Development of a DC-DC Converter for an Electric Motorcycle

Transcription

1 ECE 4600 Group Design Project Development of a DC-DC Converter for an Electric Motorcycle by Group 08 Chris Beaudoin Tyler Hampton Melanie Dyck Christian Jegues Final report submitted in partial satisfaction of the requirements for the degree of Bachelor of Science in Electrical and Computer Engineering in the Faculty of Engineering of the University of Manitoba Academic Supervisor Dr. Shaahin Filizadeh, Ph.D, P.Eng. Department of Electrical and Computer Engineering University of Manitoba Date of Submission March 10, 2014 Copyright 2014 Chris Beaudoin, Melanie Dyck, Tyler Hampton, Christian Jegues

2

3 Abstract Electric vehicles are becoming more popular in the automotive industry due to increasing interest in renewable resources. They are more efficient than conventional fossil fuel based vehicles, utilizing a higher percentage of available energy to power the wheels. The purpose of this project was to design and implement a controlled dc-dc converter for an electric motorcycle and package all of the hardware, resulting in a working motorcycle. A working motorcycle is defined as one for which the motor speed is controllable via the throttle input. A converter topology was selected and verified through simulation. Once verified, construction and testing of the prototype began. The resulting converter is capable of powering the motor in a controlled manner. More specifically, the converter is able to successfully control the speed of the motor and limit the current by controlling the voltage applied to the terminals of the motor. - ii -

4 Contributions Assess converter topologies Assess control schemes Research high power switching alternatives Investigate gate drive circuitry Investigate regenerative braking Model the motor Verify motorcycle model Model converter Simulate complete system Interpret simulation results Simulate regenerative braking Ordering parts Build converter Build gate drive circuitry Build controller Test prototype against simulation Assembly of low power circuitry Package prototype Mount prototype to motorcycle frame Chris Beaudoin Melanie Dyck Tyler Hampton Christian Jegues Legend: Lead task Contributed - iii -

5 Acknowledgements We would like to thank the following people and entities without whom this project would never have been completed: Dr. Shaahin Filizadeh, Mr. Erwin Dirks, and Mr. Daniel Card for their advice and expertise throughout the project. Their insight proved to be invaluable time and time again. The Electrical and Computer Engineering Department for the lab space, motorcycle, equipment, and funding. The Manitoba HVDC Research Centre for donating licenses for the PSCAD/EMTDC transient simulator. Dr. Behzad Kordi and Mr. Allan McKay for a second chance. The Electrical and Computer Engineering Tech Shop for their generous donation of components and time. Mr. Dan Beaudoin and Mr. Todd Gittins for their tools, skill and effort towards the construction of the aluminum heat sink. All of our friends and family for their encouragement and support throughout this project and our degree. - iv -

6 TABLE OF CONTENTS Table of Contents Abstract ii Contributions iii Acknowledgements iv Table of Contents v List of Figures viii List of Tables x Nomenclature xi 1 Introduction Motivation General Overview Problem Definition Outline Modeling and Simulation Power Electronic Converter DC Motor Overview of Permanent Magnet DC Motors Modeling of the PMDC Motor Motorcycle Control System Design Controller Type Control Scheme Simulation Results Hardware Design Switch Type Selection v -

7 TABLE OF CONTENTS 3.2 Gate Drive Circuitry Introduction to Gate Drive Circuits Low Side Switching Configuration High Side Switching Configuration Bootstrap Circuitry Power Buffer Circuitry Complete Gate Drive Circuit Additional Components Bootstrap Circuit Buffer MOSFETs DC-DC Converters Optoisolator Decoupling Capacitors Hardware Implementation Power and Gate Drive Circuitry Breadboard Wire Wrap Board Soldering Board Control System Hardware Hall Effect Sensor Final Prototype Enclosure Performance Testing Conclusions Conclusions Contributions Future Work References 53 Appendix A Budget 55 Appendix B Hardware Components 57 - vi -

8 TABLE OF CONTENTS Appendix C Curriculum Vitae 60 - vii -

9 LIST OF FIGURES List of Figures 1.1 Electric motorcycle System overview Torque-speed plane Permanent magnet dc motor Permanent magnet dc motor model Main routine flow chart Interrupt routine flow chart Motor current ripple in steady state Top speed test - speed Top speed test - current Braking test - speed Braking test - current Transition to regenerative braking Motor current ripple during regenerative braking MOSFET power module circuitry Low side switching configuration High side switching configuration Bootstrap circuit Power buffer circuit Gate drive block diagram Placement of the pull-up resistor Inverter circuit for the optoisolator signal Power buffer schematic issue Logic signal and buffer output viii -

10 LIST OF FIGURES 4.3 Buffer output extreme ringing Hall effect sensor noise Motor current ripple Motor current limit test Motor current time limit test Duty cycle vs motor current B.1 Cerebot Mx7ck microcontroller B.2 MOSFET power module B.3 Soldering board B.4 Aluminum heat sink with mounted components ix -

11 LIST OF TABLES List of Tables 1.1 Specifications Parameters of the PMDC motor Parameters of the motorcycle model MOSFET power module ratings P-channel MOSFET ratings N-channel MOSFET ratings Hall effect sensor data A.1 Budget x -

12 LIST OF TABLES Nomenclature Symbol EMF PMDC PI ADC IGBT MOSFET IC PWM B F I L τ e K E a ω m V a R a i a L a τ l Description Electromotive force. Permanent magnet dc. Proportional integral. Analog to digital converter. Insulated-gate bipolar transistor. Metal-oxide-semiconductor field-effect transistor. Integrated circuit. Pulse width modulation. Magnetic field strength. Force. Current. Length. Total torque generated by dc motor. DC motor electromotive force (EMF) constant. Internal voltage of dc motor (back EMF). Angular velocity of dc motor shaft. Terminal voltage of dc motor. DC motor winding resistance. DC motor armature current. DC motor winding inductance. Load torque. - xi -

13 LIST OF TABLES β DC motor viscous coefficient. J Moment of inertia. E Energy. P x m v ρ A C d Power corresponding to x. Mass. Linear velocity. Density. Cross sectional area. Drag coefficient. µ rr Effective friction constant. g Gravitational constant. V GS P R T P B V f V DSS I D R DSon RX QX C min Q g I qbs(max) Q ls I Cbs(leak) V cc V f V LS Gate to source voltage of a MOSFET. Value in timer period register. Period of peripheral bus clock. Voltage. Frequency. Drain-source voltage. Drain current. Drain-source on resistance. Resistance X. Transistor X. Minimum capacitance. Gate Charge. Quiescent current. Level shift charge. Capacitor leakage current. Voltage supply. Forward voltage drop. Low side MOSFET voltage drop. - xii -

14 LIST OF TABLES V DS I DM Drain to source voltage of a MOSFET. Pulse drain current. - xiii -

15 1. Introduction Chapter 1 Introduction The introduction of this report will provide the motivation behind this project and a high level, general overview. The problem definition and specifications of the prototype will be explained. Lastly, the structure of the report will be briefly outlined. 1.1 Motivation Electric vehicles are becoming more popular in the automotive industry due to increasing interest in renewable resources. They are more efficient than conventional gasoline vehicles, utilizing a higher percentage of available energy to power the wheels [1],[2]. Efficiency can be further increased by implementing regenerative braking. Regenerative braking is a valuable addition to electric vehicles because it allows the batteries to be charged when power is not being applied to the wheels. This allows the electric vehicle to use otherwise wasted energy to charge the batteries while driving; hence, the vehicle will run for longer periods of time between full charges

16 1.2 General Overview Fig. 1.1: Electric motorcycle 1.2 General Overview The purpose of this project was to design and implement a controlled dc-dc converter for the electric motorcycle shown in Figure 1.1 and package all of the hardware, resulting in a working motorcycle. A working motorcycle is defined as one for which the motor speed is controllable via the throttle input. Firstly, a converter topology was chosen and then verified through simulation. Once verified, construction and testing of the prototype began. The resultant converter is capable of powering the motor in a controlled manner. The acceleration of the motorcycle (i.e. the torque applied by the motor) is controlled by the throttle. Though speed is not directly controlled, the rider may control the speed indirectly through adjustments to the throttle. Figure 1.2 shows a schematic of the system. Regenerative braking was considered at every step of the converter design. A model of the converter and control scheme was developed and implemented in simulation. This model is fully capable of regenerative braking. The original intention of the project was to implement regenerative braking in hardware if time permitted and the converter was designed as such. In theory, the converter prototype should be able to support regenerative braking. However, due to the unavailability of a safe testing procedure, it has not been possible to confirm regenerative braking operation of the converter prototype

17 1.2 General Overview Fig. 1.2: System overview - 3 -

18 1.3 Problem Definition Table 1.1: Specifications Feature Interpretation Requirement Rated current Output current of the converter 100A continuous and 300A for up to one minute Rated output voltage Average output voltage at the 0-48V (variable) terminals of the converter Rated input voltage Input voltage at the terminals 55V of the converter Maximum speed Maximum speed the motorcycle 60-80km/h can attain on level ground (based on simulation) Acceleration Time required for the motorcycle to reach 60km/h (based on simulation) 10-15s 1.3 Problem Definition There are many constraints involved in this project and therefore the problem is defined as follows. The converter must accept a constant 55V input from the batteries and output a controlled, variable voltage less than 48V average to the terminals of the motor. The converter s control input must use the throttle. The converter must limit the motor current to less than 300A at all times and only allow current greater than 100A for a maximum time of one minute. Furthermore, the converter must be mountable within the chassis of the motorcycle. These constraints directly lead to the specifications shown in Table Outline The remainder of this report will explain the development of the converter prototype. This will begin in the Modeling and Simulation chapter, which will discuss selection and simulation of a converter topology as well as results of the simulations. The Hardware Design chapter will detail the selection and operation of each of the hardware components. The - 4 -

19 1.4 Outline Hardware Implementation chapter will demonstrate how the prototype was assembled and tested. The Final Prototype chapter will contain the performance results of the converter and indicate how the converter was mounted into the motorcycle chassis. Throughout the project, many challenges were encountered and are discussed in their relevant sections

20 2. Modeling and Simulation Chapter 2 Modeling and Simulation 2.1 Power Electronic Converter A dc-dc converter is responsible for generating a controllable dc voltage from a fixed dc source [3]. In the context of this thesis, it must convert the fixed dc voltage of the batteries to an acceptable dc voltage that can be used to drive the dc motor of the motorcycle. A variety of converter topologies were investigated, including both full H-bridge and half bridge topologies as well as single switch topologies. The fundamentals behind dc motor operation must be reviewed in order to understand the process through which a suitable converter topology was identified. The beauty of dc motors lies in the ease with which they can be controlled. The steady state speed in either direction is controlled by the magnitude and polarity of the applied voltage, while the torque is controlled by the resulting current. The current through the motor at any given instant in time will depend on the difference between the applied voltage and the back electromotive force (EMF). Thus by varying the magnitude of the applied voltage relative to the back EMF, either positive (motoring) or negative (generating) torque can be developed. Therefore, any dc motor whose armature voltage can be controlled is - 6 -

21 2.1 Power Electronic Converter Fig. 2.1: Torque-speed plane inherently capable of four-quadrant operation, referring to the quadrants of the torquespeed plane shown in Figure 2.1 [4]. The converter topology must be capable of operating on multiple quadrants of the torque-speed plane in order to achieve a system with the capacity of both motoring and regenerative braking. Since motorcycles typically do not have a reverse gear, only twoquadrant operation is required. More specifically, the converter must be able to operate in both the forward motoring (top right) and forward generating (bottom right) quadrants. With the apparent necessity of two quadrant operation, all single switch topologies were eliminated as a potential prospect, as they can only provide a single quadrant of operation. A full H-bridge converter topology has the capability to operate in all four quadrants of - 7 -

22 2.2 DC Motor the torque-speed plane, while a half bridge converter topology is limited to only two. The half bridge converter topology was ultimately selected because not only did it provide the necessary two-quadrant operation, but it did so with reduced complexity relative to that of a full H-bridge converter topology. 2.2 DC Motor This section focuses on the dc motor that drives the motorcycle. First, an explanation of how a permanent magnet dc (PMDC) motor functions is given. Secondly, a model for the PMDC motor is developed Overview of Permanent Magnet DC Motors A simple drawing of a PMDC motor is given in Figure 2.2 to aid in the explanation. The motor has a stationary magnetic field B created by permanent magnets in the stator, and a rotor with an armature winding. Current I is created in the armature winding by applying a voltage to the armature winding s terminals. According to equation 2.1, a force F that is orthogonal to the magnetic field and the armature winding current is created on the rotor, F = B IL (2.1) where L is the length of the wire. The torque of the created force causes the rotor to rotate. Equation 2.2 demonstrates how the torque τ e of the motor is proportional to the current through the armature winding, τ e = KI (2.2) - 8 -

23 2.2 DC Motor Fig. 2.2: Permanent magnet dc motor where K is the EMF constant that is determined by the mechanical parameters of the motor. As equation 2.3 shows, the EMF constant also relates the voltage and angular velocity of the motor, E a = Kω m (2.3) where E a is the internal dc voltage of the motor and ω m is the angular velocity of the motor Modeling of the PMDC Motor A working model of the motor allows predictions to be made about the motor s operation. The model will be described by two equations. The first equation is found by studying the circuit model of the PMDC motor as seen in Figure

24 2.3 Motorcycle Fig. 2.3: Permanent magnet dc motor model By simple circuit analysis of Figure 2.3, equation 2.4 is developed and describes the voltage V a supplied to the motor, V a = Kω m + R a i a + L a d dt i a (2.4) where K is the EMF constant, ω m is the motor angular velocity, R a is the motor winding resistance, i a is the armature current, and L a is the motor winding inductance. The second equation that describes the motor model is given in equation 2.5. It describes the electrical torque τ e created by the motor, τ e = τ l + βω m + J d dt ω m (2.5) where τ l is the load torque, β is the viscous damping coefficient, ω m is the motor angular velocity, and J is the motor moment of inertia. The parameters of the PMDC motor were provided by a previous student group who worked on the same electric motorcycle; it can be seen in Table Motorcycle The motorcycle and its rider serve as the load of the motor. The dynamics of the motor s load must be modeled in order to understand how the motor will perform. A motorcycle

25 2.3 Motorcycle both stores and dissipates energy. This energy is provided primarily by the motor, stored in the form of kinetic energy, and dissipated through various means. In some cases the energy may return to the motor. Therefore, the model must be able to accept, return, and consume energy. Equation 2.6 represents this flow of energy, d dt E = P motor P air P friction (2.6) where E is the kinetic energy of the motorcycle, P motor is the power provided by or returned to the motor, P air is the power lost to air resistance, and P friction is the power dissipated due to friction. Kinetic energy is governed by equation 2.7, E = mv2 2 (2.7) where m is the mass of the motorcycle and v is the forward velocity of the motorcycle. P motor is governed by equation 2.8, P motor = Jω m d dt ω m (2.8) Table 2.1: Parameters of the PMDC motor [5] Parameter Symbol PMDC Motor Value Rated continuous current I rated 100A continuous, 300A for maximum one minute Rated voltage V rated 48V EMF constant K 0.126V/rad Armature resistance R a 0.01Ω Armature inductance L a 0.055mH Viscous damping coefficient β negligible Motor moment of inertia J 0.018kg/m

26 2.3 Motorcycle where J is the motor moment of inertia and ω m is the motor angular velocity. P air is governed by equation 2.9, P air = ρac dv 3 2 (2.9) where ρ is the air density, A is the frontal area of the motorcycle and its rider, C d is the drag coefficient, and v is the forward velocity of the motorcycle. P friction is governed by equation 2.10, P friction = µ rr mg (2.10) where µ rr is a constant that is determined by the motorcycle s tire size and pressure, m is the mass of the motorcycle, and g is the acceleration due to gravity. The following parameters of the motorcycle, in Table 2.2, were provided by a previous student group who worked on the same electric motorcycle. Table 2.2: Parameters of the motorcycle model [5] Parameter Symbol Simulation Value Total mass of motorcycle m 200kg Motor moment of inertia J kg/m 2 Air density ρ 1.25kg/m 3 Front area of motorcycle A 1m 2 Drag coefficient C d 0.6 Frictional constant µ rr Acceleration due to gravity g 9.81m/s

27 2.4 Control System Design 2.4 Control System Design This section will first explain the differences between two types of controllers: proportional integral (PI) and open loop. Then, an overview of the control scheme is given Controller Type A PI controller works by first calculating an error signal using a desired reference signal and the actual output signal. The controller seeks to minimize the error signal, thus bringing the actual output signal as close as possible to the desired reference signal. In contrast, an open loop controller simply applies a gain to the reference signal and outputs the result. There is no measurement of the actual output signal and therefore no error signal exists to minimize. Because of this, the output responds very quickly to changes in the inputs. PI controllers are often slow compared to open loop controllers. While PI controllers generally output a very accurate signal (compared to the reference), they are relatively more complex and time consuming to design. An open loop controller is very simple and easy to create but does not necessarily provide an accurate output signal. On a standard motorized vehicle, the throttle does not directly control speed, but instead controls torque (i.e. acceleration). Generally, the throttle itself does not have any markings or method to select specific values of torque and a vehicle has no meaningful way of determining what the applied torque is. Therefore, typically it is not possible to use a PI controller for torque, as an error signal cannot be generated. Additionally, the driver of the vehicle generally wishes to control the speed of the vehicle and not the torque. Speed control is accomplished indirectly by the driver. The driver selects a desired speed and compares that to the current speed. The driver adjusts the throttle until the desired speed is reached. In this way, the driver is acting as a PI controller where the error signal is generated in the driver s mind. As mentioned in section 2.2, torque is proportional to the motor current for a dc motor

28 2.4 Control System Design Current (and therefore torque) can be measured and could be used to form an error signal for a PI controller. However, given that the driver: has no way to select a specific value of torque, has no access to the torque measurement, and desires only to control the speed, a PI controller will not add any value to the control system of the motorcycle. Generally, on a motorcycle, the expectation from the driver is that the torque should change immediately with a change in the throttle. Due to these reasons, an open loop current controller is optimal since it has low complexity and a fast response. In this project, the specifications state limits on how much current is allowed, therefore the controller must impose these limits. A true open loop controller is not possible since the current must be measured to ensure it is within the limits. Therefore an open loop current controller with limited feedback was selected Control Scheme The main routine logic is shown in Figure 2.4. The main function serves only two purposes. On power up, the main function initializes all the relevant variables and modules then it repeatedly reads from the analog to digital converter (ADC) in an infinite loop. The results from the ADC represent the duty cycle input from the throttle and the motor current measurement. These values are stored in global variables such that the interrupt routine can access them. The majority of the control system lies in the interrupt routine. The specific logic of the routine is shown in Figure 2.5. At a very high level, the purpose of the interrupt routine is to impose any limits and alter the duty cycle if necessary. In more detail, first the input from the throttle is checked to ensure that it does not go out of bounds. After that, the current limit is altered depending on the previous behavior of the motor current. This is in accordance with the motor current specifications. Finally the current limit and the input from the throttle are used to make alterations to the duty cycle. The input from the throttle

29 2.4 Control System Design may be used or rejected depending on the present state of the motor current. Duty cycle ramping is implemented in the increasing direction and for the case of over limit motor currents. However, the throttle input will always be used immediately if it is attempting to decrease the duty cycle. Fig. 2.4: Main routine flow chart

30 2.4 Control System Design Fig. 2.5: Interrupt routine flow chart

31 2.5 Simulation Results 2.5 Simulation Results The converter topology and control system were implemented in the PSCAD/EMTDC transient simulator using the motor and motorcycle models discussed in sections 2.2 and 2.3. All of the simulation results in this section were obtained using a 1µs solution time step; however, the plot time steps may vary. The converter and controller models both performed as expected in forward motoring mode. Simulation predicted that the current ripple of the motor in steady state would be approximately 11A peak to peak. A waveform of the ripple captured from the simulation is shown in Figure 2.6. The motorcycle model predicts that the top speed of the motorcycle should be approximately 65km/h. This prediction takes into consideration that the motor current should be limited below 300A at all times and the continuous current below 100A. Plots of the simulation result are shown in Figures 2.7 and 2.8. It can also be inferred from Figure 2.7 that the acceleration time from rest to top speed should be approximately 10s. These results meet the specifications for simulation in Table 1.1. Motor Current [A] Time [µs] Fig. 2.6: Motor current ripple in steady state

32 2.5 Simulation Results Speed [km/h] Motor Current [A] Time [s] Fig. 2.7: Top speed test - speed Time [s] Fig. 2.8: Top speed test - current The converter model was also able to run correctly in regenerative braking mode. Figures 2.9 and 2.10 show a simulation test where the motorcycle was allowed to accelerate for 5s at which time regenerative braking mode was triggered. During this specific run of the simulation, a 100A reverse current was selected. The choice of 100A was arbitrary and the control scheme can be tuned to any desired reverse current. The plot shows that the motor current transitioned from 300A to -100A once regenerative braking was triggered. Figure 2.11 shows a zoomed in view of the motor current at the moment of transition. Speed [km/h] Time [s] Motor Current [A] Time [s] Fig. 2.9: Braking test - speed Fig. 2.10: Braking test - current

33 2.5 Simulation Results 300 Motor Current [A] Time [s] Fig. 2.11: Transition to regenerative braking The current ripple during regenerative braking is shown in Figure It is similar to the ripple present during forward motoring with a slightly larger peak to peak amplitude of 13A. It is important to note that the current ripple during regenerative braking is never in steady state due to the dynamics of the motor. 95 Motor Current [A] Time [µs] Fig. 2.12: Motor current ripple during regenerative braking For the purposes of simulation, the control system did not implement any duty cycle ramping. This means that the duty cycle was allowed to instantly change to any value which caused very fast changes in motor current. The hardware prototype implements duty cycle ramping in order to restrict the rate of change of motor current

34 2.5 Simulation Results As shown in Figures 2.7 and 2.8, the converter model can correctly limit the motor current to a selectable value and still provide reasonable acceleration characteristics. Additionally, the converter can also switch at any time to regenerative braking mode and provide a controlled, selectable charge current as shown in Figure For these reasons, the simulation results show that both the converter topology and control system are viable

35 3. Hardware Design Chapter 3 Hardware Design - Preliminaries 3.1 Switch Type Selection It was concluded in section 2.1 that a half bridge power module is needed; it could be either purchased or built. Buying the module would be more expensive than building it. However, the decision was made to buy the module because the switches within the module would be matched, which allows for greater accuracy of switching between on and off states. Next, the decision on the type of half bridge power module had to be made. The two options for a half bridge power module were an Insulated-Gate Bipolar Transistor (IGBT) or a Metal-Oxide-Semiconductor Field-Effect Transistor (MOSFET). The initial decision was to use an IGBT module because it could include the gate drive circuitry within the module, which would shorten the product development time. However, a module with the gate drive circuity included was not available to ship within the time schedule. Alternatively, MOSFETs have lower switching losses than IGBTs and building the surrounding gate drive circuitry is more cost effective; therefore, a MOSFET module was chosen. When choosing a specific MOSFET module to purchase, the two ratings that were most focused on were the voltage and current ratings of the module. The module had to be rated

36 3.1 Switch Type Selection greater than the battery bank voltage rating of 55V, and greater than the motor current rating of 300A. The module also needed to be able to switch at a frequency of 20kHz. An APTM10AM02FG MOSFET power module was purchased that suited the desired ratings; its circuitry is shown in Figure 3.1. Since the MOSFET module that was purchased did not include the gate drive circuitry, it was necessary to design and purchase the components to build the gate drive circuitry. The most relevant ratings of the module are given in Table 3.1. Fig. 3.1: MOSFET power module circuitry Table 3.1: MOSFET power module ratings [6] Parameter Label Max Ratings Drain-source breakdown voltage V DSS 100V Continuous drain current I D 495A Drain-source on resistance R DSon 2.25mΩ Switching frequency f 20kHz

37 3.2 Gate Drive Circuitry 3.2 Gate Drive Circuitry This section focuses on the gate drive circuitry for the power module. First, an introduction as to the necessity of a gate drive circuit is provided, followed by several brief explanations regarding various relevant topics, such as low and high side switching configurations, bootstrapping circuits, power buffer circuits and finally gate drive integrated circuit (IC) selection Introduction to Gate Drive Circuits In many circuits, it is often necessary to use power semiconductor devices such as MOSFETs as the primary means for switching. Switching circuits can be found in a wide variety of applications such as those found in dc-dc converters. The drive signals for these MOSFETs are typically generated by some means of low power circuitry such as a microcontroller or some other external ICs, be it a pulse width modulation (PWM) controller or a pulse generator. Regardless of the choice of low power circuitry, the resulting drive signal for the MOSFET is typically limited to a few ma of current, and a voltage level similar to those found in common logic level families (i.e. ranging from 3.3V to 5V). The gate drive circuit will accept this low power drive signal and amplify it, producing a high current drive input needed to drive the gate of the high power MOSFET [7]. An understanding of the fundamentals behind various switching configurations is required before the selection of a gate drive IC and the design of its surrounding circuitry can be discussed. With the half bridge circuit shown in Figure 3.1, note that there are both high side and low side switches present, each having different drive requirements. The switching configurations will be examined independently in order to better understand the particular drive requirements of the high and low side switches

38 3.2 Gate Drive Circuitry Low Side Switching Configuration Consider Figure 3.2 which shows the low side switching configuration. It contains both a load and an N-channel enhancement type power MOSFET. With this configuration, when the MOSFET is off there will be no current flowing through the load. The MOSFET must first be turned on in order to get current flowing through the load. The voltage at the gate must be approximately 8V with respect to the voltage at the source in order to turn the MOSFET fully on. That is to say, V GS 8V (3.1) where V GS is the gate to source voltage of the MOSFET. Since the source terminal of the MOSFET is sitting at ground potential, the MOSFET can easily be turned on by simply developing 8V or more at the gate terminal. Note that with the MOSFET fully on, it begins to act as a current sink, sinking current from the load. Fig. 3.2: Low side switching configuration High Side Switching Configuration Consider Figure 3.3 which shows the high side switching configuration that contains both a load and an N-channel enhancement power MOSFET. Just like with the low side switching

39 3.2 Gate Drive Circuitry Fig. 3.3: High side switching configuration configuration, no current will flow when the MOSFET is off. A gate voltage of approximately 8V with respect to the source voltage is required in order to fully turn the MOSFET on and allow current to flow. However, in this configuration note that the terminal of the MOSFET no longer sits at ground potential, but rather at whatever potential the load is with respect to ground. Depending on the potential of the source of the MOSFET, the drive requirements for the MOSFET can vary significantly. For example, if the source is sitting at 1V with respect to ground then only 9V or more is needed at the gate in order to turn the device on. Similarly, if the source is sitting at 40V with respect to ground then 48V or more is needed at the gate in order turn the device on. Finally, note that with the MOSFET fully on, it begins to act as a current source, sourcing current to the load [8]. There are several options available for driving MOSFETs in a high side switching configuration. These options include things such as: boosting the driving voltage at the gate by means of a boost converter or charge pump circuit, using a separate isolated power supply whose ground is isolated from the ground of the MOSFET-based circuit, or implementing a bootstrap based drive circuit. Due to the simplicity of its operation, the availability and low

40 3.2 Gate Drive Circuitry cost of the additional diode and capacitor, a bootstrap based drive circuit was selected. The operation and design of the bootstrap based drive circuit will be examined in the following section Bootstrap Circuitry Fig. 3.4: Bootstrap circuit A basic bootstrap based drive circuit is shown in Figure 3.4. When the bottom switch is closed, the capacitor is charged up to 15V through the conducting diode. Once charged, the capacitor voltage can then be placed directly across the gate and source terminals to turn the power MOSFET on, provided the control switch between the capacitor and gate terminal of the power MOSFET is closed. This control switch grants control as to whether the MOSFET is to be on or off [9] Power Buffer Circuitry The main function of the power buffer circuit is to provide the gate drive IC with a high impedance input, while providing a lower drive impedance for the power MOSFET. Figure 3.5 shows a power buffer circuit which accomplishes these tasks

41 3.2 Gate Drive Circuitry Fig. 3.5: Power buffer circuit The operation of the power buffer circuit is as follows. With the input low the leftmost P-channel and rightmost N-channel MOSFETs turn on, pulling the output node low. With the input high, the leftmost N-channel and rightmost P-channel MOSFETs turn on pulling the output high. The resistor in the middle of these opposing pairs of N-channel and P- channel MOSFETs is needed to limit the current in the event that opposing MOSFETs are both on during a transition period due to changes in the input. Furthermore, with the bootstrap capacitors in place to provide the necessary drive current for the power MOSFET, the rightmost P-channel MOSFET will accommodate this surge of current while keeping the gate drive IC protected by the rest of the power buffer circuit Complete Gate Drive Circuit The gate drive IC selected to accommodate the switching signals was the high low side IR2110 driver from International Rectifier. This particular IC was selected due to its increasing popularity and widespread success with electric vehicle related applications. With the configuration shown in Figure 3.6, both the high and low side switches for a half bridge power MOSFET configuration can be successfully driven

42 3.3 Additional Components Fig. 3.6: Gate drive block diagram 3.3 Additional Components The selection of two of the main components, the MOSFET power module and the gate drive circuitry, have been discussed previously. This section will focus on the hardware selection for the remaining components of the project: the bootstrap circuit, the buffer MOSFETs, the dc-dc converters, the optoisolator circuitry, and the decoupling capacitors Bootstrap Circuit The purpose of a bootstrap circuit has been explained in section 3.2.4, and now the process of selecting the bootstrap circuit components is discussed. The selection of the capacitor is governed by equation 3.2; C min ( ) 2 2Q g + I qbs(max) f + Q ls + I Cbs(leak) f (3.2) V cc V f V LS where C min is the minimum value of the bootstrap capacitor, Q g is the gate charge of the high side MOSFET, I qbs(max) is the quiescent current for the high side drive circuitry, f is the switching frequency, Q ls is the level shift charge required per cycle, I Cbs(leak) is the bootstrap capacitor leakage current, V cc is 15V, V f is the forward voltage drop across the bootstrap diode, and V LS is the voltage drop across the low side MOSFET. A value of

43 3.3 Additional Components C min was calculated to be approximately 0.5µF. An electrolytic 10µF capacitor was used in parallel with a ceramic 0.1µF capacitor, giving an equivalent capacitance of 10.1µF. The parallel ceramic capacitor was used to filter out the high frequency portions because electrolytic capacitors do not perform well at high frequencies. Both of these capacitors were used not only because of the size requirement, but because they were available at no cost. The last component of the bootstrap circuit is the diode. The bootstrap diode must have a sufficient reverse voltage rating, current rating, and recovery time. The required current rating is the product of the MOSFET gate charge and the switching frequency, which is 27.2mA. The 1N4149 diode was chosen because it meets all the requirements [10] Buffer MOSFETs The purpose of a power buffer circuit has been explained in section 3.2.5, and now the process of selecting the power buffer MOSFETs is discussed. The P-channel and N-channel MOSFETs must be able to accept a 15V gate to source voltage. The part numbers of the MOSFETs that were purchased are IRFD9220PbF (P-Channel) and IRFD110PbF (N- Channel). Their relevant ratings are in Table 3.2 and Table 3.3. Table 3.2: P-channel MOSFET ratings [11] Parameter Symbol Maximum Rating Drain-source voltage V DS -200V Gate-source voltage V GS ±20V Continuous drain current I D -0.56A Pulsed drain current I DM -4.5A

44 3.3 Additional Components Table 3.3: N-channel MOSFET ratings [12] Parameter Symbol Maximum Rating Drain-source voltage V DS 100V Gate-source voltage V GS ±20V Continuous drain durrent I D 1.0A Pulsed drain current I DM 8.0A DC-DC Converters The only voltage supply that is available on the motorcycle is the 55V battery bank; however, 15V and 5V supplies are needed for the logic system and the low power circuitry. To achieve this, two dc-dc converters and a voltage regulator are required. One dc-dc converter must be able to take a 55V dc input from the battery bank and output a dc voltage of 15V. The model chosen was the DE03S4815A because it is rated for an input range of 36V-75V and an output voltage of 15V. The converter has a maximum rated current output of 200mA and an internal power dissipation of 2.5W, both of which suit its application [13]. Another dc-dc converter must be able to take a 15V input from the first dc-dc converter and output an isolated dc voltage of 5V for the microcontroller. The model chosen was the PB01S1505A because it is rated for an input range of 13.5V-16.5V and an output voltage of 5V. The voltage regulator must also be able to take a 15V input from the first dc-dc converter and output a dc voltage of 5V for the low power circuitry. The model chosen was the LM7805C because it is rated for an input voltage of up to 35V

45 3.3 Additional Components Optoisolator The need for an optoisolator was discovered in the hardware implementation phase, which will be discussed in section The purpose of an optoisolator is to provide isolation between the microcontroller and the high power circuitry, such that the high voltages do not affect the microcontroller receiving the signal. The HCPL2531 optoisolator was purchased. A pull-up resistor ensures that the voltage level of a line is never floating, such that when the other components on the line are inactive the voltage will be pulled up to V cc. A pull-up resistor is needed between the output pin of the optoisolator and a dc voltage of 5V. A smaller resistor is preferable because of capacitative effects of the input pins. A resistor value of 2.2kΩ was chosen. This value was chosen because it is the smallest value that is large enough to fully transition to signal low. Figure 3.7 shows the placement of the pull-up resistor. Fig. 3.7: Placement of the pull-up resistor The need for an inverter stage following the optoisolator was also discovered in the hardware implementation phase. The optoisolator inverts the signal from the microcontroller, such that a low signal input to the optoisolator would send a high signal to the gate drive circuitry. Compensation for the inversion must be done to prevent the batteries from shorting through the high power MOSFET module. A simple inverter circuit is shown in Figure 3.8, which was implemented for both signals from the optoisolator

46 3.3 Additional Components Fig. 3.8: Inverter circuit for the optoisolator signal The same P-channel and N-channel MOSFETS were used for the inverter stage as were purchased for the power buffer circuit because the ratings suited the application. The MOSFETs were already available because extra parts were ordered for the buffer Decoupling Capacitors The purpose of a decoupling capacitor on a dc voltage supply is to remove the high frequencies from the dc voltage level, and therefore help to provide a more precise dc supply. A large electrolytic capacitor and a small ceramic capacitor used in parallel is good practice for decoupling; this is because an electrolytic capacitor can provide a large capacitance relative to its physical size and a ceramic capacitor can perform well under high frequencies. Therefore, a 10µF electrolytic capacitor and a 0.1µF capacitor were placed across the output terminals of both dc-dc converters, as well as across the output terminals of the voltage regulator. The capacitor sizes were chosen because they provided a large enough capacitance, and they were available at no cost

47 4. Hardware Implementation Chapter 4 Hardware Implementation, Testing and Verification 4.1 Power and Gate Drive Circuitry In this section, the breadboard design and testing will be discussed first. Next the wire wrap board, the soldering board and their purposes will be discussed Breadboard A breadboard allowed for flexibility while testing. Initially each buffer circuit was built and tested individually. A diagram in an application note, as can be seen in Figure 4.1, was used to build the buffer circuits [10]. When initially testing the buffer circuits, there was overheating of the MOSFETs and the output was not what was expected. When looking at Figure 4.1 it can be seen that the top P-channel MOSFETs are drawn incorrectly, which caused the problem. The MOSFETs were flipped, and connected correctly as was previously seen in Figure 3.5. Once the MOSFETs were connected correctly, the input was initially grounded which resulted in 0V at the output as expected. When 5V was given to the input,

48 4.1 Power and Gate Drive Circuitry Fig. 4.1: Power buffer schematic issue Fig. 4.2: Logic signal and buffer output the output of the buffer was 15V, as expected. A function generator was then used as the input. The output once again followed the input at a greater output voltage. There was slight ringing as seen in Figure 4.2, but it was small enough that it should not be an issue with the power MOSEFT module. Next, the gate drive chip was connected to the two buffer circuits. Two function generator inputs were needed for this test because switching of the lower buffer was needed to charge the capacitor and the switching of the top buffer was needed to check that the circuit topology worked. The two input signals to the gate drive chip should never be high

49 4.1 Power and Gate Drive Circuitry at the same time. Since the buffers were not yet connected to the power MOSFET module, it was not critical to make sure both input signals were not high for testing. The final circuit ensures that both buffer outputs will never be high at the same time. This would close both power MOSFET switches, essentially shorting the battery bank. The test with the gate drive chip and buffer circuits was successful. The outputs of each individual buffer circuit followed the input of 5V at a higher voltage of 15V. A half power test was completed with the breadboard design. Two 12V batteries were connected in series to give an output voltage of 24V. This was to test that the low power circuitry was working as expected; therefore, the full 48V was not needed at this time. In this setup, the function generator was controlling the duty cycle since isolation had not yet been implemented to protect the microcontroller. The power MOSFET module was connected to the motor and switched by the breadboard. The breadboard board was being supplied by an external 15V supply. Initially, two gate drive chips were damaged when doing this. A decoupling capacitor was connected to the lower power switch in order to reduce transients when making the final connections and this solved the problem. With the decoupling capacitor this test was successful, the motor turned and went faster with a higher duty cycle as expected. During this test the power MOSFET module was getting warm. There was no heat sink at the time and it was determined that a heat sink would be necessary before performing any additional testing. The gate drive IC previously selected is not isolated, thus an optoisolator is required in order to separate the low and high power circuitry. The optoisolator needed a pull-up resistor at the output to work properly. This created an inverted signal at the output. Initially an inverted signal was given to the input so a low input to the optoisolator would give a high output to the power module and vice versa. This was tested and worked as expected. Although it was realized that a problem could occur if the microcontroller loses power during operation. If the microcontroller is off, the optoisolator would have a high

50 4.1 Power and Gate Drive Circuitry Fig. 4.3: Buffer output extreme ringing output, which gives a high signal to the power MOSFET module. A high signal would close the switch in the MOSFET module indefinitely, causing the motor to go out of control. This was not a safe design, therefore an inverter circuit was added after the optoisolator to invert the signal. The inverter includes two MOSFETs: one P-channel and one N-channel. The inverter circuit has been shown previously in Figure 3.8. When initially testing the optoisolator there was extreme ringing as seen in Figure 4.3 when the signal should be low. Decoupling capacitors across the sources had not yet been included, and they were implemented once this problem occurred. Initially the capacitors were electrically connected to the source, but not physically close, and that did not solve the problem. The decoupling capacitors must be physically close to reduce the loop length from the power to the ground on the chip, since loops act as antennas and input noise into the system. Once the capacitors were moved physically closer, the ringing problem was solved

51 4.1 Power and Gate Drive Circuitry Wire Wrap Board Initially the wire wrap board did not include all of the low power circuitry. It was built in stages, starting with the components that had been tested and worked as required. It was the first step in moving away from the breadboard. It allowed for more reliability as well as saved time when setting up, as connections no longer needed to be checked. A soldering board was always looked at as a potential next step. Having the wire wrap board allowed for a backup of the low power circuitry and was a good intermediate step, as it was easier to make changes to if need be. All of the components were mounted in sockets, which allowed for component values to be changed if need be, as well as allowed for the other components to be removed if they were needed in the future soldering board circuit. The wire wrap board included the gate drive chip, as well at the top and bottom buffers. A few extra sockets were added in order to power the components from an external source. It was tested successfully and worked as the breadboard circuit had previously. It was determined that the low power circuitry would have to be powered from the 48V battery bank. Some components needed 15V while others needed 5V. The microcontroller also needed an isolated ground from the rest of the low power circuitry. The two voltage converters and the voltage regulator were implemented into the design to meet the voltage and isolation requirements. A 48V-15V converter was used to get the desired 15V from the batteries. An isolated 15V-5V converter was used for the microcontroller, and a nonisolated 15V-5V regulator was used for the rest of the components needing 5V. This was built on the breadboard for testing. This was never implemented on the wire wrap board, since the initial wire wrap design did not account for the amount of space these components needed. It was directly implemented on the soldering board

52 4.1 Power and Gate Drive Circuitry Soldering Board A soldering board was chosen as a final design for the low power circuitry. This allowed for even more reliability than the wire wrap board. A printed circuit board was not chosen due to its high cost and inflexibility in the event that changes need to be made. All components, except decoupling capacitors, were mounted in soldering sockets. This allowed for components to be easily be removed and changed in the event they were damaged during testing or for design adjustments. The buffer circuit as well as the gate drive chip were soldered and tested before anything else. This circuitry was finalized before the addition of the optoisolator and the voltage converters. With this initial circuitry on the soldering board, a system test at 48V was completed to ensure that the circuitry was working as expected. Since the top switch of the power MOSFET module will only be used for regenerative braking, the gate and drain were tied together to ensure that the switch did not close. This test included: the voltage converters and optoisolator circuitry on a breadboard, the circuitry on the soldering board, microcontroller, power MOSFET module, 48V battery bank, the motor, and a potentiometer as the throttle. Initially the throttle was not easy to access, so a potentiometer on a breadboard was used. The batteries were powering the microcontroller, low power circuitry, and motor through the power MOSFET module. No outside power sources were used and the system worked as expected. The motor initially was at rest when everything was connected with the potentiometer at the lowest resistance. When the resistance of the potentiometer was increased, the duty cycle of the PWM was increased by the microcontroller. The low power circuitry gave the high power output of the PWM to the power MOSFET module, which in turn, caused the motor to rotate. Changes in the resistance were reflected in the speed of the motor. After the success of this test, the rest of the circuitry was soldered to the board. Many isolated tests were done on the soldering board, to be sure every part was working

53 4.2 Control System Hardware as expected. These tests were the same tests that were done on the breadboard in section Control System Hardware The microcontroller acts as the brain and main control system for the electric motorcycle. The switches within the half bridge converter are controlled by the microcontroller through the use of both the PWM and ADC modules. The PWM module is needed in order to output the necessary PWM waveforms for the switching of the converter, while the ADC module is responsible for digitizing the analog voltage signals from both the throttle potentiometer and Hall effect sensor. This digitized information is then analyzed by the microcontroller and used to readjust and refine the PWM output signals according to the particular constraints imposed by the given control scheme. Since the PWM module is responsible for generating the PWM waveforms used in the switching of the converter, these waveforms must be generated at the desired switching frequency for which the converter was designed. The operation of the PWM module must be examined in order to verify this. The PIC32MX795 microcontroller has five output compare units that can be used to generate PWM outputs. The output compare units work in conjunction with the timer to generate the required PWM signal. More specifically, the PWM period is specified by writing to the timer y period register (PRy), and the PWM duty cycle is specified by writing to the Output Compare x Secondary Register (OCxRS) [14]. The equations for calculating the PWM period and frequency are as follows, PWM Period = (P R + 1)T P B (TMR Prescale Value) (4.1) PWM Frequency = 1 P W MP eriod (4.2)

54 4.2 Control System Hardware where P R is the value of the given timer period register, T P B is the period of the peripheral bus clock and (TMR Prescale Value) is the prescalar value of the associated timer for the given PWM output. From equations 4.1 and 4.2, it can be seen that the frequency of the PWM output depends on each of the three variables mentioned previously. Given that the frequency of the peripheral bus clock is 80MHz and the desired switching frequency is 20KHz, the desired switching frequency was obtained by selecting appropriate values for the timer period register and associated timer prescalar value. The operation of the PWM module was initially verified by programming a fixed switching frequency and duty cycle and verifying the output waveform meets the necessary requirements via an oscilloscope. Once the ADC was successfully implemented, the functionality of the PWM was further tested by varying the duty cycle of the output waveform via a potentiometer. The ADC module is responsible for digitizing the analog voltage values from both the throttle and the Hall effect sensor. The PIC32MX795 microcontroller has a 10-bit ADC that can accommodate up to sixteen analog inputs. The converted values produced by the ADC will be in the range of 0 to 1023 [14]. One of the issues encountered during the development of the microcontroller was related to the use of global variables in multiple threads. The threads in question include both the main and ISR subroutines. Due to the limitations of the XC32 compiler, whenever the global variable in question was accessed in both main and interrupt subroutines, the data stored within this variable would become corrupted. This would then cause the program counter to point at a location in memory in which there was no source code, subsequently crashing the program. This problem was circumvented by simply disabling interrupts whenever the shared global variable was to be accessed by either subroutine. By disabling the interrupts, the data stored in the variables was kept intact

55 4.3 Hall Effect Sensor 4.3 Hall Effect Sensor A Hall effect sensor measures a magnetic field and outputs a voltage in relation to that. In this project, the Hall effect sensor is used to measure the motor current. The motor current must be limited according to the specifications. The Hall effect sensor is needed to measure the current, and the voltage output of the Hall effect sensor is the input into the microcontroller. The microcontroller will limit the duty cycle of the PWM waveform if the current is exceeding the designated threshold. It needs a voltage source of 5V to operate. An input of 0A should correspond to an output voltage of 2.5V. Initially the Hall effect sensor was put on the wire going from the battery bank to the motor during the first half power test with 24V. It appeared there was an offset when it was measured. An ammeter was used to monitor the current as well and that value could be compared to the output of the Hall effect sensor. Since the current going to the motor is low during testing, around 10-20A, the current could be discontinuous due to the current ripple from PWM. The current ripple was measured to be around 14A when testing with a current probe. Since the motor current had a large ripple, a different test needed to be performed to determine the Hall effect sensor was working properly. The Hall effect sensor was tested with one 12V battery and a carbon stack as the resistance. Since this was a dc current, current ripple was not an issue. The current transformer was used to measure the current to compare to the voltage output of the Hall effect sensor. Table 4.1 shows the corresponding voltage and current measurements. Equation 4.3 shows the relationship between the current input and the voltage output of the Hall effect sensor. Table 4.1 and equation 4.3 show that the Hall effect sensor is working as expected. V = I (4.3) where V represents the voltage in volts, while I represents the current in amps

56 4.3 Hall Effect Sensor Table 4.1: Hall effect sensor data Current [A] Measured Voltage [V] Calculated Voltage [V] Fig. 4.4: Hall effect sensor noise A system test with the low power circuitry, power MOSFET module, batteries, motor and function generator was performed again. The Hall effect sensor was once again tested in the system. When there was zero current the voltage output from the Hall effect sensor was around 2.5V as expected. When the duty cycle of the function generator was increased and the motor started turning the Hall effect sensor had a mean voltage of 2.8V at 15A. This was a higher mean voltage than expected, but that was likely attributed to the amount of ringing and noise as seen in Figure 4.4. The oscillations shown in the figure were present despite implementation of low pass filtering. The ground of the Hall effect sensor was the same ground as the battery. The next step

57 4.3 Hall Effect Sensor was to isolate the Hall effect sensor ground from the battery ground since battery grounds can input noise into the system. An isolated 15V-5V voltage converter was used to isolate the ground (as well as the microcontroller) from the battery ground. Testing with the entire system resulted in a less noisy signal than before, but still too high to be acceptable. The noise was greater than 100mV. A capacitor and an inductor were used to low pass filter, but no change in the noise was seen. Voltage converters can create a lot of noise since they are switching at a high frequency to lower the voltage. Since the noise was in the MHz range, this was likely the problem. All of the components, except the 48V-15V and 15V-5V converters were taken off of the breadboard. With both converters connected there was noise on the ground. When the 15V-5V converter was removed, the noise was no longer there. Therefore it was determined that the 15V-5V isolated converter was creating the noise in the Hall effect sensor. A number of possible solutions were implemented. The Hall effect sensor power and ground were connected directly to the microcontroller to help reject common mode noise. Having the Hall effect sensor ground physically closer to the ground that is measuring the output is what helps to reduce the common mode noise. All wires were cut as short as possible, since long wires can act as antennas that can pick up noise around them. Connecting wires were twisted together to reduce noise. If the two wires are not physically close together, noise can be input into one wire more than the other, creating differential noise. The noise is added to the input of the component to which the wires are connected. Twisting the wires also reduces stray inductances and closes loops to reduce the noise. If they are twisted together, they both receive the same noise (common mode noise) and therefore the difference between the two is the same as when there is no noise. A toroid around the wires going to the Hall effect sensor was also implemented. A toroid helps to reduce the amount of MHz noise that would come onto the wires. The toroid is most effective when it is placed on the wires closest to the source of the noise. The toroid

58 4.3 Hall Effect Sensor increases the common mode inductance of the wires, which reduces the noise [15]. During testing it was determined that noise was being input into the oscilloscope itself. This was noticed when the signal would change drastically just by bending the wire of the probe. This proved that the oscilliscope output was not representative of the actual signal. All of the noise reduction techniques listed above were still implemented to be sure the output of the Hall effect sensor was not noisy. Instead of measuring the output on the oscilloscope, the output was connected directly to the microcontroller. The debug mode was used on the microcontroller so different values could be watched. When there was no current going through the Hall effect sensor, the program was halted multiple times to see what value the Hall effect sensor was outputting; it was consistent. The next test was to implement logic into the microcontroller program. If the Hall effect sensor value was over a certain amount, the PWM would be limited so the current could not go over 20A, and therefore the motor could not speed up anymore. The motorcycle was loaded for this test, as high motor speeds would be needed to reach 20A when unloaded. During this test, the motor would no longer speed up at a certain point, no matter how much the potentiometer was turned. This concludes that the Hall effect sensor was working as expected

59 5. Final Prototype Chapter 5 Final Prototype 5.1 Enclosure When electrical components are added to a vehicle they must be contained within some form of an enclosure. Depending on the type of components and where on the vehicle they will be mounted, an enclosure serves several purposes. The enclosure may provide a solid structure to which components can be mounted as well as protect the components from various disturbances. These disturbances can vary and may include temperature, water, dirt or even people. However, it is beyond the scope of this project for the enclosure to provide protection from water and dirt. In this project, the vehicle is a motorcycle which may subject the converter to large vibration and also imposes size constraints. Furthermore, the converter produces significant amounts of heat. Therefore, the enclosure must provide a secure surface, it must be an efficient heat sink, and it must fit within the frame of the motorcycle. The constraint of heat sinking means that the enclosure must be constructed out of metal. However, a heavy enclosure will impede the performance of the vehicle, and since the converter components are not very heavy, a light metal such as aluminum is a suitable

60 5.2 Performance Testing material. It is for these reasons that the enclosure was chosen to be primarily a large aluminum block. Due to the size constraint, not all of the components could fit onto the aluminum block; therefore, an additional aluminum plate was used. All of the components requiring heat sinking were mounted on the aluminum block with thermal paste to provide efficient thermal contact. The microcontroller has no heat sink requirement and therefore was mounted on the aluminum plate. 5.2 Performance Testing The finalized prototype was tested to establish its performance characteristics. The purpose of measuring the performance of the converter is to compare the simulation results with the measured results in order to validate the simulation. Additionally, it must be demonstrated that the converter meets the physical specifications in Table 1.1. The simulation specifications are discussed in section 2.5 of this report. Many simple tests consisting of only attempting to spin the motor on the motorcycle using the converter were performed. Four 12V batteries were connected in series to produce a 48V supply and were used as the power input to the converter. The 48V supply is negligibly different from the expected 55V supply since both are well below the rating of the converter. Therefore, the 48V supply is sufficient to draw conclusions about the performance of the converter. A potentiometer was used as the throttle input. The motor of the motorcycle was used as the load. Without performing any measurements, it was observed that the converter could easily spin the wheel using the batteries as a power source. Additionally the speed of the motor could be altered by adjusting the duty cycle input. When the duty cycle was increased, the speed of the motor increased. These series of tests did not gather any numerical data; however, these tests proved that the converter design meets the input and output voltage specification. The first test to gather data with which to evaluate the converter was as follows. The

61 5.2 Performance Testing Fig. 5.1: Motor current ripple motorcycle was run under no load using the converter at 30% duty cycle. The current ripple was measured using a current probe. The waveform was captured on an oscilloscope and is shown in Figure 5.1. The plot shows that the current ripple is approximately a triangular waveform with a 14A peak to peak magnitude. As shown in Figure 2.6, the model of the system predicted that the motor current would be a triangular waveform with an 11A peak to peak magnitude. It can be inferred from the difference between the model prediction and measurement that the motor time constant parameter is slighter smaller than expected. However, the difference is minor and the model prediction is consistent with the measured result. It was not possible to perform any full scale current tests due to the unavailability of rated cabling. Therefore, only scaled current tests were performed. The current limiting specification was tested using altered limits in two separate experiments. In both cases, a potentiometer was used as the throttle input and the motor of the motorcycle was used as the load. In the first experiment, the current limiting capability of the controller was tested. The

62 5.2 Performance Testing Fig. 5.2: Motor current limit test maximum current limit was set to 12A. The motor was powered through the converter and the duty cycle was set to 15%. The loading on the motor was increased steadily. At constant duty cycle and with no control, it is expected that the current would increase proportionally to the load as per equation 2.2. However, the controller successfully maintained the current below 12A by automatically decreasing the duty cycle without changing the input from the potentiometer. This is shown in Figure 5.2. The figure shows the current is steady for the first 5 seconds. The waveform then increases for 5 seconds and then remains steady for about 25 seconds. During those 25 seconds, the load was increasing but the controller was maintaining the current at a 12A mean. The test successfully proved that the controller is capable of limiting the motor current to a selectable value. Therefore the maximum current limit specification is met. In the second experiment, the time limiting capability of the controller was tested. The continuous current limit was set to 12A, the maximum current limit was set very high, the time limit was set to 10 seconds and the limiting time was also set to 10 seconds. The motor was powered through the converter and the duty cycle was set to 15%. It was

63 5.2 Performance Testing Fig. 5.3: Motor current time limit test expected that the controller would adjust the duty cycle to meet the continuous limit after the motor current was above the continuous limit for the specified time. The resultant current waveform from the test is shown in Figure 5.3. For the first 30 seconds from the beginning of the test (not shown in the figure), the motor was allowed to run with no load so that the motor current was below the continuous limit. The motor ran for the first 30 seconds of the test with no change in motor current. This means that as expected, the controller was not restricting any motor current below the continuous limit. At the 10 second mark shown on the figure, the load was increased and then held steady such that the motor current was above the continuous limit. It can be seen that the motor current suddenly drops to 12A, stays there for 10 seconds, and then returns to the previous value. This proves that the controller is capable of detecting over limit currents and imposing an altered current limit after a specified time. Therefore the continuous current specification is met. An attempt was made to determine the relationship between the duty cycle and the average motor current in steady state. The relationship to current depends on the supply

64 5.2 Performance Testing Average Motor Current [A] Duty Cycle [%] Fig. 5.4: Duty cycle vs motor current voltage and the motor speed in addition to the duty cycle. Therefore this was done by setting the duty cycle then applying load such that the motor speed was brought to a specific value. The plot in Figure 5.4 shows the relationship for a motor angular speed of 220RPM. This relationship will be different if the supply voltage changes or for a different motor angular speed however as long as those two parameters are held constant then the relationship should be linear (neglecting frictional forces). The range of motor current was kept small due to the low rating of the available cables, however it appears to be approximately linear for small duty cycles. Since air resistance was not a factor during this test, this is the expected result. Linear regression techniques produce a coefficient of determination equal to 98%. The coefficient of determination is a measure of how well a data set fits a regression line with the maximum possible score being 100%

65 6. Conclusions Chapter 6 Conclusions, Contributions, and Future Work 6.1 Conclusions During the course of this report, each stage of the project has been outlined. All of the choices behind the selection of topologies have been rationalized in Chapters 2 and 3. The model that was developed on the basis of these choices has been shown to meet the specifications outlined in the introduction. The evolution of the prototype has been demonstrated in Chapter 4 and the performance of said prototype has been evaluated in Chapter 5. Chapter 5 also demonstrated how the converter prototype meets the specifications. From all of this analysis, it can be concluded that the converter prototype and its model meet all the requirements that have been set. The converter model is fully capable of operating in both forward motoring and regenerative braking modes. The torque, and thereby speed, of the physical motorcycle can be controlled via the throttle. Current flowing through the motor can also be limited. If the current is above the continuous current threshold longer than a predetermined amount of time, the controller will alter the current thresholds to further

66 6.2 Contributions limit the current. All of the components are powered via the battery bank of the motorcycle, and can be packaged within the frame. The project as a whole can be considered a success. 6.2 Contributions This project has made several contributions to its field. It has verified the possibility of an all electric motorcycle. Although it could not be tested, regenerative braking was implemented as part of the design. Reusing energy and minimizing losses is important in electrical engineering. This project has also produced four better engineers who now have practical experience. 6.3 Future Work There are a number of things that can still be done to improve the converter as it stands now. While the converter itself should be capable of regenerative braking there are still additions that can be made to that feature. There is currently no user interface to trigger regenerative braking. Additionally there has been no effort to interface the converter with the battery management system. Creating that connection would allow that converter to optimize the charge current. The issue of proper heat sinking has not been evaluated thoroughly and additional studying may be required to ensure safe operation of the converter for extended periods of time

67 REFERENCES References [1] (2013, Sept. 15) All-electric vehicles (evs). [Online]. Available: fueleconomy.gov/feg/evtech.shtml#end-notes [Accessed: Februrary 26, 2014]. [2] (2008, Apr. 12) About ev s. [Online]. Available: aspx [Accessed: Februrary 26, 2014]. [3] S. Filizadeh, Dc to dc converters(dc choppers), Class Notes, [4] A. Hughes, Electric Motors and Drives, Fundamentals, Types and Applications, 3rd ed. Elsevier, [5] U. Annakkage, M. Annakkage, and K. Samarasekera, Developing a Controller for an Electric Motorcycle Driven by a Permanent Magnet Brushed DC Motor, University of Manitoba, [Online]. Available: Files/ Archive/2012/ Final Reports/G12%20Final%20Report% pdf [Accessed: September 1, 2013]. [6] APTM10AM02FG Datasheet, Microsemi, July [Online]. Available: APTM10AM02FG.html [Accessed: October 1, 2013]. [7] S. Mahbub. (2012, December 22) Low-side mosfet drive circuits and techniques 7 practical circuits. Blog. [Online]. Available: low-side-mosfet-drive-circuits-and 23.html [Accessed: Feb 2014]. [8] S. T. Mahbub. (2012, February 24) N-channel mosfet high-side drive: When, why and how. [Online]. Available: 02/n-channel-mosfet-high-side-drive-when.html [Accessed: Feb 2014]. [9] D. W. Hart, Power Electronics. McGraw-Hill, [10] Application Note AN-978, International Rectifier, March [Online]. Available: [Accessed: February 28, 2014]

68 REFERENCES [11] IRFD9220PbF Datasheet, International Rectifier. [Online]. Available: alldatasheet.com/datasheet-pdf/pdf/227587/irf/irfd9220pbf.html [Accessed: December 1, 2013]. [12] IRFD110PbF Datasheet, International Rectifier. [Online]. Available: [Accessed: December 1, 2013]. [13] DE03S/D Series, Delta Electronics. [Online]. Available: com/datasheet-pdf/view/546253/delta/de03s4815a.html [Accessed: January 3, 2014]. [14] Microchip, Cerebot MX7cK Board Reference Manual, Digilent, n. d. [15] V. P. Arafiles, Reducing high frequency conducted noise, in IEEE International Symposium on Electromagnetic Compatibility, vol. 1, August 2002, pp

69 A. Budget Appendix A Budget Table A.1 outlines the budget for the project. It is important to note that items 8 through 12 are included in the budget but have been provided by the University of Manitoba at no cost and are therefore not included in the sub-total. The total cost comes to $ Table A.1: Budget Part Number Supplier Quantity Unit Cost ($ CAN) Sub Total ($ CAN) Item Item Description 1 Half bridge APTM10AM02FG DigiKey module 2 N-channel IRFD110PBF-ND DigiKey MOSFET 3 P-channel MOSFET IRFD9110PBF- ND DigiKey Optoisolator HCPL2531-ND DigiKey Disconnect A27795CT-ND DigiKey connectors 6 DC-DC DE03S4815A DigiKey converter 7 DC-DC converter ND DigiKey Microchip PIC32 Cerebot MX7cK board PIC32MX795F512L Diligent Inc

70 9 Resistors, diodes, capacitors, etc. 10 Electric motorcycle 11 Battery pack set (batteries, charge, BMS, meters) 12 Drive system kit (motor, controller, throttle, contactor, wire, connectors, fuse) Total Cost various Native GPR-S Developers Package GBS 48V 60Ah Package EMC-R Drive Kit ME0708 ECE Tech Shop Electric Motorsport Electric Motorsport Electric Motorsport various

71 B. Hardware Components Appendix B Hardware Components The microcontroller used in this project is the Microchips Cerebot Mx7ck as seen in Figure B.1. It was available at no cost from the University of Manitoba. Fig. B.1: Cerebot Mx7ck microcontroller

72 The MOSFET power module that was chosen was the APTM10AM02FG as seen in Figure B.2. It cost $ as seen in the budget in Appendix A. A misconnection when testing damaged the initial module. A second module was purchased which is what caused the project to go over budget. Fig. B.2: MOSFET power module The soldering board can be seen in Figure B.3. The left side includes (from top to bottom): 48V-15V converter, optoisolator, resistors, two MOSFETs to invert one optoisolator output, 15V-5V isolated converter, and two MOSFETs to invert the second optoisolator output. The right side top socket is the circuitry for the top buffer while the bottom socket is the circuitry for the bottom buffer and a 15V-5V regulator to power the non-isolated components. The middle socket holds the gate drive IC. Decoupling capacitors can be seen throughout. Fig. B.3: Soldering board

73 Figure B.4 shows how the components were mounted on the aluminum block. Included in the picture: the MOSFET power module, soldering board, and relays. Fig. B.4: Aluminum heat sink with mounted components