systems The battery stores energy in the electrochemical form and is the most widely used device for energy storage application. Other systems of energy storage are: • flywheel • compressed air • superconducting coil • ultracapacitors 26 Chapter 4 Energy Storage Systems 27 Each technology has characteristics that are suited for particular applications. For the base station, an ultracapacitor/battery hybrid storage bank offers a simple and reliable system. Batteries have high energy densities and ultracapacitors offer long cycle life and wide operating temperatures. Additionally, unlike, mechanical storage systems, such as, flywheels or compressed air, the battery/capacitor bank is simpler to interface with PV panels and the electrical load. The following sections discuss the advantages and limitation of batteries and ultracapacitors in detail. To determine the best selection for the sea lion project a thorough investigation into battery technology is conducted. 4.2 Battery Technology Batteries have high energy density that is critical for supplying power when solar energy is unavailable for long periods of time. There is no ideal battery that gives the optimum performance under all operating conditions. However, many advances have been made in battery technology resulting in the development of different chemistry and designs, each having favorable characteristics under specific applications. Operating temperature, discharge requirements, cycle life, capacity and voltage retention are all factors in selecting the correct battery. 4.2.1 The Battery Cell A battery's energy is stored in the electrochemical form. When a chemical reaction occurs the energy is released in the electrical and thermal form. Unfortunately, the thermal form is an unusable energy, a waste that reduces the conversion efficiency. The basic unit of the battery is the electrochemical cell. The internal construction of a basic battery cell is shown in Figure 4-2 [21]. The cell has three parts, a negative electrode, a positive electrode, and an electrolyte. The electrode plates are connected to an electrical load via external terminals mounted on the battery's casing. The cell stores electrochemical energy at low electrical potentials. The cell's voltage depends on the battery chemistry, for example, lead-acid cells have a nominal cell voltage at 2.1 V , while lithium-ion cells are at 3.7 V. This is known as the open circuit voltage. Chapter 4 Energy Storage Systems 2 8 Figure 4-2 Inside view of a battery cell The amount of free energy stored by the cell depends on the amount of the chemical reactants in the cell. Cells can release energy because the active materials have a higher free energy than the reaction products. A circuit is formed when an external load is connected across the positive and negative electrodes of a charged cell. With the circuit established, a chemical reaction occurs between the electrolyte and each electrode producing free electrons at the negative electrodes. The stored chemical energy is released because the positive electrodes accept the electrons through the load. Internally, the circuit is completed by the ionic current carriers of the electrolyte. Current stops when the reaction at either electrodes runs out of active material or the electrolyte can no longer supplies ions. During charging, the directions of the reactions are reversed. 4.2.2 Capacity The ability of a cell to hold energy is called its capacity (C). Capacity is measured in Ampere-hours (Ah), and is determined through a constant current discharge of a fully charged cell. As the cell discharges, the time is measured from the start of discharge until the potential across the loaded cell has reached a specific cutoff voltage, meaning it can deliver C amperes for one hour or C/n amperes for n hours. The battery is made of numerous electrochemical cells connected in a series-parallel combination to obtain a desired operating voltage and/or current. The higher the battery's voltage, the larger the number of cells required in series. The battery's rating is stated in terms of the average voltage during discharge and the Amp-hour (Ah) capacity it can deliver before the voltage drops below the specified limit. Discharging a battery too low can cause damage. Thus, a safe level cutoff voltage is chosen. For instance, a discharge limit of Chapter 4 Energy Storage Systems 29 1.75 V/cell is chosen as a safe limit for lead-acids. The product of the battery's voltage and the Ah is the energy rating in Watts-hour (Wh) that the battery can deliver to a load from the fully-charged condition. The battery charge and discharge rates are described in terms of its capacity. For example, charging a lOOAh battery at C/10 rate means charging at 10 A. Discharging that battery at C/2 rate means draining at 50 A. At this rate, the battery will be discharged in 2 hours. 4.2.3 Battery Energy Losses There are two causes of losses in a battery: the electrode polarization effects and electrical resistance in the cell current's path. Polarization losses occur because reactants from the electrolyte must be transported to the surface of the electrode before they can react with the electrode's active materials. The internal resistance of the cell produces a voltage proportional to current, an IR loss. The internal resistance of the battery and electrode polarization both oppose current flow and the combination of this opposition is called the internal impedance or equivalent series resistance (ESR) of the cell. Energy that is lost through the internal impedance is dissipated as heat. When a cell is discharging, the polarization and I 2 R losses decrease the battery's open circuit voltage. When charging a cell, the charging source must supply a voltage greater than the open circuit voltage of the cell to overcome the polarization and IR losses. The losses are gradually dependant on the current. As shown in Figure 4-3, the losses are reduced when the discharge current decreases [16]. At lower discharge rates, a cell capacity increases because it delivers more of its stored chemical energy to the load and less to heat. > 'o > _ Open circuit voltage Operating voltage IRDrop Polarization Loss Discharge Current (A) Figure 4-3 The polarization and electrical resistance losses in batteries Chapter 4 Energy Storage Systems 30 4 . 3 Battery Types There are two basic types of batteries: • The primary battery, which converts the chemical energy into the electrical energy. The electrochemical reaction in the primary battery is nonreversible, it can not be recharged. A discharged primary battery is discarded. • The secondary battery, also known as the rechargeable battery. The electrochemical reaction in the secondary battery is reversible. After a discharge, it can be recharged by adding a direct current from an external source. This type of battery converts the chemical energy into electrical energy in the discharge mode and vice versa in the charge mode. 4.3.1 Primary Batteries Primary batteries were investigated as a backup energy supply for the base station. These batteries offer very good charge retention, excellent long term shelf life and constant capacity over a wide temperature range. Primary batteries are excellent for long storage, onetime use applications such as emergency standby. Since primary batteries cannot be recharged. They will eventually need to be replaced. For this system, primary batteries are not as effective as rechargeable batteries and offer no significant advantage. 4.3.2 Secondary Batteries Secondary batteries can be recharged repetitively making them attractive for the sea lion project. Secondary batteries are available in many different technologies. Six of the most common types of rechargeable electrochemistry were examined. They are as follows [22]: • lead-acid • vented industrial nickel-cadmium (NiCad) • sealed nickel cadmium • nickel-metal hydride (NiMH) • lithium-ion (Li-ion) • lithium-polymer (Li-poly) An overview of their characteristics is summarized as follows: Chapter 4 Energy Storage Systems 31 Lead-Acid This the most common type of rechargeable battery used because of its maturity and high performance over cost ratio. Although it has the least energy density by weight and volume, lead-acid batteries are most common due to their wide availability, low cost and known performance characteristics. When a lead-acid battery discharges, water and lead sulfate are formed, the water dilutes the sulfuric acid electrolyte, and the specific gravity of the electrolyte decreases with the decreasing state of charge. The recharging reverses the reaction in the lead and lead dioxide is formed at the negative and positive plates, respectively, restoring the battery into its originally charged state. The lead-acid battery comes in various versions. The shallow cycle version is used in automobiles where a short burst of energy is drawn from the battery to start the engine. On the other hand, the deep-cycle version is suitable for repeated charge/discharge cycles. Most energy storage applications require deep-cycle batteries [23]. Vented Nickel Cadmium The vented NiCad cell has positive electrodes made of cadmium and the negative electrodes of nickel hydroxide. They have several advantages over lead-acid batteries that make them attractive for use in stand-alone PV systems. Their benefits include longer life, deeper cycle life, less leakage current, survivability from excessive discharges and excellent low temperature capacity. Nickel-cadmium can operate at -40 °C and lose only approximately 20 % of their capacity, while lead acid can lose up to 80 % of its capacity. However, at -40 °C, the NiCad is limited to a discharge rate of 0.2 C. This type of battery is used in heavy duty industrial applications such as mining vehicles, railway signaling, emergency power and diesel engine starting. The nominal voltage for a NiCad cell is 1.2 V , and 2.1 V for a lead-acid cell [24]. The voltage of a NiCad cell remains relatively stable until the cell is almost completely discharged, where the voltage drops off dramatically. Currently, vented NiCads can be five to ten times more expensive than lead-acid. Due to the raw material cost, they will never be as low as lead-acid. Moreover, cadmium has recently come under strict environmental regulation pushing their cost even higher and limiting the availability for large capacity NiCad batteries. Chapter 4 Energy Storage Systems 32 Sealed Nickel Cadmiums Sealed NiCad batteries incorporate a chemistry to prevent a buildup of pressure in the cell caused by gassing during overcharge. As a result, cells can be sealed and require.no physical maintenance. However, this type of chemistry suffers a memory effect in which a battery that is repeatedly discharged to only a percentage of its rated capacity will eventually memorize this cycle pattern and limit further discharge, resulting in a loss of capacity. Regular discharge cycles are required to prevent the memory effect. Finally, sealed NiCad technology has poor long term voltage retention. A large self discharging rate loses approximately 20 % of capacity per month. The leakage current increases at higher temperatures. Nickel-Metal Hydride The N i M H is an extension of sealed NiCad technology, and offers an improvement in energy density over that in NiCad. The major construction difference is that the anode is made of a metal hydride. This eliminates the environmental concerns of cadmium. Another performance improvement is that it has negligible memory effect. The N i M H , however, is less capable of delivering high peak power, has higher self-discharge rate at approximately 30 % of the capacity per month. N i M H is susceptible to damage due to overcharging. N i M H do not perform well in cold temperatures and have limited performance at -20 °C. Lithium-Ion Lithium-ion technology offers three times the energy density over lead-acid. Moreover, the lithium-ion has higher cell voltage of 3.5 V. This requires fewer cells in series for a given battery voltage. On the negative side, the lithium electrode reacts with liquid electrolyte, creating a sort of oxidized film. Every time the cell cycles, the lithium is stripped away, a free metal surface is exposed to the electrolyte and a new film is formed. To compensate, the cell uses thick electrodes, adding into the cost. Lithium technology is expensive. The lithium-ion electrochemistry is also sensitive to damage from overcharging. Therefore, it requires more elaborate charging circuitry with adequate protection. Lithium-Polymer This is a lithium battery with solid polymer electrolytes. It is constructed with a film of metallic lithium bonded to a thin layer of solid polymer electrolyte. The solid polymer enhances Chapter 4 Energy Storage Systems 33 the cell's specific energy by acting as both the electrolyte and the separator. Moreover, the metal in solid electrolyte reacts less than it does with liquid electrolyte. However, its low temperature performance is inferior to lithium-ions. 4.3.3 Summary of Battery Technology The following tables summarize the advantage and disadvantage of the six most common ten rechargeable batteries. Table 6 Lead acid battery Advantage • High current output • No memory effect • Can be tailored for particular use (deep cycle, high discharge) • Low cost • Widely available • Low self discharge • Easy charging requirements • Easy state of charge indication • Good temperature performance • Available in maintenance free designs » High charge efficiency Disadvantage • Dangerous materials used (lead, sulfuric acid) • Heavy • Low energy density • Reduced capacity as temperature drops • Cannot be stored in discharged state • Deep discharging reduces life • Hydrogen emissions in some designs • Requires maintenance • Low cycle life • Maintenance in some designs Table 7 Vented NiCad battery Advantage Disadvantage High current output Flat discharge profile, nearly constant at 1.2 V until almost flat Excellent long-term charge retention High capacity Longest cycle life Rugged, reliable, copes very well with abuse such as heavy loads Long life even at low depth of charge Can be stored in any state of charge Excellent low temperature capacity and performance Environmental concern with the use of Cadmium Maintenance of electrolyte Five times the cost of lead acid Requires a more complicated charging system Chapter 4 Energy Storage Systems 34 Table 8 Sealed NiCad battery Advantage No maintenance Long cycle life Good temperature performance Long life even at low depth of charge Can be stored in any state of charge Disadvantage Environmental concern with the use of Cadmium Memory effect Requires batteries conditioning to prevent memory effect High leakage current High initial cost Less cycle life then vented NiCad Table 9 N i M H battery Advantage • Higher energy capacity than NiCad • No maintenance • Cadmium-free • Long cycle life • Can be stored in any state of charge Disadvantage • Lower power density than NiCad • Poor charge retention, extremely large leakage current • High initial cost • Unable to operate at low temperatures Table 10 Lithium ion battery Advantage • Medium cycle life • High energy density • High cell voltage . Light • No memory effect Disadvantage • Moderate self-discharge rate • Expensive • Complicated charging system Table 11 Lithium polymer battery Advantage • High energy density . Light • Longer cycle life than Lithium Ion Disadvantage • Poor low temperature performance • Moderate self-discharge rate • Expensive Chapter 4 Energy Storage Systems 35 4.4 Battery Selection In terms of battery performance, NiCad technology is the best choice for the sea lion project. The technology offers deep cycle life, very low current leakage, survivability from excessive discharges and excellent capacity retention at low temperatures. However, the amount of Cadmium required in large batteries and the environmental concerns makes vented NiCads expensive and difficult to acquire. Federal Battery, a local battery supplier quoted $2000 (Cdn) for a 12 V, 102 Ah vented NiCad battery manufactured by Saft. Refer to Appendix D for Federal Battery's contact information. Unfortunately, cost is the main reason why NiCads are not used in this system. While far from perfect, lead-acid batteries offer good performance over a wide temperature range when compared to other types and are used in the sea lion project. Lead acid technology is the most common type of rechargeable battery because of its maturity and high performance over cost ratio. 4.5 Lead-Acid Battery Technology There are many types of lead-acid batteries and most lead-acid batteries used in small PV power systems are similar in appearance to automotive batteries. These box-shaped batteries are constructed using prismatic cells. In the prismatic cell, the electrodes are shaped into large flat plates. Prismatic cell design is used when large, high capacity batteries are required. Cells designed for high current discharges have many thin plates to maximize the amount of plate area. Cells with fewer, thicker plates are stronger and have greater storage capacity. Most metal conductors react with acid. Pure lead resists corrosion but is too soft and weak to be used as a grid for most applications. Lead alloys are used to strengthen the plate structure. Antimony is a common grid alloy metal used. However, it has the unfortunate characteristic of migrating from the positive plate grid to the negative plate's active material. Antimony deposit on the negative plate causes a faster self-discharging rate and increases the water loss during charging. New grid formulations use calcium as the lead alloy. Tin is sometimes used with calcium. Calcium strengthens and stiffens the grid. Tin increases corrosion resistance and strength. Batteries with lead calcium grids are better in small solar power storage systems than other types because less water is lost from the electrolyte during charging and self-discharge rates are lower. Chapter 4 Energy Storage Systems 36 4.5.1 Plate Limited Versus Electrolyte Limited One of the primary characteristics of a lead-acid battery design is the balance between the amount of active material in the plates and the number and concentration of the electrolyte ions. The discharge reaction of a battery stops when either of the plates run out of active material (plate-limited) or when the electrolyte runs out of ions (electrolyte-limited). This affects how the battery behaves in different applications. For example, a plate-limited battery is easier to recharge after a very deep discharge because the higher concentration of acid in the electrolyte makes it a better electrical conductor than an electrolyte limited battery which has lost most of its sulfate ions from the electrolyte. A discharged plate-limited battery is less likely to freeze in cold weather than an electrolyte-limited battery because the higher acidity of the plate-limited battery electrolyte lowers its freezing temperature. On the other hand, an electrolyte-limited battery has reduced plate stress. This is an advantage when batteries are frequently cycled. 4.5.2 Flooded and Valve Regulated Lead Acid Batteries Prismatic batteries are classified whether or not the electrolyte is sealed inside the battery or is accessible for servicing through vents in each cell. Vented lead-acid batteries have free flowing electrolyte and are called wet or flooded cells. Some lead-acid batteries have immobile electrolytes and are called valve regulated lead-acid (VRLA) batteries. There are two ways to immobilize the electrolyte. Gel-cell electrolytes are set with a gelling agent and absorbed glass mat (AGM) batteries use micro-glass fiber matting to hold their electrolytes like a sponge. Gel cells are plate limited and A G M are electrolyte limited. A G M batteries offer lower impedance to electrolyte ion flow than gel cells. A G M offers better high current rate performance and are used with high current loads. However, for deep-cycle applications, gel cell batteries offer the greatest number of charge/discharge cycles before failure. A G M cells have half the cycle life of gel cells. A gel cell has no liquid electrolyte and can operate at lower temperatures than A G M or wet acid batteries. A battery charging at a fast rate, or connected to a charging source at too high a voltage, can release gases due to overcharge. In wet lead-acid batteries these gases are lost and require periodic maintenance to restore the lost water. When charging a lead acid battery beyond its 100 % state of charge, the charging current will electrolyze the water in the electrolyte and generate oxygen and hydrogen gas to be vented. Hydrogen is a concern since a 4 % mixture of Chapter 4 Energy Storage Systems 37 hydrogen in air is explosive. V R L A batteries recycle these gases. V R L A incorporates an oxygen recombination cycle which occurs at the negative plate and minimizes the generation of hydrogen gas. Generally, gel and A G M batteries have about 20 % less capacity, cost about two times more, and have a shorter cycle life than comparable flooded lead acid batteries. However, V R L A batteries do not need watering, are safer (no acid spill), can be placed in a variety of positions, have a slower self-discharge characteristic, and have lower ESR than flooded batteries [25]. 4.5.3 Battery Life The lifespan of a battery will vary considerably with its construction design and how it is cycled, charged, and maintained. The theoretical cycle life of V R L A technology is presented in Figure 4-4, [26]. At a 25 % depth of discharge (DOD), a battery has a life of 2250 cycles. Even i f the battery is cycled 15 times a month, it may not last 12.5 years. As explained in the following sections, a battery will age from grid corrosion and sulfation. 1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 Theoret ica l Life C y c l e s Figure 4-4 The V R L A battery's cycle life 4.5.4 Grid corrosion Over time, the acidic electrolyte gradually eats the positive plate. Thicker plates are directly related to longer life, so with other things being equal, a battery with thicker plates will last longer. Batteries are typically manufactured with a design life of 5, 7, 12, 20 and 25 years. Chapter 4 Energy Storage Systems 38 4.5.5 Sulfation Lead sulfate is formed due to the chemical reaction between the lead plates and the sulfuric acid during normal discharge. During recharging, the lead sulfate is dissolved back into the electrolyte, but a small fraction adheres to the battery plates and hardens into the lead sulfate. The sulfate can not be dissolved back and the result is a gradual loss of electrolyte and plate material. 4.5.6 Charging Lead-Acid Batteries To obtain optimum life from batteries, proper recharging is important. In the charging process, dc power is used to reform the active chemicals of the battery to their high energy charge state. A lead-acid battery can generally be charged at any rate that does not produce excessive gassing, overcharging or high temperatures. However, the battery's IR losses are less at lower charge currents. As the battery becomes charged, its voltage increases. If charging goes beyond the battery's capacity, gassing occurs. Charging is influenced by the ambient temperature, battery voltage and depth of charge. The following table is the full charge voltage at different temperatures for the 6 cell lead acid battery [22]. To prevent gassing, the charge current should be reduced to reasonable values as the battery reaches full charge. Some ac ripple with the direct charging current is tolerable because a battery has a large capacitance. However, a large ripple will cause damage due to excessive heating. Table 12 Full charge battery voltage at different temperatures Electrolyte 6 Cell Full charge Correction Factor Temperature (°C) Voltage (V) (V) 50 13.8 -0.54 40 13.98 -0.36 30 14.19 -0.15 25 14.34 0 20 14.49 0.15 10 14.82 0.48 0 15.24 0.9 -10 15.9 1.56 -20 17.82 3.48 Charging voltage is very critical in V R L A batteries because they are recombinant batteries. The recombination of oxygen and hydrogen replaces the moisture in the battery. Chapter 4 Energy Storage Systems 39 Therefore the battery is maintenance free and never needs watering. The sealing vent is important to the battery's performance. The cell must maintain a positive internal pressure, otherwise the recombination of the gasses will not take place and the cell wil l dry out. The valve must safely release any excess pressure that is produced during excessive overcharging. As the battery overcharges, the battery capacity decreases because less gas is left inside the battery to recombine [27]. 4.6 Lead-Acid Applications Lead-acid batteries are designed for three different applications. They are starting/ lighting/ignition (SLI), deep-cycle service and float service. 4.6.1 Starting/Lighting/Ignition Battery The SLI battery is optimized to produce high short duration currents required to start an internal combustion engine. It has many thin, highly porous plates so that large areas of active material can react simultaneously and produce high peak currents for very short time periods. The grids and internal conductors are large to minimize internal resistance and IR loss. The disadvantages for producing large currents is reduced storage capacity, poor deep discharge performance and lose of capacity after a few deep discharges. This type of battery is typically electrolyte limited. Automotive starting is the largest application for SLI batteries. The battery is usually discharged less than 1 % of its rated capacity to start a combustible engine and then constantly trickled by the alternator. 4.6.2 Deep-Cycle Battery The deep-cycle battery is intended as the primary power source and is designed to provide long discharge profiles at moderate currents. A large deep-cycle battery is also known as the traction battery because of its application in electric vehicles such as forklifts and golf carts. Deep-cycle requires large storage capacity and deep discharging recovery. Most deep-cycle batteries are plate limited. The batteries are designed with fewer, thicker, and denser plates than in SLI batteries so that as much active material as possible is packed into each cell. The grids are designed for strength and this often means using antimony alloys. Deep-cycle service causes stress on the positive plate's active material, so separators often include glass fiber mats to prevent plate shedding. Chapter 4 Energy Storage Systems 40 4.6.3 Float Service Battery The float service battery, also known as the stationary battery, operates as an emergency backup supply. For the majority of the time, the battery remains idle until the primary source fails. Long life, high charge retention and minimum maintenance are the characteristics most desired for float service. The battery's plates are made of lead or lead calcium so that self-discharge and electrolyte dry-out are minimized. Stationary batteries have large volumes of low specific gravity electrolyte to reduce corrosion and increase the time before electrolyte water replacement is required. A stationary battery is typically electrolyte limited. 4.7 The Trojan T-105 and The Prevailer PV30H Battery The sea lion application requires batteries with a deep cycle profile and minimum maintenance. A high current demand is not necessary, so SLI batteries are not needed. Stationary batteries offer great charge retention and minimal maintenance. However, their low specific gravity makes them vulnerable to freezing. Vented wet deep-cycle traction batteries, such as the Trojan T-105 are almost as inexpensive as car batteries and are much more robust. They are widely available and often used in PV systems. The T-105 is a deep discharge, 6 V battery with a 225 Ah capacity. 4.7.1 Cost From Federal Battery, the T-105 is quoted at $82.30 (Cdn.) This battery is preferred for the sea lion project because of its popularity, known performance characteristics and low purchasing cost. However, maintenance is required because water is lost as the battery cycles. As well, gassing inside the base station's sealed enclosure creates an explosion risk. Therefore, the Prevailer Dryfit PV30H is chosen. It is a 12 V gel type battery with a capacity of 95 Ah. The battery is priced at $90 (Cdn) from Federal Battery. Prevailer is manufactured by Sonnenschein (a division of Exide.) The cost for this gel type battery is low. A market surplus and its construction using cheaper materials makes the PV30H priced matched with the Trojan. The PV30H is a marine grade, deep cycle battery suitable for solar panel applications. It has a five year design life. A higher quality gel battery such as the Sonnenschein A400 is designed for 12 years but costs 2.5 times more ($225 Cdn each) for a 100 A h capacity. Chapter 4 Energy Storage Systems 41 4.7.2 Discharge Rate Batteries will self discharge over time. Gel batteries have the lowest discharging rate for all lead acid batteries. At room temperature, the PV30H wil l lose approximately 2 % per month. This rate increases with temperature and age. 4.7.3 Temperature Performance Climate data from Annette, A K , reveals a worst case temperature of -19 °C (refer to section 2.3.) Table 13 shows the electrolyte freezing point at different DOD. Table 13 Freezing point of DOD Depth of Discharge (%) Freezing Temperature (°C) 0 ^ 6 0 8 -57.4 15 -52.2 26.5 -39.5 38 -26.7 _ 60 -15.0 At a 25 % DOD, the PV30H can survive down to -40 °C without damage. As a battery's temperature gets colder, its capacity decreases. The capacity versus temperature is shown in Figure 4-5. At -30 °C, the battery will lose approximately 70 % of its capacity [28]. -50 -40 -29 -18 -7 4 16 26 Temperature (°C) Figure 4-5 The temperature versus capacity performance Chapter 4 Energy Storage Systems 42 4.7.4 Battery Sizing The capacity of a battery system is determined by two factors: • How much energy is needed? • How long must the battery supply this energy? Renewable energy systems should store between 7 and 14 days of reserved energy to compensate for bad weather and 10 days is chosen. The following table [19] indicates that three PV30H batteries will provide the necessary capacity for 10 days, at a 25 % discharge limit, compensated to the lowest operating point of -30 °C. The total cost for the batteries is $270 (Cdn.) Table 14 Battery sizing chart 1. Average Watt-hours/day 48 Wh/d 2. Average daily Amp hours (12 V battery) 4 Ah/day 3. Number of days of required storage capacity 10 days 4. The amount of Amp hours storage needed 40 A h 5. The discharge limit for the batteries 25 % limit 6. Divide line 3 by line 4 160 7. Loss of capacity from winter temperature @ -30 °C This enables the battery capacity to endure cold temperatures. 70 % loss 8. Multiply line 5 by line 6. This is the total battery capacity needed. 272 Ah 9. Divide line 8 by battery capacity: 95Ah 2.86 -> 3 batteries 4.8 UltraCapacitor Technology Ultracapacitors are essentially, high cycle life, high energy capacitors that stores energy electrostatically by polarizing an electrolytic solution. Though it is an electrochemical device (also known as an electrochemical double-layer capacitor) there are no chemical reactions involved in its energy storing mechanism. This mechanism is highly reversible, allowing the ultracapacitor to be charged and discharged hundreds of thousands of times [29]. hi applications requiring long term service, it has been shown that ultracapacitors are a practical solution [30]. Chapter 4 Energy Storage Systems 43 4.8.1 Long Life In terms of energy density, ultracapacitors are positioned between battery and electrolytic capacitor technology meaning an ultracapacitor stores more energy than a conventional capacitor of similar size and a battery will store more energy than the same size ultracapacitor. The energy in batteries is stored in chemical compounds and is converted directly into electric energy by means of an electrochemical oxidation-reduction reaction. This chemical reaction on the electrodes has a major influence on the aging of batteries. The charge/discharge cycles affects the life of batteries. In capacitors, the electric energy is stored directly as a positive or negative charge on the plates without any reaction on the electrode surfaces. With capacitors, there is no chemical contamination or oxidation on the electrodes. Capacitors rely on an electrostatic effect, which is purely physical and highly reversible. Ultracapacitors have a charge/discharge life rated over 500,000 cycles. Their operation life far exceeds the lives of batteries. As long as capacitors are within their operational voltage limits, there is no risk of overcharging and no need for accurate charge and time monitoring. 4.8.2 Discharge/Charge Capability An ultracapacitor can be charged to any voltage within its voltage rating. There is no minimum voltage requirement. A battery's life is closely related to its depth of discharge. Figure 4-6 demonstrates the charging profile of a battery and ultracapacitor [3]. To maintain life, a battery should not be discharged to less than 25 % of its capacity. Capacitors are not theoretically limited to a minimum discharge and can deliver all their energy without degradation. However, for applications that require a fixed power, it is not practical to discharge the voltage below 50 % due to the increasingly high currents required. For this project, the purpose is to utilize the maximum energy stored in the capacitor bank. The discharge is governed by the system's capability to handle the lowest voltage point. Section 4.8.9 discusses in more detail the energy capacity of the capacitor bank. A capacitor has the capability to charge in seconds, whereas, a battery often requires hours. Ultracapacitors will accept as much charge as possible and a limiting circuitry is necessary to protect the charging system from high inrush current. Chapter 4 Energy Storage Systems 44 Battery Charging Ultracap • — • Time Figure 4-6 The capacitor and battery charging characteristics 4.8.3 Modeling Ultracapacitor's ESR and EPR The capacitor's application dictates the type of capacitor model used for analysis. If the capacitor is in a system, where it is cycled slowly over seconds/minutes, than the classical equivalent circuit of Figure 4-7 can be employed [31]. The classical equivalent circuit is comprised of three components, the capacitance (C), the ESR and the equivalent parallel resistance (EPR). There is an energy loss associated with the charge/discharge cycle, which is a function of the ESR and charge acceptance. A battery has larger ESR and suffers from polarization losses. It is less efficient at storing charge than a capacitor (refer to section 4.2.3.) The EPR models the current leakage effect and will impact long term energy storage performance. The EPR is typically very large, resulting in a small leakage and is often neglected for discharges on the order of a few seconds. Figure 4-7 The ultracapacitor classical equivalent circuit model Chapter 4 Energy Storage Systems 45_ 4.8.4 Temperature Performance One of the major problems with batteries is a loss of performance in cold environments. Low temperatures reduce energy capacity, slows cycling rate and may cause damage if the electrolyte freezes. Some batteries are designed for very low temperature operation. Nonetheless, battery performance significantly degrades as temperature decreases. Since capacitors do not dependant on chemical reactions, they operate reliably down to —40 °C. A capacitor's ESR increases as temperature decreases. Based on 20 °C, on average, the ESR doubles at -40 °C [32]. However, as mentioned, a capacitor's ESR is small when compared to a battery with similar capacity. 4.8.5 Environment Ultracapacitors require fewer resources and materials to manufacture than batteries. Capacitors contain no acid, hazardous substances or heavy metals such as cobalt, lead or cadmium. In large ultracapacitor cells, aluminum accounts for 25 % to 30 % of the total weight which is easily recyclable. In comparison with batteries, ultracapacitors are considered more environmentally friendly. 4.8.6 Disadvantage of Ultracapacitors For a given size, batteries hold more energy than a capacitor. The primary disadvantage of ultracapacitors is their relatively low energy density limiting their use to applications in which relatively small quantities of energy are required before the capacitors require charging. An ultracapacitor cell has a maximum voltage of approximately 2.5 V . Many applications require cells connected in series to achieve a higher voltage. Series require balancing circuitry to prevent individual cells from exceeding their voltage ratings (next section). As well, ultracapacitor technology is fairly new and presently has a large purchasing cost. However, capacitors contain inexpensive materials and require fewer resources to manufacture than batteries. As the technology advances, the cost steadily declines, dropping more than half every two years. In 2001, a Maxwell PC2500, 2700 F capacitor cost $600 (Cdn) per cell. Today (2005), each cell cost $60 (Cdn.) Chapter 4 Energy Storage Systems 46 4.8.7 Voltage Balance Voltage distribution across series stacked capacitors is ideally a function of capacitance. If each component is identical, the voltage divides evenly. For example, 15 V across six capacitors should charge each cell to 2.5 V. However, variations in capacitance, from manufacturing and over the life of the product, wil l result in a voltage imbalance. Some cells charge less than 2.5 V , while others exceed the over-voltage limit [33], [34]. After the stack has been held at a constant voltage for a period of time, the voltage distribution then becomes a function of leakage current. The cells with higher leakage discharge faster resulting in a lower voltage. The total voltage across the stack is constant and must be redistributed across each component causing some cells to over charge. Cell balancing is an effective method for managing series stacked capacitors. There are two techniques for cell balancing passive and active. Figure 4-8 is a diagram of passive cell balancing. Passive balancing compensates for variations in leakage current and capacitance by placing a bypass resistor in parallel with each cell. A l l the parallel resistors are the same value and are typically sized at 10 times the average leakage current of the cell. The cells with higher voltages will discharge through the parallel resistance at a higher rate than the cells with lower voltages. This will distribute the stack voltage evenly across the entire series of capacitors. Passive balancing is most appropriate in low duty-cycle applications that can tolerate higher leakage. The benefits to this balancing method are simplicity and low cost. The drawbacks are a slow stabilizing response and a higher power loss due to 10 times the additional leakage current. Figure 4-8 Passive circuitry for voltage balancing Chapter 4 Energy Storage Systems 47 Active balancing circuitry is required in high duty-cycle applications or when low losses are necessary. Active circuitry has higher cost than passive balancing. However, it is used in the sea lion project because it offers lower power losses. Figure 4-9 shows a simplified diagram of an active balancing circuit using comparators. Figure 4-9 Active circuitry for voltage balancing The voltage across each capacitor is compared and charge is balanced to equalize the cells. Maxwell offers an integration kit that supplies the mounting brackets, fasteners and balancing circuitry for connecting capacitors in series. The active circuitry is capable of providing up to 300 mA of current to balance adjacent cells. When the cells are balanced, the circuit draws less than 50 uA. 4.8.8 Ultracapacitor Selection The capacitor energy storage bank is a module consisting of six Maxwell PC2500 Booststrap double layer capacitors. The capacitors are packaged in a durable, lightweight, hermetically-sealed aluminum encasement. Each capacitor has accessible terminals and is able to cycle hundreds of thousands without a reduction in performance. Their operating temperature is between -40 °C and 65 °C. They have a capacity of 2700 F and a maximum voltage of 2.5 V. To interface with the 12 V battery, the capacitors are stacked in a series configuration. Therefore, the stack's maximum voltage limit is 15 V and the equivalent capacitance is 450 F. Chapter 4 Energy Storage Systems 48 4.8.9 Energy Capacity Energy stored in a capacitor is a function of its voltage and capacitance such that „ CV2 (4) 2 The capacitor bank can be charged to any voltage within its voltage rating (V M AX) of 15V. At VMAX, the bank's capacity is 14 Wh or 50.6 kJ. At any other voltage level, the stored energy is expressed as a ratio of the current energy level divided by the total capacity CV2 y CURRENT % Capacity = ^ c c m r = \ x 100 = YsHMEHL x 1 00. (5) E CV V '-'MAX ^ MAX ' MAX 2 The voltage versus capacity is plotted in Figure 4-10. Due to the voltage squared term, the graph shows an interesting relationship between the voltage and the energy. At 15V, the capacitor is fully charged, a decrease of 5 V releases more than 50 % of the energy. Below 5 V , 90 % of the energy has been released. For the sea lion project, the capacitor storage system can operate between 4.5 V to 14.5 V . The total energy is 11.9 Wh (42.8 kJ.) If the base station operates at an average of 2 W, then the capacitor bank can provide 5.9 hours of operation. The leakage current is a time dependence of the self-dissipation of the capacitor. The PC2500 has a typical leakage current of 6 mA for a 72 h test at 25 ° C . Similar to batteries, the discharge rate is temperature dependant and declines as the temperature decreases [35]. 100 5 80 o a: LU / / / / y . / ,/ . / ' / / 0 5 10 15 Capacitor Voltage (V) Figure 4-10 The percentage of energy versus capacitor voltage Chapter 4 Energy Storage Systems 49 4.9 Energy Bank Topology To take advantage of the capacitor's wide voltage range, it is decoupled from the system by a dc/dc converter as shown in the figure below. Figure 4-11 The energy bank topology The capacitor stores surplus energy from the solar panel. The converter is a power regulator and is designed to prevent the capacitor from drawing excessive battery current. The converter is bidirectional. As the capacitor discharges, its voltage decreases, however, since it is decoupled, the battery's voltage is maintained [36]. The following chapter discusses power converters. Chapter 5 Power Converters Power converters transform power into a useful form. Depending on the application, converters are designed to maximize or regulate the energy flow. For instance, a M M P T is a type of converter used to exploit as much as energy from a solar panel by loading it at its power point. Another converter regulates the energy flow between the ultracapacitor and battery bank. Only dc/dc converters are used in the sea lion project. There are two basic methodologies for accomplishing dc/dc power conversion: linear regulation and switchmode conversion. 5.1 Linear Regulators Linear regulators are step down voltage converters. They reduce a higher input voltage down to a lower output voltage by linearly controlling the conductivity of a series pass power device (BJTs, MOSFETs) in response to changes on the output load [37]. This technique offers low output noise and excellent regulation characteristics. However, linear regulators operate the power device in the linear (ohmic) region with the load current flowing through it and the input/output voltage difference across it. The power loss is given by Significant heat is generated from dissipated energy, especially i f the input/output voltage difference is large. Due to this power inefficiency, the linear regulator is not used in the sea lion project. 5.2 Switchmode Power Converters Switching power converters offer higher efficiencies because they operate power switching devices in the full-on/full-off states. Furthermore, switchmode converters are not limited to down conversions. They offer the options to: buck, boost and/or invert the output. Unfortunately, switchmode converters radiate more electrical noise (EMI) and have poorer regulation characteristics than linear regulators. However, with proper design techniques, the drawbacks of switchmode can be reduced to acceptable limits. There are many different switchmode power topologies. The most commonly used types are: • Buck: used to reduce a dc voltage to a lower dc voltage 50 Chapter 5 Power Converters 51 • Boost: provides an output voltage that is higher than the input • Cuk: an inverted output voltage that is less than or greater than the input • Non-inverting Buck/Boost: an output voltage that is less than or greater than the input • Flyback: an isolated output voltage that is less than or greater than the input, it also offers multiple outputs 5.2.1 The Buck Converter The buck converter used in this thesis is a non-isolated down converter used to reduce a dc voltage down to a lower dc voltage of the same polarity. Buck converters are frequently used in battery powered applications such as laptops, battery regulators and cell phones [38]. The topology contains a low number of components offering simplicity and reduced cost. The drawback of this buck converter is a direct dc path from input to output in the event of a shorted element. Isolated converters offer safety and protection but require a transformer. Isolation is not mandatory and due to the application's low voltage range, the buck converter is selected. 5.2.2 The Buck Topology The Buck topology is the most basic switchmode topology and is illustrated in Figure 5-1(a) and (b). It uses switching to reduce the output voltage. The switching action produces a rectangular waveform at Vs as is shown in Figure 5-1 (c). The average output voltage is the area underneath the on-time divide by the period (Ts). The on-time's duty cycle is represented by D. The inductor and capacitor are used to form a low pass filter. These elements are large enough to significantly remove nearly all the ac components. To analyze the steady state, the inductor volt-second and capacitor charge balance methods are used. The analysis is described in detail based on references [38], [39]. Chapter 5 Power Converters 52 a) b) Switch Switch Position 1 (Swl) Position 2 (Sw2) c) v s Sw2 Swl 0 +-1 D T S T s Figure 5-1 Buck converter topology and switching action As a rule of thumb, the low pass filter's natural frequency should be at least one decade less than the switching the frequency. For the sea lion project, the power converter output is connected to the storage bank. The storage bank has an extremely large capacitance value and its natural frequency is small. At steady state, the output voltage V L is shown in Figure 5-2 (a) and its current I L is Figure 5-2 (b). When the switch is in position 1, the inductor's voltage is the difference between the input and output voltage. The current ramps up until the switch engages into position 2. The inductor's voltage instantaneously changes to - V 0 u t and the current falls. At steady state, the inductor's volt-second will balance and the rise and fall times are repetitive. The ripple is represented by AI L , and is symmetrical about the average current I d c . Chapter 5 Power Converters 53 a) i V L b) Ii 1 M A X J-dc AM IN V,v- V, I N - V O U T V -V 1 IN OUT -v , O U T V, OUT I •> t AIt DT S Figure 5-2 The inductor's voltage and current signal AI L Switch Position 1 D = OnTime VL=L Ts dt di = 2AIL dt = DTS V -V -V L IN v C (change in IL) OUT V -V - L 2 ^ -m °UT DT, L = V, OUT 2AI, DT, (?) (9) (11) (length o f interval) (13) (constant voltage) (14) (15) (16) Switch Position 2 D'=l-D -V - 1 y OUT 2ML DT, L = V, OUT 2AI, DT, (8) (10) (12) Chapter 5 Power Converters 54 The equation to determine the inductor's value is derived from the figure above. Equation 16 calculates the inductor's size such that a desired current ripple A I l is a small fraction of the dc component. 5.2.3 IRMS Calculations Due to the switching action, the current flowing through the converter contains a dc and an ac component. The I R M s calculation for the inductor, switch 1 and switch 2 unfolds as follows: Inductor RMS Current In Figure 5-2 (b), the inductor's current I L is equal to the average current IdC plus the slope of the ripple. The slope (di/dt) is dependant on L, thus, I L and ILRMS are \dc-M + ^ t ,0<t<DTs (SW\) (17) I , + A I 2 A I _ _ ,DTs<t<D (SW2) "c (l-D)Ts s ' Lrms 1 rTS I\dt (18) 1 fO T J O ldc AI + 1 DZ s J dt+ £ •rs [ , . j 2AI I.+AI- t s — ]DTs\ dc (l-D)Ts (t-DT5) dt Al' High Side Switch (Switch 1) rms Current Switch 1 connects the inductor to the source. The current flowing through switch 1 is equal to (17) for 0 < t < D T S and 0 elsewhere. Its IRMS value is lSW\rms Dx 11 + AI 2\ (19) Chapter 5 Power Converters 55 Low Side Switch (switch 2) rms Current Switch 2 places the inductor parallel to the output capacitor. The current flowing through switch 2 is equal to (17) for D T S < t < T s and 0 elsewhere. Its RMS value is 5.2.4 Capacitor Bank dc/dc Converter The capacitor bank is able to draw large amounts of battery current and cannot have an instantaneous voltage applied across its terminals. Therefore, the bank's energy flow is regulated through the dc/dc converter. The converter is designed to utilize its filter inductor as a current buffer and can control the current using a hysteresis limiting technique. Where the hysteresis bandwidth HBAND equals twice the ripple current The inductor's current I L and its voltage V L are plotted in Figure 5-3 (a) and (b). To observe the converter's operation, A I L is exaggerated for clarity. In this design, the ripple is much smaller, approximately 3.5 % of the operating current. When switch 1 turns on, IL begins to ramp. The ramping slope depends on the inductor's value and the larger the value, the slower the rise. When the current reaches the upper limit, switch 1 is turned off. Switch 2 turns on to maintain the current's path. The capacitor will accumulate the charge causing its voltage to increase and the current to drop. When the current reaches its lower limit, the process repeats: switch 1 turns on and switch 2 turns off. The capacitor's voltage (VOUT) is monitored. When the maximum voltage limit (14.5 V ) is reached, charging is completed and the converter turns off. VOUT is plotted in Figure 5-3 (c). To discharge the capacitor, the switching action is reversed. Switch 2 becomes switch 1 and vice versa. The converter becomes a boost converter, since the capacitor voltage (now VIM) will be equal or less than the battery's voltage (VOUT )• The hysteresis control remains the same, except the current direction is also reversed. The capacitor is discharged to 4.5 V , releasing about 90 % of its capacity. The capacitor provides 11.9 W (42.8 kJ) of energy has a cycle life greater than 500,000 times. Section 4.8 discusses the ultracapacitor in more depth. (20) HBAND - 2AIL- (21) Chapter 5 Power Converters 56 Figure 5-3 The capacitor bank's charging characteristics Chapter 5 Power Converters 57 5.2.5 Maximum Switching Frequency Since the converter is hysteresis controlled, the switching frequency is a variable. Minimum switching occurs at the start or end of charging because it takes time for the inductor to store or release its energy. However, when the converter is operating within the hysteresis band of 2 A l , then the system is in its "steady state." At this state, during each cycle, the converter's input energy is transferred to its output. The volt-second area in position 1 equals position 2 (Figure 5-3 (b).) Thus, the maximum frequency occurs when the voltages of the capacitor and battery banks are equal. Using the volt-seconds method, f M A X can be calculated as follows. Within the hysteresis band, the volt-seconds are equal {VIN-VOUT)DTs=VourxD'Ts. (22) For the switch to be at the maximum, the time components must be at a minimum. This occurs when on/off times are equal D = D' (23) and since we have D + D'=l (24) D = l (25) 2 Thus, (22) becomes V -V =V ' IN ' OUT ' OUT (26) VouT = V-f- (27) The max frequency is T = T MAX 1 f ' (28) J MAX Substitute (25), (27), and (28) into (16) and solving for f M A x yields f (29) JMAX 4 L M -Chapter 5 Power Converters 58 5.3 Switch-Mode Power Inductors The primary function of the inductor in a switching regulator is to pass direct current and filter out components at the switching frequency and its harmonics. The value determined for Alt is somewhat arbitrary and it can be adjusted to obtain a practical value for the inductor. The size of the inductor is also dependant on the converter's switching frequency. At higher frequencies, the filter inductor does not have to store as much energy because it's being replenished more often. A smaller inductor means a reduced cost. However, the inductor's core losses (eddy currents, hysteresis) and the switching losses in semiconducting devices increase with operating frequency. For this application, the frequency can be selected and 50kHz is chosen for fiviAx. Rearranging (29), the inductor's size is chosen, such that the current ripple is a small fraction of the full load inductor current dc component IAVG v L = IN ~4{FMAXML) <3°) Two different types of core materials are commonly used for inductors used in switching regulators: powder cores and ferrite cores. Material selection depends on cost, volume, size/space limitations, and winding capabilities. Each material type has its advantages as described below: 5.3.1 Powder Cores Powder cores have a distributed air gap structure eliminating the fringing flux that occurs in a discrete gap design. This material gives a soft saturation, allowing them to be used at high dc levels without saturating. They can store considerable energy, which leads to reduced size and weight and are therefore used in inductor applications [40]. However, at high frequencies, powder cores have more loss than ferrite. Powder materials have a low resistivity leading to larger eddy currents. As well, compared to ferrite materials, powder cores have smaller permeability and require more copper turns which results in increased winding resistance, to achieve the required inductance. 5.3.2 Ferrite Cores Ferrite cores offer the advantages of lower cost, larger permeability and less loss. In this application, at low frequencies, the inductors are dominated by the copper losses and the ferrite Chapter 5 Power Converters 59 core's higher permeability reduces the number of copper turns. Powder ferrite cores have a low saturation density. Thus, to avoid saturation, a discrete air gap is added. An air gap also minimizes temperature effects. As the ambient temperature fluctuates, the permeability of ferrite cores also changes. The air gap has a constant permeability and it helps stabilize the inductor's value. The larger the gap, the less the temperature effect, however, at a penalty of additional copper turns [41], [42]. 5.3.3 Inductor Losses In a typical switching dc/dc converter, i f the inductor is not properly designed, significant energy can be lost in the filter. The dissipated power arises from two sources: the losses associated with the inductor core and those associated with the inductor's winding. In this project, the inductor is dc biased. If the ripple current is kept small, the inductor has small core losses and the dc copper losses dominate. Winding Losses Power loss in the windings is due to the winding's dc and ac resistance described by ^WINDING_LOSS ~ ^dc^dc + rms) Rac' (31) From (18) the rms value of the ac current is and the dc resistance equals rms = S (32) _ Length R d c - P — • (33) Where: • Length = length of wire • Aw - Cross area of the wire • p is the resistivity of the winding material For copper p is p = \. 724x10-8[l + 0 . 0 0 4 2 ( r ° C - 2 0 ° C ) ] . % ( 3 4 ) With increasing frequency, more current is present in the outer layers of the wire, due to the skin effect (5). The skin effect is where alternating current tends to avoid travel through the center of a solid conductor, limiting itself to conduction near the surface. This effectively limits the available cross-sectional conductor area to carry alternating electron flow, increasing the Chapter 5 Power Converters 60 equivalent ac resistance of that conductor above what it would normally be for dc. The phenomenon is caused by a changing i(t) within the conductor. Induced eddy currents flow in a manner that tends to oppose the ac flux. The eddy current reduces the net current density in the center of the conductor and increases the net current density near the surface. The resultant is an ac and dc resistance as shown in Figure 5 -4 . Cross sectional area of conductor available for conducting dc current: "dc resistance" Cross sectional area of the same conductor at low frequency: "ac resistance" Cross sectional area of the same conductor at high frequency Figure 5 -4 The skin effect Skin depth 5 is the distance / shown in Figure 5 -4 . It is described by (35) S = Where for copper, u, = u,0= permeability = 47ie-7 H/m. The equivalent ac resistance is R"=P8x{d-S) (36) To decrease the copper losses, four threads of 18 gauge magnetic wire were braided together to form a 12 gauge "Litz" wire. Core Loss Within the core, energy is lost due to the changing magnetic field during each switching cycle. Energy is required to change the walls of the magnetic domains and the loss results in heat. This is called hysteresis loss and varies as a function of AB. Since this application drives the core with dc, the loss occupies the first quadrant of a B-H loop. The second type of core loss Chapter 5 Power Converters 61 is due to eddy currents. Magnetic core materials are iron alloys making them electrical conductors. As a result, changing magnetic fields causes electrical eddy currents to flow within the core material. These eddy currents cause I 2R loss in the core's material. Manufactures of inductor cores typically include curves, at various frequencies, for core loss versus flux density. Such curves are plotted based on sinusoidal excitation bipolar flux swings. When estimating the core loss for dc-dc converters that operate with unipolar flux swings and rectangular applied voltages, the core losses can be approximate using the converter's fundamental switching frequency and one-half the ripple's peak-to-peak flux density [43]. 5.4 Inductor Design The converter is designed to operate with a maximum operating frequency of 50 kHz. The ripple current is chosen at approximately Al|_ « 0.1 mA. At f M A X = 50 kHz, AI L = 0.1 A, and V i n = 12 V , the inductor is calculated using (30) as L = = 0.6 mH. (37) 4 x 5 0 x l 0 3 x 0 . 1 The inductor is designed to operate at a maximum of 5 A. T D K Corporation and Magnetics Incorporation manufacture ferrite cores specifically designed for switch mode power inductors. TDK's PC40 and Magnetics' R and F type ferrite materials are all suitable products. Magnetics' products are easier to acquire and cores using the F material are analyzed. The F material has a higher permeability than its R material counterpart. From Magnetic's datasheet [42], the F material can operate with a magnetic flux density (B) equal to 0.3 T without saturating. The material's permeability (pr) is nominally: 3000 @ 23 °C. The permeability changes with temperature: 1300 @ -30 °C to 4000 @ 100 °C. An air gap is added to the core to stabilize the permeability and to permit a maximum current of 6 A without saturation. To optimize the inductor's design, a magnetic circuit analysis is performed on all the different core shapes available from Magnetics. The geometric dimensions for each core are inserted into a spreadsheet. The core/winding losses, air gap, inductance variation are calculated for all geometries. From the results, an optimized core can be selected. The equations used for the analysis are derived below. The derivations assume that the magnetic fields are uniform and contained within the core. Proximity and fringing are neglected. Figure 5-5 illustrates the inductor's physical model and its electrical equivalent circuit. Chapter 5 Power Converters 62 Effective length of the magnetic path / s 0 © © 0 © 0 0 ® X X X X X X X X <F=9{[ \ - J F, < Rs = + uuA„ o s e s < > R„ =• (a) (b) Figure 5-5 The inductance (a) physical model and (b) electrical equivalent circuit The inductance L is L = N2 = N2Mo Rs+R • + - (38) Ms A A S Where the number of turns N can be found by rearranging (38) such that N = L Mo To avoid saturation we from the equality BMAX = B g = B S -(39) (40) Where the subscripts g and s denote the air gap and ferrite core respectively. Subscript x represents both g and s Bx=/xH N*IMAX =Hsls+Hlg Substitution of (41) into (42) » =uHr >NxI B f i \ MAX X XX MAX Mo + L \Ms J (41) (42) (43) likewise using (39) for N in (43) L I. K M0{M5A AgJ xl B MAX MAX Mo \MS (44) Chapter 5 Power Converters 63 rearrange (44) and solve for l g yields L = JMAX LAeu0us - 2BMAX AeAgls + ^(IMAX LAeu0us) + (2IMAXBMAX) Lu0uJs(Ag - Ae) 2BMjAeAgus (45) where: Ae = effective magnetic area lg = air gap length Ag = air gap area u0 = permeability=4nl0'7 (H/m) us — ferrite permeability B = magnetic flux density (T) BMAX — maximum flux density H = magnetic field (A/m) IMAX = maximum current 5.4.1 Inductor Selection The dimensions for Magnetic's inductor cores are shown in Table 16. Using the equations above, the inductor's properties (Table 15) are inserted into a spreadsheet and the losses for all Magnetic's products are determined at the PV's maximum power point. Table 17 shows the resulting top 15 cores with the lowest losses. Table 15 Magnetic's inductor properties Temp (°C) 20 Inductance (mH) 0.6 Resistivity (Q/cm) 1.72x10"° Frequency (kHz) 50 Skin depth (cm) 2.96xl0"2 Max current (A) 5 Wire gauge 12 BMAX (Telsa) 0.28 Wire diameter (cm) 0.213 IDC(A) 3.7 (as max 4000 Ripple Current (Al) 0.1 us min 1300 A B (Telsa) 5.6x10"3 uo 1.26xl0"6 Saturation 6 A Chapter 5 Power Converters 64 Table 16 Magnetic's inductor core dimensions C o r e No . Par t No . C o r e S h a p e Magne t i c Pa th (cm) C o r e A r e a (cm 2 ) V o l u m e (mm 3 ) C e n t e r Wid th (mm) C e n t e r Length (mm) Cen te r A r e a (cm 2 ) 1 4 9 9 2 8 E C E - E 27.4 7.38 202 27 .5 27 .5 7.56 2 4 9 9 2 5 U C U - U 30.84 6.45 199 25 .4 25.4 6.45 3 4 9 9 3 8 E C E - E 14.8 5.4 79.8 37 .5 14 5.25 4 4 5 5 3 0 E C E - E 12.3 4 .17 51.4 24.61 15.62 3.84 5 4 7 2 2 8 E C E - E 13.7 3.68 50.3 19 19 3.61 6 4 5 5 2 8 E C E - E 12.3 3.5 43.1 20.6 16.8 3.46 7 4 5 7 2 4 E C E - E 10.7 3.49 36.7 18.8 18.8 3.53 8 4 6 4 0 9 E C E - E 7.74 5.16 40 50.8 10.16 5.16 9 4 6 4 1 O E C E - E 8.02 5.16 41 .4 50.8 10.16 5.16 10 4 8 0 2 0 E C E - E 18.5 3.89 72.1 19.8 19.8 3.08 11 4 4 9 2 4 E C E - E 10.4 2.57 26.7 15.62 15.62 2.44 12 4 6 0 1 6 E C E - E 11 2.48 27.2 15.62 15.62 2.44 13 44721 E C E - E 8.9 2.39 21 .3 15.67 15.67 2.46 14 4 4 0 2 2 E C E - E 9.84 2.37 23.3 20 11.9 2.38 15 45021 E C E - E 9.29 2.25 20 .9 14.6 14.6 2.13 Table 17 Top 15 cores with the lowest losses C o r e No . Air G a p (mm) N m H total wire length (cm) dc W i re (mQ) a c Wi re (mQ) I R d c L o s s (mW) IR a c L o s s (mW) C o r e L o s s (mW) PTOT (mW) 1 0.20 15 0.66 165.00 7.98 16.70 86 .94 0.06 1.09 88.08 2 0.24 17 0.63 172.72 8.36 17.48 91 .00 0.06 1.07 92 .13 3 0.39 21 0.64 216 .30 10.47 21.89 113.97 0.07 0.43 114.47 4 0.57 28 0.61 225 .29 10.90 22 .80 118.70 0.08 0.28 119.05 5 0.61 30 0.62 228 .00 11.03 23.08 120.13 0.08 0.27 120.48 6 0.64 31 0.60 231 .88 11.22 23 .47 122.17 0.08 0.23 122.48 7 0.63 31 0.63 233 .12 11.28 23 .60 122.83 0.08 0.20 123.10 8 0.43 21 0.61 256 .03 12.39 25 .92 134.90 0.09 0.22 135.20 9 0.43 21 0.61 256 .03 12.39 25 .92 134.90 0.09 0.22 135.21 10 0.68 35 0.63 277 .20 13.41 28 .06 146 .05 0.09 0.39 146.53 11 0.94 44 0.61 274.91 13.30 27 .83 144.85 0.09 0.14 145.08 12 0.94 44 0.60 274.91 13.30 27 .83 144 .85 0.09 0.15 145.09 13 0.94 44 0.61 275 .79 13.34 27 .92 145.31 0.09 0.11 145.52 14 0.97 46 0.63 293 .48 14.20 29.71 154.63 0.10 0.13 154.85 15 1.09 51 0.62 297.84 14.41 30 .15 156.93 0.10 0.11 157.14 Chapter 5 Power Converters 65_ 5.4.2 Inductor Construction As predicted from theory, by choosing a low switching frequency, the dc resistance calculated in Table 17 contributed more than 96 % of the losses. The Cores 1 and 2 are efficient because their magnetic paths are large and their wire lengths are short. However, the cores cost $30 (US) per core, thus, making them too expensive. Core 7 is chosen because 3-6 were not available from the distributor. Core 7 cost $4 (US) per core. At the time of purchase, the distributor required a minimum order of 300 bobbins. Unfortunately, only two bobbins were needed for prototyping and to save cost, the bobbins were fabricated from high density polyethylene (HPDE.) The HPDE material was obtained from shampoo bottles. For core 7, the air gap is calculated to be 0.63 mm. Paper was used as the air gap spacer. The inductor is inserted into a buck converter as shown in Figure 7-14. At a switching frequency of 50 kHz and a current of 3.73 A, the inductor's losses are measured at 142.3 mW. A saturation test is also performed and the resulting plot is shown in Figure 5-6. The plot is current versus time. The plot illustrates the inductor's di/dt slope. Saturation does not depend on time. However, in the plot, the saturation point is observed when the slope dramatically changes at approximately 6 A. This matches the saturation limit specified in Table 15. The spikes at the beginning are due to contact bounce from the experimental results. Figure 5-6 The inductor's saturation curve Chapter 6 Semiconductor Switches As power conversion relies more on switched mode applications, semiconductor manufacturers are continuously improving technology to create products that approach the ideal switch. The ideal switch would have: • zero resistance or forward voltage drop in the on-state • infinite resistance in the off-state • switch with infinite speed • would not require any input power to make it switch Unfortunately, solid-state switching technology is not perfect. The designer needs to have an understanding of available technology to choose components that best suit their application. Good engineering is about finding the best compromise. To select a suitable device for the sea lion power project, the following sections give a comparison of power switches. The challenge is obtaining a device with a high enough breakdown voltage, fast switching speed while maintaining negligible switching losses and minimum on-resistance. The tradeoff is between breakdown voltage and on-resistance. The breakdown voltage of a semiconductor switch is a function of the doping level across its reverse-biased p-n junction. Obtaining a high breakdown voltage requires low doping concentration and hence, a high resistive region. This is particular with majority carrier switching devices, such as, the MOSFET and Schottky diode. Their on-resistance is directly dependant on their rated voltage. However, for minority carrier devices, the minority carriers are injected into the lightly doped, high resistance region by the forward biased p-n junction. The resultant is a high concentration of carriers which reduces the on-resistance of the device. Therefore, minority carrier devices exhibit lower on-resistance than comparable majority-carrier devices. The p-n diode, the bipolar junction transistor family (BJT, IGBT) and the thyristor family (SCR, GTO, MCT) are all minority carrier devices [44]. In minority-carrier devices, the benefit of reduced on-resistance comes with the disadvantage of decreased switching speeds. The conducting state of a semiconductor switch is controlled by the presence or absence of charge quantities applied to the device's control terminal. The turn on and turn off switching times are the times required to insert or remove this 66 Chapter 6 Semiconductor Switches 67 charge. The total amount of controlling charge in minority carrier devices is much greater than the charge required to control an equivalent majority carrier device. Although the mechanism for inserting and removing the controlling charge of switching devices can differ, minority carrier devices exhibit switching times that are significantly longer than those of majority carrier devices. Therefore, as shown in Figure 6-1, majority carrier devices find application at lower voltage levels and higher switching frequencies, while the reverse is true of minority carrier devices. Voltage 5000V 1000V 300V SCRs, BJTs** IGBT IGBT IGBT MOSFET** IGBTS MOSFETS MOSFETS MOSFETS MOSFETS 25kHz 100kHz •Frequency ** The choice of switch depends on the application requirements Figure 6-1 Switching device selection at different frequencies and voltages A great variety of solid state devices is available on the market. Some of the more commonly used devices are as follows: • bipolar junction transistor (BJT) • metal-oxide semiconducting field effect transistor (MOSFET) • insulated gate bipolar transistor (IGBT) • silicon controlled rectifier (SCR), also known as thyristor • gate turn off thyristor (GTO) For specific applications, the choice depends on the system's power, voltage, current and frequency requirements. Each technology has its own advantages in its own applications. BJTs Chapter 6 Semiconductor Switches 68 and SCR's have high voltage ratings. SCRs have the highest and are used for very high power applications (>1000 V). SCRs are slow and their drive circuitry can be complicated. 6.1 Bipolar Junction Transistors A BJT is a current controlled device where the device is turned on and off by controlling the current (IB) into its control terminal. A BJT's drive circuitry can also be complex because the base current must be continuously supplied in a quantity sufficient to maintain saturation. This current is proportional to the on-current flowing through the switch. Since the on-current can vary according to load conditions, the drive current must be sensitive to variable load conditions. Base currents of one-tenth of the collector current are typical to keep B JTs in saturation. When a BJT is conducting current, a forward voltage drop is developed across the device terminals (collector to emitter) which give rise to conduction losses. BJTs have other performance and control limitations. Since BJTs are minority carrier device, they have relatively slow switching characteristics, limiting their maximum operating switching frequencies to the lower end of the kHz range. As well, BJTs are negative temperature coefficient devices. The switch is liable to thermal runaway: as the current increases, the temperature rises, boosting the gain of the transistor. The gain increases current and temperature and this positive feedback destroys the device. 6.2 Metal-Oxide Semiconducting Field Effect Transistors In general, the control driver for a MOSFET is easier to design. A MOSFET has high input impedance making its turn-on gate current negligible when compared to a BJT. The device is turned on/off by controlling the voltage across the gate-source junction (VGs)- This greatly simplifies the drive circuitry. MOSFETs remain fully on by simply keeping the gate voltage above a certain threshold. MOSFETs have fast switching characteristics, allowing operation into the MHz range. They also have a positive temperature coefficient allowing them to be paralleled to further increase their current handling capability and decrease their on-resistance (Roson)-For low voltage applications, power MOSFETs offer extremely low on-resistance. When a MOSFET is fully turned on, it exhibits a resistor like behavior and develops a forward voltage drop (IR drop) across its terminals (drain to source.) Similar to BJTs, this gives rise to conduction losses within the device. If the current is small, MOSFETs are more efficient because their IR drop is less than the turn-on voltage of an equivalent BJTs, IGBTs or SCRs. However, as Chapter 6 Semiconductor Switches 69 the current increases, the IR drop rises and eventually, it exceeds the turn-on voltage of other switches. This is illustrated in Figure 6-2. The voltage drop of a MOSFET and IGBT switch is plotted as the current through the devices increase. To equally compare switches, both switches were matched for identical ratings. A value of 600 V , 20 A is chosen because IGBTs are not manufactured with smaller voltage ratings. At lower values, where MOSFETs offer superior efficiency, the MOSFET has smaller conduction losses than the IGBT. Compared to other switches, MOFSETs are constrained to relatively low breakdown voltages. The voltage limit of metal oxide junction is bounded to 1000-1200 V. This confines the use of power MOSFETs in high voltage applications. Vds 0 I 1 & Vce o n @ 25°C vs. Drain/Collector Current (Components rated @ 600V, 22A) 10 UJ >8 a! D O s-Q V A w> 4 > ol IRFP22N60K MOSFET IRG4RC20F IGBT 10 Current (A) (log scale) Figure 6-2 MOSFET and IGBT conduction loss 6.3 Insulated Gate Bipolar Transistors An IGBT is a hybrid device, derived from MOSFET and BJT technologies. It was designed as a spin-off from MOSFETs to extend high speed performance into high voltage and high power applications while keeping the low forward voltage drop. Similar to MOSFETs, IGBTs are voltage controlled devices and have high input impedances. The operation of the IGBT is very similar to that of the power MOSFET. However, the resistance of an IGBT's n-doped region is much smaller than a MOSFET. Therefore, an IGBT's current ratings jumps five to ten times for a chip with the same area compared to a power MOSFET. IGBTs are faster Chapter 6 Semiconductor Switches 70 than BJTs but are still relatively slower and have higher switching losses than MOSFET. This is partially due to a tail current effect they exhibit when the device turns off. 6.4 Switch Selection BJT and IGBTs (bipolar switches) do not have ability to block reverse voltages and should be used in such a way that they are only required to block forward voltages. Bipolar switches are a single quadrant switch. They are only able to conduct positive voltages and positive currents. MOSFETs have the advantage of two quadrates. They are able to conduct negative current. For this application, MOSFETs offer less switching and conduction losses than other switches. The following sections explain the characteristics and selection of MOSFETs used in the sea lion project. 6.5 The Characteristics of the Power MOSFET To determine which MOSFET is optimal for the sea lion application, an understanding of MOSFETs is required. The following section gives an insightful description on how power MOSFETs operate and provides the equations required minimizing losses. A detailed explanation of the semiconducting physics is beyond the scope of this thesis. MOSFETs are fabricated in the form of arrays. This means that a single power MOSFET device is a parallel combination of thousands of individual cells, each cell being a MOSFET in itself. If there is no input on the gate terminal, the junction between the n region and p island between the drain and source is reverse biased. There is no significant source-drain current flow and the drain sits at the positive supply voltage. The only current flowing from the source to drain is the reverse leakage current, which is negligibly small. If a positive voltage, higher than the threshold value, typically 3 V is applied to the gate, the electric field creates an inversion layer linking the source and drain region. This provides a current path. As VQS continues to increase, the inversion area intensifies and reduces the resistance between the source and drain. To allow for manufacturing variations, the threshold window is usually defined between 2.1 V to 4.0 V for standard type MOSFETs and 1.0 V to 2.0 V for logic level types. 6.5.1 The MOSFET Operating Regions The MOSFET is off (cutoff region) when VQS is less than the threshold voltage VGS(TH)-When the MOSFET is driven above the gate's threshold voltage, it is in the ohmic region if Chapter 6 Semiconductor Switches 71 VCS-VCS(TH)>VDS>0. (46) In the ohmic region, the MOSFET behaves like a resistor and the current through the device is directly proportion to the applied V D S . IDS <* VDS (47) As VDS increases, the current also increases. However, when VDS exceeds the condition in (46), the MOSFET enters the saturation region. Once the saturation region is reached, further increase in V D s will only cause a larger voltage drop across the device and raise the power dissipation without an increase in current. The drain current is now independent of V D s and depends only on VGS- The drain current can be approximated by I DS = K^GS ~ ^GS(TH) ) (48) where K is a constant that depends on the device's geometry. In switch-mode power applications, V D s is usually greater than VGS- TO insure full turn-on and minimize the voltage drop, the gate voltage is over driven: VGS » VGS(TH) and (48) simplifies to IDS=KxV2s. (49) Based on (46)-(49), the following figure illustrates the MOSFET's operating regions. * Ohmic Region, Figure 6-3 The MOSFET's operating regions. Chapter 6 Semiconductor Switches 72 6.5.2 Parasitic Capacitances and Inductance MOSFETs have faster switching characteristics than bipolar devices because they are majority carriers. They have no excess minority carriers that must be moved into or out of the device as it turns on or off. As a result, the switching characteristics of the power MOSEFET are governed by its capacitance. The MOSFET's gate impedance is very large because the electrode layer is insulated from the n-p layer by a silicon dioxide layer (glass). While this significantly minimizes the gate's current (IGATE)> the capacitance between the drain and source layers increases. To turn the device on/off these capacitances have to be charged and discharged. The charge rate depends on the current sinking/sourcing performance of the drive circuit. The circuit model of Figure 6-4 illustrates the parasitic capacitances and inductance of the power MOSFET. Most MOSFET's datasheets do not refer to the C G D , C G s, and C Ds, but to the input gate's capacitance dss, the MOSFET's output capacitance Coss and the feedback capacitance CRSS [45]. Dr ive C i rcu i t r y Drive Circuit Resistance i — Vv \ Drive Cgd I Stray - L -Jnductence^ Cgs Stray Inductance 5 Cds [ Stray i Inductance Ciss" = Cos + C G D Coss = C D s + C D G C R S S = C G D Figure 6-4 The MOSFET's parasitic elements The parasitic capacitance depends on the device's physical dimensions, generally, the larger the MOSFET, the larger its capacitance. There is a tradeoff between lower on-resistance versus capacitance. Larger MOSFETs have larger cross sectional area, resulting in lower Roson-At the same time, the switching performance degrades due to the capacitances. The challenge is to find the optimal point for best efficiency [46]. The p-n depletion layer contributes to the MOSFET's capacitance. Since the depletion layer varies with the voltage, the capacitance is not Chapter 6 Semiconductor Switches 73 constant. The capacitors' values are a function of the voltage. For Qss, the most significant change in capacitance occurs in CGD because VDG has the largest voltage change. To insure full turn on, a MOSFET must be charged beyond the threshold voltage VGS(TH)-The gate driver must provide a high enough current to charge the equivalent gate capacitance CEI within the time required by the system design. Care should be exercised not to exceed the maximum voltage rating of VGS- Even i f the applied gate voltage is kept below the maximum rated gate voltage, the stray inductance shown in Figure 6-4, coupled with the gate capacitance, may generate ringing voltages that could destroy the oxide layer. Increasing the output impedance of the gate driver, dampens the ringing, however, it slows the charging rate. 6.6 Power Losses The power loss in any MOSFET is the combination of the switching losses (Psw) and the MOSFET's conduction losses (PCOND)-PT0T = Psw + PC0ND (50) 6.6.1 Conduction Losses Calculating conduction losses is straightforward: I2R losses in the MOSFET multiplied by the MOSFET's on-time, at the correct operating temperature Pcom=lL*RDs(l + *Tj)xD. (51) The on resistance is temperature dependant and R.DS(on)(\ + AT}) accounts for the change in Roson at the MOSFET's junction temperature (Tj). Typically manufacturers normalize Roson at 25 °C. This value is unrealistic because the power losses significantly escalate the MOSFET's temperature. To account for the resistance change, manufacturers include a datasheet of Roson versus Tj (Figure 6-5) [43]. A junction temperature of about 105 °C is usually a good starting point for commercial applications. At this temperature, the on resistance is typically 1.35 times larger than at 25 °C [48]. Chapter 6 Semiconductor Switches 74 Figure 6-5 The MOSFET'S normalized R D S o n 6.6.2 MOSFET Switching Losses As the MOSFET turns on/off, it undergoes two losses: the crossover losses caused by the crossover of V D s and IDS and the gate losses caused by charging/discharging the device's capacitance. The Gate Charge Loss The energy required to charge a MOSFET's input capacitance is lost during each switching cycle. When calculating the gate loss, the most common error is using the input gate's capacitance value (Ciss) from the MOSFET's datasheet [49]. Qss is a variable capacitance and depends on the applied gate voltage. Refer to Section 6.5.2. The gate's voltage is overdriven and Ciss is higher than the value obtained from the datasheet. The amount of total gate charge (QTOT) necessary for MOSFET turn-on is QTOT = QGS + QGD +QOD- (52) Where • Qcs is the gate-to-source capacitor charge • QGD is the gate-to-drain capacitor charge • QOD is the "overdrive charge" to insure full turn on Shown in Figure 6-6, QTOT is determined from the gate-charge plot typical supplied on MOSFETs' datasheets [50]. Chapter 6 Semiconductor Switches 75 20 > W) 16 03 -4—* > 8 12 l-l o 00 •a o > l D = 7 5 A s= 4 4 V -VD S= 28V -r / / / 0 20 40 60 80 100 QTOT Total Gate Charge (nC) Figure 6-6 The gate's charge plot Once QTOT is known, the power loss due to charging and discharging the gate capacitance is as follows. The drive current is the total charge divided by the charging time QTOT I DRIVE (53) TOT The drive current is related to the MOSFET's input capacitance QTOT _ ^ISS^GS T T Cancel common terms QTOT ~ ^issues _ The power loss can be found using the capacitor energy equation (54) (55) C V ^ISS' GS 'Charge P J Ch arg e Ch arg e ~ ECharge X fsw ^ISS^GS ft SW Ch arg e (56) (57) (58) Chapter 6 Semiconductor Switches 76 p _ QroT^Gsfsw (59) charge ~ 2 The charging loss occurs twice per switching cycle through charging and discharging ^Charge = QrOT^GsfsW (60) Crossover Loss To determine the crossover loss, the switching behavior of the high side (SW1) power MOSFET is simulated in Pspice. As shown in Figure 6-7 (a), a constant current source is applied at the MOSFET's gate (IGATE) and a load current is placed at the drain (IDS)- The driver's output impedance is R G - Figure 6-7 illustrates the path of IGATE as it flows through the parasitic capacitances. The switching characteristics of V D S and IDS are plotted in Figure 6-8 (a). VGS is plotted in Figure 6-8 (c). When the driver is initiated, IGATE flows in the directions shown in Figure 6-7 (a). Qss begins to charge and VGS increases linearly. Ideally, no current flows until V G s reaches the threshold voltage. However, there is a small leakage current between the source and drain. As VGS exceeds the threshold boundary, IDs climbs rapidly, Figure 6-8 (b). At this point, the parasitic inductance will cause VGS to deviate from its original slope (section 6.5.2.) Inductance in series with the source develops an induced voltage as the current increases. This voltage counteracts the applied gate voltage and slows down the voltage rise. Similarly, V D s will start to fall because of stray inductance in series with the drain. To avoid cross conduction, before SW1 is turned on, SW2 is turned off and SW2's intrinsic diode briefly supports the current. As IDs reaches the current flowing in the inductor (IDC-AIL), the diode is no longer providing a current path and goes into reverse recovery. The recovery charge creates a hump. This is evident at the end of timeframe T 2, in Figure 6-8 (a). After reverse recovery, the diode can now support the voltage and VDS falls rapidly. During this time, the MOSFET is in the active region because V D s is greater than VGS. According to (49), IDS is proportional to VGS and a constant IDS clamps VGS. In Figure 6-7 (c), the entire gate current now flows through CQD , causing both V G D and V D s to fall. As V D S falls, the MOSFET is between the on/off state and the device cross conducts. The product of VDS and IDs is the crossover loss shown in Figure 6-8 (b). The area underneath this curve is the total energy lost. This loss occurs twice per switching cycle: turn-on and turn-off. The gate's voltage (VGS) is dependant on the accumulated charge (Qx) stored in Qss- Therefore, the curve of VGS is identical to the charge Chapter 6 Semiconductor Switches 77 plot in Figure 6-6. Even though V G s is fixed, IGATE is still rising. The voltage across IGATE (VGATE) is ^GATE = VGS + I GATE X RG • (61) Since all of IGATE flows through CGD, the rate that CGD charges is V -V j _ • GATE ' GS GATE „ (6z) G T CGD x L\VCD GATE -~ " • (63) At Setting (62) equal to (63) yields W G D 1 oc • A ; R, (64) Equation 64 states that the change of VDG is dependant on the driver's output impedance. Decreasing the impedance will improve the charge rate and minimize the cross conductions. As V G D charges, the input capacitance increases and the slope of V D s is reduced. V D s continues to fall and the MOSFET traverses from the saturation region to the ohmic region (T4). Once V D S enters the ohmic region, VGS becomes undamped and continues its rise to the maximum gate voltage ( V G G ) . V D s completes its drop to VDS(ON). The MOSFET's final on-state voltage is equal to YDS(ON)=IDSXRON- (65) Chapter 6 Semiconductor Switches 78 Figure 6-7 The gate's current path Chapter 6 Semiconductor Switches 79 VDS and IDS Leakage Current b) c) VGG V' T H N D Gate Charge • Crossover Power L o s s | p M A X Gate Charge • G ate Voltage - - - --y-, _ ^  ? . / <^  n ^ ^ ^ T QGD ^ VOL) ^ T l T2 T3 T4 Gate Charge • T5 Figure 6-8 MOSFET turn on behavior Chapter 6 Semiconductor Switches 80 Power Dissipation due to cross conduction The following equations are derived from Figure 6-8 (a). The cross conduction can be determined from geometry of V D D and IDc rCROSS ~ 2 Jsw' V 0 0 / The time duration T 2 or T3 are a ratio of the charge and the gate current. Subscript x represent either T 2 or T3 such that T x = j ^ . (67) '•GATE Substitute of (67) into (66) P _ V D D X I D D (Qi+Qi) f rCROSS ~ ~j J sw (68) ^•l GATE The values of Q 2 and Q3 are unavailable from the gate charge plot. However, after time interval T 4 , VDS « VDS(ON)- Thus, an approximation is QCROSS = QGS QGD-Conduction occurs twice per period and (68) becomes p _ VDP X I DP JQcROSS ) f / 7 r . N rCROSS ~ j J SW 'GATE Equation (70) indicates that the best way to reduce the crossover loss is to increase IDRIVE and supply the largest amount of gate charge possible in the shortest time. When Switch 1 turns on, the high side MOSFET begins to conduct current from 0 A to (IDc-AI L). Thus, IDD for switch 1 is I D D = I d c - M L . (71) Likewise IDD for switch 2 is 100=1*+^. (72) Chapter 6 Semiconductor Switches 81 Total MOSFET Loss Finally, the MOSFET's total power loss is the sum of the switching and conduction losses ^TOTAL ~ ^CROSS ^CHARGE ~*~ ^COND (73) fsw+RDsoXlrJ*D(\ + LATX (74) 6.7 High Side Losses (Switch 1) When switch 1 conducts, PTOTAL becomes fsw + RDSo„ ) 2 x D(\ + ATj). (75) 6.8 Low Side Losses (Switch 2) The low side switching losses are less than the high side because the voltage across SW2 is equal to the smaller diode drop across it (V F ) fsw + R-DSon (/siV2rms )2(l + ATjX\-D). (76) 6.9 MOSFET Selection To suit a range of applications, MOSFETs are available with numerous ratings. To match the optimal device for the sea lion project, the entire product line of International Rectifier's TO-220, 55 V MOSFETs are analyzed. For each device, the following parameters are entered into a spreadsheet: QTOT, QCROSS, VF and RDSOIV Using the high side and low side equations, the losses are calculated at the P V ' s maximum operating point. Table 18 provides the testing parameters. The results for the top 13 most efficient MOSFETs are shown in Table 19. The IRFZ46Z MOSFET is determined to have the best efficiency for SW1 and SW2. Even though it has a relatively higher Roson, its low input capacitance makes it suitable for this application. TOTAL V D P X I DP X QcROSS ,{-) y j. ^^LTOT'GS * DRIVE CROSS TOTAL + QTOT ^GS PRIVE V F x { l d c + I , L ) x Q CROSS TOTAL + QTOT ^GS Chapter 6 Semiconductor Switches 82 Table 18 Operating parameters at the PV's maximum power point Parameter Units AIL 0.1 A IDC 4.2 A Ipv 2.9 A Swl l r m s 3.49 A R M S Sw2 l r m s 2.34 A R R N S V P V 17.3 V Duty: 0.69 Frequency 50 kHz V G S 14.5 V Gate drive current 0.5 A AT (temp coefficient) 0.35 Table 19 MOSFET selection Product RON (mfl) QTOT (nC) Q 4 (nC) Max Current (A) V F (V) Swi PcROSS (mW) S W 2 PcROSS (mW) PcHARGE (mW) Swi PcOND (mW) S \v2 PcOND (mW) Swi PTOT (mW) S\v2 PTOT (mW) I R F Z 4 6 Z 13.6 37 17 51 1.3 120.67 9.51 26.83 154.44 31.17 301.93 67.51 I R F Z 4 4 Z 13.9 37 17 51 1.2 120.67 8.78 26.83 157.84 31.86 305.33 67.46 I R F Z 4 8 Z 11 62 22 61 1.3 156.16 12.31 44.95 124.91 25.21 326.02 82.47 IRLZ44Z 13.5 85 18 51 1.3 127.76 10.07 61.63 153.30 30.94 342.69 102.64 IRL3705Z 8 130 28 75 1.3 198.74 15.66 94.25 90.85 18.34 383.84 128.25 I R F Z 4 4 N 17.5 55 24 49 1.3 170.35 13.43 39.88 198.72 40.11 408.95 93.41 I R F Z 4 6 N 16.5 65 25 53 1.3 177.45 13.98 47.13 187.37 37.82 411.94 98.93 I R F Z 4 8 N 14 70 30 64 1.3 212.94 16.78 50.75 158.98 32.09 422.67 99.62 IRF1010Z 7.5 80 40 75 1.3 283.92 22.38 58.00 85.17 17.19 427.09 97.57 IRLZ44N 22 81 20 47 1.3 141.96 11.19 58.73 249.82 50.43 450.51 120.34 IRF1010N 11 90 40 85 1.3 283.92 22.38 65.25 124.91 25.21 474.08 112.84 IRF3205Z 6.5 100 50 75 1.3 354.90 27.97 72.50 73.81 14.90 501.21 115.37 I R F Z 3 4 N 40 30 5 29 1.6 35.49 3.44 21.75 454.23 91.68 511.47 116.88 Chapter 7 Circuit Implementation and Evaluation The base station is digitally controlled using the low cost, Microchip's 18F2510 microcontroller (PIC18F.) The controller has a nominal clock speed of 20 MHz. However, to minimize power consumption, it operates at a reduced 4 MHz. The PIC18F's power consumption increases at a higher operating frequency and at a larger voltage supply. Consumption also varies over temperature, but this relationship is nonlinear. The microcontroller will only consume 44mW, with no peripherals, operating at its maximum limits of: 5.5 V , 20 MHz, at 85 °C. At half the clock speed, the power drops by 40 %. The power supply's schematic is shown in Appendix B. Voltage measurements are sensed using the PIC18F's analog to digital (A/D) converter. Ten input channels are multiplexed into one A/D. To maintain accuracy, the A/D uses a dedicated voltage reference (4.096 V) supplied by the LT1461 precision regulator. When necessary, a resistive network scales down the voltage measurement between 0 V to 5 V. To minimize power loss, the resistive values are large. An ultra low power, CMOS buffer is placed in series to provide a low impedance for the A/D converter. Currents are converted to a voltage signal using a 10 mQ, film strip, 0.5 W, 1 % resistor. The MAX472 is a bi-directional amplifier used to magnify VSENSE- The amplifier has a gain of 100 with an approximate 2 % error at 2.9 A. The amplifier is unique because it can sense bi-directional current with one power supply, yet avoid the penalty of lost resolution. The amplifier has two outputs, a voltage proportional to the current's magnitude and a sign pin. The base station's internal climate is monitored using the AD7414 digital temperature probe. The system is designed to have an operating temperature range that exceeds the outside climate. However, i f the base station's internal temperature surpasses its normal operating range, the system enters sleepmode. The temperature probe communicates with the microcontroller via the I C protocol and offers a one shot temperature measurement with 15 LIW power consumption during idle state. With the exception of the battery and capacitor storage bank, all the circuit's components used in the base station are capable of operating from -40 °C up to 125 °C with low drift. 83 Chapter 7 Circuit Implementation and Evaluation 84 The primary function of the microcontroller is to monitor and control the energy flow from the solar panel to the load. The system is capable of supplying on average 48 Wh per day. If necessary, the battery bank can provide very high power peaks. To minimize battery usage, power is supplied to the load using the following hierarchy: 1. PV panel 2. The capacitor bank 3. The battery bank This hierarchy is implemented into the microcontroller as shown in Figure 7-1. A controller is designed to operate the solar panel at its maximum power point. When the solar panel is unable to provide sufficient energy, the capacitor bank can supply 11.9Wh (42.8 kJ) of energy via a boost converter. Based on the power budget (section 2.6), this is 5.9 hours of continuous operation. When the PV's open voltage is larger than the battery's voltage, a maximum power point tracker attempts to operate the PV at its power knee. The microcontroller monitors the base station's power consumption and directs surplus energy into the storage bank. The battery's voltage and temperature are also monitored. As described in section 4.5.6, the microcontroller is capable of determining the battery's full charge voltage using a lookup table. When the battery reached its capacity, the excess energy is then redirected into the capacitor bank. The PV is shifted away from its power point after the capacitor bank is charged. The microcontroller regulates the PV to only supply the consumption energy. The maximum power point controller is discussed in the following sections. Chapter 7 Circuit Implementation and Evaluation 85 Start up Check Input Power (P) Sleep Micropic Check Load Demand (L) 1 No Provide Load Power Measure Battery SOC Provide Load Power Yes Check Cap SOC r No Check Battery SOC Yes Discharge Battery No available | power Sleep PIC18F I Provide Load Power Figure 7-1 Microcontroller state diagram Chapter 7 Circuit Implementation and Evaluation 86 7.1 Maximum Power Point Tracker A maximum power point tracking (MPPT) improves the solar panel's output power by operating the panel at its knee point on the V-I curve (section 3.4.) The optimum power point is achieved because the converter attempts to match the load's impedance at the solar panel's power point. There are many different techniques to establish the power point and the base station utilizes a digital controlled, hill climbing method. 7.1.1 Implementation of the Hi l l Climbing MPPT The tracker is a synchronous, buck type, dc/dc converter that is digitally controlled using the PIC18F. The MPPT's functional diagram and control loop is shown in Figure 7-2. The search routine is implemented by changing the duty cycle at a fixed step interval. At each step change, the controller compares the PV's output power before and after the iteration. From Figure 3-2, the power point for a PV can be estimated at approximately 78 % of its open voltage. This is not the exact location but a good starting point (VSTART) for the tracker to begin its search. The duty cycle of the P W M applied to the dc/dc converter is set to 0 % and the PV's open voltage (VOPEN) is measured. The duty is then calculated by measuring the battery voltage V B ATT- The following equation is used to determine the starting point "START Q 7 S x V • VI) \J. I O A V o p E N The MPPT is designed to operate in the continuous mode. The following table describes the nominal operating conditions for the MPPT. Table 20 Nominal operating conditions Symbol Definition Value Ipp maximum power point current 2.9 A v P P maximum power point voltage 17.3 V VOPEN open voltage 21.8 V WMAX maximum power yield 50 W VBATT nominal battery voltage 12 V D duty @ maximum power point 69.3 % Chapter 7 Circuit Implementation and Evaluation 87 P V P A N E L M P P T V, IN ^ r C „ V 0 PIC18F 'BATT i P W M ! i Generator f Duty i A 7 0 ! Controller (dP/dV=0?) -V, REF Controller Solar Panel VsTART ^ D Duty G IBATT ^. — w W i Figure 7-2 MPPT functional diagram and control loop The P-VandP-D Curve The maximum power point (MPP) is located on the power-voltage (P-V) curve in Figure 3-2 when dP_ dV = 0. (78) The microcontroller searches for the power point by changing the duty cycle of a P W M generator that drives the MPPT. The P V s output power is P(VPV) = VPVIPV. (79) Where: Vpy — PVs output voltage PPV = PVs output power Since the duty cycle is a ratio between VBATT and Vpy D BATT V PV By rearrange the duty cycle equation v y _ 'BATT D (80) (81) Chapter 7 Circuit Implementation and Evaluation 88 and substitute (81) into (79), the P(D) is a function of P(D) = V ^ ^ . ( 8 2 ) To determine i f only one maximum peak exist on the power-duty (P-D) curve, the P-V curve can be modeled using a simple parabolic approximation having the function P(VPV)^-l[VPV~Ax]2+A2. (83) Where A i and A 2 are constants. Equation 83 has one maximum point at dP, PV 0^-2[VPV - 4 ] * 0 dVPV " L P V " (84) Vpy«j-. (85) The P-D is modeled by substituting (81) into (83) p(D)*-[^-Ar+A2. Equation 86 has one maximum point at (86) dPpy w Q (8V) dDpv - 2 [ ^ - 4 ] * 0 (88) D * - ^ Z Z L . (89) A For battery charging applications, the MPPT's output voltage can be assumed constant. Thus, only one solution satisfies (88) D^—- (90) A The P-D curve has one maximum peak. It is shown in Figure 7-3 [51]. Chapter 7 Circuit Implementation and Evaluation 89 Maximum Power Point AD 0% Duty Cycle 100% Figure 7-3 Photovoltaic P-D curve Step Size The MPPT is discrete because the controller has a step size with a finite resolution. This is a limitation since the power point may exist between the incremental steps and (78) will not be satisfied. However, (78) can approach zero i f the step size (AD) is small enough. When AD « 1 then dP(D) _ P(D)-P(D-AD) ) Q d(D) ~ AD (91) Since VBATT is assumed constant then the P(D) function can be written in terms of the battery's voltage and current P(D) = IBA7T(D)xVBA7T. (92) Thus, by substituting (92) into (91), the battery's charging current can be used to directly control the duty cycle « VBATT P ) ~ I BATT CD - AD)] -» 0. ( 9 3 ) The microcontroller's P W M operates at 50 kHz and has a resolution of 62.5 mV. This enables AD = 1.25 %. Since the size is small, the controller can locate the power point within a reasonable proximity. The MPP is found when the following condition is valid IBATT(D-AD) <IBArT(D)<IBA1T(D + AD). (94) Chapter 7 Circuit Implementation and Evaluation 90 7.1.2 MPPT Logic Figure 7-4 shows the logic flow for the MPPT algorithm. As mention in the previous sections, the hill climbing start position begins at 78 % of VOPEN- The controller then seeks out the knee's true position by increasing the duty at fixed increments of 1.25 %. Start YES Duty = 0 Sense VOPEN, VBATT Calculate Duty @ 78% VQPEN NO Incr++ Set Duty Reverse = 0 Incr = 0 Sense IBATT Duty = (Duty + AD) Sense I B A TT A D = - A D NO YES > T m A n v K N O , 4 tL<r IBATT ^ 1BATT(1>AJJ) ! j>———• Found++ 5 min delay Return to Start YES Figure 7-4 MPPT logic flow Chapter 7 Circuit Implementation and Evaluation 91 At each increment, the controller compares the change in power by measuring the battery's current before and after the step change. If the current increases, then the controller continues to track in the forward direction. If the current decreases as the duty increases, then the MPP is in the opposite direction. The controller will reverse direction by decrementing the duty cycle (AD = -AD.) Regardless of the tracking direction, the controller is searching along the P-D curve in Figure 7-3. Eventually, it will reach the power point. However, based on (94), the controller can not determine the maximum point unless it compares the change in current with the next step. If the controller is truly at the MPP, then the new current measurement is less. Equation 94 becomes valid and the controller knows it found the peak. Equation 94 assumes that the solar insolation remains constant throughout the tracking process. During each tracking interval, three situations can cause (94) to become valid: 1. The tracker found the MPP 2. A false reading caused by noise 3. The solar insolation has changed To ensure the MPP is found, when (94) becomes true, the controller is programmed to reverse direction and continue the iterations. If the tracker is truly at the MPP, then based on the logic, at this location, it will oscillate ±AD. The MPPT is designed to oscillate several times to insure that it is at the maximum (Reverse = Found.) If noise caused a false reading or if the MPP has changed, then the controller will not oscillate and it will reorient itself and continue to track the maximum power. To minimize noise errors, the battery's current is measured by averaging 16 samples. To approximate (78), the step size is minimized. However, during tracking, when the MPP shifts, the small step size increases the time to find the power point. Therefore, there is a finite number of iteration. If the statement: (Incr = M A X ? ) becomes TRUE, then the system resets itself to establish a new starting point closer to the peak. Figure 7-5 is a simplified demonstration of the MPPT searching in the forward direction. To observe the MPPT operation, the searching point and step size are exaggerated for clarity. The step size is represented by the asterisk (*). In reality, the step size is much smaller and the starting point closer to the knee. The solid black dot represents the power point. It is not necessarily located the middle of the step size. Chapter 7 Circuit Implementation and Evaluation 92 c cu I O So la r Pane l ' s Current V e r s u s Vo l tage R e v e r s e = F o u n d ? Vo l tage (V) Figure 7-5 MPPT illustration Step 1: VOPEN and VBATT are measured. The searching point is determined as 78 % of VOPEN and the duty is set. Step 2: IBATT is measured. The duty cycle is incremented by AD = 1.25 % . Step 3 : IBATT is measured and compared to step 2. IBATT > [IBATT - AD] = TRUE. Increment duty. Step 4: IBATT is measured and compared to step 3. IBATT > [IBATT - AD] = FALSE. Decrement duty. Step 5: IBATT is measured and compared to step 4. IBATT > [IBATT - AD] = True. [Reverse=Found] = True? Yes stop. No, increment duty and goto step 4. 7.1.3 Input Capacitance The MPPT's draws input current in rectangular pulses. The chopping waveform causes the PV panel to shifts on its 1-V curve. To stabilize the system, a capacitor in parallel with the supply minimizes the voltage ripple and reduces peak current drawn from the input. The following sections determine the capacitor's RMS current and the minimum value to reduce the input voltage ripple to 0.5 % of V P P . Chapter 7 Circuit Implementation and Evaluation 93 Input Capacitor's RMS Current Calculating the input capacitor's rms current (Icrms) is necessary to prevent failure from internal heating within the device. The following figure is a plot of the capacitor's current during one period. Ir Figure 7-6 The input capacitor's current plot The capacitor's current is defined as follows \PP-\L+L\l \}PP 1 - - * V DTsJ for 0 < t <DT S for D T S < t < T s Using (18), the capacitor's RMS current is calculated as I Crms 1 lPp +D\ f 2 Al2 lL2-2lPPlL+ — V i Ipp T AT D T S : T s Ih+Al I L - A I (95) (96) Substitution of I, = — into (96) gives D Crms •pp 1-7J D + • A I x D Placing the variables from Table 18 and Table 20 into (97) yields I = 2 9' 1-0.69 0.69 0.1x0.69 = 1.95,4 . (97) Chapter 7 Circuit Implementation and Evaluation 94 Input Capacitance Value The input capacitor's minimum value is found from the following equation Rearranging (98) yields C W ^ = I . (98) IN A T IxAT C w = — • (99) From Figure 7-6, AT is the duty cycle's off time A T - { l ~ D ) A T ~ ~ P ' (100) 1 sw When the high MOSFET is off, the capacitor's current is a constant source supplied from the solar panel ! = (101) The voltage ripple is 0.5 % of Vp P A F = 2x 0.005 xVPP. (102) Substitution of (80), (100) and (102) into (99) yields lOOx \ P P YBATT\ m v F V v 1 0 3 ) ' PP1 SW ' PP Placing the variables from Table 18 and Table 20 into (103) gives a capacitor value of r ^ 100x2.9 n 12 Cm- (! ) = 103 juF. (]r)A\ 17.3x50* 17.3 l i U 4 j The minimum amount of capacitance to maintain a 0.5 % voltage ripple is 103 uF. However, a 103 uF capacitor is only rated up to 500 m A r m s . Thus, to provide sufficient power dissipation and to be conservative, two 470 uF aluminum electrolytic L X Z capacitors are placed in parallel with the PV panel. The L X Z series are extremely low impedance capacitors designed for switch mode power converters. Each component is rated at 63 V and can handle 2090 m A r m s current (@100 kHz) [52]. The total input capacitance is 940 uF. Chapter 7 Circuit Implementation and Evaluation 95 7.1.4 Settling Time Both the MPPT's input capacitance and the output inductance reduces the system's response time. For each duty cycle change, the MPPT must wait for the system to settle. At the maximum power point, the system's impedance matches the P V s output impedance RPP=^- = 6Q (105) IPP The input capacitors' time constant is calculated as Tc = RxC = 940xlO~6x6 = 5.64 ms. (106) The inductor's time constant is T l = L / R = 9 4 0 x l 0 ' 6 =i51fls. (107) 6 The capacitor has a larger effect on the setting time (ts). The settling time is calculated using the following equation 5r(VA~VC2) . VA ' 0 ° 8 ) Where Vci is the initial capacitor voltage and V C 2 is the final capacitor voltage. The step size can be written as AD={Vci~Vc2). (109) Therefore, during tracking when AD = 1.25 %, the minimum tracking setting time is ts > 5r x AD ts > 5x1800/^x0.0125 = \\2.5/JS. ( 1 1 0 ) At each iteration, the MPPT must wait at least 112.5 ps for the system to settle. At the starting point, the P V s open voltage is: 21.8 V . 5T(VC. ~VC7) ' ' a k < n i ) 5xl800xl0" 6x(21.8-0.78x21.8) , t > = 2ms. 21.8 The settling time for the starting point is 2 ms. Chapter 7 Circuit Implementation and Evaluation 96 7.1.5 MPPT Discontinuity In case of low insolation, the converter may enter the discontinuous conduction mode. The MPPT still attempts to find the power point. However, the system is designed for continuous operation and the optimal power point may become difficult to achieve. To improve the converter's efficiency, a synchronous converter is used in the MPPT. If the converter becomes discontinuous, the reversal current drawn from the battery can destroy the low side MOSFET. Thus, the MPPT circuitry has the ability to sense when the inductor's current reaches zero. The MAX472 current sensor has a sign pin that indicates the current's direction. When current in the inductor is zero, the pin latches low, triggering an interrupt within the 18FPIC. The low side MOSFET is then disabled. During the next cycle, the microcontroller will reset the interrupt so that the low side MOSFET can turn on when the high side turns off. The MPPT converter uses the inductor discussed in section 5.4. In case the duty cycle approaches 100 %, the MOSFET driver has a built in internal charge pump for dc operation and can provide a continuous high side state. This is also convenient for comparing the power difference between enabling the MPPT and connecting the PV directly to the battery bank. 7.1.6 MPPT Result In this PV application, the direct duty cycle controller simplified the tracking system. Since V B ATT is constant, this type of hill climbing method can be applied. The system automatically accounts for the effects of temperature and irradiation variations because the power is tracked while the PV panel operates. The hill climbing technique does have several disadvantages: slow tracking, finite step size, oscillation around the MPP and sensitivity to noise. However, the algorithm used to control the converter is designed to minimize these limitations. The following figure compares the difference between operating the BP350 PV panel at its power point, versus, connecting it directly to a deep cycle, 12 V G E L type battery. It is observed that the MPPT increased the output power by an average of 20.1 %. It is worth mentioning that this improvement is not a constant. It is dependant on the PV's output power and the battery's voltage. These variables are not fixed and can shift dramatically due to a change in load, battery voltage, temperature and/or insolation. The converter's circuitry is designed for efficiency. Switching components operate at a low frequency and are optimized to reduce loss. The converter's efficiency is measured at Chapter 7 Circuit Implementation and Evaluation 97 96.3 % in synchronous mode and 93.4 % when the low side MOSFET is disabled. This achievement is the result of properly selected MOSFETs, a well designed inductor and the implementation of low power, microcontroller based circuitry. 0) o -»—. CD CO - MPPT — Direct Connection :00 AM 11:30 AM 12:00 PM 12:30 PM 1:00 PM 1:30 PM 2:00 PM Time (hrs) Figure 7-7 The BP350's performance using the MPPT 7.2 Implementation of the Capacitor Bank dc/dc Converter The converter's theory is explained in chapter 5. As mentioned the capacitor's energy bank stores up to 11.9 Wh (42.8 kJ) of energy and operates between 4.5 V and 14.5 V. The functional diagram of the converter's controller is shown in Figure 7-8. The functions are structured as follows: Maintains the capacitor voltage between 4.5 V and 14.5 V Provides the current's hysteresis band that regulates the current from 400 mA to 5 A Outputs a voltage as a function of the capacitor's current Synchronous MOSFET gate driver Voltage limiter: Current limiter: Current sense: MOSFET driver: Direction controller: Controls the capacitor's energy flow as an input or an output Chapter 7 Circuit Implementation and Evaluation 98 PV (jP P a n e l 1 M P P T VBATT{t) I_ H i g h S i d e r*h 600u 0 V c a p Low S i d e Voltage Limiter i4) Vmi V c a p ( v ) V m a x ( ! ) ~7x A X T MOSFET ? t ) ^ Driver -r ' i + Direction Controller —j Fo rward=0 B a c k w a r d = l Imax /"+\ Hysteresis Current Limiter : 450F ABS (I) Current Sense Figure 7-8 UltraCapacitor converter's functional diagram To minimize components, the hysteresis current limiter and directional controller is integrated within the PIC18F. The microcontroller has an onboard comparator with a programmable inversion output. When the comparator output is in the noninvert mode, the dc/dc converter becomes a synchronous B U C K and the capacitor charges. When the inverter is enabled, the converter is in the synchronous boost mode and the capacitor discharges. The current is regulated by comparing the inductor's current to a reference value. A 100 mA ripple tolerance is added using positive feedback. This is shown in Figure 7-9 (a). The reference voltage is calculated as follows Chapter 7 Circuit Implementation and Evaluation 99 I REF I PV •I (112) Where I P V = P V s Current ILOAD = Current from the battery to the load The current through the inductor is controlled by means of a fixed hysteresis control loop shown in Figure 7-9 (b). The hysteresis width is the current ripple. The microcontroller calculates I R E F and generates a voltage representing I R E F using the AD5301 digital to analog (D/A) converter. The MAX472 current sensor provides a voltage signal directly proportional to the capacitor's current (ICAP)- The comparator compares the signal to IREF and generates a variable frequency. Using (29), the maximum frequency is calculated to be 50 kHz. a) AD5301 D2A 'PV" 'LOAD ^ IN OUT Invert 1 Nonlnvert 0 V(Current Sense) MOSFET Driver b) ILOAD Fixed Hysteresis: 200mV Figure 7-9 Capacitor current limiting topology and control loop The microcontroller's comparator has a response time of 600 ns and an offset voltage of 10 mV. The disadvantage of the hysteresis controller is that the current ripple is a function of the entire system's response time and not only the comparator. If the switching frequency is increased, the controller may become too slow. As well, the controller is sensitive to noise. Switching spikes or transients can cause premature triggering. Careful track layout is required to minimize the effect. Finally, the hysteresis band makes it incapable to operate at currents less than 200 mA. Chapter 7 Circuit Implementation and Evaluation 100 Figure 7-10 is a measurement of the converter regulating a charging current at 1 A. Using the MAX472 sensor, the current's magnitude is displayed on channel 1. As mentioned, the waveform is a 1:1 voltage representation of the sensed current flowing through the converter's inductor. The current ripple is approximately 200 mAPk_Pkand has an averaged dc value of 1.04 A. The comparator's output is displayed on channel 2. This signal is the input source for the MOSFET driver. The comparator compares the current's waveform (CHI) to V R E F - 1 V . The reference voltage is programmable and enables the current to vary from 200 mA up to 5 A. When the comparator's output is high (5 V ) , Swi turns on and the current increases. At 1.16 A, the current reaches the upper hysteresis limit and the comparator goes low. Swi is disabled and Sw2 turns on. As the capacitor bank stores the charge, the current ramps down. The capacitor bank can charge up to 14.5 V . However, this upper limit depends on the battery's state of charge. Due to the MOSFET's body diodes, the capacitor bank can not exceed the battery's voltage. The spikes on (Chi) are switching noise coupled into the oscilloscope's probe. In the figure, the 1 A charging current is chosen because the current's ripple and dc average are clearly shown in one caption. Current Sense from Max472 (Chi) 1 = V MOSFET Drive Signal (Ch2) • / M r T O W W ' , ' , ! " " ! ' ' , T , " ! M " • GND GND Chi 200 mV Ch2 5 V M 10 (aS Figure 7-10 The converter's current sense and drive signal at 1 A Chapter 7 Circuit Implementation and Evaluation 101 In Figure 7-11, the converter is operating at the photovoltaic's maximum power current of 2.9 A (Chi). To capture a change in the capacitor's voltage (Ch2), a large time scale is required (500 ms/div). This resulted in under sampling and Channel l 's measurement is not a true representation of the ripple frequency. However, the averaged current is accurate at 2.9 A. VcAP (Ch2) Current Signal (Chi) — 1 = V GND GND Chi 1 A Ch2 2 V M 500 ms Figure 7-11 The capacitor's converter operating in buck mode. To discharge the capacitor, the comparator's invert bit is enabled and the waveforms shown in Figure 7-10 are reversed. The MOSFETs also exchange switching positions; Sw2 becomes the high side MOSFET. When the comparator's output is low, Swi is activated and the capacitor discharges its stored energy into the inductor. The inductor's current increases to the upper threshold of the hysteresis band causing the comparator's output to change state and engage Sw2- To maintain the current flow, the inductor is forced to boost its voltage and the energy is transferred into the battery. The inductor's current then ramps down to the lower limit and the cycle repeats. The system will discharge the capacitor to 4.5 V . Figure 7-12 is a caption showing the capacitor's decreasing voltage and its constant averaged current. Chapter 7 Circuit Implementation and Evaluation 102 In theory, as long as the current is greater than the hysteresis band, the synchronous converter does not enter the discontinuous mode. However, at 200 mA, the converter will enter discontinuous mode causing a reversal in current. To avoid this problem, the current direction is monitored. This is captured on channel 3 of Figure 7-13. The measurements shown are in buck mode. When the current goes to zero (Ch2), the MAX472's sign pin changes state (Ch3.) The low side MOSFET is immediately disabled by engaging the high side switch (Chi.) There are two plots in Figure 7-13 because the oscilloscope used for measurements cannot save three waveforms per caption. To clearly indicate the zero crossing, the current is shown in both plots as Ch2. To avoid both MOSFETs conducting simultaneously, whenever there is a transition, the LT1158 driver follows a logical sequence to turn off one switch and turn on the other. When turn-off is initiated, VQS is monitored until it decreases below the turn-off threshold. Afterwards, the other MOSFET is turned on. Chapter 7 Circuit Implementation and Evaluation 103 Figure 7-13 The discontinuous mode signal to prevent reverse current 7.2.1 Start Mode The MAX472 is chosen because it is a bidirectional sensor with a full scale output (0 V - 5 V) in both directions. The benefit is a good resolution, where the sensor's output is directly proportional to the hysteresis. By connecting the MAX472 to the ultracapacitor, the current in the inductor is sensed regardless which MOSFET conducts. However, the MAX472's Chapter 7 Circuit Implementation and Evaluation 104 sensing pins must be 1.5 V higher than its output. Therefore, it can not measure the capacitor's charging current i f VCAP<(l.5 + ICAP). (113) The capacitor bank operates between 4.5 V to 14.5 V and at this range, the current sensor functions. Unfortunately, due to the leakage current, the situation arises when the capacitor bank must charge from a 0 V state. Below 4.5 V , the sensor's operation is problematic. Especially if the bank is fully discharged, then the sensor is rendered useless. A low duty cycle, open loop soft start was considered to charge the capacitor bank up to its operating voltage. With such a large capacitance value connected to the dc/dc converter, the output is effectively a short circuit. The current is limited only by the duty cycle and the converter's non ideal resistance. This resistance is small, estimated at 200 mQ. It is also temperature dependant. Therefore, a current sensor is also mounted onto the battery side. This sensor is only used during the start charge mode because its location cannot monitor the low side MOSFET. Synchronous mode is disabled when the capacitor's voltage is below 4.5 V . The MAX472's output pin is a high impedance current source that can be connected to other MAX472's output pins for current summing. When the converter is in the start mode, the current sensor connected to the capacitor is offline. When the convert operates normally, the battery's sensor is disabled and synchronous mode is enabled. 7.2.2 Capacitor Converter's Efficiency The efficiency of the capacitor converter is slightly less than that of the MPPT. The capacitor's converter is bi-directional and hysteresis controlled. It has extra circuitry and two sense resistors to regulate the current. These components contribute to the higher loss. However, the V I N of the capacitor converter is less then the MPPT which minimizes the additional loses. Using the operating conditions from Table 20, the capacitor converter's efficiency is measured at 95.4 % in synchronous mode and 92.1 % during soft start. 7.3 System Implementation As mentioned, the base station systems are controlled using the 18F2510 microcontroller. The controller's code is programmed in C language and complied using the CCS 18F complier. The code is included in Appendix C. Using the MAX232 IC, the microprocessor is interfaced to Chapter 7 Circuit Implementation and Evaluation 105 a PC via the serial port. The microcontroller is fully automated. However, it is also configured so a user can manually control various systems. For instance, the MPPT can be enabled or disabled. Using the keyboard as an input device, the capacitor bank can be charged or discharged with any desired current between from 400 mA up to 5 A. The base station's input power (the PV supply) is also adjustable by entering a fixed duty cycle. As well, the PC enables the microcontroller to display real time measurements onto a monitor. This information can be saved to create plots such as Figure 7-7. In addition, the microcontroller is programmed to notify problems, such as a communication error with the temperature probe. Finally, the system can be connected to the satellite transmitter via the serial port. The base station's internal temperature, battery/capacitor state of charge and the amount of power generation or consumption can be remotely monitored or controlled via the duplex link. For prototyping, the power supply's digital controller, MPPT and dc/dc converters are constructed as shown in Figure 7-14. The microcontroller and its digital components are located on the breadboard. This digital controlling system operates on 5 V logic obtained with the MC4063 switch mode power supply (refer to schematic in appendix B.) During operation, the digital circuitry consumes 130 mW (the 5 V supply output current was measured at Idc = 25.6 mA.) To minimize noise, the MPPT and converter's converter power circuitry (MOSFETs, drivers, low pass filters) are soldered onto a prototype board. The dc inductors described in chapter 5, are shown on the left of the photo. Next to the inductors is the serial port. The monitoring system and the energy banks are shown in Figure 7-15. A gel type battery with a lower capacity then specified in Table 14 was sufficient to test the power supplies ability to charge and discharge the battery bank. Similarly, a smaller capacitor bank (right side of photo) is also used for testing. The 450 F capacitance is too large for prototyping. To investigate the capacitor's dc/dc response at different states of charge, a 3 F capacitor bank provided quicker charge/discharge cycles. The smaller bank also permits the capture of the capacitor's voltage changing as shown in Figure 7-11 and Figure 7-12. The photovoltaic used for testing is shown in Figure 7-16. There are two 50 W BP350 photovoltaics in the photo. Only one is needed to provide the base station's power budget. The BP350 specifications are included in Table 5 of chapter 3. The PVs are located on top of the Macleod Electrical Engineering building at the University of British Columbia. Figure 7-14 The power supply's digital controller, MPPT tracker and capacitor converter Figure 7-15 The power supply with its energy storage banks Chapter 7 Circuit Implementation and Evaluation 107 Figure 7-16 UBC's two BP350W photovoltaic panels In the final design, the total weight of the power supply is approximately 95 kg. The three PV30H lead acid batteries are the bulk weight (90 kg). The capacity bank weighs 4.5 kg. The storage banks can supply sufficient power for wild life monitor system for 10 continuous days. Component pricing is shown in Table 21. The total cost of the power supply is $1043 (Cdn). This amount does not include the base station's housing, transceiver or satellite phone. Table 21 Component pricing Component Price (Each, $ Cdn) Quantity Total Cost ($ Cdn) PV30H Battery Capacitor Bank BP350 PV Panel 90 60 331 3 6 1 270 360 331 Microcontroller 12 1 12 Inductor 10 2 20 Additional Electronics 50 N / A 50 (resistors, caps, semiconductors) Total: $1043 Chapter 8 Conclusion The power supply is designed to be an autonomous system that provides the energy needs for a radio based, wildlife monitoring station. The purpose of this research is to extend the battery's service life and improve the base station's immunity to cold climates. In this thesis, two methods were used to accomplish the objective. The first undertaking was to extensively research, design and implement high efficient components to minimize battery demand. As a result, the PV, its electronics and the load are well matched for the application. The next task was to incorporate a battery/capacitor bank to store energy. By integrating ultracapacitor technology to create a hybrid energy storage system, the battery cycling is minimized. The ultracapacitors can cycle hundreds of thousands times without of reduction in performance. Therefore, the capacitor bank is the primary energy storage system. When the renewable source can not supply sufficient energy to the electrical load the capacitor will discharge as the primary storage system. The battery bank is required i f the capacitor bank becomes depleted. When the energy in both storage banks becomes low, the base station enters a sleep mode until the renewable energy returns. The capacity of the energy storage banks depends on three factors: the base station's energy consumption, the duration that the banks must supply this energy and the amount of available solar energy. The power supply operates continuous throughout the year. It is designed to function even during the shortest daylight hours in December. From the project's description in chapter 2, the base station's power is 2 W averaged over a weekly window. Insolation data obtained for the coast of south Alaska and British Columbia determined that a 50 W PV (the BP350) provides sufficient energy for the base station. To compensate for intermittency in solar power, the storage bank are sized to deliver sufficient energy for 10 days, at a 25 % discharge limit, compensated to the lowest operating point o f -30 °C. The power supply's output voltages are 12 V (V B ATT) and 5 V . The 5 V supply is a constant voltage (±1 %) with a 750mA output current capacity. The 12 V is the unregulated battery's output voltage and can vary from approximately 10.5 V up to 16 V. This range depends on the battery's depth of charge and its temperature. If required, the bank is capable of delivering 100W of peak power. Based on 108 Chapter 8 Conclusion 109 climate data from Table 1, the base station is capable to operate between the temperatures o f -30 °C to 45 °C. Beyond this range, it enters the sleep mode state. Chapter three discusses photovoltaic systems. In this chapter, the equations to determine the tilt angle for maximizing the PV's aperture are derived. Equation 2 calculates the angle at a specific day of the year as a function of the PV's latitude. The angle is set in December to compensate the PV's output power for the month's short daylight hours. To further enhance the PV's output, an MPPT interfaces the solar panel to the battery bank. As described in section 7.1.1, when using a hill climbing tracking technique, measurements observed an improvement of 20 %. In chapter four, a detailed investigation into battery technology determined that a lead acid, gel type battery is the most practical solution for this application. The Dryfit PV30H 12 V battery is recommended. It is a deep cycle, marine grade battery priced at $90 (Cdn) from Federal battery (Appendix D.) It has a capacity of 95 Ah and a five year design life. A higher quality gel battery, such as the Sonnenschein A400 is suitable for 12 years. It has the same equivalent capacity, but costs 2.5 times more. The base station requires three PV30H batteries for its storage bank. As mentioned, the bank provides sufficient energy for 10 days, at a 25 % discharge limit. As shown in Figure 4-4, this threshold extends battery life by minimizing damage from a deep discharge. The battery is capable of surviving a temperature from -45 °C to 65 °C. High efficient inductors are explored in Chapter 5. A theoretical model is developed to help optimize the core's efficiency. Using a spreadsheet, the model is applied to different core geometries. The resultant is a low loss inductor with a fixed 600 uH over the base station's operating temperature range. When connected to a buck converter, at 3.7 A , with a 50 kHz switching frequency, the measured losses are 142.32 mW. As shown in Figure 5-6, the inductor can handle up to 6A before saturating. In chapter 6, an investigation of MOSFET technology is conducted. The losses of the MOSFET in a buck converter topology are analyzed in great detail. The switching behavior is studied and equations modeling the high side and low side losses are derived. Similar to the previous chapter, the equations are inserted into a spreadsheet and the entire product line of International Rectifier's TO-220, 55V MOSFETs are analyzed. Using (75) and (76), the losses Chapter 8 Conclusion 110 are calculated at the solar panel's maximum operating point. The IRFZ46Z MOSFET is determined to have the best efficiency for the high and low side switch. Using the components shown in the schematic (Appendix B) a system is designed to hybridize the ultracapacitor bank and battery bank. The PIC18F2510 microprocessor controls the MPPT and monitors the energy flow. The ultracapacitor bank is decoupled from the battery bank by a dc/dc converter. This allows the capacitor bank to have a variable voltage range of 4.5 V up to 14.5 V. The capacitor's converter is hysteresis current limited. The converter avoids discontinuous mode by monitoring the current direction using the MAX472 current sensor. Once the current attempts to fall below the zero crossing, the high side MOSFET turns on preventing a reverse current flow. The capacitor's converter efficiency is measured at 95.4 % for synchronous and 92.1 % in soft start. However, the converter uses additional components to regulate the current, thus, its power loss is slightly more than the MPPT. The MPPT's efficiency is 96.3 % in synchronous mode and 93.4% when the low side MOSFET is disabled. Nonetheless, both converters are very efficiency due to properly selected MOSFETs, a well designed inductor and the implementation of low power, microcontroller based circuitry. In this project, it is possible to construct the energy storage entirely from ultracapacitors. However, ultracapacitor technology is expensive and for a given physical size, batteries hold more energy. The major drawbacks of capacitors are their relatively low energy density and high purchasing price. As a result, the storage capacity is largely limited by its economical cost. To maintain an inexpensive base station, the capacitor energy storage bank can only contain a module of six Maxwell PC2500 capacitors. This capacitor module provides an energy supply of 11.9 Wh (42.8 kJ.) Since the base station operates at 2 W, the capacitor bank can provide 5.9 hours of operation. Presently (2006), the capacitor bank costs $360 (Cdn.) However, capacitors contain inexpensive materials and require less resource to manufacture than batteries. As the ultracapacitor technology progresses, the cost is steadily declining, dropping more than half every two years. In 2001, a Maxwell PC2500, 2700 F (2.5 V) capacitor cost $600 (Cdn) per cell. Today each cell cost $60 (Cdn.) While researching this project, Maxwell introduced an improved product and outdated their PC2500. The new MC2600 capacitor is physical smaller, offers less leakage current, a higher threshold voltage, and double the guaranteed cycling life (one million cycles.) Therefore, as ultracapacitor technology advances, the economical limitation is decreased and the dependency for batteries in this type of application is also reduced. Chapter 9 Future Work 111 Chapter 9 Future Work At the present time, the power supply is a working prototype. It operates successfully at ideal conditions where the system's efficiency is compared to theoretical values. However, testing the power supply at its temperature extremes has not been performed. Although the design has accounted for temperature effects, long term usage and aging, the system has not operated under these conditions. Furthermore, the watchdog timer in the microprocessor is disabled for prototyping. After testing, it is recommended to enable this function as a safety feature. The reverse shoot through current that occurs in discontinuous mode of the synchronous converters is controlled via software interrupt. To improve reliability, protection should be hardware dedicated. The fuses connected to the positive terminals of the storage banks do provide short circuit protection, however, additional active short circuit protection is considered for future work. Implementing this system using the 18F2510 microcontroller was an alternative design approach. Although this microcontroller can perform most of the tasks, external components such as digital to analog converters, comparators, op amps and voltage regulators were required for additional functions. Migrating into DSP technology is an improvement since these functions are integrated in one package design. To conserve energy, idle components should enter a sleep mode. For instance, the temperature probe is programmed as a one shot measurement. It then goes into a low power, inactive state. Most of the integrated circuits selected for this project have a logic shutdown pin. An algorithm from the microcontroller can be designed to engage this feature. Components without this function can be upgraded by having logic MOSFET switches connected between their power pin and the dc bus. The equations that predict the MOSFET losses are only derived for the buck topology. Even though the capacitor's converter is low loss, the equations do not account for operation in boost mode. Modeling the MOSFETs for this condition could reveal a method to improve efficiency. An enhanced topology for the capacitor's converter is also suggested. The converter's hysteresis controller is prone to noise and the hysteresis voltage thresholds make it ineffective at currents less than 400mA. The switching speed is also limited by the controller's response time. If the switching frequency is increased, the controller becomes too slow. For future work, a Chapter 9 Future Work 112 charge pump system is considered as an alternative topology. To transfer the energy between the storage banks, this design has an intermediate, small value capacitor between the battery and the ultracapacitor. During half a cycle, the capacitor charges from the input and for the remaining period, it discharges into the output. Finally, implementing an algorithm to operate the MPPT in discontinuous mode could improve the PV's power output during a low light condition. As well, designing a system so that the solar panel can physically follow the sun is also considered. Currently, in this prototype, a battery blocking diode is placed to prevent the battery discharging into the solar panel. While using a Schottky diode minimizes the turn-on drop. A P-MOSFET with the appropriate driving circuit can further improve the efficiency. Bibliography [1] United States Department of Energy. "Averages of Solar Radiation for Each of 360 Months, 1961-1990, Annette, Alaska, W B A N N o . 26451" Solar Radiation Data Manual for Flat-Plate and Concentrating Collectors (1990), [2005] Available at: http: //rredc. nrel. go v/so lar/pubs/redbook/ [2] Google Maps. "Map Data," [2005], Available at http://maps.google.com/ [3] EPCOS A G Corporate. "UltraCap Double Layer Capacitors," A New Energy Storage Device For Peak Power Application, (2002), [2005], Available at: http://www.epcos.com [4] Paumanok Publications Inc. "Supercapacitors: Is This the Future of the Battery? Is This the Future of the Battery?," Passive Component Industry Magazine, September/October, pp. 8-24, 1999 [5] Microchip Technology Incorporated. "Power Considerations," PICmicro Microcontrollers, (2002 September), [2005], Available at: http://wwl.microchip.com/downloads/en/DeviceDoc/power.pdf [6] GlobalStar. "Product Brochure" GSP 1620 - Duplex Modem, (2004 August), [2005], Available at: http://common.globalstar.com/docs/gspl620_brochure_en_CA.pdf [7] GlobalStar. "Product Brochure" Simplex Transmitter Unity, (2004 August), [2006], Available at: http ://comm on. globalstar.com/docs/simplex.pdf [8] Marine Mammal Research Unit, "Development of a Sub-dermal RF Identification Tag for the Endangered Steller Sea Lion (Eumetopias jubatus)" Project Progress Report, Electrical and Computer Engineering, Chemical and Biological Engineering, University of British Columbia (2004) [9] Rittal Communication Systems. "CS Outdoor Solutions", (2003 September). [2006], Available at: http://www.rittal.com/ [10] J.P. Holman Heat Transfer. 5 t h ed. McGraw-Hill, New York, 1981 [11] H. Piggott, Windpower Workshop, T. Kirby, Ed., Machynlleth , Wales: Centre for Alternative Technology, 2001. [12] P. Gippe, Wind Energy Basics A guide to Small and Micro Wind Systems, Vermont: Chelsea Green Publishing Company, 1999 [13] M . Patel, Wind and Solar Power Systems, Florida: C R C Press L L C , 1999 113 Bibliography 114 [14] P. Richard, "How Photovoltaics Are Tested & Rated," Home Power #23. June/July, pp. 20-23, 1991 [15] P. Horowitz and W. Hi l l , The Art of Electronics, New York: Cambridge University Press, 1989 [16] P. McChesney, "Solar Electric Power for Instruments at Remote Sites", Open File Report 00-128, Department of Interior U.S. Geological Survey, Vancouver W A : University of Washington Geophysics Program, 2000 [17] P. Richard and S.Coleman, "PV Module Angles," Home Power #36, August/September, pp. 14-16, 1993 [18] B. Chowdhury and B. Rahman, "Analysis of interrelationships between photovoltaic power and battery storage for electric utility load management." Power Systems, IEEE Transactions on Power Systems,Volume 3, Issue 3, pp: 900 - 907 Aug. 1988 [19] Go Solar Company , "To Calculate Solar Battery Bank Sizing" [2005], Available at: http://www.solarexpert.com/Pvbattery.html [20] Sierra Solar System, "Array Sizing Worksheet Form" [2005], Available at: http://www.sierrasolar.com/design/array [21] W. Young, "Emergency Power Systems with Photovoltaics" PV Disaster Training Workshop Manual, Presented at the Governor's Hurricane Conference, May 2000 [22] D. Linden and T. Reddy, Handbook Of Batteries. Toronto: McGraw-Hill, 2002 [23] Arizona Solar Center, "Handbook of Secondary Storage Batteries and Charge Regulators in Photovoltaic Systems," Final Report. (2003), [2005] Available at: http://www.azsolarcenter.corri/technology/technical/batteries.html [24] I. Buchman, "Discharging at high and low temperature", Cadex Electronics Inc, [2005] Available at: http://www.batteryuniversity.com/print-partone-15 .htm [25] Trojan Battery Company, "Battery Selection," [2005] Available at: http: / /www. troj an-battery. com/Tech- Support/F AQ/B atteryS election, aspx [26] J. Le, "Trojan T105 Battery Specification," T-105 Life Cycle, Trojan Battery Company, (2005) [27] M K Battery Company, "Downloads and Technical Reference," [2005] Available at http://www.mkbattery.com/techref_faq.php [28] J. Le, "Trojan T105 Temperature VS Capacity," T-105 Performance, Trojan Battery Company, (2005) Bibliography 115 [29] Maxwell Technologies, "Maxwell Technologies Ultracapacitors," Electrical Double Layer Bootstrap Capacitors, [2005] Available at: http ://www.maxwell. com/ultracapacitors/index .html [30] L. Palma, P Enjeti and J Howze, "An approach to improve battery runtime in mobile applications with supercapacitors," Power Electronics Specialist Conference, 2003. PESC '03. 2003 IEEE 34th Annual Volume 2, 15-19, pp: 918-923 June 2003 [31] R. Nelms and R. Spyker, "Classical equivalent circuit parameters for a double-layer capacitor. "Aerospace and Electronic Systems, IEEE Transactions on Volume 36, Issue 3, Part 1, pg: 829 - 836 July 2000 [32] Maxwell Technologies, "PC2500 Datasheet," Electrical Double Layer Bootstrap Capacitors, [2005] Available at: http://www.maxwell.com/pdf/uc/datasheets/PC2500.pdf [33] Maxwell Technologies, "AN-002 Cell Balancing," Application Note, [2005] Available at: http://www.maxwell.com/pdf/uc/app_notes/AN-002 Cell_balancing.pdf [34] D. Linzen,, S. Buller, E Karden and R. De Doncker, "Analysis and evaluation of charge-balancing circuits on performance, reliability, and lifetime of supercapacitor systems" Industry Applications Conference, 2003. 38th IAS Annual Meeting. Conference Record of the Volume 3, 12-16 pg: 1589 - 1595 Oct. 2003 [35] United States Council for Automotive Research, "FreedomCAR Ultracapacitor Test Manual" US Department of Energy, September 2004 [36] A . Rufer and P. Barrade, " A supercapacitor-based energy storage system for elevators with soft commutated interface" Industry Applications Conference, 2001. Thirty-Sixth IAS Annual Meeting. Conference Record of the 2001 IEEE Volume 2, pp:1413-1418, Sept/Oct. 2001 [37] Y . Panarin, "Power Bipolar Junction Transistors and Power Darlingtons," Dublin: Dublin Institute of Technology, Electronic and Communication Engineering, K235 (2005 June 30) [2005] Available at: http://www.electronics.dit.ie/staff/ypanarin/K235-notes.htm [38] R. Erickson and D. Maksimovic, Fundamentals of Power Electronics, Second Ed, Norwell, Massachusetts: Kluwer Academic Publishers, 2001 [39] N . Mohan, T. Undeland and W. Robbins, Power Electronics Converters, Applications and Design, Toronto: John Wiley & Sons Inc., 1995 [40] L. Dixon "Magnetics Design for Switching Power Supplies" Texas Instruments, Course Material, 2001 March 15, Dallas Texas Bibliography 116 [41] G. Slemon and A. Straughen, Electrical Machines, Toronto: Addison Wesley Publishing Company, 1982 [42] Magnetics Inc, "Ferrite Core Design Manual 2004," Materials, pp 3.1 3.11 (2004), [2005] Available at: http://www.mag-inc.com/ferrites/ferrites.asp [43] T. Eichson, "Estimating Inductor Losses Easily in Power Supply Designs," Power Electronics Technology, April, pp 14-24, 2005 [44] J. Takesuye and S. Deuty, "Introduction to Insulated Gate Bipolar Transistors" Motorola Semiconductor Application Note A N 1541, Motorola [45] Philips Semiconductors, "Introduction to Power Electronics," Philips Semiconductors Applications ,Hazel Grove, 1994 [46] D. Tulbure, "Introduction to Power Mosfets" Micronote Series 901, Microsemi Santa Ana [47] International Rectifier, "LRF3707z Datasheet," Hexfet Power MOSFET, International Rectifier, E l Segundo, California: (2003), [2005] Available at: http://www.irf.com/product-info/datasheets/data/irG707z.pdf [48] Maxim, Dallas Semiconductors "Power Supply Engineer's Guide to Calculate Dissipation for MOSFETs in High-Power Supplies" Electronic Design, Penton Media, October 2002. [49] Microchip Technology Incorporated. "Driving Power MOSFETs in High-Current, Switch Mode Regulators," AN786. (2002), [2005], Available at: wwl.microchip.com/downloads/en/AppNotes/00786a.pdf [50] International Rectifier, "IRF3305 Datasheet," Hexfet Power MOSFET, International Rectifier, El Segundo, California: [2005] Available at: http://www.irf.com/product-info/datasheets/data/irfl010z.pdf [51] W. Xiao and W. Dunford, " A Modified Adaptive Hi l l Climbing MPPT Method for Photovoltaic Power Systems." Power Electronics Specialists Conference, 2004. PESC 04. 2004 IEEE 35th Annual Volume 3, 20-25 June 2004, pp.1957 - 1963 Vol.3 [52] United Chem-Con, " L X Z Datasheet," Miniature Aluminum Electrolyte Capacitor, Illinois: [2005] Available at: http://www.chemi-con.com/files/LXZ0703.pdf Appendix A Base Stations locations Table 22 Base station locations Location Longitude Latitude Carmanah Point -124.78157 48.61624 Mara Rock -125.48500 48.88139 Long Beach -125.72028 49.04250 Raphael Point -126.23527 49.31389 Barrier Islands -127.53250 50.02056 O'Leary Islands -127.65472 50.10389 Solander Island -127.94472 50.11972 Cape Scott -128.42195 50.78611 Maggot Island -128.78778 50.80305 Beresford Island -128.77167 50.79778 Sartine Island -128.90500 50.82195 Triangle Island -129.08444 50.87444 Ashby Point -127.92333 50.94806 Virgin Rocks -128.20667 51.28917 Pearl Rocks -128.00305 51.37194 Gosling Rocks -128.46638 51.87056 Mclnnes Island -128.72362 52.27083 Steele Rock -129.37361 52.47306 Ashdown Island -129.21889 53.05972 North Danger Rocks -130.34723 53.25945 Bonilla Island -130.62138 53.47694 Reef Island -131.48500 52.87500 Cape St. James -130.99194 51.91861 South Tasu Head -132.08055 52.70472 Moresby Islands -132.36528 52.97667 Cone Head -132.72417 53.38250 Joseph Rocks -133.13333 53.82306 Langara Island -133.02444 54.26750 Anthony Island -131.24445 52.10556 Warrior Rocks -130.85333 54.07444 R1 560k 1% 00 D B 9 T O P C U A R T rille Digital Controlling Circuitry Size Document Number }ev Dale: Sunday. January 22, 2006 $heet ol 2 53 a X! W 3 P Appendix B Schematic 119 =q H «q i q r - K C Q L U + ' + Di m g S 8 ™ g h- o; CO UJ + « + Q; U ) Q L L U ^ U 1 > Q go°<"« g [ f l > f f l U J L L ± 0 [ I l m i — > C4 O OJ 0> >> F o 3 o •fcj £ U co wer Numb v, Jan o CM 0) "O E CU o H Q N 1 • Appendix C Software Code //18F2510 MICROCONTROLLER #include<18F2510.h> llllllllllllllllllllllllllllllllllllllllllllllllllll FUSES llllllllllllllimiHIIIIIIIIIIIIIHIIIIIIIIIIIHIIIIII #FUSES NOWDT //No Watch Dog Timer #FUSES X T //Crystal osc <= 4mhz #FUSES N O F C M E N //Fail-safe clock monitor disabled #FUSES NOBROWNOUT //No brownout reset #FUSES BORV20 //Brownout reset at 2.0V #FUSES PUT //Power Up Timer #FUSES N O S T V R E N //Stack full/underflow will not cause reset #FUSES N O D E B U G //No Debug mode for ICD #FUSES L V P //Low Voltage Programming on B3(PIC16) or B5(PIC18) #FUSES NOWRT //Program memory not write protected #FUSES NOIESO //Internal External Switch Over mode disabled #FUSES NOEBTR //Memory not protected from table reads #FUSES NOEBTRB //Boot block not protected from table reads #FUSES M C L R //Master Clear pin #FUSES NOPROTECT //Code not protected from reading #FUSES NOCPB //No Boot Block code protection #FUSES NOWRTB //Boot block not write protected #FUSES NOWRTC //configuration not registers write protected #FUSES N O F C M E N //Fail-safe clock monitor disabled /*PORTB pins are configured as analog input channels on RESET*/ #FUSES P B A D E N ///////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// ///////////////////////////////////////////////////COmiGURATlOW/////////////////////////////////////////////// 11 AID #device ADC=10 //Analog to Digital lObit Resolution //clock 4MHz #use delay(clock=4000000) //RS232 9.6kbs, No Parity, 8 bits #use rs232(baud=9600,parity=N,xmit=PrN_C6,rcv=PIN_C7,bits=8) //I2C #use i2c(MASTER, SDA=PIN_C4, SCL=PIN_C3) //USE FAST PORTA and PORTB (manual define) #use fast_io(B) #use fast_io(a) 120 Appendix C Software Code 121 //FILES TO INCLUDE #include <input.c> #include <stdlib.h> / /DEBUGGING: /* Two options: - MPPTRECOPvD for testing and Debugging - N O R M A L for normal operation*/ #defme N O R M A L / / M E M O R Y ADDRESS M A P S / /COMPARATOR CONTROL REGISTER INVERSION BIT //(controls the Capacitor Power Flow) #bit CM1CON0 = 0xFB4.5 //inversion bit for comparator 2 #byte C M C O N = 0xFB4 //comparator registry //PORTB B Y T E #byte PORTB =0xF81 iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiniiiiiiiiiiiiiiiiimmm / /ADC CHANNELS: #define B A T T _ C H A N N E L 11 #defme P V V O L T C H A N N E L 9 #define P V _ C U R R E N T _ C H A N N E L 1 #defme C A P V O L T C H A N N E L 8 #defmeVREF CALIB C H A N N E L 3 //Battery's Voltage A D C channel / /PVs voltage A D C channel / /PVs current A D C Channel //CAP's A D C channel / /VREF C A L I B R A T I O N C H A N N E L //ON | OFF #defme OFF 0 #defineON 1 / /CAPACITOR C H A R G E R #define CAP_SENSE PIN_C0 //Capacitor sense side enable pin #define BATT_SENSE PLN_C5 //Battery sense side enable pin #define C A P D R I V E R P L N B 7 //Capacitor Driver enable #defme C H A R G E 1 //Capacitor charge direction (synch, discount) #define DISCHARGE 0 //Capacitor discharge direction #defme COMP_ON AO_A3_A1_A2_OUT_ON_A4_A5 //comparator enable pins #defme COMP_OFF NC_NC_NC_NC //comparator disable #defme COMP2 PIN_A5 //comparator 2 OUTPUT PIN //INTERRUPT PINS #defme CAP_IPIN 5 //EPIN is #5 on PORTB, disables synchronous mode. #define CAP_VPIN 6 //VPIN is #6 on PORTB, monitors cap voltage #defme C A P V O L T A G E O F F S E T 1.5 //Cap voltage must be 1.5 above current Appendix C Software Code 122 //Delay Time #define START_POINT_DELAY 250 //250ms for charging/discharging input capacitor #define M P P T J D E L A Y 200 //200us for system to stabilize after a duty step change #define D E L A Y _ T I M E 1600 //Delay when displaying information onto the monitor //MPPT CONTROL DEFINITIONS #defme MPPT_HUMP 0.78 //MPPT approximate: 78% of open voltage #defme I N T E R V A L _ V A L U E 1 //MPPT Scanning interval /*Stabilizes oscillation when MPPT is between the tracking interval*/ #define MPPT_FOUND 6 //power point oscillation amount #define LIMIT 80 // how many times to search when tracking MPPT / /TEMPERATURE CONTROL DEFINITIONS #defme THERMO 0b 10010000 //Thermo Address #define A V G _ M A X 8 //the M A X amount of measuring times to average the temp value / /DUTY CONTROL / /PWM 100% full on DUTY=(Value/80)* 100 Resolution is 1.25% #defme F U L L _ O N 80 /^Conversion Duty factor to change from D=100% to the up P W M , refer to notebook*/ #define Duty_Factor 80 / /COMPARATOR INVERSIONS #defme C2INV 1 //Comparator 2 Inversion #define C2NON 0 //Comparator 2 Noninversion #define C1LNV 0x8 //Comparator 1 Inversion #define C1NON 0 //Comparator 1 Noninversion //DIGITAL to A N A L O G /* I2C D A C sensor (AD5301) Address: ObOOOllOOO*/ #DEFINE I D A C 1 ObOOO 11000 /* I2C D A C sensor (AD5301) Address: ObOOOllOOO*/ #DEFINE V D A C 2 ObOOO 11010 #defme RESOLUTION 256 / /DAC RESOLUTION (8bits) #DEFINE M A X _ O N 4 //fully turns on the D A C to highest voltage / /TEMPERATURE PROBE /*I2C temperature sensor (MCP9802A0) Address: (0bl0010000)*/ #DEFINE T E M P P R O B E 0b 10010000 / /ADC CONVERSION FACTORS #define PV_FACTOR 6 //resistor ladder: (160+800)/160 Factor =6 #defme BATT_FACTOR 5 //resistor ladder:(160+640)/160 Factor =5 #defme CAP_FACTOR 4 //resistor ladder: (160+480)/160 Factor =4 Appendix C Software Code 123 #define C U R R E N T F A C T O R 1 //Current sensor factor /*ADC FACTOR, (1023/4.096)=249.7558 (refer to L O G BOOK)*/ #define A D C _ F A C T O R 249.7558 #define R E F V O L T A G E 4.096 //REF VOLTAGE=4.096V #defme A 2 D M A X V A L U E 1023 //A2D maximum value lllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllll^ iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiniiniimj^cTiomiiin void intialize_pic(); //initializes the pic void i2c_DAC_init(unsigned int dac); //initialize I2C D A C void i2c_temp_init(); //initialize the temperature sensor void clear_scr(); //clears the screen void calibrate_vref(); //calibrates A2D's Vref void read_temp(); //gets the system's temperature //gets the max voltage based on the system's temperature void get_bat_max_volt(); void hill_climb(); //hill climbing algorithms void read_battery_voltage(); //get the battery's terminal voltage void read_cap_voltage(); //get the cap terminal voltage void read_pv_voltage(); //get the PV voltage void read_pv_current(); //get the P V current //takes a A D C measurement and averages it 16 times unsigned long read_averaging(); void set_batt_duty(); //manual sets the PV duty cycle void DAC_Output(unsigned int dac, float value); //write to the I2C D A C void cap_menu(); //prints the cap charging menu options void print_menu(); //prints command menu to monitor /*sets the duty cycle, provides a delay for system to stabilize and gets the resulting power*/ float getjpvpower(unsigned long duty, unsigned int delay); void timerl_isr(); void cap_control(); ///////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// signed int temperature; //System Temperature unsigned long pv_duty; //PV duty unsigned long cap_duty; / /CAP D U T Y signed int interval; //interval step size value float max_volt; //maximum battery voltage from temperature look up table float batt_volt; //Battery voltage float pv_current; //PV current float cap_current; //cap current float capjvolt; //ultracap voltage float pvjvolt; //PV panel's voltage Appendix C Software Code 124 unsigned intl6 timer_count; //timer for MPPT int vref_calib_offset; //Vref offset value... how much off is V R E F from 4.906V unsigned int control_state; //portb state int cap_direction; //charging direction for capacitor //PWM2 //Software P W M driven: Values for D2A conversion static B O O L E A N pinState; //checks the state //lookup table: battery voltage unsigned int batvolt_table[]={162, 161, 160, 156, 154, 151, 149, 147, 145, 143, 141, 139, 138, 137, 136, 135, 134, 133}; //lookup table: temperature signed int temp_table[] = {-40, -35, -30, -25, -20, -15, -10, -5, 0, 5, 10, 15, 20, 25, 30, 35, 40, 45}; char command; //keyboard input command IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIH IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIM main() { unsigned long value; char string[5]; //variable for keyboard commands /^current temp variable, in case abort, don't want capacitor charge current to get value*/ float temp; delay_ms(100); //delay startup for 100ms so peripherals can start up. #ifdef N O R M A L clear_scr(); //clear screen #endif //print Microprocessor's name printf("\n\rWelcome to Majesus Bot, Microprocessor 18F2510\n\r"); #ifdef N O R M A L printfC'IN TESTING & PROTOTYPE M O D E W ) ; #endif initialize_pic(); //initialize pic print_menu(); //print menu to monitor //INTERRUPTS EXT_LNT_EDGE(0, H_TO_L); disable_interrupts(int_timerO); enable_interrupts(int_rda); enable_interrupts(int_rb); enable_interrupts(int_EXT); //trigger high to low //enable timerO interrupt //enable keyboard input //enable portb interrupts //enable battery current interrupt Appendix C Software Code 125 control_state=portb; enable_interrupts(int_comp); clear_interrupt(int_ext); enableinterrupts(global); cap_direction=l; command=OFF; //read portb before enabling interrupts //prevent battery current interrupt from false setting //FOR M P P T R E C O R D TESTING #ifdef MPPT_RECORD //display measurements to monitor printf("IN M P P T R E C O R D I N G MODE\n\r"); //set command to zero //infinite loop //wait for timer clock to interrupt //call hill_climb(); command=0; while(TRUE) { while(! command); hill_climb(); printf("%f ",batt_volt); printf("%f ",pv_volt); printf("%f ",pv_current); printf("%lu ",pv_duty); //simulate no MPPT, PV connected directly to battery set_pwml_duty(FULL_ON); set_pwm2_duty(FULL_ON); delay_ms(START_POINT_DELAY); delay_ms(START_POINT_DELAY); read_battery_voltage(); read_pv_voltage(); read jpv_current(); printf("%f ",batt_volt); printf("%f ",pv_volt); printf("%f ",pv_current); //shutdown and repeat in 5minutes set_pwm 1 _duty(OFF); set_pwm2_duty(OFF); command=0; } #endif //stop driver's enable //stop driver's enable //set command pack to zero, and wait for timer while(TRUE) { command=getc(); switch(command) { case'd': { //get command from keyboard //check command //set duty automatically Appendix C Software Code 126 printf("Type duty cycle then press enter: \n\r"); gets(string); //get string value temp=atof(string); //convert to integer printf("DUTY Entered: %2.2fA.\n\r", temp); pv_duty=(unsigned long)(temp*DUTY_FACTOR); //cast to pv_duty se t_bat t_duty(); //set duty cycle break; } case 'c': //charge cap { do { printf("CAP C H A R G E MODE. Type current value (between 0.4 and 3.5A) then press enter: \n\r"); gets(string); temp=atof(string); printf("Current Entered: %2.2fA. Press '1' to accept, else press any key to reset.\n\n\r", temp); command=getc(); } while ((command != '1') | (temp > 3.5) | (temp < .4)); cap_cuiTent=temp; printf("Cap current is now %2.2fA.\n\r", cap_current); printf("Direction? 1 = CHARGE, 0 = DISCHARGE \n\r" ); gets(string); //get string? cap_direction=atoi(string); //convert to integer printf("Direction is: %u\n\r", cap_direction); cap_control(); break; } { DAC_Output(I_DAC 1, OFF); //output_low(CAP_DRIVER); break; } case 'h': //call hill climbing routine { printf("STEP SIZE value? (AS INTEGER)"); //integer value? gets(string); //get string? interval=atoi(string); //convert to integer printf("STEP SIZE is: %u\n\r", interval); hill_climb(); //call function break; Appendix C Software Code 127 } } } } lllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllll ////////////////////////////////////////INTERRUPT SERVICE ROUTINES/////////////////////////////////// lllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllll/lllllllllllllllllllllllllllllllllllllllllllllllllllllll I* This ISR is called every 65.536ms (4*65536/4Mhz). At every 5minutes, the power point is found*/ #int_timerO void clock_isr() { timer_count++; //increment timer every 65.546ms #ifdefNORMAL printf('TN TIMER_COUNT %lu", timer_count); #endif if(timer_count==2000) //2000=5min, time has expired { #ifdef N O R M A L printf("IN I S R 2 W ) ; #endif command='h'; //call hil lcl imbing routine out of ISR timer_count=0; //reset timer_count } / /BATTERY R E V E R S A L CURRENT SERVICE ROUTINE //disables synchronous mode #INT_EXT void Battery_reverse() { //set enable PIN to follow input PIN (synchronous disabled) set_p wm 1 _duty(p v_duty); #ifdef N O R M A L printf("\n\rln B A T T REVERSE interrupt!\t B A T T E R Y SYNCHRONOUS DRIVE D I S A B L E D W ) ; #endif control_state=portb; //read in portb clear_interrupt(int_ext); //clear external interrupt flag disable_interrupts(int_ext); //disable external interrupt } Appendix C Software Code 128 //CAP C O M P A R A T O R ISR #INT_COMP void comparator_int() { int c; //define integer c for clearing C M C O N c = C M C O N ; //clear the "change condition" by reading (or writing) the C M C O N register. printf("In comparator InterruptVn\r\n"); output_toggle(CAP_DRIVER); //toggle Cap output driver } //PORTB ISR #INT_RB void int_PORTB() { control_state=portb; //read in portb printf("lN PORTB interrupt \r\n"); printf("control_state:%x \n\r", control_state); / /CAP CURRENT R E V E R S A L if(bit_test(control_state, CAP_IPrN) != cap_direction) / / i f not true, then direction reversed. { output_HIGH(CAP_DRIVER); //Enable Capacitor Driver else if(!bitJest(control_state, CAP_VPIN)) { printf("In C A P _ V ! W ) ; } clear_interrupt(int_rb); //clear interrupt flag / / K E Y B O A R D INTERRUPT #int_rda void serial_isr() { float temp; //current temp variable, in case abort, don't want pv_current to have value. char string[5]; command=getc(); printf("In K E Y B O A R D I S R W ) ; switch(command) { case'd': { printf("Type duty cycle then press enter: \n\r"); Appendix C Software Code 129 gets(string); temp=ato f(string); printf("DUTY Entered: % 2 . 2 f A . W , temp); pv_duty=(unsigned long)(temp*DUTY_F ACTOR); //cast to pv_duty break; } case '8': { do { printf("CAP C H A R G E MODE. Type current value (between 0.4 and 3.5A) then press enter: \n\r"); printf("TO '0' to shutoff Cap charge\n\r"); gets(string); temp=atof(string); i f (command ='0') { output_low(CAP_DRIVER); break; } printf("Current Entered: %2.2fA. Press '1' to accept, else press any key to reset.\n\n\r", temp); command=getc(); } while ((command != '1') | (temp > 3.5) | (temp < .4)); pv_current=temp; //give pv_current temp variable cap_control(); //call cap control break; } case 'h': //used for debugging MPPT { printf("STEP SIZE value? (AS INTEGER)"); gets(string); interval=atoi(string); printf("STEP SIZE is: %u\n\r", interval); break; } } } Appendix C Software Code 130 /// / / / / / / / / / / / / / / / / /^ //////////////////////////^^ //////////////////////////^^ //This function initializes interrupts, P W M , ports, comparators, D2A and Temp probe void intialize_pic() { port_b_pullups(FALSE); //disable por_b_pullups. Controlled by outputs //define portb as inputs Interrupt Pins: B7,B6, BO set_tris_b(0b01111111); /* (pg 101 in MICROPIC manual) A2D Pins: B4-B1 B5 cap volt charge. B6 Battery reversal B7 output cap driver */ output_low(CAP_DRIVER); //shutdown cap driver set_tris_a(0b 10001100); //define PORTA output and input pins (pg98 in MICROPIC M A N U A L ) //enable port b pullups output_c(0); //define all outport_C as 0 output_HIGH(BATT_sense); //disable Battery side CAP current sensor output_HIGH(CAP_sense); //disable capacitor side CAP current sensor //Initialize CAP charger hysteresis comparator setup_comparator(COMP_ON); //SETUP C O M P A R A T O R AS OFF until needed2 //Setup A D C //setup_adc_ports( AN0_TO_AN3); //configure AN0 to AN3 as Analog inputs and OR'ed it with VSS_VREF: (0-VrefH) //using the operator '|' setup_adc_ports(AN0_TO_ANl 1 | VSS_VREF); setup_adc(ADC_CLOCK_DIV_8); //AdC clock, refer to table 19-1 on datasheet //SETUP TIMERS Used to measure MPP every 5 minutes (will increment every luS) setup_timer_0(RTCC_INTERNAL|RTCC_DIV_l); //disable timerl and timer3 interrupts. Not needed setup_timer_l(Tl D I S A B L E D ) ; setup Jimer_3(T3_DISABLED); /^Hardware P W M (PWM1) TIMER for switching clock of MPPT)*/ Setup_timer_2(T2_DIV_BY_l, 19, 1); //(TMR2 Prescale Value, PR2, # of resets before an interrupt): Appendix C Software Code 131 / /DISABLE Serial Port Interface (SPI) setupspi(FALSE); #ifdef N O R M A L //setup i2c for temperature probe i2c_temp_init(); #endif //PWM2 //setup PWM2 for MPPT, hill climbing (CCP1): setup_ccp2(CCP_PWM); / /CCP_PWM means Set CCP to P W M mode: //(PWM frequency=50kHz refer to page 138 for values of TMR2 and PR2) //PWM1 //setup PWM2 for cap soft start charging: setup_ccpl(CCP_PWM); / /CCP_PWM means Set CCP to P W M mode: //(PWM frequency same as PWM2. 50kHz cause it uses the same timer2) //Initialize I2C DACs i2c_DAC_init(I_DACl); //output cap's current reference i2c_DAC_init(V_DAC2); //output cap's voltage reference //calibrate A2D's V R E F offset. calibrate_vref(); //set PWMs to OFF pv_duty=OFF; set_p wm 1 _duty(OFF); set_pwm2_duty(OFF); get_b at_max_vo lt(); read_battery_voltage(); read_pv_voltage(); read_c ap_vo ltage(); pv_current=0; timer_count=0; DAC_Output(I_DACl, OFF); printf("\n\rW); //get the battery's max charging temperature //read battery's terminal voltage //read P V s terminal voltage //read cap's terminal voltage //set pv_current variable to zero //set timer to zero / /DAC Current is set to zero //INTERRUPTS EXT_INT_EDGE(0, H_TO_L);//trigger high to low for battery interrupt #ifdef M P P T R E C O R D / / M P P T R E C O R D , records the MPPT every 5 minutes enable_interrupts(int_timerO); //enable timerO interrupt #endif Appendix C Software Code 132 #ifdef N O R M A L //normal mode disable_interrupts(int_timerO);//disable interrupt, #endif // enable_interrupts(int_rda); //enable keyboard input enable_interrupts(int_rb); //enable portb interrupts enable_interrupts(int__EXT); //enable battery current interrupt control_state=portb; //read portb before enabling interrupts disable_interrupts(int_comp); clear_interrupt(int_ext); //prevent battery current interrupt from false setting enable_interrupts(global); return; } //This function prints the commands to the monitor void prin t_menu() { printf("TYPE'd' to setup battery duty cycle\n\r"); printf("TYPE 'h' to enable HillclimbVnV'); printf("TYPE 'c' to enable charge capacitor\n\r"); printf("TYPE 'o' to shut down capacitor charger\n\r"); } //This function controls the capacitor charging dc/dc converter void cap_control() { float cap_v; //cap_v=pv_current+CAP_CHARGE_OFFSET; if (cap_direction==CHARGE) { read_cap_voltage(); CM1CONO=C2NON; cap_v=cap_current+CAP_VOLTAGE_OFFSET; if (cap_volt < cap_v) { printf("CAP V O L T A G E TOO LOW. SOFT START E N A B L E D . CHARGING V O L T A G E TO: %2.2f\n\r", cap_v); cap_v=cap_v/CAP_F ACTOR; printf("WRITLNG TO V _ D A C 2 voltage:%2.2f\n\r", cap_v); //SET V D A C comparator to cap_v threshold and wait till cap V hit reference to drive interrupt DAC_Output(V_DAC2, capv); DAC_Output(I_DACl, capcurrent); output_high(BATT_sense); //disable Battery side current sensor output_low(CAP_sense); //enable capacitor side current sensor Appendix C Software Code } Else { DAC_Output(I_DACl, cap_current); outputlow(BATTsense); //enable Battery side current sensor outputHIGH(CAPsense); //disable capacitor side current sensor } } else i f (cap_direction==DISCHARGE) { CM1CONO=C2INV; if (cap_volt < cap_v) { printf("CAP V O L T A G E TOO LOW. Can't discharge: %2.2f\n\r", cap_v) output J o w(CAPJDRTVER); //disable Capacitor Driver return; } printf("CAP discharging\n\r", cap_v); output_low(BATT_sense); //disable Battery side current sensor output_HIGH(CAP_sense); //enable capacitor side current sensor } printf("Enabling CAP J D R I V E R W ) ; output HIGH(CAP_DRiVER); //Enable Capacitor Driver } //This function gets the battery's Voltage charge limit at a specific temperature void get_bat_max_volt() { int counter=0; //check to see i f temperature exceeds the max temp if (temperature >= 45) { Printf("Upper temperature limit!\n\r"); max_volt=133; } else { //while loop, repeats until becomes false, while (temp table[counter] < temperature) { Appendix C Software Code 134 / /DEBUG //prints temperature and temp_table look up table /*printf("Temp:table %d temperature %d Counter %d\n\r", temp_table[counter], temperature, counter);*/ counter++; } //convert to Maxvolt (cast batvolt_table) max_volt=((float)batvolt_table[counter]/l 0+0.01); //the 0.01 is to compensate for rounding error } #ifdef N O R M A L printf("Max charging voltage %2.1f\n\r",max_volt); #endif return; } /*This function sets the duty cycle, provides a delay for the system to stabilizes and then calculates the power*/ float get_pvpower(unsigned long duty, unsigned int delay) { set_p wm2_duty(duty); set_pwml_duty(FULL_ON); //reset E N A B L E ON delay_us(delay); //delay for input capacitor to stabilize read__battery_voltage(); read_pv_voltage(); read_p v_current(); #ifdef N O R M A L PRINTF("SETTING D U T Y TO:%lu\nV, duty); #endif return (pv_current* batt_volt); } //changes batt duty manual, for debugging purposes void set_batt_duty() { PRINTF("LN B A T T D U T Y change function..AnV); set_pwml_duty(OFF); //stop driver's enable clear_interrupt(INT_EXT); //clear E X T E R N A L interrupt enable_interrupts(LNT_EXT); //renable E X T E R N A L interrupt set_pwm2_duty(pv_duty); //give MOSFET driver it's starting point duty cycle printf("LNPUT D U T Y (BD) is %lu", pv_duty); if(pv_duty<30) Appendix C Software Code 135 set_pwml_duty(pv_duty); //will be in discontinuous else set_pwml_duty(FULL_ON); //activate MOSFET Driver delay_ms(200); read_battery_voltage(); read_pv_voltage(); read_pv_current(); printf("POWER IS: %f2.2\mr", pv_current* batt_volt); } //This function finds the PV's Maximum Power Point void hil l_climb() { float power, power_old; //safety value, minimizes the amount of M M P T search int search=0; //safety value, i f MPPT found, prevents oscillation int found=0; //shut off pv_duty and measure PV's open voltage set_pwml_duty(OFF); //stop driver's enable set_pwm2_duty(OFF); //stop driver's enable delay_ms(START_POINT_DELAY); clear_interrupt(INT_EXT); //clear E X T E R N A L interrupt enable_interrupts(INT_EXT); //renable E X T E R N A L interrupt //determine batt voltage read_battery_voltage(); //delay 400ms for Panel to stabilize read_pv_voltage(); if ((batt_volt+l) >= pv_volt) { printf("PV P A N E L V O L T A G E TO LOW, NO POWER A V A I L A B L E W ) ; return; } / /MMPT around 78% (MPPT_HUMP) of open voltage. Determine D U T Y pv_duty=(unsigned long int)(batt_volt*DUTY_FACTOR)/(pv_volt*MPPT_HUMP); set_pwm2_duty(pv_duty); //give MOSFET driver it's starting point duty cycle //enable DRIVER set_p wm 1 _duty(FULL_ON); delay_ms(START_POINT_DELAY); //wait for system to stabilize Appendix C Software Code 136 power=get_pvpower(pv_duty, 0); #ifdef N O R M A L printf("POWER AT H U M P is %f\n\r\nV, power); #endif while(search<LIMIT) //safety limit, prevent an infinite search loop { / / i f interval is positive, VLN is decreasing. pv_duty+=interval; //save current power as old power_old=power; //get power new value power=get_pvpower(pv_duty, MPPT_DELAY) ; #ifdefNORMAL printf("POWER A T search interval (%d) is %f\n\r\n\r", search, power); delay_ms(DELAY_TIME); #endif if (power<power_old) { found++; //increment Found interval=(-l interval); //search the other direction #ifdef N O R M A L printf("Going wrong way... (IN the IF state) FOUND Number: (%d). Power is % f \n\r\n\r", found, power); delay_ms(DEL AY_TIME); #endif / / i f FOUND==MMPT_FOUND, then System is one after the M M P T (this code prevents constant oscillation) if (found==MPPT_FOUND) { pv_duty+=interval; //reset pv_duty to old value; get_pvpower(pv_duty, M P P T _ D E L A Y ) ; //set power back to older value since it is greater. #ifdef N O R M A L printf("FOUND=MPPT_FOUND... power is %f\n\r\n\r", power); #endif return; } } Appendix C Software Code 137 search++; //increment search } } / /AVERAGING FUNCTION. Take 16 measurements and A V E R A G E S them, unsigned long ADC_averaging() { int count=0; //initialize count as zero unsigned long measurement=0; //initialize measurement as zero while(count < 16) { measurement += read_adc(); //read from A D C channel and add it to measurement count++; //increment historycount } measurement = measurement » 4; //divide by 16 / /DEBUG //PRINTS the M E A S U R E M E N T V A L U E //printf("Measurement is: %ld\n\r ", measurement); return measurement; } /*Due to noise, determines how much incorrect the A2D measures VREF. Calibrates the A2D by this offset*/ void calibrate_vref() { float temp; set_adc_channel(VREF_CALIB_CHANNEL); //select A D C mux to V R E F C A L I B C H A N N E L #ifdef N O R M A L printf("Calibrating A2D VREF..At"); #endif temp=ADC_averaging(); //measure V R E F vref_calib_offset=(unsigned int)(A2D_MAX_VALUE-temp); //cast the measurement #ifdef N O R M A L printf("VREF M E A S U R E D AS: %2.2f\nV, temp); printf("VREF Calibrated at an offset of: %d\nV", vref_calib_offset); #endif } Appendix C Software Code 138 //Reads the Battery's Voltage void read_battery_voltage() { set_adc_channel(BATT_CHANNEL); //select A D C mux to B A T T channel bat t_volt=((float)ADC_averaging() + vref_calib_offset); //get battery voltage, cast it and add offset / /ADC conversion and multiply the B A T T F ACTOR.. . refer to lab book //add + 0.001 for floating rounding. //printf("Battery Voltage before conversion: %2.2f\n\r", batt_volt); batt_volt=(batt_volt/ADC_FACTOR)*BATT_FACTOR+0.001; #ifdef N O R M A L printf("\t\t\t\tBattery Voltage: %2.2f\nV", batt_volt); #endif } //Reads the P V s Voltage void read_pv_voltage() { set_adc_channel(PV_VOLT_CHANNEL); //select A D C mux to P V channel pv_volt=((fioat)ADC_averaging() + vref_calib_offset); //get pv_voltage, cast it and add offset / /ADC conversion and multiply the P V F A C T O R . . . refer to lab book //add +0.001 for floating rounding. pv_volt=(pv_volt/ADC_FACTOR)*PV_FACTOR+0.001; #ifdef N O R M A L printf("\t\t\t\tPV Voltage: %2.2f\n\r", pv_volt); #endif } //Reads the Capacitor's Voltage void read_cap_voltage() { set_adc_chamrel(CAP_VOLT_CHANNEL); //select A D C mux to CAP channel /*get Capacitor voltage, cast it and add offset*/ cap_volt=((float)ADC_averaging() + vref_calib_offset); / /ADC conversion and multiply the C A P F ACTOR... refer to lab book //add +0.001 for floating rounding. cap_volt=(cap_volt/ADC_FACTOR)*CAP_FACTOR+0.001; #ifdef N O R M A L Appendix C Software Code 139 printf("\t\t\t\tCap_voltage: %2.2f\n\r", cap_volt); #endif } //Reads the P V s Current void read_pv_current() { set_adc_channel(PV_CURRENT_CHANNEL); //get pv_current, cast it and add offset pv_current=((float)ADC_averaging() + vref_calib_offset); / /ADC conversion and multiply the CURRENTJF ACTOR.. . refer to lab book //add +0.001 for floating rounding. pv_current=(pv_current/AI)C_FACTOR)*ClJRRENT_FACTOR+0.001; #ifdef N O R M A L printf("\t\t\t\tPV_current: %2.2f\n\r", pv_current); #endif } //Clears the screen void clear_scr() { printf("\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\r"); } //Initialize I2C 8b'it D A C (AD5301), and set its output to 0V. void i2c_DAC_init(unsigned int dac) { i2c_start(); // i2c D A C sensor Address (ObOOO 11000),: write to AD5301 if(i2c_write(dac)==l) { #ifdef N O R M A L printf("D2A sensor error check D A C : %ux. CONFIGURATION ABORTED, \n\r\n", dac); #endif i2c_stop(); return; } #ifdef N O R M A L printf("Configuring I2C D A C : %ux... Setting to 0V ou tpu tW, dac); Appendix C Software Code 140 #endif /*DAC INPUT 16bit SHIFT REGISTER FORMAT: [x x PD1 PDO D7 D6 D5 D4 D3 D2 DI DO x x x x] (x are don't cares)*/ i2c_write(0b00000000); // Set power mode to: Normal Operation and write 0V to D A C i2c_write(0b00000000); //change configuration to no one-shot, no shutdown and 9bit resolution i2c_stop(); #ifdef N O R M A L printf("I2C D A C : %ux configured.\n\n\r", dac); #endif delay_ms(l); //delay 1ms for D A C to setup. } //Write to I2C 8bit (AD5301). void DAC_Output(unsigned int dac, float value) { unsigned int out_high, out_low; out_high=(unsignedint)((value*RESOLUTION)/REF_VOLTAGE); //D=2A8*x/Vref / /DAC INPUT 16bit SHIFT REGISTER FORMAT: [x x PD1 PDO D7 D6 D5 D4 D3 D2 DI DO x x x x] (x are don't cares) out_low=out_high; out_high=out_high»4; //right shift upper data four times /*include the first nibble (set power on) Last nibble is the M S B N I B B L E data.*/ out_high=out_high & ObOOOOl 111; /*left shift lower data four times. LSB Nibble of this byte is don't cares.*/ out_low=out_low « 4; i2c_start(); // i2c D A C sensor AddressOOOl 1000,: write to AD5301 if(i2c_write(dac)==l) { #ifdef N O R M A L printf("D2A sensor error Can't write to D A C . Check chip number: %ux.\n\r",dac); #endif i2c_stop(); return; } i2c_write(out_high); // OUTPUT high data Appendix C Software Code 141 i2c_write(out_low); //output low i2c_stop(); #ifdef N O R M A L printf("Voltage written to DAC# %ux as: %2.2fdac, value); #endif } //Setup the temperature probe void i2c_temp_init() { i2c_start(); // i2c temperature sensor Address (Ob 10010000), configure: write to MCP9802A0 if(i2c_write(TEMP_PROBE)==l) { #ifdef N O R M A L printf("Temp sensor error... Temp Probe not found.\n\r"); #endif i2c_stop(); return; } #ifdef N O R M A L printf("Temp sensor configuring...\n\r"); #endif i2c_write(0b01); // Point to control registry i2c_write(0b00000000); //change configuration to no one-shot, no shutdown and 9bit resolution i2c_stop(); //change temp pointer to Temp register i2c_start(); i2c_write(0bl0010000); i2c_write(0b00); // Point to Temperature register i2c_stop(); #ifdef N O R M A L printf("Temp sensor configured\n\n\r"); #endif delay_ms(500); //delay 0.5 seconds for temperature probe to setup. read_temp(); //read temperature Appendix C Software Code 142 } / /READ the T E M P E R A T U R E from the probe void read_temp() { //INITALIZE V A R I A B L E S int 16 avg_temp=0; //final temperature value (avg 8 times) int history_count=0; //tracks amount of times averaged #ifdef N O R M A L printf("Taking temp measurement \n\r "); #endif while (history_count<AVG_MAX) I i2c_start(); // i2c temperature sensor Address, read from MCP9802A0 if(i2c_write(TEMP_PROBE)==l) { #ifdef N O R M A L printf("TEMP SENSOR ERROR... Temp Probe Can't read measurement.W); #endif i2c_stop(); return; } avg_temp=(i2c_read() + avg_temp); / /DEBUG //Print avg_temp //printf("avg_temp: %ld Celsius\n\n\r", avg_temp); i2c_read(); //remove 9th bit i2c_stop(); history_count++; delay_ms(40); //convert to temperature: avg_temp/(8) and convert to signed integer (8bit) temperature=(int)(avg_temp»3); } #ifdef N O R M A L printf("SYSTEM T E M P E R A T U R E IS: %d Celsius\n\r", temperature); #endif } 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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


Related Items