Thesis_HovensPrevoo.

Thesis_HovensPrevoo.
Bachelor Graduation Project
Maximum Power Point Tracking
Topology, sensor and switch design
Authors
Y.K.L.M. Prevoo
M.G.P. Hovens
Supervisor
1323113
1361503
Dr. J. Popovic
Delft University of Technology
Faculty of EEMCS
June 26, 2012
Preface
This bachelor thesis is part of a report that consists of three different theses.
Together, they form a technical report for the bachelor graduation project
Maximum Power Point Tracking for Electric Automotive applications. This
project was proposed by the Nuon Solar Team for the bachelor Electrical
Engineering at the University of Technology Delft. Together with this report, a business plan was written for the course High Tech Startups together
with an ethical essay for the sub course Ethiek in bedrijf.
This project will extensively explain the technical challenges that come with
realizing a maximum power point tracker for solar vehicles. The titles of the
three theses in the project are:
• Maximum Power Point Tracking
Topology, sensor and switch design
• Converter Design for Maximum Power Point Tracking
• Maximum Power Point Tracking: Algorithm and Software
Development
We would like to thank:
dr. Jelena Popovic, supervisor of our bachelor graduation project, for her
technical as well as non-technical support.
Milos Acanski MSc, for his contributions to the project and help with the
simulation models.
Kasper Zwetsloot, for his assistance with building, testing and debugging
the prototypes.
The Nuon Solar Team for providing funding, data equipment and facilities
for testing.
Delft, June 26, 2012
Yves Prevoo
Max Hovens
i
Summary
The Nuna 6 is a solar racing vehicle that solely relies on solar energy from
the sun. Every year, the team seeks ways to improve the performance of
their car. One way to improve the performance is to maximize the power
output of the solar panels on top of the car. Maximizing the power output
can be done with a ’Maximum Power Point Tracker’. The aim of this thesis
was to develop an improved, distributed ’Maximum Power Point Tracking’system which optimizes the power efficiency of the solar panel array of the
Nuna 6 solar racing vehicle.
To prove that the proposed distributed topology is more power efficient when
compared to a central tracking topology, simulations of the total Nuna 6
electrical system were performed. Based on the simulation results, together
with Nuna 6 specifications, a DC-DC boost converter was designed. Validation of the design was done by simulation with the Nuna 6 model. After
validation, a breadboard proof-of-concept was built. The proof-of-concept
was successfully tested and compared with earlier simulations. The system
design process was evaluated and recommendations for further study and
future real-life implementations were formulated.
The simulation results prove that the proposed distributed tracking system
is as much as 40% more efficient in large insolation differences and 10%
in small insolation differences. The system excels when insolation differs,
however it is slightly less efficient when used with equal insolation on every
panel. The developed proof of concept demonstrates a functioning maximum
power point tracker and DC-DC boost converter. The power efficiency of
the boost converter was found to be between 95.8% and 98.5%, with an
efficiency of 97.1% for the rated input power of 200W .
ii
Contents
Preface
i
Summary
ii
1 Introduction
1
2 Problem Analysis
2.1 Problem definition . . . . .
2.2 Requirements . . . . . . . .
2.2.1 Required conditions
2.2.2 Prefered conditions .
2.2.3 Optional conditions
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
3
3
5
5
5
6
3 Related Research
3.1 Distributed MPPT approach .
3.2 Power converters . . . . . . . .
3.3 Sensors . . . . . . . . . . . . .
3.3.1 Current measurement .
3.3.2 Voltage measurement .
3.4 Switching elements and drivers
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
7
7
8
9
9
11
12
. . . . .
division
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
15
15
16
18
22
26
26
29
34
.
.
.
.
.
4 Topology & function formulation
4.1 System topology and simulation . . . .
4.1.1 System topology and functional
4.1.2 Simulation . . . . . . . . . . .
4.1.3 Comparison simulation . . . . .
4.2 Subsystem functions and concepts . .
4.2.1 Sensors . . . . . . . . . . . . .
4.2.2 Switching element . . . . . . .
4.2.3 Gate driver . . . . . . . . . . .
iii
CONTENTS
5 Subsystem design
5.1 Sensors . . . . . . . .
5.1.1 Current sensor
5.1.2 Voltage sensor
5.2 Switching element . .
5.3 Gate driver . . . . . .
6 Proof of concept
6.1 Implementation . . . .
6.2 Sensors . . . . . . . .
6.2.1 Current sensor
6.2.2 Voltage sensor
6.3 Switching element . .
6.4 Gate driver . . . . . .
iv
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
38
38
38
44
45
47
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
51
51
53
54
54
54
54
7 Evaluation
7.1 Efficiency Measurements
7.2 Gate driving . . . . . .
7.3 Sensors . . . . . . . . .
7.4 Control circuitry . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
57
57
59
63
64
8 Conclusion & Recommendations
65
Bibliography
70
Glossary
71
List of figures
74
List of tables
75
Appendix A - MOSFET Calculator
76
Appendix B - Model Plotter
82
Appendix C - Brief of requirements
84
Appendix D - IRFSL4620 Efficiency Measurements
88
Chapter 1
Introduction
Every other year, the Nuon Solar Team of the University of Technology
Delft participates in a solar-powered car race, called the world solar challenge. New developments in solar energy technology drive the Nuon solar
team to further improve the performance of their vehicle. However, harvesting solar energy comes with its own challenges. Solar panels are highly
dependent on the conditions they are placed in. A few characteristics that
influence the energy efficiency are temperature, humidity, incidence angle
of the sunbeams and shading [1]. These phenomena cause the point at
which the power produced is maximized (known as maximum power point,
or MPP) to change, which could lead to a decrease in energy production of
as much as 30% [2].
A major technological challenge is to track the maximum power point. This
is done with so called maximum power point trackers (MPPTs). Since a
few years, MPPTs are being developed to keep track of this power point.
However, due to high research costs involved and the relatively low market
adaptation, MPPTs are expensive and in most cases, not suited for application in a solar racing vehicle, because of the special needs of the racing
vehicle. The Nuon solar team uses two of these MPPTs in their cars. This
is a cost-effective decision, but has major disadvantages concerning energy
production, since multiple solar cells have to be connected to one MPPT
and therefore it can only track the worst power-point of the available cells.
To overcome this issue, the solar team needs to have an individual MPPT
per two panels to effectively maximize the energy production.
This thesis describes the process of the design and realization of the overall
topology of the MPPT, the switching device and the associated driver for
1
CHAPTER 1. INTRODUCTION
2
the MPPT module. A proof of concept demonstrates the viability of the
proposed system.
Composed of eight chapters in total, this thesis will first focus on the
analysing the problem and the overall topology of the MPPT. In chapter 2
the problem is defined, simulated and the boundary conditions are set by
use of the Nuna 6. Chapter 3 describes relevant research in this field and
shows the current state-of-art. Chapters 4 and 5 describe the design process
that was used to derive the optimal solution and justifies the concepts that
were chosen to be used in the MPPT. Chapter 6 will present the proof of
concept. In chapter 7 the design process will be evaluated, recommendations
in accordance with the evaluation are formulated in chapter 8.
Chapter 2
Problem Analysis
This chapter will consist of a thorough analysis of the problem in section
2.1, after which the requirements for the MPPT will be set in section 2.2.
2.1
Problem definition
To convert sunlight into energy, the Nuon Solar Team makes use of solar
panels that are 22.3% efficient. Although Australia is a sun-drenched country, clouds occasionally cover the sun from the solar panels, decreasing the
light intensity on the solar panels. Another effect that affects the performance of a solar panel is temperature.
In figure 2.1, it can be seen that the power density of a solar panel decreases
with increasing temperature. From figure 2.2 can be derived that the power
density significantly decreases when there is less light shining on the solar
panel.
The weather phenomenon is a hard phenomenon to predict. Since the power
directly relates to the performance of the Nuon Solar Team at the world solar
challenge, maximizing the output power of the solar panels is an important
goal for the Nuon Solar Team. In the current system, the entire PV area,
which is divided into 10 separate ’strings’ of 10 serially connected panels
each, is connected to two Maximum Power Point Trackers. Each maximum
power point tracker is only able to track one power point, and it will be
limited by the string that performs the worst. This means that when only
one of the 5 strings on one of the power point trackers receives less light, all
strings on that tracker are operated below their optimum power point, thus
wasting energy that could be used to win the race.
3
CHAPTER 2. PROBLEM ANALYSIS
4
In this thesis, the main question is:
How to efficiently maximize the output power of the solar panels for the
Nuna 6 racing vehicle, by using multiple Maximum Power Point Trackers?
Figure 2.1: Temperature dependancy of a solar panel, source: [3]
Figure 2.2: Light intensity dependancy of a solar panel, source: [3]
CHAPTER 2. PROBLEM ANALYSIS
2.2
5
Requirements
In this section, the requirements for the MPPT will be set. First the required
conditions are set. These conditions must be met in order to produce a viable
MPPT. The prefered conditions are the conditions that are not necessary
for the project, but these conditions are common in MPPTs and should be
added when possible. Optional conditions are the conditions for the end
product, which will be put on the market, these conditions have the least
priority. A more detailed overview of the requirements can be found in
appendix C.
2.2.1
Required conditions
• The MPPT must have an efficiency of minimal 95%.
• The MPPT has to be installed in fivefold in Nuna’s car, resulting
in a nominal input power between 20W and 100W and an absolute
maximum of 200W.
• The MPPT accepts input voltage between 60V and 120V and provides
an output voltage between 120V and 160V.
• The MPPT provides plug-and-play connectivity, thus every team member of Nuna can install it.
• The MPPT needs to work independent, but must be able to react to
externally given commands.
• The MPPT is maintenance-free during operation.
2.2.2
Prefered conditions
• The MPPT needs to communicate with the car using CAN-bus communication on a sond-only basis, providing additional information on
at least:
– Power in [W]
– Power out [W]
– Input Voltage [V]
– Output Voltage [V]
– Output Current [A]
CHAPTER 2. PROBLEM ANALYSIS
6
• An external reset switch is needed for resetting the MPPT during an
infinite loop or error.
• Extra safety systems, such as fuses to prevent overcharging or limit
short circuit damage are applied to the design.
• Input and output and other high-voltage connections need to be propperly shielded.
2.2.3
Optional conditions
• The condition of the MPPT can be controlled visually using multiple
leds.
• The design is robust, light and compact.
• The MPPT will work in the conditions given in Australia during the
World Solar Challange.
• All the used components meet the specifications as provided by Veolia
World Solar Challenge.
• The case and connections are dust-sealed and are shock and vibration
proof.
Chapter 3
Related Research
In this chapter the state-of-art is explored. First we will look at distributed
maximum power point tracking in section 3.1, then in section 3.2 a general
approach to power converters. Next, we will focus on the current and voltage sensors used in these systems in section 3.3, followed by the switching
element in section 3.4 and the driver in section 3.4.
The Nuna Solar Racing car operates 6m 2 of PV panels that output a certain
DC voltage. The engine and battery system of the car operate at another,
higher, voltage. Furthermore, to operate the PV panels efficiently, they need
to be kept at a specific voltage and current, using an algorithm. In order to
set the PV panels at their maximum power point and provide the engine and
batteries with the required voltage, a power conversion device is needed. To
minimize the effects of partial shading and differences in solar incident angle,
the total PV area is divided into several smaller segments. Each segment
will be equipped with its own tracking unit and power converter.
3.1
Distributed MPPT approach
A drawback of solar panel arrays is the loss of power due to partial shading
of the panels or mismatches in the PV modules themselves. Under ideal
conditions, these losses are minimal or absent [4]. In these ideal conditions,
the maximum power point for the whole array is the same. Under partial shading conditions there exist multiple power points, where the central
tracker can only track one. Not all panels are operated at maximum power,
and energy is lost. Distributed Maximum Power Point Tracking (DMPPT)
aims to overcome these issues by giving each panel its own tracker, ensuring
efficient operation of all panels [5]. In essence, going from a central Power
7
CHAPTER 3. RELATED RESEARCH
8
Point Tracker to DMPPT requires only the installation of more trackers on
the same array. Having multiple trackers can increase cost, but these trackers require a lower power rating and therefore can be cheaper per piece than
a large tracker. In most systems, the added cost of extra power electronics
is outweighed by the gain in energy output [6]
3.2
Power converters
Power converters exist in several forms, and are ultimately derived from a
general form of power supply. An overview of power supply technology is
given in figure 3.1.
Figure 3.1: Classification of power supply technologies, source: [7]
From figure 3.1 we see that two classes of regulators exist, linear and switching regulators. Although linear regulators have excellent characteristics
when it comes to bandwidth and noise, they can only step-down a voltage, [7], thus rendering them useless for our application where an increase
in voltage is needed. Switching regulators come in three forms: switchedcapacitor, resonant and Pulse Width Modulation (PWM ) regulators.
Switched capacitor regulators, or charge-pumps, can be used to step-up
a voltage without the need for inductors and with relative simple circuit
topologies [8]. These types of converters see much use in level translators in
digital communication buses and low power integrated step-up converters,
however they are not suited for medium or high-power applications due to
limitations on the capacitors [9, 10].
PWM regulators regulate their output voltage by hard switching a transistor between the on and off states, whereby the duty cycle of the switch
CHAPTER 3. RELATED RESEARCH
9
control signal determines the output voltage. A magnetic field in an inductive element is used to store and release energy, which also allows some
types of PWM regulators to step-up the input voltage [7]. More efficient
PWM converters are constructed by using soft-switching techniques, reducing the losses in the switch. Examples are converters that rely on resonance
to switch under Zero-Voltage or Zero-Current conditions or snubber circuits
that accomplish the same task. While these techniques can greatly improve
converter efficiency, they also increase the complexity [11, 12, 13, 14, 15, 16,
17, 18].
3.3
Sensors
In order to calculate the maximum power point, we need to measure the
current and voltage of the solar panel(s) that are connected to the maximum
power point tracker. To do this, we will use voltage- and current sensors.
In order to pick the right sensor, we will briefly describe the possibilities.
3.3.1
Current measurement
Current can be measured in three different ways: resistive, optically isolated
resistive and magnetically [19]. Each way of measuring current has its own
advantages and disadvantages. The three ways of measuring current are
compared in table 3.1 on a few key points.
Type
Resistive
Optically
isolated
resistive
Magnetic
Table 3.1: Common current
Current
Isolated
Accuracy
range
Very low - No
High
High
Medium - Yes
Low
High
Medium
Medium Very High
Yes
Medium
measuring methods [19]
AC Response
Non-Intrusive
Cost
Medium - High
No
Low
Low - Medium
No
Medium
Medium - High
Yes
High
Resistive Resistive measurement or shunt measurement is a way of measuring current with the use of Ohms law. Shunt resistors are low value
resistances that are placed in the path of the current. Because of this current a voltage is created across the terminals of the shunt, proportional to
CHAPTER 3. RELATED RESEARCH
10
the current value. Shunt resistors are low cost and easy to use. Often, a
differential amplifier amplifies the voltage across the shunt to increase the
signal-to-noise ratio. This setup can be found in figure 3.2. However, since
the shunt is in-circuit, the efficiency of the system is reduced [20].
Figure 3.2: Shunt current sensor, source: [20]
Optically isolated resistive Optically isolated resistive current measurement is used in high-current systems (typically >100kA). Due to complexity,
size and price, optical sensors are not really a feasible option for us, so it
will not be further investigated here.
Magnetically
Magnetic current measuring can be done several ways [21]:
• Current transformer
• Rogowski coil
• Search-coil magnetometer
• Flux-Gate magnetometer
• Magneto diode
• Hall sensor
The first five measurement techniques have in common that they can only
measure AC and therefore these forms are merely given as an example of
ways to measure current. Since our system only operates on DC, we will
not further discuss these measurement techniques.
CHAPTER 3. RELATED RESEARCH
11
Hall sensor The Hall sensor is a sensor that is based on the Hall effect.
Discovered by Edwin H.Hall in 1879, it is the effect that a voltage is induced
perpendicular to a current that is passing through a metal sheet, when a
magnetic field is passing through the plate. The voltage is a result of the
Lorentz force that acts on each individual charge particle that passes through
the sheet; this force is perpendicular to the current and magnetic field.
3.3.2
Voltage measurement
Like current measurement, voltage measurement can be done in several
ways. In table 3.2 several ways of measuring the voltage are given.
Table 3.2: Common voltage measuring methods [22]
Class
Operating principle
Subclass Application
field
Electromagnetic Interaction
between Moving
DC
magnets and magnetic magnets
fields
Electrodynamic Interaction
between DC and AC
currents
Electrostatic
Electrostatic
interac- DC and AC
tions
Thermal
Current thermal effects Direct
DC and AC
action
Induction
Magnetic induction
Analog
DC and AC
Electronic
Signal processing
Analog
DC and AC
and
Digital
As we can see from table 3.2 we can measure voltage in six different ways.
However, electromagnetic, electrodynamic, electrostatic, thermal and inductive measurement of voltage are no feasible options for us, since they require
mostly large components, moving parts or large currents, which in our case,
are hardly applicable.
As shown in table 3.2, there are two ways to represent a measure of voltage:
analogue and digital. In our MPPT, an analogue-to-digital converter (ADC)
will be used. However, an ADC only accepts voltages in the range of 0 − 5V .
Since our input voltage will be 60 − 120V we will have to scale the voltage,
CHAPTER 3. RELATED RESEARCH
12
prior to sending it into the ADC. Voltage scaling can be done in several ways:
voltage division using resistors and an opto-coupled floating transformer.
Voltage divider A voltage divider is a simple serial connection between
two resistors (shown in figure 3.3). Due to Kirchhoffs law, the voltage is
divided between the resistors. This setup is simple to make, but it will
reduce our resolution by a factor 2.
Figure 3.3: Voltage division
Optocoupled floating transformer An opto-coupled floating transformer
first transforms DC to AC with an opto-coupler. This is then fed into a transformer that will decrease the primary voltage to a smaller voltage, a little
bit larger than the desired voltage. At the secondary side of the transformer,
the voltage will be rectified and divided again to give the appropriate value.
The trick of this transformer is in the part that it uses a floating voltage
as reference. Because of this, it will only decrease the voltage range that is
informative to be measured.
3.4
Switching elements and drivers
Central to any of the mentioned power converters is the switching element.
For any converter the choice is between a Bipolar Junction Transistor (BJT )
Metal-Oxide-Semiconductor Field-Effect Transistor (MOSFET ), Insulated
Gate Bipolar Transistor (IGBT ) or a Thyristor. Thyristors are not considered as they are not used in practice for our type of application. The
choice between a BJT, MOSFET and IGBT remains and is made based
on rated output power, voltage, operating frequency and efficiency: IGBTs
are good devices for low-frequency, high-voltage, high-power applications [7]
where the limit between low and high-power applications is generally put
at 500 W [23]. Most IGBTs can be operated up to around 50 kHz, making
them unsuitable for high frequency converters [24, 25]. Figure 3.4 shows the
CHAPTER 3. RELATED RESEARCH
13
preferred operating regions of IGBT or MOSFET switches, not taking the
operating power into account.
Figure 3.4: Preferred operating regions for MOSFET and IGBT, source:
[23]
Figure 3.5 shows the operating regions of MOS devices versus BJT’s. Figures
3.4 and 3.5 clearly show the advantage of MOSFET over IBGT and BJT
devices, therefore, in this thesis we will focus on MOSFETS as switching
devices as they are preferred in the operating range of the converter as
described in Chapter 2.
Figure 3.5: Preferred operating regions for MOSFET and BJT, source: [26]
CHAPTER 3. RELATED RESEARCH
14
Gate driver A drive circuit is needed to quickly switch the switching device between its on and off states, minimizing the time spent in the device’s
active region with relative high dissipative losses [27, 28, 29]. Control for the
switch will be provided by a microcontroller operating at logic-level. The
drive circuit acts as an interface between the digital logic and the MOSFET, amplifying the control signal and providing isolation between the two
devices. The MOSFET is a voltage controlled device, requiring a constant
voltage on the gate to remain conductive, however the presence of parasitic
capacitances Cgd and Cgs greatly influences the behaviour of the MOSFET,
requiring significant current to rapidly charge/discharge, needed to achieve
the necessary fast switching times [30]. Currently, the use of dedicated gate
driver chips is widespread, as they are easily available, relatively cheap and
provide good performance [31, 29].
Chapter 4
Topology & function
formulation
In this chapter the development of a first-order quantitative model for the
electrical power system of the Nuna 6 Solar Racing Vehicle is described.
With this model, different topologies can be simulated in order to find the
most efficient solution regarding maximum energy output using a new type
of MPPT + DC/DC converter. Section 4.1 describes the Nuna 6 system
topology and describes the development of the model. In section 4.2, the
system topology from 4.1 is broken down into sub-systems and their functions are explained. Then the chapter continues describing the process with
which concept solutions for these functions are made and present equations
used to justify the choices made in the final design.
4.1
System topology and simulation
In this section, the overall topology and the division into subsystems and
further into functional elements will be made. A simple schematic overview
of the Nuna 6 is shown in figure 4.1. For reasons of simplicity, only relevant
modules are shown. The different components will briefly be discussed.
15
CHAPTER 4. TOPOLOGY & FUNCTION FORMULATION
4.1.1
16
System topology and functional division
Figure 4.1: Total system overview of the Nuna 6
Solar panel The solar panel is the source of energy for the Nuna 6. Characteristics of the solar panel directly influence the performance of the Nuna
6. For the simulation, the Simulink model for a standard solar cell is used,
which is defined according to the specifications of the solar cells of the Nuna.
Motor To drive, the Nuna uses an electric motor to propel the vehicle.
This motor consumes most of the power, however it can also feed power
back into the system using regenerative braking.
Battery As energy storage device, a battery is used. For simulation, a
standard battery model is used, defined according to the specifications of
the Nuna 6 battery.
Inverter The inverter, also called motor controller, converts the DC voltage from the battery pack into a three-phase AC voltage to drive the motor.
It also allows power to flow from the motor into the battery, to enable regenerative braking.
MPPT Finally, the MPPT is modelled using submodules. The submodules are defined by their function. First to measure the current and voltages, a sensor submodule is implemented. Secondly, a boost converter is
modelled, which will up convert the voltage. The controller is a submodule
that is in between the boost converter and the sensor submodules. The
CHAPTER 4. TOPOLOGY & FUNCTION FORMULATION
17
controller controls the conversion done in the boost converter, according to
the information it receives from the sensor submodule.
To correctly control the converter, the controller needs a reliable and accurate measurement of the voltage and current of the input and output of the
solar panel and MPPT, respectively. The sensor module can be divided in
two elements (shown in figure 4.2). The two elements represent an individual function, which has its own specifications. These specification will be
further developed in section 4.2.
Figure 4.2: Division of sensor submodule into functional elements
To convert the voltage from the level of the battery, a converter is needed
that can convert voltage up. To do this, a boost-converter is used. A
boost-converter consist of four elements that each have their own function,
as shown in figure 4.3. A switching element, switches an energy storage
element that in turn, stores the energy that is delivered by the solar panel.
Next a rectifying element rectifies the voltage that is created by the storage
element. However, since the characteristic is AC, it needs to be smoothed
to create a steady voltage at the output.
Figure 4.3: Division of boost converter submodule into functional elements
As last submodule of the MPPT, the controller is the brain of the MPPT.
CHAPTER 4. TOPOLOGY & FUNCTION FORMULATION
18
It is fed the information of the sensor submodule, with which it calculates
the maximum power point and then generates a pulse that controls the time
the switching element of the boost-converter will stay on. The first stage in
the information processing is the conversion from the analogue signal of the
sensors to a readable, digital signal. Through an algorithm, the maximum
power point is calculated. The last element, the PWM Generator then
creates the signal that controls the boost-converter element, the switch. A
schematic overview of the elements of the controller is given in figure 4.4.
Figure 4.4: Division of controller submodule into functional elements
4.1.2
Simulation
In order to test the MPPT before building it, a simulation is done of the
real world scenario. In this simulation, assumptions are made to decrease
the simulation time and increase the simplicity of the simulation. These assumptions are made because of the short time in which we have to perform
the simulation. First, a model of the current setup of the Nuna 6 is developed, secondly a model with the proposed, new MPPT setup is constructed.
Both models only differ in the number and type of MPPTs used and the
arrangement of the solar panels. The current setup of the Nuna 6 contains
DriveTek MPPTs and its model will be referred to as Nuna 6 + DriveTek.
The MPPT designed in this thesis is called SolarMax. The model of the
proposed system will therefore be referred to as Nuna 6 + SolarMax. Both
models will be compared to each other based on performance in their power
output.
Software For the simulations, Matlab and more specifically Simulink was
used. Simulink was chosen for its comprehensive component libraries, flexibility and familiarity.
Model In this section, a model for the current Nuna 6 is developed. This
model will be used as a reference for improvements. As we have seen in
CHAPTER 4. TOPOLOGY & FUNCTION FORMULATION
19
figure 4.1, there are four Simulink submodels to be developed: a model for
the solar panels, the MPPT, the battery and the motor. With each submodel
the assumptions and simplifications will be given.
Solar Panels Solar panels are an array of serially connected solar cells.
The Nuna solar panels have been modelled using the standard Simulink solar
cell model. The parameters to this model have been adjusted to match the
actual Nuna 6 cell parameters, which are SunPower C60 silicon cells. There
are 25 panels on the Nuna 6, comprised of 8 different dimensions. The most
common panel on the car, with 10 of the 25 panels being of this type, has
been modelled. It contains 84 cells. Figure 4.5 shows a part of this setup.
The only input to this system is the solar insolation in W/m2 .
Figure 4.5: Partial solar panel model
Assumptions & simplifications
• The Simulink solar cell model is correct and applicable.
• The SunPower C60 cells can be modelled by fitting the datasheet parameters to the Simulink model.
• The solar insolation is constant over the panel.
• Temperature variations are not considered.
DriveTek MPPT The Nuna 6 currently contains a commercially available DriveTek V4 DC-DC Converter and maximum power point tracker.
The components for this power converter have been reverse engineered from
the system and put into the model. It should be noted that the output capacitors are much larger than what would be expected based on a minimal
CHAPTER 4. TOPOLOGY & FUNCTION FORMULATION
20
ripple current [32]. This is possibly done in order to keep the output voltage as stable as possible. There could be some room for improvement here
in the DriveTek design. Originally, the switching device in the converter
was modelled using the SPICE model for the actual DriveTek MOSFET.
This however proved to be computationally complex and it was replaced by
a simple switch. The switching losses in the MOSFET are not taken into
account, however the average conduction losses are modelled.
Figure 4.6: DriveTek V4 Converter model
The algorithm that controls the PWM signal is proprietary and unknown.
A Perturb & Observe algorithm [33] will be used in the simulations. This is
a highly popular algorithm as it is relatively simple yet very efficient.
Assumptions & simplifications
• The switching losses are left out.
• The control algorithm is replaced by Perturb & Observe.
• The inductor series resistance is assumed to be 0.2 Ω.
SolarMax MPPT The SolarMax MPPT is the DC-DC converter and
maximum power point tracker discussed in this thesis. Its performance
will be compared to the current system. The DC-DC converter has been
modelled after design data provided by the authors and [32, 33]. As with
the DriveTek, the exact MOSFET model has been removed.
CHAPTER 4. TOPOLOGY & FUNCTION FORMULATION
21
Figure 4.7: SolarMax Converter model
In order to keep the simulations practical, and to be able to compare the
different systems, the same MPP algorithm as the DriveTek (Perturb & Observe) has been implemented. This way differences in the converter and the
distributed MPPT setup can be simulated, but not the actual differences
in performance of the different algorithms. For a detailed analysis of these
performances, we refer to [33].
Assumptions & simplifications
• The switching losses are left out.
• The control algorithm is by Perturb & Observe.
Battery The Nuna 6 battery system consists of 38 serially connected battery strings, with each string containing 12 parallel batteries. In total, 456
Panasonic NCR18650A Li-Ion batteries are used. The battery bank has
been modelled using Simulink’s Li-Ion battery model, where the parameters
have been matched to the Nuna 6 setup, using [34, 35].
Motor The Nuna 6 uses a 3-phase AC motor to propel the vehicle. To convert a battery pack DC voltage to the motor 3-phase AC, a motor controller
is used. Initially this system was modelled in a Simulink model containing the motor, modelled after the Nuna motor specifications and the motor
controller. This system performed somewhat erratically and proved to be
much too slow to simulate as part of a larger system. Therefore, the DC-AC
conversion and motor were greatly simplified. Data provided by the Nuna
team showed that under steady-state (cruise-speed) conditions, the average
power consumption of the motor was about 1.1kW . With a specified battery voltage of 160V , the steady-state DC resistance of the motor can be
CHAPTER 4. TOPOLOGY & FUNCTION FORMULATION
22
calculated. The motor submodel has thus been reduced to a single resistor,
greatly speeding up the simulations, but eliminating the possibility to simulate vehicle acceleration or regenerative braking scenarios.
Assumptions & simplifications
• DC-AC conversion and advanced motor model have been left out.
• The vehicle has a constant cruise speed on a level surface.
• No transient analysis possible.
• The DC resistance of the motor is 25Ω.
4.1.3
Comparison simulation
In order to compare the current system and the proposed new setup, a
comparison simulation has been done. The main problem of the current
setup is that there are only two DriveTek MPPTs in the vehicle. Half
of the solar panels are connected to one DriveTek MPPT. With different
insolation on parts of the car, each panel might have a different maximum
power point. Since each MPPT can only track one of these power points, the
total efficiency decreases. By using multiple SolarMax trackers we decrease
the amount of panels attached to each tracker and therefore can operate
more panels on the true maximum power point. This simulation is meant
to show the differences in power generation with the different setups.
Nuna 6 + DriveTek The Nuna 6 + DriveTek model consists of the Nuna
6 car elements, described in the previous paragraph, with two DriveTek
trackers each with half of the solar panels connected to them.
Nuna 6 + SolarMax The Nuna 6 + SolarMax model has the same Nuna
6 car model as Nuna 6 + DriveTek, only it has four SolarMax trackers, each
with 14 of the solar panels attached.
Scenarios Both models have been combined into one large model, and two
scenarios have been simulated to compare the performance of both Nuna 6
+ Drivetek and Nuna 6 + SolarMax. These models have been put together
in one Simulink file for easy comparison, since they share the same inputs.
Both models are run completely independently from each other within this
CHAPTER 4. TOPOLOGY & FUNCTION FORMULATION
23
simulation, they only share common insolation input signals. This model is
shown in figure 4.8.
Figure 4.8: The complete combined model
The first scenario has equal insolation of 1000W/m2 for all solar panels on
the car. The results can be seen in figure 4.9.
Figure 4.9: Output power for equal insolation of 1000W/m2
We see that in the case of equal insolation, the DriveTek system performs
CHAPTER 4. TOPOLOGY & FUNCTION FORMULATION
24
better than the SolarMax system. This is to be expected, as with equal insolation on all panels there is only one power point. The DriveTek system has
only two MPPTs and therefore has lower converter losses than the SolarMax
system, with its four MPPTs. We see that the SolarMax system tracks the
power point faster than the DriveTek, but the stable output power of the
DriveTek system is about 10% higher than that of SolarMax. Although the
higher speed of SolarMax could be an advantage, this case proves that for
equal insolation a central power point tracking solution would be best.
The second scenario has different insolation for half of the panels on the car.
Originally, the idea was to simulate the insolation as a differing function of
time for multiple different parts of the car, and then compare which system,
Nuna 6 + DriveTek or Nuna 6 + SolarMax would yield the most energy.
This would closely match the real-world scenario where partial shading (both
slow and fast, e.g. clouds and roadside trees) and differing incident angles
play an important role. However, it soon became apparent that this was
computationally too demanding, with simulations taking multiple days to
complete. To keep the simulation times manageable, the car has been split
into two different segments of multiple panels, each with its own constant
insolation. With the insolations set at 1000W/m2 and 2000W/m2 , we arrive
at the results in figure 4.10.
Figure 4.10: Output power for insolation 1000 and 2000W/m2
We see in this case that after the tracking system has stabilized, the output
power of the SolarMax system is roughly 40% higher in this simulation. It
can also be seen that the DriveTek system takes longer to track the maximum
power point. This could be due to the fact that there are more solar panels
connected to one tracker. Although the differences in favour of the SolarMax
CHAPTER 4. TOPOLOGY & FUNCTION FORMULATION
25
in this case are obvious, and the chosen insolation values are typical for race
conditions, situations in which the insolation on one half of the car is twice
as high as on the other half, might not be very common. Insolation values
differing by a factor two could be of importance in areas with lots of trees or
shrubs along the road. Under these conditions, the benefit of the SolarMax
system are clear.
Another simulation with smaller differences in insolations, set at 750W/m2
and 1000W/m2 respectively, has been done. The results are shown in figure
4.11.
Figure 4.11: Output power for insolation 750 and 1000W/m2
We see that the DriveTek system tracks the power point slowly and eventually settles at a total output power that is 10% lower than that of SolarMax.
These results prove the point that when the differences in insolation are less
extreme (compared to the case above, which had 1000 and 2000W/m2 ), the
gain in efficiency is also less with a distributed system. However, a 10% gain
in efficiency is very significant for the Nuna team and can easily mean the
difference between winning and losing a race. Also, the faster tracking of
the SolarMax system is significant in areas with quickly changing lighting
conditions.
Summary These simulations prove that the proposed distributed maximum power point tracking topology indeed is more efficient compared to
the current Nuna 6 + DriveTek setup. Larger differences in insolation on
different parts of the car lead to bigger differences between the Nuna 6 +
DriveTek and Nuna 6 + SolarMax systems. With the insolation on the car
being 1000W/m2 and 2000W/m2 (a difference of 100%), the gain in output
CHAPTER 4. TOPOLOGY & FUNCTION FORMULATION
26
power is as high as 40%. With the insolation at 750W/m2 and 1000W/m2
(a difference of 25%), the gain in power is 10%, which clearly is a significant
increase for a high performance racing vehicle such as the Nuna.
4.2
4.2.1
Subsystem functions and concepts
Sensors
In this section, a more detailed explanation of the current and voltage sensors functions and topologies will be given. In the first part, consideration
regarding the current sensor will be made. This will be followed by the
topologies of the voltage sensor.
Current sensor
In this section, multiple topologies of current sensing will be presented and
discussed.
Current sensor problem As can be seen from section 4.1, the central
brain of the MPPT will be a microcontroller. This controller only accepts
inputs as voltages, not as currents, therefore, directly inserting current into
a microcontroller is not an option. This presents our first constrained: the
current needs to be converted into a readable, quantity: a voltage. Another
restriction of the current sensor is that it should be as efficient as possible.
Methods As described in chapter 3, current sensing can be done in several ways. The first way to measure current is by using an indirect quantity
that is (ideally) linear related to the current.
An example of a linear relationship between current and another quantity is
Ohms law, which relates voltage to current through a resistor. This method
of current measuring satisfies the first requirement stated in the first paragraph, it converts the current into a voltage. Power dissipation in a resistor
is represented by the formula:
P = I2 ∗ R
(4.1)
This equation shows that the power dissipation is quadratic related to the
current and linear proportional to the resistance value. Since the input
current is fixed, the resistance value has to be picked very small. This,
however, presents another dilemma. Since the resistor will be picked very
CHAPTER 4. TOPOLOGY & FUNCTION FORMULATION
27
small, the voltage drop across the resistor will be very small. Since the
voltage has to be transported into the microcontroller, it is a very small
signal, which means that it is very susceptible to noise, which is hard to
measure with an ADC.
Signal-to-noise To overcome the issue of low signal-to-noise ratio, the
voltage can be amplified using an amplifier. Multiple scenarios can be evaluated to do this. However, there exist a multitude of off-the-shelf ICs that
are specifically designed to measure current, which operate on an external
shunt resistor and subsequently amplifies the voltage to logic level (5V ).
Selection of the implementations will be discussed in chapter 5.
Summary
sensor.
A summary of the function and requirements of the current
Function: measuring the input and output current of the MPPT and making this value available for a microcontroller.
Requirements:
• High accuracy
• Low power dissipation
• Translate to readable voltage for microcontroller (0 − 5V range)
Voltage sensor
In this paragraph, there will be made a general specification of the voltage
sensor and the functions it should accomplish.
Voltage sensing problem Measuring voltage is easier to accomplish than
measuring current, because it does not have the restriction of having to be
translated to a different quantity, as was the case with current measurement.
This greatly simplifies things. With voltage measurement there is, however,
another challenge that has to be overcome. Microcontrollers only accept
voltage inputs ranging from 0 to 5V . The output voltage levels of the MPPT
range from 60 to 120V . This makes a direct coupling of the solar panel to
the microcontroller impossible. To come up with a solution, we have to do
the inverse operation as we did with the current sensor; we have to divide
CHAPTER 4. TOPOLOGY & FUNCTION FORMULATION
28
the voltage. Voltage division can be done in two ways: DC-DC conversion
and resistor division.
Methods DC-DC conversion could be used to convert the 120V to a lower,
usable value. This would mean that a DC-DC converter has to be used, this
is impractical due to complexity and time issues and therefore not further
discussed.
Voltage divider As last option, the voltage divider remains. This simple
but practical solution has its advantages and drawbacks. The main advantage of a voltage divider is that it is very easy to implement, since only what
is needed are two resistors. The disadvantages of a resistor divider are that
it can be inefficient in terms of power consumption and even more disadvantageous is the reduction in resolution. This reduction of resolution occurs
because our voltage always remains between two values, 60V to 120V . This
means that the output of the voltage divider (as shown in figure 4.12) ranges
only from 2.5V 5V , which means that the resolution is reduced by a factor
of two, since the range between 0V 2.5V is never used.
Figure 4.12: Voltage divider
Summary
sensor.
A summary of the function and requirements of the voltage
Function: measuring the input and output voltage of the MPPT and making this value available for a microcontroller.
Requirements:
• High accuracy
• Low power dissipation
• Translate to readable voltage for microcontroller (0-5V range)
CHAPTER 4. TOPOLOGY & FUNCTION FORMULATION
4.2.2
29
Switching element
The switching element in DC-DC boost converter is preferably an ideal
switch that switches between different current paths in the converter, as
shown in figure 4.13.
Figure 4.13: Boost converter topology, source: [36]
As this ideal switch does not exists, a real replacement for it has to be found
that approximates the ideal conditions as close as possible. As we have seen
in chapter 3, the choice for a switching device was already narrowed down
by the available literature. A MOSFET is the device of choice because of its
superior characteristics in the operating range of our converter. The process
of designing, or selecting, a proper MOSFET is described below.
MOSFET Considerations When selecting a MOSFET, we first need to
look at the requirements the device has to meet in order to function properly.
Some theoretical background is given, and device selection is based on two
separate groups of requirement: Requirements that ensure proper operation
(Operating requirements) and requirements that reduce device losses as much
as possible (Performance requirements). The basic requirements can be
specified as a few fast rules that every device must absolutely meet: The
MOSFET needs to be able to handle the peak inductor current. The DrainSource voltage VDS needs to be in excess of the output voltage + diode
drop. The gate-source turn on voltage of the MOSFET, VGS needs to be
less than the input voltage (and never higher than the maximum allowed
CHAPTER 4. TOPOLOGY & FUNCTION FORMULATION
30
VGS as specified by the manufacturer), to ensure that the voltage applied
to the gate pin can actually turn on the MOSFET. Power dissipated in the
device must be within the specified region. In order to ensure safe operation
of the MOSFET, it needs to be operated within its specific safe operating
area (SOA). The SOA is determined by three quantities intrinsic to the type
of MOSFET used: Maximum drain current Id , breakdown voltage Vbr and
junction temperature Tj [37] . The above parameters represent the bare
minimum characteristics. However, to get a good design, we must ensure
that the losses in the device are as low as possible. The MOSFET switch
presents 2 losses in the circuit: switching losses and conduction losses. After
the losses have been calculated, the thermal properties of the FET have to
be taken into account, in order to ensure that the junction temperature does
not exceed the rated value for the calculated dissipated power.
Switching losses and switching speed The switching behaviour of the
MOSFET is largely affected by the various internal capacitors, shown in
Figure 4.14.
Figure 4.14: MOSFET with junction capacitance, source: [30]
The shown capacitances are non-linear parasitic capacitances influencing the
device’s operation. Specifically Cgd (also referred to as Miller capacitance,
since its variations are caused by the Miller effect [38] and Cgs are strongly
non-linear as a function of the input voltage, as shown in Figure 4.15.
CHAPTER 4. TOPOLOGY & FUNCTION FORMULATION
31
Figure 4.15: Variation of Cgd and Cgs as a function of VDS , source: [30]
When the MOSFET is switched, Cgd and Cgs must be charged or discharged
through the gate. This has strong implications on the choice for a proper
gate driver circuit, as will be explained in the paragraph about drivers [30].
Switching losses in MOSFETs result from current flowing through the MOSFET at the same time that a voltage is across the device’s terminals, during
the turn on and turn off times of the MOSFET [27]. Small turn on and
turn off times are thus required, not only to achieve the required switching
speed, dictated by the operating frequency fsw , but also to keep the losses
to a minimum. One way to achieve this is to ensure low capacitances, where
Cgd and Cgs are of particular importance [30]. The choice for a particular
MOSFET will partially be based on the values for the capacitances Cgd and
Cgs , however MOSFET manufacturers do not provide these values, but
rather specify them as:
Cgd = Crss
Cgs = Ciss − Crss
Cds = Coss − Crss
(4.2)
With Coss , Crss and Ciss device parameters, provided by the datasheet [30].
To incorporate these various effects in choosing the right MOSFET, the total
gate charge, QG , has become the figure of merit. QG makes for a handy,
one-figure first look when selecting devices. Care must however be taken not
to oversimplify things, since the test conditions provided by manufacturers
might deviate and assumptions surrounding gate charge, specifically varying
gate voltage, are only valid for specific conditions [31, 39]. Nevertheless, QG
is a good tool in helping engineers efficiently selecting the right MOSFET.
The following equations elaborate this: Charge is dictated by:
CHAPTER 4. TOPOLOGY & FUNCTION FORMULATION
Q=I ∗T
32
(4.3)
Since f = 1/T where T is the period time, we can write:
Q∗f =I
(4.4)
With (4.3) and (4.4) we can calculate the current needed to charge the gate
capacitance of the FET. Since losses are the product of voltage and current,
a high gate charge will lead to high dissipative losses in the driving circuit
(see also next paragraph).
A first-order approximation of power loss due to switching in the MOSFET is given by [30, 31]:
PSW M =
1
∗ VDS ∗ ID ∗ fsw ∗ (trise + tf all )
2
(4.5)
With ID drain current, fsw switching frequency and ton , tof f the on and off
times where finite values of ID and VDS co-exist, respectively.
Conduction losses Once the MOSFET has switched on, the MOSFET
presents a small dc resistance between its Drain and Source terminals. This
is the MOSFETs Drain-Source on resistance or RD S(on). Again, this needs
to be as low as possible. Instantaneous power dissipation in the ON-state
of the MOSFET is given by [27]
2
PCM (t) = ID
(t) ∗ RDS(on)
(4.6)
Integrating 4.6 over Tsw gives the average of the conduction losses [40]:
Z Ts w
1
2
PCM =
PCM (t) dt = RDS(on) ∗ IDrms
(4.7)
Tsw 0
With IDrms the rms value for the on state current, which for a boost converter topology is defined as [40]:
2
2
IDrms
= D ∗ Iin
(4.8)
We see that the conduction losses greatly depend on RDS(on) , which is
therefore an important parameter. MOSFET manufacturers reduce the onresistance RDS(on) of the MOSFET by constructing many parallel conduction paths between the drain and source. Thus, like connecting resistors
CHAPTER 4. TOPOLOGY & FUNCTION FORMULATION
33
in parallel, the RDS(on) comes down with more parallel paths. However, in
connecting drain-source paths in parallel, a negative effect is that the gatesource capacitance Cgs is also connected in parallel, so a low RDS(on) , and
hence low conduction loss, implies a higher gate source capacitance with accompanying switching losses. Thus the MOSFET that is chosen should be
a compromise between these two characteristics. In addition, high current
MOSFETs tend to come in much larger packages, so meeting the ideal of
low RDS(on) and low Cgs might violate a space requirement specification.
Another factor of importance is the fact that the RDS(on) significantly increases with higher junction temperatures, because of an increase of charge
carrier collisions, due to higher atom energies, and thus larger vibrations,
in the semiconductor lattice structure [27]. Typical MOSFETS have a temperature coefficient from 0.35% to 0.5% [41]. The RDS(on) at the operating
temperature for the pessimistic temperature coefficient of 0.5% is given by
[41]:
RDS(on)hot = RDS(on)spec ∗ (1 + 0.005 ∗ (TJ(hot) − Tspec ))
(4.9)
where Tspec is the temperature at which RDS(on)spec is specified. TJ(hot) is
the designer’s assumption of the MOSFET junction temperature.
Lastly, the RDS(on) of a MOSFET inherently increases with an increase
in breakdown voltage, because of the inverse relationships between breakdown voltage and RDS(on) due to epitaxial thickness and doping levels [30]
where low doping and high thickness represent a high breakdown voltage,
but poor RDS(on) and vice versa.
Leakage losses Next to both mentioned losses, there are also leakage
(blocking) losses. When the MOSFET is in the off-state, it carries a small
leakage current. As this current is in the µA range for a typical device [40, 7]
these losses are usually neglected.
Loss budget The complete loss in a MOSFET is the sum of conduction
and switching losses, neglecting the leakage losses and can now be computed
using equations 4.7 and 4.5:
PM
PM
= Pcond + PSW
2
= RDS(on)hot ∗ IDrms
+
1
2
∗ VDS ∗ IDrms ∗ fsw ∗ (ton + tof f )
(4.10)
CHAPTER 4. TOPOLOGY & FUNCTION FORMULATION
34
Thermal management The lost power from 4.10 is dissipated to heat
that increases the junction temperature TJ . For each MOSFET, the maximum junction temperature, TJmax , is specified. Care has to be taken to ensure that this temperature is not exceeded, for silicon devices, TJmax is usually between 150 ◦ C and 200 ◦ C [7]. The temperature rise of the MOSFET
junction relative to the ambient temperature, ∆TJA , is defined as [42, 7]:
∆TJA = θJA ∗ PD
(4.11)
with PD the dissipated power and θJA the thermal junction-to-ambient resistance. When a heatsink or other cooling element is applied, θJA is [7]:
θJA = θJC + θCS + θSA
(4.12)
where θJC is the manufacturer specified MOSFET junction-case thermal resistance, θCS the case-to-sink thermal resistance and θSA the sink-to-ambient
thermal resistance.
The maximum power dissipation for a given TJmax and TA is given by
[7]:
PDmax =
TJmax − TA
θJA
(4.13)
In order to calculate whether natural convection provides enough cooling, or
if a heatsink is required, we can use 4.13 to calculate if the required θJA can
be met by the device. PDmax must be calculated using 4.7 and 4.9, where
assumptions for TJhot have to be made.
Device selection The selection procedure for a particular device can now
be based on the previous theory. First, the device has to meet the Operating
requirements, and from the possible choices that remain, a trade-off between
the Performance requirements has to be made. The Rds(on) and Qg hereby
form the focus point, as they provide a relatively easy way to compare
devices and they characterize the conduction losses and the switching losses
in an obvious way. Using the design parameters, the dissipated power and
the junction temperature can be derived. Based on this, a suitable MOSFET
and possibly a heatsink can be selected.
4.2.3
Gate driver
The gate driver (also called switch driver or driver) provides the interface
between the control circuitry and the DC-DC converter. Its function is to
CHAPTER 4. TOPOLOGY & FUNCTION FORMULATION
35
ensure that the MOSFET is switched on and off fast enough based on the
digital control signal. The MOSFET and driver are closely related and in
their functionality strongly dependent on each other. The characteristics
of the driver have to be carefully matched to those of the device it drives.
Here we will present the procedure used to achieve this. As we have seen in
chapter 3, the ability of the driver to charge/discharge the gate capacitance
Cg is of utmost importance for achieving acceptable switching speed as well
as limiting the switching losses. The typical MOSFET gate charge QG , CG
and power required to charge this capacitance are related as [39, 29, 28]:
QG = CG ∗ VGS
(4.14)
1
2
∗ CG ∗ VGS
∗ fsw
(4.15)
2
First-order approximation for the power dissipated in the driver[39]:
PG =
2
PD = CG ∗ VGS
∗ fsw
(4.16)
The time required to charge CG , using a constant charge current, Icharge :
QG = Icharge ∗ tcharge
(4.17)
with tcharge the charge/discharge time, as dictated by the switching time.
This method is flawed however, as it does not take into account intrinsic
properties of the driver circuit and assumes a constant charging current,
Icharge , which is not the case. As we have seen in Figure 4.15, the capacitances change during the charge cycle. As VDS decreases, the capacitance
increases and the challenge for the driver circuit is to both charge Cgd and
Cgs while their capacitance increases rapidly. As a rule of thumb, it is advised to select a driver circuit with a peak current rating Ipeak of at least
twice the calculated Icharge . We will see some enhancements to this crude
method later in this chapter.
Switch driver for Boost Converter In the case of a simple boost converter as is described in this thesis, the switch element is directly tied to
ground, allowing for a relative simple low-side gate driver circuit, eliminating the need for bootstrapping/charge pumping or other level shifting
techniques associated with high-side driving [37]. Drivers can be designed
from the ground up, using discrete elements, or an off-the-shelf IC driver
can be chosen. IC drivers excel in convenience, compactness, low propagation delays and reliability, but lack the flexibility of a purpose-built circuit
CHAPTER 4. TOPOLOGY & FUNCTION FORMULATION
36
which exactly matches the requirements [28]. A typical MOSFET driver is
depicted in figure 4.16, showing a typical topology using a totem-pole output stage, using a P-MOSFET for charging and N-MOSFET for discharging
the gate connected to the output.
Figure 4.16: Microchip T442X 9A MOSFET Driver, source: [29]
In the transition state between charging and discharging, current flowing
through Q1 is divided between the gate of the output MOSFET and Q2,
leading to so-called shoot-through current, which in turn leads to dissipation
usually named cross-over energy. When the gate is being charged, the circuit
of figure 4.16 can be simplified to figure 4.17.
Figure 4.17: Microchip T442X charging external gate, source: [29]
This model allows us to revisit 4.16, making a more thorough analysis for
selecting the right driver possible, based on a time constant approach [29]:
CHAPTER 4. TOPOLOGY & FUNCTION FORMULATION
tcharge = (Rdriver + Rgate ) ∗ CG ∗ T C
37
(4.18)
Where Rdriver is the RDS(on) of the drivers output stage P-MOSFET, Rgate
any resistance between the driver output and the MOSFET gate and TC,
the number of required time constants, which will range between 1 (63%
charged) to 3 (95% charged), depending on the threshold voltage VT H of
the selected power MOSFET. It can be seen that through 4.3 VGS is taken
into account, making for a more accurate selection based not only on Ipeak ,
which is a rather simple figure used for easy comparison, but also on the
required Rdriver and VGS .
With this more advanced driver model the dissipated power loss in the driver
can be specified further compared to equation 4.16 [31]:
PD = PD(on) + PD(of f )
(1 − D) ∗ ROL ∗ VCC ∗ QG ∗ fsw
ROL + Rgate
D ∗ ROH ∗ VCC ∗ QG ∗ fsw
PD(on) =
ROH + Rgate
PD(of f ) =
(4.19)
(4.20a)
(4.20b)
Where:
D = Duty cycle
VCC = Driver supply voltage
ROH = Driver output resistance with output high.
ROL = Driver output resistance with output low.
Combining equations 4.18, 4.17 and 4.19 allows us to combine all important
properties of the MOSFET and driver and enables to choosing of a device
that meets the specifications on Icharge , Ipeak , Rdriver , tcharge and PD .
Chapter 5
Subsystem design
In this chapter the design choices are presented and explained, based on the
theoretical concepts from chapter 4. In a sense, this chapter describes the
translation between theory and the practical application in order to build a
functioning proof of concept. The designs are presented in the same order
as the subsystems were developed in chapter 4. In section 5.1 the sensor
modules will be designed, where after in sections 5.2 and 5.3, the switching
element and gate driver will be designed.
5.1
Sensors
In the previous chapter, the overall functions of every subsystem were defined
and the requirements for each subsystem were set. In this chapter, based
on the requirements and functions, design choices are made and possible
implementations are considered.
5.1.1
Current sensor
As stated in chapter 4 the current sensor requirements were:
• High accuracy
• Low power dissipation
• Translate to readable voltage for microcontroller (0 − 5V range)
In the previous chapter, considerations where made about certain topologies.
Here the topologies will be investigated further and the full implementation
will be realized.
38
CHAPTER 5. SUBSYSTEM DESIGN
39
To attain high accuracy current sensing with a shunt resistor, a high value
resistor would be most effective, since high value resistors create a high
voltage drop across the shunt resistor, which results in a high signal-to-noise
ratio. As we have seen from the previous chapter, this presents a dilemma
with the power dissipation. A trade-off will have to be made. In practice,
low shunt resistor values are used. The voltage across the shunt is amplified
by an amplifier, which is subsequently fed into the microcontroller.
Constraints The first design challenge is to choose for high-side or lowside injection. High-side injection puts the shunt resistor into the hot wire of
the system, whereas low-side injection puts the shunt into the ground wire
of the system. High-side injection poses the problem that the shunt resistor
and consequently the IC have to deal with large common-mode signals. On
the other hand, low-side injection creates an undesirable resistor into the
ground path, which creates a floating ground, which means that it needs a
common-mode input range of below zero.
Due to ground complications with low-side injection, high-side is preferred.
To implement the current sensor circuit, a decision can be made to implement and off-the-shelf IC or to implement a discrete operational amplifier.
Implementing an discrete operational amplifier has some major disadvantageous [43]:
• Low input resistance
• Common-Mode Rejection Ratio (CMRR) decreases very rapidly when
resistors are not matched very well, which requires high precision resistors.
The Common Mode Rejection Ratio (CMMR) can be expressed in the following equation:
Ad
CM RR = 20 ∗ log10 (
)
(5.1)
Acm
Where Ad is the differential gain factor and Acm is the common mode gain
factor, which is typically much smaller than the differential gain factor. For
CMRR applies: the higher, the better.
Another constraint that is placed on the IC selection is the voltage the
IC works on. In the brief of requirements in chapter 3 it is stated that the
maximum input voltage will be 120V maximum and the output voltage will
be 160V . So an IC is needed that can cope with these voltage levels.
CHAPTER 5. SUBSYSTEM DESIGN
Name
AD8212
INA210
HV7801
40
Table 5.1: Current Sense IC overview
Voltage range (V )
CMRR (dB)
+7 to > +500 with ext.
transistor
−26 to +26
−0.5 to +450
80 at 10kHz
High/Lowside
High-side
140
Not specified
Low-side
High-side
IC selection The ICs that can be used are shown in table 5.1.1. Since
high-side injection is preferred over low side injection, the INA210 is only
to be taken as back-up solution. The two remaining ICs are further investigated.
The HV7801 is a high-side injection current monitor IC, which transfers a
current measurement voltage value to its ground referenced output. This IC
has a very wide input voltage range, high accuracy, low power consumption,
ease of use, and low cost [44]. However, the CMRR is no-where specified,
so this option is secondary to the AD8212.
The AD8212 is a high voltage, current monitor. It accurately amplifies
a small differential input voltage in the presence of large common-mode
voltages up to > 500V with an external PNP transistor. The current output
of the device is proportional to the input differential voltage. The desired
gain can be set with an external resistor. The feature that makes the AD8212
very useful is high voltage operation, which is achieved by using an external
high voltage breakdown PNP transistor. In this configuration, the commonmode range of the AD8212 is equal to the breakdown of the external PNP
transistor. Therefore, operation at several hundred volts is easily achieved.
Because of the suitable characteristics, the AD8212 is a good choice; some
calculations need to be made to incorporate this device into the MPPT.
AD8212 A load current flowing through the external shunt resistor produces a voltage at the input terminals. As shown in figure 5.1, amplifier A1
responds by causing Q1 to conduct current through Resistor R1 to equalize
the potential at both the inverting and non-inverting inputs of A1.
The emitter-current through Q1 (Iout ) is proportional to the input voltage
(Vsense ) and therefore is proportional to the current (Iload ) through the shunt
(Rshunt ). By use of an external resistor the output current (Iout ) is converted
CHAPTER 5. SUBSYSTEM DESIGN
41
to a voltage. The output voltage value is dependent on the input/output
gain equation desired in the application.
Figure 5.1: AD8212 Topology, source: [45]
Shunt resistor
range [45]:
Calculating the resistor values for the required operating
gm = 1000 µA
V
Rout = 100kΩ for both in and output sensor, as given in [45].
Iout
Vsense
Vout
Vout
= gm ∗ Vsense
= Iload ∗ Rshunt
= Iout ∗ Rout
∗Rout
= Vsense
gm
Iload,in = 3, 33A at 60V .
Iload,out = 1, 66A at 120V .
Using equation 5.2, Rshunt can be determined.
(5.2)
CHAPTER 5. SUBSYSTEM DESIGN
Vout
Rshunt
=
=
Vsense ∗Rout
gm
Vout ∗gm
Iload ∗Rout
42
= 5V
(5.3)
Rshunt,in = 0, 015Ω
Rshunt,out = 0, 030Ω
Power rating calculation:
Pshunt
2
= Ishunt
∗ Rshunt
Pshunt,in = 3, 332 ∗ 0, 015
Pshunt,out = 1, 662 ∗ 0, 030
= 0.166W
= 0.083W
(5.4)
External transistor The AD8212 has a 5V regulator. This regulator
ensures that COM (which is the most negative of all the terminals) is always
5V less than the supply voltage (V+). Assuming a V+ of 120V maximum,
it follows that the voltage at COM is:
V+ − 5V = 115V
(5.5)
The base-emitter junction of Q2, in addition to the internal Vbe , makes the
collector of Q1 approximately equal to:
115V + 2 ∗ Vbe(Q2) = 115V + 1.2V = 116.2V
(5.6)
This voltage appears across Q2. The voltage across Q1 then is:
120V − 116.2V = 3.8V
(5.7)
This way, Q2 resists 116.2V and Q1 only 3.8V , which is below the breakdown
voltages for each transistor.
Bias resistor Operating the AD8212 with an external transistor, means
that Ibias increases as the input V+ increases. To chose the correct value for
Rbias , figure 5.2 is used.
CHAPTER 5. SUBSYSTEM DESIGN
43
Figure 5.2: AD8212 Bias resistor chart, source: [45]
Since the maximum input voltage will be 120V , the resistor value for Rbias,in
should be about 130kΩ. The maximum output voltage is 160V which means
that we need a Rbias,out of 170kΩ.
Ibias,in,max
=
120V −5V
Rbias
=
120−5
130kΩ
= 884µA
Ibias,in,min
=
60V −5V
Rbias
=
60−5
130kΩ
= 423µA
Ibias,out,max =
160V −5V
Rbias
=
160−5
170kΩ
= 912µA
Ibias,out,min
120V −5V
Rbias
=
120−5
170kΩ
= 676µA
(5.8)
=
From [45], it is recommended to maintain Ibias between 200µA and 1mA to
ensure that the circuit operates as expected. With a Rbias of 130kΩ, the
operation range is 423 − 884µA for the input sensor and 676 − 912µA for the
output sensor. These values are well in the specified recommended range.
Shown in figure 5.1, Q2 can be a FET or a bipolar PNP transistor. The
PNP transistor is less expensive, however Iout is reduced by the current lost
through the base of the Q2. This will lead to an output error in the voltage
over Rout . Fortunately, the AD8212 incorporates a current compensation
circuit, which compensates for the loss of current in the base of the PNP
transistor. The choice for a suitable PNP transistor is based on a high
breakdown voltage, Vceo , (> 120V ). We chose the BF423ZL1G due to its
CHAPTER 5. SUBSYSTEM DESIGN
44
high Vceo , its low price and high availability.
5.1.2
Voltage sensor
As stated in chapter 4 the current sensor requirements were:
• High accuracy
• Low power dissipation
• Translate to readable voltage for microcontroller (0 − 5V range)
A decision was made in chapter 4 to incorporate an voltage divider because
of ease of design. However, it came with the disadvantage of reducing the
resolution by a factor 2.
There are a few design challenges when designing a voltage divider. As
stated above power consumption must be reduced to a minimum, the resistors used should be linear over a sufficient voltage and current range, should
be linear with increasing temperature and should be rated higher than the
maximum power level, for safety reasons.
Resistor values Calculating the resistor values of the voltage divider, a
constraint on the power dissipation has to be made. The power dissipated
by a resistor is given by:
U2
(5.9)
R
From this equation, it is apparent that higher resistor values dissipate less
power. As acceptable value, 0.05% of the total power was chosen. This is
about 0.1W .
As stated in chapter 4 the maximum output voltage should be 5V , because of the microcontroller. This means that we have to scale down the
voltages of the input and output of the MPPT, by a factor:
Pdiss =
cu,in
cu,out
23 ∗ R1
31 ∗ R3
1
= (R1R+R
2)
3
= (R3R+R
4)
= R2
= R4
=
=
5
120
5
160
(5.10)
With the power requirement stated as above (0, 1W ), we can calculate the
resistor values.
CHAPTER 5. SUBSYSTEM DESIGN
Pdiss,in
=
Pdisso ut =
45
U2
(R1 +R2 )
U2
(R3 +R4 )
R1 = 6kΩ R2 = 138kΩ
R3 = 8kΩ R4 = 248kΩ
(5.11)
(5.12)
Power rating These resistors should all have a power rating of > 0.1W ,
with the smallest possible tolerance. For the resistors, the choice was made
for 0, 25W resistors with a tolerance of 0, 1%.
Since the microcontroller is very susceptible to higher voltages, a decision
was made to put Zener diodes across R1 and R3 with a breakdown voltage
of 5V . If the input or output voltage increase to higher values than 120V
and 160V respectively, the total voltage will fall across resistors R2 and R4 .
This has consequences for the power rating of R2 and R4 . When choosing for
0, 25W , which have good tolerance factors, the maximum allowable voltage
is:
p
Uin,max = pPmax,R2 ∗ R2 = 185V
(5.13)
Uout,max =
Pmax,R4 ∗ R4 = 248V
This means that for the input voltage a 50% of increase above the maximum
voltage is survivable. For the output voltage this is 55%.
5.2
Switching element
With the theory developed in the previous chapter a well founded decision
for the switching element can now be made. First, the operating requirements are set, secondly the performance requirements and lastly the practical considerations in selecting a device that meets the requirements are
discussed.
Operating requirements These parameters, summarized in table 5.2,
represent the absolute minimum the device must cope with.
• The minimum breakdown voltage VDS(br) must be at least equal to the
maximum output voltage, VO . This value is derived from the Nuon
Solar Team specifications.
CHAPTER 5. SUBSYSTEM DESIGN
46
• The continuous drain current ID must be at least equal to the peak
switch current Isw(peak) , calculated in cooperation with the other design teams [33, 32].
• The total turn-on time tton and turn-off time tof f determine the maximum switching frequency. With the switching frequency set at 100
kHz, the total cycle time of the MOSFET is maximum 10 µs. We will
see that this is not a limitation, as all devices can be operated at much
higher frequencies.
Table 5.2: MOSFET Operating requirements
Parameter
Minimum
Typical
Drain-Source breakdown voltage VDS(br) 160 V
200 V
Continuous Drain Current ID
3.9 A
5A
Total cycle time tcycle
no limit
-
Maximum
no limit
no limit
10 µs
Performance requirements In the design process we saw that the total
losses in a MOSFET depend on a trade off between several parameters. The
RDS(on)hot , QG , and thermal properties are of most importance.
Device selection All components were selected from Farnell, on request
from the Nuon Solar Team. Using their parametric search function, all
components that do not match the Operating requirements were filtered
out. Also, all components not in stock or in surface-mount packages were
not considered, due to time limitations. The remaining components need
to be selected on their Performance requirements. For the selection they
are judged based on a RDS(on) , QG , thermal and dissipation properties. To
speed up this process, a Matlab program was developed. The code to this
script can be found in Appendix A. This program contains all variables relevant to our boost-converter system. It reads the MOSFET parameters and
calculates all the equations presented in chapter 4. Some devices failed to
meet the thermal requirements, such as operating in a 60 ◦ C ambient temperature, where others presented too high RDS(on) etcetera. From all possible
options, three devices came out as best options, they are summarized in 5.3.
CHAPTER 5. SUBSYSTEM DESIGN
Table 5.3: MOSFET Selection [46, 47, 48]
No heatsinks, TA = 60◦ C
Device
RDS(on)
QG (nC) θJA
Pcond
PSW
(mΩ)
( ◦ C/W ) (W )
(W )
STF20NF20 100
28
62.5
0.84
1.21
IRFSL4620
63.7
25
40
0.78
0.64
IRFU220N
600
15
110
7.25
0.48
◦
◦
With heatsink, TA = 60 C, θHA = 8.97( C/W )
Device
RDS(on)
QG (nC) θJA
Pcond
PSW
(mΩ)
( ◦ C/W ) (W )
(W )
STF20NF20 100
28
15.0
0.84
0.90
IRFSL4620
63.7
25
11
0.55
0.78
IRFU220N
600
15
13.5
6.48
0.48
47
PDiss
(W )
2.10
1.42
7.73
∆TJA
(◦ C)
> 115
56.9
> 115
PDiss
(W )
1.74
1.33
6.93
∆TJA
(◦ C)
26.1
14.6
76.3
Table 5.3 has been calculated using worst-case scenario’s and with standard
heatsinks. Greyed out cells indicate that the TJ would rise above TJ(max) ,
destroying the device. We see that heatsinking of the MOSFETS will be
necessary. This keeps the losses lower, a simple aluminium heatsink, θHA =
8.97( ◦ C/W ), has been chosen that fits the MOSFET packages. Based on
these calculations, the device with the lowest losses, PDiss could have been
chosen. Because the MOSFET needs to be carefully matched to the driver,
the final choice can only be made after the losses in the driver have been
evaluated.
5.3
Gate driver
With the choice for the Power MOSFET narrowed down to the three options
in table 5.3, the driver can be chosen.
Gate-Source voltage The first decision that has to be made is the gate
drive voltage, VGS . This is of importance as equation 4.16 shows that the
2 . On the MOSFET side, a
dissipated power in the driver increases with VGS
higher VGS leads to a lower RDS(on) and thus lower losses. However, it also
creates a higher QG [49], thus leading to even higher driver losses. Lastly,
a low VGS limits the maximum drain current ID . The choice for VGS is
therefore a typical engineering trade-off. For all the MOSFETs in table 5.3,
a VGS above 8V does not limit the ID . With a minimum of 8V there are also
no problems with the turn-on threshold voltage VT H . For these MOSFETs
CHAPTER 5. SUBSYSTEM DESIGN
48
the RDS(on) does not decrease significantly for a VGS above roughly 10V
[49, 50]. Unfortunately the manufacturers do not provide exact figures for
this. Summarized in 5.4 are these figures for the different devices. For
calculation of PDiss(driver) , equation 4.16 has been used.
Table 5.4: VGS Selection [46, 47, 48]
fsw = 100kHz, TJ = 25 ◦ C
VGS (V ) RDS(on) (mΩ) QG (nC)
PDiss(driver) (W )
STF20NF20
8
n/a
10
100
12
n/a
IRFSL4620
8
n/a
10
63.7
12
n/a
IRFU220N
8
n/a
10
600
12
n/a
25
28
32.5
0.020
0.028
0.039
18
23
27.5
0.014
0.023
0.033
10
15
17
0.010
0.015
0.021
Based on these figures, a VGS of 10V has been chosen, as a trade-off between
losses in the driver and the MOSFET.
Ipeak and Rdriver As we have seen in chapter 4, the process of driver
selection continues with the peak driver current. With equation 4.17 the
assumed constant charge current is calculated for the devices. The gate
1
charge time, tcharge is set at 100
of a single period of the switching frequency
[50]:
tcharge =
1
1
∗
100 fsw
(5.14)
Which, for fsw = 100kHz is 100ns. The rule of thumb is to choose a
driver with at least twice the calculated Icharge as its peak value. Table 5.5
summarizes the results.
CHAPTER 5. SUBSYSTEM DESIGN
Table 5.5: Driver peak current selection
fsw = 100kHz, tcharge = 100ns, VGS = 10V
Device
QG (nC)
Icharge (A)
STF20NF20 28
0.28
IRFSL4620
23
0.23
IRFU220N
15
0.15
49
[46, 47, 48]
Ipeak (A)
0.56
0.46
0.30
The second requirement on the driver was based on the maximum driver
output resistance, Rdriver (equation 4.18). This method requires a time
constant T C to be set. With a VGS of 10V , a T C of 1 means that the gate
charge would be charged up to 6.3V . All MOSFETs are already fully turned
on at this point, but for a safety margin a T C of 2 has been chosen.
Table 5.6: Driver Rdriver selection [46, 47, 48]
fsw = 100kHz, tcharge = 100ns, VGS = 10V , T C = 2
Device
CG (nF )
Rgate (Ω)
Rdriver(max) (Ω)
STF20NF20
IRFSL4620
IRFU220N
2.8
2.3
1.5
n/a
2.6
n/a
15.1
19.1
30.7
Table 5.6 shows the results. For the STF20NF20 and IRFU220N, no Rgate
is specified. The results, shown in gray, are based on an assumed value of
Rgate , equal to that of the IRFSL4620. We see that the values are all quite
high (above 15Ω), there will be no problems meeting this in a driver.
Like with the MOSFET, the Farnell database was used to select useful
devices, based on rated output current, output resistance, availability and
IC package. Based on these results a MIC4420 driver has been chosen. It
provides a much higher current output than needed (6A peak), has low
output resistance and drives the MOSFET gate to within a few mV of its
VCC . This means that a 10V supply to the MIC4420 is enough to provide
the MOSFET with a VGS of 10V . The calculated power dissipation in
combination with each device is given in table 5.7. Equation 4.19 was used,
which closely matches the values produced in table 5.4, especially at the
higher TA . Note that the output resistance is strongly dependent on the
temperature. Worst case values have been given as well.
CHAPTER 5. SUBSYSTEM DESIGN
50
Table 5.7: Driver and MOSFET combination: Driver selection [46, 47, 48,
51]
fsw = 100kHz, VCC = 10V , VGS = 10V
MIC4420, ROH = 2.3 (Ω), ROL = 3.0 (Ω), TA = 25 ◦ C
Device
QG (nC)
PDiss(driver) (W )
STF20NF20 28
0.022
IRFSL4620
23
0.018
IRFU220N
15
0.012
MIC4420, ROH = 5.0 (Ω), ROL = 5.0 (Ω), TA = 125 ◦ C
Device
QG (nC)
PDiss(driver) (W )
STF20NF20
IRFSL4620
IRFU220N
28
23
15
0.028
0.023
0.015
Summary Choices for the MOSFET and driver have been presented and
justified. We have seen that relative to the MOSFET, the power losses in the
driver are small. For the combination of MOSFET and driver, typically only
about 1.5% of the energy is dissipated in the driver. With the remaining
98.5% being dissipated in the MOSFET, it can easily be justified to choose
the MOSFET with the lowest losses. From table 5.3 the lowest loss device
is the IRFSL4620. In combination with the MIC4420 driver, the total losses
amount to 1.36W . These are the devices that will be implemented in the
prototype.
Chapter 6
Proof of concept
In this chapter, the proof of concept and the practical implementation of
the total MPPT system. First, in section 6.1 an overview will be given
how the proof of concept was built and what the overall complications were.
Secondly, in section 6.2, 6.3, 6.4 a more specific description of the individual
subsystems is given.
6.1
Implementation
To test the proof of concept, a breadboard was used for the sensors, gate
driver and microcontroller. The reason for this is that it is easy to use
and changes can be made very easily. However, there is a drawback: breadboards can only conduct 1A of current, which makes them unsuitable for the
boost-converter part, which is rated up to 3.33A. For the boost-converter, a
stripboard (Veroboard) was used because of the higher conduction capability. Unfortunately, this requires soldering which may cost some more time
when changes need to be made.
In figure 6.1, the boost-converter, voltage and current sensor and the gate
driver are shown. In table 6.2 the numbers are matched with their corresponding elements.
51
CHAPTER 6. PROOF OF CONCEPT
52
Figure 6.1: Proof of concept, practical implementation
Table 6.1:
#
1
2
3
4
5
6
7
8
9
10
11
12
Practical implementation
Element
Inductor
Input capacitor
Output capacitor
MOSFET with heatsink
Output shunt resistor
Input shunt resistor
Diode
Gate driver
Ouput voltage sensor
Input voltage sensor
Input current sensor
Output current sensor
overview legend
Value
112µH
6.8µF
6.8µF
0.03Ω
0.015Ω
MIC4420
AD8212
AD8212
CHAPTER 6. PROOF OF CONCEPT
6.2
53
Sensors
In this section, the practical sensor implementation will be discussed. First
we will look at the current sensor and the difficulties implementing it, followed by the implementation of the voltage sensor.
Figure 6.2: Practical implementation of the sensor modules
#
IC1
IC2
T1,T2
B1,B2
O1, O2
R11, R12, R2
R3, R4
Z1, Z2
Table 6.2: Sensor implementation legend
Element
Value
Input current monitor
AD8212
Output current monitor
AD8212
PNP Transistor
BF423 / BC178
Bias resistors
In: 30kΩ, Out: 170kΩ
Output resistors
In: 26.1kΩ, Out: 33kΩ
Input voltage sensor resistors
R11: 137kΩ, R12: 1kΩ, R2: 6kΩ
Output voltage sensor resistors R3: 248kΩ, R4: 8kΩ
Zener diodes
1N4733A
CHAPTER 6. PROOF OF CONCEPT
6.2.1
54
Current sensor
As described in chapter 5, for the current sensor the IC AD8212 was chosen for its accuracy and ease of implementation. The implementation of
the current sensor using the AD8212 IC was straightforward. Following the
scheme in chapter 5, the circuit was connected. A picture is shown in figure
6.2. The AD8212 is only available in MSOP (Minimal Small Outline Package), which cannot be inserted into the breadboard directly. The IC was
soldered to an DIP (Dual In-line Package) which can easily be inserted into
the breadboard.
6.2.2
Voltage sensor
In chapter 5, a design implementation for the voltage sensor was chosen.
Due to simplicity, accuracy and ease of use, the voltage divider was chosen
as most suitable option. The practical implementation is shown in figure 6.2.
The design is not comprehensive, just two resistors in series, with a zener
diode parallel to R2 and R4 for the input and output sensors, respectively.
6.3
Switching element
The IRFSL4620 MOSFET was chosen as the preferred device, however in order to compare its performance, the two other MOSFETS that were considered have been tested as well. This ’luxury’ is possible because of the relative
low cost of MOSFETS when compared to other components.The MOSFET
is not mounted directly to the prototype board, but rather mounted in a
socket. This allows for easy replacement in case of a failure, but also allows
swapping one type of MOSFET with another quickly. Each MOSFET has
been equipped with a clip-on heatsink, selected in chapter 6.
6.4
Gate driver
The gate driver MIC4420 was implemented in the prototype. figure 6.3
shows the setup, table 6.3 shows the numbered parts.
CHAPTER 6. PROOF OF CONCEPT
55
Figure 6.3: Implementation of the gate driver
#
1
2
3
4
Table 6.3: Driver overview legend
Element
Value
MOSFET gate connection Supply voltage
0 − 12V
µC PWM signal
Decoupling capacitor
100nF
After installation of the driver, testing could begin. It soon became apparent that there were some problems with the input signal from the microcontroller, which was corrupted with noise due to the length of the cable.
Re-routing the cable solved this. Next, measurements on the output revealed large oscillations on the switching waveforms. It is suspected that
this is due to interactions between the MOSFET gate capacitance CG , the
driver capacitance CD and the drive wire inductance Lw . Decreasing the
drive wire length was not an option since the MOSFET could not be placed
closer to the driver. Placing the driver on the stripboard close to the driver
was problematic, since a MOSFET failure also destroys the driver. This
would mean that each time a MOSFET fails a lot of soldering has to be
done on the driver and its peripherals. In order to damp the oscillations,
a small external resistor of 5Ω was installed between the driver output and
the MOSFET gate. A disadvantage of this resistor is that it creates a RC
time, which causes the current flowing into the gate to be delayed, which
CHAPTER 6. PROOF OF CONCEPT
56
will cause the MOSFET to switch slower. Figure 6.4 shows the results. The
white waveform (snapshot) shows the switching waveform without the external resistor, the coloured lines represent the situation with the 5Ω resistor.
Figure 6.4: Gate drive and drain-source voltage waveform with STF20NF20
with (coloured) and without (white) external 5Ω resistor.
We see that the oscillations in VGS have been damped to a large extent,
where also the VDS reaches its final value quicker and is generally much
more stable.
Chapter 7
Evaluation
In this chapter, the proof of concept, described in chapter 6 will be evaluated.
Evaluation is done by use of measurements. In section 7.1, the efficiency of
the MPPT with different MOSFETs is evaluated. Followed by an evaluation
of the gate switching behaviour in section 7.2. Lastly the power consumption
of the control circuitry will briefly be discussed in section 7.4. This chapter
is ended with a summary of the evaluations.
7.1
Efficiency Measurements
The efficiency measurements where done with the three selected MOSFETs
of chapter 5. For each MOSFET, measurements were done at different duty
cycles of the gate driver and different voltages at the input. In figures 7.1,
7.2, ??, for the IRFSL4620, STF20NF20 and IRFU220NBPF, respectively,
the efficiency measured at different values of the duty cycle are plotted
against the input power.
During testing, the input voltage was varied between 50V and 120V . First,
we used a 0 − 70V power supply with a maximum current of 22A. In order
to take measurements above 70V a 0 − 600V supply that is current limited
to 2.6A, was used. Due to the 2.6A input current limit, some measurements
in the 70 − 120V range could not be performed. All measurement data from
the IRFSL4620 can be found in Appendix D.
IRFSL4620 Figure 7.1 shows that the efficiency improves at higher input
power. Also, the efficiency seems to be linear increasing with increasing
input power. The 40% duty-cycle measurement seems to diverge a bit from
its linear behaviour, this is probably due to inaccurate measurements. This
57
CHAPTER 7. EVALUATION
58
is in-line with the expectations.
Figure 7.1: Efficiency vs. Input power at different duty-cycles IRFSL4620,
100Ω load
STF20NF20 In figure 7.2, the efficiency of the second MOSFET is plotted
against the input power. What can be seen is that the efficiency is again
linear with the input power, except for the case of 20% duty-cycle, where it
seems to diverge. The overall efficiency of the STF20NF20 is lower than the
IRFSL4620 until about 100W where it is the same. Again, this is in-line
with our expectations.
Figure 7.2: Efficiency vs. Input power at different duty-cycles STF20NF20,
100Ω load
CHAPTER 7. EVALUATION
59
IRFU220NBPF The efficiency of the IRFU220NBPF seems linear with
the input power. However, at a duty cycle > 50% the efficiency seems to
decrease and the MOSFET fails. During testing, two MOSFETs of this type
were destroyed. This was somewhat unexpected. Expectations where that
the MOSFET would last longer.
7.2
Gate driving
This section will describe the results that were obtained by measuring the
gate voltage Vgs (shown in yellow) and the drain-to-source voltage Vds (shown
in green). The measurements that are shown in this section are all done
with an external Rgate resistor, which had a value of 5Ω. First the results of
the IRFSL4620 will be presented, followed by the STF20NF20 and last the
IRFU220NPBF.
IRFSL4620 In figure 7.3, the Vgs and the Vds are shown. Vgs the voltage
1
of the gate driver increases to 10V in about 50
ns and then drops, because
of the gate capacitance (Cg ). This oscillation is constant with a period of
2,5
50 ns. When Vgs reaches its third peak, Vds drops rapidly in about 20ns,
where after it remains fairly constant at 0V . Vgs keeps oscillating heavily
from 5 − 17V at first but damps after 200ns to about 9 − 11V . Table 7.1
shows the measured values and the specified datasheet values. The datasheet
values have been specified at different parameter (VDS , ID , Rgate ) values,
and this table allows us to compare the timing delays of the MOSFET under
different operating conditions.
Figure 7.3: Turn-on of IRFSL4620, Vgs = 10V , Vin = 60V , Rgate = 5Ω
CHAPTER 7. EVALUATION
Table 7.1: Measured quantities
Parameter
Measured (ns)
Turn on delay time 25
Rise time
15
Turn off delay time 40
Fall time
10
60
IRFSL4620
Datasheet (ns) [47]
13, 4
22, 4
25, 4
14, 8
Turn-off is shown in figure 7.4. At first, Vgs is stable at 10V . When the
Vgs drops to 0V the Vds response is delayed by 45ns, where after the Vds
increases to Vin in 10ns. After Vds rose and Vgs fell, there is some transient
behaviour, which damps out after about 250ns.
Figure 7.4: Turn-off of IRFSL4620, Vgs = 10V , Vin = 60V , Rgate = 5Ω
STF20NF20 Similar switching behaviour when compared to the IRFSL4620
can be seen with the STF20NF20 MOSFET (shown in figure 7.5). As with
the previous MOSFET, the voltage of the gate driver increases for in about
1
50 ns to the intended 10V then drops, because of the gate capacitance (Cg ).
Vds decreases in 25ns to 0V where it remains fairly constant. Vgs on the
other hands remains oscillating heavily between 3 − 17V for about 45ns,
after which it remains reasonably stable.
CHAPTER 7. EVALUATION
61
Figure 7.5: Turn-on of STF20NF20, Vgs = 10V , Vin = 60V , Rgate = 5Ω
Table 7.2: Measured quantities
Parameter
Measured (ns)
Turn on delay time 15
Rise time
15
Turn off delay time 50
Fall time
10
STF20NF20
Datasheet (ns) [46]
15
30
40
10
The off switching behaviour is different from the IRFSL4620. Vgs damps
much slower than the previous MOSFET, in about 50ns. We also see in
figure 7.6 that the MOSFET is turned on again when it is not supposed to.
This is due to oscillations in the drive voltage rising above the threshold.
Table 7.2 again shows the timing values. Here we see that the timings
more closely match the specified values in the datasheet, despite differing
operating conditions.
CHAPTER 7. EVALUATION
62
Figure 7.6: Turn-off of STF20NF20, Vgs = 10V , Vin = 60V , Rgate = 5Ω
IRFU220 Compared to the other MOSFETs, we see that the IRFU220
has the ’nicest’ switching behaviour, with the least delay times and relative
small oscillations. Table 7.3 shows the timing signals. We see that the specified values closely match the measured values. The specified timing values
for this MOSFET are generally much lower than both other devices, a result
of the low (15nC) gate charge. Unfortunately, this MOSFET comes in a relative small I-Pak package and we had problems fitting it to the heatsinks.
Although test results were somewhat promising for this MOSFET, the cooling problems led to a few destroyed devices.
Figure 7.7: Turn-on of IRFU220, Vgs = 10V , Vin = 40V , Rgate = 5Ω
CHAPTER 7. EVALUATION
63
Table 7.3: Measured quantities IRFU220
Parameter
Measured (ns) Datasheet (ns) [48]
Turn on delay time 10
6, 4
Rise time
10
11
Turn off delay time 25
20
Fall time
8
12
Figure 7.8: Turn-off of IRFU220, Vgs = 10V , Vin = 40V , Rgate = 5Ω
7.3
Sensors
The current and voltage sensors, designed in chapter 5 were tested. The
voltage sensor, which consists of a simple voltage divider acted almost as
expected. However, the Zener diode decreased the accuracy significantly.
The Zener diodes were removed.
The current sensor caused more problems. The chosen transistor in chapter
5 (the BF423) was replaced by a different PNP transistor (BC178), because
it did not go into conduction mode. Apart from this problem, the current
sensor is very noisy. This is due to the very small signal, which is in the
order of a few 100mV . This can be solved by using shielded lines for the
current sensors’ input.
CHAPTER 7. EVALUATION
7.4
64
Control circuitry
The gate is driven by the MIC4420 driver chip. This driver is in turn
controlled by the PWM signal from the ATmega 168 microcontroller [33].
In chapter 5 we calculated that the power dissipation of the driver would be
0.023W for this setup. The proof-of-concept is connected to a 10V voltage
supply. The ATmega requires 5V , which for practical reasons is provided by
a Low Drop Output (LDO) regulator. The total dissipation of the driver,
LDO and microcontroller combined has been measured at 0.3W . With the
LDO dissipating 0.15W and the ATmega 0.1W , we arrive at a measured
dissipation of 0.05W for the gate driver. This is higher than calculated,
however this could be attributed to inaccurate measurements. Although
the total dissipation of the control circuit is small compared to that of the
converter, some improvement could be made by replacing the LDO with a
more efficient 10V to 5V DC-DC converter. This is however beyond the
scope of this thesis.
Summary We have measured the efficiency of the DC-DC converter and
looked at the MOSFET behaviour in this system. Converter efficiencies for
all devices was found to be > 95% under almost all circumstances. It was
shown that the device chosen in chapter 5, the IRFSL4620, indeed is the
most efficient MOSFET of the three tested devices. All three MOSFETs
meet the set requirements when it comes to timing and other requirements.
The IRFU220 was not tested successfully at some duty cycles due to device failures, however a better cooling solution could fix this. This has not
been done since calculations already showed that this device will be less
efficient. Thermal properties of the MOSFETs under load have not been
measured. Thermal performance could be evaluated by measuring heatsink
temperatures (thermocouple or non-contact measurements using infrared)
and comparing them to our design values. These measurements have not
been implemented due to lack of time. Since both the IRFSL4620 and
STF20NF2 have operated under full load for long times (> 15 minutes)
without failure or excessive heat production, we assume that the proposed
heatsink functions properly.
Chapter 8
Conclusion &
Recommendations
In this thesis a distributed maximum power point tracking system for use
in the Nuna 6 solar racing vehicle was proposed. The advantages of this
system under changing lighting conditions were shown with Simulink simulations. As a part of this proposed system, a DC-DC boost converter proofof-concept has been built. For the boost converter, this thesis describes the
design process of the switching element, gate driver and sensors. The proofof-concept boost converter has been tested and proved to be 97.1% efficient
at the specified operating conditions. This is a typical value for this type of
power electronics. This value exceeds our requirements.
In cooperation with the two other theses belonging to this project, the principle of distributed tracking and the proof-of-concept of this system have
been shown to function properly. However, this is only a first step in the
process of developing a prototype that can be tested and eventually used
under race conditions. To create a practical system of multiple trackers,
further study into the optimal distributed tracking architecture (primarily
the number of trackers for maximum power output) is required. On the
boost converter module, more effort could be put into finding more efficient boost converter topologies and advanced soft-switching techniques to
further decrease converter losses. To design a reliable product for use in
the actual vehicle, we recommend further study into making an off-the-shelf
reliable, safe, robust and user-friendly product. Which meets all specific
considerations and requirements regarding the Nuna or other solar racing
vehicles.
65
Bibliography
[1] G. I. A. V. B. Omubo-Pepple, C. Israel-Cookey, “Effects of temperature, solar flux and relative humidity on the efficient conversion of solar
energy to electricity,” European Journal of Scientific Research, vol. 35,
no. 2, pp. 173–180, 2009.
[2] M. Patel, Wind and Solar Power Systems: Design, Analysis, and
Operation, ser. Mechanical Engineering. Taylor & Francis, 2006.
[Online]. Available: http://books.google.nl/books?id=8aQDw14fQnkC
[3] C.-C. Chu and C.-L. Chen, “Robust maximum power point tracking
method for photovoltaic cells: A sliding mode control approach,” Solar
Energy, vol. 83, no. 8, pp. 1370 – 1378, 2009. [Online]. Available:
http://www.sciencedirect.com/science/article/pii/S0038092X09000516
[4] R. Giral, C. Carrejo, M. Vermeersh, A. Saavedra-Montes, and
C. Ramos-Paja, “Pv field distributed maximum power point tracking
by means of an active bypass converter,” in Clean Electrical Power
(ICCEP), 2011 International Conference on, june 2011, pp. 94 –98.
[5] N. Femia, G. Lisi, G. Petrone, G. Spagnuolo, and M. Vitelli, “Distributed maximum power point tracking of photovoltaic arrays: Novel
approach and system analysis,” Industrial Electronics, IEEE Transactions on, vol. 55, no. 7, pp. 2610 –2621, july 2008.
[6] S. Poshtkouhi, V. Palaniappan, M. Fard, and O. Trescases, “A general
approach for quantifying the benefit of distributed power electronics
for fine grained mppt in photovoltaic applications using 3d modeling,”
Power Electronics, IEEE Transactions on, vol. PP, no. 99, p. 1, 2011.
[7] M. K. Kazimierczuk, Pulse-width Modulated DC-DC Power Converters.
Wiley, 2008.
66
BIBLIOGRAPHY
67
[8] M. D. Seeman, “Analytical and practical analysis of switched-capacitor
dc-dc converters,” Ph.D. dissertation, 2006.
[9] L. Pylarinos, “Charge pumps: An overview,” in Proceedings of the
IEEE International Symposium on Circuits and Systems, 2003.
[10] G. Palumbo and D. Pappalardo, “Charge pump circuits: An overview
on design strategies and topologies,” Circuits and Systems Magazine,
IEEE, vol. 10, no. 1, pp. 31–45, 2010.
[11] G. Hua and F. C. Lee, “Soft-switching pwm techniques and their applications,” in Power Electronics and Applications, 1993., Fifth European
Conference on, pp. 87–92 vol.3.
[12] C. P. Basso, Switch-Mode Power Supplies: SPICE Simulations and
Practical Designs. McGraw-Hill, 2008.
[13] K. H. Liu and F. C. Y. Lee, “Zero-voltage switching technique in dc/dc
converters,” Power Electronics, IEEE Transactions on, vol. 5, no. 3,
pp. 293–304, 1990.
[14] M. Bildgen, Resonant converter topologies. ST Microelectronics, 1999. [Online]. Available: http://www.datasheetcatalog.org/
datasheet/SGSThomsonMicroelectronics/mXyzxvt.pdf
[15] J. Bauman and M. Kazerani, “A novel capacitor-switched regenerative snubber for dc/dc boost converters,” Industrial Electronics, IEEE
Transactions on, vol. 58, no. 2, pp. 514–523, 2011.
[16] J. Marshall and M. Kazerani, “A novel lossless snubber for boost converters,” in Industrial Electronics, 2006 IEEE International Symposium
on, vol. 2, pp. 1030–1035.
[17] J. Smith, K. M. and K. M. Smedley, “Properties and synthesis of passive lossless soft-switching pwm converters,” Power Electronics, IEEE
Transactions on, vol. 14, no. 5, pp. 890–899, 1999.
[18] Y. Konishi and H. Yung-Fu, “Soft-switching buck boost converter using passive snubber composed of pulse current regenerative resonant
circuit,” in Telecommunications Energy Conference, 2007. INTELEC
2007. 29th International, pp. 886–890.
[19] P. Bode, “Application note 39 current measurement applications
handbook,” 2008. [Online]. Available: http://www.diodes.com/ files/
products appnote pdfs/zetex/an39.pdf
BIBLIOGRAPHY
68
[20] Keithley, Low level measurements handbook, precision dc current, voltage and resistance measurements. Keithley, 2004.
[21] N. Pilling, R. Holmes, and G. Jones, “Low-power optical current measurement system employing a hybrid transmitter,” Science, Measurement and Technology, IEE Proceedings -, vol. 141, no. 2, pp. 129 –134,
mar 1994.
[22] J. Webster, The Measurement, Instrumentation, and Sensors Handbook,
ser. Electrical Engineering Handbook Series. CRC Press, 1999.
[Online]. Available: http://books.google.nl/books?id=b7UuZzf9ivIC
[23] C. Blake and C. Bull, “Igbt or mosfet: Choose wisely.” [Online].
Available: www.irf.com/technical-info/whitepaper/choosewisely.pdf
[24] W. Konrong, F. C. Lee, H. Guichao, and D. Borojevic, “A comparative
study of switching losses of igbts under hard-switching, zero-voltageswitching and zero-current-switching,” in Power Electronics Specialists
Conference, PESC ’94 Record., 25th Annual IEEE, pp. 1196–1204 vol.2.
[25] Powerex,
“Cm150du-12f
datasheet.”
[Online].
Available:
http://pdf1.alldatasheet.com/datasheet-pdf/view/20687/
POWEREX/CM150DU-12F.html
[26] International Rectifier, “Power mosfet basics,” 2001. [Online].
Available: www.irf.com/technical-info/appnotes/mosfet.pdf
[27] N. Mohan, T. M. Undeland, and W. P. Robbins, “Power electronics converters, applications, and design (3rd edition),” 2003.
[28] International Rectifier, “Application note an-937:
Gate drive
characteristics an requirements for hexfet power mosfets.” [Online].
Available: www.irf.com/technical-info/appnotes/an-937.pdf
[29] Microchip Technology, “Application note an898 determining mosfet
driver needs for motor drive applications,” 2003. [Online]. Available:
http://ww1.microchip.com/downloads/en/AppNotes/00898a.pdf
[30] H. Rashid, Power Electronics Handbook: Devices, Circuits, and Applications. Elsevier, 2010.
[31] CLARE, “Mosfet/igbt drivers application note an-401,” 2012. [Online].
Available: pdfserv.maxim-ic.com/en/an/AN401.pdf
BIBLIOGRAPHY
69
[32] A. T. Sluimer and F. Nassiri Nia, “Converter design for maximum power
point tracking,” To be published.
[33] S. Moring and A. F. Pols, “Maximum power point tracking: Algorithm
and software development,” To be published.
[34] Panasonic, “NNP Series NCR18650A Datasheet,” 2009. [Online].
Available: http://industrial.panasonic.com/www-data/pdf/ACA4000/
ACA4000PE6.pdf
[35] Panasonic, “Lithium Ion Batteries Technical Handbook,” 2007.
[Online]. Available: http://industrial.panasonic.com/www-ctlg/ctlg/
qACI4000 EU.html
[36] Wikipedia, “Boost converter topology,” may 2012. [Online]. Available:
http://upload.wikimedia.org/wikipedia/en/9/9f/Boost circuit.png
[37] D. M. Robert Warren Erickson, Fundamentals of Power Electronics.
Springer, 2001.
[38] I. Budihardjo and P. G. Lauritzen, “The lumped-charge power mosfet model, including parameter extraction,” Power Electronics, IEEE
Transactions on, vol. 10, no. 3, pp. 379–387, 1995.
[39] International Rectifier, “Application note an-944: Use gate charge to
design the gate drive circuit for power mosfets and igbts.” [Online].
Available: www.irf.com/technical-info/appnotes/an-944.pdf
[40] G. Dusan, M. Purschnel, and A. Kiep, “Infineon mosfet power losses calculation using the data-sheet parameters.”
[Online]. Available: http://www.btipnow.com/library/white papers/
MOSFETPowerLossesCalculationUsingtheData-SheetParameters.pdf
[41] MAXIM, “Application note 1832: Power supply engineer’s guide
to calculate dissipation for mosfets in high-power supplies,” 2002.
[Online]. Available: http://pdfserv.maxim-ic.com/en/an/AN1832.pdf
[42] Electus, “Application note: Heatsink basics.” [Online]. Available:
http://www.electusdistribution.com.au/images uploaded/heatsink.pdf
[43] MAXIM, “Application note 746: High-side current sense measurement:
Circuits and principles,” 2001. [Online]. Available: http://pdfserv.
maxim-ic.com/en/an/AN746.pdf
BIBLIOGRAPHY
70
[44] Supertex Inc., “Hv7801 datasheet,” 2008. [Online]. Available:
www.supertex.com/pdf/datasheets/HV7801.pdf
[45] Analog Devices, “AD8212 Datasheet,” 2009. [Online]. Available:
www.analog.com/static/imported.../AD8212.pdf
[46] STMicroelectronics,
“STD20NF20/STF20NF20/STP20NF20
Datasheet,”
2009. [Online]. Available:
http://www.st.com/
internet/com/technical resources/technical literature/datasheet/
CD00150587.pdf
[47] International Rectifier, “IRFS4620PbF/IRFSL4620PbF Datasheet,”
2008. [Online]. Available:
http://www.irf.com/product-info/
datasheets/data/irfs4620pbf.pdf
[48] International Rectifier, “IRFR220N/IRFU220N Datasheet,” 2000.
[Online]. Available:
http://www.irf.com/product-info/datasheets/
data/irfr220n.pdf
[49] Texas Instruments, “Application note: Optimizing mosfet gate drive
voltage.” [Online]. Available: http://www.ti.com/litv/pdf/slua341
[50] Microchip, “Application note an799:
Matching mosfet drivers
to mosfets,” 2004. [Online]. Available: http://ww1.microchip.com/
downloads/en/AppNotes/00799b.pdf
[51] MICREL, “Mic4420/4429 datasheet,”
www.micrel.com/ PDF/mic4420.pdf
2005. [Online]. Available:
Glossary
∆T
θ
ADC
VBR
Q
Icharge
CM RR
CCM
I
DCM
Id
VDS
Rdriver
ROH
VDS
Rdriver
ROH
ROL
EM I
ESR
f
CG
QG
Rgate
CGD
Temperature rise [◦ C]
Thermal resistance
Analogue-to-digital conversion
Breakdown voltage [V]
Charge [C]
Charge current [A]
Common mode rejection ratio
Continuous conduction mode
Current [A]
Discontinuous conduction mode
Drain current [A]
Drain-source Voltage [V]
Driver output resistance [Ω]
Driver output resistance, high output [Ω]
Drain-source Voltage [V]
Driver output resistance [Ω]
Driver output resistance, high output [Ω]
Driver output resistance, low output [Ω]
Electromagnetic interference
Equivalent series resistance [Ω]
Frequency [Hz]
Gate capacitance [F]
Gate charge [C]
Gate resistance [Ω]
Gate-drain capacitance (”Miller”) [F]
71
GLOSSARY
CGS
VGS
Il
LC
IGBT
Tj
RL
IM
MPPT
M OSF ET
Ciss
RDS(on)
Coss
Crss
Vof f
CO
VO
Ipeak
PV
P
PWM
SOA
SiC
fsw
SM P S
T
TC
ZCS
ZV S
72
Gate-source capacitance [F]
Gate-source voltage [V]
Inductor current [A]
Inductor-capacitor
Insulated gate bipolar transistor
Junction Temperature [◦ C]
Load Resistance [Ω]
Maximum drain current [A]
Maximum power point tracking
Metal-oxide-semiconductor field-effect transistor
MOSFET input capacitance [F]
MOSFET on-state resistance [Ω]
MOSFET output capacitance [F]
MOSFET reverse transfer capacitance [F]
Off-state voltage [V]
Output capacitance [F]
Output voltage [V]
Peak current rating [A]
Photovoltaic
Power [W]
Pulse width modulation
Safe operating area
Silicon carbide
Switching frequency [Hz]
Switch-mode power supply
Time [s]
Time constant
Zero current switching
Zero voltage switching
List of Figures
2.1
2.2
Temperature dependancy of a solar panel, source: [3] . . . .
Light intensity dependancy of a solar panel, source: [3] . . .
4
4
3.1
3.2
3.3
3.4
8
10
12
3.5
Classification of power supply technologies, source: [7] . . . .
Shunt current sensor, source: [20] . . . . . . . . . . . . . . . .
Voltage division . . . . . . . . . . . . . . . . . . . . . . . . .
Preferred operating regions for MOSFET and IGBT, source:
[23] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Preferred operating regions for MOSFET and BJT, source: [26]
4.1
4.2
4.3
4.4
4.5
4.6
4.7
4.8
4.9
4.10
4.11
4.12
4.13
4.14
4.15
4.16
4.17
Total system overview of the Nuna 6 . . . . . . . . . . . . . .
Division of sensor submodule into functional elements . . . .
Division of boost converter submodule into functional elements
Division of controller submodule into functional elements . .
Partial solar panel model . . . . . . . . . . . . . . . . . . . .
DriveTek V4 Converter model . . . . . . . . . . . . . . . . .
SolarMax Converter model . . . . . . . . . . . . . . . . . . .
The complete combined model . . . . . . . . . . . . . . . . .
Output power for equal insolation of 1000W/m2 . . . . . . .
Output power for insolation 1000 and 2000W/m2 . . . . . .
Output power for insolation 750 and 1000W/m2 . . . . . . .
Voltage divider . . . . . . . . . . . . . . . . . . . . . . . . . .
Boost converter topology, source: [36] . . . . . . . . . . . . .
MOSFET with junction capacitance, source: [30] . . . . . . .
Variation of Cgd and Cgs as a function of VDS , source: [30] . .
Microchip T442X 9A MOSFET Driver, source: [29] . . . . . .
Microchip T442X charging external gate, source: [29] . . . . .
16
17
17
18
19
20
21
23
23
24
25
28
29
30
31
36
36
5.1
5.2
AD8212 Topology, source: [45] . . . . . . . . . . . . . . . . .
AD8212 Bias resistor chart, source: [45] . . . . . . . . . . . .
41
43
73
13
13
LIST OF FIGURES
6.1
6.2
6.3
6.4
7.1
7.2
7.3
7.4
7.5
7.6
7.7
7.8
Proof of concept, practical implementation . . . . . . . . . .
Practical implementation of the sensor modules . . . . . . .
Implementation of the gate driver . . . . . . . . . . . . . . .
Gate drive and drain-source voltage waveform with STF20NF20
with (coloured) and without (white) external 5Ω resistor. . .
Efficiency vs. Input power at different duty-cycles IRFSL4620,
100Ω load . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Efficiency vs. Input power at different duty-cycles STF20NF20,
100Ω load . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Turn-on of IRFSL4620, Vgs = 10V , Vin = 60V , Rgate = 5Ω .
Turn-off of IRFSL4620, Vgs = 10V , Vin = 60V , Rgate = 5Ω .
Turn-on of STF20NF20, Vgs = 10V , Vin = 60V , Rgate = 5Ω
Turn-off of STF20NF20, Vgs = 10V , Vin = 60V , Rgate = 5Ω
Turn-on of IRFU220, Vgs = 10V , Vin = 40V , Rgate = 5Ω . .
Turn-off of IRFU220, Vgs = 10V , Vin = 40V , Rgate = 5Ω . .
74
52
53
55
56
58
58
59
60
61
62
62
63
List of Tables
3.1
3.2
Common current measuring methods [19] . . . . . . . . . . .
Common voltage measuring methods [22] . . . . . . . . . . .
9
11
5.1
5.2
5.3
5.4
5.5
5.6
5.7
Current Sense IC overview . . . . . . . . . . . . . . . . .
MOSFET Operating requirements . . . . . . . . . . . .
MOSFET Selection [46, 47, 48] . . . . . . . . . . . . . .
VGS Selection [46, 47, 48] . . . . . . . . . . . . . . . . .
Driver peak current selection [46, 47, 48] . . . . . . . . .
Driver Rdriver selection [46, 47, 48] . . . . . . . . . . . .
Driver and MOSFET combination: Driver selection [46,
48, 51] . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . .
. . .
. . .
. . .
. . .
. . .
47,
. . .
40
46
47
48
49
49
6.1
6.2
6.3
Practical implementation overview legend . . . . . . . . . . .
Sensor implementation legend . . . . . . . . . . . . . . . . . .
Driver overview legend . . . . . . . . . . . . . . . . . . . . . .
52
53
55
7.1
7.2
7.3
Measured quantities IRFSL4620 . . . . . . . . . . . . . . . .
Measured quantities STF20NF20 . . . . . . . . . . . . . . . .
Measured quantities IRFU220 . . . . . . . . . . . . . . . . .
60
61
63
75
50
Appendix A - MOSFET
Calculator
close all;
clear all;
clc;
%MOSFET AND DRIVER CALCULATOR%%%%
disp('BAP NUNA Mosfet & Driver calculator ')
V DS = input('Maximum Drain−Source voltage:
if (isempty(V DS))
V DS = 160;
end
D = input('Duty cycle:
if (isempty(D))
D = 0.62;
end
Default 0.62[ ](RANGE 0−1)');
fsw = 1e3*input('Switching frequency:
if (isempty(fsw))
fsw = 100000;
end
disp('MOSFET CALCULATOR
sel = input('[0/1]');
Default 160[V]');
Default 100[kHz]');
[0] OR DRIVER CALCULATOR [1]')
%%%%%%%%%%%%%%%%%%%%%%%%%%MOSFET BEGIN%%%%%%%%%%%%%%%%%%%%%%
if(sel == 0)
disp('General parameters') %ask for input, assume defaults when nothing is input
P in = input('Input power: Default 200[W]');
if (isempty(P in))
76
APPENDIX A - MOSFET CALCULATOR
77
P in = 200;
end
V in = input('Minimum input voltage:
if (isempty(V in))
V in = 60;
end
Default 60[V]');
T Amax = input('Max allowed Ambient Temp in housing:
if (isempty(T Amax))
T Amax = 60;
end
Default 60[C]');
disp('MOSFET parameters, Defaults IRFSL4620') % DEFAULTS FOR IRFSL4620PBF
TJ max = 175
t on = 1e−9*input('MOSFET rise time:
if (isempty(t on))
t on = 22.4e−9;
end
t off = 1e−9*input('MOSFET fall time:
if (isempty(t off))
t off = 14.8e−9;
end
22.4e−9[ns]');
14.8e−9[ns]');
R DSON spec = input('On state DrainSource resistance, R DS ON:
if (isempty(R DSON spec))
R DSON spec = 63.7e−3;
end
T spec = input('Temperature at which R DS ON is specified:
if (isempty(T spec))
T spec = 25;
end
63.7e−3[Ohm]');
25[C]');
HS =input('Heatsink [1/0]?');
HSempty = isempty(HS);
if(HSempty == 1)
HS =0;
end
if(HS==0)
theta JA = input('Total Junction−to−Ambient thermal resistance:
40[C/W]');
if (isempty(theta JA))
theta JA = 40;
end
APPENDIX A - MOSFET CALCULATOR
78
elseif (HS ==1)
theta JC = input('MOSFET Junction−to−case thermal resistance:
1.045[C/W]');
if (isempty(theta JC))
theta JC = 1.045;
end
theta CS = input('Case−to−sink thermal resistance:
if (isempty(theta CS))
theta CS = 1;
end
1[C/W]');
theta SA = input('Heatsink−to−ambient thermal resistance:
if (isempty(theta SA))
theta SA = 8.97; % Indian chief heatsink
end
8.97[C/W]');
theta JA = theta JC+theta CS+theta SA
end
TJ hot = input('Guess Junction operating temp:
[C]');
while 1
I in = P in/V in;
I Drms = sqrt(D*(I in)ˆ2);
R DSON HOT = R DSON spec *(1+0.005*(TJ hot−T spec));
P CM = R DSON HOT * I Drmsˆ2;
P SWM = 0.5* V DS * I Drms *fsw*(t on+t off);
P M = P CM+P SWM;
TJ rise = P M * theta JA;
T A = TJ hot−TJ rise;
if T A<T Amax %Ambient smaller than T Amax, thus mosfet will get too hot when T Amax is
if TJ hot>TJ max
disp('error terror,. Mosfet too hot')
break
end
APPENDIX A - MOSFET CALCULATOR
79
TJ hot = TJ hot+1;
end
TJ hot
if T A>T Amax
disp('De benodigde Ambient Temperature ligt hoger dan die temperatuur in de casing.
c = input('doorgaan [1/0]?');
if (c == 1)
TJ hot = TJ hot−input('hoeveel aftrekken van TJ hot?')
end
if (c==0)
disp('TJ rise')
TJ rise
break
end
end
end
disp('Total losses in MOSFET in W')
disp('Conduction:')
disp(P CM)
disp('switching:')
disp(P SWM)
disp('Totaal:')
disp(P M)
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%MOSFET END%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%DRIVER BEGIN%%%%%%%%%%%%%%%%%
if(sel ==1)
disp('WELKOM BIJ DE DRIVER SELECTOR')
V GS = input('Gate Source Voltage:
if (isempty(V GS))
V GS = 10;
end
Default 10[V]');
Q G = 1e−9*input('Typical Gate Charge:
if (isempty(Q G))
Q G = 15e−9;
end
Default 15 [nC]');
APPENDIX A - MOSFET CALCULATOR
80
%%%% FIRST ORDER DISSIPATION%%%%%%
C G = Q G/V GS;
P DISS DRIVER = C G * V GSˆ2*fsw
P DISS DRIVER2 = Q G * V GS *fsw
%%%%%%FIRST ORDER CHARGE TIME, PEAK CHARGE CURRENT%%%%%
%%%t Charge is 1/100 of 1 switching cycle
t charge = (1/fsw)*(1/100)
i charge = Q G/t charge
%%%%% REQ ON RDRIVER%%%%%%%%
R gate = input('MOSFET Gate resistance:
if (isempty(R gate))
R gate = 2.6;
end
Default 2.6 [Ohm]');
TC = input('number of TCs to charge gate (1−3):
if (isempty(TC))
TC = 2;
end
Default 2 []');
R driver = (t charge/(C G *TC))−R gate
%%%%%%%%%%%%%%%%Advanced loss calculation%%%%%%%%%%%%%%%
disp('Advanced dissipation calculation')
R OL = input('Driver output resistance @ output LOW:
if (isempty(R OL))
R OL = 3;
end
R OH = input('Driver output resistance @ output HIGH:
if (isempty(R OH))
R OH = 2.3;
end
V CC = input('Driver Suppyl voltage:
if (isempty(V CC))
V CC = V GS;
end
Default 3 [ohm]');
Default 2.3 [ohm]');
Default equal to V GS: 10 [V]');
APPENDIX A - MOSFET CALCULATOR
P Dadv = ((R OH * V CC * Q G *fsw)/(R OL))
end
disp('Run script again? [1/0]');
restart = input('[1/0]');
if(restart == 0)
end
if(restart ==1)
mosfet calculator
end
81
Appendix B - Model Plotter
close all;
kies = input('Model draaien [0], alleen plotten [1]');
if (kies == 0)
mdl = 'MPPT';
load system(mdl);
sim(mdl);
end
%duty = max(duty cycle.signals.values);
insolation1 = max(Insolation 1.signals.values);
insolation2 = max(Insolation 2.signals.values);
figure(1)
subplot(221)
plot(Pout SolarMax.signals.values)
title('Pout SolarMax')
subplot(222)
plot(Pout DriveTek.signals.values)
title('Pout DriveTek')
subplot(223)
plot(Eout SolarMax.signals.values)
title('Eout SolarMax: Total harvested energy')
subplot(224)
plot(Eout DriveTek.signals.values)
title('Eout DriveTek: Total harvested energy')
82
APPENDIX B - MODEL PLOTTER
83
ha = axes('Position',[0 0 1 1],'Xlim',[0 1],'Ylim',[0
1],'Box','off','Visible','off','Units','normalized', 'clipping' , 'off');
text(0.5, 1,['Insolation 1: ' num2str(insolation1) ', Insolation 2: ' num2str(insolation2)]
figure(2)
subplot(211)
plot(Pout SolarMax.signals.values)
title('Pout SolarMax')
subplot(212)
plot(Pout DriveTek.signals.values)
title('Pout DriveTek')
kies2 = input('Alle data weggooien [1], data bewaren [0]')
if (kies2 == 1)
clear all;
clc;
end
clc;
Appendix C - Brief of
requirements
In this section, we will present a list of general requirements of the MPPT
system. These specifications are typical requirements that a customer might
have. The MPPT should:
• fit into the Nuna 6;
• always produce the maximum power output of a solar panel;
• communicate one-way with the solar car;
• communicate through the CAN bus protocol;
• be able to be used by anyone without knowledge of the system;
• indicate its status with LEDs
• consist of only one unit;
• be as efficient as possible;
• be easily replaceable.
These requirements are formulated in more detail in following subsections.
Use of the MPPT
[2.2.1.1] The MPPT will maximize the output power of a connected solar cell
under all conditions specified in the Usage Characteristics.
[2.2.1.2] The MPPT will be installed fivefold in the Nuna 6 solar racing vehicle,
this will be called the system.
84
APPENDIX C - BRIEF OF REQUIREMENTS
85
[2.2.1.3] The system can be installed, removed and operated by any Nuon Solar
team member, without requiring prior knowledge of its inner workings.
[2.2.1.4] The MPPT provides plug-and-play connectivity.
[2.2.1.5] Each MPPT has an external reset switch.
[2.2.1.5] The system will communicate with the vehicle.
[2.2.1.6] The MPPTs operating status can be visually inspected.
[2.2.1.7] The system will be robust, stable and safe.
[2.2.1.8] The system will be constructed as light and compact as possible.
[2.2.1.9] The system can be easily integrated into Nuna 6 and its future successors.
[2.2.1.10] Each MPPT will have a life expectancy under normal operating conditions of at least 3 years.
[2.2.1.11] The MPPT is maintenance free during its expected lifetime.
Ecological requirements
[2.2.2.1] The system wastes as little energy as possible.
[2.2.2.2] The production process has minimal environmental impact.
[2.2.2.3] All components are RoHS compliant.
Technical requirements
[2.2.3.1] A total of 5 on board MPPTs will provide Nuna 6 with more power
under equal conditions than the current power point tracking system.
[2.2.3.2] A fault in one of the MPPTs will not lead to loss of power to Nuna 6,
other than the power controlled by the faulty MPPT itself.
[2.2.3.3] The MPPT accepts input voltages between 60V and 120V .
[2.2.3.4] The MPPT connects to the Nuna 6 power bus, providing an output
voltage between 110V and 160V .
APPENDIX C - BRIEF OF REQUIREMENTS
86
[2.2.3.5] Should the output voltage rise above 160V , indicating a full battery
on the Nuna vehicle, the system will stop tracking to prevent overcharging.
[2.2.3.6] The systems nominal input power is between 20W and 100W with an
absolute maximum of 200W .
[2.2.3.7] The average efficiency of each MPPT will be greater then 95%.
[2.2.3.8] The system operates without malfunction with environmental temperatures ranging between −10 ◦ C and 80 ◦ C.
[2.3.3.9] The system operates without malfunction with environmental humidity ranging between 5% and 95%.
[2.2.3.10] Each MPPT communicates with the vehicle via the vehicle’s CAN-bus
on a send-only basis, providing (next to all communication necessary
for safe and reliable CAN-bus operation) at least: power in [W ], power
out [W ], input voltage [V ], output voltage [V ], output current [A].
[2.2.3.11] The MPPT has at least 4 status LEDs, showing operating status
and/or possible error codes.
[2.2.3.12] The MPPT provides In-System Programming (ISP) using a JTAG
interface.
[2.2.3.13] All components meet the specifications as provided by Veolia World
Solar Challenge.
[2.2.3.14] High-voltage connections must be properly shielded.
[2.2.3.15] All critical high-power connections contain correctly dimensioned fuses.
[2.2.3.16] Casings and enclosures must be constructed properly and provide adequate sealing, strength and durability to protect enclosed components.
[2.2.3.17] All connections will be dust-sealed, shock and vibration proof.
Production and installation requirements
[2.3.4.1] The MPPT and all its components are integrated in a enclosure for
easy mounting in the Nuna 6.
[2.3.4.2] Installation of the system is possible using standard tools available to
the Nuon Solar team.
APPENDIX C - BRIEF OF REQUIREMENTS
87
[2.3.4.3] Each enclosure has a label for identification purposes.
[2.3.4.4] External connections provide standard connectors for easy installation
in the vehicle: power bus connectors: Green Power Connector PC 4
HV/4-STF-7.62 Phoenix, CAN-bus connectors: Orange CAN Connector BL 3,5/4 Weidmller.
Requirements on product end-of-life
[2.2.5.1] The MPPT complies with EU WEEE guidelines for electronic waste
disposal.
[2.2.5.2] The system must be removable from the vehicle.
[2.2.5.3] Any batteries in the system must be easy to remove to be processed
separately.
Strategic and marketing requirements
[2.2.6.1] The system is sold business-to-business.
Appendix D - IRFSL4620
Efficiency Measurements
88
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement