<%BANNER%>

High Efficiency Loosely Coupled Wireless Power Transfer System via Magnetic Induction

Permanent Link: http://ufdc.ufl.edu/UFE0024707/00001

Material Information

Title: High Efficiency Loosely Coupled Wireless Power Transfer System via Magnetic Induction
Physical Description: 1 online resource (172 p.)
Language: english
Creator: Low, Zhen
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: induction, magnetic, power, wireless
Electrical and Computer Engineering -- Dissertations, Academic -- UF
Genre: Electrical and Computer Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: After the introduction of cellular technology and WiFi technology, the power cable is the last cable that has yet to be eliminated. Magnetic induction is the leading technology to achieve wireless power transfer at high efficiencies ( > 75%) with power level ranging from several microwatts to thousands of watts. Using near-field operation at frequencies below 1 MHz significantly lowers the probability of interference and RF safety issues since the wavelength is extremely long and radiation is limited. Design rules are developed to select the components of the system, which are further verified by experiments. Interoperability analysis shows that the system is sufficiently robust that coils from one platform can be used in another platform without damaging the receiver. A method of load/fault detection is proposed and verified with both hardware and software implementation. Finally, three different methods to control the unregulated receiver voltage control and power delivery is studied and it is determined that varying Cout in a single discrete step is able to allow the system to achieve sufficient control.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Zhen Low.
Thesis: Thesis (Ph.D.)--University of Florida, 2009.
Local: Adviser: Lin, Jenshan.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2010-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024707:00001

Permanent Link: http://ufdc.ufl.edu/UFE0024707/00001

Material Information

Title: High Efficiency Loosely Coupled Wireless Power Transfer System via Magnetic Induction
Physical Description: 1 online resource (172 p.)
Language: english
Creator: Low, Zhen
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: induction, magnetic, power, wireless
Electrical and Computer Engineering -- Dissertations, Academic -- UF
Genre: Electrical and Computer Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: After the introduction of cellular technology and WiFi technology, the power cable is the last cable that has yet to be eliminated. Magnetic induction is the leading technology to achieve wireless power transfer at high efficiencies ( > 75%) with power level ranging from several microwatts to thousands of watts. Using near-field operation at frequencies below 1 MHz significantly lowers the probability of interference and RF safety issues since the wavelength is extremely long and radiation is limited. Design rules are developed to select the components of the system, which are further verified by experiments. Interoperability analysis shows that the system is sufficiently robust that coils from one platform can be used in another platform without damaging the receiver. A method of load/fault detection is proposed and verified with both hardware and software implementation. Finally, three different methods to control the unregulated receiver voltage control and power delivery is studied and it is determined that varying Cout in a single discrete step is able to allow the system to achieve sufficient control.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Zhen Low.
Thesis: Thesis (Ph.D.)--University of Florida, 2009.
Local: Adviser: Lin, Jenshan.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2010-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024707:00001


This item has the following downloads:


Full Text





HIGH EFFICIENCY LOOSELY COUPLED WIRELESS POWER TRANSFER SYSTEM VIA
MAGNETIC INDUCTION





















By

ZHEN NING LOW


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2009
































2009 Zhen Ning Low



































To my parents and wife









ACKNOWLEDGMENTS

I would like to thank my advisor Dr. Jenshan Lin for this wonderful opportunity to work

under his guidance and mentoring. I have truly enjoyed working for him over the years acquiring

technical knowledge as well as other soft skills. I would also like to thank Dr. William

Eisenstadt, Dr. David Arnold and Dr. Sumi Helal for their time and being on my committee. I

would like to thank my project teammates, both past and present for which without them this

work will not be possible. They are: Joaquin Casanova for electromagnetic analysis and coil

design; Jason Taylor for PCB design and fabrication; Raul Chinga for all the late night

prototyping work; and Ashley Trowell for ferrites and shielding work. I would like to thank

WiPower and Florida High Tech Corridor for funding this project. I would like to thank Linear

Technology and Terry Decker for their support and evaluation boards. I would also like to thank

Shannon Chillingworth and all of the personnel in the EE office.

Finally, I would like to thank my wife for her encouragement and unconditional support

when things get rough. Without her, this would not be possible.









TABLE OF CONTENTS



A C K N O W L ED G M EN T S ............................................................................ 4

LIST OF TABLES .................................. ............................ 8

LIST OF FIG U RE S ................................................. 9

AB STRA CT .............. ................... ..................... ............................ 15

CHAPTER

1 IN T R O D U C T IO N ............................................................................................... ....... .......... 16

1.1 History of Wireless Power Transmission .................................... 18
1.2 M odern W wireless Power Transmission ................................................................ 20
1.3 Background Information.............................. ......... ........23
1.3.1 Fundamental Theory of Operation via Phase Response.............................. 23
1.3.2 Class D Pow er A m plifier .......................................................................... .. ...... 26
1.3.3 Class E Power Amplifier............. ........ ............... 30

2 WIRELESS POWER TRANSFER SYSTEM............. .......................... 38

2.1 Single-channel Class E Power Amplifier ....................................... 39
2.2 M ulti-channels/Stackable Class E Power Amplifier.......................... ............... 41
2.3 Inductive Coupling ....................................................................... ....... ...... ... 42
2.4 Impedance Transformation Network ...................... ................... 44
2.5 Receiver ........... ............................... ........ 47

3 DESIGN OF IMPEDANCE TRANSFORMATION NETWORK.............. ............... 50

3.1 Series-Parallel Impedance Transformation Network....... ............................... 50
3.1.1 Introduction ............... .. ........................ 50
3.1.2 Determination of Crx value................ ...... ........... 53
3.1.3 Determination of C,,ut value..................... ....... ............... 58
3.1.4 Determination of Cshunt value .................... ..... ......... 59
3.2 Parallel-Parallel Impedance Transformation Network .................................. 60
3.2.1 Introduction ............... ................... ................... ......................60
3.2.2 D eterm nation of C rx value........................................................................ .......... 61
3.2.3 D eterm nation of Ct, value......................................................................... ...... 63
3.2.4 Determination of Cshunt value .................................. ........... 67










4 WIRELESS POWER TRANSFER SYSTEM SUPPORTING MULTIPLE
RECEIVERS ........................................ 69

4.1 Inductive C oupling ............................................... 69
4.2 Switch Design.........................................72
4.3 Switch Simulation ................................. ............... 76
4.4 System Response with Receiver Switch................................................................. 79

5 EX PERIM EN TAL VER IFICA TION ................................................................ ............... 83

5.1 High Power 300W Dual Channel System using Parallel-Parallel Impedance
Transform ation N etw ork Topology ...................... .......................................83
5.2 Low Power Multiple Receivers System with decoupling Switch using Series-
Parallel Impedance Transformation Network Topology ............................................ 95

6 INTEROPERABILITY BETWEEN DIFFERENT PLATFORMS (COIL SIZES)............. 104

6 .1 T est B en ch S etu p ....................................................................................................... 10 5
6.2 Experim mental V verification ........................................ 107

7 LOAD/FAULT DETECTION AND POWER DELIVERY TRACKING ........................... 111

7.1 Load/Fault D election Schem e ........................................ 112
7.1.1 D election C ircuit............................................................................... 112
7.1.2 Detection Flowchart/Logic................................. 115
7.2 Experim mental V verification ........................................ 117
7.2.1 T est B ench and C ircuit ........................................................................... ............117
7.2 .2 E xperim mental R esults........................................................................ 122
7.3 Extension of Load/Fault Detection Scheme...................................... 130
7.3 .1 M :N C ou pling Structure ............................................................................... 130
7.3.2 Removing L,,ut from Transmitter ................................... 139

8 RECEIVER VOLTAGE CONTROL .................................................... 146

8.1 Varying Supply Voltage to Achieve Receiver Voltage Control ............................... 147
8.2 Varying Operating Frequency to Achieve Receiver Voltage Control...................... 152
8.3 Varying C,,ut to Achieve Receiver Voltage Control................................... ...... 157
8.4 Conclusion ......................................................... ........... ......... 160

9 SUMMARY AND FUTURE WORK ........................ ...... ..................... 162

9.1 Sum m ary ................... ........................................................ 162
9.2 Future Work................................................. 164









APPENDIX: FCC REGULATIONS............... ......... ....... ........166

LIST OF REFERENCES .......................................................................... 169

BIOGRAPHICAL SKETCH ......................................... 172



















































7









LIST OF TABLES


Table page

5-1 Component Values for High power 300W Dual Channel System using Parallel-
Parallel Impedance Transformation Network Topology ........................................ 84

5-2 Component Values for Low Power Multiple Receivers System with decoupling
Switch using Series-Parallel Impedance Transformation Network Topology ................ 96

6-1 Specification of the three different platforms........................................105

6-2 Coupling parameters of nine possible combinations with first three as intended pairs... 106

7-1 Component Values for Load/Fault Detection Test Bench............................................ 118

A-i FCC Part 18.307 conduction test limits .................................. 166

A-2 FCC Part 15.109 emission limits at 3 m range.............................................. 167









LIST OF FIGURES


Figure page

1-1 Block diagram of the multiple receivers wireless power transfer system using
inductive coupling. .............. ...... .......................... 17

1-2 Complex plane of inductive and capacitive load. ............. .......... .................24

1-3 Phasor diagram. .....................................................25

1-4 Class D power amplifier. ......................................... 26

1-5 Class E power amplifier. ................. ......... .......... ......... 31

2-1 Wireless power transfer system diagram. ..................................... 38

2-2 Schematic of the dual-channel Class E power amplifier ..................................... 41

2-3 Topologies for a single-element impedance transformation network...............................45

2-4 Input voltage of a half wave rectifier and a full wave rectifier................ ...............48

3-1 Windings of a 20cm x 20cm transmitting coil used for experimental verification............51

3-2 Normalized power deliver with respect to location of transmitting coil in Figure 3-1
using a receiving coil of 9 cm x 6 cm. .................. .................. ............... 52

3-3 Simplified schematic of wireless power transfer system using series-parallel
impedance transformation network and Class E transmitter. Ztx Impedance looking
into the transmitter load network. Ztxcoil Impedance looking into the transmitting
coil. Zrx Impedance looking into receiver network. RLoad is the equivalent resistance
looking into the rectifier. ......................................... 53

3-4 Resistance and reactance of Zrx versus load resistance at different with different Crx.
(50 nF, 100 nF and 150 nF) .................................................. 54

3-5 Peak resistance response looking into the transmitting coil with respect to Crx. ...............56

3-6 Resistance and reactance looking into the transmitting coil. .................. ............56

3-7 Coupling efficiency with respect to load resistance.......... .....................57

3-8 Ztx, phase response with respect to load resistance for various C,,ut capacitance values.... 58

3-9 Transistor drain voltage waveform for different load resistance. (Cshunt= 10 nF)..........59









3-10 Simplified schematic of wireless power transfer system using parallel-parallel
transformation network and Class E transmitter. Ztx Impedance looking into the
transmitter load network. Ztxcoil Impedance looking into the transmitting coil. Zrx -
Impedance looking into receiver network. RLad is the equivalent resistance looking
into the rectifier. .............. ........................................61

3-11 Optimum receiver capacitor value versus load resistance............................................61

3-12 Coupling efficiency and transformed impedance looking into the transmitting coil.........63

3-13 Reactance of Ztxcoil versus load resistance with different Ctx. ....................................... 65

3-14 Amplitude and phase of impedance of unloaded Ztx versus Ctx........................................ 66

3-15 Rtx and Xtx versus load resistance. (Ctx: 105 nF)..... ................ ......... 66

3-16 Phase of Ztx versus load resistance. (Ctx: 105 nF) ........ ......................................67

3-17 Transistor drain voltage waveform for different load resistance. (Chun,,t = 19 nF).............67

4-1 Power delivery to loads for a single receiver setup and a dual receivers setup with
one of the load fixed at 1000 ..........................................................71

4-2 Block diagram of the proposed switch................ ....... ............72

4-3 Schematic of the proposed switch circuit. ..................................... 74

4-4 Schematic of the improved proposed switch circuit. .............. ................ ............75

4-5 Schematic of the switch in Advanced System Design with a resistive as load............... 76

4-6 Switch control waveform (0 V for off and 3 V for on). A minimum of 1 V is required
to turn on the transistor. .................................................................... 77

4-7 G generated sw itch control w aveform s.......................................................................... ........78

4-8 Output waveform of the switch before the rectifier ......................................................79

4-9 Proposed half-wave rectifier receiver architecture. ....................... ............... 79

4-10 ADS schematic of test bench for receiver architecture in Figure 5-10............................ 81

4-11 Simulation results of test bench (Figure 4-10). Red: Low voltage control
signal.Black: Receiver rectified voltage. ........................................... 82

5-1 Photograph of the dual-channel Class E power amplifier. ..................................... 85

5-2 Photograph of the transmitting coil 10 turns (embedded into the table top) and
receiving coil 5 turns (placed on top). .................. .................. ... ............... 85









5-3 Power delivery (left y-axis) and efficiency (right y-axis) of the system versus load
resistance. Supply voltage: 120 V........................................... ............... 86

5-4 Transistor and inductor temperature with natural convection cooling and forced air
cooling versus supply voltage. ........... ............ ...... ............... 87

5-5 Photograph of the dual-channel Class E power amplifier with forced air cooling. ...........88

5-6 Voltage and current waveforms of the Class E transmitter ..................................... 89

5-7 Power delivered to load versus load resistance. A maximum power of 69 W occurs
approximately at 50 Q for dual -channel and maximum power of 10 W occurs at
approximately at 75 Q for single -channel. Supply voltage: 60 V........................ 89

5-8 Mode-switching operation for optimized efficiency across a wide power delivery
range. (1) Dual-channel mode for higher power, (2) Dual-channel mode switch-over
to single-channel mode when better efficiency can be obtained at a lower power
level, (3) Single-channel mode for lower power, (4) Single-channel mode switch-
over to dual-channel mode when higher power delivery is needed. Supply voltage:
60 V. ........................................ ........ 90

5-9 System efficiency versus load resistance with a maximum efficiency of 64.5% for a
single-channel system, and 76% for a dual-channel system at approximately 700.
Supply voltage: 60 V ..................................................... 9 1

5-10 Transmitter efficiency versus load resistance. Maximum transmitter efficiency
occurs across the range of 600 to 100 Q load resistance at 90% for dual -channel and
79% for single-channel. Supply voltage: 60 V...................................................92

5-11 System efficiency versus load resistance for single-channel and dual-channel modes
achieving high efficiency at high power output. It also illustrates that a single-
channel mode is more efficient at low power delivery states. Supply voltage: 60 V .......93

5-12 Transmitting coil RMS voltage versus load resistance. Supply voltage: 60 V...............93

5-13 Receiver DC voltage versus load resistance, converging to approximately 70 V in
dual-channel mode and 38 V in single-channel mode. Supply voltage: 60 V. ................94

5-14 Photograph of a test setup with two receivers with decoupling switch on the
packaged transm hitting coil. .................................................................. 96

5-15 Power delivery to the receiver with switch and system efficiency versus load
resistance for a one-to-one setup (simulated and measured)............. ............ 97

5-16 Efficiency of power delivery to the receiver with switch versus power delivered for a
1 to 1 setup (sim ulated and m easured)................................................................................... 98

5-17 Comparison between receiver with switch and receiver without switch............................99









5-18 Comparison between single receiver standalone and dual-receiver setup with one
receiver sw itched off. ..................................................... 99

5-19 Power delivery to receiver 1 versus its load resistance at different fixed receiver 2
load resistance. ......... ........ ...............................101

5-20 Power delivery to receiver 2 versus receiver 1 load resistance at different fixed
receiver 2 load resistance. (same legends as in Figure 5-20)....................................... 101

5-21 System efficiency versus total power delivery to both receivers with the load
resistance of receiver 2 fixed and the load resistance of receiver 1 swept across the
stated range for a dual-receiver test bench. (same legends as in Figure 5-20)................. 102

5-22 Measured power delivery to a dual-receiver system with load resistance varied from
100 to 20000 ........................................ ......................... 103

6-1 System efficiency and receiver voltage vs. power delivery plot for different sized
receiving coil on a small sized transmitting coil. .................................... 107

6-2 System efficiency and receiver voltage vs. power delivery plot for different sized
receiving coil on a medium sized transmitting coil. ...................... .............. 109

6-3 System efficiency and receiver voltage vs. power delivery plot for different sized
receiving coil on a big sized transmitting coil................ ............... ..... 110

7-1 Block diagram of the proposed wireless power transfer system with detection circuit
(detecting supply current and coil voltage) .................................... 112

7-2 Schematic of coil current extraction network using a current sense resistor. ............... 114

7-3 Detection scheme flow chart for proposed system. ...................... ............... 116

7-4 Photograph of the fabricated transmitter circuit with control circuit............................. 118

7-5 Schematic of power stage of fabricated transmitter ....................... ... ............ 119

7-6 Schematic of driver stage of fabricated transmitter. ...................... ... ........... 119

7-7 Schematic of detection and control stage of fabricated transmitter.............................. 120

7-8 Efficiency-Power plot of single receiver setup (Solid black line: Small coil. Dashed
gray line: Big coil) .. .............................................. 122

7-9 Examples of different loading conditions and fault modes on coil voltage versus
supply current space. ........................................... 123

7-10 Efficiency-power plot of three sets of dual load measurements and two sets of single
load measurements using a combination of big and small receiving coils.................. 125









7-11 Transmitter coil voltage and supply current space illustrating three different zones:
no-load, single load, and dual loads. .......... ......................... 126

7-12 Transmitter coil voltage and supply current space diagram illustrating three different
zones, no-load, safe and fault. .......... .............................127

7-13 Linear relationship between supply current and power delivered to the loads for all 5
sets of measurements. Solid dashed line: y = 0.095x + 0.055............... ........ 129

7-14 Power delivery error distribution plot. Calculated power delivery is based on
measured supply current. Solid dashed line: ideal error free calculation...................... 129

7-15 Examples of M:N coupling structures. .................................. 131

7-16 Measured power delivery of a dual receiver setup on a dual transmitting coil test
bench with both receivers on a single coil leaving the other coil unloaded..................... 132

7-17 Measured power delivery of a dual receiver setup on a dual transmitting coil test
bench with one receiver on each transmitting coil................... .................133

7-18 Efficiency-power plot of two sets of dual load measurements and two sets of single
load measurements. (Green lines: Single receiver, Blue lines: Dual receivers on
single transmitting coil, Red lines: Dual receivers on dual transmitting coil) ................. 134

7-19 Transmitter coil voltage and supply current space. (Green diamond: No load, Black
lines: Single receiver, Blue lines: Dual receivers on single transmitting coil, Read
lines: Dual receiver on dual transmitting coil) ........................ ............. 134

7-20 Linear relationship between supply current and power delivered to the loads for all 4
sets of measurements. Solid dashed line: y = 0.076x + 0.048................ ... ... 135

7-21 Power delivery error distribution plot. Calculated power delivery is based on
measured supply current. Solid dashed line: ideal error free calculation...................... 136

7-22 Transmitter coil voltage and supply current space with fault tests. (Green lines: Fault
tests without valid receiver on one of the coil, Purple lines: Fault tests with a valid
receiver on one of the transmitting coils) ........................................ 137

7-23 Different methods of driving a 2:N structure............... ...................... 139

7-24 Block diagram of wireless system without L,,ut...... .......................................... 140

7-25 Transmitter coil voltage and supply current space with fault tests with Lout removed.
(Green lines: Fault tests without valid receiver on one of the coil, Purple lines: Fault
tests with a valid receiver on one of the transmitting coils)............................. 141









7-26 Waveform of the transmitter circuit without L,,ut when fault conditions overlaps with
valid load conditions. (Red: Drain voltage, Orange: Coil voltage, Green: Coil
Current) ................ ....... .................. ............... 142

7-27 Spectrum of transmitter coil voltage without L,,ut when fault conditions overlaps with
valid load conditions. ........................................... 142

7-28 Waveform of the transmitter circuit with L,,ut when fault conditions overlaps with
valid load conditions. (Red: Drain voltage, Orange: Coil voltage, Green: Coil
Current) ................ ....... .................. ............... 143

7-29 Spectrum of transmitter coil voltage with L,,ut when fault conditions overlaps with
valid load conditions. ........................................... 144

8-1 Block diagram of wireless power system ................... ....... ........ 146

8-2 Efficiency-power plot for supply voltage from 18 V to 30 V in steps of 1 V............... 148

8-3 Receiver Voltage-Power plot for supply voltage from 18 V to 30 V in steps of 1 V...... 149

8-4 Linear relationship between supply current and power delivered to the load for
supply voltage from 18 V to 30 V in steps of 1 V. ....................... .......... ..... 151

8-5 Efficiency-power plot for operating frequency from 240 kHz to 250 kHz in steps of 2
k lHz. ......................................... ........ 153

8-6 Receiver voltage-power plot for operating frequency from 240 kHz to 250 kHz in
steps of 2 kHz. .............................................. 153

8-7 Efficiency-power plot for operating frequency from 235 kHz to 240 kHz in steps of 1
k lHz. ......................................... ........ 155

8-8 Receiver voltage-power plot for operating frequency from 235 kHz to 240 kHz in
steps of 1 kHz. ............................................. 156

8-9 Linear relationship between supply current and power delivered to the load for
operating frequency from 235 kHz to 250 kHz in steps of 1 kHz. ................................. 156

8-10 Efficiency-power plot for different C,,ut values.................................... 158

8-11 Receiver voltage-power plot for different C,,ut values. .................................. ...... 158

8-12 Linear relationship between supply current and power delivered to the load for Cout
values of 15 nF, 16 nF, 17.2 nF and 18.2 nF ............................... ...... 159









Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

HIGH EFFICIENCY LOOSELY COUPLED WIRELESS POWER TRANSFER SYSTEM VIA
MAGNETIC INDUCTION

By

Zhen Ning Low

August 2009

Chair: Jenshan Lin
Major: Electrical and Computer Engineering

After the introduction of cellular technology and WiFi technology, the power cable is the

last cable that has yet to be eliminated. Magnetic induction is the leading technology to achieve

wireless power transfer at high efficiencies (>75%) with power level ranging from several

microwatts to thousands of watts. Using near-field operation at frequencies below 1 MHz

significantly lowers the probability of interference and RF safety issues since the wavelength is

extremely long and radiation is limited.

Design rules are developed to select the components of the system, which are further

verified by experiments. Interoperability analysis shows that the system is sufficiently robust that

coils from one platform can be used in another platform without damaging the receiver. A

method of load/fault detection is proposed and verified with both hardware and software

implementation. Finally, three different methods to control the unregulated receiver voltage

control and power delivery is studied and it is determined that varying Cout in a single discrete

step is able to allow the system to achieve sufficient control.









CHAPTER 1
INTRODUCTION

A wireless power transfer system is any system that has the capability to transfer electrical

power from a power source to an electrical load without any interconnecting wires. A wireless

power transfer system will have three basic building blocks:

1. Transmitter: DC power is converted to AC power which is used to drive the transmitting
antenna (coil).

2. Energy Transfer: The AC power is transmitted through space to some point where a
receiving unit is located.

3. Receiver: The transmitted AC power is collected and converted to DC by a rectification
process.

Wireless power systems fall into two main categories, medium to long range, where the

coverage is greater or equal to a typical Personal Area Network (PAN) which is typically a 10 m

radius, and short range, where the coverage is localized within the vicinity of the transmitting

device (typically a 5" distance). Although there have been attempts [1-3] to achieve long range

power delivery via far-field techniques, the efficiency and power delivery is insufficient to

charge even a typical cellular phone overnight. In order to provide power comparable to a typical

wall mounted DC supply, the system would most likely violate RF safety regulations [4] or use a

large number of transmitters resulting in an impractical implementation. Therefore, far-field

techniques are most suitable for very low power applications (< 100 mW) unless they are used in

less regulated environments such as military or space exploration. Inductive coupling has been

one of the leading candidates in achieving wireless power transfer at power levels ranging from

several microwatts to several kilowatts. Its operating range is limited as power delivery and

efficiency degrades rapidly with increasing distance between the transmitting and receiving unit.

Using near-field operation at frequencies below 1 VIMHz significantly lowers the probability of

interference and RF safety issues since the wavelength is extremely long and radiation is limited.









Unlike far-fields techniques, near-field techniques are extremely sensitive to the loading

conditions of the receivers) as well as the number of receivers. The block diagram of the

wireless power transfer system using inductive coupling is show in Figure 1-1.


S) I

Figure 1-1. Block diagram of the multiple receivers wireless power transfer system using
inductive coupling.

Other methods of wireless power transmission include using means of optics such as

laser. Although, it is easy to focus a laser beam, energy conversion from electricity to laser is

extremely inefficient. In addition, current state of the art photovoltaic cells used to convert the

received laser to electrical energy have an efficiency of typically less than 50%.









1.1 History of Wireless Power Transmission

The beginning of wireless power transmission can be dated back to 1868 when James

Clerk Maxwell developed the classical electromagnetic theory. Maxwell's equations unify

previous unrelated observations of electricity, magnetism and optics into one consistent theory.

Later in 1888, Heinrich Rudolf Hertz verified the existence of electromagnetic radiation by

building the first radio transmitter.

The first significant breakthrough in wireless power transmission technology happened in

1897 when Nikola Tesla filed his first patents based on the Wardenclyffe Tower, which is also

known as Tesla Tower in Colorado Springs. According to [2], Tesla resonated a 3 feet diameter

copper ball on top of the 200 feet power with 300 kW of power from the Colorado Springs

Electric Company.

There were various attempts after the Second World War to make an efficient transmission

of large amount of power wirelessly but they have failed. This is because the device technology

during that time was insufficient to either generate sufficient amount of power or effectively

rectify the received microwave power. In 1964, William C. Brown invented the "Amplitron" [5]

which is commonly known as the crossed-field amplifier. The amplifier is capable of an output

of 400 kW at efficiency of 70%. With the development of silicon Schottky-barrier diodes by

Hewlett-Packard Associates, high efficiency high power wireless power transfer was made

possible. In 1975, the largest wireless power transfer was demonstrated at the Venus Site of JPL

Goldstone Facility. 450 kW of power was transmitted across one mile using a 26m diameter

parabolic antenna [2]. The 3.4 m x 7.2 m receiving rectenna array achieved a rectified DC power

of30 kW at 82.5% efficiency. The system efficiency was only 6.67% efficiency without taking

the transmitter into consideration. Therefore, most of the power loss was still due to radiation.

Generating a focused electromagnetic beam so that all of the energy can be projected onto a









small area remains a challenging task. The only two ways to achieve an efficient power transfer

over a long distance is to increase the number of transmitting antenna elements so as to achieve a

more focused beam or increase the number of elements of the receiving rectenna. Repeater

stations can be used to reduce the radiation loses, but the conversion losses to retransmit the

power can be so high that it is not feasible.

A more feasible alternative to the long range power transmission is via inductive coupling.

However, the operating range is limited to the proximity of the transmitter. The roots of

inductive coupling date back to 1831 when Michael Faraday demonstrated that in order to induce

a current on a conductor the magnetic field needs to be changing [6]. Faraday did not continue to

pursue the work in this area and left it for others to pick up from where he left off. Among many

who have tried to study the phenomena was Joseph Henry. All the experiments were conducted

by manually turning on and off DC supplies, until the introduction of dynamo in 1860s making

AC supply available. Sir William Grove was the first to connect a transformer to an AC source to

generate high voltage for his laboratory work.

One of the earliest successful demonstrations of a consumer product using inductive

coupling with considerable power is the inductive cooker. In 1971, during the National

Association of Home Builders convention in Houston, TX Westinghouse Electric Company

demonstrated single burner range using a 25 kHz operating frequency [7].

On the other end of the power band, General Electric introduced the rechargeable cordless

electric toothbrush in 1961, mitigating the risk of an electric shock [8]. The power transfer is

achieved by splitting a transformer core into two, for which one of the core is located into the

"base station" and the other is located in the toothbrush. Energy can be coupled to the toothbrush

by placing it onto the "base station" with proper alignment.









The challenge of a robust inductive coupling is to be able to transfer energy to a receiving

device with considerable lateral degree of freedom. In addition, portable electronics are getting

very compact and lightweight, and consumers are unwilling to accept the huge heavy ferrite core.

By using an air core transformer and with the receiver much smaller than the transmitter the

coupling is extremely poor. The only solution is to increase the operating frequency to enhance

the coupling but it is only in the recent years that power transistors are capable of achieving high

power and high speed making considerable amount of power via magnetic induction possible. In

addition, losses through the windings were significant due to skin depth and proximity effect

until the introduction of Litz wires.

1.2 Modern Wireless Power Transmission

The first company to start working on inductive coupling wireless power system was

Splashpower [9] in 2001. In order to avoid the receiving device from "blocking" the vertical

field, Splashpower developed a unique coil design which enables the transmitting platform to

transmit a horizontal field in both X and Y direction. This enables the receiver to be insensitive

to location and rotation. However, with the introduction of many compact and low profile

devices such as Apple's iPod and Motorola's Razr, modem consumer electronics are more

sensitive to the thickness of the device. Therefore, the allowable extra cross sectional area on the

receiving device is almost non-existent. In addition being one of the early adopters of the

technology, the Splashpower system operates in tens of kilohertz making the system more

inefficient than current solutions which operate at hundreds of kilohertz. Splashpower was

acquired by Fulton Innovation also known as eCoupled in 2008.

eCoupled's wireless power technology was a by-product of its parent company Alticor's

eSpring water purifier [10]. Approximately 10 years ago engineers at eCoupled were trying to

prevent corrosion and electrical shock hazard to the ultraviolet lamp; in the end they developed a









wireless power system to solve their problem. The system designed by eCoupled involves

complex control schemes and costly communication link to achieve resonance operation.

Although the cost of the system is reasonable to the unique water purifier, it is considered to be

high for cost sensitive consumer electronics which have very low profit margin. However,

eCoupled has successfully demonstrated a power delivery up to 1000 W grilling a piece of steak

on a George Foreman grill as well as powering a 2000 W blender.

ConvenientPower [11] is brainchild of Professor Ron Hui from the City University of

Hong Kong. His first technical journal was dated back to 2005 which is the first published for

portable electronics. Professor Hui's most significant contribution to the wireless power space is

the design concept of generating an even magnetic field across a large area. Although, 7 journal

papers were published and 6 patents were awarded, little was known about the system as there

hasn't been any official demonstration conducted to the press. Available literature indicates that

the system is a variation of magnetic induction proposed by Splashpower, eCoupled and

WiPower [14] using the same underlying principles.

PowerMat [12] which is based in Israel is the newest player in the wireless power space.

Neither publications nor patents have been found so far. Similar to ConvenientPower little

information is known about the technology. Although, there was an official demonstration to the

press during the recent Consumer Electronics Show 2009 in Las Vegas, there was little

discussion about the technology. PowerMat claimed that the innovation is more of the

communication protocol than the wireless power technology. Based on the limited information

provided at CES and the company's website, one can deduce that their technology is also a

variation of magnetic induction similar to Splashpower, eCoupled and WiPower.









WiTricity [13] was developed in a research team led by Professor Marin Solja6ic in MIT.

Two carefully aligned huge capacitively loaded coils with a diameter of 24 inches were used to

transmit power to a 60 W light bulb at 2 m away with an approximate efficiency of 40%. The

operating frequency was at 10 MHz instead of the sub megahertz range. Greater operating range

was achieved by increasing operating frequency and coil size. In addition, the system was

designed for resonance, thus the system will need complex control schemes to seek resonance

under varying load conditions.

Research work on wireless power transmission via magnetic induction for WiPower [14]

started during late 2006 at the University of Florida. Instead of seeking resonance, the system

design is based on the load impedance response taking varying load impedance into account. The

WiPower wireless power system ensures that the system's natural response matches closely with

a typical off the shelf wall wart, ensuring unconditional stability at any load impedance. The

system can be easily scaled to the required power level and power delivery up to 300 W has been

demonstrated.

PowerCast [15] founded in 2003 is the only wireless power company which took the long

range far field approach. Operating at ISM bands of 900 MHz and 2.4 GHz the system is able to

achieve considerable coverage. However, the system suffers from the fundamental problem of

power losses due to radiation resulting in mediocre power delivery. The efficiency is far worse

than Brown's experiment in 1975 where Brown used a focused beam for a point to point RF link.

Since the system requires covering a large area, a point to point system using a focused beam

will not be practical for consumer applications. To date the PowerCast system only has the

power handling capabilities of lighting LEDs or provide supplementary power to low power

wireless sensor nodes.









1.3 Background Information

The purpose of this section is to provide the background information relating to the

material presented in later chapters. This section starts with a discussion the fundamental theory

of operating via phase response and how power control can be achieved by this technique.

Following that a discussion of two popular high efficiency switch mode power amplifier

topologies, the Class D and Class E, which includes their analysis equations are presented. The

discussion will enable the reader to appreciate the advantages and disadvantages of each

topology. Using the analysis, the Class E power amplifier is found to be a more suitable

amplifier than the Class D power amplifier because it is able to achieve high efficiency operation

from a phase angle of 400 to 700 while controlling the power delivered to the receiver.

1.3.1 Fundamental Theory of Operation via Phase Response

To ensure unconditionally stable operation, the system must be able to deliver the required

power with respect to the load resistance closely matching that of a constant voltage power

source. The transmitted power should decrease when the load impedance increases vice versa.

Error in selecting component values or control scheme will result in either oscillation or

undesirable power delivery trend, that is, the transmitter will increase its output when the load

resistance increases. The switching buck regulator at the receiver will attempt to maintain its

power delivery by increasing its input resistance or decrease its duty cycle resulting in a positive

feedback. Poor efficiency will be observed due to excess power dissipated as heat and device

failure may occur due to over-voltage.

There are various methods to achieve the desirable power delivery trend. A closed-loop

system will require some form of control scheme and tunable components/supply voltage as well

as a communication link between the receiver and transmitter. This will cause the system to be

complicated, bulky in size and costly. The optimum solution to the problem is to achieve the










desired power delivery trend with respect to load resistance across a wide range of load

resistances. This can be achieved by designing the system to have an effective resistance

response or phase response looking into the transmitting coil. Since the class E power amplifier,

which has a limited range of operating load resistance, is selected for the system and it is difficult

to generate the desirable effective resistance trend while maintaining a fixed reactance using a

pair of loosely coupled coils, the technique of power delivery control exploiting phase response

will be more appropriate. Figure 1-2 shows the complex plane of inductive and capacitive load.

Since only positive resistive loads are consider, the analysis will only consider the right half

plane. The class D power amplifier [16] and class E power amplifier [18] are not suitable to drive

capacitive loads, thus, the first quadrant of the Figure 1-2 will be the only valid region of

operation.


jX

Increasing phase
angle




Inductive load
region R


Capacitive load
region




Decreasing
phase angle





Figure 1-2. Complex plane of inductive and capacitive load.










j m S


jQ


P Re S


Figure 1-3. Phasor diagram.

Where

P: Real power. Unit: watts (W)

Q: Reactive power. Unit: volt-amperes reactive (VAr)

S: Complex power. Unit: volt-amperes (VA)

|S|: Apparent power. Unit: volt-amperes (VA)

4: Phase angle of the load.

Figure 1-3 shows the phasor diagram to illustrate the concept of power delivery versus

phase response. The real power delivered to the load is reduced when the magnitude of the

apparent is kept relatively the same while increasing its phase angle to be more inductive in the

anti-clockwise direction. The system can make use of such delivery trend to achieve a desirable

power delivery profile with increasing load resistance. The trend can be achieved by using a

series of reactive networks before the transmitting coil and after the receiving coil to shape the

impedance response accordingly. The magnitude of the apparent power can also be varied to

improve the power delivery trend.









1.3.2 Class D Power Amplifier

VDD




Q1
Zout
Clock 1 1 out




Q2

ZLoad

Clock 2


Figure 1-4. Class D power amplifier.

Driving the transmitting coil is the most crucial part of the system. In order to achieve high

efficiency, only switch mode amplifier topologies will be considered for the system. Since the

system should be kept simple with minimum component count, only Class D and Class E power

amplifier topologies are considered. The class D power amplifier will be discussed in this section

whereas the class E power amplifier will be discussed in the following section.

Class D power amplifiers have been traditionally considered as the most practical high-

frequency switch mode amplifiers. The main advantage of a Class D power amplifier is that the

voltage stress across the transistors is equal to the supply voltage making them suitable for either

high voltage or high speed. A circuit of the Class D power amplifier consisting of two transistors

Q1 and Q2 which can be modeled as two bidirectional switches, an LC output filter and a load

ZLoad is shown in Figure 1-4. It was found that the class D power amplifier is able to drive both

resistive and inductive loads but not capacitive loads [17]. The body-drain diode of the MOSFET

can be used as an anti-parallel diode in case of an inductive load.









The analysis of the class D power amplifier is based on Figure 1-4 with the following four

assumptions.

1. The transistor and diode forms an ideal resistive switch. All parasitic capacitance and
switching time are neglected. In addition, the turn on resistance remains constant.

2. The turn on resistance of the Q1 is sufficiently small that voltage drop across Q1 can be
neglected. The turn on resistance ofQ1 can be moved to be part of Zout.

3. The reactive components L and C are not sensitive to frequency.

4. The loaded quality factor of the output LC filter is high enough that the current through
ZLoad is sinusoidal.

The impedance Zout looking into the load network can be described by Equation 1-1


Z =Z +wjL+R +-+
out Load L P arasitic j+ C
R .+R
C P arasitic turn on transistor
=R +R +R + (1-1)
Load L P arasitic C Parasitic
(x C +LC -1
R + +jc9 Load ------
turn on transistor C


According to Figure 1-4, it can be inferred that the input voltage to the load network is a

square wave from ground to VDD neglecting the losses through the turn on resistance of the

transistor. The square wave can be described by Equation 1-2.

V for 0 < ot < z
DD
v (1-2)
out 0 for r < ctK 2


It is assumed that only the fundamental tone is delivered to the load. The fundamental tone

of voltage v,,ut is described in Equation 1-3 from Fourier analysis. The peak voltage is

approximately 0.64 VDD and the RMS voltage is approximately 0.45 VDD.

2V
v DD sinmt (1-3)
out 7









The current through transistor Q1 can be described as


I out sin(mt -) for 0 < cot < (1-4)
i = out (1-4)
0 for r < ct <2i

By integration Equation 1-4 across a single period, we can obtain the DC component of the

supply current.


IDD OU Jsin(wct- )d(t)
I cos2
=out (1-5)

2V R

2Z2 2
D out

If the load is a purely resistive load, the DC supply current can be expressed by Equation

1-5 by setting Z,,ut of Equation 1-6 to Rout.

2V
I DD (1-6)
DD 2
out

The DC input power can be then described by Equation 1-8 from Equation 1-5.


2V 2R
P DD out (1-8)
IN Z 2
out

Therefore, the DC input power of a purely resistive load can be expressed by Equation 1-9

by setting Z,,ut of Equation 1-8 to Rout.


2V 2
P = DD (1-9)
IN z2R
out

Iout can be easily derived from Equation 1-3.










I = DD (1-10)
out out


By using the same concept as above, a resistive loaded class D power amplifier will have

an output current as shown in Equation 1-11.


out 2DD (1-11)
,out RD
out

Power delivered to the load can be then derived by using ohm's law using Equation 1-11.

Since we are only interested in the real power delivery, real power delivery can be calculated by

neglecting all imaginary terms of the apparent power delivery in Equation 1-12.


I Z
P out Load
out(apparent) 2
2V
2V 2Z
DD Load (1-12)

out
2V
2V 2R
p DD Load
out(real) z2 t 2
out

Power loss of the class D power amplifier can be described in Equation 1-13. The three

major factors contributing to the power loss of a class D power amplifier are the parasitic

resistances of the series inductor and capacitor as well as the turn on resistance of both

transistors.


South (L P parasitic RC Parasitic turn -on transistor (1
loss 2

From Equation 1-14, it can be seen that an ideal class D power amplifier is able to achieve

100% efficiency which is desirable for this application.









P
out(real)
P
in (1-14)

Load
L P arasitic C P arasitic turn on transistor Load

1.3.3 Class E Power Amplifier

The class E power amplifier [17] consists of a single switch with a load network. The main

advantage of the class E is that it only consists of a single transistor and does not require a pair of

out-of-phase gate drive for the transistors as shown in Figure 2-3 for the class D. Therefore, a

clock signal with sufficient drivability can be directly fed to the gate of the transistor without a

gate driver to control the driving signal to the high side and low side of the amplifier (Class D).

However, the voltage stress on the transistor of an ideal class E power amplifier is 3.562 times

the supply voltage, making the power amplifier unattractive at high supply voltage. For example,

if the supply voltage of an optimized class E power amplifier is 48V, the drain voltage can swing

up to approximately 170 V. Therefore, by providing a 80% safety margin, one need to use a

transistor with a breakdown voltage of approximately 220 V. In addition, the high drain voltage

might cause higher noise levels in the system due to poor PCB design and if appropriate noise

reduction is not implemented. Reference [17] stated three objectives for the transistor's drain

voltage and current waveform to achieve class E operation.

1. The rise of the voltage across the drain of the transistor should only occur when the
transistor is turned off

2. The drain voltage across the transistor should be zero at the time when the transistor is
turned on. (Zero Voltage Switching operation ZVS)

3. The gradient of the drain voltage should be zero at the time the transistor is turned on.
(Zero Derivative Switching ZDS)










VDD


Figure 1-5. Class E power amplifier.

The detail analysis of an ideal class E power amplifier is based on Figure 1-5 which was

first discussed in [18] [19] by Frederick H. Raab with the following five assumptions.

1. The RF choke LDC is sufficiently large that the current flowing through has negligible AC
components.

2. The series LC filter of LI and C2 has sufficiently high Q that the output current to the
load can be considered to be a sinusoid (single tone). Therefore, it can be assumed that
the filter is an open circuit at higher harmonic frequencies.

3. The shunt capacitance of the transistor is independent of the drain voltage or small
enough to be neglected.

4. The transistor is an ideal switch and has only 3 different states

a. On state with zero resistance.

b. Off state with infinite resistance.

c. Transition state for which the response time is zero.

5. The load will always be a resistive load. Any extra reactance can be added to the series
LC filter.

Using assumption 2 the output voltage and currents across the load should be sinusoidal

(single tone, free of higher order harmonics) and can be described by Equation 1-15 and 1-16

respectively.









Load (t) = c sin(cot + q)


Ld (mt) = R sin (mt + (1-16)
Load

The drain voltage should be also a sinusoid when the transistor is turned off but with a

phase shift due to extra reactance X introduced by the series LC filter. Therefore, the drain

voltage is expressed in Equation 1-17 as the summation of the voltage drop across the load and

the series LC filter.



vdr, (cmt)= csin(mt+ +)+ RL-a sin(mot+ )(1-17)
Load
= csin (t + )

Where,


S~ ~ 1 C2




1
CC2
c =c + qta (C= pc (1-18)





Loao


When switch is off, the drain voltage is produced by the charging of capacitor Cl by the

difference in current between the supply current and the current driving the load network.

Therefore the drain voltage can be obtained by integrating the across the window when the

switch is turned off as shown in Equation 1-20.


(1-15)










drain Ct) C C U
cot

1 (Ot c
J I C-- sin(u +)
JL'> 7Load
12 (1-20)

CC ( c .
c t- + +cos in (( y) +

B 2 BBR
Load Load

Where B is co C and y is half of the transistor turn off time.

The magnitude of the fundamental tone can be obtained by Fourier integral. Since, the

drain voltage waveform is not symmetrical around 7t/2, there will be an unknown phase offset

with respect to the load voltage waveform.

c I =i2zv (cotis cot+ COtd
c1 drain 0m1 \

C -2 \ c ( \
=-2 ry + + zBR sin (y -i) cos sin y
LiB 2Load

+-B -2sinq1 siny+2 +0) cos 1 sin + 2y sin 0lcosy

c [sin(2 + /) sin2y- 2ysin v/
2xiBR
Load

Using Equation 1-21 we can solve for c by substituting pc for ci and rearranging to obtain

Equation 1-22.

2y sin y cosz1l +(2y cosy -2sin y)sin1l
CC Load 1
crBRLoadP + sin (20 + 7) sin 2y -y sin y + 2(y -) cos 1 sin y (1-22)

= ICCRLoadh (q, V, y, B, R, p)









Since the fundamental tone at the drain is defined by a sinewave of phase (pi, there can be

no cosine or quadrature component with respect to phase pi. Replacing the sine term with a

cosine term in Equation 1-21, Equation 1-23 is obtained.


O=-I2v dcott+cos( t 1) daot
= Jf drain (mt)c dot

C y--- 01) sin( ) 2 sin sin y
nBV 2 1 nBR 1
I r Load
,cc o 2z L ad(1-23)
+ -2 cos siny-2 + sin1siny+2ysin 1cosy

c 2 ye cosuy
S- cos (20 + V/) sin 2y + Co
2.irBR 'z-BR
Load Load

Therefore, by rearranging the terms in Equation 1-23 and solving for c, Equation 1-24 is

obtained.

2y sin 1sin y -2y cos q cos y + 2 cos 1 sin y
CC Load 2nq1 -n
-2sin( -y)sin y sin sin 2y cos (2q + /) + y cos y/ (1-24)

C = ICC RLoadg( Oq'

Since both Equation 1-22 and Equation 1-24 describes c, by substituting Equation 1-22

into Equation 1-24, Equation 1-25 is obtained.

g(, V, y) = h(, y, y,B,R, p) (1-25)

In order to solve equation 1-25, we need to cross multiple the denominators of the two

functions g and h and break down the higher order terms of sine and cosine to their first order

equivalent. Since the only unknown in Equation 1-25 is ), after rearranging the terms we can

obtain a solution for 4 by either Equation 1-26 or Equation 1-27 and they provide a single

consistent answer.









(2y sin y cos / + 2y cosy sin / 2 sin y sin /) -2 sin sin2)

+ (2y sin y sin/ 2y cosy cos y/ + 2 sin y cos q/)
(zBRp y sin y/ + sin / sin y cos y)

+(2ysin ysin/ -2ycos ycos/+2sin ycos/) 2 sin2 ycosi

-(2y sin y cos q/ + 2y cos y sin/ 2 sin y sin i/)

-1 (y cosy sin y cosy cosiV/)
(p= tan
(2y sin y cos / + 2y cos y sin V/ 2 sin y sin i)
(zBRp y sin i/ + sin q/ sin y cos y)
+ (2y sin y sin 2y cos y cos y/ + 2 sin y cos q/)
(y cosi sin y cos y cos q/)



(1-26)








(2y sin y sin/ 2y cos y cos y/ + 2 sin y cos /y)

(BRp y sin / + sin / sin y cosy + 2 cos /sin2 y

-(2y sin y cos y/ +2y cos y sin/ 2 sin y sin q/)

tan (y cosy + 2 sin2 y sin / sin y cosy cos yi
= tan
(2y sin y cos y/ + 2y cos y sin/ 2 sin y sin q/)

(zBRp y sin / + sin / sin y cosy + 2 cos /sin2 y

+(2y sin y sin/-2y cosy cos +2sin y cos/) -2sin/sin2 y)

+ (2y sin y sin/ 2y cosy cos y/ + 2 sin y cos q/)

(y cos/ + 2 sin2 ysin/ sin y cos ycos) (1-27)
(2ysincos 2ycossin sinsin sin(1-27)
-(2y sin y cos +2y cos ysin- 2sin y sinI) 2(sin2 ycos ()









Once we obtained 4 from either Equation 1-26 or Equation 1-27, we can obtain either g or

h by substituting 4 into Equation 1-22 or Equation 1-24 to get c. Since we assume that there is no

voltage drop across RF choke, we can assume that the integrated drain voltage across one period

is equal to the value of the DC supply voltage as shown in Equation 1-28 where Rdc is the input

resistance of the class E power amplifier looking into from the supply voltage port.

V I= 02v (ot)dot
cc 2r drain


CC 2_ fy--+gsin(5-y) +ott+gcos(mt+0) ot
27zB z _J 2 (1-28)

I
L 2y2+2ygsin( -y) -2gsin siny
27B \ [
CC dc

Using Equation 1-15, Equation 1-22, Equation 1-24 and Equation 1-28. Power delivery

and efficiency can be derived as followed.

Rd =- 2y2 +2ygsin( -y)) -2gsin siny] (1-29)


1 c2
P -
o 2R
Load
I 2 2R
CC L oad (1-30)
2
V 2 2R
cc Load
2R2
dc

P. VI
I cc CC
V 2 (1-31)
cc
Rd
dc










P
0
P
I
S2g2 (1-32)
cc Load
2R2
dc

Although, it is known that an optimized ideal single-ended class E power amplifier

requires its switching transistor to have a 3.562 times higher breakdown voltage than that of the

Class D, it is often overlooked that the output power of an optimized ideal single-ended Class E

power amplifier as shown in Equation 1-33 from [20] is 2.847 times higher than that of typical

optimized ideal Class D power amplifier as shown in Equation 1-34.

2
8 VC C
P 8
out ClassE z2 4 R

V 2
=0.5768 cc
R (1-33)


v2
2 VCC
out ClassD 2 R

2
= 0.2026 cc
R (1-34)


From Equation 2-33 and Equation 2-34 we can conclude that in order to achieve similar

output power, the Class D power amplifier requires a supply voltage that is 1.687 times higher

than that of Class E power amplifier. When supply voltage is constrained, a Class E transmitter

power amplifier is preferred to a class D transmitter because of higher output power at the same

supply voltage.










CHAPTER 2
WIRELESS POWER TRANSFER SYSTEM

Impedance Impedance
Class E Power
S las wer- transformation transformation -
Amplifier
network network
I Receiver


+ ll 121 + | _2* \ __ Voltage

+ Detection and |
S- Control V, V2





Inductive Coupling Device under
charge/power



Figure 2-1. Wireless power transfer system diagram.

The system diagram of the proposed wireless power transfer system via magnetic

induction is shown in Figure 2-1. Energy transfer between the transmitter and receiver is

achieved by inductive coupling using a pair of air core coils. The class E power amplifier

discussed in Chapter 1 under background information is used as an inverter to convert the input

DC power to high frequency AC power of several hundred kilohertz. In order to achieve the

desirable power delivery trend and impedance looking into the transmitter's load network, an

impedance transformation network is inserted before the transmitting coil and after the receiving

coil. The received AC power is converted to DC by the receiver's rectifier. A full wave rectifier

can be used to reduce device stress on the diodes or a half wave rectifier can be used to reduce

the receiver PCB footprint. Depending on loading conditions, the rectified DC voltage after the

rectifier varies across a wide range. Therefore, a voltage regulator is used to regulate the









fluctuating DC voltage to a stable DC voltage before it is used to power the receiving device. A

detection scheme using a micro-controller and sensing circuits is required to ensure the

transmitter is in its nominal operating conditions. Such a detection scheme is proposed and

discussed in Chapter 7. Control signals can be used to disable the transmitter or tune the system

power delivery profile to enhance system efficiency and robustness. Techniques to control power

delivery profile as well as receiver voltage are discussed in Chapter 8.

2.1 Single-channel Class E Power Amplifier

The class E power amplifier is an elegant design with minimum number of components.

Theory of operation of the class E power amplifier is discussed in Chapter 1. The amplifier

consists of three main sections, the DC feed network (typically a single choke inductor), the

switch (typically a single N channel MOSFET) and the load network. The network from Cl

towards the load in Figure 1-5 can be considered part of the load network and will be discussed

in Chapter 3.

It is preferred that the choke inductor of the DC feed network (LDc) to have a sufficiently

large reactance so that it can be considered as an open circuit for the AC signal. Since

impedances looking into the load network under most operating conditions will not exceed 50 Q,

an effective impedance of 500 Q or better for the DC feed network is sufficient. Therefore, for a

system operating at 250 kHz, the minimum inductance value should be 330 iH. Although, it is

desirable to have the inductance value as high as possible, the parasitic resistance will be large

resulting in poor efficiency and heating issues. In addition, an inductor with a larger inductance

will typically of a larger size making the transmitter bulky.

The switch is the core of the class E power amplifier. Selecting the correct switch is very

important, it needs to be sufficiently fast, have a low turn on resistance and have a sufficiently

high breakdown voltage. To achieve a high breakdown voltage, the most practical candidate for









the switch is the N channel power MOSFET. The theoretical maximum voltage across the switch

for a class E power amplifier is slightly more than 3.5 times of the supply voltage if it is

optimized to drive a fix load. As seen from Figure 3-9 and Figure 3-17, the drain voltage can be

up to 20% higher than the theoretical maximum while driving a varying load. Therefore, the

maximum voltage across the switch should be approximately 4.2 times of the supply voltage

during nominal operation instead of 3.5 times. However, the selected MOSFET should have a

break down voltage of at least 5.5 times of the supply voltage to achieve a 0.8 factor of safety

margin for various fault conditions. A wireless power transfer system with a supply voltage of

15V will require a transistor with a breakdown of 100 V or better. Although, a higher breakdown

voltage will ensure the system to be more robust, a transistor with a higher breakdown voltage

tends to be slower in speed and have a larger turn on resistance. In an ideal case the voltage

across the transistor should be 0 V when a large current is driven across the transistor in its turn

on state. To achieve this operating condition, the turn on resistance of the transistor needs to be

sufficiently small. This also helps in reducing energy being dissipated as heat across the turn on

resistance. A turn on resistance of 0.10 or less is a good starting point The capacitance between

the drain and source of a MOSFET varies with the drain voltage. To ensure than the fluctuation

of drain to source capacitance with respect to the drain voltage can be neglected, the variation

should be much smaller than Cl (typically <5% of Cl). If the gate to source capacitance is too

large, a gate driver is required to provide sufficient drivability to effectively turn on and off the

transistor. Finally, the transistor must be able to switch sufficiently fast to reduce switching

losses, the rise time and fall time of the transistor should be better than 1% of the period of the

operating frequency (better than 40 ns for operating frequency of 250 kHz).










2.2 Multi-channels/Stackable Class E Power Amplifier


VDD



SLDC 1

LI 1 0 C2

Q1_ 1



Clock 1, T 1_




LDC 2
LD -2 ZLoad
L 2


Q1C



| ~ Clock 2


Figure 2-2. Schematic of the dual-channel Class E power amplifier.

In addition to the typical single-ended class E transmitter, a multiple channel class E

transmitter with independent gate drive is shown in Figure 2-2 (Dual-channel). The dual-channel

class E allows the system to shut down one of the channels to switch to a lower power mode so

that the receiver's regulator responds by reducing its input resistance. The analysis of the dual-

channel class E is straightforward. If both the channels are enabled, the equivalent inductance

across both L1 is L1/2, while if one channel is enabled, the equivalent inductance is simply L1

making Z,,ut more inductive while keeping the real components of Zout the same. If Zout is more

inductive, more reactive power and less real power is transmitted to ZLoad reducing power

delivery. Likewise, the equivalent value of Cshun,,t is doubled for a dual-channel configuration.









The dual-channel class E transmitter can be easily scaled to multiple channels, but the size

increases along with a higher probability of device mismatches. As the number of channels

increases, the problems caused by device mismatches might far outweigh the benefits. It was

found empirically that the number of channels should be kept at a maximum of three. The

observed efficiency improvement was negligible when the number of channels was increased

from three to four and the efficiency started to degrade at five channels, predominately due to

variations in inductance value. Although this technique enables the system to operate at a higher

efficiency, the tradeoff is that the power delivered to the load network is fed across a single

inductor instead of two for a dual-channel case, thus the equivalent parasitic resistance of Lo,, is

doubled for a dual-channel topology, reducing the potential enhancement of the system

efficiency as power loss via the inductor remains the same in regardless of the number of

channels.

2.3 Inductive Coupling

Power transfer for the system is achieved via magnetic induction between two air core

coils. Vertical magnetic field is the dominate component of power transfer because both the

transmitting and receiving coils have large surface area. By having a larger surface area, the coil

is able to intercept more of the magnetic flux and receive more power. In addition, modern

portable electronic devices are extremely thin in profile having limited cross sectional area for

any magnetic flux to cut through the cross sectional area. Appropriate shielding [21] at the

expense of weight and thickness can be used to make the system more robust in environments

where the system's magnetic field is likely to interact with other nearby objects. However,

shielding is not the focus of this dissertation and it is assumed that the system will be working in

an environment free of structures and materials that will significantly affect the system

performance or is sufficiently shielded to begin with.









Although it would be ideal for both the transmitting coil and the receiving coil to be of the

same size to ensure maximum coupling as shown in [22] [24], a practical system will tend to

use a receiver coil significantly smaller than the transmitting coil. This allows a user to freely

place a device in any orientation without intentionally "docking" it as well as to place multiple

devices on the transmitting coil. Therefore, in order to achieve consistent power delivery and

impedance response regardless of the receiving coil's location, the transmitting coil must be able

to generate a magnetic field that has relatively even distribution. This is achieved by using the

method proposed in [25]. Alternative methods are also proposed in [26] [27].

The voltage and current of the transmitting coil and the receiving coil can be described

using Equation 2-1, Equation 2-2 and Equation 2-3 from [22] [27] neglecting any second order

effects such as skin depth and proximity effect which can be mitigated by using Litz wires.

V jwM11I1j + jwM12I2 (2-1)

V2 = jmM21I + jmM22 2 (2-2)


M 12 = k M1M22 (2-3)

Where

Vi is the voltage at the transmitting coil. (Figure 2-1)

11 is the current at the transmitting coil. (Figure 2-1)

V2 is the voltage at the receiving coil. (Figure 2-1)

'2 is the current at the receiving coil. (Figure 2-1)

M11 is the self inductance of the transmitting coil.

M22 is the self inductance of the receiving coil.

M12 = M21 is the mutual inductance of the two coils.

k is the coupling coefficient between the two coils.









By Ohm's law:

Z. =R + jX
tx tx tx
V_ (2-4)
I1

Z R + jX
rx rx rx
V2 (2-5)
12

Using Equation 2-1, Equation 2-2 and Equation 2-4 and assuming a time-harmonic

operation with frequency co, impedance looking into the transmitting coil for a single receiver is

derived as Equation 2-6.


C2M2R Co 2 coM + X
tx 2 2 11 R2 )2
rx(\ 22 rx) rx OM22 +rx (2-6)

2.4 Impedance Transformation Network

The purpose of impedance transformation network on the primary and secondary sides of

the coupling is to achieve maximum power transmission and efficiency by operating within the

optimum impedance range looking into the transmitter load network [29] over a wide range of

load resistances.

In consideration of size and efficiency, capacitors instead of resistors and inductors should

be used for the network. This is because resistors dissipate power and the size of a low loss

inductor is generally large. Although, a multi-element transformation network might achieve a

better response, for simplicity and low component count, the system uses a single-element

transformation network. The four possible topologies of the single-element transformation

network are shown in Figure 2-3.









I--| --- r ---- --- 0 0--11 --- --|--
Cx Crx Ctx Cx
0 ----- J _----- 0 0 ---- ) --I-0
A B
0 -i-- o 0
Ctx C x Ctx C x

C D

Figure 2-3. Topologies for a single-element impedance transformation network. A) Series-Series
Topology, B) Series-Parallel Topology, C) Parallel-Series Topology and D) Parallel-
Parallel Topology

Fundamentally, a series capacitor only introduces a negative reactance and does not change

the real part of the impedance. On the other hand, a parallel capacitor changes both the real and

imaginary parts of the impedance. To simplify the analysis, the receiver input impedance is

modeled using a variable resistor load and Equation 2-7 illustrates the transformation performed

by the parallel capacitor.

R aC R 2
Z rx rx rx
rx l+m2C 2R 2 1+2C 2R 2 (2-7)
rx rx rx rx

Equation 2-7 shows that the resistance R, is "compressed" by a factor of

1 / (1 + c2Crx2R2) Thus, the equivalent resistance R, decreases with increasing load

resistance. At high load resistance, the transformed resistance is small. Therefore, a significant

part of the received power is dissipated across the receiving coil as heat. This phenomenon is

desirable if the receiver is in a state that requires very little power or during trickle charge.

Therefore, it has a decouplingg" effect regulating the power delivery with increasing load

resistance. However, this should occur if and only if the transmitter is designed to output limited

power under this operation condition because heating can become a problem if too much power









is being dissipated across the receiving coil. Due to the parallel capacitor, a reactive term jXrx is

introduced to the equation. The reactive term j Xrx decreases nonlinearly from null with

increasing load resistance with an asymptote of -1/ )C,r This is used to compensate the

receiving coil's self inductance.

From Equation 2-6 it can be observed that the resistance looking into the transmitter coil is

reduced significantly if the resistance looking from the receiver coil into the receiver is

increased. Rt, is further reduced because the mutual inductance is relatively small. If the total

resistance looking into the transmitting coil is mainly the parasitic resistance of the transmitting

coil, limited power is transmitted to the receiver as most of the power is dissipated across the

transmitting coil as heat. Therefore, it is preferred for a wireless power transmission system

using loosely coupled coils to have a parallel capacitor across the receiving coil. By substituting

Equation 2-7 into Equation 2-6, the expression of impedance looking into the transmitting coil

with a parallel capacitor across the receiving coil is shown in Equation 2-8.



z 12 12 +0l+2CR 2R2 +
tx 2 ) r2
R m2 oCR
I+02C2RM2 K 22 1 +2 2R2


O2M2 fM CR
12 22 1+22R22

j O2 11 2
SR m oM )CR (2-8)
+2 C w2R2 )M22 l+oC2R2 j (2-8)

For the transmitter transformation network, a series or parallel topology can be used. To

maintain an ideal efficiency above 95%, the allowable variation of load resistance of an ideal

class E amplifier should be kept within +55% and -37% [18]. If the variation of resistance









looking into the transmitter coil is too large, it is preferred that a parallel capacitor is used instead

of a series capacitor to "compress" the resistance. A suitable capacitor value is needed to ensure

that the transmitter does not suffer immediate failure when there is no receiving coil, and to

produce an increasing reactance trend with increasing load resistance so as to ensure the

preferred power delivery trend.

On the other hand, an appropriate receiver capacitance value can also be selected so that

the resistance looking into the transmitting coil is kept within a reasonable bound but not too

small to impact on the coupling efficiency and to produce an increasing reactance trend with

increasing load resistance so as to ensure the preferred power delivery trend. A suitable series

transmitting capacitor is then needed to translate the reactance looking into the transmitting coil

to cancel part of the self inductance of the transmitting coil. Based on the above analysis, both

the parallel-parallel and series-parallel impedance transformation network topologies can be used

for the system.

2.5 Receiver

Since more receivers will be paired up with a single transmitter and most receivers are

intended to be integrated into compact portable electronics, the receiver of the wireless power

transfer system needs to be low cost and compact in size. Therefore, the receiver will only

consist of a rectifier to convert the AC power to DC power and a voltage regulator to ensure a

stable DC voltage is use to power the device. The effort of attaining universality for cell phone

power port still faces huge resistance from the OEMs. By having a proprietorship

communication protocol for a wireless power transfer system will work against a universal

charging/power platform to reduce waste and power consumption. Therefore, a communication

link should be avoided unless it is the last resort and the wireless power receiver is integrated

into the device tapping into its existing wireless communication system.









Reverse recovery time of the diodes used for the rectifier is critical because the operating

frequency is much higher than the typical 50/60 Hz AC power lines. A good rule of thumb is to

select a diode with a reverse recovery time less than 1% of the operating frequency's period. For

example, a 200 kHz system should use a rectifying diode with a reverse recovery not slower than

50 ns. It is easy to find schottky diodes with negligible reverse recovery time at breakdown

voltages below 100 V for low to medium power applications (< 100W). However, for high

power applications either an ultrafast recovery diode or schottky diodes in series are required to

ensure they are not damaged by the reverse biased voltage.

20-

10--

0-

S -10

-20

1 -30

-40 1 1 1 1 1
1.830 1.831 1.832 1.833 1.834 1.835 1.836 1.837 1.838 1.839 1.840

time, msec

Figure 2-4. Input voltage of a half wave rectifier and a full wave rectifier.

Figure 2-4 shows the input voltage of both a half wave rectifier and a full wave rectifier

where the flat section of the waveform is when the rectifying diode conducts charging the

charging holding capacitor used to smooth out the voltage ripples at the output of the rectifier.

The diodes for a full wave rectifier conducts during both the positive and negative cycle of the

waveform whereas the diodes for the half wave rectifier only conducts during either cycle

depending on the orientation of the diode. For the specific load condition shown in Figure 2-4,









the full wave rectifier requires a diode with a breakdown voltage of 20 V and better to operate.

On the other hand the half wave rectifier will require a diode with a breakdown voltage of 40 V

and better to operate which is a lot higher than a full wave rectifier. Although, a half wave

rectifier helps to reduce the size of the receiver PCB by reducing the number of diodes from four

to one, it will require a diode of a much higher voltage rating to operate. Under most situations, a

full wave rectifier is preferred for high power applications where a slight increase in PCB size is

not an issue.

A switching buck regulator is preferred over low dropout regulators for this application

because the input voltage swings across a considerable range and high efficiency must be

maintained. In order to keep the receiver compact the output inductor of the switching regulator

should be kept below 10 iH, which can be achieved by selecting regulators with switching

frequencies above 500 kHz. Although, increasing the switching frequency will result in the

voltage regulator being more compact, the efficiency starts to degrades losing power to heat.









CHAPTER 3
DESIGN OF IMPEDANCE TRANSFORMATION NETWORK

The design of the proposed wireless power system starts by setting constraints of the

dimensions of the transmitting and receiving coil as well as operating frequency. This chapter

presents the design rules for two systems using the parallel-parallel impedance transformation

network and the series-parallel impedance transformation network achieving desirable power

delivery profile. The parallel-parallel impedance transformation network design rule is based on

the dual-channel transmitter topology to achieve high power transfer while the series-parallel

impedance transformation network design rule is based on the single-channel transmitter

topology to achieve compact low power design. A different operating frequency is used to

illustrate that the design rule applies to a wide frequency range.

3.1 Series-Parallel Impedance Transformation Network

3.1.1 Introduction

An operating frequency of 240 kHz is used for the design of the series-parallel impedance

transformation network topology. Since high efficiency schottky diodes are used for the receiver

rectifier, the resistance looking into the rectifier should be very close to the resistance after the

rectifier as very little energy is lost. Therefore, to simplify the analysis, the load resistance is

defined as the equivalent resistance looking into the rectifier instead of after the rectifier. It is

desirable for the receiving coil to be much smaller than the transmitting coil, but efficiency and

power transfer capabilities start to degrade significantly due to poor coupling if the receiver is

too small. Therefore, it is preferred to keep the coupling coefficient k above 0.1. To minimize

space usage as well as ease of integration into the target device, the receiving coil is typically

tightly wound. However, due to the requirement of ensuring a consistent coupling coefficient

regardless of position and orientation the windings of the transmitting coil are very different










from the receiving coil. The gaps between each turn of the transmitting coil are spaced in a non

uniform manner with the windings more spaced out when it gets to the middle of the coil so as to

achieve even field distribution. Thus, achieving consistent performance regardless of the

placement of receiving coil.

2o, Is


E 12
0
C.O

'I)
0
CL8


0 2 4 6 8 10 12 14 16 18 20
x position (cm)

Figure 3-1. Windings of a 20cm x 20cm transmitting coil used for experimental verification.

Figure 3-1 shows the windings of a 13 turns 20 cm x 20 cm transmitting coil from [25]

which will be use for later experimental verification, the last 3 turns are overlapped onto each

other. Designing of the transmitting coil windings to achieve an even magnetic field distribution

is not the scope of this dissertation. The normalized power delivery with respect to location of

the transmitting coil in Figure 3-1 using a tightly wounded receiving coil of 9 cm x 6 cm with 6










turns is shown in Figure 3-2 using the centre of the receiving coil as reference. The receiver is

13.5% of the size of the transmitter. Power delivery variation is kept within 5% with a standard

deviation of 2.2%. Since there is no obvious distribution of power delivery trend with respect to

receiver location, it can be assumed that the generated magnetic field is even and the variations

are due to measurement errors.

20 I I I 1.05


18- 1.04


16 1.03


14- 1.02


E 12 -l1.01

2 lor -1

0
L8 8- 0.99


6- 0.98


4 0.97


2 0.96


0 1 I 0.95
0 2 4 6 8 10 12 14 16 18 20
x position (cm)

Figure 3-2. Normalized power deliver with respect to location of transmitting coil in Figure 3-1
using a receiving coil of 9 cm x 6 cm.

Key parameters of the coils including self inductances, mutual inductance, and parasitic

resistances can be extracted by measuring the fabricated coil with an impedance analyzer or

analyzing with electromagnetic simulation tools. The coils were fabricated using 100/40 round

served Litz wires for the experiment to mitigate proximity effect and skin effect. The 100/40

round served Litz wires consists of 100 strands of 40 gauge wires insulated from each other. The









self inductance of the transmitting coil is 45.3 pH with a parasitic resistance of 0.5 Q. The self

inductance of the receiving coil is 5.2 .iH with a parasitic resistance of 0.1 Q. Mutual inductance

between the coils is 2.8 .iH with a coupling coefficient of 0.1824. Measurements of the coils

were measured using the HP4192A LF Impedance Analyzer. The design rule for the series-

parallel impedance transformation network will be based on Figure 3-3. Although C,,t and Ct,

are two separate capacitors, Ct, will be considered to be part of C,,t.




1+ -L 7, co --------c

S-txcoil Z-rx
SLout

shunt Ctxc i R/oad
AC


Transformation network /


Figure 3-3. Simplified schematic of wireless power transfer system using series-parallel
impedance transformation network and Class E transmitter. Zt, Impedance looking
into the transmitter load network. Ztxcoil Impedance looking into the transmitting
coil. Zrx Impedance looking into receiver network. RLoad is the equivalent
resistance looking into the rectifier.

3.1.2 Determination of Cr value

The design of the system starts from the receiver looking into the load. The typical

impedance response for different parallel capacitors is shown in Figure 3-4. The capacitance

value is selected based on the inductance of the receiving coil as well as the mutual inductance

between the coils. Although, it will be desirable to achieve a maximum resistance looking into

the transmitting coil across a wide range of load resistances [29], the resistance variation looking

into the transmitting coil might be too large, thus requiring a shunt capacitor across the









transmitting coil to "compress" the resistance resulting in the parallel-parallel impedance

transformation network topology which will be discussed later. Therefore, it will be more

practical to select a capacitor value that will generate the desirable resistance range looking into

the transmitter load network Ztx as shown in Figure 3 -3.

,12
10
S-C = 5OnF
S6 ----C- = 100nF
S4 .* ,...C = 150nF










-20
0 200 400 600 800 1000
Load Resistance (Q)

-5
S-10V
c --------- ...C = 100nF
t \1... .C = 150nF
W-20-
_'-0 200 400 600 800 1000
Load Resistance (Q)

Figure 3-4. Resistance and reactance of Z, versus load resistance at different with different C,.
(50 nF, 100 nF and 150 nF)

Shifting of the reactance value to achieve a desirable phase response so as to achieve a

desirable power delivery profile can be done by varying Cout or Lout in Figure 3-3. In order to

determine the range of resistance looking into the transmitting coil an appropriate Lout value

needs to be selected first. The class E transmitter requires a minimum loaded Q of 1.7879 to

operate [20]. There are two factors that affect the decision of Lout. For the same loaded Q value it

will be desirable to have Lout as large as possible so that the resistance looking into the

transmitting coil will be larger. Therefore, the parasitic resistance of the transmitting coil can be

neglected. However, if Lout is too large the parasitic resistance of the inductor will be relatively









large unless a better inductor of lower parasitic resistance is used. However, when the parasitic

resistance of the inductor of the same inductance value gets lower, the size of the inductor also

gets bigger. The lowest loss inductor will be an air core inductor using Litz wire but its size will

be larger. In addition, based on Equation 3-1 if the resistance looking into the transmitting coil is

too large limited power will be delivered to the receiver. On the other hand, if Lout is too small

the maximum value of the resistance looking into the transmitting coil will be limited. Therefore,

with a small resistance looking into the transmitting coil, the parasitic resistance of L,,ut and the

transmitting coil will get more significant affecting the system efficiency and power delivery as

most of the voltages will drop across the parasitic resistance and not across the reflected

resistance by the receiver looking into the transmitting coil.


P= --cos0


V2R

Z 2 (3-1)
tx

For an operating frequency of 240 kHz the system will be able to operate well with a Lout

value from 6.8[tH to 22[tH depending on the parasitic resistance of the transmitting coil, the

parasitic resistance of L,,ut and C,,ut as well as size constrain. For this design a 10 [iH inductor

(RL-5480-5-10 from Renco) is selected. The inductor has low loss, 0.16 Q of parasitic resistance

at 240 kHz and is considerable small in size (15.875 mm diameter and 17.78 mm height).

However, due to de-rating at higher operating frequency the effective inductance of the inductor

is 9.5 [tH instead of 10 [tH at 240 kHz. The inductance will further decrease with higher current

and temperature. This is predominately due to the temperate sensitive nature of most ferrous

cores used in inductors.










ZUIl 86.4nF-in-resonance


86.4nF in resonance
10- >
with receiving coil
00

90-

80-

70 -

60-

50-
First solution
40 at73nF


0 10 20 30 40 50 60 70 80


90 100 110 120 130 140 150 160 170 180 190 200

Crx (nF)


Figure 3-5. Peak resistance response looking into the transmitting coil with respect to Crx.


S31--------------------------03------------------------
80- ,-80 -
75 75-
70/ 70-
S65 65.
60- 60 -
U 55 55
50 50-
r 45- -Resistance 45 -Resista
Z. 40- *--Reactance .40- Reactar
u 35- 35
S 30- 30
*5i 25 ; 25
AA
S20 20
15c- p 15-
10- 10

0 O0
0 200 400 600 800 1000 0 200 400 600 800
Load Resistance (Q) Load Resistance (Q)

A

Figure 3-6. Resistance and reactance looking into the transmitting coil. A) with receiver
capacitance of 73 nF and B) with receiver capacitance of 97 nF.


nce
ice










1000


B


-"""Max resistance looking into the TX coil
-8Q resistance line


Second solution
at 97nF


..........


y . .,,,~~~~~~~I~


I










Although both receiver capacitance values provide the same resistance trend looking into

the transmitting coil, the reactance trend is different. Using a capacitance value of 73 nF before

the resonance capacitance value of 86.4 nF results in an increasing trend of reactance with

increasing load resistance converging at approximate 82 Q. On the other hand, a capacitance

value of 97 nF will result in a decreasing trend of reactance with increasing load resistance

converging at approximately 51 Q. According to Equation 3-1, increasing the reactance while

keeping the resistance relatively the same will decrease the power delivery. Therefore, in order to

obtain the desirable trend of decreasing power delivery with respect to increase load resistance

the first solution of 73 nF before the resonance capacitor value with the receiving coil is selected.

100oo




70






LU
30 -

20-
..Transmicter Efficiency
10 ***Receiver Efficiency
Total Efficiency
0 100 200 300 400 500 600 700 800 900 1000
Load Resistance (Q)

Figure 3-7. Coupling efficiency with respect to load resistance.

Based on the selected receiver capacitance value, the efficiency of the coupling with

respect to load resistance shown in Figure 3-7 can be calculated using the parasitic resistance of

the coil. Coupling efficiency peaks at close to 90% at a load resistance of 30 Q. Although

efficiency rolls off to an approximate 36% at 1 kM, power delivered at the resistance is










extremely low. The gradual degradation in receiver efficiency is desirable as it helps to regulate

the power during trickle charge. Power delivered by the transmitter remains consistent at high

load resistances because the equivalent load impedance Zxt looking into the transmitter load

network does not change much.

3.1.3 Determination of Cout value





80 --C





0 -C =5n

0 out nF

C0ut = 1nF

--C =9lnF
00



=12nF

0 100 200 300 400 500 600 700 800 900 1000
Load Resistance (Q)

Figure 3-8. Ztx phase response with respect to load resistance for various C,,ut capacitance values.

Figure 3-8 shows the phase response of Ztx with respect to load resistance for different C,,ut

values with increasing phase angle with increasing capacitance value. The Class E power

amplifier does not perform well under capacitive loads and high efficiency is achieved at a range

of phase angle from 400 to 700 [18]. Therefore, any value 7 nF and above can be used for Cout.

Since the coil inductance is large, the phase response is sensitive to the component values. This

can be verified by the large phase response swing observed in Figure 3-8 when C,,ut is changed

from 6 nF to 7 nF. Since C,,ut should be selected to achieve maximum power delivery and










stability, C,,ot is selected to be 8 nF because the variation in phase response when C,,ut changes

from 7 nF to 8 nF and from 8 nF to 9 nF is less than from 6 nF to 7 nF. This ensures that the

fluctuation in power delivery due to component tolerance is limited. However, a higher Cout can

be selected to limit the power delivery as shown in Equation 3-1.

3.1.4 Determination of Cshunt value

Once the values of the inductors and capacitors in the transmitter load network and the

receiver network are determined, the remaining step is to determine Cshunt to achieve ZVS and

ZDS operation so as to minimize switching losses. The optimum Cshunt value can be determined

using the equations derived in Chapter 2 and [18]-[19] which are implemented in Matlab code.

The optimum Cshunt is found to be 10 nF and the variation of transistor drain voltage versus load

resistance is shown in Figure 3-9. It can be seen that the transistor drain voltages are kept very

close to zero when the transistor is being switched on at a phase of 1800. In addition, the

negative voltages do not occur because the built in diode will start to conduct and restrict the

voltage at the negative of its turn on voltage which is around -1.3V.



0.8

MW0.6





S-Lca.J resist=lin.:e Oomn
E -Load resistance 40ohm
o 0-
E -Load resistance 60ohm
Load resistance 80ohm
-0.2- -Load resistance 100ohm
Load resistance 250ohm
-Load resistance 1000ohm
0 20 40 60 80 100 120 140 160 180
Phase (Degrees)


Figure 3-9. Transistor drain voltage waveform for different load resistance. (Cshunt =10 nF).









3.2 Parallel-Parallel Impedance Transformation Network

3.2.1 Introduction

Instead of 240 kHz, an operating frequency of 134 kHz is used for this design rule.

Operating frequency should not affect the design rules unless it is extremely high, for example

more than 5 MHz. Similar to the series-parallel impedance transformation network design rule,

the load resistance is defined as the equivalent resistance looking into the rectifier instead of after

the rectifier. Key parameters of the coils including self inductances, mutual inductance, and

parasitic resistances can be extracted by measuring the fabricated coil with an impedance

analyzer or analyzing with electromagnetic simulation tools. A pair of coils was fabricated using

16 AWG magnetic wires instead of Litz wires for the experiment. The transmitting coil is 21 cm

by 21 cm with 10 turns while the receiving coil is 13 cm by 13 cm with 5 turns tightly wounded

together. The transmitting coil is designed with the appropriate spacing between the turns to

achieve 5% variation of the received power at all different locations. Therefore, it can be

reasonably assumed that the coupling is constant regardless of receiving coil position. The self

inductance of the transmitting coil is 31.95 [iH with a parasitic resistance of 0.30. The

receiving coil is 12.52 [iH with a parasitic resistance of 0.2. Mutual inductance between the

coils is 7.454 [iH with a coupling coefficient of 0.373. Measurement of the coils was measured

using the HP4192A LF Impedance Analyzer. Since the coils are fabricated using magnetic wires

instead of Litz wires, the actual parasitic resistance of both coils during power transfer will be

larger than the low voltage signal measured values. This is due to both proximity and skin depth

effects. The design rule for the parallel-parallel impedance transformation network will be based

on Figure 3-10 which has an extra capacitor Ctx shunt across the transmitting coil relative to

Figure 3 -3.





















ACt





Figure 3-10.


Cshunt ""'- I Rload




Transformation network

Simplified schematic of wireless power transfer system using parallel-parallel
transformation network and Class E transmitter. Zt, Impedance looking into the
transmitter load network. Ztxcoil Impedance looking into the transmitting coil. Zrx
- Impedance looking into receiver network. RLoad is the equivalent resistance
looking into the rectifier.


3.2.2 Determination of C, value


25 50 75 100 125 150
Load Resistance (Q)


175 200 225


Figure. 3-11. Optimum receiver capacitor value versus load resistance.


135

132.5

u. 130

^ 127.5

-o 125

r 122.5

E 120
E 117.5

O 115
112.5


110i


I .












-


250


I









The capacitance value is selected based on the inductance of the receiving coil as well as

the mutual inductance between the coils. It is ideal to achieve a maximum resistance looking into

the transmitting coil across a wide range of load resistances. However, in reality, the optimum

capacitance value is different for their respective load resistances. In addition, there is no closed

form analytical solution. Differentiating the real part of Equation 2-8 and solving for the

optimum receiver capacitance value is not straightforward. Therefore, a simple parameter sweep

of Matlab code is used to sweep through a range of receiver resistances and capacitance values to

extract the optimum value.

Based on the coil parameters, Figure 3-11 shows the optimum Crx value versus load

resistance. The optimum Crx value decreases rapidly from 135 nF to 113 nF with increasing load

resistance. The typical operation of a switching voltage regulator used for this application does

not present a very low resistance at its input. For example, it can be safely concluded that to

power a typical USB enabled device at 5 V, 500 mA (input resistance of 10) the regulator

input resistance should not drop below 250 by assuming that the regulator has 100% efficiency

and the minimum input regulation voltage is 8 V. Since it is likely that the regulator will operate

with load resistance between 25D to 100 Q during high power transfer, it is be important to

achieve high efficiency across this impedance range. 113 nF is chosen as the preferred receiver

capacitance value.

Figure 3-12 shows the coupling efficiency between the coils and the impedance looking

into the transmitting coil (Ztxcoil). The efficiency is calculated using the ratio of the power

delivered to the resistance Rtxcoil over the power delivered to both the parasitic resistance and the

effective resistance Rtxcoil. The above method is used to determine the transmitting and receiving

coil efficiencies. The coupling efficiency is the combination of both the transmitting coil and









receiving coil efficiencies. It can be seen that the efficiency of the transmitting coil remains high

for all cases, which is desirable. This is because the transmitter puts out a large amount of power

and should have a higher efficiency to mitigate heat loss. The gradual degradation in receiver

efficiency is desirable as it helps to regulate the power during trickle charge. This can be seen in

later analysis that the power delivered by the transmitter remains consistent at high load

resistances. This is because the equivalent load impedance Ztx looking into the transmitter load

network does not change much.

-100 .-- -------------- ------- --------------- -------

'80 8
c 70
0) ---Transmitting coil efficiency
60 B -""Receiving coil efficiency
U -Coupling efficiency
O 25 50 75 100 125 150 175 200 225 250
Load Resistance (Q)
70
760 --- R
5 0 ,X

U 3 tx -



0 25 50 75 100 125 150 175 200 225 250
Load Resistance (Q)

Figure 3-12. Coupling efficiency and transformed impedance looking into the transmitting coil.

3.2.3 Determination of Ctx value

Selecting an appropriate transmitting coil parallel capacitor value requires to fulfill two

constraints. First, the system must not fail catastrophically when the receiving coil is removed

from the transmitting coil (Zt, should not be capacitive and preferably has a phase angle of

greater than 400). Although it is possible to implement a load detection scheme as discussed in









Chapter 7 to turn off the transmitter and reduce unloaded power losses, it is still desirable for the

unloaded power consumption to be minimal to begin with. By repeatedly stressing the

components with excessive voltage and current will cause them to have a shorter life cycle. In

addition, lower power consumption during unloaded condition, will further reduce power

consumption during the standby mode.

Limiting unloaded power loss can be achieved by ensuring the unloaded Ztx has equivalent

impedance similar to the case with a high load resistance (high impedance with large phase

angle). From the schematic of the Class E circuit in Figure 3-10, it can be deduced that most of

the power loss is due to the transmitting coil and inductor parasitic resistances as they are in the

path of power transfer. Therefore, one way to reduce the unloaded power loss is to use an

inductor with lower parasitic resistance or Litz wire for the transmitting coil.

For the second constraint, the reactance of the transformed impedance must have an

increasing trend with respect to the load resistance in order to achieve an increasing phase

response. Harmonics rejection as well as any phase shifting to bring the impedance of the

transmitter load network to the appropriate range is realized by Cout and Lout. Cout and Lout are

selected based on the operating frequency. Since inductors are typically larger than capacitors, it

is not recommended to put more than one inductor on each channel. For the selected operating

frequency, L,,ut is selected to be 100 |iH (50 |iH equivalent single channel) with a parasitic

resistance of 1.30 at 134 kHz and C out is selected to be 68nF. Since inductors have typically

poorer tolerance than capacitors, they can be placed next to each other so that the mutual

inductance between the inductors will force the current in the inductors to be synchronized. C,,ut

can be tuned to vary the power delivery profile after the design is completed. If C is too low Z
out tx









phase will be too small affecting the ZVS operation. If C,,ot is too high the large Ztx phase will

limit power delivery.

80
S-C = 10 nF
Ci
60 -C = 20 nF
Cr40 -C = 30 nF
40
S-C= 40 nF
S20 -C = 50 nF
o L\ -C = 60 nF
N /-C N = 80 nF

4--
0 -20 -CN = 100 nF

C -40
-5
U -60

-8
0 200 400 600 800 1000
Load Resistance (Q)

Figure 3-13. Reactance of Ztxcoil versus load resistance with different Ctx.

120

90------- ......---------------------------

60





)W 0
C -30-
E


---Transmitter load network phase
-Transmitter load network impedance
150 200 2!
C (nF)









Figure 3-14. Amplitude and phase of impedance of unloaded Ztx versus Ctx.

From Figure 3 -13, it can be determined that Ctx must be at least 60 nF to ensure proper

phase response. Figure 3-14 shows the amplitude and phase of the unloaded Zt, with different

Ctx. Since the class E power amplifier does not work well when driving capacitive load [18] and

might even result in a system failure due to heating, Ct, values between 44 nF to 93 nF should be

avoided. Based on both conditions, it can be concluded that Ctx should be above 93 nF. Although

it would be ideal to have a large Ct, value so that the unloaded power loss is minimized, the

variation of the load network reactance decreases, as seen in Figure 3-13, when the value of

capacitance increases. If the variation of both resistance and reactance are small, the phase shift

across the load resistance is small, resulting in little variation in the power delivery versus load

resistance. If the power delivered to the receiver is not reduced to a manageable level at high

load resistance, power will be dissipated in the receiving coil creating heating problems. Based

on a minimum magnitude of unload Zt, of 10 Q, a capacitance value of 105 nF is selected for Ctx.

14 I I I I I
-R
-12 ---X "

010

C8

I R








10 25 50 75 100 125 150 175 200 225 250
Load Resistance (Q)

Figure 3-15. Rt, and Xtx versus load resistance. (Ctx: 105 nF)










I I I I I I I I


25 50 75 100 125 150
Load Resistance (0)


175 200 225


Figure 3-16. Phase of Zt versus load resistance. (Ctx: 105 nF)

3.2.4 Determination of Cshmunt value


-Load resistance 20Q
-Load resistance 40Q
-Load resistance 60Q
-Load resistance 100Q
-Load resistance 250Q
-Load resistance 500Q
-Load resistance 1000Q
40 60 80 100
Phase (Degrees)


Figure 3-17. Transistor drain voltage waveform for different load resistance. (Cshunt = 19 nF).


0n


a)

a)

I-r
65
-o

(0

CL55


250


W 0.

0



o 0.*
-0
N
(U
E
0
Z


180


LLil


.~-I


0









The method to determine Cshunt is the same as the design rule of the series-parallel

impedance transformation network topology. The optimum Cshunt is found to be 19 nF and the

variation of transistor drain voltage versus load resistance is shown in Figure 3-17. If the system

is to be reduced to a single channel topology the optimum Cshunt value will be double of 19 nF,

38 nF. On the other end the optimum Cshunt value for a triple channel topology will be 12.67 nF.

The optimum value of Cshunt can be easily scaled to the desired number of channels.









CHAPTER 4
WIRELESS POWER TRANSFER SYSTEM SUPPORTING MULTIPLE RECEIVERS

4.1 Inductive Coupling

The analysis of a 1:N coupling structure for a single transmitting coil delivering power to

multiple receivers extends from the analysis in Chapter 2. Since the receivers are intended to be

integrated into portable devices, it is highly unlikely that the receivers will be overlapped. In

addition due to the physical constraint of the devices, they should be sufficiently spaced apart.

Therefore, the mutual inductance between the receiving coils can be neglected as the coupling

between the receiving coils will be significantly weaker than the coupling between the

transmitting coil and receiving coils.

The voltage and current characteristics of the transmitting coil and X number of receiving

coils can be described using the following equations:

X
V jcM llI + jc M NIN (4-1)
N=I

VN = jmNM I + jM NNIN (4-2)


MIN = kN M 1MNN (4-3)

Where

Vi is the voltage at the transmitting coil. (Figure 1-1)

Ii is the current at the transmitting coil. (Figure 1-1)

VN is the voltage at N receiving coil. (Figure 1-1)

IN is the current at N receiving coil. (Figure 1-1)

M11 is the self inductance of the transmitting coil.

MNN is the self inductance of N receiving coil.









M1N = MN1 is the mutual inductance of the transmitting coil and Nth receiving coil.

kN is the coupling coefficient between the transmitting coil and Nth receiving coil.

X is the total number of receivers.

By Ohm's law:

Z R +jX
tx tx + tx

V (4-4)



rxN rxN rxN
VN (4-5)
N

Using Equation 4-1, Equation 4-2 and Equation 4-4 and assuming a time-harmonic

operation with frequency co, the impedance looking into the transmitting coil for multiple

receivers is derived as Equation 4-6.

tx

N= R2 +(coM +X ,
rxN \ NN rxN
(4-6)
X c02M2 (oMM +X N)
+j coMI IN \ NN rxN 2
N= R2 + M O +X + )
N rxN \ NN rxN)

A typical buck switching regulator requires a higher input voltage to operate and tends to

"amplify" the load resistance. The "amplification" of load resistance will tend to "choke" the

other receivers in a multiple receivers setup especially when one of the receivers is in a high

resistance/trickle charge state. The trend can be observed from Equation 4-6 and Equation 3-1

which also can be observed from Figure 4-1. In order to achieve considerable power delivery

when one of the receiving devices is fully charged, it needs to decouple itself from the system.









The decoupling can be achieved by a switch either in series of the power path to break the

connection or in parallel with the receiving coil to create a short across the receiving coil so that

the transmitter does not see the transformed impedance due to the receiver.


5 i Power Delivered
to Receiver 1
4 (Single RX)
ac 4 -- -- -

Power Delivered
W --3 to Receiver 1
(dual RX with
Receiver 2 @
> 2
0 ~f1000 ohms)
Power Delivered
____ to Receiver 2
w '(dual RX with
O Receiver 2 @
C0 0_********************_* 1000 ohms)

10 100 1000

Receiver 1 Load Resistance (92)

Figure 4-1. Power delivery to loads for a single receiver setup and a dual receivers setup with
one of the load fixed at 1000C.

Since the receiver will be a portable device such as cellular phone, mp3 player, etc, the

receiver switch used to decouplee" the receiver needs to be compact and able to be driven by a

low voltage, e.g. not more than 3 V. Although, most electromechanical switches are able to

tolerate large voltages and currents, they are typically large for portable electronics and generate

a "clicking" sound during switching which is not acceptable. Off the shelf solid state switches

are typically designed for 50/60 Hz AC line application. They are relatively larger in size and do

not offer sufficient isolation for hundreds of kilohertz signals. It is possible to find switches

which operate at high frequencies but the power handling starts to drop with increase in

frequency as shown in [30] unless novel materials are used as shown in [31] which will make the

switch expensive. In addition, it is difficult to control the switch with voltages lower than the










voltage being switched using a simple transmission gate topology or switch transistors. A new

switch architecture which is an extension of a transmission gate switch is proposed to control a

large voltage/current signal with a low voltage control signal. The discussion of the switch

circuit in this section is independent of the decoupling architecture.

4.2 Switch Design

The block diagram is shown in Figure 4-2. The proposed switch should be able to handle

voltages up 25 Vrms and current up to 2 Arms with an operating frequency up to 1 MHz. In

addition, the proposed design should not use any inductors so that it can be easily integrated into

an IC or single package solution with the voltage regulator.



Negative voltage Switch
rectification control Control
network Network





|ip- A |9 Power Switch O A
voltage voltage




Positive voltage Switch
S ^ rectification control
network Network



Figure 4-2. Block diagram of the proposed switch.

The power switch block in Figure 4-2 uses a typical transmission gate architecture which is

a NMOS and a PMOS in parallel. A schottky diode must be added to either before or after the

transistor to counter the effect of the body diode of the power MOSFET. The schottky diode

selected must have power handling comparable to the body diode of the transistor. The control

signals to the gate of both of the transistors are provided via their respective switch control









network. Two rectification circuits extract the maximum voltage and minimum voltage of the

input AC voltage. The maximum voltage and minimum voltage are used as an input for the

respective switch control network in a cross-coupled topology. Based on the control signal

provided by the receiver, the switch control network will switch between the maximum voltage

and the minimum voltage to either turn the transmission gate on or off. Schematic of the

proposed switch circuit is shown in Figure 4-3. A single package dual N and P channel MOSFET

(IRF7343) from International Rectifier is used. The transistors have an absolute Vgs of 20 V and

an absolute Vds of 55 V. The peak continuous drain current of the N channel MOSFET is 4.7 A

and P channel MOSFET is 3.4 A. Turn on resistance for both transistors are typically better than

0.1 Q. Therefore, the transistors are able to handle considerable amount of power at high

efficiency. Rise time and fall time of both transistors are better than 22 ns, giving it a fast

response time. The input capacitor of both transistors is typically better than 750 pF which

makes driving the transistor feasible. The Cgd of both transistors are less than 100pF and the Cds

of both transistors are less than 125 pF, reducing the leakage current at high frequency when the

switch is turned off. MBRA340T3 is selected for both the rectification network diode and switch

because it is able to handle voltages up to 40 V and currents up to 3 A. In addition, it has a small

forward voltage drop of 0.45 V and being a schottky diode it has negligible reverse recovery

time.

Notation for resistors and capacitors are in the form ofRXX and CXX. The number

after the underscore is used to differentiate between the two switch control networks which are

similar, namely channel 1 for the P channel MOSFET of the transmission gate and channel 2 for

the N channel MOSFET of the transmission gate. Value for Cl is 100 nF, C2 is 10 nF, RI is 10

k and R2 is 47 kQ.










Switch control network
d-- -
IRF7343n
R2 2 I
Control R2




M I
I -I






_O I|


MBSwitch c
F MBRA340T3
AC in


40" % r- MBRA340T3

ItF7343p




0 ( C C2_1
SII

2.1,


Sio r cCt1 nwork

F in -43 o s R2 |









S w i J]ntroI 7et I Aro









Switch control network
IRF7343n
Co R2 2
Control R 2

I I
I n













AC in


SMBRA34- -
0 D C2 1









0
0 o |
MBRA340I






















Switch control network


Figure 4-4. Schematic of the improved proposed switch circuit.
4 fs^ ^









?r | J I









Figure 4-4. Schematic of the improved proposed switch circuit.












An improved version of the proposed switch is to replace the diode in series with the


transmission gate transistors with a transistor as shown in Figure 4-4. By doing so, the power


signal will not suffer a forward voltage diode drop, reducing the losses through the switch.


However, the switch control network must ensure that it is able to drive the extra gate loading the


two rectification networks. Since the diodes used have a low forward drop voltage, and for


simplicity, the switch will use the schematic on Figure 4-3 instead of Figure 4-4.


4.3 Switch Simulation






de c R3 C irf7
C1 R3 "P
4ODE4 Vnr3V C=100 nF R=lOO Ohm C4nF 4 R
odel=mbra343 C=10 nF
R=47 kOhm







11- ---de-4d R
(%)Vdc=O V Diode
DIODE2 i
Am FV irf7343n Model=m bra340t3 X -


d 23Diode
D E V1 DIODE1
irf7343p Model=mbra340t3W -- fri 7 R
Diode X2 Vi, 'R
R=50 Ohm
Melba343 +' VtPulse R0 Om
SRC2
,Vow-D V
Vhigh=3 V
R R 1 DelayO nsec
R2 R Edge=linear
R=10mODOhrn irF7343n R=47 kOhn VWdih=5 msec
CX3 Period=10 msec
C-- C2
C C=10 lnF k TRANSIENT
C=10D nF Vg DiodeModel
yu T Tran mbra34Dr0t3
Tran1
StopTimre=20 msec
MaxnimeStep=100 nsec



Figure 4-5. Schematic of the switch in Advanced System Design with a resistive as load.









Simulation and verification of the switch is done using Advanced System Design by

Agilent. The simulation schematic to analyze the performance of the switch with a 50 Q resistive

load is shown in Figure 4-5. Transistor and diode model in the simulation are obtained from the

manufacturer. Therefore, the simulation results should match the performance of the fabricated

circuit.

4







0


Swit;h co itrol oltage

0 2 4 6 8 10 12 14 16 18 20

Time (ms)

Figure 4-6. Switch control waveform (0 V for off and 3 V for on). A minimum of 1 V is
required to turn on the transistor.

Figure 4-6 shows with switch control waveform which is 0 V for off state and 3 V for on

state at a duty cycle of 50% and a frequency of 100 Hz. Based on the transistor used (IRF7343),

the switch is able to operate at turn on voltages as low as 1 V. Figure 4-7 shows the switch

control voltage for each respective channel. The turn off response time for channel 1 is

approximately 630 |^S and turn off response time for channel 2 is approximately 700 [iS. The

turn on response time for channel 1 is approximately 60 |^S and turn on response time for

channel 2 is approximately 70 [iS. Therefore, the switch can operate up to 1 kHz switching

frequency, which is way beyond the application of the circuit which is to turn off the receiver









when the device is fully charged. The turn on time is significantly faster than the turn off time

because the voltage across Cl and C2 are the same when the switch is turned off and when the

switch transits from the off stage to on stage C2 is charged/discharged via the low resistance path

through the transistor whereas when the switch transits from the on stage to the off stage C2 is

charged/discharged via resistor RI which increases the time constant significantly. Response

time can be improved by decreasing RI at the expense of power loss through the switch.

Decreasing either Cl or C2 also helps to improve the response time. However, the ripples on the

rectified voltage might become significant and affect the operation of the circuit.

30-
Genrated swit h co trol aveform c annel 1

10
;) 0
0s -10
-20
-30-
0 2 4 6 8 10 12 14 16 18 20

Time (ms) A
30- I I
Generated swit h co trol vaveform channel 2
20-

S10

c-i0:
So -

00 -10
-20

-30- -
0 2 4 6 8 10 12 14 16 18 20

Time (ms) B

Figure 4-7. Generated switch control waveforms. A) Channel 1 (P channel MOSFET of the
transmission gate) and, B) Channel 2 (N channel MOSFET of the transmission gate).









20-


0


o
10 Switch

aveform
-20 1
0 2 4 6 8 10 12 14 16 18 20

Time (ms)

Figure 4-8. Output waveform of the switch before the rectifier.

Figure 4-8 shows the output AC waveform of the switch across the 50 Q resistive load.

The switch is capable of breaking the high voltage/current AC signal path with reasonable turn

on and turn off time with the time off time slightly slower than the turn on time. The reason for

the difference in timing is the same reason for those of the switch control network waveform.

4.4 System Response with Receiver Switch





Receiving Switch Crx Cdc RLoad
Coil




Figure 4-9. Proposed half-wave rectifier receiver architecture.

Figure 4-10 shows the proposed receiver architecture. A half wave rectifier is used instead

of a full wave rectifier because the circuit will be more compact using a single diode instead of

four. Another benefit is that the rectification process only suffers a single diode forward voltage









drop instead of two. In addition, the operating frequency will be in the range of hundreds of

kilohertz, thus it will not require a very large charge holding capacitor. Instead of using the

switch in the power path, the switch is placed shunt across the coil. There are two reasons in

using such a topology. Firstly, by shorting the receiving coil, the receiver coil sees a short.

Therefore, Rrx and Xrx in Equation 2-7 will be extremely small that their value can be assumed to

be zero.

Let Xrx in Equation 2-7 be zero.


c2M2 R r0M3.2 M
Z -2M12 rx 2 +j OM 1 -12 222 (4-7)
R1 + (cOM R1 + (coM
rx \ 22 rx 22






S22

Substituting Equation 2-3 into Equation 4-8.


Ztx = j MI Mk2M11) (4-9)

In order for the system to be able to support multiple devices and provide sufficient lateral

freedom, it is reasonable for the coupling coefficient to be much less than 0.25. By assuming the

coupling coefficient to be 0.25, k2 will be 0.0625 which is much lower than 1. Therefore, if the

receiving coils are shorted under loosely coupled condition, the transmitter only sees the self

inductance of the transmitting coil. However, if the coupling coefficient between the coils starts

to increase there will be a reduction in self inductance of the transmitting coil. This will cause Zt,

to be less inductive and potentially causing the class E power amplifier to be no operating under

ZVS/ZDS condition.










The second reason is that the switch's natural state is open when no voltage is applied to

the control port. Therefore, the receiver will allow power to pass through when the control port is

left floating. This is critical especially when the battery on the receiver is fully drained and is

unable to control the switch. By using the proposed architecture in Figure 4-9, the receiver does

not require any bootstrapping even when there is no power, because the switch will be in a

naturally on state.

Figure 4-10 shows the test bench to verify the performance of the switch when used in this

specific scenario. A similar clock control source is used as well. The switch circuit in Figure 4-5

is converted to the modular block XI. Instead of using a full Class E driver, the transmitter is

simplified as an AC current source with 1A peak current at 240kHz. The transmitter coil is

modeled as inductor LI and receiving coil as inductor L2, mutual inductance is modeled via

"Mutual l" block.


MUTND
TRANSIENT
,1 14 Diode Model
ItSine L Mutual Tran mbra340t3
SRC1 Li Mutuall Tran1
IdC-O mA L=45.3 uH M=2.8 uH StopTime=20 msec
Amplitude-1 A R-0.5 Ohm Inductorl="L1" MaxTimeStep=200 nsec
Freq-240 kHz Inductor2="L2"
Delay-O rsec
Damping-0
Phase-0


Vdc
Diode
DIODE1
L C VtPulse Model-mbra340t3 C R
SL2 C1 SRC2 VCt C2 R1
SL=5.2 uH C=75.0 nF fLVtowO V I s,,,,.,,* C=4.7 uF R=50 Ohm


Figure 4-10. ADS schematic of test bench for receiver architecture in Figure 5-10.












16 .- 3.0
14
-2.5
12
> 10- -2.0 <

> 8 1.5 <
6-
1.0
4-
2 -0.5

0 0.0
0 2 4 6 8 10 12 14 16 18 20

time, msec

Figure 4-11. Simulation results of test bench (Figure 4-10). Red: Low voltage control signal.
Black: Receiver rectified voltage.

Figure 4-11 shows the simulation results of the test bench of Figure 4-10 with the control

voltage in red and output rectified voltage in black. As predicted in switch standalone

simulations, the switch closing time is faster than the opening time, resulting in a faster

decoupling response time. Although the receiver is supposed to be fully decoupled when the

switch closes, a 0.5V DC voltage still can be observed at the load. This is because the switch is

not an ideal switch and a potential drop will be observed across the parasitic resistance of the

transmission gate transistor as well as the series diodes used to counter the effect of the body

diode. The 0.5V DC voltage should not be a concern because it is not sufficient to turn on the

voltage regulator which is typically used to provide a stable DC voltage. From the simulation

results, it can be concluded that the switch circuit can be used in scenarios where a control

voltage is significantly smaller than the input AC voltage at high frequencies regardless of

topologies.









CHAPTER 5
EXPERIMENTAL VERIFICATION

5.1 High Power 300W Dual Channel System using Parallel-Parallel Impedance
Transformation Network Topology

Based on the parallel-parallel impedance transformation network topology design rule, a

dual-channel Class E transmitter test system capable of delivering nearly 300 W with a supply

voltage of 120 V is fabricated using the IRFP21N60L HEXFET power MOSFET from

International Rectifier. A full wave rectifier with a shunt charge holding capacitor at the output

using MUR420 from Vishay is fabricated to convert the AC power to DC power. Since the

forward voltage drop is 0.875 V and the reverse recovery is 30 ns, power loss due to the voltage

drop and reverse recovery is small compared to the amount of power delivered to the load. Load

resistance in this section is the equivalent resistance looking into the regulator or device being

charged/powered as shown in Figure 1-1 instead of the equivalent resistance looking into the

rectifier as shown in Figure 3-3.

In order to reduce losses through parasitic resistance, low loss polypropylene capacitors are

used. To achieve a balance between size and efficiency, a 100 [H inductor (1140-101K-RC) by

Bourns Jw Miller is selected for Lou,,. Since most of the loss of the transmitter is due to the

parasitic resistance of Lou,, a larger and more efficient inductor can be used if space permits.

Table 5-1 shows the calculated value of each component with respect to the actual component

value used in the experimental setup. The calculated values obtain from Chapter 3 are initially

used and further tuned to achieve optimum power delivery and efficiency across a wide range of

load resistance. Most of the values follow closely to the calculated value from Matlab. The only

exception is Cshunt. The main reason for such a discrepancy is that the equation used in [18]-[19]

assumes the transistor to be an ideal switch. Therefore, while calculating the drain voltage, the

built-in diode in the transistor was not taken into account. The other parameters of the transistor









do not have any significant effect on the calculated values since the rise and fall times of the

transistor are significantly faster than the switching time and the drain-to-source capacitance is

less than 1 nF. In addition, the turn-on resistance of the transistor is extremely small. Another

reason is that during the calculation, the DC feed inductor (LDc) is assumed to be infinitely large,

which is not true in experiment.

Table 5-1. Component Values for High power 300W Dual Channel System using Parallel-
Parallel Impedance Transformation Network Topology
Calculated Experimental % Variation
Crx 113 nF 115 nF +1.8%
Ct, 105 nF 100 nF -4.8%
Cout 68 nF 68 nF 0%
Lout 100 [H 100 pH 0%
Lout parasitic resistance 1.3 Q
Cshunt 19 nF 15 nF -21%
TX coil inductance 31.95 |iH
TX coil dimension 21 cm x 21 cm
TX coil parasitic resistance 0.32 Q
RX coil Inductance 12.52 |JH
RX coil dimension 13cm x 13 cm
RX coil parasitic resistance 0.20 Q
Mutual Inductance 7.475 pH
Coupling Coefficient 0.374
LDC 500 pH

The fabricated dual-channel transmitter with a dimension of 10 cm x 8.5 cm is shown in

Figure 5-1. The two inductors of L,,ut occupy a significant amount of space due to the

requirement of low parasitic resistance so as to maintain high efficiency in power delivery.

Figure 5-2 shows the transmitting coil (21 cm x 21 cm) embedded inside a table top case, and the

receiving coil (13 cm x 13 cm) placed on top of the case. The actual separation between the two

coils is 10 mm. The setup enables the user to have a large degree of translational freedom both in

the X and Y direction. It should be noted that there is no ferrite core in either the transmitting

coil or receiving coil.






















A,


Figure 5-1. Photograph of the dual-channel Class E power amplifier.


N


Figure 5-2. Photograph of the transmitting coil -
receiving coil 5 turns (placed on tol


10 turns (embedded into the table top) and









Using a 120V power supply, power delivery of 295 W to a 51 load with a DC voltage of

121.5 V and current of 2.43 A is achieved. The input current from the power supply is 3.25 A.

The end-to-end system efficiency is 75.7%. Peak drain voltage is 460 V which is 25% lower than

the rated maximum voltage of the transistor. Figure 5-3 shows the efficiency and power delivery

of the 120 V system versus load resistance. Although the maximum power of 295 W occurs at 50

Q load resistance, a high efficiency of at least 77% is achieved across the range of 60 Q to 140

Q, which matches the optimum range of power delivery above 200 W. The power delivered is

proportional to the square of the supply voltage. The power delivery of the system can be

increased by increasing the supply voltage as long as the DC power supply driving system is able

to provide sufficient power and the drain voltage across the transistor stays within its breakdown

voltage.

300 80

C 250 ----- ---------- 70

-' 200 ---------------- 60



o ***
0) 100 ------------------ 40 w

I50- -- 30

0 20
0 250 500 750 1000 1250 1500 1750 2000

Load Resistance (0)

Power Delivered to Load (W) ..... System Efficiency (%)


Figure 5-3. Power delivery (left y-axis) and efficiency (right y-axis) of the system versus load
resistance. Supply voltage: 120 V.










140
130
120
110
100
90
80
70
60
50
40
30
20


20


30 40 50 60 70 80 90 100 110 120

Supply Voltage (V)

-Transistor Temperature (natural convection cooling)
-Inductor Temperature (natural convection cooling)
-Transistor Temperature (forced cooling)
-Inductor Temperature (forced cooling)


Figure 5-4. Transistor and inductor temperature with natural convection cooling and forced air
cooling versus supply voltage.

The transmitter is able to operate under natural convection cooling with up to 95 W of

power delivery to the load using a supply voltage of 65 V. Power levels above 95 W requires

forced air cooling on the transmitter to keep the temperature of the transistors below 70 'C.

Although most of the loss due to heat occurs at the transistor or inductor, a temperature increase

at the inductor is not critical because it is a passive component and more resilient to heat.

However, this will cause a drop in inductance value which will decrease the load network phase

angle increasing the power delivery slightly. The slight increase in power delivery should be

taken care by the receiver's regulator. Temperatures of both the transistor and inductor for

natural convection cooling and forced air cooling were measured using the Fluke 62 mini


- 4-4-4-4-4-4-




;-

-W- --- -









infrared thermometer at extremely close proximity is shown in Figure 5-4. A 12 V DC dual ball

bearing fan of size 90 mm x 90 mm x 25 mm with a speed of 2700 rpm achieving an air flow of

44 cfm is used for forced air cooling (shown in Figure 5-5). Since the power consumption of the

fan is only 2.4 W, it can be neglected for the efficiency calculations. In addition, the heat sinks

on the transistors further enhance the heat dissipation capabilities. As seen in Figure 5-4, the

transistor temperature reaches a practical limit of 75 'C for a supply voltage of 120 V. Although

higher power can be achieved by increasing the supply voltage using a higher power output

power supply, more effective heat dissipation methods are required to prevent the transistors

from overheating.

A compact system operating without forced air cooling is often preferred. Therefore, the

following measurements based on a supply voltage of 60 V operating under natural convection

cooling are presented. Figure 5-6 shows the transmitting coil current and voltage when it is

driving a 50 Q load at the receiver. It can be seen that the current is lagging the voltage. As seen

from the drain waveform, ZVS operation is achieved.




















Figure 5-5. Photograph of the dual-channel Class E power amplifier with forced air cooling.
































Figure 5-6. Voltage and current waveforms of the Class E transmitter.


0 250 500 750 1000 1250 1500 1750 2000

Load Resistance (0)


^-Both channels enabled


***** Single channel enabled


Figure 5-7. Power delivered to load versus load resistance. A maximum power of 69 W occurs
approximately at 500 for dual -channel and maximum power of 10 W occurs at
approximately at 75 Q for single -channel. Supply voltage: 60 V.









Figure 5-7 shows the power delivery versus load resistance of both dual-channel (solid

line) and single-channel (dotted line) modes. Power delivery for both modes peak at about 5(D

load resistance, the dual channel mode peak power is close to 70 W while the single channel

mode is at 10 W. The system efficiency versus load resistance is shown in Figure 5-9 which

verifies that the resistance looking into the network of the transmitting coil in parallel with the

shunt capacitor decreases with increasing load resistance. This results in a larger voltage drop

across the parasitic resistance of the inductor and a lower efficiency.


0 100


,-Both channels enabled


500


***** Single channel enabled


Figure 5-8. Mode-switching operation for optimized efficiency across a wide power delivery
range. (1) Dual-channel mode for higher power, (2) Dual-channel mode switch-over
to single-channel mode when better efficiency can be obtained at a lower power
level, (3) Single-channel mode for lower power, (4) Single-channel mode switch-
over to dual-channel mode when higher power delivery is needed. Supply voltage:
60 V.


200 300 400

Load Resistance (0)










80

70

60
U
2 50

U 40
E
30

20

10


0 250 500 750 1000 1250 1500 1750 2000

Load Resistance (0)

Both channels enabled *****... Single channel enabled


Figure 5-9. System efficiency versus load resistance with a maximum efficiency of 64.5% for a
single-channel system, and 76% for a dual-channel system at approximately 700.
Supply voltage: 60 V.

The efficiency of the single-channel mode is approximately 10-15% lower than the

efficiency of the dual-channel mode for the same load resistance because the current is flowing

through a single Lout inductor instead of a pair of them. The parasitic resistance is doubled, thus

resulting in lower system and transmitter efficiencies as shown in Figure 5-9 and Figure 5-10

respectively. However, when the system enters light load mode or trickle charge mode, it is

desirable to switch to the single-channel mode. It can be seen from Figure 5-11 that the system

efficiency is approximately 15% higher than the dual-channel mode for delivering the same

amount of power below 10 W. Instead of operating at high load resistance for a dual-channel

mode resulting in high receiver DC voltage as shown in Figure 5-13, it is possible to achieve

similar power delivery at much lower load resistance for a single-channel mode, resulting in a

lower receiver DC voltage and higher system efficiency. In addition, a typical buck regulator has


* S

S










higher DC-DC efficiency when the input voltage is lower. Therefore, a load detection scheme is

required to determine the switch-over point from dual-channel mode to single-channel mode. It

can be seen from Figure 5-7 that a power delivery of 10 W occurs at 5000 load resistance in the

dual-channel mode making it a good switch-over point to single-channel mode ((2) in Figure 5-

8). It can be seen that a 5000 load resistance translates to an approximate RMS voltage of 20 V

across transmitting coil for the dual-channel mode as shown in Figure 5-12. Likewise, if the

power requirement for the single-channel mode is too high, it is required to switch to the dual-

channel mode. It can be inferred from Figure 5-7 that the switch-over point to dual-channel mode

is approximately 750 ((4) in Figure 5-8) where the efficiency peaks in single-channel mode,

which translates to a RMS coil voltage of 22 V in Figure 5-12.


500


750 1000 1250

Load Resistance (0)


^-Both channels enabled


1500


1750 2000


***** Single channel enabled


Figure 5-10. Transmitter efficiency versus load resistance. Maximum transmitter efficiency
occurs across the range of 600 to 100 Q load resistance at 90% for dual -channel
and 79% for single-channel. Supply voltage: 60 V.


* S

* .
* S __


250



























0 10 20 30 40 50 60 70

Power Delivered to Load (W)


- Both channels enabled


Figure 5-11.









o



+*-
0
(,



U
o


U,

I-


***** Single channel enabled


System efficiency versus load resistance for single-channel and dual-channel
modes achieving high efficiency at high power output. It also illustrates that a
single-channel mode is more efficient at low power delivery states. Supply voltage:
60 V.


100

90

80

70

60

50

40

30

20

10


0 250 500 750 1000 1250 1500 1750 2000
Load Resistance (0)

Both channels enabled ..... Single channel enabled

Figure 5-12. Transmitting coil RMS voltage versus load resistance. Supply voltage: 60 V.











Figure 5-13 shows the receiver DC voltage versus load resistance for both the dual-channel


mode and the single-channel mode. It can be seen that there is no over voltage issue as the

voltage starts to converge to a value of approximately 70 V when the load resistance is high. The

receiver open-circuit voltage is 73.4 V and open-circuit power consumption is 10 W in the dual-

channel mode. However, during single-channel operation, the receiver open-circuit voltage is

only 38.3 V and its open-circuit power consumption is only 4 W which is 6 W less than that of


the dual-channel mode. Therefore, it is preferred to perform load detection using the single-

channel mode and increase the output power by enabling the dual-channel mode if more power is

required to reduce standby power consumption. By controlling the maximum receiver DC


voltage, the requirement for the receiver's regulator is relaxed enabling the system designer a

wider range of selection.


75

265

b 55
(,
0 45
U
o 35
L1

25

a 15

5


0 250 500 750 1000 1250 1500 1750 2000

Load Resistance (0)


Both channels enabled *****... Single channel enabled


Figure 5-13. Receiver DC voltage versus load resistance, converging to approximately 70 V in
dual-channel mode and 38 V in single-channel mode. Supply voltage: 60 V.


- ^^ ----- -


*
S-- -- -

*
*
*...
0__ __ __ __ __ __ __ __ __ __ __ __ __ __ __ __
S[-------









5.2 Low Power Multiple Receivers System with decoupling Switch using Series-Parallel
Impedance Transformation Network Topology

The Class E transmitter test system operating at 240 kHz was fabricated using the

IRLR/U3410 power MOSFET. A 13-turn 20 cm x 20 cm transmitting coil which was described

in the design rules and two 6-turn 9 cm x 6 cm receiving coils were used. The platform is capable

of simultaneously charging up to four independent devices. Although it is possible to extend the

experiment to more than two receivers, for better understanding of the system response and

easier analysis, experiments of charging two receivers were conducted. A Matlab code was also

written based on the equations derived in [18]-[19] to study the efficiency and power delivery.

All measurements and simulation results are based on a 12V power supply. The 12V power

supply is selected because the supply voltage is readily available from the DC supply plugs in

vehicles and several other AC-DC converters.

Table 5-2 shows the value of each component used in the experiment. Component values

are selected by matching the closest available component value and further tuned to achieve

optimum performance. Crx is selected to be 75 nF. Since the switch contributes 3.5 nF of

capacitance and the rectifier contributes another 3.5 nF of capacitance, a 68 nF capacitor is used

to achieve an effective capacitance of 75nF. Capacitance contributed to the receiver due to the

switch and rectifier is measured using the HP4192A LF Impedance Analyzer in small signal

operation. Therefore, the actual capacitance under large signal operation will be slightly different

depending on the load conditions and voltage at the switch and rectifier. In order to reduce losses

and heat through the parasitic resistance of the capacitors, low loss polypropylene capacitors are

used. Alternatively, COG/NPO surface mount capacitors can be used as well. However, they are

physically small and tend to heat up faster than the larger leaded polypropylene capacitors

making them not suitable for high power applications.









Table 5-2. Component Values for Low Power Multiple Receivers System with decoupling
Switch using Series-Parallel Impedance Transformation Network Topology
Calculated Experimental % Variation
Crx 73nF 75nF +2.7%
Cout 8nF 9.4nF +17.5%
Lout 9.5 [H 9.5 [H 0%
Lo,,ut parasitic resistance -0.160
Cshunt 10nF 12nF +20%
TX coil inductance 45.3 [H
TX coil dimension 20cm x 20cm
TX coil parasitic resistance 0.50
RX coil Inductance 5.2[tH
RX coil dimension 9cm x 6cm
RX coil parasitic resistance 0.10
Mutual Inductance 2.8 [iH
Coupling Coefficient 0.182
LDC 500 LIH


Transmitter


Figure 5-14. Photograph of a test setup with two receivers with decoupling switch on the
packaged transmitting coil.










Figure 5-14 shows the photograph of the test setup with two receivers on the packaged

transmitting coil. The vertical separation between the transmitting coil and receiving coils is

about 2 mm. The locations of receivers on the transmitting coil were fixed by the blue tapes to

ensure the same conditions for all measurements.


6

5

4

3

2

1

0


0-0


o
CL <-


250


- Efficiency (measured)
- Power Delievered (mea


500 750 1000
Load Resistance (()
Efficiency (simulated)
sured) Power Delievered (simulated)


Figure 5-15. Power delivery to the receiver with switch and system efficiency versus load
resistance for a one-to-one setup (simulated and measured)

Comparison of power delivery to the receiver and system efficiency versus load resistance

of a one-to-one setup between the simulated results and the experimental results is shown in

Figure 5-15. Measurements are conducted by connecting the receiver to a rheostat for which the

resistance is varied manually in predetermined steps. The simulation and measured trend of a

single receiver setup agree well with peak measured power at around 4.75 W. There is a slight

discrepancy between the measured and simulated efficiencies because the transistor and DC feed

inductor are assumed to be ideal in the simulation model. In addition, the effect of the build in









body diode of the transistor is also not taken into consideration. The assumption affects the

calculated supply current and the calculated efficiency. Power delivery is not affected because it

depends on Zx rather than the transistor during nominal operation. A comprehensive plot

presenting the performance of the system in Figure 5-16 shows the system efficiency versus

power delivery.


80

70

60

S50

S40

a 30

20

10


0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

Power Delivered to Load (W)

Efficiency (measured) Efficiency (simulated)

Figure 5-16. Efficiency of power delivery to the receiver with switch versus power delivered for
a 1 to 1 setup (simulated and measured)

Figure 5-17 compares the performance of the receiver with and without the switch with

power delivery capabilities of both receivers are nearly the same. They peak at around 4.75 W

with the efficiency of the receiver with switch slightly lowered by 1 to 2%. The impact of the

switch in the receiving mode is minimal because it is in shunt with the receiver. In addition, it

can be concluded that the leakage through the switch is negligible. Figure 5-18 compares the

performance of a single receiver with the decoupling switch architecture and dual-receiver setup

with one of the receivers decoupled from the transmitter. The efficiency degrades by an average


-- -- -------



- ^ - ~-_- -










of 5% and no more than 10% overall even though the second receiver is turned off. Although the

receiver is decoupled from the system, the switch circuitry still has some turn-on resistance when

it attempts to short the receiving coil. Therefore, a small amount of power is still dissipated

across the switch.


Q5U

70

g 60

U 50


u 30

20

10


Figure 5-17.










w


Co

80

70

60

50

40

30

20

10


0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5

Power Delivered to Load (W)
Receiver without Switch -Receiver with switch

mparison between receiver with switch and receiver without switch.






I- -- -
-e 00 -wo



-------------- ------ --


0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

Power Delivered to Load (W)
-Receiver 2 off Receiver 1 standalone


5.5 6.0


Figure 5-18. Comparison between single receiver standalone and dual-receiver setup with one
receiver switched off









To study the power delivery of a dual-receiver platform, the load resistance of one of the

receivers (receiver 2) is fixed while the load resistance of the other receiver (receiver 1) is swept

across the range of 100 to 2000 Q in 15 discrete steps (10 0, 15 Q, 20 Q, 25 Q, 30 Q, 40 Q, 50

Q, 75 Q, 100 Q, 150 Q, 200 Q, 250 Q, 500 Q, 1000 Q and 2000 Q) in an experiment using the

same method for the single receiver measurement. Figure 5-19 shows the power delivery to the

receiver 1 versus its load resistance at different receiver 2 load resistance values, while Figure 5-

20 shows the power delivery to receiver 2 versus receiver 1 load resistance at different fixed

receiver 2 load resistance values. When the load resistance of receiver 2 is kept above 50 Q, the

variation of power delivery to receiver 1 due to receiver 2 is limited, but peak power drops to

around 2.5 W. In addition, power delivery to receiver 2 also stays consistent regardless of the

load resistance or power delivery to receiver 1, as long as the load resistance of receiver 1 stays

above 50 Q. To reduce the dependency of the receivers on each other due to a single driving coil,

the minimum load resistance should be greater than 50 Q. However, this will result in a

reduction of maximum power delivered to the load by the system. The dependency of the

receivers is due to the collective impedance looking into the transmitting coil due to the multiple

receivers and not the mutual inductance between the receiving coils. The minimum load

resistance can be designed by selecting an appropriate receiver regulator and setting the

appropriate power delivery profile by changing Ctx, or the supply voltage. This will set the

unregulated input voltage before the regulator to achieve the specified load resistance looking

into the regulator while at its maximum power delivery condition so that the load resistance will

not fall below minimum value mitigating the power choking effect. The experimental

verification and analysis is limited to two receivers but similar trends are expected for multiple

receivers of three or more.












5 -77- -- --------













10 100 1000

Receiver 1 Load Resistance (Q)
Receiver 2 off Receiver 2 @ 10 ohm Receiver 2 @ 15 ohm
Receiver 2 @ 20 ohm Receiver 2 @ 25 ohm Receiver 2 @ 30 ohm
Receiver2 @ 40 ohm ***** Receiver2 @ 50 ohm ***** Receiver2 @ 75 ohm
***** Receiver2 @ 100 ohm ***** Receiver2 @ 150 ohm ***** Receiver2 @ 200 ohm
***** Receiver 2 @ 250 ohm -Receiver2 @ 500ohm -Receiver 2 @ 1000 ohm
Receiver 2 @ 2000 ohm -Receiver 2 OC -Receiver 1 standalone

Figure 5-19. Power delivery to receiver 1 versus its load resistance at different fixed receiver 2
load resistance.


10 100 1000

Receiver 1 Load Resistance (0)

Figure 5-20. Power delivery to receiver 2 versus receiver 1 load resistance at different fixed
receiver 2 load resistance. (same legends as in Figure 5-19)









Once the fully charged receiver (receiver 2) is decoupled from the system using the switch

circuit, power delivery to the other receiver (receiver 1) increases significantly. Therefore, the

switch circuit can be used to prevent the receiver that is fully charged to "choke" the other

receiver of the power it requires. This will mitigate the effect of reduced charge rate for the

receiving devices so that the system will be able to deliver sufficient power to the receiver.


Figure 5-21 shows the system efficiency versus total power delivered to the loads with

receiver 2 fixed at a specific load resistance while sweeping the resistance of receiver 1 from

100 to 20000. System efficiency is above 55% for power delivery above 2 W. Although the

efficiency starts to degrade significantly at lower power delivery, the absolute system power loss

is low. Therefore, no heating issues were observed during the experiment. All components were

operating below 360C.

85

75

65

55 ---

45

35

25

15 --------------

5
0 1 2 3 4 5 6 7 8
Power delivered to Loads (W)

Figure 5-21. System efficiency versus total power delivery to both receivers with the load
resistance of receiver 2 fixed and the load resistance of receiver 1 swept across the
stated range for a dual-receiver test bench. (same legends as in Figure 5-19)









Figure 5-22 shows the measured power delivery space of receiver 1 and receiver 2. Under

all loading conditions the guaranteed power delivery is approximately 2W. The guaranteed

power delivery can also be observed in Figure 5-19 and Figure 5-20 in another form as discussed

previously. Therefore, the system is capable of delivering 2W of power under all conditions

which is close to the specified power delivery of2.5W in the design. Higher power delivery can

be achieved by reducing the capacitance of Ct, slightly or by increasing the supply voltage. Care

must be taken during reduction of Ct, to keep the phase of Ztx within the high efficiency/low loss

operation above 400.

5 1 I I I I I T I


0 1 2 3 4 5 6


Power Delievered to Receiver 1 (W)


Figure 5-22. Measured power delivery to a dual-receiver system with load resistance varied
from 100 to 20000.









CHAPTER 6
INTEROPERABILITY BETWEEN DIFFERENT PLATFORMS (COIL SIZES)

The design rules of the wireless power system are presented in Chapter 3. They are further

verified with experimental results in Chapter 5. However, the presented design rules and results

only verify the operation of a specific one to one device pairing between the transmitting coil and

receiving coil. This does not include operation between another pair of coils of different sizes

and might result in a cross platform interoperability issue. It might not be a critical problem if

power delivery is insufficient, but it will be a major problem if it results in damaging the

receiver. For a practical system, it is expected that the same transmitter for a certain cell

phone/device to be able to work for all other different cell phones/device which might have a

different receiving coil size due to its physical dimensions. In addition, it will be desirable if the

cell phone's transmitter is be able to transfer power to a smaller device which requires less power

such as an mp3 player or a Bluetooth headset. As companies start to expand their product lines or

try to differentiate from one another, there will be transmitters with different sized transmitting

coils. Therefore, it is critical to ensure there is some level of interoperability between the

different platforms. Without interoperability between the different platforms, the consumer will

be forced to purchase a different transmitter for each of the product he or she owns, not

achieving the purpose of a universal charging platform to reduce power consumption and

wastage. Regardless of providing sufficient power to drive the receiver, the transmitter should

not damage the receiver via overvoltage at the input of the receiver's voltage regulator. The

interoperability can be further enhanced by creating a wireless power standards body to regulate

the different systems. A wireless power consortium is being set up by Phillips in December 2008

but has yet to gain sufficient momentum to ensure some form of interoperability between

different products and platforms.









6.1 Test Bench Setup

The interoperability study is performed by using three different sets of transmitting and

receiving coils with components selected using the series-parallel topology design rules in

Chapter 3. Transmitter and receiver units are then interchanged between the respective platform

without changing any component or automatic tuning to study the interoperability of the system.

Table 6-1 shows the sizes of the different pairs of coils and their respective components values,

for consistency a L,,ut of 10 gH is used for all the platforms. Crx being in shunt with the receiving

coil gets smaller with the increase of receiving coil size as the self inductance of the receiving

coil increases. C,,t which is in series with the transmitting coil gets smaller when the transmitting

coil increases in size as the self inductance of the transmitting coil increases. This is because

more negative reactance is required to offset the increase in self inductance of the transmitting

coil. Since the Ztx for all 3 cases are designed to be relatively close to each other, Cshunt is kept

the same for all three different platforms.

Table 6-1. Specification of the three different platforms
Value
Small platform
Transmitting coil size 11 cm x 8 cm (88 cm2)
Receiving coil size 5 cm x 4 cm (20 cm2)
Crx 103.3 nF
Cout 27 nF
Cshunt 15 nF
Medium platform
Transmitting coil size 16 cm x 16 cm (256 cm2)
Receiving coil size 6 cm x 5.5 cm (33 cm2)
Crx 94.7 nF
Cout 10 nF
Cshunt 15 nF
Big platform
Transmitting coil size 20 cm x 20 cm (400 cm2)
Receiving coil size 9 cm x 6 cm (54 cm2)
Crx 68 nF
Cout 9 nF
Cshunt 15 nF









Table 6-2. Coupling parameters of nine possible combinations with first three as intended pairs
TX Size RX Size L1 (uH) L2 (uH) M (uH) K Size Ratio
Small Small 10.96 3.70 2.21 0.347 4
Medium Medium 40.53 4.10 2.28 0.176 8
Big Big 46.00 5.45 2.81 0.178 7
Small Medium 10.96 4.10 3.05 0.455 3
Small Big 10.96 5.45 4.66 0.603 2
Medium Small 40.53 3.70 1.65 0.135 13
Medium Big 40.53 5.45 3.45 0.232 5
Big Small 46.00 3.70 1.37 0.105 20
Big Medium 46.00 4.10 1.88 0.137 12

With three different platforms, it is possible to come up with nine possible combinations as

shown in Table 6-2. The original pairs of the transmitting and receiving coils for the three

platforms designed using the design rules in Chapter 3 are shown as the first three rows. From

Equation 2-40, it can be inferred that the self inductance of the receiving coil is compensated by

Crx and has little effect on the coupling between the coils when they are swapped between each

other. By looking at the real part of Equation 2-6, it is independent of the transmitting coil's self

inductance. Since most of the effects of the receiving coil's self inductance is being compensated

by Crx the power transfer is more dependent on the mutual inductance between the pair of

transmitting and receiving coils used. Similar analysis can be used for the imaginary part of Ztx

as C,,ut is used to cancel out most of the effects of self inductance of the transmitting coil. Other

than the setup with a small transmitting coil and a big transmitting coil, the mutual inductances

between the transmitting and receiving coil are kept to around 2 iH. Once the appropriate

capacitor and inductor values are selected via the design rule presented in Chapter 3, the Class E

transmitter is rather robust under most loading conditions. No control mechanism is required to

achieve correct power delivery trend and high efficiency. The variation of mutual inductance

should be acceptable to still maintain relatively high efficiency power transfer with the correct

power delivery trend with respect to load resistance.









6.2 Experimental Verification

Using the nine different combinations of the three platforms on a one transmitter to one

receiver setup, nine different data sets were collected. The supply voltage is selected to be 12 V

but the analysis should be independent of the supply voltage. For this experiment, the setups

should be able to deliver at least 5 W of power and a receiver voltage of not more than 25 V and

not less than 8 V for nominal operation of the voltage regulator to achieve a 5 V regulated

output. The results are described using the notation of transmitting coil size as the first word and

the receiving coil size as the second word. Therefore, it means that a small sized receiving coil is

being placed on top of a medium sized transmitting coil for the notation "medium small".

80 21

^ 70 ..... 20

60 ---------------- 19 >






150 --------- --------- ----- 16


3 0 2 4 6 8 10
E 0 ---------~----~'r----------------- 13 >



0 2 4 6 8 10

Power Delievered to Load (W)

Small Small (Efficiency) Small Medium (Efficiency)
Small Big (Efficiency) -Small Small (Receiver Voltage)
-Small Medium (Receiver Voltage)- -Small Big (Receiver Voltage)


Figure 6-1. System efficiency and receiver voltage vs. power delivery plot for different sized
receiving coil on a small sized transmitting coil.









The results are presented in three different graphs each for a different transmitting coil

size. All results using the small receiving coil are presented in red, medium receiving coil are

presented in blue and big receiving coil are presented in green. Dashed lines are used for the

unregulated receiver voltage (right axis) while solid lines are used for system efficiency

excluding the voltage regulator (left axis).

Figure 6-1 shows the experimental results of three different sized coils being placed on a

transmitter with a small transmitting coil. When a larger receiving coil is placed on top of the

small transmitting coil the real part of Ztx becomes larger, reducing the losses through the

transmitting coil's parasitic resistance and L,,out's parasitic resistance. Therefore, the setup for

"small medium" is more efficient than "small small". However, when the real part of Ztx

becomes too large due to high mutual inductance between the coils (4.66 |^H for "small big"

setup), the phase angle of Ztx becomes too low that the class E transmitter. Therefore, the class E

transmitter is working at the borderline case of ZVS/ZDS with increased losses at the transmitter

across the switching transistor. This occurs when the green efficiency line starts to deviate from

the blue line with increase in power delivery (decreasing in phase angle) at approximately 1 W

power level. However, the efficiency is still acceptable and no significant heating issue is

observed. The receiver voltage under a specific loading condition is kept approximately the same

for all three cases with the maximum unregulated receiver voltage at slightly less than 21 V and

the minimum unregulated receiver voltage above 13 V, no over-voltage was observed. The lower

unregulated receiver voltage for the "small big" setup is due to the same reason for the lower

efficiency phenomena as more power is being dissipated at the transmitter than transmitted to the

receiver as useful power. All three setups are able to achieve at least the stated 5 W of power

delivery to the receiver.










80

70

60

50

w 40
E
30

20

10


0 1 2 3 4 5 6 7 8 9 10

Power Delievered to Load (W)

Medium Small (Efficiency) Medium Medium (Efficiency)
Medium Big (Efficiency) Medium Small (Receiver Voltage)
Medium Medium (Receiver Voltage) Medium Big (Receiver Voltage)


Figure 6-2. System efficiency and receiver voltage vs. power delivery plot for different sized
receiving coil on a medium sized transmitting coil.

Figure 6-2 shows the same experimental results as Figure 6-1 but with the transmitting coil

replaced with the medium sized transmitting coil. Based on the previous observation with "small

medium" the trend of increasing efficiency of "medium big" is expected. With the decrease in

mutual inductance between the coils, the real part of Ztx is reduced for this setup resulting in

more energy being dissipated across the parasitic resistance of Lout and the transmitting coil.

Therefore, the trend of reduction on efficiency is also expected for the "medium small" setup. In

addition due to the same reasons less power is expected to be transferred to the receiver. All

three setups are able to achieve at least 5 W of power delivery with "medium small" slightly

above 5 W. The unregulated receiver voltage is still kept within the nominal operating range of 8

V to 25 V.


24

22

20
>
18

16

14 i

12

10









80 20

70 18



S50 1 2 3614

2 40 12

30 ____ 10 w



10 "6------
0 1 2 3 4 5 6 7

Power Delievered to Load (W)

Big Small (Efficiency) Big Medium (Efficiency)
Big Big (Efficiency) Big Small (Receiver Voltage)
Big Medium (Receiver Voltage) Big Big (Receiver Voltage)


Figure 6-3. System efficiency and receiver voltage vs. power delivery plot for different sized
receiving coil on a big sized transmitting coil.

Following the same approach, the medium sized transmitting coil is replaced with a big

sized transmitting coil and results are show in Figure 6-3. The efficiency and power delivery

degrades for both the smaller receiving coils. The coupling is so weak for the "big small" setup

that very little power (up to 2 W) is being transmitted to the receiver. In addition, the unregulated

receiver voltage drops below 8 V which is too low to achieve voltage regulation for a 5 V output.

Since eight out of nine of the combinations except "big small" meet the specification of at least 5

W of the power delivery and unregulated receiver voltage range of 8 V to 25 V, we can infer that

a desirable range of size ratio between the receiving coil and transmitting coil is approximately

1:2 to 1:12. This also translates to a coupling coefficient of 0.137 to 0.603.









CHAPTER 7
LOAD/FAULT DETECTION AND POWER DELIVERY TRACKING

The proposed near-field wireless power transfer system is sensitive to nearby conductive or

magnetic objects as both the mutual inductance and the self inductances of the transmitting and

receiving coils will be affected. Although it is possible to shield the transmitting coil from

interference behind or beneath it, the shield does not prevent a user from potentially damaging

the system by placing objects such as a metal sheet on the transmitting coil or simply flipping the

transmitting coil over a metal table. Therefore, to ensure robust operation of the system, a

method of fault mode detection must be implemented so that the transmitter circuitry will not be

damaged due to undesirable actions by the user.

In addition to protecting the transmitting platform from being damaged by conductive or

magnetic objects, it is also desirable to reduce power consumption by turning off the transmitter

when no valid receiving device is placed on the transmitting platform. The system will wake up

for a very short period of time which is long enough for it to reach steady state. The time is

predominately dependant on the self inductance of the transmitting coil and operating frequency.

It is typically in the range of 0.5 ms to 5.0 ms. Probing the system with a rate of 1 Hz to 2 Hz

speed is sufficiently fast for the user not to feel much latency. Therefore, the worst case duty

cycle is 1% and is sufficient to reduce no load power consumption considerably. Although, one

could use a communication link to perform authentication and handshaking, there would be a

considerable increase in cost and component count. An alternative is to detect the system loading

condition by the voltages and currents of the transmitter. To ensure low power consumption and

cost, the voltages and currents to be detected must be either DC or be converted to DC so that a

low speed analog-to-digital convertor (ADC) can be used to accurately extract the information

and convert to digital domain.









7.1 Load/Fault Detection Scheme


7.1.1 Detection Circuit


SRioad


Transformation network I


Figure 7-1. Block diagram of the proposed wireless power transfer system with detection circuit
(detecting supply current and coil voltage)

Figure 7-1 shows the block diagram of the proposed wireless power transfer system with

detection circuit. The class E transmitter operates at 240 kHz. There are three parameters that can

be extracted from the wireless power transfer system without establishing a communication link

between the transmitter and receiver. They are namely, the coil voltage, the supply current, and

the coil current.

Coil voltage is extracted by rectifying the coil voltage via a high impedance path using a

half wave rectifier. R0coil and C0coil shown in Figure 7-1 are used to smooth out the rectified

voltage. Rcoil is also used to regulate the current flow in the rectifying diode to prevent sudden

and large current spikes from damaging the components. The voltage is then stepped down via a









potential divider using Rdividerl and Rdivider2 to a workable voltage without damaging the input port

of the micro-controller. To mitigate loading effects and reduce high frequency noise, a buffer

using a low speed operational amplifier, e.g., LM324, in the voltage follower configuration can

be used before the micro-controller's ADC port. Higher order low pass can be achieved by

adding appropriate capacitors to the operational amplifier as a simple first order low pass filter.

Supply current is extracted from the circuit via the current sense resistor Rsense in Figure 7-

1. The Rsense resistor is located at the class E transmitter ground before returning to the system

ground, instead of locating at the high side before LDC. Therefore, any voltage drop across the

resistor will be referenced to the system ground instead of the supply voltage. Since the voltage

drop will be extremely small (0.1V or less), a non-inverting amplifier using an operational

amplifier LM324 can be used to amplify the voltage, mitigating any loading effects and reducing

high frequency noise. Similar to the extraction of coil voltage, extra low pass filter can be added

by the same technique discussed.

Measurements of the coil current can be realized by using either current sense transformers

or current sense resistors. Current sensing transformers are typically large in size and operate at

frequencies lower than 100 kHz, making them impractical for this system. A current sense

resistor can be added to the low side of the coil to measure the voltage drop across the resistor.

This is more practical than placing the current sense resistor on the high side as shown in Figure

7-2. This is because extremely high tolerance resistors are needed for the potential dividers if

they are placed on the high side. Detecting a high frequency AC current with respect to ground is

difficult because the ground of such a high voltage and current system is extremely noisy relative

to the high side of the transmitting coil. The ground noise will contain frequencies including the

operating frequency making it not possible to mitigate its effects via filtering. Unlike measuring









the supply current, which does not require rectification, both the high frequency voltage signal

and the ground noise will be rectified. Therefore, it is not possible to perform low pass filtering

to mitigate the effects of ground noise using a low speed operational amplifier. Since the ground

noise is unpredictable and changes according to loading conditions, it is difficult to perform

precise and stable measurement of the coil current. For the above reasons, only the coil voltage

and supply current will be extracted.

iRsense
AC Out so iF -e7bfehco idtrDiudttnfh

R1 H R1 L

D ,collI
Microcontroller
TX Coil
Differentia
=-I col Amplifier R2 H R2 L


Ground T ICOI



Figure 7-2. Schematic of coil current extraction network using a current sense resistor.

A DC switch shown in Figure 7-1 before the choke inductor LDC is used to turn off the

transmitter under no-load mode or fault mode which includes over current. A low turn-on

resistance PMOS transistor using a low power NMOS transistor to pull down its gate voltage can

be used as the switch. Additional over current protection can be implemented by using a

polymeric positive temperature coefficient device (PPTC) at the supply as a second level

protection. Reverse polarity voltage protection can also be implemented by adding a reverse

biased diode in shunt with the supply voltage. If a reverse voltage is applied, the diode will cause

a short circuit and the PPTC will be activated to disconnect the supply path. Over-voltage

protection of sudden spiking can be implemented by using a transient voltage suppression (TVS)









diode commonly referred as a transorb across the supply. The TVS diode can also be doubled as

the diode for reverse voltage protection. To prevent thermal runaway, a thermistor can be located

next to the transistor of the power stage. By using the thermistor as part of a potential divider on

which the supply voltage is applied, the temperature can be tracked by reading the voltage across

the thermistor using the micro-controller's ADC.

7.1.2 Detection Flowchart/Logic

The detection scheme flowchart is shown in Figure 7-3. It can be implemented using a low

cost micro-controller, such as 16F688 by Microchip. It is found experimentally that the system

reaches a steady state after it is being powered on for 1 ms. This is when a decision to power on

or off the transmitter can be made. The time to steady state is dependent on the operating

frequency as well as the self inductance of the transmitting coil.

The no-load and safe modes follow a similar logic flow. The only difference is that the

transmitter is powered down if no-load mode is detected whereas the transmitter is powered up if

the safe mode is detected. Both modes will probe the circuit for supply current and coil voltage

after each predetermined X seconds to determine its operating mode, (a reasonable number for X

is 1). Increasing X will incur higher latency to make the system response slow and decreasing X

will incur higher no-load power consumption. The operating modes are determined by the supply

voltage and coil voltage space, which will be discussed later. The thresholds are dependent on

the supply voltage, transmitter's component values, and transmitting coil parameters. The

dependence of the parameters on the receiver is found to be weak. Therefore, the thresholds

work independent of receivers as well. Hysteresis needs to be implemented in the code so that

the system does not oscillate when its operating mode is at the borderline case. The fault counter

Z is reset if the system ends up in either mode.












































Figure 7-3. Detection scheme flow chart for proposed system.

When the coil voltage and supply current is excessive, the system enters its fault mode. A

common cause of fault mode is when a huge piece of metal is placed over the transmitting coil,

which decreases the total effective inductance of the transmitting coil significantly. When this

happens, the class E ZVS/ZDS operation is no longer valid. If the transmitter is not powered

down immediately, the transistor will be damaged due to excessive power dissipating as heat.

The delay to probe the system to determine if the cause of the fault has been rectified is increased

by a factor A on top of the original X delay. The delay is increased with increasing occurrence of









fault so that the system does not need to repeatedly stress the system within such a short window

of time. Once the number of tries reaches N, the system will enter into the fatal fault mode, and

the only way to exit from the mode is to perform a hard reset which involves disconnecting the

DC supply of the system.

Other fault modes which are not covered by the flowchart include thermal run away mode

due to excessive heating at the power stage and battery fault mode for a one-to-one system. As

will be shown in the experimental verification, it is possible to track the charge/power received

status of a receiving unit for a one-to-one power system. If the trend of the power delivered over

time deviates from the expected trend, the system will enter a battery/receiver fault mode which

also requires a hard reset. This also prevents possible damage when a user places a non-

compliant receiver of a different charging profile on the transmitter. Brownout and over-voltage

at the supply can also be detected by a supply voltage monitoring network so that the transmitter

can be powered down under fault mode. A supply monitoring network is simply a potential

divider to drop the supply voltage to the range of the micro-controller's ADC so that the micro-

controller can make a decision based on the ADC's input.

7.2 Experimental Verification

7.2.1 Test Bench and Circuit

The wireless power transfer system is fabricated based on the component values shown in

Table 7-1 which was selected using design rules presented in Chapter 3. The transmitting coil

used is similar to the one used in Figure 3-1. To ensure that the proposed detection scheme can

be applied to the same transmitter platform regardless of the receiver size, two different receiver

sizes were used in the experimental verification. All coils were fabricated using 100/40 round

served Litz wires. The supply voltage for the system is selected to be 12 V and can be varied

depending on the power level requirements. Supply voltage should not be considered as a factor









during the analysis. The fabricated transmitter shown in Figure 7-4 has a size of 15 cm x 2 cm. It

is designed to be long and narrow so that it can be placed beside the transmitting coil as a single

integrated unit. The low power detection and control block is located away from the high voltage

power stage to reduce noise and coupling effects. The power input jack is located between the

two blocks for the same reasons.


Detection and Class E Transmitter

Control (Power stage)
Figure 7-4. Photograph of the fabricated transmitter circuit with control circuit.


Table 7-1. Component Values for Load/Fault Detection Test Bench


Value


Coil Specifications
TX coil inductance
TX coil dimension
TX coil parasitic resistance
RX coil inductance (big RX)
RX coil dimension (big RX)
RX coil parasitic resistance (big RX)
Mutual inductance (big RX)
Coupling coefficient (big RX)
RX coil inductance (small RX)
RX coil dimension (small RX)
RX coil parasitic resistance (small RX)
Mutual inductance (small RX)
Coupling coefficient (small RX)
Circuit Specifications
LDC
Crx (big RX)
Crx (small RX)

Lout
Lout parasitic resistance
Cshunt


45.3 gH
20 cm x 20 cm
0.5 Q
5.45 gH
9 cm x 6 cm
0.235 Q
2.81 gH
0.178
4.00 gH
6 cm x 5.5 cm
0.22 Q
1.88 gH
0.140


500 gH
68 nF
95 nF
9.4 nF
10 gH
0.16 Q
12 nF











Fl, PPTC Resettable Fusel.5A hold, 3.OA trip
MINISMDC150F/24-2



U10F
DC +- S LM3480IM3-
Supply < IN 5.0 OUT -
(12V) \ / Control ^
U) C? GND Supply (5V)
1---- ^ --


S. i of p ffiver





Figure 7-5. Schematic of power stage of fabricated transmitter.


C2, 9.4nF
H 7Coil High
Side


Figure 7-6. Schematic of driver stage of fabricated transmitter.






















Cdecoupling

-1 h


R14, 9k


Figure 7-7. Schematic of detection and control stage of fabricated transmitter.

Figure 7-5 to Figure 7-7 shows the schematic of the fabricated transmitter. Figure 7-5

shows the power stage which handles the over current protection, reverse voltage protection and

on/off control of the transmitter. Figure 7-6 shows the driver stage which includes the class E

transmitter and clock. Figure 7-7 shows the detection and control stage. The transmitter has two

levels of protection, hardware protection and software controlled protection.









Hardware protection is used as a last resort protection for which the thresholds are set to be

higher than the software controlled protection. It should only be triggered when the micro-

controller (U3) in Figure 7-7 fails. Over current protection is realized by a PPTC fuse, Fl as

shown in Figure 7-5 which has a 1.5 A hold and 3.0 A trip. The transient voltage suppression

(TVS) diode, Dl in Figure 7-5 serves two purposes. It helps to eliminate any transient voltage

spikes due to discharge from the energized coil during powering down, and protects the

transmitter from over-voltage. In addition, if the user accidentally connects the DC supply in

reverse polarity it will attempt to short out the supply and load excessive current. The fuse Fl

will be activated to break the connection protecting the transmitter.

An intermediate stage of over current protection is added to the transmitter via disabling

the PWM clock, U2 in Figure 7-6. This is achieved by pulling the CS pin to high. Values of R15

and R16 in Figure 7-7 are selected to shut down the clock when supply current exceeds 1.2 A

which is higher than the software determined maximum current of 0.85A as shown in Figure 7-

12.

The transmitter should rely on software control protection for its nominal operation. This is

achieved by turning on and off the PMOS transistor Q2 via a low voltage control generated by

the micro-controller (U3) using a NMOS transistor to pull down the gate voltage of Q2 via RI

and R2. It is important that the regulator for the detection and control stage Ul is placed before

the Q2 transistor so that the detection and control stage is still in operation even when power is

being cut to the driver stage. Using a quad operational amplifier LMV324IDR as shown in

Figure 7-7, the micro-controller (U3) is able to read in 4 different parameters of the system. They

are coil voltage (U4A), supply voltage (U4B), driver temperature (U4C) and driver stage current

(U4D).









Coil voltage and driver stage current extraction has already been discussed in the previous

section. The thresholds are predetermined by sweeping the loads as shown in Figure 7-12. By

extracting the supply voltage, the transmitter is able to prevent over-voltage conditions as well as

under-voltage conditions that might cause the system to deviate from its nominal operation. The

system is set to a supply voltage operating range of 9 V to 15 V. In addition to detecting the

supply voltage, the transmitter is able to vary its thresholds accordingly to make it more robust.

The thresholds can be stored in the flash memory of the micro-controller (U3). A 0.5V resolution

is sufficient to prevent any false alarm or damage to the system. Finally, a thermistor (RI 1) is

placed next to the class E driver's transistor to monitor its temperature. The system will shut

down when temperature exceeds 750C and only resume nominal operation when the temperature

drops below 600C.

7.2.2 Experimental Results

90

80 ----

70 --



X- 40

S40 --

30 -

20 -

10 -
0 1 2 3 4 5 6

Power Delivered to Load (W)
Small Coil ---Big Coil

Figure 7-8. Efficiency-Power plot of single receiver setup (Solid black line: Small coil. Dashed
gray line: Big coil)










Figure 7-8 shows the efficiency-power plot of a single receiver setup using the big and

small receivers being placed on the same transmitting coil. Since both of the receiving coils are

optimized for operation on the same transmitting coil, the efficiencies of both receivers are

similar for the respective power level. Better than 60% efficiency can be achieved for power

delivery level above 1.5 W, keeping absolute power loss of the system low at all times, which is

important to ensure the system does not overheat. Due to stronger coupling, the big receiving

coil has a slightly higher efficiency as well as power delivery. Power delivery is about 6 W using

the big coil and about 5 W using the small coil.

Copper plate sliding
ZVS starts to fail laterally to cover TX coil
250

225


E d
> 175

150 Q

125
-0 s small Coil Transistor #
100 cm x 5.5 cm failure '
0 No Load
0 75 Big Coil
9 cmxm x 6 cm
50 -
Copper plate
25 moving closer to
TX coil vertically
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Supply Current (A)
Figure 7-9. Examples of different loading conditions and fault modes on coil voltage versus
supply current space.

Figure 7-9 shows the coil voltage and supply current space diagram which is used to

determine the different operating modes. The system enters the no-load mode when the operating

condition is in the vicinity of the diamond shape. A margin of 10 V and 50 mA can be use to

frame the no load operating point so that the system will be more robust under component









variation. The safe zone is at lower coil voltage (less than 80 Vrms) and current (less than 0.85 A)

region where two receivers are at their nominal operation without being over loaded. Therefore,

safe zone is located at the bottom left corner of the coil voltage and supply current space as any

excessive voltage or current will damage the transmitter. The small receiving coil generates a

larger transmitting coil voltage because the same amount of power delivery is required over a

smaller receiving area, which requires a stronger magnetic field. Two different potential fault

scenarios, a large copper plate being brought closer to the transmitting coil and a large copper

plate being slid over the transmitting coil, are shown in Figure 7-9. Although, there are other

fault scenarios such as placing a smaller copper sheet in the middle of the transmitting coil, the

two scenarios tested should be able to provide sufficient understanding on how the system will

react under various fault modes. The solid dashed line shows the trend of coil voltage and supply

current moving clockwise when the distance between the transmitting coil and a larger copper

sheet becomes smaller. ZVS/ZDS starts to fail at the voltage inflection point as more energy is

dissipated across the transistor instead of being transferred to the load (copper pate). If the

system does not shut down immediately the transistor will be damaged due to heating. The

hollow dashed line shows an increasing coil voltage and supply current when the overlapped area

between the transmitting coil and a large copper plate increases. The trend of the line is expected

to follow the same as the experiment which the copper plate is brought closer to the transmitting

coil. However, the coil voltage and supply current is so excessive that the transistor will be

damaged due to non ZVS/ZDS operation. Therefore, measurements were not carried out beyond

2 A supply current. Since high coil voltage will always lead to over-voltage problems at various

points of the circuit and damage the components, it is not important to study the trend of the

curve during fault operation as the priority is to shut down the system as quickly as possible.










90

8 0 ...........

70

O 60

S50-





20

10
0 1 2 3 4 5 6 7
Power Delivered to Loads (W)

Figure 7-10. Efficiency-power plot of three sets of dual load measurements and two sets of
single load measurements using a combination of big and small receiving coils.

Three sets of measurements were carried out on the system using two loads of load

resistance ranging from 200 to 4000 Q in 14 discrete steps. The three sets are two big coils, two

small coils, and one big coil with one small coil. The pair of receiving coils are placed side by

side in the middle of the transmitting coil with a 2 cm gap between each other. The efficiency-

plots of the three measurements are shown in Figure 7-10. In addition, two single-receiver

measurements are also shown in Figure 7-10 for comparison. DC-DC efficiency is above 70%

for power delivery levels above 3W, crossing 85% for some load points. The spread of efficiency

is approximately 5% regardless of the number of receivers. All load conditions achieved

ZVS/ZDS operation, maintaining high efficiency operation with minimum power loss. Since the

efficiency of the system is well bounded regardless of loading conditions and their sizes, the

spread in the coil voltage and supply current space should be limited as well.









80

No Load -
72
168
S64 Single Load .






no44 sigl Nload an Dual Loads
4 0 *Op


352



no-load, single load, and dual loads.






Figure 7-11 shows the transmitting coil voltage and supply space for a single load (big coil

and small coil) and dual loads. Although, there is some overlapping between the 2 spaces, it is

possible to detect the number of loads for most of the loading conditions. As shown in Figure 7-

11 a sharp transition will be observed when an extra load is being placed on or removed from the

transmitting coil. Therefore, the system can easily detect if an additional receiver is placed on the

transmitting coil or being removed from it by tracking the transmitter coil voltage and supply

current over time. It is also possible to detect the number of loads when the transmitter is

powered on with valid loads on the transmitting coil because during the initial power-on states of

most electronic devices, they will not draw much power. Therefore, the power delivery will

slowly ramp up during the first few seconds when the transmitter is powered on. Due to this

intrinsic characteristic of most electronic devices, the system will typically observe an initial coil

voltage of 40 Vrms to 48 Vrms for dual loads and 52 Vrms to 58 Vrms for single loads. Under the










worst case situation, if it is not possible to differentiate the number of loads being placed on the

transmitter, the system can track the loading conditions via techniques such as Markov Chains

over a specific period of time and determine its loading conditions. This requires large amount of

training data to perform load pattern analysis which is beyond the scope of this dissertation.

250
Invalid zone because supply
225 current Is not high 010 ugh lto
I g e An v It lhigh -.uil 4 mll. 1 o ^K
200 -0

175


Undkesrable zone
S125 Operation might
cause thermal
100 NO run away 0*
75 i ZVS not achieved






0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Supply Current (A)
Figure 7-12. Transmitter coil voltage and supply current space diagram illustrating three
different zones, no-load, safe and fault.

Summarizing the above discussions, Figure 7-12 shows the transmitter coil voltage and

supply current space indicating the various zones. The no-load zone is enclosed by red circle and

the safe zone is enclosed by blue lines. The no-load is basically a point solution with a radius for

system tolerance purposes which can be simplified into a rectangle or square. The zone can be

described using two points in the space or four data points for which each point is described

using the transmitter coil voltage and supply current or simply the four sides of the

rectangle/square. The safe zone has two steps. For supply current below 0.3 A, the transmitting

coil voltage should be between 32 Vrms and 64 Vrms whereas for supply current above 0.3 A, the









transmitting coil voltage should be between 32 Vrms and 80 Vrms. Any supply current greater than

0.85 A is considered as a state of over current for this system and determined as fault mode. Five

data points are required to describe the safe zone. In fact, any operating condition not within the

safe or no-load zone is considered to be in the fault zone. Therefore, each unique supply voltage

and operating frequency will require nine data points to describe its coil voltage and supply

current space. The purple zone on the top left corner of Figure 7-12 is an invalid zone because

the supply current is too low to generate such voltages of those magnitudes and would never

occur. ZVS/ZDS operation is still valid to some extent in the area above the safe zone. However,

the transmitting coil voltage is very high and might damage components. In addition, excessive

transmitter coil voltage might lead to thermal run away on some components especially the

transistor. The brown area on the right of the safe zone is when the transmitter no longer operates

under the ZVS condition. When this happens, large amount of energy is dissipated across the

transistor as the high voltage across the drain of the transistor and high current through the

transistor are no longer orthogonal in time.

An interesting linear relationship between the supply current and the power delivered to

the load is shown in Figure 7-13. The results from the five sets of measurements can be

consistently described using the equation y = 0.095x + 0.055 where the y-axis is supply current

to the transmitter and the x axis is the power delivered to the load. Based on the 12 V supply

voltage, for every 1 W of power at the receiver the transmitter requires an input power of 1.14

W. Once the system is operating in the safe zone, using the proposed equation, Figure 7-14

shows the plot of calculated delivered power via measured supply current to the transmitter

versus the actual delivered power. The results show relatively good agreement at all power levels

with an average error of -0.08 W and standard deviation of 0.16 W for 616 sets of reading.




























5 6


Power Delivered to Loads (W)


Figure 7-13.


Linear relationship between supply current and power delivered to the loads for all
5 sets of measurements. Solid dashed line: y = 0.095x + 0.055.


0 1 2 3 4 5 6 7


Measured Power Delivery (W)


Figure 7-14.


Power delivery error distribution plot. Calculated power delivery is based on
measured supply current. Solid dashed line: ideal error free calculation.


0 1


3 4


7 8









Although the experimental verification shows the feasibility of the proposed load/fault

mode detection scheme, the thresholds for different modes and the equation to track the power

delivery are highly dependent on the supply voltage and operating frequency. Therefore, a

voltage sensing mechanism is implemented to detect the supply voltage and make appropriate

adjustments to the thresholds. The operating frequency is assumed to be stable for this case by

using 1% tolerance components for the RC timing network. A more effective method is to use a

crystal reference to generate the clock frequency; a low cost 30ppm crystal will generate a clock

signal with negligible frequency offset. However, by using this technique the frequency will be

locked to a single frequency or multiple/fraction of the operating frequency making fine

frequency resolution control not possible.

7.3 Extension of Load/Fault Detection Scheme

7.3.1 M:N Coupling Structure

The analysis of the load/fault detection scheme is confined only to single transmitting coil

architecture. As shown in Figure 5-23, placing two receivers concurrently onto a single

transmitting coil will "choke" the power delivery, as the power delivery of one receiver is

dependent on the loading condition of the other receiver. This observation is supported by

Equation 4-6 for which the real part ofZtx increases with each additional receiver while the

reactance of Ztx remains approximately the same. Therefore, the power delivery remains

approximately the same but is being split between two receivers. Instead of using a single

transmitting coil, a M:N coupling structure shown in Figure 7-15 is proposed. The M:N coupling

structure can be implemented to support multiple portable devices with a one-to-one pairing or

cover a large area by using a large array of transmitting coils to power compact sized receivers.

In depth analysis of the M:N coupling structure is beyond the scope of the discussion as the

focus will be on system level. For the following analysis and experimental verification, the focus








will be on a M:N coupling structure supporting two portable devices. A single class E transmitter

is used to drive two similar transmitting coils (15.5 cm x 17.5 cm) being placed beside each

other. Two identical receiving coils (9 cm x 6 cm) are being used. The design rules in Chapter 3

are used to select the components.


-x


I


A B

Figure 7-15. Examples of M:N coupling structures. A) Supporting two portable devices and. B)
Large array of transmitting coils covering a large area.

Figure 7-16 shows the power "choking" delivery trend of a dual receiver setup when both

receivers are being placed on a single transmitting coil while leaving the other coil unloaded. The

power delivery is dependent on the other receiver with power level above 1W is observed, which

is similar to the trend in Figure 5-23. The "choke" point is lower than that of Figure 5-23 because

the bigger receiving coil results in stronger coupling. A stronger coupling between the

transmitting and receiving coils will enhance the dependency of power delivery between two

receivers being placed on a single transmitting coil. Figure 7-17 shows the power delivery trend


" .

se...**"...e


Trnmitn


Trnmitn


Transmitting coil A



Transmitting coil 6


Transmitting
coi I ZZ


Transmitting1
coiml AZ










of a dual receiver setup when one receiver is being placed on each transmitting coil. The power

"choking" phenomena is not observed for this case, power delivery to each receiver is

independent of one another. This is because the coupling between each transmitting coil is very

weak as they are placed next to each other instead of overlapping each other. Therefore, it can be

assumed to be two independent single transmitting coils to a single receiving coil system. Power

delivery to the individual receivers is not balanced because the coils are hand wounded with

tolerance of self inductance as high as 10%. This problem will be eliminated during mass

production as the coils will be machine wounded or PCB fabricated with much higher tolerance.

As shown, by using the M:N coupling structure the problem of power chockingg" can be

eliminated. To ensure that the design is feasible, further analysis on the efficiency, validity of

transmitter coil voltage and supply current space as well as power delivery.


4.0

3.5
r*

3.0
83
g 2.5
0

* 1.5
S1
1


-S- -. -- -


-
J0.5



0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

Power Delivered to Receiver 2 (W)


Figure 7-16. Measured power delivery of a dual receiver setup on a dual transmitting coil test
bench with both receivers on a single coil leaving the other coil unloaded.


























0 1 2 3 4 5 6 7 8 9

Power Delivered to Receiver 2 (W)


Figure 7-17. Measured power delivery of a dual receiver setup on a dual transmitting coil test
bench with one receiver on each transmitting coil.

Figure 7-18 shows the efficiency-power plot for both test results of Figure 7-16 and Figure

7-17 including the results for a single receiver setup. Efficiency is high for both cases achieving

a better than 75% efficiency for power level above 4 W. Efficiency follows the same trend for

setup with single receiver and dual receivers on a single transmitting coil confirming the design

rules presented in Chapter 3 can be applied to a M:N coupling structure. Slight efficiency

improvement is observed for the setup with a receiver is placed on each of the individual coils.

This can be explained by the fact that, although one of the transmitting coil is not loaded, current

is still running through the coil. The current will cause energy to dissipate across the coil's

parasitic resistance as high level of reactive power is being driven into the unloaded coil.

However, if the coil is loaded, the resistance seen by the coil due to its load is much larger than

the parasitic resistance of the coil and the losses can be neglected for this case, improving the

efficiency of the system.































Figure 7-18.


0 4 -1 -- --
0


0- -- -- -

0

0

0

0

0
0 2 4 6 8 10 12 14 16 18

Power Delievered to the Load(s) (W)


Efficiency-power plot of two sets of dual load measurements and two sets of single
load measurements. (Green lines: Single receiver, Blue lines: Dual receivers on
single transmitting coil, Red lines: Dual receivers on dual transmitting coil)


110
105
100
95
90
85
80
75
70
65
60
55
50


0.2 0.4


Figure 7-19.


0.6 0.8
Supply Current (A)


Transmitter coil voltage and supply current space. (Green diamond: No load, Black
lines: Single receiver, Blue lines: Dual receivers on single transmitting coil, Read
lines: Dual receiver on dual transmitting coil)









Figure 7-19 shows the transmitter coil voltage and supply current space for various loading

conditions which is similar to Figure 7-12. Therefore, it is possible to extend the same load

detection technique for a M:N coupling structure. The discrepancies shown in Figure 7-19 for the

single receiver case is due to the tolerance of the hand wounded transmitting coil as discussed

earlier. It is observed that the two different conditions of loading the transmitting coils with two

receivers resulted in two distinctive zones in the transmitter coil voltage and supply current

space. By using techniques of pattern recognition and sufficient training sample size, the

transmitter is able to determine the number of receivers on each individual coil. Although there

are some overlaps between a single receiver setup and the setup with two receivers on a single

transmitting coil, the trends of the lines are distinctively different following a different charge

profile trend. The trends can be modeled into different sets of Markov Chains for which over

time the state of the system can be determined.

1.4










0.4 ---- --------------

0.2 -
0 I
2- 0.8 ----- -.4-------







0 2 4 6 8 10 12 14 16 18

Power Delievered to Load(s) (W)


Figure 7-20. Linear relationship between supply current and power delivered to the loads for all
4 sets of measurements. Solid dashed line: y = 0.076x + 0.048.










18 -

0 16



12 16

010








0
I< 4 10 -- -- --h- --



0 --- --- --



0 2 4 6 8 10 12 14 16 18

Measured Power Delivered to Load (W)


Figure 7-21. Power delivery error distribution plot. Calculated power delivery is based on
measured supply current. Solid dashed line: ideal error free calculation.

Figure 7-20 shows that the linear relationship between the supply current and power

delivered to the load(s) still holds for a M:N coupling structure regardless of the placement of the

two receivers. The linear relationship can be described by the equation y= 0.067x + 0.048. Figure

7-21 shows the power delivery error distribution plot using the equation to calculate the

delivered power. The average error is found to be 0.017W and standard deviation error is found

to be 0.183W out of a maximum of 18W power delivery. This shows a very small spread in

errors which verified the feasibility of using the power tracking scheme on a M:N coupling

structure. Similar trends are expected for higher number of transmitting and receiving coils

attached to the class E transmitter. However, as more transmitting coils are attached in parallel to

a single transmitter, the effective inductance will be reduced causing the circuit to be more

sensitive. Increasing the transmitting coil size or number of turns will help to mitigate the effect.










300
W 275
> 250
S225
200 ------ --
o 175
= 150
O 125
100 -
575 --------
v, 50
S 25


0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0
Supply Current (A)

Figure 7-22. Transmitter coil voltage and supply current space with fault tests. (Green lines:
Fault tests without valid receiver on one of the coil, Purple lines: Fault tests with a
valid receiver on one of the transmitting coils)

Five different common fault scenarios are being studied for the M:N coupling structure.

They all involved a large piece of copper sheet (30 cm x 30 cm) sliding across the coil or being

brought to close proximity on the same plane. All of the fault lines green and purple follow a

clockwise trend with increasing severity of the fault (proximity of copper sheet or amount of

overlap). The green lines in Figure 7-21 show three different fault scenarios without any valid

receiver being placed on one of the transmitting coils. The detection scheme is capable of

detecting a fault condition when a valid load is not placed on one of the transmitting coil (green

line), but there are some overlaps between the load conditions and fault conditions when a valid

load is being place on one of the transmitting coils while a piece of copper sheet is being slide

across or brought closer to the unloaded transmitting coil (purple line). Although the overlapping

region is small, an analysis is required to determine if it is acceptable for a "minor" fault to be

considered safe.









The analysis starts with looking at what happens when a sheet of copper is partially

overlapped onto a transmitting coil. Due to the close proximity of the copper or any electric

conductor of sufficient mass, the self inductance of the coil will be reduced. The reduction is

due to the proximity and amount of overlap. For this setup, two coils are driven in parallel.

Therefore, with the reduction of self inductance of one of the transmitting coils, due to a copper

sheet in close proximity, the total effective inductance looking into the pair of coils will be

reduced as well. The reduction in effective inductance of the dual parallel structure will cause a

reduction in the phase angle of Zt, making Ztx more capacitive. A more capacitive Ztx will result

in higher power being transferred or higher unregulated receiver voltage. This observation is

similar to decreasing C,,ut which will be discussed in Chapter 8. Although a slight overlap

between the unloaded transmitting coil and a piece of copper sheet may seem to of not a major

problem, it will cause the voltage before the voltage regulator of the valid load to increase.

Therefore, if the maximum input voltage of the voltage regulator used on the receiver has

insufficient margin, the voltage regulator will be damaged. To prevent the voltage regulator from

being damaged, the voltage regulator must have a high input voltage tolerance which makes the

receiver bulking and costly. It will also limit the selection of voltage regulators available to the

designer.

A potential solution to the problem is to design the transmitter unit so that each

transmitting coil is driven by an independent transmitter using a master clock and a master

micro-controller as shown in Figure 7-22. Load/fault detection is done by each individual pair of

transmitters. Since the coupling between neighboring coils is weak, the system can be assumed

to consist of multiple 1:1 or 1 :N transmitting systems which the load/fault detection scheme is

proven to be effective earlier in this chapter. Depending on the load conditions, the transmitters








can be turned on and off accordingly, making it more energy efficient when one of the coils is

unloaded. In addition, if a fault occurs on one of the coils it can be shut down while power

delivery is not affected by the other transmitting coil. Since each transmitting coil will require

their respective set of class E transmitter circuitry, it will increase the board size, components

count and potentially increase the cost. However, each class E transmitter will most probably be

required to drive a single load instead of multiple loads. Thus, the power handling of each

transmitter will be reduced, allowing the use of smaller and more compact components. Due to

the reduction in component size and power handling, making each transmitter more compact and

cheaper offset the cost of multiple units.


Trnmitn coi A


A B

Figure 7-23. Different methods of driving a 2:N structure. A) Original M:N design for the
experimental verification and B) proposed improved M:N design.
7.3.2 Removing L,,out from Transmitter
Since a Class E power amplifier only requires a phase angle of Zt to be from 400 to 700, it

is possible to remove Lout from the transmitter load network using part of the leakage inductance

(imaginary part of Equation 4-6) of the transmitting coil as Lout (shown in Figure 7-24) as long as

the phase angle can be maintained within the high efficiency range across all receiver load

resistances. Therefore, the design of the transmitting and receiving coils must be integrated as

part of the design rule and can no longer be treated as a generic linear block. Since L,,ut is the


Trnmtigci








component with the highest loss in the class E transmitter, removing Lout will improve the

efficiency by at least 5%. In addition, when the transmitting coil is not loaded, a significant

amount of power is lost across the inductor and the transmitting coil. As Litz wire is being used

to fabricate the transmitting coil, most of the losses will be across L,,out. Therefore, by removing

Lout from the system, the no load power consumption can be reduced significantly as well.

However, doing so will cause the higher order harmonics content to be stronger across the coil as

Lo,,ut also serves as a low pass mechanism. The radiation efficiency of the coil improves with the

increase in frequency, resulting in the higher order harmonics being transmitted into free space.

Therefore, care must be taken during the design to ensure the emission is low enough to pass the

FCC part 18 test, this is important especially for high power system.









SCshunt out Ccrx Rload
AC



Figure 7-24. Block diagram of wireless system without Lout.

Using the same experimental setup as the M:N coupling structure, Lout is removed and Cout

is tuned to compensate for the missing positive reactance by decreasing its capacitance value.

The same fault test procedures are carried out on the test bench with Lout removed so that a fair

analysis can be made. Results of the fault tests using the same format as Figure 7-22 are

presented in Figure 7-25.










290
270
> 250
230
S210
O 190
= 170
0
- 150
E 130
* 110
90
70
1- cmn


-00
- -^-^ ^ _^
-^^ -


0 0.5 1 1.5 2 2.5 3 3.5 4

Supply Current (A)

Figure 7-25. Transmitter coil voltage and supply current space with fault tests with Lout
removed. (Green lines: Fault tests without valid receiver on one of the coil, Purple
lines: Fault tests with a valid receiver on one of the transmitting coils)

Similar initial trend of the test results shown in Figure 7-22 was observed for all the five

different fault tests with the lines going in a clockwise direction. However, the lines follow an

upward trend while looping back and moving into the low voltage low current region instead of

the downward trend as show in Figure 7-22. This results in the certain conditions of the fault

mode overlapping with the safe or valid load region. All of the conditions involved the copper

sheet being in close proximity or almost complete overlap onto the transmitting coil. This

condition is a common case when the transmitting coil is accidentally flipped over onto a

metallic table top or an unshielded transmitting coil being placed onto of a metallic table top. By

removing Lout from the transmitter using existing circuitry, the system will be unable to

differentiate between a valid load and a fault mode under certain conditions. Therefore, to ensure

robust operation while removing Lout, modifications of the detection circuit are required.









III I U,


Figure 7-26.


High 2nd order
harmonics
during fault
mode


Waveform of the transmitter circuit without Lout when fault conditions overlaps
with valid load conditions. (Red: Drain voltage, Orange: Coil voltage, Green: Coil
Current)


Figure 7-27. Spectrum of transmitter coil voltage without Lout when fault conditions overlaps
with valid load conditions.


/ Transistor
switched on.

u-J

' T T i j I .









To better understand the mechanism which cause this upward trend to happen, the

transmitting coil voltage and its spectrum was captured using an oscilloscope. As seen from

Figure 7-26, the second harmonic is much stronger than the fundamental tone for both the

transmitting coil voltage and coil current with the transmitting coil voltage at approximately 150

V peak to peak. This is supported by the measured spectrum of the transmitter coil voltage

shown in Figure 7-27. The fundamental tone is approximately 27 dBV while the second

harmonic is almost 31 dBV. The second harmonic can be considered as detection noise as the

load/fault detection scheme neglects the effects of higher order harmonics. With a stronger

second order harmonic on top of the fundamental tone, the transmitter coil voltage will be larger

than expected, explaining why the transmitting coil voltage is larger than expected or has an

upward trend into the valid load region under certain fault conditions.

















/ / / T/
-------------------?----^"----^u ----^v----x










Figure 7-28. Waveform of the transmitter circuit with L,,ut when fault conditions overlaps with
valid load conditions. (Red: Drain voltage, Orange: Coil voltage, Green: Coil
Current)
































Figure 7-29. Spectrum of transmitter coil voltage with Lout when fault conditions overlaps with
valid load conditions.

Figure 7-28 and Figure 7-29 show the same waveforms as Figure 7-25 and Figure 7-26 but

with Lout added back to the transmitter circuit. The higher order harmonic are being suppressed

by L,,out. The second harmonic is reduced by approximately 10dB at 19 dBV with the fundamental

tone remained at approximately 27 dBV. The transmitting coil voltage is reduced significantly

from 150 V peak to peak to 70 V peak to peak. This brings the transmitting coil voltage below

the safe zone lower threshold.

Although removing L,,ut is able to improve efficiency and reduce heating, higher harmonics

contents which are filtered by L,,ut are allowed to pass. This affects the load/fault detection

scheme as detection noise. The error in detection can also be mitigated by using a simple low

pass RC filter before the detection diode to reduce the higher order harmonics. This can be

realized by adding an appropriate capacitor to ground between R6 and D2 of Figure 7-7. If

needed, a higher order low pass filter can be added in the coil voltage detection path to further









reduce the higher order harmonics. By adding the low pass filter, the original detection scheme

can be used without causing any detection error. However, the higher order harmonics voltage

across the transmitting coil will still be a potential issue during the emission test as the same

transmitting coil will become a better radiator with the increase of operating frequency.









CHAPTER 8
RECEIVER VOLTAGE CONTROL

Although, the system is able to reduce the power delivered to the receiver when the

effective load resistance increases, the unregulated DC voltage at the receiver will increase with

increasing load resistance. The swing in receiver voltage is not important when the power level

of the system is low as the voltage regulator at the receiver is able to tolerate such fluctuations.

However, the range of voltage swing increases with increasing power rating of the system. A

buck regulator will have poorer efficiency with a higher input voltage and tend to be larger in

size as external switch transistors are required to perform the regulation. In addition, if a high

input voltage is required for the receiver voltage regulator, the choices for the system designer

will be limited and BOM cost will be higher as well. Therefore, it will be desirable for the

unregulated voltage at the receiver to be bounded to a certain range.



1+ LD Ztx


Lout Cout
AC Cshunt Rrx R/oad





Figure 8-1. Block diagram of wireless power system.

From the simplified block diagram of the system as shown in Figure 8-1, there are limited

parameters of the system that one can vary. Most probably the structure of the coil will be fixed,

any attempt to vary the coil structure in terms of turns and area will make the system

complicated. Capacitor Crx can be tuned to control the received power level. There are two ways

that Crx can be tuned; it can be achieved by either using a varactor or switching between banks of









capacitors. Since the system is intended to operate with the frequency range of less than 1 MHz,

it will take a considerable amount of capacitance to achieve an appropriate tuning range.

Although, it is possible to achieve it with switchable banks of capacitors, it is not practical for

the current implementation unless a high voltage and current switch is realized in the form of an

integrated circuit. In addition, it is desirable to keep the receiver as simple as possible and move

all the complexity to the transmitter as a transmitter will typically be used to power multiple

receivers concurrently or at different instances. There are three possible ways to achieve voltage

control of the unregulated receiver voltage on the transmitter. They are varying supply voltage,

varying operating frequency and varying C,,t. Each option will be studied with experimental

results from a one-to-one setup test bench. The test bench used in the experiment is intended for

high power applications (up to 80W) operating at 240 kHz and a supply voltage of 30V with the

transmitting coil of 25 cm x 35 cm and the receiving coil is 12 cm x 12 cm. The findings should

apply across all power levels and coil sizes with different level of tolerances.

8.1 Varying Supply Voltage to Achieve Receiver Voltage Control

The most straightforward way to achieve power control and receiver voltage control is to

control the supply voltage of the transmitter. Experimental results by sweeping the supply

voltage from 18 V to 30 V in steps on 1 V as shown in Figure 8-2 and Figure 8-3 further verify

the feasibility of supply voltage control to achieve power control and receiver voltage control.

From Figure 8-2, high efficiency is achieved across the whole range of supply voltages with

efficiency peaking at different power level depending on the supply voltage. If the system is able

to control the supply voltage from very low voltages such 6 V up to 30 V in sufficiently small

steps, the system will be able to achieve better than 80% system efficiency regardless of loading

conditions. However, this is not practical because it will require the system to be driven from a

variable output voltage regulator instead of directly from a DC source. This will significantly









impact the system efficient especially when the supply voltage is very low e.g. 9 V. This is

because a buck regulator becomes less efficiency when the difference between the input and

output voltages increases. In addition, a high efficiency high power voltage regulator with a

variable output is complex and costly. It is typically achieved by using an off the shelf variable

voltage regulator and replacing one of the feedback resistors with a digital rheostat. The digital

rheostat is then controlled by the micro-controller which is used for load and fault mode

detection.

85 30V

80 ~- --29V
75 28V
-27V
70 -26V
> 65 --- 25V
60 --------------------------- -24V
55 -23V
6- 55 --------------------- ---
-22V
-21V
45 -- -- -- -C -- 20V
40 -19V
18V
0 10 20 30 40 50 60 70 80 90 100

Power Delivered to Load (W)

Figure 8-2. Efficiency-power plot for supply voltage from 18 V to 30 V in steps of 1 V.

Either the load tracking technique proposed in Chapter 7 or a communication link can be

used to track the loading condition for a one to one load system. The supply voltage can be

varied to optimize for the specific load condition. This technique can be applied to the other two

following methods of controlling receiver voltage. Each supply voltage will have a different

power delivery profile with respect to supply current. However, since the relationship between

supply current and power delivered can be approximated by a linear relationship, the profile is









condensed into two data points, the gradient and intercept. As discussed in Chapter 7, the no-load

requires four data points. The safe zone which was also discussed in Chapter 7 requires

additional 5 data points. Therefore, each voltage point will require 11 data points and if each data

point takes up 2 bytes of memory, each supply voltage point will require 22 bytes. A reasonable

system will require a supply voltage resolution from 20 to 50 points, requiring 440 bytes or 1100

bytes of memory space on the micro-controller.



S75 ---- -2V


0 65-27V


u 55

50 23V
45 22V
S3 -- ---- ------ -21V

35 _20V
-19V
0 10 20 30 40 50 60 70 80 90 100

Power Delivered to Load (W)

Figure 8-3. Receiver voltage-power plot for supply voltage from 18 V to 30 V in steps of 1 V.

As shown in Figure 8-3, the unregulated receiver voltage decreases with decreasing supply

voltage at the same power level. Therefore, the receiver voltage can be kept within a tight

bounded range for power delivery across a large power band. Depending on the acceptable

voltage ripples and with sufficient resolution achieved on the supply voltage control, it is

possible for the system to achieve wireless power transfer without a voltage regulator on the

receiver. This is only true if and only if the coupling coefficient remains the same for all possible









placements. For a practical system, such a requirement will not be realistic unless the receiving

coil is large enough to average the magnetic field variation or a locking mechanism is in place to

ensure a fix placement. If so it will defeat the purpose of providing some level of mobility for a

wireless power system unless it is used in some unique applications. Therefore, to ensure 100%

stability without the system drifting in the wrong direction and still attain some level of mobility,

a communication link is required to close the loop.

From Figure 8-3 it can be seen that if the system is to achieve a power handling of up to 40

W, the lowest possible voltage to achieve such a scheme without a voltage regulator on the

receiver will be 42.5V (supply voltage 20 V line). The lowest voltage for such a power level is

high because of high leakage inductances on both the transmitting and receiving coils. Such high

leakage inductances are due to poor coupling coefficient which results from the size ratio

between the transmitting and receiving coil (transmitting coil is 5 times larger than receiving

coil). In order to bring the voltage lower, the receiving coil needs to become larger as relative to

the transmitting coil (reduce leakage inductances) or have less turns (trading voltage for current

at the expense of resistive losses in the receiving coil). This brings another set of constraints in

the system design embedding the coil design into an integrated part of the system design. Such

design constraints are unique to specific applications which are beyond the scope of this

discussion and can be covered as potential future work for more tightly coupled one to one

wireless power transmission system.

Figure 8-4 shows the linear relationship between the supply current and power delivered to

the load for different supply voltage (18V to 30V in steps of IV). The y-intercept for all the

supply voltages remains approximately the same, but the gradient decreases with increasing

supply voltage. The trend can be explain by the fact that for the same power level at similar









efficiency, the supply current decreases with increasing supply voltage to maintain similar power

level and efficiency. The difference will become more significant with the increase of power

delivered to the load resulting in a fan out plot from the y-intercept. Therefore, each supply

voltage will require a new data set to describe the trend. It is possible to use a single y-intercept

for all supply voltages if the tolerance for power tracking is relaxed. The difference will not be

significant if the power handling of the system is sufficiently large (>50 W).

4.0 30V

3.5 29V

3.0_ __ ______ -28V
-27V
2.5
-26V
2.0 ---- ------ -25V

.2 1.5 24V
C-- 23V
S1.0 -- -- --
22V
0.5 --- ---- ------- -----
521V
0.0 20V


0 10 20 30 40 50 60 70 80 90 100 19V

Power Delivered to Load (W) 18

Figure 8-4. Linear relationship between supply current and power delivered to the load for
supply voltage from 18 V to 30 V in steps of 1 V.

In conclusion, varying the supply voltage is an effective way to control the power delivery

as well as the unregulated receiver voltage. With sufficient resolution, the system is able to

deliver a stable receiver voltage for which the voltage regulator can be removed if and only if the

coupling coefficient is sufficiently constant for all placement or both the transmitting and

receiving coil is fixed in position. Since it is meaningless to design a wireless power system

without any freedom in placement and it is also not possible to design a transmitting coil with a

100% even field distribution, the solution is not practical without a communication link to









provide a feedback loop. By adding a voltage regulator between the DC supply and the

transmitter to control the supply voltage, the system will incur additional power losses which

might result in thermal and efficiency issues. In addition, the complexity, size and cost added to

control the supply voltage can be considerable. Therefore, varying supply voltage to control

unregulated voltage and power delivery is an effective technique but not practical.

8.2 Varying Operating Frequency to Achieve Receiver Voltage Control

Varying the operating frequency can be used to indirectly tune to impedance looking into

the transmitter load network after Cshunt as shown in Figure 8-1. The impedance looking into the

transmitter load network (Ztx) can be simplified into a series network of a capacitor, an inductor

and a load resistance under any load condition. To ensure the effective Ztx is inductive, the

operating frequency is selected to be higher than the self resonance frequency of the LC network

not taking the load resistance into account. Therefore, Ztx will be less inductive with decreasing

operating frequency, decreasing the phase of Ztx, thus, increasing power delivery as well as

unregulated receiver coil voltage, vice versa. However, the above assumption is true if and only

ifZVS/ZDS operation still holds at the specification load impedance without changing Cshunt in

Figure 8-1. Although making Cshunt tunable will improve the frequency dynamic range of a

frequency tunable class E power amplifier, it will also increase its complexity. If the ZVS/ZDS

operation is not valid, there will be an increase in device stresses and degradation in efficiency.

Due to the large amount of data and ease of using the nominal operating frequency as a

reference point, the analysis is split into two different frequency ranges. The first frequency

range (Figure 8-5 and Figure 8-6) will be for operating frequency above the selected operating

frequency in steps of 2 kHz up to 250 kHz and the second frequency range (Figure 8-7 and

Figure 8-8) will be for operating frequency below the selected operating frequency in steps of 1

kHz down to 23 5 kHz.










90



70 --- ----------- --
-240kHz
6 0 -- - -
SI/ |--242kHz
=50 t----------------- -- --
Q If -244kHz
40 -1------------------
S40 --246kHz
30 --------- -- -- ------
248kHz
20 -- -- -- -- -- -- -- -- -- 250kHz

10 -- -- -

0 10 20 30 40 50 60 70 80 90

Power Delievered to Load (W)


Figure 8-5. Efficiency-Power plot for operating frequency from 240 kHz to 250 kHz in steps of
2 kHz.

80


e 75 -

7 80 1t- ------------- -240kHz

24k...... 2 Hz
65
X0-244kHz
60
S60-246kHz

55 -------------- 248kHz

0_ 250kHz

0 10 20 30 40 50 60 70 80 90

Power Delivered to Load (W)


Figure 8-6. Receiver voltage-power plot for operating frequency from 240 kHz to 250 kHz in
steps of 2 kHz.









As shown in Figure 8-5, the power delivered to the load decreases with increasing

operating frequency because Zx, becomes more inductive decreasing real power and increasing

reactive power as discussed in Chapter 1. The unregulated receiver voltage decreases marginally

with increasing operating frequency up to 244 kHz. However, as seen from Figure 8-5 the

system efficiency starts to degrade at frequencies above 244 kHz. This is due to ZVS/ZDS

operation does not hold anymore. The non-ideal waveform causes the switching losses across the

drain to increase. The switching losses are predominately caused by the transistor being turned

on before drain voltage drops to zero due to a shorter period with the increase in frequency. This

results in the charges from Cshunt are being dumped across the transistor during turn on at high

currents.

The unregulated receiver voltage also starts to become larger at lighter load especially

during open circuit conditions when the operating frequency is increased to above 244 kHz as

shown in Figure 8-6. Although Ztx is more inductive with the increase in operating frequency

limiting the power delivery, the coupling is also increased increasing the voltage induced on the

receiver. As shown in Equation 2-6 the increase of coupling with respect to operating frequency

is a square law effect. Therefore, there are two competing mechanisms affecting the power

delivery and unregulated receiver voltage working against each other. In addition to the non

ZVS/ZDS operation, Ztx, swings over a larger range with the same components on the transmitter

and receiver load resistance at higher operating frequencies. Therefore, the range and value of

unregulated receiver voltages increases with operating frequency as shown in Figure 8-6.

As shown in Figure 8-7, the power delivered to the load increases with decreasing

operating frequency because Zt, becomes more capacitive with decreasing frequency. The

unregulated receiver voltage increases marginally with decreasing operating frequency down to









235 kHz and below. This trend is expected with the increase in power delivery. It can be

observed from Figure 8-7 that the system efficiency starts to degrade much slower than

increasing the operating frequency because ZVS/ZDS operation still holds. Although ZVS/ZDS

operation does still hold, the zero crossing occurs earlier with decreasing frequency. With the

decrease in operating frequency the period of each cycle gets longer but the drain waveform

remains similar, causing the zero crossing of the drain waveform to occur earlier before the

transistor is turned on. Minimum amount of power is still lost via the built in diode of the

transistor causing a slight drop in system efficiency. The power loss via the built in diode can be

reduced by adding an external diode such as a schottky diode with a faster transition time and

smaller turn on voltage. The analysis is only valid if ZVS/ZDS operation still holds while the

operating frequency is decreased.


90









40 -------------------

30

20


235kHz

236kHz
-237kHz

238kHz

-239kHz
-240kHz


0 10 20 30 40 50 60 70 80 90 100

Power Delievered to Load (W)


Figure 8-7. Efficiency-power plot for operating frequency from 235 kHz to 240 kHz in steps of
1 kHz.















-235kHz

236kHz
- 237kHz

238kHz
-239kHz
-240kHz


0 10 20 30 40 50 60 70 80 90 100

Power Delievered to Load (W)


Figure 8-8. Receiver voltage-power plot for operating frequency from 235 kHz to 240 kHz in
steps of 1 kHz.


0 10 20 30 40 50 60 70 80 90 100

Power Delievered to Load (W)


-235k

236k

-237k

5238k

-239k

-240k

-242k

-244k

-246k

-248k

-250k


Figure 8-9. Linear relationship between supply current and power delivered to the load for
operating frequency from 235 kHz to 250 kHz in steps of 1 kHz.









Figure 8-9 shows the linear relationship between the supply current and power delivered to

the load for different operating frequencies (235 kHz to 250 kHz in steps of 1 kHz). The y-

intercept and gradient of all the operating frequencies are different. Therefore, each operating

frequency will require a different set of data set to describe the relationship. The fixed losses

increases with operating frequency because the drain waveform deviates from the ZVS/ZDS

conditions when the transistor is turned on earlier before the drain waveform settled down at

ground. Therefore, more power is dissipated as heat across the transistor drawing a higher

current regardless of loading conditions. This results in an increase of the y-intercept.

In conclusion, using frequency control as a form of power and unregulated receiver voltage

control is not the optimum way. The useable frequency range to achieve high efficiency is

limited unless the value of Cshunt can be tuned as well. This will increase the complexity of the

system as two different parameters are being tracked concurrently. Although, the power delivery

is reduced with increasing operating frequency, the maximum unregulated receiver voltage

during light load actually increases after the frequency crosses a certain threshold which is 244

kHz for this setup. Therefore, the scheme is only able to reduce power delivery and not reduce

high voltage device stresses for all loading conditions. In addition, if the operating frequency is

varied across a considerable range, FCC regulations will require the system to be tested for each

operating frequency.

8.3 Varying C,,t to Achieve Receiver Voltage Control

The last method to control the receiver voltage is to vary the value of Cout. By increasing

the C,,ut value, Zt, will be made more inductive, reducing real power delivery to the receiver. The

reduction in power delivery will reduce the unregulated receiver voltage for a specific load

condition. This technique ensures a more consistent performance with respect to loading

conditions as compared to that of varying operating frequency because the frequency is fixed.










The trend of unregulated receiver voltage as shown in Figure 8-11 shows a predictable trend as

relative to the one shown in Figure 8-6. In addition, by using a fixed operating frequency instead

of a variable operation frequency will make the FCC tests more straightforward.


- 18.2nF

- 17.2nF

-16 nF

15nF


0 10 20 30 40 50 60 70 80 90 100
Power Delievered to Load (W)

Figure 8-10. Efficiency-power plot for different Cout values.


Ou


t 70
7 65----------- --------
65
s 60- ------------
S55----------------
S60


S45-------------
40
S35-- -----------
30

0 10 20 30 40 50 60 70 80 90 100

Power Delievered to Load (W)

Figure 8-11. Receiver voltage-power plot for different Cout values.


158


- 18.2nF

- 17.2nF

-16 nF

15nF


S










Cot can be varied either by continuous tuning using a varactor or switching a bank of

capacitors. Using a varactor across high AC voltage and current is not practical. Therefore, the

more practical solution is to use discrete capacitors and relays to switch between the different

capacitance values. Traditionally, high power relays are noisy and bulky but power relay

technology has improved significantly that they are becoming more compact and does not

produce much audible noise. An example is the P2 V23079 relay by Tyco electronics. Its

dimension is 15 mm x 7 mm x 10 mm and is able to handle up to 5A current. The experimental

results shown the Figure 8-8 and Figure 8-9 is based on a 2 relays design with the base capacitor

value at 15 nF. The two extra capacitors' values are 1 nF and 2.2 nF. Therefore, it is able to

achieve four different C,,ut values at 15 nF, 16 nF, 17.2 nF and 18.2 nF. The number of steps are

two to the power of the number of relays, thus more relays can be added to increase the

resolution. However, it is shown later than a single relay setup might be sufficient.


4.0 -

3.5-------------- ---

3.0-

2.5 --------------
18.2nF
2.0 ------------------
lu 17.2nF
1.5 -
CL i 16 nF
1.0
0.515 nF
0.5 -------------- -

0.0 l
0 10 20 30 40 50 60 70 80 90

Power Delievered to Load (W)


Figure 8-12. Linear relationship between supply current and power delivered to the load for Cout
values of 15 nF, 16 nF, 17.2 nF and 18.2 nF.









Figure 8-12 shows the linear relationship between the supply current and power delivered

to the load for difference Cout values (15 nF, 16 nF, 17.2 nF and 18.2 nF). Since all of the lines

are overlapping one another, they can be approximated to a common y-intercept and gradient

value regardless of C,,ut value. The only different between them is the length of the line which

does not affect the load tracking algorithm. Therefore, by varying Cout to control unregulated

receiver voltage and power delivery, a single set of data is sufficient to accurately track the

power delivery of the wireless power system. This will simplify the system significantly.

As shown in Figure 8-10, the improvement in efficiency becomes less significant with the

increase in C,,ut value; little difference is observed for C,,ut values of 17.2 nF and 18.2 nF. The

unregulated receiver voltage value decreases significantly with increasing Cout value, for

example, the unregulated receiver voltage decreases an average of 15 V for power level up to 50

W when the C,,ut is increased from 15 nF to 16 nF while limiting the maximum power delivery

from 85 W to 50 W. For simplicity, it is still feasible to have only a single relay switching

between two capacitance values of 15 nF and 16 nF. The maximum unregulated receiver voltage

will be kept below 65 V with a single transition between the two states. In addition, a single set

of values is sufficient to describe the supply current and power delivery relationship. This makes

tracking and control simple and straightforward while keeping memory usage to the minimum.

By reducing the number of times the relay switches, it will also extend the lifetime of the relay.

8.4 Conclusion

By studying the three different techniques of varying power delivery and unregulated

receiver voltage, the simplest and most cost effective technique is to use a single relay to control

between two different C,,ut values. Although, varying the supply voltage is an effectively

technique, the complexity and added cost is too high to justify unless the system is able to

completely remove the regulator at the receiver. However, by doing so, a communication link is









required to ensure the receiver's voltage is stabilized and not enter into a positive feedback loop.

Finally, varying the operating frequency is not a good technique to control the unregulated

receiver voltage due to the limited ZVS/ZDS operating frequency range and sensitivity of the

system at increased frequency. Another method which was not discussed is the perform duty

cycling on the transmitter. Although, it is a simple and straightforward technique, due to the

large inductances of the transmitting coil, it will take considerable amount of time for the

transmitter to reach steady state. Therefore, the duty cycling speed of the transmitter is limited.

With a slow duty cycling speed, the receiver's DC charging holding capacitor at the input of the

regulator will need to be significantly larger in capacitance value to provide a smooth DC input

voltage to the regulator, making it not practical.









CHAPTER 9
SUMMARY AND FUTURE WORK

9.1 Summary

A new technique of designing a wireless power transfer system using the class E mode of

operation for power transmission via inductive coupling is proposed. The history of wireless

power transmission, background information is presented in Chapter 1. The system level design

of the wireless power transfer system is discussed in Chapter 2. The design rules for two

different impedance transformation network topologies, the series-parallel topology and parallel-

parallel topology have been presented in Chapter 3. Instead of using complex detection schemes

and variable tank circuits to seek resonance and high efficiency, the system is designed to

achieve the desired power delivery profile via its natural response across a wide range of load

resistances. A multi-channel topology is also proposed to achieve high efficiency for different

power ranges. In addition, a Matlab program is developed to automatically generate the value for

each component. Extending the work of a one-to-one system to support multiple receivers, a

switch architecture is proposed in Chapter 4 to decouplee" the fully charged receiver from the

system so that power delivery to the other receiver can be improved. The switch is able to pass

high power signals as well as sustain high input voltages. Simulation results of the switch using

Advanced System Design by Agilent verified the performance of the switch.

A dual channel system using the parallel-parallel impedance transformation network

topology is fabricated and tested to verify the design. The experimental results are presented in

Chapter 5. The fabricated system is capable of delivering nearly 300 W with forced air cooling

and the power delivery can be varied via its supply voltage. Higher power delivery can be

achieved if a power supply with higher output and transistors with higher breakdown voltages

are used. With natural convection cooling, the system achieves a maximum power delivery of 69









W with end-to-end system efficiency of 74%. Another lower power system supporting multiple

receivers using a single transmitting coil is also presented in Chapter 5. The system is designed

for portable consumer products which are typically charged from a USB port at 5V, 500mA.

Experimental results show the power choking phenomena or fish plot when two or more loads

are placed on a single transmitting coil. However, when the fully charged device is decoupled

from the system using the switch, power delivery to the other receiver is resume to normal.

It is found from the interoperability analysis in Chapter 6 that the system is sufficiently

robust to power receivers designed for other platforms as long as the size ratio between the

receiving coil and transmitting coil is kept within the range of 1:2 to 1:12. This enables a

transmitting unit designed to power a specific model of cell phone to work for other cell phones

or even mp3 players. Therefore, the consumer is not required to purchase a different transmitting

unit or each device he owns, making it a universal system, reducing waste and power

consumption.

A method of load/fault detection is presented in Chapter 7 to protect the transmitter from

fault modes ranging from minor to fatal which will potentially damage the transmitter. An

example of a fatal fault will be having a huge piece of metal sheet being placed on top of the

transmitting coil affecting the self inductance of the transmitting coil. In addition, the load

detection system is able to detect if a valid receiver is placed on the transmitting coil, powering

down when the transmitter is not in use. This will reduce the no load power consumption

significantly making it more energy efficient. A method of power delivery tracking scheme is

also proposed to track the power delivery without a communication link. This enables the

transmitter to track the charge profile of the receiver so that it can switch between different

power handling mode or detect a damaged device or battery.









Finally, three different techniques to control power delivery as well as the unregulated

receiver voltage are studied in Chapter 8. Although, varying the supply voltage is an effective

way to perform power control, the implementation is too complex. Since a voltage regulator is

able to tolerate a large range of input voltages, varying C,,ut in a single discrete step is more than

sufficient to provide control to the power delivery and unregulated receiver voltage so as not to

damage the receiver regulator. More steps can be added to bring down the voltage if needed.

In conclusion, depending on the requirements, the system can be reconfigured to transmit

power wirelessly to different devices for a wide variety of applications. This technology can be

applied to rugged electronics to enable the creation of hermetically sealed units and to eliminate

the problem of charging port contamination and corrosion for operation in potentially harsh

conditions. In environments where sparking and arching hazards exist, this technology can be

applied to eliminate an electronic device's external metallic contacts. This technology can also

be implemented in our everyday portable devices from cell phones to laptops for the convenience

of the everyday consumer.

9.2 Future Work

Further extension of the load/fault detection scheme can be explored using more robust

pattern recognition techniques to determine the number of receivers a transmitter is powering as

well as their load profile. Although, the work presented is limited to close proximity of not more

than 5 cm range and power handling of up to 300 W, the design rules and concepts presented in

this dissertation laid the groundwork on how a wireless power system using magnetic induction

can be designed. Using the same principles the system can be expanded to support power

delivery for longer range and higher power levels e.g. 1 foot distance and >1 kW of power

handling. This will enable many new technologies. One of which is contactless electric vehicle

charging for which vehicles can be charged both in the garage at home or at work with a coil

164









embedded into the ground. The system will not require the owner to make an effort to connect a

charging cable. The same technology can be applied to powering robotics at factories or

warehouses so that they do not need to carry heavy batteries or to power small in-body or on-

body sensors to monitor a patient vital signs. Therefore, there are various applied research topics

that can be proposed base on this work. In addition, using the same pair of coils, near field high

speed communication systems with data rates beyond 1 Gbps can be designed using wideband

carrierless communication system.









APPENDIX A
FCC REGULATIONS

FCC regulations require any wireless power system via magnetic induction to pass 3

separate tests. They are Part 18.307 power line conduction test, Part 18.305 emission test and

Part 15.109 part 15B test. Each individual test is intended for different purposes and will be

discuss in the following sections.

Part 18.307 Conduction Test

The conduction test is to ensure that limited amount of power supply noise is being fed

back into the power grid. The limits of the conduction test for various frequency bands are

shown in Table A-i [32]. The noise can be in the form of differential mode noise and common

mode noise. However, since a typical DC power will have charging holding and decoupling

capacitors along the power path, the predominant noise being fed back into the power system is

common mode noise. Measurements of the conduction test are carried at the AC port of the DC

power supply. The DC power supply is connected to a line impedance stabilization network

(LISN) before connecting into the power grid. The LISN is fundamentally an electronic noise

filter network so that the noise from the device under test (DUT) is not fed back into the power

grid. The filtered noise is output from the LISN typically via a 50 Q port which can be connected

to a spectrum analyzer or an oscilloscope to analyze the noise signal.

Table A-1. FCC Part 18.307 conduction test limits
Conducted limit (dBIV)
Frequency of emission (MHz) Quasi-peak Average
Quasi-peak Average
0.15 -0.5 66 to 56* 56 to 46*
0.5 -5 56 46
5-30 60 50
*Decreases with the logarithm of the frequency









Part 18.305 Emission Test

FCC part 18 section 305 states that any device operating at non-ISM frequency below 500

kHz is allowed to have a maximum field strength of 15 V/m at a distance of 300 m. The

emission limits are more relaxed if an ISM frequency is used (25[iV/m). However, the worst case

of operating in a non-ISM frequency is assumed.

It is not practical to conduct RF tests at 300 m range. Therefore, most measurements are

conducted using a 10 m range or a 3 m range. By rule of thumb each decade reduction in

distance will result in a 40 dB increase in field strength. Therefore the maximum field strength is

converted to 82.6 dB V/m at a 10 m range and 103.5 dB V/m at a 3 m range. The measurement

is typically performed using an antenna with a known correction factor so that the output

spectrum into a 50 Q port can be converted from dByV to its equivalent field strength of

dByV/m. For a non-radiating system the field strength limit specified by FCC is extremely high

and any wireless power system should be able to meet it unless the power level delivered is in

excess of 1 kW.

Part 15.109 Part 15B Test

Table A-2. FCC Part 15.109 emission limits at 3 m range
Frequency of Emission (MHz) Field Strength (iV/m) Field Strength (dByV/m)
38-88 100 40.0
88-216 150 43.5
216-960 200 46.0
Above 960 500 54.0

Since the wireless power transfer system neither sends data nor have a communication

link, it is considered to be an unintentional radiator falling under Part 15B. The emission limits

of Part 15B of various frequency bands at 3 m range are shown in Table A-2 [33]. The intention

of Part 15B is to ensure that unintentional radiation is low enough not to interfere to surrounding

electronics equipment. Most unintentional radiations are due to fast switching edges of clock









signals and high frequency noise riding on the supply bus. Radiation due to high speed clock

edges can be easily suppressed by using a nulling resistor between the output of the gate driver

and the gate of the switching transistor to slow down the clock signal. Supply noise can be easily

suppressed by adding more decoupling capacitors with the appropriate self resonance frequency

(SFR) and equivalent series resistance (ESR) specification.









LIST OF REFERENCES


1. C. E. Greene, D. W. Harrist, J. G. Shearer, M. Migliuolo, G. W. Puschnigg,
"Implementation of an RF power transmission and network," U.S. Patent Application,
US2007/0191075A1, Jan 23rd, 2007.

2. W. C. Brown, "The History of Power Transmission by Radio Waves," IEEE
Transactions on Microwave Theory and Techniques, vol. 32, pp. 1230 1242, Sep 1984.

3. W. C. Brown, E. E. Eves, "Beamed microwave power transmission and its application to
space," IEEE Transactions on Microwave Theory and Techniques, vol. 40, pp. 1239 -
1250, Jun 1992.

4. IEEE Std C95.1, 2005 Edition, IEEE Standard for Safety Levels ii iih Respect to Human
Exposure to Radio Frequency Electromagnetic Fields, 3 kHz to 300 GHz.

5. J. F. Showrow, G. H. MacMaster and W. C. Brown, "The super power CW amplitron,"
Microwave Journal, pp. 52, Oct 1964.

6. J. W. Coltman 'The Transformer," IEEE Industrial Applications Magazine, Jan Feb
2002, pp. 8 -15.

7. Inductive cooker [Internet]. Wikipedia (US); [updated 2009 May 25; cited 2009 May 27].
Available from http://en.wikipedia.org/wiki/Inductioncooker

8. Electric Toothbrush [Internet]. Wikipedia (US); [updated 2009 May 16; cited 2009 May
27]. Available from http://en.wikipedia.org/wiki/Electrictoothbrush

9. Cut Loose. Wireless power for portable devices [Internet]. SplashPower Limited (UK);
[updated 2009 May 27; cited 2009 May 27]. Available from
http://www. splashpower.com/

10. Welcome to the Evolution of Wireless Power [Internet]. eCoupled/Fulton Innovation
(US); [updated 2009 May 21; cited 2009 May 27]. Available from
http://www.ecoupled.com/

11. Leading and Partnering in Wireless Power [Internet]. ConvenientPower (HK); [updated
2009 May 27; cited 2009 May 27]. Available from http://www.convenientpower.com/

12. Wireless Charging [Internet]. PowerMat (Israel); [updated 2009 Jan 7; cited 2009 May
27]. Available from http://www.pwrmat.com/

13. Wireless Electricity Delivered Over Distance [Internet]. WiTricity (US); [updated 2009
Jan 7; cited 2009 May 27]. Available from http://www.witricity.com/

14. Wireless Power [Internet]. WiPower (US); [updated 2009 Apr 27; cited 2009 May 27].
Available from http://www.wipower. com/









15. Wireless Power for a Wireless World [Internet]. PowerCast (US); [updated 2009 May 21;
cited 2009 May 27]. Available from http://powercastco.com/

16. M. K. Kazimierczuk "Class D voltage-switching MOSFET power amplifier", IEE
Proceedings-B, Vol 138, No. 6, Nov 1991, pp. 285 296.

17. N. 0. Sokal and A. D. Sokal "Class E A New Class of High Efficiency Tuned Single-
Ended Switching Power Amplifier" IEEE Journal of Solid State Circuits, Vol. SC-10,
No. 3, pp. 168 176, June 1975.

18. F. H. Raab, "Effects of circuit variations on the class E tuned power amplifier," IEEE
Journal of Solid-State Circuits, vol. 13, pp. 239 247, Apr 1978.

19. F. H. Raab, "Idealized operation of the class E tuned power amplifier," IEEE
Transactions on Circuits and Systems, vol. 24, pp. 725 735, Dec 1977.

20. Nathan 0. Sokal, "Class-E RF Power Amplifiers", QEX, pp. 9 20, Jan 2001.

21. X, Liu, S. Y. R. Hui, "An Analysis of a Double-layer Electromagnetic Shield for a
Universal Contactless Battery Charging Platform," in Proc. IEEE 36th Power Electronics
Specialists Conference, 16th June 2005, pp. 1767 1772.

22. C. Wang, G. A. Covic, 0. H. Stielau, "Power transfer capability and bifurcation
phenomena of loosely coupled inductive power transfer system," IEEE Transactions on
Industrial Electronics, vol. 51, pp. 148 157, Feb. 2004.

23. C. Wang, G. A. Covic, 0. H. Stielau, "Investigating an LCL load resonant inverter for
inductive power transfer applications," IEEE Transactions on Power Electronics, vol. 19,
pp. 995 1002, July 2004.

24. C. Wang, 0. Stielau, G. A. Covic "Design consideration for a contactless electric vehicle
battery charger, IEEE Transactions on Industrial Electronics, vol. 52, pp. 1308 1314,
Oct. 2005.

25. J. J. Casanova, Z. N. Low, J. Lin, R. Tseng, "Transmitting Coil Achieving Uniform
Magnetic Field Distribution for Planar Wireless Power Transfer System," in Proc. IEEE
Radio and Wireless Symposium, 18th 22nd January, 2009.

26. X. Liu, S. Y. R Hui, "Optimal design of a hybrid winding structure for planar contactless
battery charging platform," IEEE Transactions on Power Electronics, vol. 23, pp. 455 -
463,Jan. 2008.

27. R. Laouamer, M. Brunello, J. P. Ferrieux, 0. Normand, N. Bucheit, A multi-resonant
converter for non-contact charging with electromagnetic coupling," in Proc. 23rd
International Conference on Electronics, Control and Instrumentation, Nov 1997, vol. 2,
pp. 792 797.









28. G. Gwon, D. Park, S. Choi, S. Han, "Wireless charger decreased in variation of charging
efficiency," International patent application, PCT/KR2006/001706, 4th May 2006.

29. Z. N. Low, R. A. Chinga, R. Tseng, J. Lin, "Design and Analysis of a Loosely Coupled
Planar Wireless Power Transfer System using Magnetic Induction," IEEE Transactions
on Industrial Electronics, vol. 56, pp. 1801 1812, May 2009.

30. Minsik Ahn, Chang-Ho Lee, Laskar, J., "CMOS High Power SPDT Switch using
Multigate Structure", in Proc. IEEE International Symposium on Circuits and Systems,
25th 27th May 2007, pp. 3283 3286.

31. M. Yu, R. J. Ward, G. M. Hegazi, "High power RF switch MMICs development in GaN-
on-Si HFET technology", in Proc IEEE Radio and Wireless Symposium, 22nd 24th Jan
2008, pp. 855 585.

32. United States Code of Federal Regulations (CFR) Chapter 47, Part 18, Section 307, 1st
October 2001.

33. United States Code of Federal Regulations (CFR) Chapter 47, Part 15, Section 109, 10th
July 2008.









BIOGRAPHICAL SKETCH

Zhen Ning Low was bored in Singapore. He attended the Nanyang Technological

University in Singapore earning the bachelor's degree in electrical and electronic engineering in

2005 under the accelerated bachelors programme. He worked as an intern at the Institute for

Infocomm Research for half a year developing Zigbee wireless sensor networks. During his

undergraduate, he joined Positioning and Wireless Technology Centre as an undergraduate

research working on GPS and UWB position location systems and authored two conference

papers and two journal papers. After obtaining his bachelor's degree in February 2005, he joined

the Institute for Infocomm Research as a Research Engineer to continue his work on Zigbee

sensor networks and UWB position location system. In August 2006, he came to the University

of Florida to pursue his Ph.D. in electrical engineering. For his Ph.D. research he worked under

the guidance of Dr Jenshan Lin in the Radio Frequency Circuits and Systems group working on

wireless power transfer systems. He has 18 publications in technical journals and conferences

and 6 patent applications.





PAGE 1

1 HIGH EFFICIENCY LOOSELY COUPLED WIRELESS POWER TRANSFER SYSTEM VIA MAGNETIC INDUCTION By ZHEN NING LOW A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2009

PAGE 2

2 2009 Zhen Ning Low

PAGE 3

3 To my parents and wife

PAGE 4

4 ACKNOWLEDGMENTS I would like to thank my advisor Dr Jenshan Lin for th is wonderful opportunity to work under his guidance and mentoring. I have truly enjoyed working for him over the years acquiring technical knowledge as well as other soft skills. I would also like to thank Dr William Eisenstadt, Dr David Arnold and Dr Sumi Helal for their time and being on my committee. I woul d like to thank my project team mates both past and present for which without them this work will not be possible. They are : Joaquin Casanova for electromagnetic analysis and coil design; Jason Ta ylor for PCB design and fabrication; Raul Chinga for all the late night prototyping work ; and Ashley Trowell for ferrites and shielding work I would like to thank WiPower and Florida High Tech Corridor for funding this project. I would like to thank Linea r Technology and Terry Decker for their support and evaluation boards. I would also like to thank Shannon Chillingworth and all of the personnel in the EE office. Finally, I would like to thank my wife for her encouragement and unconditional support when t hings get rough. Without her this w ould not be possible.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................................... 4 LIST OF TABLES ................................................................................................................................ 8 LIST OF FIGURES .............................................................................................................................. 9 ABSTRACT ........................................................................................................................................ 15 CHAPTER 1 INTRODUCTION ....................................................................................................................... 16 1.1 History of Wireless Power Transmission ........................................................................ 18 1.2 Modern Wireless Power Transmission ............................................................................ 20 1.3 Background Information ................................................................................................... 23 1.3.1 Fundamental Theory of Operation via Phase Response ..................................... 23 1.3.2 Class D Power Amplifier ..................................................................................... 26 1.3.3 Class E Power Amplifier ...................................................................................... 3 0 2 WIRELESS POWER TRANSFER SYSTEM .......................................................................... 38 2.1 Single -channel Class E Power Amplifier ........................................................................ 39 2.2 Multi -channels/Stackable Class E Power Amplifier ....................................................... 41 2.3 Inductive Coupling ............................................................................................................ 42 2.4 Impeda nce Transformation Network ............................................................................... 44 2.5 Receiver ............................................................................................................................. 47 3 DESIGN OF IMPEDANCE TRANSFORMATION NETWORK .......................................... 50 3.1 Series -Parallel Impedance Transformation Network ...................................................... 50 3.1.1 Introduction ........................................................................................................... 50 3.1.2 Determination of Crx value ................................................................................... 53 3.1.3 Determination of Cout value .................................................................................. 58 3.1.4 Determination of Cshunt value ............................................................................... 59 3.2 Parallel Parallel Impedance Transformation Network ................................................... 60 3.2.1 Introduction ........................................................................................................... 60 3.2.2 Determination of Crx value ................................................................................... 61 3.2.3 Determination of Ctx value ................................................................................... 63 3.2.4 Determination of Cshunt value ............................................................................... 67

PAGE 6

6 4 WIRELESS POWER TRANSFER SYSTEM SUPPORTING MULTIPLE RECEIVERS ............................................................................................................................... 69 4.1 Inductive Coupling ............................................................................................................ 69 4.2 Switch Design .................................................................................................................... 72 4.3 Switch Simulation ............................................................................................................. 76 4.4 System Response with Receiver Switch .......................................................................... 79 5 EXPERIMENTAL VERIFICATION ........................................................................................ 83 5.1 High Power 300W Dual Channel System using Parallel Parallel Impedance Transformation Network Topology ................................................................................ 83 5.2 Low Power Multiple Receivers System with decoupling Switch using Series Parallel Impedance Transformation Network Topology ............................................... 95 6 INTEROPERABILITY BETWEEN DIFFERENT PLATFORMS (COIL SIZES) ............. 104 6.1 Test Bench Setup ............................................................................................................. 105 6.2 Experimental Verification .............................................................................................. 107 7 LOAD/FAULT DETECTION AND POWER DELIVERY TRACKING ........................... 111 7.1 Load/Fault Detection Scheme ........................................................................................ 112 7.1.1 Detection Circuit ................................................................................................. 112 7.1.2 Detection Flowchart/Logic ................................................................................. 115 7.2 Experimental Verification .............................................................................................. 117 7.2.1 Test Bench and Circuit ....................................................................................... 117 7.2.2 Experimental Results .......................................................................................... 122 7.3 Extension of Load/Fault Detection Scheme .................................................................. 130 7.3.1 M:N Coupling Structure ..................................................................................... 130 7.3.2 Removing Lout from Transmitter ....................................................................... 139 8 RECEIVER VOLTAGE CONTROL ...................................................................................... 146 8.1 Varying Supply Voltage to Achieve Receiver Voltage Control .................................. 147 8.2 Varying Operating Frequency to Achieve Receiver Voltage Control ......................... 152 8.3 Varying Cout to Achieve Receiver Voltage Control ...................................................... 157 8.4 Conclusion ....................................................................................................................... 160 9 SUMMARY AND FU TURE WORK ..................................................................................... 162 9.1 Summary .......................................................................................................................... 162 9.2 Future Work ..................................................................................................................... 164

PAGE 7

7 APPENDIX : FCC REGULATIONS ............................................................................................ 166 LIST OF REFERENCES ................................................................................................................. 169 BIOGRAPHICAL SKETCH ........................................................................................................... 172

PAGE 8

8 LIST OF TABLES Table page 5 1 Component Values for High power 300W Dual Channel System using Parallel Parallel Impedance Transformation Network Topology ..................................................... 84 5 2 Component Values for Low Power Multiple Receivers System with decoupling Switch using Series -Parallel Impedance Transformation Network Topology ................... 96 6 1 Specification of the three different platforms ..................................................................... 105 6 2 Coupling parameters of nine possible combinations with first three as intended pairs ... 106 7 1 Component Values for Load/Fault Detection Test Bench ................................................. 118 A 1 FCC Part 18.307 conduction test limits .............................................................................. 166 A 2 FCC Part 15.109 emission limits at 3 m range ................................................................... 167

PAGE 9

9 LIST OF FIGURES Figure page 1 1 Block diagram of the multiple receivers wireless power transfer system using inductive coupling. ................................................................................................................. 17 1 2 Complex plane of inductive and capacitive load. ................................................................ 24 1 3 Phasor diagram. ...................................................................................................................... 25 1 4 Class D power a mplifier. ....................................................................................................... 26 1 5 Class E p ower a mplifier. ........................................................................................................ 31 2 1 Wireless power transfer system diagram. ............................................................................. 38 2 2 Schematic of the dual -channel C lass E power amplifier. .................................................... 41 2 3 Topologies for a single -element impedance transformation network. ................................ 45 2 4 Input voltage of a half wave rectifier and a full wave rectifier. .......................................... 48 3 1 Windings of a 20cm x 20cm transmitting coil used for experimental verification. ........... 51 3 2 Normalized power deliver with respect to location of transmitting coil in Figure 3 1 using a receiving coil of 9 cm x 6 cm. .................................................................................. 52 3 3 Simplified schematic of wireless power transfer system using series -parallel impedance transformation network and Class E transmitter. Ztx Impedance looking into the transmitter load network. Ztxcoil Impedance looking into the transmitting coil. Zrx Impedance looking into receiver network. RLoad is the equivalent resistance looking into the rectifier. ....................................................................................................... 53 3 4 Resistance and reactance of Zrx versus load resistance at different with different Crx. (50 nF, 100 nF and 150 nF) ................................................................................................... 54 3 5 Peak resistance response looking into the transmitting coil with respect to Crx. ............... 56 3 6 Resistance and reactance looking into the transmitting coil. .............................................. 56 3 7 Coupling efficiency with respect to load resistance. ............................................................ 57 3 8 Ztx phase response with respect to load resistance for various Cout capacitance values. ... 58 3 9 Transistor drain voltage waveform for different load resistance. ( Cshunt = 10 nF). ............ 59

PAGE 10

10 3 10 Simplified schematic of wireless power transfer system using parallel -parallel transformation network and Class E transmitter. Ztx Impedance looking into the transmitter load network. Ztxcoil Impedance looking into the transmitting coil. Zrx Impedance looking into receiver network. RLoad is the equivalent resistance looking into the rectifier. ..................................................................................................................... 61 3 11 Optimum receiver capacitor value versus load resistance. .................................................. 61 3 12 Coupling efficiency and transfo rmed impedance looking into the transmitting coil. ........ 63 3 13 Reactance of Ztxcoil versus load resistance with different Ctx. ............................................. 65 3 14 Amplitude and phase of impedance of unloaded Ztx versus Ctx. ......................................... 66 3 15 Rtx and Xtx versus lo ad resistance. (Ctx: 105 nF) .................................................................. 66 3 16 Phase of Ztx versus load resistance. (Ctx: 105 nF) ................................................................ 67 3 17 Transistor drain voltage waveform for different load resistance. ( Cshunt = 19 nF). ............ 67 4 1 Power d elivery to l oads for a single receiver setup and a dual receivers setup with one of the load fixed at 1000 ............................................................................................ 71 4 2 Block diagram of the proposed switch. ................................................................................. 72 4 3 Schematic of the proposed switch circuit. ............................................................................ 74 4 4 Schematic of the improved proposed switch circuit. ........................................................... 75 4 5 Schematic of the switch in Advanced System Design with a resistive as load. ................. 76 4 6 Switch control waveform (0 V for off and 3 V for on). A minimum of 1 V is required to turn on the transistor. ......................................................................................................... 77 4 7 Generated switch control waveform s ................................................................................... 78 4 8 Output waveform of the switch before the rectifier. ............................................................ 79 4 9 Proposed half -wave rectifier receiver architecture. ............................................................. 79 4 10 ADS schematic of test bench for receiver ar chitecture in Figure 5 10. .............................. 81 4 11 Simulation results of test bench (Figure 4 10). Red: Low voltage control signal.Black: Receiver rectified voltage. .............................................................................. 82 5 1 Photograph of the dual channel Class E power amplifier. .................................................. 85 5 2 Photograph of the transmitting coil 10 turns (embedded into the table top) and receiving coil 5 turns (placed on top). ............................................................................... 85

PAGE 11

11 5 3 Power delivery (left y axis) and efficiency (right y axis) of the system versus load resistance. Supply voltage: 120 V. ........................................................................................ 86 5 4 Transistor and inductor temperature with natural convection cooling and forced air cooling versus supply voltage. .............................................................................................. 87 5 5 Photograph of the dual channel Class E power amplifier with forced air cooling. ........... 88 5 6 Voltage and current waveforms of the Class E transmitter. ................................................ 89 5 7 Power delivered to load versus load resistance. A maximum power of 69 W occurs approximately at 50 -channel and maximum power of 10 W occurs at ap proximately at 75 -channel. Supply voltage: 60 V. ...................................... 89 5 8 Mode -switching operation for optimized efficiency across a wid e power delivery range. (1) Dual -channel mode for higher power, (2) Dual -channel mode switch -over to single -channel mode when better efficiency can be obtained at a lower power level, (3) Single -channel mode for lower power, (4) Single -channel mode switch over to dual -channel mode when higher power delivery is needed. Supply voltage: 60 V. ........................................................................................................................................ 90 5 9 System efficiency versus load resistan ce with a maximum efficiency of 64.5% for a single -channel system, and 76% for a dual channel system at approximately 70 Supply voltage: 60 V. ............................................................................................................ 91 5 10 Transmitter efficiency versus load resistance. Maximum transmitter efficiency occurs across the range of 60 -channel and 79% for single -channel. Supply voltage: 60 V. .................................................................... 92 5 11 System efficiency versus load resistance for single -channel and dual -channel modes achieving high efficiency at high power output. It also illustrates that a single channel mode is more efficient at low power delivery states. Supply voltage: 60 V. ....... 93 5 12 Transmitt ing coil RMS voltage versus load resistance. Supply voltage: 60 V. ................. 93 5 13 Receiver DC voltage versus load resistance, converging to approximately 70 V in dual -channel mode and 38 V in single -channel mode. Supply voltage: 60 V. .................. 94 5 1 4 Photograph of a test setup with two receivers with decoupling switch on the packaged transmitting coil. .................................................................................................... 96 5 15 Power delivery to the receiver with switch and system efficiency versus load resistance for a one-to -one setup (simulated and measured) ............................................... 97 5 16 Efficiency of power delivery to the receiver with switch versus power delivered for a 1 to 1 setup (simulated and measured) .................................................................................. 98 5 17 Comparison between receiver with switch and receiver without switch. ........................... 99

PAGE 12

12 5 18 Comparison between single receiver standalone and dual receiver setup with one receiver switched off. ............................................................................................................. 99 5 19 Power delivery to receiver 1 versus its load resistance at different fixed receiver 2 load resistance. ..................................................................................................................... 101 5 20 Power delivery to receiver 2 versus receiver 1 load resistance at different fixed receiver 2 load resistance. (same legends as in Figure 5 20) ............................................ 1 01 5 21 System efficiency versus total power delivery to both receivers with the load resistance of receiver 2 fixed and the load resistance of receiver 1 swept ac ross the stated range for a dual receiver test bench. (same legends as in Figure 5 20) ................. 102 5 22 Measured power delivery to a dual r eceiver system with load resistance varied from 10 ...................................................................................................................... 103 6 1 System efficiency and receiver voltage vs. power delivery plot for different sized receiving coil on a small sized transmitting coil. ............................................................... 107 6 2 System efficiency and receiver voltage vs. power del ivery plot for different sized receiving coil on a medium sized transmitting coil. .......................................................... 109 6 3 System efficiency and receiver voltage v s. power delivery plot for different sized receiving coil on a big sized transmitting coil. ................................................................... 110 7 1 Block diagram of the proposed wi reless power transfer system with detection circuit (detecting supply current and coil voltage) ........................................................................ 112 7 2 Schematic of coil current extraction network using a current sense resistor. .................. 114 7 3 Detection scheme flow chart for proposed system. ........................................................... 116 7 4 Photograph of the fa bricated transmitter circuit with control circuit. ............................... 118 7 5 Schematic of power stage of fabricated transmitter. .......................................................... 119 7 6 Schematic of driver stage of fabricated transmitter. .......................................................... 119 7 7 Schematic of detection and control stage of fabricated transmitter. ................................. 120 7 8 Efficiency Power plot of single receiver setup (Solid black line: Small coil. Dashed gray line: Big coil) ............................................................................................................... 122 7 9 Examples of different loading conditions and fault modes on coil voltage versus supply current space. ............................................................................................................ 123 7 10 Efficiency -power plot of three sets of dual load measurements and two sets of single load measurements using a combination of big and small receiving coils. ...................... 125

PAGE 13

13 7 11 Transmitter coil voltage and supply current space illustrating three different zones: no -load, single load, and dual loads. ................................................................................... 126 7 12 Transmitter coil voltage and supply current space diagram illustrating three different zones, noload, safe and fault. ............................................................................................. 127 7 13 Linear relationship between supply current and power delivered to the loads for all 5 sets of measurements. Solid dashed line: y = 0.095x + 0.055. .......................................... 129 7 14 Power delivery error distribution plot. Calculated power delivery is based on measured supply current. Solid dashed line: ideal error free calculation. ........................ 129 7 15 Examples of M:N coupling structure s. ............................................................................... 131 7 16 Measured power delivery of a dual receiver setup on a dual transmitting coil test bench with both receivers on a single coil leaving the other coil unloade d. .................... 132 7 17 Measured power delivery of a dual receiver setup on a dual transmitting coil test bench with one receiver on each trans mitting coil. ............................................................ 133 7 18 Efficiency -power plot of two sets of dual load measurements and two sets of single load measurements. (Green lines: Single receiver, Blue lines: Dual receivers on single transmitting coil, Red lines: Dual receivers on dual transmitting coil) ................. 134 7 19 Transmitter coil voltage and supply current space. (Green diamond: No load, Black lines: Single receiver, Blue lines: Dual receivers on single transmitting coil, Read lines: Dual receiver on dual transmitting coil) ................................................................... 134 7 20 Linear relationship between supply current and power delivered to the loads for all 4 sets of measurements. Solid dashed line: y = 0.076x + 0.048. .......................................... 135 7 21 Power delivery error distribution plot. Calculated power delivery is based on measured supply current. Solid dashed line: ideal error free calculation. ........................ 136 7 22 Transmitter coil voltage and supply current space with fault tests. (Green lines: Fault tests without valid receiver on one of the coil, Purple lines: Fault tests with a valid receiver on one of the transmitting coils) ........................................................................... 137 7 23 Different methods of driving a 2:N structure .................................................................... 139 7 24 Block diagram of wireless system without Lout. ................................................................. 140 7 25 Transmitter coil voltage and supply current space with fault tests with Lout removed. (Green lines: Fault tests without valid receiver on one of the coil, Purple lines: Fault tests with a valid receiver on one of the transmitting coils) .............................................. 141

PAGE 14

14 7 26 Waveform of the transmitter circuit without Lout when fault conditions overlaps with valid load conditions. (Red: Drain voltage, Orange: Coil voltage, Green: Coil Current) ................................................................................................................................. 142 7 27 Spectrum of transmitter coil voltage without Lout when fault conditions overlaps with valid load conditions. ........................................................................................................... 142 7 28 Waveform of the transmitter circuit with Lout when fault conditions overlaps with valid load conditions. (Red: Drain voltage, Orange: Coil voltage, Green: Coil Current) ................................................................................................................................. 143 7 29 Spectrum of transmitter coil voltage with Lout when fault conditions overlaps with valid load conditions. ........................................................................................................... 144 8 1 Block diagram of wireless power system. .......................................................................... 146 8 2 Efficiency -p ower plot for supply voltage from 18 V to 30 V in steps of 1 V. ................. 148 8 3 Receiver Voltage-Power plot for supply voltage from 18 V to 30 V in steps of 1 V. ..... 149 8 4 Linear relationship between supply current and power delivered to the load for supply voltage from 18 V to 30 V in steps of 1 V. ............................................................ 151 8 5 Efficiency -p ower plot for operating frequency from 240 kHz to 250 kHz in steps of 2 kHz. ....................................................................................................................................... 153 8 6 Receiver v oltage -p ower plot for operating frequency from 240 kHz to 250 kHz in steps of 2 kHz. ...................................................................................................................... 153 8 7 Efficiency -p ower plot for operating frequency from 235 kHz to 240 kHz in steps of 1 kHz. ....................................................................................................................................... 155 8 8 Receiver v oltage -p ower plot for operating frequency from 235 kHz to 240 kHz in steps of 1 kHz. ...................................................................................................................... 156 8 9 Linear relationship between supply current and power delivered to the load for operating frequency from 235 kHz to 250 kHz in steps of 1 kHz. ................................... 156 8 10 Efficiency -p ower plot for different Cout values. ................................................................. 158 8 11 Receiver v oltage -p ower plot for different Cout values. ...................................................... 158 8 12 Linear relationship between supply current and power delivered to the load for Cout values of 15 nF, 16 nF, 17.2 nF and 18.2 nF. ..................................................................... 159

PAGE 15

15 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy HIGH EFFICIENCY LOOSELY COUPLED WIRELESS POWER TRANSFER SYSTEM VIA MAGNETIC INDUCTION By Zhen Ning Low August 2009 Chair: Jenshan Lin Major: Electrical and Computer Engineering After the introduction of cellular technology and WiFi technology, the power cable is the last cable that has yet to be eliminated. Magnet ic induction is the leading technology to achieve wireless power transfer at high efficiencies (>75%) with power level ranging from several microwatts to thousands of watts. Using near -field operation at frequencies below 1 MHz significantly lowers the pro bability of interference and RF safety issues since the wavelength is extremely long and radiation is limited. Design rules are developed to sele ct the components of the system which are further verified by experiments. Interoperability analysis shows tha t the system is sufficiently robust that coils from one platform can be used in another platform without damaging the receiver. A method of load/fault detection is proposed and verified with both hardware and software implementation. Fi nally, three differe nt methods to control the unregulated receiver voltage control and power delivery is studied and it is determined that varying Cout in a single discrete step is able to allow the system to achieve sufficient control.

PAGE 16

16 CHAPTER 1 INTRODUCTION A wireless power transfer system is any system that has the capabilit y to transfer electrical power from a power source to an electrical load withou t any interconnecting wires. A wireless power transfer system will h ave three basic building blocks: 1 Transmitter: DC po wer is converted to AC power which is used to drive the transmitting antenna (coil). 2 Energy Transfer: The AC power is transmitte d through space to some point where a receiving unit is located. 3 Receiver: The transmitted AC power is collected and converted t o DC by a rectification process. Wireless power systems fall into two main categories, medium to long range, where the coverage is greater or equal to a typical Personal Area Network (PAN) which is typically a 10 m radius and short range, where the covera ge is localized within the vicinity of the transmitting device (typically a 5 distance). Although there have been attempts [ 1 3 ] to achieve long range power delivery via far -field techniques, the efficiency and power delivery is insufficient to charge eve n a typical cellular phone overnight. In order to provide power comparable to a typical wall mounted DC supply, the system would most likely violate RF safety regulations [ 4 ] or use a large number of transmitters resulting in an impractical implementation. Therefore, far -field techniques are most suitable for very low power applications (< 100 mW) unless they are used in less regulated environments such as military or space exploration. Inductive coupling has been one of the leading candidates in achieving wireless power transfer at power levels ranging from several microwatts to several kilowatts. Its operating range is limited as power delivery and efficiency degrades rapidly with increasing distance between the transmitting and receiving unit. Using near -field operation at frequencies below 1 MHz significantly lowers the probability of interference and RF safety issues since the wavelength is extremely long and radiation is limited.

PAGE 17

17 Unlike far -fields techniques, near -field techniques are extremely sensitiv e to the loading conditions of the receiver(s) as well as the number of receivers. The block diagram of the wireless power transfer system using inductive coupling is show in Figure 1 1. Inverter DC-AC Regulator and device under charge /power Impedance Transformation network Impedance Transformation network CouplingAC-DC Rectifier I2I1 V1V2+ +Regulator and device under charge /power Impedance Transformation network AC-DC Rectifier VN+... -IN Figure 1 1 Block diagram of the multiple receivers wireless power transfer system using inductive coupling. Other methods of wireless power transmission include using means of optics such as laser. Although, it is easy to focus a laser beam, energy conversion from electricity to laser is extremely inefficient. In addition, current state of the art photovoltaic cells used to convert the received laser to electrical energy ha ve an efficiency of typically less than 50%.

PAGE 18

18 1.1 History of Wireless Power Transmission The beginning of wireless pow er transmission can be dated back to 1 8 68 when James Clerk Maxwell developed the classical electromagnetic theory. Maxwells equations unify previous unrelated observations of electricity, magnetism and optics into one consistent theory. Later in 1888, Hei nrich Rudolf Hertz verified the existence of electromagnetic radiation by building the first radio transmitter. The first significant breakthrough in wireless power transmission technology happen ed in 1897 when Nikola Tesla filed his first patents based on the Wardenclyffe Tower which is also known as Tesla Tower in Colorado Springs According to [ 2 ] Tesla resonated a 3 feet diameter copper ball on top of the 200 feet power w ith 300 kW of power from the Colorado Springs Electric Company. There were variou s attempts after the Second World War to make an efficient transmi ssion of large amount of power wireless ly but they have failed. This is because the device technology during that time was insufficient to either generate sufficient amount o f power or effec tively rectify the received microwave power. In 1964, William C. Brown invented the Amplitron [5] which is commonly known as the crossed-field amplifier. The amplifier is capable of an output of 400 kW at efficiency of 70%. With the development of silicon Schottky-barrier diodes by HewlettPackard Associates, high efficiency high power wireless power transfer was made possible. In 1975, the largest wireless power transfer was demonstrated at the Venus Site of JPL Goldstone F acility. 450 kW of power was transmitted across one mile using a 26m diameter parabolic antenna [2]. The 3.4 m x 7.2 m receiving rectenna array achieved a rectified DC power of 30 kW at 82.5% efficiency. The system efficiency was only 6.67% efficiency with out taking the transmitter into consideration. Therefore, most of the power loss was still due to radiation. Generating a focused electromagnetic beam so that all of the energy can be projected onto a

PAGE 19

19 small area remains a challenging task The only two ways to achieve an efficient power transfer over a long distance is to increase the number of transmitting antenna elements so as to achieve a more focus ed beam or increase the number of elements of the receiving rec t e nna. R epeater stations can be used to red uce the radiation loses but the conversion losses to retransmit the power can be so high that it is not feasible. A more feasible alternative to the long range power transmission is via inductive coupling. However, the operating range is limited to the pr oximity of the transmitter. The roots of inductive coupling date back to 1831 when Michael Faraday demonstrate d that in order to induce a current on a conductor the magnetic field needs to be changing [6]. Faraday did not continue to pursu e the work in thi s area and left it for others to pick up from where he left off. Among many who have tried to study the phenomena was Joseph Henry. All the experiments were conducted by manually turning on and off DC supplies until the introduction of dynamo in 1860s making AC supply available. Sir William Grove was the first to connect a transformer to an AC source to generate high voltage for his laboratory work. One of the earliest successful demonstrations of a consumer p roduct using inductive coupling with considerable power is the inductive cooker. In 1971, during the National Association of Home Builders convention in Houston, TX Westinghouse Electric Company demonstrated single burner range using a 25 kHz operating fre quency [7]. On the other end of the power band, General Electric introduced the rechargeable cordless electric toothbrush in 1961, mitigating the risk of an electric shock [8]. The power transfer is achieve d by splitting a transformer core into two, for w hich one of the core is located into the base station and the other is located in the toothbrush. Energy can be coupled to the toothbrush by placing it onto the base station with proper alignment.

PAGE 20

20 The challenge of a robust inductive coupling is to be a ble to transfer energy to a receiving device with considerable lateral degree of freedom. In addition, portable electronics are getting very compact and lightweight and consumers are unwilling to accept the huge heavy ferrite core. By using an air core tr ansformer and with the receiver much smaller than the transmitter the coupling is extremely poor The only solution is to increase the operating frequency to enhance the coupling but it is only in the recent years that power transistors are capable of achi eving high power and high speed making considerable amount of power via magnetic induction possible. In addition, losses through the windings were significant due to skin depth and proximity effect until the introduction of Litz wires. 1.2 Modern Wireless Power Transmission The first company to start working on inductive coupling wireless power system was Splashpower [9] in 2001. In order to avoid the receiving device from blocking the vertical field, Splashpower developed a unique coil design which enab les the transmitting platform to transmit a horizontal field in both X and Y direction. This enables the receiver to be insensitive to location and rotation. However, with the introduction of many compact and low profile devices such as Apples iPod and Motorolas Razr, modern consumer electronics are more sensitive to the thickness of the device. Therefore, the allowable extra cross sectional area on the receiving device is almost non -existent In addition being one of the early adopters of the technology, the Splashpower system operates in tens of kilohertz making the system more inefficient than current solutions which operate at hundreds of kilohertz Splashpower was acquired by F ulton Innovation also known as eCoupled in 2008. eCoupleds wireless power technology was a by -product of its parent company Alticors eSpring water purifier [10] Approximately 10 years ago engineers at eCoupled were trying to prevent corrosion and electrical shock hazard to the ultraviolet lamp; in the end they developed a

PAGE 21

21 wireless power system to solve their problem The system designed by eCoupled involves complex control schemes and costly communication link to achieve resonance operation Although the cost of the system is reasonable to the unique water purif ier, it is considered to be high for cost sensitive consumer electronics which ha ve very low profit margin. However, eCoupled has successfully demonstrated a power delivery up to 1000 W grilling a piece of steak on a George Foreman grill as well as powerin g a 2000 W blender ConvenientPower [11] is brainchild of Professor Ron Hui from the City University of Hong Kong. His first technical journal was dated back to 2005 which is the first published for portable electronics. Professor Huis most significant co ntribution to the wireless power space is the design concept of generating an even magnetic field across a large area. Although, 7 journal papers were published and 6 patents were awarded, little was known about the system as there hasnt been any official demonstration conducted to the press. A vailable literature indicates that the system is a variation of magnetic induction proposed by Sp lashpower, eCoupled and WiPower [14] using the same underlying principles. PowerMat [12] which is based i n Israel is th e newest player in the wireless power space. Neither publications nor patents have been found so far. Similar to ConvenientPower little information is known about the technology Although, there was an official demonstration to the press during the recent Consumer Electronics Show 2009 in Las Vegas, there was little discussion about the technology PowerMat claimed that the innovation is more of the communication protocol than the wireless power technology. Based on the limited information provided at CES a nd the companys website, one can deduce that their technology is also a variation of magnetic induction similar to Splashpower, eCoupled and WiPower.

PAGE 22

22 WiTricity [13] was developed in a research team led by Professor in MIT. Two carefully aligned huge capacitively loaded coils with a diameter of 24 inches were used to transmit power to a 60 W light bulb at 2 m away with an approximate efficiency of 40%. The operating frequency was at 10 MHz instead of the sub megahertz range. Greater operating range was achieved by increasing operating frequency and coil size. In addition, the system was designed for resonance, thus the system will need complex control schemes to seek resonance under varying load conditions. Research work on wireless power transmission via magnetic induction for WiPower [14] started during late 2006 at the University of Florida. Instead of seeking resona nce, the system design is based on the load impedance response taking varying load impedance into account. The WiPower wireless power system ensures that the systems natural response matches closely with a typical off the shelf wall wart, ensur ing uncondi tional stability at any load impedance. The system can be easily scaled to the required power level and power delivery up to 300 W has been demonstrated. PowerCast [15] founded in 2003 is the only wireless power company which took the long range far fiel d approach. Operating at ISM bands of 900 MHz and 2.4 GHz the system is able to achieve considerable coverage. However, the system suffer s from the fundamental problem of power losses due to radiation resulting in mediocre power delivery. The efficiency is far worse than Browns experiment in 1975 where Brown use d a focused beam for a point to point RF link Since the system requires covering a large area, a point to point system using a focused beam will not be practical for consumer applications To date the PowerCast system only has the power handling capabilities of lighting LEDs or provide supplementary power to low power wireless sensor nodes

PAGE 23

23 1.3 Background Information The purpose of this section is to provide the background information relating to th e material presented in later chapters. This section starts with a discussion the fundamental theory of operating via phase response and how power control can be achieved by this technique. Following that a discussion of two popular high efficiency switch mode power amplifier topologies, the Class D and Class E, which includes their analysis equations are presented The discussion will enable the reader to appreciate the advantages and disadvantages of each topology. Using the analysi s, the Class E power amplifier is found to be a more suitable amplifier than the Class D power amplifier because it is able to achieve high efficiency operation from a phase angle of 40 to 70 while controlling the power delivered to the receiver 1.3.1 F undamental Theory of Operation via Phase Response To ensure unconditionally stable operation, the system must be able to deliver the required power with respect to the load resistance closely matching that of a constant voltage power source The transmitted power should decrease when the load impedance increases vice versa. Error in selecting component values or control scheme will result in either oscillation or undesirable power delivery trend, that is, the transmitter will increase its output w hen the load resistance increases. The switching buck regulator at the receiver will attempt to maintain its power delivery by increasing its input resistance or decrease its duty cycle resulting in a positive feedback. Poor efficiency will be observed due to excess power dissipated as heat and device failure may occur due to over -voltage. There are various methods to achieve the desirable power delivery trend. A closed-loop system will require some form of control scheme and tunable components/supply volt age as well as a communication link between the receiver and transmitter. This will cause the system to be complicated bulky in size and costly. The optimum solution to the problem is to achieve the

PAGE 24

24 desired power delivery trend with respect to load resist ance across a wide range of load resistances. This can be achieved by designing the system to have an effective resistance response or phase response looking into the transmitting coil. Since the class E power amplifier which has a limited range of operat ing load resistance is selected for the system and it is difficult to generate the desirable effective resistance trend while maintaining a fixed reactance using a pair of loosely coupled coils, the technique of power delivery control exploiting phase res ponse will be more appropriate. Figure 1 2 shows the complex plane of inductive and capacitive load. Since only positive resistive loads are consider, the analysis will only consider the right half plane. The class D power amplifier [16] and class E power amplifier [18] are not suitable to drive capacitive loads, thus, the first quadrant of the Figure 1 2 will be the only valid region of operation. Figure 1 2. Complex plane of inductive and capacitive load.

PAGE 25

25 Figure 1 3. Phasor diagram Where P: Real power. Unit: watts (W) Q: Reactive power. Unit: volt amperes reactive (VAr) S: Complex power. Unit: volt amperes (VA) |S|: Apparent power. Unit: volt amperes (VA) : Phase angle of the load. Figure 1 3 shows the phasor diagram to illustrate the concept of power delivery versus phase response. The real power delivered to the load is reduced when the magnitude of the apparent is kept relatively the same while increasing its phase angle to be more inductive in the anti -clockwise direction The system can make use of such delivery trend to achieve a desirable power delivery profile with increasing load resistance. The trend can be achieved by using a series of reactive networks before the transmitting coil and after the receiving coil to shape the impedance resp onse accordingly. The m agnitude of the apparent power can also be varied to improve the power delivery trend.

PAGE 26

26 1.3.2 Class D Power Amplifier Q1 Q2 Clock 1 Clock 2 VDD L ZLoad C Zout Figure 1 4 Class D power a mplifier. Driving the transmitting coil is the most crucia l part of the system. In order to achieve high efficiency, only switch mode amplifier topologies will be considered for the system. Since the system should be kept simple with minimum component count, only Class D and Class E power amplifier topologies are considered. The class D power amplifier will be discussed in this section whereas the class E power amplifier will be discussed in the following section. Class D power amplifiers have been traditionally considered as the most practical highfrequency swit ch mode amplifiers. The main advantage of a Class D power amplifier is that the voltage stress across the transistors is equal to the supply voltage making them suitable for either high voltage or high speed. A circuit of the Class D power amplifier consis ting of two transistors Q1 and Q2 which can be modeled as two bidirectional switches, an LC output filter and a load ZLoad is shown in Figure 1 4 It was found that the class D power amplifier is able to drive both resistive and inductive loads but not capacitive loads [17]. The body-drain diode of the MOSFET can be used as an anti parallel diode in case of an inductive load.

PAGE 27

27 The analysis of the class D power amplifier is based on Figure 1 4 with the following four assumptions. 1 The transistor and diode for ms an ideal resistive switch. All parasitic capacitance and switching time are neglected. In addition, the turn on resistance remains constant. 2 The turn on resistance of the Q1 is sufficiently small that voltage drop across Q1 can be neglected. The turn on resistance of Q1 can be moved to be part of Zout. 3 The reactive components L and C are not sensitive to frequency. 4 The loaded quality factor of the output LC filter is high enough that the current through ZLoad is sinusoidal. The impedance Zout looking into the load network can be described by Equation 1 1 1 _P _P _P _P 1 ZZjLR outLoadLarasitic jC RR Carasiticturnontransistor RRR LoadLarasiticCarasitic XCLC Load Rj turnontransistor C (1 1) According to Figure 1 4 it can be inferred that the input voltage to the load network is a square wave from ground to VDD neglecting the losses through the turn on resistance of the transistor. The square wave can be described by Equation 1 2. for 0 0 for 2 Vt DD v out t (1 2) It is assumed that only the fundamental tone is delivered to the load. The fundamental tone of voltage vout is described in Equ ation 1 3 from Fourier analysis. The peak voltage is approximately 0.64 VDD and the RMS voltage is approximately 0.45 VDD. 2 sin V DD vt out (1 3)

PAGE 28

28 The current through transistor Q1 can be described as sin for 0 1 0 for 2 Itt out i Q t (1 4) By integration Equation 1 4 across a single period, we can obtain the DC component of the supply current. sin()() 0 2 cos 2 2 2 I out I tdt DD I out VR DDout Z out (1 5) If the load is a purely resistive load, the DC supply current can be expressed by Equation 1 5 by setting Zout of Equation 1 6 to Rout. 2 2 V DD I DD R out (1 6) The DC input power can be then described by Equation 1 8 from Equation 1 5. 2 2 2 2 VR DDout P IN Z out (1 8) Therefore, the DC input power of a purely resistive load can be expressed by Equation 1 9 by setting Zout of Equation 1 8 to Rout. 2 2 2 V DD P IN R out (1 9) Iout can be easily derived from Equation 1 3.

PAGE 29

29 2 V DD I out Z out (1 10) By using the same concept as above, a resistive loaded class D power amplifier will have an output current as shown in Equation 1 11. 2 V DD I out R out (1 11) Power delivered to the load can be then derived by using ohms law usin g Equation 1 11. Since we are only interested in the real power delivery, real power delivery can be calculated by neglecting all imaginary terms of the apparent power delivery in Equation 1 -12. 2 () 2 2 2 2 2 22 () 2 2 IZ outLoad P outapparent VZ DDLoad Z out VR DDLoad P outreal Z out (1 12) Power loss of the class D power amplifier can be described in Equation 1 13. The three major factors contributing to the power loss of a class D power amplifier are the parasitic resistances of the series inductor and capacitor as well as the turn on resis tance of both transistor s 2 _P _P 2 IRRRoutLarasiticCarasiticturnontransistor P loss (1 13) From Equation 1 14, i t can be seen that an ideal class D power amplifier is able to achieve 100% efficiency which is desirable for this application.

PAGE 30

30 () PP P outreal P in R Load RRRR LarasiticCarasiticturnontransistorLoad (1 14) 1.3.3 Class E Power Amplifier The class E power amplifier [17] consists of a single switch with a load network. The main advantage of the class E is that it only consists of a single transistor and does not require a pair of out of -phase gate drive for the trans istors as shown in Figure 2 3 for the class D. Therefore, a clock signal with sufficient drivability can be directly fed to the gate of the transistor without a gate driver to control the driving signal to the high side and low side of the amplifier (Class D). However, the voltage stress on the transistor of an ideal class E power amplifier is 3.562 times the supply voltage, making the power amplifier unattractive at high supply voltage. For example, if the supply voltage of an optimized class E power amplifier is 48V, the drain voltage can swing up to approximately 170 V. Therefore, by providing a 80% safety margin, one need to use a transistor with a breakdown voltage of approximately 220 V. In addition, the high drain voltage might cause higher noise levels in the system due to poor PCB design and if appropriate noise reduc tion is not implemented. Reference [17] stated three objectives for the transistors drain voltage and current waveform to achieve class E operation. 1 The rise of the voltage across the drain of the transistor should only occur when the transistor is turned off. 2 The drain voltage across the transistor should be zero at the time when the transistor is turned on. (Zero Voltage Switching operation ZVS) 3 The gradient of the drain voltage should be zero at the time the transistor is turned on. (Zero Derivative S witching ZDS)

PAGE 31

31 Q1 Clock VDD L1 ZLoad C2 Zout LDC C1 Figure 1 5 Class E power a mplifier. The detail analysis of an ideal class E power amplifier is based on Figure 1 -5 which was first discussed in [18] [19] by Frederick H. Raab with the following five assumpti ons. 1 The RF choke LDC is sufficiently large that the current flowing through has negligible AC components. 2 The series LC filter of L1 and C2 has sufficiently high Q that the output current to the load can be considered to be a sinusoid (single tone). There fore, it can be assumed that the filter is an open circuit at higher harmonic frequencies. 3 The shunt capacitance of the transistor is independent of the drain voltage or small enough to be neglected. 4 The transistor is an ideal switch and has only 3 differe nt states a On state with zero resistance. b Off state with infinite resistance. c Transition state for which the response time is zero. 5 The load will always be a resistive load. Any extra reactance can be added to the series LC filter. Using assumption 2 the output voltage and currents across the load should be sinusoidal (single tone free of higher order harmonics ) and can be described by Equation 1 15 and 1 16 respectively.

PAGE 32

32 sin vtct Load (1 15) sin c itt Load R Load (1 16) The drain voltage should be also a sinusoid when the transistor is turned off but with a phase shift due to extra reactance X introduced by the series LC filter. Therefore, the drain voltage is expressed in Equation 1 17 as the summation of the voltage drop ac ross the load and the series LC filter 1 1 2 sin sin sin11 cjL C vtct t drain R Load ct (1 17) Where, 2 1 1 2 1 1 2 L C cc c R Load (1 18) 1 1 1 2 tan 1 L C R Load (1 19) When switch is off, the drain voltage is produced by the charging of capacitor C1 by the difference in current between the supply current and the current driving the l oad network. Therefore the drain voltage can be obtained by integrating the across the window when the switch is turned off as shown in Equation 1 20.

PAGE 33

33 1 1 1 1 sin 2 sin 2 cos tvt iudu drain C C t o t c I udu CC BR Load y I c CC yy B BR Load I c CCtt BBR Load (1 20) Where B is C1 and y is half of the transistor turn off time. T he magnitude of the fundamental tone can be obtained by Fouri er integral Since the drain voltage waveform is not symmetrical around /2, there will be an unknown phase offset with respect to the load voltage waveform 1 2 sin 0 11 2 sincossin 1 11 2 2sinsin2cossin2sincos 1111 2 sin2sin22sin 2 cvttdt drain I c CC yyy B BR Load I CC y yyy B c yy BR Load (1 21) Using Equation 1 21 we can s olve for c by substituting c for c1 and rearranging to obtain Equation 1 22. 2sincos2cos2sinsin 11 1 sin2sin2sin2()cossin 1 2 ,,,,, yyyyy cIR CCLoad BR yyy y Load IRhyBR CCLoad (1 22)

PAGE 34

34 Since the fundamental tone at the drain is defined by a sinewave of phase 1, there can be no cosine or quadrature component with respect to phase 1. Replacing the sine term with a cosine term in Equation 1 21, Equation 1 23 is obtained 1 2 0 cos 0 1 sin2sinsin 11 2 2cossin2sinsin2sincos 1111 2 cos cos2sin2 2 vttdt drain I c CC y yy B BR Load I CC y yyy B c yc y BR BR Load Load (1 23) Therefore by rearranging the terms in Equation 1 23 and solving for c, Equation 1 24 is obtained 2sinsin2coscos2cossin 111 1 2sinsinsinsin2cos2cos 1 2 ,, yyyyy cIR CCLoad yyyy cIRgy CCLoad (1 24) Since both Equation 1 22 and Equation 1 2 4 describes c, by substituting Equation 122 into Equation 1 24, Equation 1 25 is obtained. ,,,,,,, gyhyBR (1 25) In order to solve equation 1 25, we need to c ross multiple the denominators of the two functions g and h and break down the higher order terms of sine and cosine to their first order equivalent Since the only unknown in Equation 1 25 is a fter rearranging the terms we can obtain a solution for b y e ither Equation 1 26 or Equation 1 27 and they provide a single consistent answer.

PAGE 35

35 2 2sincos2cossin2sinsin2sinsin 2sinsin2coscos2sincos sinsinsincos 2 2sinsin2coscos2sincos2sincos 2sincos2cossin2sinsin cossincoscos 1 tan 2s yyyyy y yyyyy BRy yy yyyyyy yyyyy yyy y incos2cossin2sinsin sinsinsincos 2sinsin2coscos2sincos cossincoscos yyyy BRy yy yyyyy yyy (1 26) 2sinsin2coscos2sincos 2 sinsinsincos2cossin 2sincos2cossin2sinsin 2 cos2sinsinsincoscos 1 tan 2sincos2cossin2sinsin 2 sinsinsincos2cossin yyyyy BRy yy y yyyyy yyyy yyyyy BRy yy y 2 2sinsin2coscos2sincos2sinsin 2sinsin2coscos2sincos 2 cos2sinsinsincoscos 2 2sincos2cossin2sinsin2sincos yyyyy y yyyyy yyyy yyyyyy (1 27)

PAGE 36

36 Once we obtained from either Equation 1 26 or Equation 1 27, we can obtain either g or h by substituting into Equation 1 22 or E quation 1 24 to get c. Since we assume that there is no voltage drop across RF choke, we can assume that the integrated drain voltage across one period is equal to the value of the DC s upply voltage as shown in Equation 1 28 where Rdc is the input resistance of the class E power amplifier looking into from the supply voltage port. 1 2 0 2 2 sin cos 22 2 2 22sin2sinsin 2 Vvtdt ccdrain y I CCygytgtdt B y I CC yygygy B IR CCdc (1 28) Using E quation 1 15, E quation 1 22, E quation 1 24 and E quation 1 28. Power delivery and efficiency can be derived as followed. 1 2 22sin2sinsin 2 R yygygy dc B (1 29) 2 1 2 22 2 22 2 2 c P o R Load IgR CCLoad VgR ccLoad R dc (1 30) 2 PVI iccCC V cc R dc (1 31)

PAGE 37

37 22 22 P o P i VgR ccLoad R dc (1 32) Although, it is known that an optimized ideal single -ended class E power amplifier requires its switching transistor to have a 3.562 times higher breakdown voltage than that of the Class D, it is often overlooked that the output power of an optimized ideal single -ended Class E power amplifier as shown in Equation 1 33 from [20] is 2.847 times higher than that of typical optimized ideal Class D power amplifier as shown in E quation 1 34. 2 8 2 4 2 0.5768 V CC P outClassE R V CC R (1 33) 2 2 2 2 0.2026 V CC P outClassD R V CC R (1 34) From Equation 2 33 and Equation 2 34 we can conclude that in order to achieve similar output power, the Class D power amplifier requires a supply voltage that is 1.687 times higher than that of Class E power amplifier. When supply voltage is constr ained, a Class E transmitter power amplifier is preferred to a class D transmitter because of higher output powe r at the same supply voltage.

PAGE 38

38 CHAPTER 2 WIRELESS POWER TRANS FER SYSTEM Class E Power Amplifier Voltage Regulator Impedance transformation network Impedance transformation network Inductive CouplingReceiver I2I1 V1V2+ +Detection and Control Device under charge /power Figure 2 1. Wireless power transfer system diagram. The system diagram of the proposed wireless power transfer system via magnetic induction is shown in Figure 2 1. Energy transfer between the transmitter and receiver is achieved by inductive coupling using a pair of air core coils. The class E power amplifier discussed in Chapter 1 under background information is used as an inverter to convert the input DC power to high frequency A C power of several hundred kilohertz. In order to achieve the desirable power delivery trend and impedance looking into the transmitters load network, an impedance transformation network is inserted before the transmitting coil and after the receiving coi l. The received AC power is converted to DC by the receivers rectifier. A full wave rectifier can be used to reduce device stress on the diodes or a half wave rectifier can be used to reduce the receiver PCB footprint. Depending on loading conditions, the rectified DC voltage after the rectifier varies across a wide range. Therefore, a voltage regulator is used to regulate the

PAGE 39

39 fluctuating DC voltage to a stable DC voltage before it is used to power the receiving device. A detection scheme using a micro -con troller and sensing circuits is required to ensure the transmitter is in its nominal operating conditions. Such a detection scheme is proposed and discussed in Chapter 7. Control signals can be used to disable the transmitter or tune the system power deliv ery profile to enhance system efficiency and robustness. Techniques to control power delivery profile as well as receiver voltage are discussed in Chapter 8. 2.1 Single -channel Class E Power Amplifier The class E power amplifier is an elegant design with minimum number of components. Theory of operation of the class E power amplifier is discussed in Chapter 1. The amplifier consists of three main sections, the DC feed network (typically a single choke inductor), the switch (typically a single N channel MOSFET) and the load network. The network from C1 towards the load in Figure 1 5 can be considered part of the load network and will be discussed in Chapter 3. It is preferred that the choke inductor of t he DC feed network (LDC) to have a sufficiently large reactance so that it can be considered as an open circuit for the AC signal. Since impedances looking into the load network under most operating conditions will not exceed 50 an effective impedance o f 500 or better for the DC feed network is sufficient. Therefore, for a system operating at 250 kHz, the minimum inductance value should be 330 H. Although, it is desirable to have the inductance value as high as possible, the parasitic r esistance will be large resulting in poor efficiency and heating issues. In addition, an inductor with a larger inductance will typically of a larger size making the transmitter bulky. The switch is the core of the class E power amplifier. Selecting the correct switch is very important, it needs to be sufficiently fast, ha ve a low turn on resistance and ha ve a sufficiently high breakdown voltage. To achieve a high breakdown voltage, t he most practical candidate for

PAGE 40

40 the switch is the N channel power MOSFET. The theoretical maximum voltage across the switch for a class E power amplifier is slightly more than 3.5 times of the supply voltage if it is optimized to drive a fix load. As seen from Figure 3 9 and Figure 3 17, the drain voltage can be up to 20% higher tha n the theoretical maximum while driving a varying load. Therefore, the maximum voltage across the switch should be approximately 4.2 times of the supply voltage during nominal operation instead of 3.5 times However, the selected MOSFET should have a break down voltage of at least 5.5 times of the supply voltage to achieve a 0.8 factor of safety margin for various fault conditions A wireless power transfer system with a supply voltage of 15V will require a transistor with a breakdown of 100 V or better. Al though, a higher breakdown voltage will ensure the system to be more robust, a transistor with a higher breakdown voltage tends to be slower in speed and have a larger turn on resistance. In an ideal case the voltage across the transistor should be 0 V whe n a large current is driven across the transistor in its turn on state. To achieve this operating condition, the turn on resistance of the transistor needs to be sufficiently small. This also helps in reducing energy being dissipated as heat across the tur n on resistance. A turn on resistance of 0.1 The capacitance between the drain and source of a MOSFET varies with the drain voltage. To ensure than the fluctuation of drain to source capacitance with respect to the drain voltage can be neglected, the variation should be much smaller than C1 (typically <5% of C1). If the gate to source capacitance is too large, a gate driver is required to provide sufficient drivability to effectively turn on and off the transistor. Finall y, the transistor must be able to switch sufficiently fast to reduce switching losses, the rise time and fall time of the transistor should be better than 1% of the period of the operating frequency (better than 40 ns for operating frequency of 250 kHz).

PAGE 41

41 2 .2 Multi -channels/Stackable Class E Power Amplifier Q1_1 Clock 1 VDD L1_1 ZLoad C2 LDC_1 C1_1 Q1_2 Clock 2 L1_2 LDC_2 C1_2 Figure 2 2. Schematic of the dual channel C lass E power amplifier. In addition to the typical single ended class E transmitter, a multiple channel class E transmitter with in dependent gate drive is shown in Figure 2 2 (Dual -channel). The dual channel class E allows the system to shut down one of the channels to switch to a lower power mode so that the receivers regulator responds by reducing its input resistance. The analysis of the dual channel class E is straightforward. If both the channels are enabled, the equivalent inductance across both L1 is L1/2, while if one channel is enabled, the equivalent inductance is simply L1 making Zout more inductive while keeping the real components of Zout the same. If Zout is more inductive, more reactive power and less real power is transmitted to ZLoad reducing power delivery. Likewise, the equivalent value of Cshunt is doubled for a dual -channel c onfiguration.

PAGE 42

42 The dual channel class E transmitter can be easily scaled to multiple channels, but the size increases along with a higher probability of device mismatches. As the number of channels increases, the problems caused by device mismatches might f ar outweigh the benefits. It was found empirically that the number of channels should be kept at a maximum of three. The observed efficiency improvement was negligible when the number of channels was increased from three to four and the efficiency started to degrade at five channels, predominately due to variations in inductance value. Although this technique enables the system to operate at a higher efficiency, the tradeoff is that the power delivered to the load network is fed across a single inductor ins tead of two for a dual -channel case, thus the equivalent parasitic resistance of Lout is doubled for a dual channel topology, reducing the potential enhancement of the system efficiency as power loss via the inductor remains the same in regardless of the n umber of channels. 2.3 Inductive Coupling Power transfer for the system is achieved via magnetic induction between two air core coils. Vertical magnetic field is the dominate component of power transfer because both the transmitting and receiving coils ha ve large surface area. By having a larger surface area, the coil is able to intercept more of the magnetic flux and receiv e more power. In addition, modern portable electronic devices are extremely thin in profile having limited cross sectional area for an y magnetic flux to cut through the cross sectional area. Appropriate shielding [ 21] at the expense of weight and thickness can be used to make the system more robust in environments where the system's magnetic field is likely to interact with other nearby objects. However, shielding is not the focus of this dissertation and it is assumed that the system will be working in an environment free of structures and materials that will significantly affect the system performance or is sufficiently shielded to begi n with

PAGE 43

43 Although it would be ideal for both the transmitting coil and the receiving coil to be of the same size to ensure maximum coupling as shown in [ 22] [24], a practical system will tend to use a receiver coil significantly smaller than the transmit ting coil. This allows a user to freely place a device in any orientation without intentionally docking it as well as to place multiple devices on the transmitting coil. Therefore, in order to achieve consistent power delivery and impedance response regardless of the receiving coils location, the transmitting coil must be able to generate a magnetic field that has relatively even distribution. This is achieved by using the method proposed in [ 25]. Alternative methods are also proposed in [26] [27]. The voltage and current of the transmitting coil and the receiving coil can be described using E qu ation 2 1 Equation 2 2 and Equation 2 3 from [22] [27] neglecting any second order effects such as skin depth and proximity effect which can be mitigated by using Litz wires. 1111122 VjMIjMI (2 1 ) 2211222 VjMIjMI (2 2 ) 121122 MkMM (2 3 ) Where V1 is the voltage at the transmitting coil (Figure 2 1) I1 is the current at the transmitting coil (Figure 2 1) V2 is the voltage at the receiving coil (Figure 2 1) I2 is the current at the receiving coil (Figure 2 1) M11 is the self inductance of the transmitting coil M22 is the self inductance of the receiving coil M12 = M21 is the mutual inducta nce of the two coils k is the coupling coefficient between the two coils

PAGE 44

44 By Ohms law: 1 1 ZRjX txtxtx V I (2 4 ) 2 2 ZRjX rxrxrx V I (2 5 ) Using Equation 2 1 Equation 2 2 and Equation 2 4 and assuming a time -harmonic impedance looking into the transmitting coil for a single receiver is derived as Equation 2 6 22 22 1222 12 11 22 22 22 22 MMX MR rx rx Z jM tx RMX RMX rx rx rx rx (2 6 ) 2.4 Impedance Transformation Network The purpose of impedance transformation network on the primary and secondary sides of the coupling is to achieve maximum power transmission and efficiency by operating within the optimum impedance range looking into the transmitter load network [29] over a wide range of load resistances In consideration of size and efficiency, capacitors instead of resistors and inductors should be used for the network. This is because resistors dissipate power and the size of a low loss inductor is generally large. Althou gh, a multi -element transformation network might achieve a better response, for simplicity and low component count, the system uses a single -element transformation network. The four possible topologies of the single -element transformation network are shown in Fig ure 2 3

PAGE 45

45 Ctx Crx Ctx Crx Crx Ctx Ctx Crx A B C D Figure 2 3 Topologies for a single -element impedance transformation network. A) Series -Series Topolog y, B) Series -Parallel Topolog y C) Parallel -Series Topology and D) Parallel Parallel Topology Fundamentall y, a series capacitor only introduces a negative reactance and does not change the real part of the impedance. On the other hand, a parallel capacitor changes both the real and imaginary parts of the impedance. To simplify the analysis, the receiver input impedance is modeled using a variable resistor load and Equation 2 7 illustrates the transformation performed by the parallel capacitor. 2 222222 11 R CR rx rxrx Zj rx CRCR rxrx rxrx (2 7) Equation 2 7 shows that the resistance Rrx is compressed by a factor of 2221/(1 )rxrxCR Thus, the equivalent resistance Rrx decreases with increasing load resistance. At high load resistance, the transformed resistance is small. Therefore, a significant part of the received power is dissipated across the receiving c oil as heat. This phenomenon is desirable if the receiver is in a state that requires very little power or during trickle charge. Therefore, it has a decoupling effect regulating the power delivery with increasing load resistance. However, this should oc cur if and only if the transmitter is designed to output limited power under this operation condition because heating can become a problem if too much power

PAGE 46

46 is being dissipated across the receiving coil. Due to the parallel capacitor, a reactive term jXrx i s introduced to the equation The reactive term jXrx decreases nonlinearly from null with increasing load resistance with an asymptote of 1/rxC This is use d to compensate the receiving coils self inductance. From Equation 2 6 it can be observed that the resistance looking into the transmitter coil is reduced significantly if the resistance looking from the receiver coil into the receiver is increased. Rtx is further reduced because the mutual inductance is relatively small. If the total resistance looking into the transmitting coil is mainly the parasitic resistance of the transmitting coil, limited power is transmitted to the receiver as most of the power is dissipated across the transmitting coil as heat. Therefore, it is pre ferred for a wireless power transmission system using loosely coupled coils to have a parallel capacitor across the receiving coil. By substituting Equation 2 7 into E q uation 2 6 the expression of impedance looking into the transmitting coil with a parall el capacitor across the receivin g coil is shown in Equation 2 8. 22 12 222 1 2 2 2 22 222 222 11 2 22 1222 222 1 11 2 2 2 22 222 222 11 R M CR Z tx R CR M CR CR CR MM CR jM R CR M CR CR (2 8) For the transmitter transformation network, a series or parallel topology can be used. To maintain an ideal efficiency above 95%, the allowable variation of load resistance of an ideal class E amplifier should be kept within +55% and 37% [ 18]. If the variation of resistance

PAGE 47

47 looking into the transmitter coil is too large, it is preferred that a parallel capacitor is used instead of a series capacitor to compress the resistance. A suitable capacitor value is needed to ensure that the transmitter does not suffer immediate failure when there is no receiving coil, and to produce an increasing reactance trend with increasing load resistance so as to ensure the preferred power delivery trend. On the other hand, an appropriate receiver capacitance value can also be selected so that the resistance looking into the transmitting coil is kept within a reasonable bound but not too small to impact on the coupling efficiency and to produce an increasing reactance trend with increasing load resistance so as to ensure the preferred power del ivery trend A suitable series transmitting capacitor is then needed to translate the reactance looking into the transmitting coil to cancel part of the self inductance of the transmitting coil Based on the above analysis, both the parallel -parallel and s eries parallel impedance transformation network topolog ies can be used for the system. 2.5 Receiver Since more receivers will be paired up with a single transmitter and most receivers are intended to be integrated into compact portable electronics, the re ceiver of the wireless power transfer system needs to be low cost and compact in size Therefore, the receiver will only consist of a rectifier to convert the AC power to DC power and a voltage regulator to ensure a stable DC voltage is use to power the de vice. The effort of attaining universality for cell phone power port still faces huge resistance from the OEMs. By having a proprietorship communication protocol for a wireless power transfer system will work against a universal charging/power platform to reduce waste and power consumption Therefore, a communication link should be avoided unless it is the last resort and the wireless power receiver is integrated into the device tapping into its existing wireless communication system.

PAGE 48

48 Reverse recovery time of the diodes used for the rectifier is critical because the operating frequency is much higher than the typical 50/60 Hz AC power lines. A good rule of thumb is to select a diode with a reverse recovery time less than 1% of the operating frequencys perio d. For example a 200 kHz system should use a rectifying diode with a revers e recovery not slower than 50 ns It is easy to find schottky diodes with negligible reverse recovery time at breakdown voltages below 100 V for low to medium power applications (< 100W). However, for high power applications either an ultrafast recovery diode or schottky diodes in series are required to ensure they are not damaged by the reverse biased voltage. Figure 2 4. Input voltage of a half wave rectifier and a full wave rectifier. Figure 2 4 shows the input voltage of both a half wave rectifier and a full wave rectifier where t he flat section of the waveform is when the rectifying diode conducts charging the charging holding capacitor use d to smooth out the voltage ripples at the output of the rectifier. The diodes for a full wave rectifier conducts during both the positive and negative cycle of the waveform whereas the diodes for the half wave rectifier only conducts during either cycle de pending on the orientation of the diode. For the specific load condition shown in Figure 2 4 1.831 1.832 1.833 1.834 1.835 1.836 1.837 1.838 1.839 1.830 1.840 -30 -20 -10 0 10 -40 20 time, msec Full_Wave_Input_Voltage Half_Wave_Input_Voltage

PAGE 49

49 the full wave rectifier requires a diode with a breakdown voltage of 20 V and better to operate On the other hand the half wave rectifier will require a diode wi th a breakdown voltage of 40 V and better to operate which is a lot higher than a full wave rectifier Although, a half wave rectifier helps to reduce the size of the receiver PCB by reducing the number of diodes from four to one it will require a diode o f a much higher voltage rating to operate. Under most situations a full wave rectifier is preferred for high power applications where a slight increase in PCB size is not an issue. A switching buck regulator is preferred over low dropout regulators for this application because the input voltage swings across a considerable range and high efficiency must be maintained. In order to keep the receiver compact the output inductor of the switching regulator should be kept below 10 H which can be achieve d b y selecting regulators with switching frequencies above 500 kHz. Although, increasing the switching frequency will result in the voltage regulator being more compact, the efficiency starts to degrades losing power to heat

PAGE 50

50 CHAPTER 3 DESIGN OF IMPEDANCE TR ANSFORMATION NETWORK The design of the proposed wireless power system starts by setting constraints of the dimensions of the transmitting and receiving coil as well as operating frequency. This chapter presents the design rules for two systems using the pa rallel -parallel impedance transformation network and the series -parallel impedance transformation network achieving desirable power delivery profile. The parallel -parallel impedance transformation network design rule is based on the dual channel transmitte r topology to achieve high power transf er while the series parallel impedance transformation network design rule is based on the single -channel transmitter topology to achieve compact low power design. A different operating frequency is used to illustrate that the design rule applies to a wide frequency range. 3.1 Series -Parallel Impedance Transformation Network 3.1.1 Introduction An operating frequency of 240 kHz is used for the design of the series parallel impedance transformation network topology Since high efficiency schottky diodes are used for the receiver rectifier, the resistance looking into the rectifier should be very close to the resistance after the rectifier as very little energy is lost. Therefore, to simplify the analysis, the load resistan ce is defined as the equivalent resistance looking into the rectifier instead of after the rectifier. It is desirable for the receiving coil to be much smaller than the transmitting coil, but efficiency and power transfer capabilities start to degrade sign ificantly due to poor coupling if the receiver is too small. Therefore, i t is preferred to keep the coupling coefficient k above 0.1. To minimize space usage as well as ease of integration into the target device, the receiving coil is typically tightly wou nd. However, due to the requirement of ensuring a consistent coupling coefficient regardless of position and orientation the windings of the transmitting coil are very different

PAGE 51

51 from the receiving coil. The gaps between each turn of the transmitting coil a re spaced in a non uniform manner with the windings more spaced out when it gets to the middle of the coil so as to achieve even field distribution Thus, achiev ing consistent performance regardless of the placement of receiving coil. Figure 3 1 Windings of a 20cm x 20cm transmitting coil used for experimental verification. Fig ure 3 1 shows the windings of a 13 turns 20 cm x 20 cm transmitting coil from [25] which will be use for later experimental verification, the last 3 turns are overlapped ont o each other. Designing of the transmitting coil windings to achieve an even magnetic field distribution is not the scope of this dissertation The normalized power delivery with respect to location of the transmitting coil in Fig ure 3 1 using a tightly wounded receiving coil of 9 cm x 6 cm with 6

PAGE 52

52 turns is shown in Fig ure 3 2 using the centre of the receiving coil as reference The receiver is 13.5% of the size of the transmitter. Power delivery variation is kept within 5% with a standard deviation of 2.2 %. Since there is no obvious distribution of power delivery trend with respect to receiver location, it can be assumed that the generated magnetic field is even and the variations are due to measurement errors. Figure 3 2 Normalized power deliver with respect to location of transmitting coil in Fig ure 3 1 using a receiving coil of 9 cm x 6 cm. Key parameters of the coils including self inductances, mutual inductance, and parasitic resistances can be extracted by measuring the fabricated coil with an impedance analyzer or analyzing with electromagnetic simulation tools. The coils were fabricated using 100/40 round served Litz wires for the experiment to mitigate proximity effect and skin effect. The 100/40 round served Litz wires consists of 100 strands of 40 gauge wires insulated from each other. The

PAGE 53

53 self inductance of the transmitting coil is 45.3 H with a parasitic resistance of 0.5 inductance of the receiving coil is 5.2 H with a parasitic resistance of 0.1 between th e coils is 2.8 H with a coupling coefficient of 0.1824. Measurements of the coils were measured using the HP4192A LF Impedance Analyzer. The design rule for the series parallel impedance transformation network will be based on Figure 3 3. Although Cout an d Ctx are two separate capacitors, Ctx will be considered to be part of Cout. Cshunt Ctx AC Crx Rload Zrx Ztxcoil ZtxLoutLDC Transformation network Cout Figure 3 3 Simplified schematic of wireless power transfer system using series-parallel impedance transformation network and Class E transmitter. Ztx Impedance looking into the transmitter load network. Ztxcoil Impedance looking into the transmitting coil. Zrx Impedance looking into receiver network. RLoad is the equivalent resistance looking into the rectifier. 3.1.2 Determination of Crx value The design of the system starts from the receiver looking into the load. The typical impedance response for different parallel capacitors is shown in Fig ure 3 4 The capacitance value is selected based on the inductance of the receiving coil as well as th e mutual inductance between the coils. Although, it will be desirable to achieve a maximum resistance looking into the transmitting coil across a wide range of load resistances [ 29], the resistance variation looking into the transmitting coil might be too large thus requiring a shunt capacitor across the

PAGE 54

54 transmitting coil to compress the resistance resulting in the parallel parallel impedance transformation network topology which will be discussed later Therefore, it will be more practical to select a c apacitor value that will generate the desirable resistance range looking into the transmitter load network Ztx as shown in Figure 3 3 Figure 3 4 Resistance and reactance of Zrx versus load resistance at different with different Crx. (50 nF, 100 nF and 150 nF) S hifting of the reactance value to achieve a desirable phase response so as to achieve a desirable power delivery profile can be done by varying Cout or Lout in Figure 3 3 In order to determine the range of resistance looking into the transmit ting coil an appropriate Lout value needs to be selected first. The class E transmitter requires a minimum loaded Q of 1.7879 to operate [2 0 ]. There are two factors that affect the decision of Lout. For the same loaded Q value it will be desirable to have Lout as large as possible so that the resistance looking into the transmitting coil will be larger. Therefore, the parasitic resistance of the transmitting coil can be neglected. However, if Lout is too large the parasitic resistance of the inductor will b e relatively

PAGE 55

55 large unless a better inductor of lower parasitic resistance is used. However, when the parasitic resistance of the inductor of the same inductance value gets lower the size of the inductor also gets bigger. The lowest loss inductor will be a n air core inductor using Litz wire but its size will be larger In addition, based on Equation 3 1 if the resistance looking into the transmitting coil is too large limited power will be delivered to the receiver. On the other hand, if Lout is too small t he maximum value of the resistance looking into the transmitting coil will be limited. Therefore, with a small resistance looking into the transmitting coil, the parasitic resistance of Lout and the transmitting coil will get more significant affecting the system efficiency and power delivery as most of the voltages will drop across the parasitic resistance and not across the reflected resistance by the receiver looking into the transmitting coil 2 cos2 2 V P Z tx VR tx Z tx (3 1) For an operating frequency of 240 kHz the system will be able to operate well with a Lout value from 6.8H to 22H depending on the parasitic resistance of the transmitting coil, the parasitic resistance of Lout and Cout as well as size constrain. For this design a 10 H inductor (RL 54805 10 from Renco) is selected. The inductor has low loss, 0.16 of parasitic resistance at 240 kHz and is considerable small in size (15.875 mm diameter and 17.78 mm height). However, due to de rating at higher operating frequenc y the effective inductance of the inductor is 9.5 H instead of 10 H at 240 kHz. The inductance will further decrease with higher current and temperature. This is predominately due to the temperate sensitive nature of most ferrous cores used in inductors.

PAGE 56

56 Fig ure 3 5 Peak resistance response looking into the transmitting coil with respect to Crx. Fig ure 3 6 Resistance and reactance looking into the transmitting coil. A) with receiver capacitance of 73 nF and B) with receiver capacitance of 97 nF

PAGE 57

57 Although both receiver capacitance values provide the same resistance trend looking into the transmitting coil, the reactance trend is different. Using a capacitance value of 73 nF before the resonance capacitance value of 86.4 nF results in an increasing trend of reactance with increasing load resistance converging at approximate 82 On the other hand, a capacitance value of 97 nF will result in a decreasing trend of reactance with increasing load resistance converging at approximately 51 According to Equation 3 1 increasing the reactance while keeping the resistance relatively the same will decrease the power delivery. Therefore, in order to obtain the desirable trend of decreasing power delivery with respect to increase load resistance the first solution of 73 nF before the resonance capacitor value with the receiving coil is selected. Figure 3 7 Coupling efficiency with respect to load resistance. Based on the selected receiver capacitance value, the efficiency of the coupling with respect to load resistance shown in Fig ure 3 7 can be calculated using the parasitic resistance of the coil. Coupling efficiency peaks at close to 90% at a load resistance of 30 Although efficiency rolls off to an approximate 36% at 1 k power delivered at the re sistance is

PAGE 58

58 extremely low. The gradual degradation in receiver efficiency is desirable as it helps to regulate the power during trickle charge. Power delivered by the transmitter remains consistent at high load resistances because the equivalent load imped ance Ztx looking into the transmitter load network does not change much. 3.1.3 Determination of Cout value Fig ure 3 8 Ztx phase response with respect to load resistance for various Cout capacitance value s Fig ure 3 8 shows the phase response of Ztx wi th respect to load resistance for different Cout values with increasing phase angle with increasing capacitance value. The Class E power amplifier does not perform well under capacitive loads and high efficiency is achieved at a range of phase angle from 40 to 70 [18]. Therefore, any value 7 nF and above can be used for Cout. Since the coil inductance is large, the phase response is sensitive to the component values This can be verified by the large phase response swing observed in Fig ure 3 8 when Cout is changed from 6 nF to 7 nF. Since Cout should be selected to achieve maximum power delivery and

PAGE 59

59 stability, Cout is selected to be 8 nF because the variation in phase response when Cout changes from 7 nF to 8 nF and from 8 nF to 9 nF is less than from 6 n F to 7 nF This ensures that the fluctuation in power delivery due to component tolerance is limited. However, a higher Cout can be selected to limit the power delivery as shown in Equation 3 1 3.1.4 Determination of Cshunt value Once the values of the in ductors and capacitors in the transmitter load network and the receiver network are determined, the remaining step is to determine Cshunt to achieve ZVS and ZDS operation so as to minimize switching losses. The optimum Cshunt value can be determined u sing the equations derived in Chapter 2 and [18][19] which are implemented in Matlab code. The optimum Cshunt is found to be 10 nF and the variation of transistor drain voltage versus load resistance is shown in Fig ure 3 9 It can be seen that the transis tor drain voltages are kept very close to zero when the transistor is being switched on at a phase of 180. In addition, the negative voltages do not occur because the built in diode will start to conduct and restrict the voltage at the negative of its tur n on voltage which is around 1.3V. Fig ure 3 9 Transistor drain voltage waveform for different load resistance. ( Cshunt = 10 nF).

PAGE 60

60 3.2 ParallelParallel Impedance T ransformation Network 3.2.1 Introduction Instead of 240 kHz, a n operating frequency of 134 kHz is used for this design rule Operating frequency should not affect the design rules unless it is extremely high, for example more than 5 MHz. Similar to the series -parallel impedance transformation network design rule the load resistance is defined as the equivalent resistance looking into the rectifier instead of after the rectifier. Key parameters of the coils including self inductances, mutual inductance, and parasitic resistances can be extracted by measuring the fabricated coil with an impedance analyzer or analyzing with electromagnetic simulation tools. A pair of coils was fabricated using 16 AWG magnetic wires instead of Litz wires for the experiment. The transmitting coil is 21 cm by 21 cm with 10 turns while the receiving coil is 13 cm by 13 cm with 5 turns tightly wounded together. The transmitting coil is designed with the appropriate spacing between the turns to achieve 5% variation of the received power at all different locations. Therefore, it can be reas onably assumed that the coupling is constant regardless of receiving coil position. The self inductance of the transmitting coil is 31.95 H with a parasitic resistance of 0.32 receiving coil is 12.52 H with a parasitic resistance of 0.2 inductance between the coils is 7.454 H with a coupling coefficient of 0.373. Measurement of the coils was measured using the HP4192A LF Impedance Analyzer. Since the coils are fabricated using magnetic wires instead of Litz wires, the actual parasitic re sistance of both coils during power transfer will be larger than the low voltage signal measured values. This is due to both proximity and skin depth effects. The design rule for the parallel -parallel impedance transformation network will be based on Figur e 3 10 which has an extra capacitor Ctx shunt across the transmitting coil relative to Figure 3 3

PAGE 61

61 Cshunt Cout AC Ctx Crx Rload Zrx Ztxcoil ZtxLoutLDC Transformation network Fig ure 3 10. Simplified schematic of wireless power transfer system using parallel -parallel transformation network and Class E transmitter. Ztx Impedance looking into the transmitter load network. Ztxcoil Impedance looking into the transmitting coil. Zrx Impedance looking into receiver network. RLoad is the equivalent resistance looking into the rectifier. 3.2.2 Determination of Crx value Fig ure 3 1 1 Optimum receiver capacitor value versus load resistance.

PAGE 62

62 The capacitance value is selected based on the inductance of the receiving coil as well as the mutual inductance between the coils. It is ideal to achieve a maximum resistance looking into the transmitting coil across a wide range of load resistances. However, in reality, the optimum capacitance value is different for their respective load resistances. In addition, there is no closed form analytical solution Differentiatin g the real part of Equation 2 8 and solving for the optimum receiver capacitance value is not straightforward. Therefore, a simple parameter sweep of Matlab code is used to sweep through a range of receiver resistance s and capacitance value s to extract the optimum value. Base d on the coil parameters, Figure 3 1 1 shows the optimum Crx value versus load resistance. The optimum Crx value decreases rapidly from 135 nF to 113 nF with increasing load resistance. The typical operation of a switchi ng voltage regulator used for this application does not present a very low resistance at its input. For example it can be safely concluded that to power a typical USB enabled device at 5 V, 500 mA (input resistance of 10 the regulator input resistance should not drop below 25 100% efficiency and the minimum input regulation voltage is 8 V. Since it is likely that the regulator will operate with load resistance between 25 achieve high efficiency across this impedance range. 113 nF is chosen as the preferred receiver capacitance value. Fi gure 3 12 shows the coupling efficiency between the coils and the impedance looking into the transmitting coil (Ztxcoil). The efficiency is calculated using the ratio of the power delivered to the resistance Rtxcoil over the power delivered to both the par asitic resistance and the effective resistance Rtxcoil. The above method is used to determine the transmitting and receiving coil efficienc ies The coupling efficiency is the combination of both the transmitting coil and

PAGE 63

63 receiving coil efficienc ies It can be seen that the efficiency of the transmitting coil remains high for all cases, which is desirable. This is because the transmitter puts out a large amount of power and should have a higher efficiency to mitigate heat loss. The gradual degradation in rec eiver efficiency is desirable as it helps to regulate the power during trickle charge. This can be seen in later analysis that the power delivered by the transmitter remains consistent at high load resistances. This is because the equivalent load impedance Ztx looking into the transmitter load network does not change much. Fig ure 3 1 2 Coupling efficiency and transformed impedance looking into the transmitting coil 3.2.3 Determination of Ctx value Selecting an appropriate transmitting coil parallel ca pacitor value requires to fulfill two constraints. First, the system must not fail catastrophically when the receiving coil is removed from the transmitting coil (Ztx should not be capacitive and preferably has a phase angle of greater than 40 ). Although it is possible to implement a load detection scheme as discussed in

PAGE 64

64 Chapter 7 to turn off the transmitter and reduce unloaded power losses, it is still desirable for the unloaded power consumption to be minimal to begin with B y repeatedly stres sing the components with excessive voltage and current will cause them to have a shorter life cycle. In addition, lower power consumption during unloaded condition, will further reduce power consumption during the standby mode. Limiting unloaded power los s can be achieved by ensuring the unloaded Ztx has equivalent impedance similar to the case with a high load resistance (high impedance with large phase angle). From the schematic of the C lass E circuit in Fig ure 3 10, it can be deduced that most of the po wer loss is due to the transmitting coil and inductor parasitic resistances as they are in the path of power transfer. Therefore, one way to reduce the unloaded power loss is to use an inductor with lower parasitic resistance or Litz wire for the transmitting coil For the second constraint, the reactance of the transformed impedance must have an increasing trend with respect to the load resistance in order to achieve an increasing phase response. Harmonics rejection as well as any phase shifting to bring the impedance of the transmitter load network to the appropriate range is realized by Cout and Lout. Cout and Lout are selected based on the operating frequency. Since inductors are typically larger than capacitors, it is not recommended to put more than o ne inductor on each channel. For the selected operating frequency, Lout is selected to be 100 H (50 H equivalent single channel) with a parasitic resistance of 1.3 out is selected to be 68nF. Since inductors have typically poorer tolera nce than capacitors, they can be placed next to each other so that the mutual inductance between the inductors will force the current in the inductors to be synchronized. Cout can be tuned to vary the power delivery profile after the design is completed. I f Cout is too low Ztx

PAGE 65

65 phase will be too small affecting the ZVS operation. If Cout is too high the large Ztx phase will limit power delivery. Fig ure 3 1 3 Reactance of Ztxcoil versus load resistance with different Ctx.

PAGE 66

66 Fig ure 3 1 4 Amplitude and phase of impedance of unloaded Ztx versus Ctx. From Fig ure 3 1 3 it can be determined that Ctx must be at least 60 nF to ensure proper phase response. Figure 3 1 4 shows the amplitude and phase of the unloaded Ztx with different Ctx. Since the class E power amplifier does not work well when driving capacitive load [1 8 ] and might even result in a system failure due to heating, Ctx values between 44 nF to 93 nF should be avoided. Based on both conditions, it can be concluded that Ctx s hould be above 93 nF. Although it would be ideal to have a large Ctx value so that the unloaded power loss is minimized, the variation of the load network reactance decreases, as seen in Fig ure 3 1 3 when the value of capacitance increases. If the variatio n of both resistance and reactance are small, the phase shift across the load resistance is small, resulting in little variation in the power delivery versus load resistance. If the power delivered to the receiver is not reduced to a manageable level at hi gh load resistance, power will be dissipated in the receiving coil creat ing heating problems. Based on a minimum magnitude of unload Ztx of 10 capacitance value of 10 5 nF is selected for Ctx. Fig ure 3 1 5 Rtx and Xtx versus load resistance (Ctx: 105 nF)

PAGE 67

67 Fig ure 3 1 6 Phase of Ztx v ersus load resistance. (Ctx: 105 nF) 3.2.4 Determination of Cshunt value Figure 3 17. Transistor drain voltage waveform for different load resistance. ( Cshunt = 19 nF).

PAGE 68

68 The method to determine Cshunt is the same as the design rule of the series -parallel impedance transformation network topology. The optimum Cshunt is found to be 19 nF and the variation of transistor drain voltage versus load resistance is shown in Figure 3 1 7 If the system is to be reduced to a s ingle channel topology the optimum Cshunt value will be double of 19 nF, 38 nF. On the other end the optimum Cshunt value for a triple channel topology will be 12.67 nF. The optimum value of Cshunt can be easily scaled to the desired numbe r of channels.

PAGE 69

69 CHAPTER 4 WIRELESS POWER TRANSFER SYSTEM SUPPORT ING MULTIPLE RECEIVERS 4.1 Inductive Coupling The analysis of a 1:N coupling structure for a single transmitting coil delivering power to multiple receivers extends from the analysis in Chapter 2. Since the r eceivers are intended to be integrated into portable devices, it is highly unlikely that the receivers will be overlapped. In addition due to the physical constraint of the devices they should be s ufficiently space d apart Therefore, the mutual inductance between the receiving coils can be neglected as the coupling between the receiving coils will be significantly weaker than the coupling between the transmitting coil and receiving coils. The voltage and current character istics of the transmitting coil and X number of receiving coils can be described using the following equations: 1111 1 1 X VjMIjMI NN N (4 1 ) 11 VjMIjMI NNNNN (4 2 ) 111 MkMM NNNN (4 3 ) Where V1 is the voltage at the transmitting coil (Figure 1 1) I1 is the current at the transmitting coil (Figure 1 1) VN is the voltage at N receiving coil (Figure 1 1) IN is the current at N receiving coil (Figure 1 1) M11 is the self inductance of the transmitting coil MNN is the self inductance of N receiving coil

PAGE 70

70 M1N = MN1 is the mutual inductance of the transmitting coil and Nth receiving coil kN is the coupling coefficient between the transmitting coil and Nth recei ving coil X is the total number of receivers By Ohms law: 1 1 ZRjX txtxtx V I (4 4 ) ZRjX rxNrxNrxN V N I N (4 5) Using Equation 4 1, Equation 4 2 and Equation 4 4 and assuming a time -harmonic pedance looking into the transmitting coil for multiple recei vers is derived as Equation 4 6. 22 1 2 2 1 22 1 11 2 2 1 MR X NrxN Z tx N RMX rxNNNrxN MMX X NNNrxN jM N RMX rxNNNrxN (4 6) A typical buck switching regulator requires a higher input voltage to operate and tends to amplify the load resistance. The amplification of load resistance will tend to choke the other receivers in a multiple receivers setup especially when one of the receiver s is in a high resistance/trickle charge state. The trend can be observed from Equation 4 6 and Equation 3 1 which also can be observed from Figure 4 1. In order to achieve considerable power delivery when one of the receiving devices is fully cha r ge d it needs to decouple itself from the system.

PAGE 71

71 The decoupling can be achieved by a switch either in series of the power path to break the connection or in parallel with the receiving coil to create a short across the receiving coil so that the transmitter does not see the transformed impedance due to the receiver Figure 4 1 Power d elivery to l oads for a single receive r setup and a dual receivers setup with one of the load fixed at 1000 Since the receiver will be a portable device such as cellular phone, mp3 player, etc t he receiver switch used to decouple the receiver needs to be compact and able to be driven by a low voltage, e.g. not more than 3 V. Although, most electromechanical switches are able to tolerate large voltages and currents, they are typically large for portable electronics and generate a clicking sound during switching which is not acceptable. Off the shelf solid state switches are typically designed for 50/60 Hz AC line application T hey are relatively larger in size and do not offer sufficient isolation for hundreds of kilohertz signals. It is possible to find switches which operate at high fr equencies but the power handling starts to drop with increase in frequency as shown in [30] unless novel materials are used as shown in [31] which will make the switch expensive. In addition, it is difficult to control the switch with voltages lower than t he 0 1 2 3 4 5 10 100 1000Power Delievered to Loads (W)Receiver 1 Load Resistance ( ) Power Delivered to Receiver 1 (Single RX) Power Delivered to Receiver 1 (dual RX with Receiver 2 @ 1000 ohms) Power Delivered to Receiver 2 (dual RX with Receiver 2 @ 1000 ohms)

PAGE 72

72 voltage being switched using a simple transmission gate topology or switch transistors. A new switch architecture which is an extension of a transmission gate switch is proposed to control a large voltage/current signal with a low voltage control signal. The discussion of the switch circuit in this section is independent of the decoupling architecture. 4.2 Switch Design The block diagram is shown in Figure 4 2. The proposed switch should be able to handle voltages up 25 Vrms and current up to 2 Arms with an operating frequency up to 1 MHz. In addition, the proposed design should not use any inductors so that it can be easily integrated into an IC or single package solution with the voltage regulator. Power Switch Switch control Network Switch control Network Negative voltage rectification network Positive voltage rectification network Input AC voltage Output AC voltage Control Figure 4 2 Block diagram of the proposed switch. The power switch block in Figure 4 2 uses a typical transmission gate architecture which is a NMOS and a PMOS in parallel. A schottky diode must be added to either before or after the transistor to co unter the effect of the body diode of the power MOSFET. The schottky diode selected must have power handling comparable to the body diode of the transistor. The control signals to the gate of both of the transistors are provided via their respective switch control

PAGE 73

73 network. Two rectification circuits extract the maximum voltage and minimum voltage of the input AC voltage. The maximum voltage and minimum voltage are used as an input for the respective switch control network in a cross -coupled topology. Based on the control signal provided by the receiver, the switch control network will switch between the maximum voltage and the minimum voltage to either turn the transmission gate on or off. Schematic of the proposed switch circuit is shown in Figure 4 3. A si ngle package dual N and P channel MOSFET (IRF7343) from International Rectifier is used. The transistors have an absolute Vgs of 20 V and an absolute Vds of 55 V. The peak continuous drain current of the N channel MOSFET is 4.7 A and P channel MOSFET is 3. 4 A. Turn on resistance for both transistors are typically better than 0.1 Therefore, the transistors are able to handle considerable amount of power at high efficiency. Rise time and fall time of both transistors are better than 22 ns giving it a fast response time. The input capacitor of both transistors is typically better than 750 pF which makes driving the transistor feasible The Cgd of both transistors are less than 100pF and the Cds of both transistors are less than 125 pF, reducing the leakage current at high frequency when the switch is turned off. MBRA340T3 is selected for both the rectification network diode and switch because it is able to handle voltages up to 40 V and current s up to 3 A. In addition, it has a small forward voltage drop of 0.45 V and being a schottky diode it has negligible reverse recovery time. Notation for resistors and capacitors are in the form of RX_X and CX_X. The number after the underscore is used to differentiate between the two switch control networks which are similar, namely channel 1 for the P channel MOSFET of the transmission gate and channel 2 for the N channel MOSFET of the transmission gate. Value for C1 is 100 nF, C2 is 10 nF, R1 is 10 k and R2 is 47 k

PAGE 74

74 IRF7343p IRF7343n IRF7343n MBRA340T3 MBRA340T3 C1_1 C2_1 MBRA340T3 MBRA340T3 AC in IRF7343p C1_2 C2_2 R1_2 R1_1 AC out IRF7343n R2_2 Control IRF7343p Control 0/3V R2_1 Switch control network Switch control network SwitchPositive voltage rectification network Negative voltage rectification network Figure 4 3 Schemat ic of the proposed switch circuit.

PAGE 75

75 IRF7343p IRF7343n IRF7343n C1_1 C2_1 MBRA340T3 MBRA340T3 AC in IRF7343p C1_2 C2_2 R1_2 R1_1 AC out IRF7343n R2_2 Control IRF7343p Control 0/3V R2_1 Switch control network Switch control network SwitchPositive voltage rectification network Negative voltage rectification network IRF7343n IRF7343p Figure 4 4 Schematic of the improved proposed switch circuit.

PAGE 76

76 An improved version of the proposed switch is to replace the diode in series with the transmission gate transistors with a trans istor as shown in Figure 4 4. By doing so the power signal will not suffer a forward voltage diode drop, reducing the losses through the switch. However, the switch control network must ensure that it is able to drive the extra gate loading the two rectification network s Since the diodes used ha ve a low forward drop voltage, and for simplicity, the switch will use the schematic on Figure 4 3 instead of Figure 44. 4.3 Switch Simulation Figure 4 5 Schematic of the switch in Advanced System Design with a resistive as load.

PAGE 77

77 Simulation and verifi cation of the switch is done using Advanced System Design by Agilent. The simulation schematic to analyze the performance of the switch with a 50 resistive load is shown in Figure 4 5. Transistor and diode model in the simulation are obtained from the ma nufacturer. Therefore, the simulation results should match the performance of the fabricated circuit. Figure 4 6 Switch control waveform (0 V for off and 3 V for on). A minimum of 1 V is required to turn on the transistor. Figure 4 6 shows with switch control waveform which is 0 V for off state and 3 V fo r on state at a duty cycle of 5 0% and a frequency of 100 Hz. Based on the transistor used (IRF7343), the switch is able to operate at turn on voltages as low as 1 V. Figure 4 7 shows the switch control voltage for each respective channel. The turn off response time for channel 1 is approximately 630 S and turn off response time for channel 2 is approximately 700 S. The turn on response time for channel 1 is approximately 60 S and turn on response time for channel 2 is approximately 70 S. Therefore, the switch can operate up to 1 kHz switching frequency, which is way beyond the application of the circuit which is to turn off the receiver

PAGE 78

78 when the device is fully charged. The turn on time is significant ly faster than the turn off time because the voltage across C1 and C2 are the same when the switch is turned off and when the switch transits from the off stage to on stage C2 is charged/discharged via the low resistance path through the transistor whereas when the switch transits from the on stage to the off stage C2 is charged/discharged via resistor R1 which increases the time constant significantly. Response time can be improved by decreasing R1 at the expense of power loss through the switch. Decreasin g either C1 or C2 also helps to improve the response time. However, the ripples on the rectified voltage might become significant and affect the operation of the circuit. Figure 4 7 Gener ated switch control waveform s A) C hannel 1 (P channel MOSFET of the transmission gate) and, B) C hannel 2 (N channel MOSFET of the transmission gate).

PAGE 79

79 Figure 4 8 Output waveform of the switch before the rectifier. Figure 4 8 shows the output AC waveform of the switch across the 50 resistive load. The switch is capable of breaking the high voltage/current AC signal path with reasonable turn on and turn off time with the time off time slightly slower than the turn on time. The reason for the difference in timing is the same reason fo r those of the switch control network waveform. 4.4 System Response with Receiver Switch Figure 4 9 Proposed half -wave rectifier receiver architecture. Figure 4 10 shows the proposed receiver architecture. A half wave rectifier is used instead of a ful l wave rectifier because the circuit will be more compact using a single diode instead of four. Another benefit is that the rectification process only suffers a single diode forward voltage

PAGE 80

80 drop instead of two. In addition, the operating frequency will be in the range of hundreds of kilohertz, thus it will not require a very large charge holding capacitor. Instead of using the switch in the power path, the switch is placed shunt across the coil. There are two reasons in using such a topology. Firstly, by shorting the receiving coil, the receiver coil sees a short. Therefore, Rrx and Xrx in E quation 2 7 will be extremely small that their value can be assumed to be zero Let Xrx in E quation 2 7 be zero. 22 32 12 1222 11 22 22 22 22 MR MM rx Z jM tx RMRM rx rx (4 7 ) Next let Rrx in E quation 4 7 to be zero. 2 12 11 22 M ZjM tx M (4 8 ) Substituting E quation 2 3 into E quation 4 8 2 1111 ZjMkM tx (4 9 ) In order for the system to be able to support multiple devices and provide sufficient lateral freedom, it is reasonable for the coupling coefficient to be much less than 0.25. By assuming the coupling coefficient to be 0.25, k2 will be 0.0625 which is much lower than 1. Therefore, if the receiving coils are shorted under loosely coupled condition, the transmitter only sees the self inductance of the transmitting coil. However, if the coupling coefficient between the coils starts to increase there will be a reduction in self inductance of the transmitting coil. This will cause Ztx to be less inductive and potentially causing the class E power amplifier to be no operating under ZVS/ZDS condition.

PAGE 81

81 The second reason is that the switchs natural state is open whe n no voltage is applied to the control port. Therefore, the receiver will allow power to pass through when the control port is left floating. This is critical especially when the battery on the receiver is full y drained and is unable to control the switch. By using the proposed architecture in Figure 4 -9 the receiver does not require any bootstrapping even when there is no power, because the switch will be in a naturally on state. Figure 4 1 0 shows the test bench to verify the performance of the switch whe n used in this specific scenario. A similar clock control source is used as well. The switch circuit in Figure 4 5 is converted to the modular block X1. Instead of using a full Class E driver, the transmitter is simplified as an AC current source with 1A p eak current at 240kHz. The transmitter coil is modeled as inductor L1 and receiving coil as inductor L2, mutual inductance is modeled via Mutual1 block. Figure 4 1 0 ADS schematic of test bench for receiver architecture in Figure 5 10.

PAGE 82

82 Fig ure 4 1 1 Simulation results of test bench (Fig ure 4 1 0 ). R ed: Low voltage control signal. Black: Receiver rectified voltage Fig ure 4 1 1 shows the simulation results of the test bench of Fig ure 4 1 0 with the control voltage in red and output rectified voltage in black As predicted in switch standalone simulations the switch closing time is faster than the opening time, resulting in a faster decoupling response time. Although the receiver is suppose d to be fully decoupled when the switch closes, a 0.5V DC voltage still can be observed at the load. This is because the switch is not an ideal switch and a potential drop will be observed across the parasitic resistance of the transmission gate transistor as well as the series diodes used to counter the effect of the b ody diode. The 0.5V DC voltage should not be a concern because it is not sufficient to turn on the voltage regulator which is typically used to provide a stable DC voltage. From the simulation results, it can be concluded that the switch circuit can be use d in scenarios where a control voltage is significantly smaller than the input AC voltage at high frequencies regardless of topologies.

PAGE 83

83 CHAPTER 5 EXPERIMENTAL VERIFIC ATION 5.1 High P ower 300W Dual Channel System using Parallel -Parallel Impedance Transfo rmation Network Topology Base d on the parallel -parallel impedance transformation network topology design rule a dual -channel Class E transmitter test system capable of delivering nearly 300 W with a supply voltage of 120 V is fabricated using the IRFP21N60L HEXFET power MOSFET from International Rectifier. A full wave rectifier with a shunt charge holding capacitor at the output using MUR420 from Vishay is fabricated to convert the AC power to DC power. Since the forward voltage drop is 0.875 V and the re verse recovery is 30 ns, power loss due to the voltage drop and reverse recovery is small compared to the amount of power delivered to the load. Load resistance in this section is the equivalent resistance looking into the regulator or device being charged /powered as shown in Fig ure 1 1 instead of the equivalent resistance looking into the rectifier as shown in Figure 3 3 In order to reduce losses through parasitic resistance, low loss polypropylene capacitors are used. T o achieve a balance between size a nd efficiency, a 100 H inductor ( 1140101K RC) by Bourns Jw Miller is selected for Lout. Since most of the loss of the transmitter is due to the parasitic resistance of Lout, a larger and more efficient inductor can be used if space permits. Table 5 1 shows the calculated value of each component with respect to the actual component value used in the experimental setup. The calculated values obtain from Chapter 3 are initially used and further tuned to achieve optimum power delivery and efficiency acros s a wide range of load resistance. Most of the values follow closely to the calculated value from Matlab. The only exception is Cshunt. The main reason for such a discrepancy is that the equation used in [ 18] [19] assumes the transistor to be an ideal swit ch. Therefore, while calculating the drain voltage, the built in diode in the transistor was not taken into account. The other parameters of the transistor

PAGE 84

84 do not have any significant effect on the calculated values since the rise and fall times of the tra nsistor are significantly faster than the switching time and the drain to -source capacitance is less than 1 nF. In addition, the turn on resistance of the transistor is extremely small. Another reason is that during the calculation, the DC feed inductor ( LDC) is assumed to be infinitely large, which is not true in experiment. Table 5 1 Component Values for High power 300W Dual Channel System using Parallel Parallel Impedance Transformation Network Topology Calculated Experimental % Variation C rx 113 nF 115 nF +1.8% Ct x 105 nF 100 nF 4.8% C out 68 nF 68 nF 0% L out 100 H 100 H 0% L out parasitic resistance 1.3 C shunt 19 nF 15 nF 21% TX coil inductance 31.95 H TX coil dimension 21 cm x 21 cm TX coil parasitic resistance 0.32 RX coil Inductance 12.52 H RX coil dimension 13cm x 13 cm RX coil parasitic resistance 0.20 Mutual Inductance 7.475 H Coupling Coefficient 0.374 L DC 500 H The fabricated dual -channel transmitter with a dimension of 10 cm x 8.5 cm is shown in Fig ure 5 1 The two inductors of Lout occupy a significant amount of space due to the requirement of low parasitic resistance so as to maintain high efficiency in power delivery. Fig ure 5 2 shows the transmitting coil (21 cm x 21 cm) embedded inside a table top case, and the receiving coil (13 cm x 13 cm) placed on top of the case. The actual separation between the two coils is 10 mm. The setup enables the user to have a large degree of translational freedom both in the X and Y direction It should be noted that there is no ferrite core in either the transmitting coil or receiving coil

PAGE 85

85 Figure 5 1 Photograph of the dual -channel Class E power amplifier. Fig ure 5 2 Photograph of the tra nsmitting coil 10 turns (embedded into the table top) and receiving coil 5 turns (placed on top)

PAGE 86

86 Using a 120V power supply, power delivery of 295 W to a 50 121.5 V and current of 2.43 A is achieved. The input current from the power supply is 3.25 A. The end -to -end system efficiency is 75.7%. Peak drain voltage is 460 V which is 25% lower than the rated maximum voltage of the transistor. Fig ure 5 3 shows the efficiency and power delivery of the 120 V system versus load resis tance. Although the maximum power of 295 W occurs at 50 The power delivered is proportional t o the square of the supply voltage. The power delivery of the system can be increased by increasing the supply voltage as long as the DC power supply driving system is able to provide sufficient power and the drain voltage across the transistor stays withi n its breakdown voltage. Fig ure 5 3 Power delivery (left yaxis) and efficiency (right yaxis) of the system versus load resistance. Supply voltage: 120 V. 20 30 40 50 60 70 80 0 50 100 150 200 250 300 0 250 500 750 1000 1250 1500 1750 2000System Efficiency (%) Power Delivered to Load (W)Load Resistance ( ) Power Delivered to Load (W) System Efficiency (%)

PAGE 87

87 Figure 5 4 Transistor and inductor temperature with natural convection cooling and forced air cooling versus supply voltage. The transmitter is able to operate under natural convection cooling with up to 95 W of power delivery to the load using a supply voltage of 65 V. Power levels above 95 W require s forced air cooling on the transmitter to keep the temperat ur e of the transistors below 70 C. Although most of the loss due to heat occurs at the transistor or inductor, a temperature increase at the inductor is not critical because it is a passive component and more resilient to heat. However, t his will cause a drop in inductance value which will decrease the load network phase angle increasing the power delivery slightly. The slight increase in power delivery should be taken care by the receivers regulator. Temperatures of both the transistor a nd inductor for natural convection cooling and forced air cooling were measured using the Fluke 62 mini 20 30 40 50 60 70 80 90 100 110 120 130 140 20 30 40 50 60 70 80 90 100 110 120Temperature ( C)Supply Voltage (V) Transistor Temperature (natural convection cooling) Inductor Temperature (natural convection cooling) Transistor Temperature (forced cooling) Inductor Temperature (forced cooling)

PAGE 88

88 infrared thermometer at extremely close proximity is shown in Figure 5 4 A 12 V DC dual ball bearing fan of size 90 mm x 90 mm x 25 mm with a speed of 2700 rpm achieving an air flow of 44 cfm is used for forced air cooling (shown in Fig ure 5 5 ). Since the power consumption of the fan is only 2 .4 W, it can be neglected for the efficiency calculations. In addition, the heat sinks on the transistors further enhance the heat dissipation capabilities. As seen in Figure 5 4 the transistor temperature reaches a practical limit of 75 C for a supply voltage of 120 V. Although higher power can be achieved by increasing the supply voltage using a higher power output power supply, more effective heat dissipation methods are required to prevent the transistors from overheating. A compact system operating without forced air cooling is often preferred. Therefore, the following measurements based on a supply voltage of 60 V operating under natural convection cooling are presented. Figure 5 6 shows the transmitting coil current and voltage when it is driving a 50 can be seen that the current is lagging the voltage. As seen from the drain waveform, ZVS operation is achieved. Fig ure 5 5 Photograph of the dual -channel Class E power amplifier with forced air cooling. Air Flow

PAGE 89

89 Fig ure 5 6 Voltage and current waveforms of the Class E transmitter. Figure 5 7 Power delivered to load versus load resistance. A maximum power of 69 W occurs approximately at 50 -channel and maximum power of 10 W occurs at approximately at 75 -channel. Supply voltage: 60 V. 0 10 20 30 40 50 60 70 0 250 500 750 1000 1250 1500 1750 2000Power Delivered to Load (W)Load Resistance ( ) Both channels enabled Single channel enabled

PAGE 90

90 Fig ure 5 7 shows the power delivery versus load resistance of both dual -channel (solid line) and single -channel (dotted line) modes. Power delivery for both modes peak at about 50 load resistance the dual channel mode peak power i s close to 70 W while the single channel mode is at 10 W. The system efficiency versus load resistance is shown in Fig ure 5 9 which verifies that the resistance looking into the network of the transmitting coil in parallel with the shunt capacitor decreases with increasing load resistance. This results in a larger voltage drop across the parasitic resistance of the inductor and a lower efficiency. Fig ure 5 8 Mode -switching operation for optimized efficiency across a wide power delive ry range. (1) Dual -channel mode for higher power, (2) Dual -channel mode switch -over to single -channel mode when better efficiency can be obtained at a lower power level, (3) Single -channel mode for lower power, (4) Single -channel mode switch over to dual -c hannel mode when higher power delivery is needed. Supply voltage: 60 V. 0 5 10 15 20 25 30 0 100 200 300 400 500 600Power Delivered to Load (W)Load Resistance ( ) Both channels enabled Single channel enabled 1 2 3 4

PAGE 91

91 Fig ure 5 9 System efficiency versus load resistance with a maximum efficiency of 64.5% for a single -channel system, and 76% for a dual channel system at approximately 70 Supply voltage: 60 V. The efficiency of the single -channel mode is approximately 10 15% lower than the efficiency of the dual -channel mode for the same load resistance because the current is flowing through a single Lout inductor instead of a pair of the m. T he parasitic resistance is doubled, thus resulting in low er system and transmitter efficiencies as shown in Fig ure 5 9 and Figure 5 10 respectively. However, when the system enters light load mode or trickle charge mode, it is desirable to switch to th e single -channel mode. It can be seen from Fig ure 5 -11 that the system efficiency is approximately 15% higher than the dual -channel mode for delivering the same amount of power below 10 W. Instead of operating at high load resistance for a dual channel mod e resulting in high receiver DC voltage as shown in Fig ure 5 13, it is possible to achieve similar power delivery at much lower load resistance for a single -channel mode, resulting in a lower receiver DC voltage and higher system efficiency. In addition, a typical buck regulator has 10 20 30 40 50 60 70 80 0 250 500 750 1000 1250 1500 1750 2000System Efficiency (%)Load Resistance ( ) Both channels enabled Single channel enabled

PAGE 92

92 higher DC DC efficiency when the input voltage is lower. Therefore, a load detection scheme is required to determine the switch -over point from dual -channel mode to single -channel mode. It can be seen from Fig ure 5 7 that a powe r delivery of 10 W occurs at 500 dual -channel mode making it a good switch-over point to single -channel mode ((2) in Fig ure 5 8 ). It can be seen that a 500 across t ransmitting coil for the dual -channel mode as shown in Figure 5 12. Likewise, if the power requirement for the single -channel mode is too high, it is required to switch to the dual channel mode. It can be inferred from Fig ure 5 7 that the switch -over point to dual -channel mode is approximately 75 ure 5 8 ) where the efficiency peaks in single -channel mode, which translates to a RMS coil voltage of 22 V in Figure 5 12. Fig ure 5 10. Transmitter efficiency versus load resistance. Maximum transmitter efficiency occurs across the range of 60 -channel and 79% for single -channel. Supply voltage: 60 V. 60 65 70 75 80 85 90 95 0 250 500 750 1000 1250 1500 1750 2000Transmitter Efficiency (%)Load Resistance ( ) Both channels enabled Single channel enabled

PAGE 93

93 Fig ure 5 11. System efficiency versus load resistance for single -channel and dual -channel modes achieving high efficiency at high power output. It also illustrates that a single -channel mode is more efficient at low power delivery states. Supply voltage: 60 V. Fig ure 5 12. Transmitting coil RMS voltage versus load resistance. Supply voltage: 60 V. 15 20 25 30 35 40 45 50 55 60 65 70 75 80 0 10 20 30 40 50 60 70System Efficiency (%)Power Delivered to Load (W) Both channels enabled Single channel enabled 10 20 30 40 50 60 70 80 90 100 0 250 500 750 1000 1250 1500 1750 2000Transmitting coil RMS Voltage (V)Load Resistance ( ) Both channels enabled Single channel enabled

PAGE 94

94 Fig ure 5 13 shows the receiver DC voltage versus load resistance for both the dual channel mode and the single -channel mode It can be seen that there is no over voltage issue as the voltage starts to converge to a value of approximately 70 V when the loa d resistance is high. The receiver open -circuit voltage is 73.4 V and open-circuit power consumption is 10 W in the dual channel mode. However, during single -channel operation, the receiver opencircuit voltage is only 38.3 V and its open-circuit power consumption is only 4 W which is 6 W less than that of the dual channel mode. The refore, it is preferred to perform load detection using the single channel mode and increase the output power by enabling the dual -channel mode if more power is required to reduce standby power consumption By controlling the maximum receiver DC voltage, the requirement for the receivers regulator is relaxed enabling the system designer a wider range of selection Fig ure 5 13. Receiver DC voltage versus load resistanc e, converging to approximately 70 V in dual -channel mode and 3 8 V in single -channel mode. Supply voltage: 60 V. 5 15 25 35 45 55 65 75 0 250 500 750 1000 1250 1500 1750 2000Receiver DC Voltage (V)Load Resistance ( ) Both channels enabled Single channel enabled

PAGE 95

95 5.2 Low Power Multiple Receivers System with decoupling Switch using Series -Parallel Impedance Transformation Network Topology The Class E tran smitter test system operating at 240 kHz was fabricated using the IRLR/U3410 power MOSFET. A 13 turn 20 cm x 20 cm transmitting coil which was described in the design rules and two 6-turn 9 cm x 6 cm receiving coils were used. The platform is capable of si multaneously charging up to four independent devices. Although it is possible to extend the experiment to more than two receivers, for better understanding of the system response and easier analysis, experiments of charging two receivers were conducted. A Matlab code was also written based on the equations derived in [ 18] [19 ] to study the efficiency and power delivery. All measurements and simulation results are based on a 12V power supply. The 12V power supply is selected because the supply voltage is rea dily available from the DC supply plugs in vehicles and several other AC DC converters. Table 5 2 shows the value of each component used in the experiment. Component values are selected by matching the closest available component value and further tuned t o achieve optimum performance. Crx is selected to be 75 nF. Since the switch contributes 3.5 nF of capacitance and the rectifier contributes another 3.5 nF of capacitance, a 68 nF capacitor is used to achieve an effective capacitance of 75nF. Capacitance contributed to the receiver due to the switch and rectifier is measured using the HP4192A LF Impedance Analyzer in small signal operation. Therefore, the actual capacitance under large signal operation will be slightly different depending on the load conditions and voltage at the switch and rectifier. In order to reduce losses and heat through the parasitic resistance of the capacitors low loss polypropylene capacitors are used. Alternatively, C0G/NP0 surface mount capacitors can be used as well. However, they are physically small and tend to heat up faster than the larger leaded polypropylene capacitors making them not suitable for high power applications.

PAGE 96

96 Table 5 2 Component V alues for Low Power Multiple Receivers System with decoupling Switch using Series -Parallel Impedance Transformation Network Topology Calculated Experimental % Variation C rx 73nF 75nF +2.7% C out 8nF 9.4nF +17.5% L out 9.5H 9.5H 0% L out parasitic resistance 0.16 C shunt 10nF 12nF +20% TX coil inductance 45.3H TX coil dimension 20cm x 20cm TX coil parasitic resistance 0.5 RX coil Inductance 5.2 H RX coil dimension 9cm x 6cm RX coil parasitic resistance 0.1 Mutual Inductance 2.8 H Coupling Coefficient 0.182 L DC 500 H Fig ure 5 1 4 Photograph of a test setup with two receivers with decoupling switch on the packaged transmitting coil.

PAGE 97

97 Figure 5 1 4 shows the photograph of the test setup with two receivers on the packaged transmitting coil. The vertical separation between the transmitting coil and receiving coils is about 2 mm. The locations of receivers on the transmitting coil were fixed by the blue tapes to ensure the same conditions for all measurements. Figure 5 1 5 Power delivery to the receiver with switch and system efficiency versus load resistance for a one -to -one setup (simulated and measured) Comparison of power delivery to the receiver and system efficiency versus load resistance of a one -to -one setup between the simulated results and the experimental results is shown in Fig ure 5 1 5 Measurements are conducted by connecting the receiver to a rheostat for which the resistance is varied manually in predetermined steps. The simulation and measured trend of a single receiver setup agree well with peak measured power at around 4.75 W. The re is a slight discrepancy between the measured and simulated efficiencies because the transistor and DC feed inductor are assumed to be ideal in the simulation model. In addition, the effect of the build in 0 1 2 3 4 5 6 0 10 20 30 40 50 60 70 80 0 250 500 750 1000Power Delivered to Load (W) Efficiency (%)Load Resistance ( ) Efficiency (measured) Efficiency (simulated) Power Delievered (measured) Power Delievered (simulated)

PAGE 98

98 body diode of the transistor is also not taken i nto consideration. The assumption affects the calculated supply current and the calculated efficiency. Power delivery is not affected because it depends on Ztx rather than the transistor during nominal operation. A comprehensive plot presenting the perform ance of the system in Fig ure 5 1 6 shows the system efficiency versus power delivery. Fig ure 5 1 6 Efficiency of power delivery to the receiver with switch versus power delivered for a 1 to 1 setup (simulated and measured) Fig ure 5 1 7 compares the performance of the receiver with and without the switch with p ower delivery capabilities of both receivers are nearly the same. They peak at around 4.75 W with the efficiency of the receiver with switch slightly lowered by 1 to 2%. The impact of the switch in the receiving mode is minimal because it is in shunt with the receiver. In addition, it can be concluded that the leakage through the switch is negligible. Fig ure 5 18 compares the performance of a single receiver with the decoupling swi tch architecture and dual receiver setup with one of the receivers decoupled from the transmitter. The efficiency degrades by an average 10 20 30 40 50 60 70 80 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0Efficiency (%)Power Delivered to Load (W) Efficiency (measured) Efficiency (simulated)

PAGE 99

99 of 5% and no more than 10% overall even though the second receiver is turned off. Although the receiver is decoupled fr om the system, the switch circuitry still has some turn on resistance when it attempts to short the receiving coil. Therefore, a small amount of power is still dissipated across the switch. Fig ure 5 1 7 Comparison between receiver with switch and recei ver without switch Fig ure 5 1 8 Comparison between single receiver standalone and dual -receiver setup with one receiver switched off. 10 20 30 40 50 60 70 80 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5Efficiency (%)Power Delivered to Load (W) Receiver without Switch Receiver with switch 10 20 30 40 50 60 70 80 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0Efficiency (%)Power Delivered to Load (W) Receiver 2 off Receiver 1 standalone

PAGE 100

100 To study the power delivery of a dual receiver platform, the load resistance of one of the receivers (receiver 2) is fixed while the load resistance of the other receiver (receiver 1) is swept across the range of 10 discrete steps (10 15 an experiment using the same method for the single receiver measurement. Fig ure 5 19 shows the power delivery to the receiver 1 versus its load resistance at different receiver 2 load resistance values, while Fig ure 5 2 0 shows the power delivery to receiver 2 versus receiver 1 load resistance at different fixed receiver 2 load resistance values. When the load resistance of receiver 2 is kept above 5 0 va riation of power delivery to receiver 1 due to receiver 2 is limited but peak power drops to around 2.5 W. In addition, power delivery to receiver 2 also stays consistent regardless of the load resistance or power delivery to receiver 1, as long as the lo ad resistance of receiver 1 stays abo ve 5 0 the minimum load resistance should be greater than 5 0 However, this will result in a reduction of maximum power delivered to the load by the system. The dependency of the receivers is due to the collective impedance looking into the transmitting coil due to the multiple receivers and not the mutual inductance between the receiving coils. The minimum load resistance can be de signed by selecting an appropriate receiver regulator and setting the appropriate power delivery profile by changing Ctx, or the supply voltage. This will set the unregulated input voltage before the regulator to achieve the specified load resistance looki ng into the regulator while at its maximum power delivery condition so that the load resistance will not fall below minimum value mitigating the power choking effect The experimental verification and analysis is limited to two receivers but similar trends are expected for multiple receivers of three or more.

PAGE 101

101 Fig ure 5 19. Power delivery to receiver 1 versus its load resistance at different fixed receiver 2 load resistance. Figure 5 2 0 Power delivery to receiver 2 versus receiver 1 load resistance at different fixed receiver 2 load resistance (same legends as in Figure 5 19) 0 1 2 3 4 5 6 10 100 1000Power Delievered to Receiver 1 (W)Receiver 1 Load Resistance ( ) Receiver 2 off Receiver 2 @ 10 ohm Receiver 2 @ 15 ohm Receiver 2 @ 20 ohm Receiver 2 @ 25 ohm Receiver 2 @ 30 ohm Receiver 2 @ 40 ohm Receiver 2 @ 50 ohm Receiver 2 @ 75 ohm Receiver 2 @ 100 ohm Receiver 2 @ 150 ohm Receiver 2 @ 200 ohm Receiver 2 @ 250 ohm Receiver 2 @ 500 ohm Receiver 2 @ 1000 ohm Receiver 2 @ 2000 ohm Receiver 2 OC Receiver 1 standalone 0 1 2 3 4 5 10 100 1000Power Delievered to Receiver 2 (W)Receiver 1 Load Resistance ( )

PAGE 102

102 Once the fully charged receiver (receiver 2) is decoupled from the system using the switch circuit, power delivery to the other receiver (receiver 1) increases significantly. Th erefore, the switch circuit can be used to prevent the receiver that is fully charged to choke the other receiver of the power it requires This will mitigate the effect of reduced charge rate for the receiving devices so that the system will be able to deliver sufficient power to the receiver. Figure 5 2 1 shows the system efficiency versus total power delivered to the loads with receiver 2 fixed at a specific load resistance while sweeping the resistance of receiver 1 from 10 y is above 55% for power delivery above 2 W. Although the efficiency starts to degrade significantly at lower power delivery, the absolute system power loss is low. Therefore, no heating issues were observed during the experiment. All components were opera ting below 36 C. Fig ure 5 2 1 System efficiency versus total power delivery to both receivers with the load resistance of receiver 2 fixed and the load resistance of receiver 1 swept across the stated range for a dual receiver test bench. (same legends as in Fig ure 5 19) 5 15 25 35 45 55 65 75 85 0 1 2 3 4 5 6 7 8Efficiency (%)Power delivered to Loads (W)

PAGE 103

103 Fig ure 5 2 2 shows the measured power delivery space of receiver 1 and receiver 2. Under all loading conditions the guaranteed power delivery is approximately 2W. The guaranteed power delivery can also be observed in Fig ure 5 19 and Figu re 5 2 0 in another form as discussed previously. Therefore, the system is capable of delivering 2W of power under all conditions which is close to the specified power delivery of 2.5W in the design. Higher power delivery can be achieved by reducing the capacitance of Ctx slightly or by increasing the supply voltage. Care must be taken during reduction of Ctx to keep the phase of Ztx within t he high efficiency/low loss operation above 40 Fig ure 5 2 2 Measured power delivery to a dual receiver system with load resistance varied from 10 0 1 2 3 4 5 0 1 2 3 4 5 6Power Delievered to Receiver 2 (W)Power Delievered to Receiver 1 (W)

PAGE 104

104 CHAPTER 6 INTEROPERABILITY BET WEEN DIFFERENT PLATF ORMS (COIL SIZES) The design rules of the wireless power system are presented in Chapter 3 They are further verified with experimental results in Chapter 5. However, the presented design rules and results only verify the operation of a specific one to one device pairing be tween the transmitting coil and receiving coil This does not include operation between another pair of coils of different sizes and might result in a cross platform interoperability issue It might not be a critical problem if power delivery is insufficie nt but it will be a major problem if it results in damaging the receiver. For a practical system, it is expected that the same transmitter for a certain cell phone /device to be able to work for all other different cell phones /device which might have a different receiving coil size due to its physical dimensions. In addition, it will be de sirable if the cell phones transmitter is be able to transfer power to a small er device which requires less power such as an mp3 player or a Bluetooth headset As companies start to expand their product line s or try to differentiate from one another, there will be transmitters with different sized transmitting coils. Therefore, it is critical to ensure there is some level of interoperability between the differen t platforms. Without interoperability between the different platforms, the consumer will be forced to purchase a different transmitter for each of the product he or she owns, not achieving the purpose of a universal charging platform to reduce power consum ption and wastage. Regardless of providing sufficient power to drive the receiver the transmitter should not damage the receiver via overvoltage at the input of the receivers voltage regulator. The interoperability can be further enhanced by creating a w ireless power standards body to regulate the different systems. A wireless power consortium is being set up by Phillips in December 2008 but ha s yet to gain sufficient momentum to ensure some form of interoperability between different products and platform s

PAGE 105

105 6.1 Test Bench Setup The interoperability study is performed by using three different sets of transmitting and receiving coils with components selected using the series parallel topology design rules in Chapter 3 Transmitter and receiver units are then interchanged between the respective platform without changing any component or automatic tuning to study the interoperability of the system. Table 6 1 shows the siz e s of the different pairs of coils and their respective components values, for consistency a Lout of 10 H is used for all the platforms. Crx being in shunt with the receiving coil gets smaller with the increase of receiving coil size as the self inductance of the receiving coil increases. Cout which is in series with the transmitting coil gets smaller when the transmitting coil increase s in size as the self inductance of the transmitting coil increases. This is because more negative reactance is required to offset the increase in self inductance of the transmitting coil. Since the Ztx for all 3 cases are designed to be relatively close to each other, Cshunt is kept the same for all three different platforms. Table 6 1 Specification of the three different platforms Value Small platform Transmitting coil size 11 cm x 8 cm (88 cm 2 ) Receiving coil size 5 cm x 4 cm (20 cm 2 ) C rx 103.3 nF C out 27 nF C shunt 15 nF Medium platform Transmitting coil size 16 cm x 16 cm (256 cm 2 ) Receiving coil size 6 cm x 5.5 cm (33 cm 2 ) C rx 94.7 nF C out 10 nF C shunt 15 nF Big platform Transmitting coil size 20 cm x 20 cm ( 400 cm 2 ) Receiving coil size 9 cm x 6 cm (54 cm 2 ) C rx 68 nF C out 9 nF C shunt 15 nF

PAGE 106

106 Table 6 2 Coupling parameters of nine possible combinations with first three as intended pairs TX Size RX Size L1 (H) L2 (H) M (H) K Size Ratio Small Small 10.96 3.70 2.21 0.347 4 Medium Medium 40.53 4.10 2.28 0.176 8 Big Big 46.00 5.45 2.81 0.178 7 Small Medium 10.96 4.10 3.05 0.455 3 Small Big 10.96 5.45 4.66 0.603 2 Medium Small 40.53 3.70 1.65 0.135 13 Medium Big 40.53 5.45 3.45 0.232 5 Big Small 46.00 3.70 1.37 0.105 20 Big Medium 46.00 4.10 1.88 0.137 12 With three different platforms, it is possible to come up with nine possible combinations as shown in Table 6 2 The original pairs of the transmitting and receiving coils for the three platforms designed using the design rules in Chapter 3 are shown as the first three rows From E quation 2 40, it can be inferred that the self inductance of the receiving coil is co mpensated by Crx and has little effect on the coupling between the coils when they are swapped between each other. By looking at the real part of E quation 2 6 it is independent of the transmitting coils self inductance. Since most of the effects of the r eceiving coils self inductance is being compensated by Crx the power transfer is more dependent on the mutual inductance between the pair of transmitting and receiving coil s used. Similar analysis can be used for the imaginary part of Ztx as Cout is used to cancel out most of the effects of self inductance of the transmitting coil O ther than the setup with a small transmitting coil and a big transmitting coil, the mutual inductances between the transmitting and receiving coil are kept to around 2 H. Once the appropriate capacitor and inductor values are selected via the design rule presented in Chapter 3 t he Class E transmitter is rather robust under most loading conditions N o control mechanism is require d to achieve correct power deliver y tren d and high efficiency. T he variation of mutual inductance should be acceptable to still maintain relatively high efficiency power transfer with the correct power delivery trend with respect to load resistance.

PAGE 107

107 6.2 Experimental Verification Using the nine d ifferent combinations of the three platforms on a one transmitter to one receiver setup, nine different data sets were collected. The supply voltage is selected to be 12 V but the analysis should be independent of the supply voltage. For this experiment, t he setups should be able to deliver at least 5 W of power and a receiver voltage of not more than 25 V and not less than 8 V for nominal operation of the voltage regulator to achieve a 5 V regulated output The results are described using the notation of t ransmitting coil size as the first word and the receiving coil size as the second word. Therefore, it means that a small sized receiving coil is being placed on top of a medium sized transmitting coil for the notation medium small. Figure 6 1 System efficiency and receiver voltage vs. power delivery plot for different sized receiving coil on a small sized transmitting coil. 13 14 15 16 17 18 19 20 21 0 10 20 30 40 50 60 70 800 2 4 6 8 10Unregulated Receiver Voltage (V) System Efficiency (%)Power Delievered to Load (W) Small Small (Efficiency) Small Medium (Efficiency) Small Big (Efficiency) Small Small (Receiver Voltage) Small Medium (Receiver Voltage) Small Big (Receiver Voltage)

PAGE 108

108 The results are presented in three different graphs each for a different transmitting coil size. All results using the small rec eiving coil are presented in red, medium receiving coil are presented in blue and big receiving coil are presented in green. Dashed lines are used for the unregulated receiver voltage (right axis) while solid lines are used for system efficiency excluding the voltage regulator (left axis). Figure 6 1 shows the experimental results of three different sized coils being placed on a transmitter with a small transmitting coil. When a larger receiving coil is placed on top of the small transmitting coil the real part of Ztx becomes larger, reducing the losses through the transmitting coils parasitic resistance and Louts parasitic resistance. Therefore, the setup for small medium is more efficient than small small. However, when the real part of Ztx becomes t oo large due to high mutual inductance between the coils ( 4 66 H for small big setup), the phase angle of Ztx becomes too low that the class E transmitter Therefore, the class E transmitter is working at the borderline case of ZVS/ZDS with increase d lo sses at the transmitter across the switching transistor This occurs when the green efficiency line starts to deviate from the blue line with increase in power delivery (decreasing in phase angle) at approximately 1 W power level. However, the efficiency i s still acceptable and no significant heating issue is observed The r eceiver voltage under a specific loading condition is kept approximately the same for all three cases with the maximum unregulated receiver voltage at slightly less than 21 V and the min imum unregulated receiver voltage above 13 V no over -voltage was observed. The lower unregulated receiver voltage for the small big setup is due to the same reason for the lower efficiency phenomena as more power is being dissipated at the transmitter t han transmitted to the receiver as useful power All three setup s are able to achieve at least the stated 5 W of power delivery to the receiver.

PAGE 109

109 Figure 6 2 System efficiency and receiver voltage vs. power delivery plot for different sized receiving coil on a medium sized transmitting coil. Figure 6 2 shows the same experimental results as Figure 6 1 but with the transmitting coil replaced with the medium sized transmitting coil. Based on the previous observation with small medium t he trend of increasing efficiency of medium big is expected. W ith the decrease in mutual inductance between the coils, the real part of Ztx is reduced for this setup resulting in more energy being dissipated across the parasitic resistance of Lout and the transmitting co il. Therefore, the trend of reduction on efficiency is also expected for the medium small setup. In addition due to the same reasons less power is expected to be transferred to the receiver. All three setup s are able to achieve at least 5 W of power deli very with medium small slightly above 5 W. The unregulated receiver voltage is still kept within the nominal operating range of 8 V to 25 V. 10 12 14 16 18 20 22 24 10 20 30 40 50 60 70 800 1 2 3 4 5 6 7 8 9 10Unregulated Receiver Voltage (V) System Efficiency (%)Power Delievered to Load (W) Medium Small (Efficiency) Medium Medium (Efficiency) Medium Big (Efficiency) Medium Small (Receiver Voltage) Medium Medium (Receiver Voltage) Medium Big (Receiver Voltage)

PAGE 110

110 Figure 6 3 System efficiency and receiver voltage vs. power delivery plot for different sized receiving coil on a big sized transmitting coil. Following the same approach, the medium sized transmitting coil is replaced with a big sized transmitting coil and results are show in Figure 63 The efficiency and power delivery degrades for both the smaller receiving c oils. The coupling is so weak for the big small setup that very little power (up to 2 W) is being transmitted to the receiver. In addition, the unregulated receiver voltage drops below 8 V which is too low to achieve voltage regulation for a 5 V output. Since eight out of nine of the combinations except big small meet the specification of at least 5 W of the power delivery and unregulated receiver voltage range of 8 V to 25 V we can infer that a desirable range of size ratio between the receiving coil and transmitting coil is approximately 1:2 to 1:12. This also translates to a coupling coefficient of 0.137 to 0.603. 6 8 10 12 14 16 18 20 10 20 30 40 50 60 70 800 1 2 3 4 5 6 7Unregulated Receiver Voltage (V) Efficiency (%)Power Delievered to Load (W) Big Small (Efficiency) Big Medium (Efficiency) Big Big (Efficiency) Big Small (Receiver Voltage) Big Medium (Receiver Voltage) Big Big (Receiver Voltage)

PAGE 111

111 CHAPTER 7 LOAD/FAULT DETECTION AND POWER DELIVERY T RACKING The proposed near -field wireless power transfer system is sensitive to nearby conductive or magnetic objects as both the mutual inductance and the self inductances of the transmitting and receiving coils will be affected Although it is possible to shield the transmitting coil from interferences behind or beneath it, the shield doe s not prevent a user from potentially damaging the system by placing objects such as a metal sheet on the transmitting coil or simply flipping the transmitting coil over a metal table. Therefore, to ensure robust operation of the system, a method of fault mode detection must be implemented so that the transmitter circuitry will not be damaged due to undesirable actions by the user In addition to protecting the transmitting platform from being damaged by conductive or magnetic objects, it is also desirable to reduce power consumption by turning off the transmitter when no valid receiving device is placed on the transmitting platform. The system will wake up for a very short period of time which is long enough for it to reach steady state. The time is predomi nately dependant on the self inductance of the transmitting coil and operating frequency. It is typically in the range of 0.5 ms to 5.0 ms. Probing the system with a rate of 1 Hz to 2 Hz speed is sufficiently fast for the user not to feel much latency. The refore, the worst case duty cycle is 1% and is sufficient to reduce no lo ad power consumption considerably. Alt hough, one could use a communication link to perform authentication and handshaking, there would be a considerable increase in cost and component count. An alternative is to detect the system loading condition by the voltages and current s of the transmitter. To ensure low power consumption and cost, the voltages and currents to be detected must be either DC or be converted to DC so that a low speed analog -to -digital convertor (ADC) can be used to accurately extract the information and convert to digital domain.

PAGE 112

112 7.1 Load/Fault Detection Scheme 7.1.1 Detection Circuit Cshunt Cout Crx Rload LoutLDC Transformation network Dvcoil Rdivider1 Rdivider2 Cvcoil Amplifiersupply current Buffercoil voltage a1 1 a2 2 b1 5 Vcc1 0 GND 0 Micro-controller Rsense Rvcoil Vcc In 1 In 2 Out 1 1 2 3 4 Figure 7 1 Block diagram of the proposed wireless pow er transfer system with detection circuit (detecting supply current and coil voltage) Figure 7 1 shows the block diagram of the proposed wireless power transfer system with detection circuit. The class E transmitter operates at 240 kHz. There are three par ameters that can be extracted from the wireless power transfer system without establishing a communication link between the transmitter and receiver. They are namely, the coil voltage, the supply current, and the coil current. Coil voltage is extracted by rectifying the coil voltage via a high impedance path using a half wave rectifier. Rvcoil and Cvcoil shown in Figure 7 1 are used to smooth out the rectified voltage Rvcoil is also used to regulate the current flow in the rectifying diode to prevent sudden and large current spikes from damaging the components The voltage is then stepped down via a

PAGE 113

113 potential divider using Rdivider1 and Rdivider2 to a workable voltage without damaging the inp ut port of the micro -controller To mitigate loading effects and reduce high frequency noise, a buffer using a low speed operational amplifier, e.g., LM324, in the voltage follower configuration can be used before the micro -controllers ADC port. Higher or der low pass can be achieved by adding appropriate capacitors to the operational amplifier as a simple first order low pass filter. Supply current is extracted from the circuit via the current sense resistor Rsense in Figure 7 1. The Rsense resistor is loc ated at the class E transmitter ground before returning to the system ground, instead of locat ing at the high side before LDC. Therefore, any voltage drop across the resistor will be referenced to the system ground instead of the supply voltage. Since the voltage drop will be extremely small (0.1V or less), a noninverting amplifier using an operational amplifier LM324 can be used to amplify the voltage, mitigating any loading effects and reducing high frequency noise. Similar to the extraction of coil volt age, extra low pass filter can be added by the same technique discussed. Measurements of the coil current can be realized by using either current sense transformers or current sense resistors. Current sensing transformers are typically large in size and op erate at frequencies lower than 100 kHz, making them impractical for this system. A current sense resistor can be added to the low side of the coil to measure the voltage drop across the resistor. This is more practical than placing the current sense resis tor on the high side as show n in Figure 7 2. This is because extremely high tolerance resistors are needed for the potential dividers if they are placed on the high side. Detecting a high frequency AC current with respect to ground is difficult because the ground of such a high voltage and current system is extremely noisy relative to the high side of the transmitting coil The ground noise will contain frequencies including the operating frequency making it not possible to mitigate its effects via filterin g. Unlike measuring

PAGE 114

114 the supply current, which does not require rectification, both the high frequency voltage signal and the ground noise will be rectified. Therefore, it is not possible to perform low pass filtering to mitigate the effects of ground noise using a low speed operational amplifier Since the ground noise is unpredictable and changes according to loading conditions, it is difficult to perform precise and stable measurement of the coil current. For the above reasons, only the coil voltage and s upply current will be extracted. TX Coil R2_H R1_H R1_L R2_L AC Out Ground Microcontroller Differential Amplifier Rsense Dicoil Cicoil Figure 7 2 Schematic of coil current extraction network using a current sense resistor. A DC switch shown in Figure 7 1 before the choke inductor LDC is used to turn off the transmitter under no -load mode or fault mode which includes over current. A low turn -on resistance PMOS transistor using a low power NMOS transistor to pull down its gate voltage can be used as the switch. Additional over current protection can be implemented by using a polymeric positive temperature coefficient device (PPTC) at the supply as a second level protection. Reverse polarity voltage protection can also be implemented by adding a reverse biased diode in shunt with the supp ly voltage. If a reverse voltage is applied, the diode will cause a short circuit and the PPTC will be activated to disconnect the supply path. Over voltage protection of sudden spiking can be implemented by using a transient voltage suppression (TVS)

PAGE 115

115 diod e commonly referred as a transorb across the supply. The TVS diode can also be doubled as the diode for reverse voltage protection. To prevent thermal runaway, a thermistor can be located next to the transistor of the power stage. By using the thermistor a s part of a potential divider on which the supply voltage is applied, the temperature can be tracked by reading the voltage across the thermistor using the micro -controller's ADC. 7.1.2 Detection Flowchart/Logic The detection scheme flowchart is shown in Figure 7 3. It can be implemented using a low cost micro -controller, such as 16F688 by Microchip. It is found experimentally that the system reaches a steady state after it is being powered on for 1 ms This is when a decision to power on or off the transm itter can be made. The time to steady state is dependent on the operating frequency as well as the self inductance of the transmitting coil. The no -load and safe modes follow a similar logic flow. The only difference is that the transmitter is powered down if no load mode is detected whereas the transmitter is powered up if the safe mode is detected. Both modes will probe the circuit for supply current and coil voltage after each predetermined X seconds to determine its operating mode, (a reasonable number for X is 1). Increasing X will incur higher latency to make the system response slow and decreasing X will incur higher noload power consumption. The operating modes are determined by the supply voltage and coil voltage space, which will be discussed late r The thresholds are dependent on the supply voltage, transmitters component values, and transmitting coil parameters. The dependence of the parameters on the receiver is found to be weak. Therefore, the thresholds work independent of receivers as well. Hysteresis needs to be implemented in the code so that the system does not oscillate when its operating mode is at the borderline case. The fault counter Z is reset if the system ends up in either mode.

PAGE 116

116 No Load/ Safe Fault Delay for X seconds Delay for X Z A seconds and increase fault counter Z Safe operating zone No load operating zone Fault counter < N Reset fault counter No Load/ Safe Fault Fatal fault Power down transmitter Power down transmitter Power up transmitter Power down transmitter Yes No Yes Yes No No Figure 7 3 Detection scheme flow chart for proposed system. When the coil voltage and supply current is excessive, the system enters its fault mode. A common cause of fault mode is when a huge piece of metal is placed over the transmitting coil, which decreases the total effec tive inductance of the transmitting coil significantly. When this happens, the class E ZVS /ZDS operation is no longer valid. If the transmitt er is not powered down immediately, the transistor will be damaged due to excessive power dissipating as heat. The delay to probe the system to determine if the cause of the fault has been rectified is increased by a factor A on top of the original X delay. The delay is increased with increasing occurrence of

PAGE 117

117 fault so that the system does not need to repeatedly stress the system within such a short window of time. O nce the number of tries reaches N the system will enter into the fatal fault mode, and the only way to exit from the mode is to perform a hard reset which involves disconnecting the DC supply of the system. Other fault modes which are not covered by the flowchart include thermal run away mode due to excessive heating at the power stage and battery fault mode for a one -to -one system. As will be shown in the experimental verification it is possible to track t he charge/power received status of a receiving unit for a one -to -one power system. If the trend of the power delivered over time deviates from the expected trend, the system will enter a battery/receiver fault mode which also requires a hard reset. This al so prevents possible damage when a user places a noncompliant receiver of a different charging profile on the transmitter. Brownout and over -voltage at the supply can also be detected by a supply voltage monitoring network so that the transmitter can be p owered down under fault mode. A supply monitoring network is simply a potential divider to drop the supply voltage to the range of the micro -controllers ADC so that the micro controller can make a decision based on the ADCs input. 7.2 Experimental Verif ication 7.2. 1 Test Bench and C ircuit The wireless power transfer system is fabricated based on the component values shown in Table 7 1 which was selected using desi gn rules presented in Chapter 3 The transmitting coil used is similar to the one used in Fi gure 3 1. To ensure that the proposed detection scheme can be applied to the same transmitter platform regardless of the receiver size, two different receiver sizes were used in the experimental verification. All coils were fabricated using 100/40 round se rved Litz wires. The supply voltage for the system is selected to be 12 V and can be varied depending on the power level requirements Supply voltage should not be considered as a factor

PAGE 118

118 during the analysis. The fabricated transmitter shown in Figure 7 4 h as a size of 15 cm x 2 cm. It is designed to be long and narrow so that it can be placed beside the transmitting coil as a single integrated unit. The low power detection and control block is located away from the high voltage power stage to reduce noise a nd coupling effects. The power input jack is located between the two blocks for the same reasons. Figure 7 4 Photograph of the fabricated transmitter circuit with control circuit. Table 7 1 Component Values for Load/Fault Detection Test Bench Value Coil Specifications TX coil inductance 45.3 H TX coil dimension 20 cm x 20 cm TX coil parasitic resistance 0.5 RX coil inductance (big RX) 5.45 H RX coil dimension (big RX) 9 cm x 6 cm RX coil parasitic resistance (big RX) 0.235 Mutual inductance (big RX) 2.81 H Coupling coefficient (big RX) 0.178 RX coil inductance (small RX) 4.00 H RX coil dimension (small RX) 6 cm x 5.5 cm RX coil parasitic resistance (small RX) 0.22 Mutual inductance (small RX) 1.88 H Coupling coefficient (small RX) 0.140 Circuit Specifications L DC 500 H C rx (big RX) 68 nF C rx (small RX) 95 nF C out 9.4 nF L out 10 H Lout parasitic resistance 0.16 C shunt 12 nF

PAGE 119

119 F1, PPTC Resettable Fuse1.5A hold, 3.0A trip MINISMDC150F/24-2 DC Supply (12V) a1 1 b1 2 GND 0 IN OUT GND U1 LM3480IM35.0 D1, TVS 17V 1500W SMCJ17A Driver Supply (12V) Control Supply (5V) Driver ON/ OFF Cdecoupling R1, 68k R2, 68k Q1, BSS138 Q2, NDT452A Cdecoupling Figure 7 5 Schematic of power stage of fabricated transmitter. Q3, IRLR3410 a1 1 a2 2 3 a3 4 a4 b1 b2 b3 b4 5 6 7 8 COMP FB CS RC REF VCC OUT GND U2 UCC3813-1 C2, 9.4nF L2, 10H C1, 12nF Coil High Side Coil Low Side Cdecoupling C3, 220pF R4, 12.1k L1, 500H Driver Supply (12V) R3, 330 Cdecoupling Over current protection R5, 0.1k TX Current Probe Figure 7 6 Schematic of driver stage of fabricated transmitter.

PAGE 120

120 a1 1 b1 2 Vcc1 0 a1 1 a2 2 3 a3 4 a4 b1 b2 b3 b4 5 6 7 8 a1 1 b1 2 GND 0 RC0 RC1 RC2 RC3 RC4 RC5 VDD GND RA0 RA1 RA2 RA3 RA4 RA5 Control Supply (5V) Cdecoupling Driver ON/ OFF + + + C4, 10nF R6, 10k R7, 274k R8, 4.7k R8, 47k R9, 6k R11, NTC 10k NCP18XH103J03RB R10, 6k U3 16F688 R12, 1k R13, 1k R15, 2.7k R16, 10k Coil High Side TX Current Probe DC Supply (12V) Control Supply (5V) Control Supply (5V) Over current protection D2, 1A 400V ES1G + R14, 9k C5, 100nF U4A LMV324IDR U4B LMV324IDR U4C LMV324IDR U4D LMV324IDR Cdecoupling Figure 7 7 Schematic of detection and control stage of fabricated transmitter. Figure 7 5 to Figure 7 7 shows the schematic of the fabricated transmitter. Figure 7 5 shows the power stage whi ch handles the over current protection, reverse voltage protection and on/off control of the transmitter. Figure 7 6 shows the driver stage which includes the class E t ransmitter and clock. Figure 7 7 shows the detection and control stage. The transmitter has two levels of protection, hardware protection and software controlled protection.

PAGE 121

121 Hardware protection is used as a last resort protection for which the thresholds a re set to be higher than the software controlled protection. It should only be triggered when the micro controller (U3) in Figure 7 7 fails. Over current protection is realized by a PPTC fuse, F1 as shown in Figure 7 5 which has a 1.5 A hold and 3.0 A trip. The transient voltage suppression (TVS) diode, D1 in Figure 7 5 serves two purposes. It helps to eliminate any transient voltage spikes due to discharge from the energized coil during powering down, and protects the transmitter from over -voltage. In addi tion, if the user accidentally connects the DC supply in reverse polarity it will attempt to short out the supply and load excessive current. The fuse F1 will be activated to break the connection protecting the transmitter. An intermediate stage of over c urrent protection is added to the transmitter via disabling the PWM clock, U2 in Fig ure 7 6 This is achieved by pulling the CS pin to high. Values of R15 and R16 in Figure 7 7 are selected to shut down the clock when supply current exceeds 1.2 A which is higher than the software determined maximum current of 0.85A as shown in Figure 7 1 2 The transmitter should rely on software control protection for its nominal operation. Thi s is achieved by turning on and off the PMOS transistor Q2 via a low voltage control generated by the micro -controller (U3) using a NMOS transistor to pull down the gate voltage of Q2 via R1 and R2. It is important that the regulator for the detection and control stage U1 is placed before the Q2 transistor so that the detection and control stage is still in operation even when power is being cut to the driver stage. Using a quad operational amplifier LMV324IDR as shown in Figure 7 7 the micro -controller (U 3) is able to read in 4 different parameters of the system. They are coil voltage (U4A), supply voltage (U4B), driver temperature (U4C) and driver stage current (U4D).

PAGE 122

122 Coil voltage and driver stage current extraction has already been discussed in the prev ious section The thresholds are predetermined by sweeping the loads as shown in Figure 7 1 2 By extracting the supply voltage, the transmitter is able to prevent over voltage conditions as well as under -voltage conditions that might cause the system to de viate from its nominal operation. The system is set to a supply voltage operating range of 9 V to 15 V. In addition to detecting the supply voltage, the transmitter is able to vary its thresholds according ly to make it more robust. The thresholds can be st ored in the flash memory of the micro -controller (U3). A 0.5V resolution is sufficient to prevent any false alarm or damage to the system. Finally, a thermistor (R11) is placed next to the class E drivers transistor to monitor its temperature. The system will shut down when temperature exceeds 75 C and only resume nominal operation when the temperature drops below 60 C. 7.2.2 Experiment al Results Fig ure 7 8 Efficiency Power plot of single receiver setup (Solid black line: Small coil. Dashed gray line: Big coil) 10 20 30 40 50 60 70 80 90 0 1 2 3 4 5 6Efficiency (%)Power Delivered to Load (W) Small Coil Big Coil

PAGE 123

123 Figure 7 8 shows the efficiency -power plot of a single receiver setup using the big and small receivers being placed on the same transmitting coil Since both of the receiving coils are optimized for operation on the same transmitting coil, t he efficiencies of both receivers are similar for the respective power level Better than 60% efficiency can be achieved for power delivery level above 1 .5 W keeping absolute power loss of the system low at all times which is important to ensure the system does not overheat. Due to stronger coupling, the big receiving coil has a slightly higher efficiency as well as power delivery Power delivery is about 6 W using the big coil and about 5 W using the small coil. Figure 7 9 Examples of different loading conditions and fault modes on coil voltage versus supply current space Figure 7 9 shows the coil voltage and supply current space diagram which is used to determine the different operating modes. The system enters the noload mode when the operating condition is in the vicinity of the diamond shape. A margin of 10 V and 50 mA can be use to frame the no load operating point so that the system will be more robust under component

PAGE 124

124 variation. The safe zone is at lower coil vol tage (less than 80 Vrms) and current (less than 0.8 5 A) region whe re two receivers are at their nominal operation without being over loaded Therefore, safe zone is located at the bottom left corner of the coil voltage and supply current space as any exces sive voltage or current will damage the transmitter. The small receiving coil generates a larger transmitting coil voltage because the same amount of power delivery is required over a smaller receiving area, which requires a stronger magnetic field. Two di fferent potential fault scenarios, a large copper plate being brought closer to the transmitting coil and a large copper plate being slid over the transmitting coil, are shown in Figure 7 9 Although, there are other fault scenarios such as placing a small er copper sheet in the middle of the transmitting coil the two scenarios tested should be able to provide sufficient understanding on how the system will react under various fault modes. The solid dashed line shows the trend of coil voltage and supply cur rent moving clockwise when the distance between the transmitting coil and a larger copper sheet becomes smaller. ZVS /ZDS starts to fail at the voltage inflection point as more energy is dissipated across the transistor instead of being transferred to the l oad (copper pate). If the system does not shut down immediately the transistor will be damaged due to heating. The hollow dashed line shows an increasing coil voltage and supply current when the overlapped area between the transmitting coil and a large cop per plate increases. The trend of the line is expected to follow the same as the experiment which the copper plate is brought closer to the transmitting coil. However, the coil voltage and supply current is so excessive that the transistor will be damaged due to non ZVS/ZDS operation. Therefore, measurements were not carried out beyond 2 A supply current Since h igh coil voltage will always lead to over voltage problems at various points of the circuit and damage the components it is not important to study the trend of the curve during fault operation as the priority is to shut down the system as quic kly as possible

PAGE 125

125 Figure 7 10. Efficiency -power plot of three sets of dual load measurements and two sets of single load measurements using a combination of big and small receiving coils. Three sets of measurements were carried out on the system using two loads of load resistance ranging from 20 discrete steps. The three sets are two big coils, two small coils, and one big coil with one small coil. The pair of receiving coils are placed side by side in the middle of the transmitting coil with a 2 cm gap between each other. The efficiency plots of the three measurements are shown in Fig ure 7 10. In addition, two single receiver measurements are also shown in Figure 7 10 for comparison. DC DC efficiency is above 70% for power delivery levels above 3W, crossing 85% for some load points. The spread of efficiency is approximately 5% regardless of the number of receivers. All load conditions achieve d ZVS /ZDS operation, maintaining high efficiency operation with minimum power loss. Since the efficiency of the system is well bounded regardless of loading conditions and their sizes, the spread in the coil voltage and supply current space should be limit ed as well. 10 20 30 40 50 60 70 80 90 0 1 2 3 4 5 6 7Efficiency (%)Power Delivered to Loads (W)

PAGE 126

126 Figure 7 11. Transmitter coil voltage and supply current space illustrating three different zones: no -load, single load, and dual loads. Figure 7 11 shows the transmitting coil voltage and supply space for a single load (big coil and small c oil) and dual loads. Although, there is some overlapping between the 2 spaces, it is possible to detect the number of loads for most of the loading conditions. As shown in Figure 7 11 a sharp transition will be observed when an extra load is being placed o n or removed from the transmitting coil. Therefore, the system can easily detect if an additional receiver is placed on the transmitting coil or being removed from it by tracking the transmitter coil voltage and supply current over time It is also possibl e to detect the number of loads when the transmitter is powered on with valid loads on the transmitting coil because during the initial power on states of most electronic devices, they will not draw much power. Therefore, the power delivery will slowly ram p up during the first few seconds when the transmitter is powered on. Due to this intrinsic characteristic of most electronic devices, t he system will typically observe an initial coil voltage of 40 Vrms to 48 Vrms for dual loads and 52 Vrms to 58 Vrms for single loads. Under the

PAGE 127

127 worst case situation, if it is not possible to differentiate the number of loads being placed on the transmitter, the system can track the loading conditions via techniques such as Markov Chains over a specific period of time a nd determine its loading conditions. This requires large amount of training data to perform load pattern analysis which is beyond the scope of this dissertation Figure 7 1 2 Transmitter coil voltage and supply current space diagram illustrating three different zones noload, safe and fault. Summarizing the above discussions, Figure 7 1 2 shows the transmitter coil voltage and supply current space indicating the various zones. The no -load zone is enclosed by red circle and the safe zone is enclosed by b lue lines. The no -load is basically a point solution with a radius for system tolerance purposes which can be simplified into a rectangle or square. The zone can be described using two points in the space or four data points for which each point is describ ed using the transmitter coil voltage and supply current or simply the four sides of the rectangle/square The safe z one has two steps. For supply current below 0.3 A, the transmitting coil voltage should be between 32 Vrms and 64 Vrms whereas for supply c urrent above 0.3 A, the

PAGE 128

128 transmitting coil voltage should be between 32 Vrms and 80 Vrms. Any supply current greater than 0.85 A is considered as a state of over current for this system and determined as fault mode F ive data points are required to describe the safe zone In fact, any operating condition not within the safe or no -load zone is considered to be in the fault zone Therefore, each unique supply voltage and operating frequency will require nine data points to describe its coi l voltage and supply current space. The purple zone on the top left corner of Figure 7 1 2 is an invalid zone because the supply current is too low to generate such voltages of those magnitudes and would never occur. ZVS /ZDS operation is still valid to some exten t in the area above the safe zone. However, the transmitting coil voltage is very high and might damage components. In addition, excessive transmitter coil voltage might lead to thermal run away on some components especially the transistor The brown area on the right of the safe zone is when the transmitter no longer operates under the ZVS condition. When this happens, large amount of energy is dissipated across the transistor as the high voltage across the drain of the transistor and high current th rough the transistor are no longer orthogonal in time. An interesting linear relationship between the supply current and the power delivered to the load is shown in Figure 7 1 3 The results from the five sets of measurements can be consistently described using the equation y = 0.095x + 0.055 where the y axis is supply current to the transmitter and the x axis is the power delivered to the load. Based on the 12 V supply voltage, for every 1 W of power at the receiver the transmitter requires an input power of 1.14 W. Once the system is operating in the safe zone, using the proposed equation, Figure 7 1 4 shows the plot of calculated delivered power via measured supply current to the transmitter versus the actual delivered power. The results show relatively good agreement at all power levels with an average error of 0.08 W and standard deviation of 0.16 W for 616 sets of reading.

PAGE 129

129 Figure 7 1 3 Linear relationship between supply current and power delivered to the loads for all 5 sets of measurements. Solid da shed line: y = 0.095x + 0.055. Figure 7 1 4 Power delivery error distribution plot. Calculated power delivery is based on measured supply current. Solid dashed line: ideal error free calculation. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0 1 2 3 4 5 6 7 8Supply Current (A)Power Delivered to Loads (W) 0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8Calculated Power Delivery (W)Measured Power Delivery (W)

PAGE 130

130 Although the experimental verification shows the feasibility of the proposed load/fault mode detection scheme, the thresholds for different modes and the equation to track the power delivery are highly dependent on the supply voltage and operating frequency Therefore, a voltage sensing mechanism is impl emented to detect the supply voltage and make appropriate adjustments to the thresholds. The operating frequency is assumed to be stable for this case by using 1% tolerance components for the RC timing network. A more effective method is to use a crystal r eference to generate the clock frequency; a low cost 30ppm crystal will generate a clock signal with negligible frequency offset. However, by using this technique the frequency will be locked to a single frequency or multiple/fraction of the operating freq uency making fine frequency resolution control not possible. 7.3 Extension of Load/Fault Detection Scheme 7.3.1 M:N Coupling Structure The analysis of the load/f ault detection scheme is confined only to single transmitting coil architecture. As shown in Figure 5 23, placing two receivers concurrently onto a single transmitting coil will choke the power delivery as the power delivery of one receiver is dependent on the loading condition of the other receiver This observat ion is supported by E quation 4 6 for which the real part of Ztx increases with each additional receiver while the reactance of Ztx remains approximately the same. Therefore, the power delivery remains approximately the same but is being split between two receivers. Instead of using a single transmitting coil, a M:N coupling structure shown in Figure 7 15 is proposed. The M:N coupling structure can be implemented to support multiple portable devices with a one to -one pairing or cover a large ar ea by using a large array of transmitting coil s to power compact sized receivers. In depth analysis of the M:N coupling structure is beyond the scope of the discussion as the focus will be on system level. For the following analysis and experimental verifi cation, the focus

PAGE 131

131 will be on a M:N coupling structure supporting two portable devices. A single class E transmitter is used to drive two similar transmitting coils (15.5 cm x 17.5 cm) being placed beside each other. Two identical receiving coils (9 cm x 6 cm) are being used. The design rules in Chapter 3 are used to select the components. Figure 7 15. Examples of M:N coupling structure s A) Supporting two portable devices and. B) Large array of transmitting coils cover ing a large area. Figure 7 1 6 shows the power choking delivery trend of a dual receiver setup when both receivers are being placed on a single transmitting coil while leaving the other coil unloaded. T he power delivery is dependent on the other receiver with power level above 1W is observed, which is similar to the trend in Figure 5 23. The choke point is lower than that of Figure 5 23 because the bigger receiving coil results in stronger coupling. A stronger coupling between the transmitting and receiving coils will enhance the de pendency of power delivery between two receivers being placed on a single transmitting coil. Figure 7 1 7 shows the power delivery trend

PAGE 132

132 of a dual receiver setup when one receiver is being placed on each transmitting coil. The power choking phenomena is n ot observed for this case, power delivery to each receiver is independent of one another. This is because the coupling between each transmitting coil is very weak as they are placed next to each other instead of overlapping each other. Therefore, it can be assumed to be two independent single transmitting coil s to a single receiving coil system Power delivery to the individual receivers is not balanced because the coils are hand wounded with tolerance of self inductance as high as 10%. This problem will be eliminated during mass production as the coils will be machine wounded or PCB fabricated with much higher tolerance. As shown by using the M:N coupling structure the problem of power chocking can be eliminated. To ensure that the design is feasible, fu rther analysis on the efficiency, validity of transmitter coil voltage and supply current space as well as power delivery Figure 7 1 6 Measured power delivery of a dual receiver setup on a dual transmitting coil test bench with both receivers on a s ingle coil leaving the other coil unloaded. 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0Power Delivered to Receiver 1 (W)Power Delivered to Receiver 2 (W)

PAGE 133

133 Figure 7 1 7 Measured power delivery of a dual receiver setup on a dual transmitting coil test bench with one receiver on each transmitting coil. Figure 7 1 8 shows the efficiencypower plot for both test results of Figure 7 1 6 and Figure 7 1 7 including the results for a single re ceiver setup. Efficiency is high for both cases achieving a better than 75% efficiency for power level above 4 W Efficiency follows the same trend for setup with single receiver and dual receiver s on a single transmitting coil confirming the design rules presented in Chapter 3 can be applied to a M:N coupling structure Slight efficiency improvement is observed for the setup with a rec eiver is placed on each of the individual coil s This can be explained by the fact that, a lthough one of the transmitting coil is not loaded, current is still running through the coil The current will cause energy to dissipate across the coils parasitic resistance as high level of reactive power is being driven into the unloaded coil However, if the coil is loaded, the resistance seen by the coil due to its load is much larger than the parasitic resistance of the coil and the losses can be neglected for this case improving the efficiency of the system 0 1 2 3 4 5 6 0 1 2 3 4 5 6 7 8 9Power Delivered to Receiver 1 (W)Power Delivered to Receiver 2 (W)

PAGE 134

134 Figure 7 1 8 Efficiency -power plot of two sets of dual load measurements and two sets of single load measurements. (Green line s : Single receiver, Blue line s : Dual receivers on single transmitting coil, Red line s : Dual receivers on dual transmitting coil) Figure 7 1 9 Transmitter coil v oltage and supply current space. (Green diamond: No load, Black lines: Single receiver, Blue lines: Dual receivers on single transmitting coil, Read lines: Dual receiver on dual transmitting coil) 10 20 30 40 50 60 70 80 90 0 2 4 6 8 10 12 14 16 18Efficiency (%)Power Delievered to the Load(s) (W) 50 55 60 65 70 75 80 85 90 95 100 105 110 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4TX Coil Voltave (Vrms)Supply Current (A)

PAGE 135

135 Figure 7 1 9 shows the transmitter coil voltage and supply current space for various loading conditions which is similar to Figure 7 1 2 Therefore, it is possible to extend the same load detection technique for a M:N coupling structure. The discrepancies shown in Figure 7 1 9 for the single receiver case is due to the tolerance of the hand wounded transmitting coil as discussed earlier. It is observed that the two different conditions of loading the transmitting coils with two receivers resulted in two distinctive zones in the transmitter coil voltage and supply current space. By using techniques o f pattern recognition and sufficient training sample size the transmitter is able to determine the number of receivers on each individual coil. Although there are some overlaps between a single receiver setup and the setup with two receivers on a single t ransmitting coil t he trends of the lines are distinctively different follow ing a different charge profile trend. The trends can be modeled into different sets of Markov C hains for which over time the state of the system can be determined. Figure 7 20. Linear relationship between supply current and power delivered to the loads for all 4 sets of measurements. Solid dashed line: y = 0.076x + 0.048. 0 0.2 0.4 0.6 0.8 1 1.2 1.40 2 4 6 8 10 12 14 16 18Supply Current (A)Power Delievered to Load(s) (W)

PAGE 136

136 Figure 7 2 1 Power delivery error distribution plot. Calculated power delivery is based on measured supply current. Solid dashed line: ideal error free calculation. Figure 7 20 shows that the linear relationship between the supply current and power delivered to the load(s) still hold s for a M:N coupling str ucture regardless of the placement of the two receivers. The linear relationship can be describe d by the equation y= 0. 0 67x + 0.048. Figure 7 2 1 shows the power delivery error distribution plot using the equation to calculate the delivered power The avera ge error is found to be 0.017W and standard deviation error is found to be 0.183W out of a maximum of 18W power delivery. This shows a very small spread in error s which verified the feasibility of using the power tracking scheme on a M:N coupling structure Similar trends are expected for higher number of transmitting and receiving coils attached to the class E transmitter. However, as more transmitting coils are attached in parallel to a single transmitter the effective inductance will be reduced causing the circuit to be more sensitive I ncreasing the transmitting coil size or number of turns will help to mitigate the effect 0 2 4 6 8 10 12 14 16 18 0 2 4 6 8 10 12 14 16 18Calculated Power Delivered to Load (W)Measured Power Delivered to Load (W)

PAGE 137

137 Figure 7 2 2 Transmitter coil voltage and supply current space with fault tests. (Green lines: Fault tests without valid receive r on one of the coil, Purple lines: Fault tests with a valid receiver on one of the transmitting coil s ) Five different common fault scenarios are being studied for the M:N coupl ing structure. They all involved a large piece of copper sheet (30 cm x 30 cm) sliding across the coil or being brought to close proximity on the same plane. All of the fault lines green and purple follow a clockwise trend with increasing severity of the fault (proximity of copper sheet or amount of overlap). The green lines in Figure 7 21 show three different fault scenarios without any valid receiver being placed on one of the transmitting coil s The detection scheme is capable of detecting a fault condition when a valid load is not placed on one of the transmitting coil (green line), but there are some overlaps between the load conditions and fault conditions when a valid load is being place on one of the transmitting coil s while a piece of copper sheet is being slide across or brought closer to the unloaded transmitting coil ( purple line). Although the overlapping region is small, an analysis is required to determine if it is acceptable for a minor fault to be considered safe 0 25 50 75 100 125 150 175 200 225 250 275 300 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0Transmitter Coil Voltage (Vpeak)Supply Current (A)

PAGE 138

138 The analysis starts with looking at what happens when a sheet of copper is partially overlapped onto a transmitting coil. Due to the close proximity of the copper or any electric conductor of sufficient mass, the self inductance of the coil will be reduced. The reduction is due to the proximity and amount of overlap. For this setup, two coils are driven in parallel. Therefore, with the reduction of self inductance of one of the transmitting coil s, due to a copper sheet in close proximity, the total effective inductance looking into the pair of coils will be reduced as well. The reduction in effective ind uctance of the dual parallel structure will cause a reduction in the phase angle of Ztx, making Ztx more capacitive. A more capacitive Ztx will result in higher power being transferred or higher unregulated receiver voltage. This observation is similar to decreasing Cout which will be discussed in Chapter 8 Although a slight overlap between the unloaded transmitting coil and a piece of copper sheet may seem to of not a major problem, it will cause the voltage before the voltage regulator of the valid load to increase. Therefore, if t he maximum input voltage of the voltage regulator used on the receiver has insufficient margin, the voltage regulator will be damaged. To prevent the voltage regulator from being damaged, the voltage regulator must have a high i nput voltage tolerance which makes the receiver bulking and costly. It will also limit the selection of voltage regulator s available to the designer. A potential solution to th e problem is to design the transmitt er unit so that each transmitting coil is d riven by an independent transmitter using a master clock and a master micro -controller as shown in Figure 7 2 2 Load/fault detection is done by each individual pair of transmitter s Since the coupling between neighboring coils is weak, the system can be as sumed to consist of multiple 1:1 or 1:N transmitting systems which the load/fault detection scheme is proven to be effective earlier in this chapter. Depending on the load conditions the transmitters

PAGE 139

139 can be turned on and off accordingly, making it more en ergy efficient when one of the coils is unloaded. In addition, if a fault occurs on one of the coil s it can be shut down while power delivery is not affected by the other transmitting coil. Since each transmitting coil will require their respective set of class E transmitter circuitry, it will increase the board size, components count and potentially increase the cost. However, each class E transmitter will most probably be required to drive a single load instead of multiple loads. T hus, the power handling of each transmitter will be reduced, allowing the use of smaller and more compact components. Du e to the reduction in component size and power handling, making each transmitter more compact and cheaper offset the cost of multiple un its. Figure 7 2 3 Different methods of driving a 2:N structure. A) Original M:N design for the experimental verification and B) proposed improved M:N design. 7.3.2 Removing Lout from Transmitter Since a Class E power amplifier only requires a phase angle of Ztx to be from 40 to 70 it is possible to remove Lout from the transmitter load network using part of the leakage inductance (imaginary part of E quation 4 6) of the transmitting coil as Lout (shown in Figure 724) as long as the phase angle can be maintained within the high efficiency range across all receiver load resistances. Therefore, the design of the transmitting and receiving coils must be integrated as part of the design rule and can no longer be treated as a generic linear block. Since Lout is the

PAGE 140

140 component with the highest loss in the class E transmitter, removing Lout will improve the efficiency by at least 5%. In addition, when the transmitting coil is not loaded, a significant amount of power is lost across the inductor and the transm itting coil. As Litz wire is being use d to fabricate the transmitting coil, most of the losses will be across Lout. Therefore, by removing Lout from the system, the no load power consumption can be reduced significantly as well However, doing so will caus e the higher order harmonics content to be stronger across the coil as Lout also serves as a low pass mechanism. The radiation efficiency of the coil improves with the increase in frequency, resulting in the higher order harmonics being transmitted into fr ee space. Therefore, care must be taken during the design to ensure the emission is low enough to pass the FCC part 18 test, this is important especially for high power system. Figure 7 2 4 Block diagram of wireless system without Lout. Using the same experimental setup as the M:N coupling structure, Lout is removed and Cout is tuned to compensate for the missing positive reactance by decreasing its capacitance value. The same fault test procedures are carried out on the test bench with Lout removed so that a fair analysis can be made. Results of the fault tests using the same format as Figure 7 2 2 are presented in Figure 7 2 5 Cshunt Cout AC Crx Rload ZtxLDC

PAGE 141

141 Figure 7 2 5 Transmitter coil voltage and supply current space with fault tests with Lout removed. (Green lines: Fault tests without valid receiver on one of the coil, Purple lines: Fault tests with a valid receiver on one of the transmitting coils) Similar initial trend of the test results sh own in Figure 7 2 2 was observed for all the five different fault tests with the lines going in a clockwise direction. However, the lines follow a n upward trend while looping back and moving into the low voltage low current region instead of the downward tr end as show in Figure 7 2 2 This results in the certain conditions of the fault mode overlapping with the safe or valid load region. All of the conditions involved the copper sheet being in close proximity or almost complete overlap onto the transmitting c oil. This condition is a common case when the transmitting coil is accidentally flipped over onto a metallic table top or an unshielded transmitting coil being placed onto of a metallic table top. B y removing Lout from the transmitter using existing circui try, the system will be unable to differentiate between a valid load and a fault mode under certain conditions. Therefore, to ensure robust operation while removing Lout, modifica tions of the detection circuit are required. 50 70 90 110 130 150 170 190 210 230 250 270 290 0 0.5 1 1.5 2 2.5 3 3.5 4Transmitting Coil Voltage (Vpeak)Supply Current (A)

PAGE 142

142 Figure 7 2 6 Waveform of the transmitter circuit without Lout when fault conditions overlaps with valid load conditions. (Red: Drain voltage, Orange: Coil voltage, Green: Coil Current) Figure 7 2 7 Spectrum of transmitter coil voltage without Lout when fault conditi ons overlaps with valid load conditions.

PAGE 143

143 To better understand the mechanism wh ich cause this upward trend to happen, the transmitt ing coil voltage and it s spectrum was capture d using a n oscilloscope. As seen from Figure 7 2 6 the second harmonic is much s tronger than the fundamental tone for both the transmitt ing coil voltage and coil current with the transmitting coil voltage at approximately 150 V peak to peak This is supported by the measured spectrum of the transmitter coil voltage shown in Figure 7 2 7 The fundamental tone is approximately 27 dBV while the second harmonic is almost 31 dBV. T he second harmonic can be considered as detection noise as the load/fault detection scheme neglects the effects of higher order harmonics. With a stronger second o rder harmonic on top of the fundamental tone the transmitter coil voltage will be larger than expected explaining why the transmitting coil voltage is larger than expected or has an upward trend into the valid load region under certain fault conditions. Figure 7 2 8 Waveform of the transmitter circuit with Lout when fault conditions overlaps with valid load conditions. (Red: Drain voltage, Orange: Coil voltage, Green: Coil Current)

PAGE 144

144 Figure 7 2 9 Spectrum of transmitter coil voltage with Lout when fa ult conditions overlaps with valid load conditions. Figure 7 2 8 and Figure 7 2 9 show the same waveforms as Figure 7 2 5 and Figure 7 2 6 but with Lout added back to the transmitter circuit. The higher order harmonic are being suppressed by Lout. The second harmonic is reduced by approximately 10dB at 19 dBV with the fundamental t one remain ed at approximately 27 dBV. The transmitting coil voltage is reduced significantly from 150 V peak to peak to 70 V peak to peak. This brings the transmitting c oil voltage below the safe zone lower threshold. A lthough removing Lout is able to improve efficiency and reduce heating, higher harmonics contents which are filtered by Lout are allowed to pass. This affects the load/fault detection scheme as detection no ise. T he error in detection can also be mitigated by using a si mple low pass RC filter before the detection diode to reduce the higher order harmonics. This can be realize d by adding an appropriate capacitor to ground between R6 and D2 of Figure 7 7 If n eeded, a higher order low pass filter can be added in the coil voltage detection path to further

PAGE 145

145 reduce the higher order harmonics. By adding the low pass filter, the original detection scheme can be used without causing any detection error. However, the h igher order harmonics voltage across the transmitting coil will still be a potential issue during the emission test as the same transmitting coil will become a better radiat or with the increase of operating frequency.

PAGE 146

146 CHAPTER 8 RECEIVER VOLTAGE CONTROL A lthough, t he system is able to reduce the power delivered to the receiver when the effective load resistance increases, the unregulated DC voltage at the receiver will increase with increasing load resistance. The swing in receiver voltage is not important when the power level of the system is low as the voltage regulator at the receiver is able to tolerate such fluctuations However, the range of voltage swing increases with increasing power rating of the system. A buck regulator will have poorer efficiency with a higher input voltage and tend to be larger in size as external switch transistors are required to perform the regulation. In addition, if a high input voltage is required for the receiver voltag e regulato r, the choices for the system designer will be limited and BOM cost will be higher as well. Therefore, it will be desirable for the unregulated voltage at the receiver to be bounded to a certain range. Cshunt Cout AC Crx Rload ZtxLoutLDC Figure 8 1 Block diagra m of wireless power system From the simplified block diagram of the system as shown in Figure 8 1, there are limited parameters of the system that one can vary. Most probably t he structure of the coil will be fixed, any attempt to vary the coil structure in terms of turns and area will make the system complicated. Capacitor Crx can be tuned to control the received power level. There are two ways that Crx can be tuned; it can be achieved by either using a va ractor or switching between bank s of

PAGE 147

147 capacitors. Since the system is intended to operate with the frequency range of less than 1 MHz, it will take a considerable amount of capacitance to achieve an appropriate tuning range. Although, it is possible to achieve it with switchable banks of capac itors, it is not practical for the current implementation unless a high voltage and current switch is realized in the fo r m of an integrated circuit. In addition, it is desirable to keep the receiver as simple as possible and move all the complexity to the transmitter as a transmitter will typically be used to power multiple receivers concurrently or at different instances. There are three possible ways to achieve voltage control of the unregulated receiver voltage on the transmitter. They are varying supply voltage, varying operating frequency and varying Cout. Each option will be studied with experimental results from a one to one setup test bench The test bench used in the experiment is intended for high power applications (up to 80W) operat ing at 240 kHz and a supply voltage of 30V with the transmitting coil of 25 cm x 35 cm and the receiving coil is 12 cm x 12 cm. The findings should apply across all power levels and coil sizes with different level of tolerances. 8.1 Varying Supply Voltage to Achieve Receiver Voltage Control The most straightforward way to achieve power control and receiver voltage control is to control the supply voltage of the transmitter. Experimental results by sweeping the supply voltage from 18 V to 30 V in steps on 1 V as shown in Figure 8 2 and Figure 8 3 further verify the feasibility of supply voltage control to achieve power control and receiver voltage control. From Figure 8 2, high efficiency is achieved across the whole range of supply voltages with efficiency peaking at dif ferent power level depending on the supply voltage. If the system is able to control the supply voltage from very low voltages such 6 V up to 30 V in sufficiently small steps, the system will be able to achieve better than 80% system efficiency regardless of loading conditions. However, this is not practical because it will require the system to be driven from a variable output voltage regulator instead of directly from a DC source. This will significantly

PAGE 148

148 impact the system efficien t especially when the sup ply voltage is very low e.g. 9 V. This is because a buck regulator becomes less efficiency when the difference between the input and output voltages increases. In addition, a high efficiency high power voltage regulator with a variable output is complex and costly. It is typically achieved by using a n off the shelf variable voltage regulator and replacing one of the feedback resistors with a digital rheostat. The digital rheostat is then controlled by the micro-controller which is used for load and fault mode detection. Figure 8 2 Efficiency p ower plot for supply voltage from 18 V to 30 V in steps of 1 V Either the load tracking technique proposed in Chapter 7 or a communication link can be used to track the loading condition for a one to one load system. The supply voltage can be varied to optimize for the specific load condition. This technique can be applied to the other two following methods of controlling receiver voltage. Each supply voltage will have a different power delivery profile with respect to supply current. However, since the relationship between supply current and power delivered can be approximate d by a linear relationship, the profile is 40 45 50 55 60 65 70 75 80 85 0 10 20 30 40 50 60 70 80 90 100Efficiency (%)Power Delivered to Load (W) 30V 29V 28V 27V 26V 25V 24V 23V 22V 21V 20V 19V 18V

PAGE 149

149 conde nsed into two data points, the gradient and intercept. As discussed in Chapter 7 t he noload requires four data points. The safe zone which was also discussed in Chapter 7 requires additional 5 data points Therefore, each voltage point will require 11 da ta points and if each data point takes up 2 bytes of memory, each supply voltage point will require 22 bytes. A reasonable system will require a supp ly voltage resolution from 20 to 50 points, requiring 440 bytes or 1100 bytes of memory space on the micro -controller. Figure 8 3 Receiver v oltage p ower plot for supply voltage from 18 V to 30 V in steps of 1 V As shown i n Figure 8 3, the unregulated receiver voltage decreases with decreasing supply voltage at the same power level Therefore, the receiver voltage can be kept within a tight bounded range for power delivery across a large power band Depending on the acceptable voltage ripples and with sufficient resolution achieved on the supply voltage control, it is possible for the system to achieve wirel ess power transfer without a voltage regulator on the receiver. This is only true if and only if the coupling coefficient remains the same for all possible 35 40 45 50 55 60 65 70 75 80 0 10 20 30 40 50 60 70 80 90 100Unregulated Receiver Voltage (V)Power Delivered to Load (W) 30V 29V 28V 27V 26V 25V 24V 23V 22V 21V 20V 19V 18V

PAGE 150

150 placements. For a practical system, such a requirement will not be realistic unless the receiving co il is large enough to average the magnetic field variation or a locking mechanism is in place to ensure a fix placement. If so it will defeat the purpose of providing some level of mobility for a wireless power system unless it is used in some unique appli cations Therefore, to ensure 100% stability without the system drifting in the wrong direction and still attain some level of mobility a communication link is required to close the loop. F rom Figure 8 3 it can be seen that if the system is to achieve a power handling of up to 40 W, the lowest possible voltage to achieve such a scheme without a voltage regulator on the receiver will be 42.5V (supply voltage 20 V line). The lowest voltage for such a power level is high because of high leakage inductances on both the transmitting and receiving coil s Such high leakage inductances are due to poor coupling coefficient which results from the size ratio between the transmitting and receiving coil (transmitting coil is 5 times larger than receiving coil). In orde r to bring the voltage lower, the receiving coil needs to become larger as relative to the transmitting coil (reduce leakage inductances) or have less turns (trading voltage for current at the expense of resistive losses in the receiving coil) This brings another set of constrain t s in the system design embedding the coil design into an integrated part of the system design Such design constraint s are unique to specific application s which are beyond the scope of this discussion and can be covered as potenti al future work for more tightly coupled one to one wireless power transmission system. Figure 8 4 shows the linear relationship between the supply current and power delivered to the load for different supply voltage (18V to 30V in steps of 1V). The yinte rcept for all the supply voltages remains approximately the same but the gradient decreases with increasing supply volt age The trend can be explain by the fact that for the same power level at similar

PAGE 151

151 efficiency, the supply current decreases with increas ing supply voltage to maintain similar power level and efficiency The difference will become more significant with the increase of power delivered to the load resulting in a fan out plot from the y -intercept Therefore, each supply voltage will require a new data set to describe the trend. I t is possible to use a single y intercept for all supply voltages if the tolerance for power tracking is relaxed The difference will not be significant if the power handling of the system is sufficiently large (>50 W) Figure 8 4. Linear relationship between supply current and power delivered to the load for supply voltage from 18 V to 30 V in steps of 1 V In conclusion, varying the supply voltage is an effective way to control the power delivery as well as the unr egulated receiver voltage. With sufficient resolution, the system is able to deliver a stable receiver voltage for which the voltage regulator can be removed if and only if the coupling coefficient is sufficiently constant for all placement or both the tra nsmitting and receiving coil is fixed in position Since it is meaningless to design a wireless power system without any freedom in placement and it is also not possible to design a transmitting coil with a 100% even field distribution, the solution is not practical without a communication link to 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 0 10 20 30 40 50 60 70 80 90 100Supply Current (A)Power Delivered to Load (W) 30V 29V 28V 27V 26V 25V 24V 23V 22V 21V 20V 19V 18V

PAGE 152

152 provide a feedback loop. By adding a voltage regulator between the DC supply and the transmitter to control the supply voltage the system will incur additional power losses which might result in thermal and effic iency issues. In addition, the complexity, size and cost added to control the supply voltage can be considerable. Therefore, varying supply voltage to control unregulated voltage and power delivery is an effective technique but not practical. 8.2 Varying Operating Frequency to Achieve Receiver Voltage Control Varying the operating frequency can be used to indirectly tune to impedance looking into the transmitter load network after Cshunt as shown in Figure 8 1. The impedance looking into the transmitter lo ad network (Ztx) can be simplified in to a series network of a capacitor an inductor and a load resistance under any load condition. To ensure the effective Ztx is inductive, the operating frequency is selected to be higher than the self resonance frequenc y of the LC network not taking the load resistance into account Therefore, Ztx will be less inductive with decreasing operating frequency, decreasing the phase of Ztx, thus, increasing power delivery as well as unregulated receiver coil voltage, vice vers a However, the above assumption is true if and only if ZVS/ZDS operation still holds at the specification load impedance without changing Cshunt in Figure 8 1. Although making Cshunt tunable will improve the frequency dynamic range of a frequency tunable class E power amplifier, it will also increase its complexity If the ZVS/ZDS operation is no t valid there will be an increase in device stresses and degradation in efficiency. Due to the large amount of data and ease of using the nominal operating frequency as a reference point, t he analysis is split into two different frequency ranges T he first frequency range ( Figure 8 5 and Figure 8 6 ) will be for operating frequency above the selected operating frequency in steps of 2 kHz up to 250 kHz and the secon d frequency range ( Figure 8 7 and Figure 8 8 ) will be for operating frequency below the selected operating frequency in steps of 1 kHz down to 235 kHz

PAGE 153

153 Figure 8 5 Efficiency Power plot for operating frequency from 240 kHz to 250 kHz in steps of 2 kHz Figure 8 6 Receiver v oltage p ower plot for operating frequency from 240 kHz to 250 kHz in steps of 2 kHz 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 90Efficiency (%)Power Delievered to Load (W) 240kHz 242kHz 244kHz 246kHz 248kHz 250kHz 50 55 60 65 70 75 80 0 10 20 30 40 50 60 70 80 90Unregulated Receiver Voltage (V)Power Delivered to Load (W) 240kHz 242kHz 244kHz 246kHz 248kHz 250kHz

PAGE 154

154 As shown in Figure 8 5 the power delivered to the load decreases with increasing operating frequency because Ztx becomes more inductive decreasing real power and increasing reactive power as discussed in Chapter 1 The unregulated receiver voltage decrease s marginally with increasing operating frequency up to 244 kHz However, as seen from Figure 8 5 the system efficiency starts to degr ad e at frequencies above 244 kHz. This is due to ZVS/ZDS operation does not hold anymore The nonideal waveform causes the switching losses across the drain to increase The switching losses are predominately caused by the transistor being turned on before drain voltage drops to zero due to a shorter period with the increase in frequency This results in the charges from Cshunt are being dumped across the transistor during turn on at high currents The unregulated receiver voltage also starts to become larger at lighter load especially during open circuit conditions when the operating frequency is increased to above 244 kHz as shown in Figure 8 6 Although Ztx is more inductive with the increase in operating frequency limiti ng the power delivery, the coupling is also increased increasing the voltage induced on the receiver. As shown in E quation 2 6 the increase of coupling with respect to operating frequency is a square law effect. Therefore, there are two competing mechanism s affecting the power delivery and unregulated receiver voltage working against each other In addition to the non ZVS/ZDS operation, Ztx swings over a larger range with the same components on the transmitter and receiver load resistance at higher operatin g frequencies Therefore, the range and value of unregulated receiver voltages increases with operating frequency as shown in Figure 8 6 As shown in Figure 8 7 the power delivered to the load increases with decreasing operating frequency because Ztx becomes more capacitive with decreasing frequency. The unregulated receiver voltage increases marginally with decreasing operating frequency down to

PAGE 155

155 235 kHz and below This trend is expected with the increase in power delivery. It can be observed from Figu re 8 7 that the system efficiency starts to degrade much slower than increasing the operating frequency because ZVS/ZDS operation still holds Although ZVS/ZDS operation does still hold, the zero crossing occurs earlier with decreasing frequency. With the decrease in operating frequency the period of each cycle gets longer but the drain waveform remains similar, causing the zero cross ing of the drain waveform to occur earlier before the transistor is turned on Minimum amount of power is still lost via the built in diode of the transistor causing a slight drop in system efficiency. The power loss via the built in diode can be reduced by adding an external diode such as a schottky diode with a faster transition time and small er turn on voltage. The analysis i s only valid if ZVS/ZDS operation still holds while the operating frequency is decreased. Figure 8 7 Efficiency p ower plot for operating frequency from 235 kHz to 240 kHz in steps of 1 kHz 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 90 100Efficiency (%)Power Delievered to Load (W) 235kHz 236kHz 237kHz 238kHz 239kHz 240kHz

PAGE 156

156 Figure 8 8 Receiver v oltage p ower plot for operating frequ ency from 235 kHz to 240 kHz in steps of 1 kHz Figure 8 9. Linear relationship between supply current and power delivered to the load for operating frequency from 235 kHz to 250 kHz in steps of 1 kHz. 55 60 65 70 75 80 85 0 10 20 30 40 50 60 70 80 90 100Unregulated Receiver Voltage (V)Power Delievered to Load (W) 235kHz 236kHz 237kHz 238kHz 239kHz 240kHz 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 0 10 20 30 40 50 60 70 80 90 100Supply Current (A)Power Delievered to Load (W) 235k 236k 237k 238k 239k 240k 242k 244k 246k 248k 250k

PAGE 157

157 Figure 8 9 shows the linear relationship between the supply current and power delivered to the load for different operating frequencies (235 kHz to 250 kHz in steps of 1 kHz). The yintercept and gradient of all the operating frequencies are different. Therefore, each operating frequency will req uire a different set of data set to describe the relationship. The fixed losses increases with operating frequency because the drain waveform deviates from the ZVS/ZDS conditions when the transistor is turned on earlier before the drain waveform settled down at ground. Therefore, more power is dissipated as heat across the transistor drawing a higher current regardless of loading conditions This results in an increase of the y -intercept In conclusion, using frequency control as a form of power and unregu lated receiver voltage control is not the optimum way. T he useable frequency range to achieve high efficiency is limited unless the value of Cshunt can be tuned as well This will increase the complexity of the system as two different parameters are being tracked concurrently Although, the power delivery is reduced with increasing operating frequency, the maximum unregulated receiver voltage during light load actually increases after the frequency crosses a certain threshold which is 244 kHz for this setup. Therefore, the scheme is only able to reduce power delivery and not reduce high voltage device stresses for all loading conditions In addition, if the operating frequency is varied across a considerable range, FCC regulations will require the system to be tested for each operating frequency. 8.3 Varying Cout to Achieve Receiver Voltage Control The last method to control the receiver voltage is to vary the value of Cout. By increasing the Cout value, Ztx will be made more inductive, reducing real power delivery to the receiver. The reduction in power delivery will reduce the unregulated receiver voltage for a specific load condition. This technique ensures a more consistent performance with respect to loa ding conditions as compared to that of varying operating frequency because the frequency is fixed.

PAGE 158

158 The trend of unregulated receiver voltage as shown in Figure 8 11 shows a predictable trend as relative to the one shown in Figure 8 6 In addition, by using a fix ed operating frequency instead of a variable operation frequency will make the FCC tests more straightforward. Figure 8 10. Efficiency -p ower plot for different Cout values. Figure 8 11. Receiver v oltage -p ower plot for different Cout values. 40 45 50 55 60 65 70 75 80 85 90 0 10 20 30 40 50 60 70 80 90 100Efficiency (%)Power Delievered to Load (W) 18.2nF 17.2nF 16 nF 15nF 30 35 40 45 50 55 60 65 70 75 80 0 10 20 30 40 50 60 70 80 90 100Unregulated Receiver Voltage (V)Power Delievered to Load (W) 18.2nF 17.2nF 16 nF 15nF

PAGE 159

159 Cout can be varied either by continuous tuning using a varactor or switching a bank of capacitors. Using a varactor across high AC voltage and current is not practical. Therefore, the more practical solution is to use discrete capacitors and relays to switch between the different capacitance values. Traditionally, high power relays are noisy and bulky but power relay technology has improved significantly that they are becoming more compact and does not produce much audible noise. An example is the P2 V23079 relay by Tyco electronics I ts dimension is 15 mm x 7 mm x 10 mm and is able to handle up to 5A current. The experimental results shown the Figure 8 8 and Figure 8 9 is based on a 2 relays design w ith the base capacitor value at 15 nF. The two extra capacitors values are 1 nF and 2.2 nF. Therefore, it is able to achieve four different Cout values at 15 nF, 16 nF, 17.2 nF and 18.2 nF. The number of steps are two to the power of the number of relays, thus more relays can be added to increase the resolution. However, it is shown later than a single relay setup might be sufficient. Figure 8 12. Linear relationship between supply current and power delivered to the load for Cout values of 15 nF, 16 nF, 17.2 nF and 18.2 nF. 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 0 10 20 30 40 50 60 70 80 90Supply Current (A)Power Delievered to Load (W) 18.2nF 17.2nF 16 nF 15 nF

PAGE 160

160 Figure 8 12 shows the linear relationship between the supply current and power delivered to the load for difference Cout values (15 nF, 16 nF, 17.2 nF and 18.2 nF). Since all of the lines are overlapping one another, they can be appr oximated to a common y -intercept and gradient value regardless of Cout value. The only different between them is the length of the line which does not affect the load tracking algorithm. Therefore, by varying Cout to control unregulated receiver voltage an d power delivery, a single set of data is sufficient to accurately track the power delivery of the wireless power system. This will simplify the system significantly. As shown in Figure 8 10, the improvement i n efficiency becomes less significant with the increase in Cout value ; little differen ce is observed for Cout values of 17.2 nF and 18.2 nF. The unregulated receiver voltage value decreases significantly with increasing Cout value, for example the unregula ted receiver voltage decreases a n average of 15 V for power level up to 50 W when the Cout is increased from 15 nF to 16 nF while limiting the maximum power delivery from 85 W to 50 W. For simplicity, it is still feasible to have only a single relay switch ing between two capacitance values of 15 nF and 16 nF. The maximum unregulated receiver voltage will be kept below 65 V with a single transition between the two states. In addition, a single set of values is sufficient to describe the supply current and power delivery relationship. This makes tracking and control simple and straightforward while keeping memory usage to the minimum. By reducing the number of times the relay switches, it will also extend the lifetime of the relay. 8.4 Conclusion By studying the three different techniques of varying power delivery and unregulated receiver voltage, the simplest and most cost effective technique is to use a single relay to control between two different Cout values. Although, varying the supply voltage is an effe ctively technique, the complexity and added cost is too high to justify unless the system is able to completely remove the regulator at the receiver. However, by doing so, a communication link is

PAGE 161

161 required to ensure the receivers voltage is stabilized and not enter into a positive feedback loop. Finally, varying the operating frequency is not a good technique to control the unregulated receiver voltage due to the limited ZVS/ZDS operating frequency range and sensitivity of the system at increased frequency. Another method which was not discussed is the perform duty cycling on the transmitter. Although, it is a simple and straightforward technique, due to the large inductances of the transmitting coil, it will take considerable amount of time for the transmi tter to reach steady state. Therefore, the duty cycling speed of the transmitter is limited. With a slow duty cycling speed, the receivers DC charging holding capacitor at the input of the regulator will need to be significantly larger in capacitance valu e to provide a smooth DC input voltage to the regulator, making it not practical.

PAGE 162

162 CHAPTER 9 SUMMARY AND FUTURE W ORK 9.1 Summary A new technique of designing a wireless power transfer system using the class E mode of operation for power transmiss ion via inductive coupling is proposed. The history of wireless power transmission, b ackground information is presented in Chapter 1 The system level design of the wireless power transfer system is discussed in Chapter 2. The design rules for t wo differen t impedance transformation network topolog ies the series -parallel topology and parallel parallel topology have been presented in Chapter 3 Instead of using complex detection schemes and variable tank circuits to seek resonance and high efficiency, the sy stem is designed to achieve the desired power delivery profile via its natural response across a wide range of load resistances. A multi -channel topology is also proposed to achieve high efficiency for different power ranges. In addition, a Matlab program is developed to automatically generate the value for each component. Extending the work of a one to one system to support multiple receivers, a switch architecture is proposed in Chapter 4 to decouple the fully charged receiver from the system so that po wer delivery to the other receiver can be improved. The switch is able to pass high power signals as well as sustain high input voltage s Simulation results of the switch using Advanced System Design by Agilent verified the performance of the switch. A du al channel system using the parallel parallel impedance transformation network topology is fabricated and tested to verify the design. The experimental results are presented in Chapter 5 The fabricated system is capable of delivering nearly 300 W with forced air cooling and the power delivery can be varied via its supply voltage. Higher power delivery can be achieved if a power supply with higher output and transistors with higher breakdow n voltages are used. With natural convection cooling, the system achieves a maximum power delivery of 69

PAGE 163

163 W with end -to end system efficiency of 74%. Another lower power system supporting multiple receivers using a single transmitting coil is also presented in Chapter 5. The system is designed for portable consumer products which are typically charged from a USB port at 5V, 500mA Experiment al results show the power choking phenomena or fish plot when two or more loads are placed on a single transmitting coi l. However when the fully charged device is decoupled from the system using the switch, power delivery to the other receiver is resume to normal. It is found from the interoperability analysis in Chapter 6 that the system is sufficiently robust to power r eceivers designed for other platforms as long as the size ratio between the receiving coil and transmitting coil is kept within the range of 1:2 to 1:12. This enables a transmitting unit designed to power a specific model of cell phone to work for other ce ll phones or even mp3 players Therefore, the consumer is not required to purchase a different transmitting unit or each device he own s m aking it a universal system, reducing wast e and power consumption. A method of load/fault detection is presented in Chapter 7 to protect the transmitter from fault mode s ranging from minor to fatal which will potentially damage the transmitter An example of a fatal fault will be having a huge piece of metal sheet being placed on top of the transmitting coil affecting the self inductance of the transmitting coil In addition, the load detection system is able to detect if a valid receiver is placed on the transmitting coil, powering down when the transmitter is not in use. This will reduce the no load power consumption sig nificantly making it more energy efficient A method of power delivery tracking scheme is also proposed to track the power delivery without a communication link. This enables the transmitter to track the charge profile of the receiver so that it can switch between different power handling mode or detect a damaged device or battery.

PAGE 164

164 Finally, three different techniques to control power delivery as well as the unregulated receiver voltage are studied in Chapter 8. Although, varying the supply voltage is an eff ective way to perform power control the implementation is too complex. Since a voltage regulat or is able to tolerate a large range of input voltages, varying Cout in a single discrete step is more than sufficient to provide control to the power delivery a nd unregulated receiver voltage so as not to damage the receiver regulator. More steps can be added to bring down the voltage if needed. In conclusion, d epending on the requirements, the system can be reconfigured to transmit power wirelessly to different devices for a wide variety of applications. This technology can be applied to rugged electronics to enable the creation of hermetically sealed units and to eliminate the problem of charging port contamination and corrosion for operation in potentially hars h conditions In environments where sparking and arching hazards exist, this technology can be applied to eliminate an electronic devices external metallic contacts This technology can also be implemented in our everyday portable devices from cell phones to laptops for the convenience of the everyday consumer 9.2 Future Work Further extension of the load/fault detection scheme can be explored using more robust pattern recognition techniques to determine the number of receivers a transmitter is powering as well as their load profile. Although, t he work presented is limited to c lose proximity of not more than 5 cm range and power handling of up to 300 W t he design rules and concepts presented in this dissertation laid the groundwork on how a wireless power system using magnetic induction can be designed Using the same principle s the system can be expanded to support power delivery for longer range and higher power levels e.g. 1 foot distance and >1 kW of power handling. This will enable many new technologies. One of which is contactless electric vehicle charging for which vehicl es can be charged both in the garage at home or at work with a coil

PAGE 165

165 embedded into the ground. The system will not require t he owner to make an effort to connect a charging cable. The same technology can be applied to powering robotics at factories or wareh ouses so that they do not need to carry heavy batteries or to power small in body or onbody sensors to monitor a patient vital signs. Therefore, there are various applied research topics that can be proposed base on this work. In addition, using the same pair of coils near field high speed communication systems with data rates beyond 1 Gbps can be designed using wideband carrierless communication system.

PAGE 166

166 APPENDIX A FCC REGULATIONS FCC regulations require any wireless power system via magnetic induction to pass 3 separate tests. They are Part 18.307 power line conduction test, Part 18.305 emission test and Part 15.109 part 15B test. Each individual test is intended for different purposes and will be discuss in the following sections. Part 18.307 Condu ction Test The conduction test is to ensure that limited amount of power supply noise is being fed back into the power grid. The limits of the conduction test for various frequency bands are shown in Table A 1 [32] The noise can be in the form of differential mode noise and common mode noise. However, since a typical DC power will have charging holding and decoupling capacitor s along the power pa th, the predominant noise being fed back into the power system is common mode noise. Measurements of the conduct ion test are carried at the AC port of the DC power supply. The DC power supply is connected to a line impedance stabilization network (LISN) before connecting into the power grid. The LISN is fundamentally an electronic noise filter network so that the no ise from the device under test (DUT) is not fed back into the power grid. The filtered noise is output from the LISN typically via a 50 port which can be connected to a spectrum analyzer or an oscilloscope to analyze the noise signal. Table A 1 FCC Pa rt 18.307 conduction test limits Frequency of emission (MHz) Conducted limit (dBV) Quasi peak Average 0.15 0.5 66 to 56* 56 to 46* 0.5 5 56 46 5 30 60 50 *Decreases with the logarithm of the frequency

PAGE 167

167 Part 18.305 Emission Test FCC part 18 section 305 states that any device operating at nonISM frequency below 500 kHz is allow ed to have a maximum field strength of 15V/m at a distance of 300 m. The emission limits are more relaxed if an ISM frequency is used (25V/m). However, the worst cas e of operating in a nonISM frequency is assumed. It is not practical to conduct RF tests at 300 m range. Therefore, most measurements are conducted using a 10 m range or a 3 m range. By rule of thumb each decade reduction i n distance will result in a 40 dB increase in field strength. Therefore the maximum field strength is converted to 82 6 dB V/m at a 10 m range and 103. 5 dB V/m at a 3 m range. The measurement is typically performed using an antenna with a know n correction factor so that the output spect rum in to a 50 port can be converted from dBV to it s equivalent field strength of dBV/m. For a nonradiating system the field strength limit specified by FCC is extremely high and any wireless power system should be able to meet it unless the power leve l delivered is in excess of 1 kW. Part 15. 109 Part 15B Test Table A 2 FCC Part 15.109 emission limits at 3 m range Frequency of Emission (MHz) Field Strength (V/m) Field Strength (dBV/m) 38 88 100 40.0 88 216 150 43.5 216 960 200 46.0 Above 960 500 54.0 Since the wireless power transfer system neither sends data nor have a communication link, it is considered to be an unintentional radiator falling under Part 15B. The emission limits of Part 15B of various frequency bands at 3 m range are shown in Table A 2 [33] The intention of Part 15B is to ensure that unintentional radiation is low enough not to interfere to surrounding electronics equipment. Most unintentional radiations are due to fast switching edges of clock

PAGE 168

168 signals and high fr equency noise r iding on the supply bus. Radiation due to high speed clock edges can be easily suppressed by using a nulling resistor between the output of the gate driver and the gate of the switching transistor to slow down the clock signal. Supply noise can be easily suppressed by adding more decoupling capacitors with the appropriate self resonance frequency (SFR) and equivalent series resistance ( ESR ) specification.

PAGE 169

169 LIST OF REFERENCES 1 C. E. Greene, D. W. Harrist, J. G. Shearer, M. Migliuolo, G. W. P uschnigg, Implementation of an RF power transmission and network, U.S. Patent Application, US2007/0191075A1, Jan 23rd, 2007. 2 W. C. Brown, The History of Power Transmission by Radio Waves, IEEE Transactions on Microwave Theory and Techniques vol. 32, p p. 1230 1242, Sep 1984. 3 W. C. Brown, E. E. Eves, Beamed microwave power transmission and its application to space, IEEE Transactions on Microwave Theory and Techniques vol. 40, pp. 1239 1250, Jun 1992. 4 IEEE Std C95.1, 2005 Edition, IEEE Standard for Safety Levels with Respect to Human Exposure to Radio Frequency Electromagnetic Fields, 3 kHz to 300 GHz 5 J. F. Showrow, G. H. MacMaster and W. C. Brown, The super power CW amplitron, Microwave Journal pp. 52, Oct 1964. 6 J. W. Coltman The Transformer, IEEE Industrial Applications Magazine Jan Feb 2002, pp. 8 15. 7 Inductive cooker [Internet] Wikipedia (US); [updated 2009 May 25; cited 2009 May 27]. Available from http:// en.wikipedia.org/wiki/Induction_cooker 8 Electric Toothbrush [Internet] Wikipedia (US); [updated 2009 May 16; cited 2009 May 27]. Available from http://en.wikipedia.org/wiki/Electric_toothbrush 9 Cut Loose. Wireless power for portable devices [Internet] SplashPower Limited (UK) ; [updated 2009 May 27; cited 2009 May 27]. Available from http://www.splashpower.com/ 10. Welcome to the Evolution of Wireless Power [Internet] eCoupled/Fulton Innovation (US) ; [updated 2009 May 21; cited 2009 May 27]. Available from http://www.ecoupled.com / 11. Leading and Partnering in Wireless Power [Internet]. ConvenientPower (HK); [updated 2009 May 27; cited 2009 May 27] Available from http://www.convenientpower.com/ 12. Wireless Charging [Internet] PowerMat (Isra e l ); [updated 2009 Jan 7; cited 2009 May 27] Available from http://www.pwrmat.com/ 13. Wireless Electricity Delivered Over Distance [Internet] WiTricity (US); [updated 2009 Jan 7; cited 2009 May 27] Available from http://www.witricity.com / 14. Wireless Power [Internet] WiPower (US); [updated 2009 Apr 27; cited 2009 May 27] Available from http://www.wipower.com/

PAGE 170

170 15. Wireless Power for a Wireless World [Internet] PowerCast (US); [updated 2009 May 21 ; cited 2009 May 27] Available from http://powercastco.com/ 16. M. K. Kazimierczuk Class D voltage -switching MOSFET power amplifier, IEE Proceedings -B Vol 138, No. 6, Nov 1991, pp. 285 296. 17. N. O. Sokal and A. D. Sokal Class E A New C lass of High Efficiency Tuned Single Ended Switching Power Amplifier IEEE Journal of Solid State Circuits Vol. SC 10, No. 3, pp. 168 176, June 1975. 18. F. H. Raab, Effects of circuit variations on the class E tuned power amplifier, IEEE Journal of Soli d -State Circuits vol. 13, pp. 239 247, Apr 1978. 19. F. H. Raab, Idealized operation of the class E tuned power amplifier, IEEE Transactions on Circuits and Systems vol. 24, pp. 725 735, Dec 1977. 20. Nathan O. Sokal, Class E RF Power Amplifiers, QEX, pp. 9 20, Jan 2001. 21. X, Liu, S. Y. R. Hui, An Analysis of a Double -layer Electromagnetic Shield for a Universal Contactless Battery Charging Platform, in Proc. IEEE 36th Power Electronics Specialists Conference 16th June 2005, pp. 1767 1772. 22. C. Wang, G. A. Covic, O. H. Stielau, Power transfer capability and bifurcation phenomena of loosely coupled inductive power transfer system, IEEE Transactions on Industrial Electronics vol. 51, pp. 148 157, Feb. 2004. 23. C. Wang, G. A. Covic, O. H. Stie lau, Investigating an LCL load resonant inverter for inductive power transfer applications, IEEE Transactions on Power Electronics vol. 19, pp. 995 1002, July 2004. 24. C. Wang, O. Stielau, G. A. Covic Design consideration for a contactless electric vehi cle battery charger IEEE Transactions on Industrial Electronics vol. 52, pp. 1308 1314, Oct. 2005. 25. J. J. Casanova, Z. N. Low, J. Lin, R. Tseng, Transmitting Coil Achieving Uniform Magnetic Field Distribution for Planar Wi reless Power Transfer System, in Proc. IEEE Radio and Wireless Symposium 18th 22nd January 2009. 26. X. Liu, S. Y. R Hui, Optimal design of a hybrid winding structure for planar contactless battery charging platform, IEEE Transactions on Power Electronics vol. 23, pp. 455 463, Jan. 2008. 27. R. Laouamer, M. Brunello, J. P. Ferrieux, O. Normand, N. Bucheit, A multi resonant converter for non -contact charging with electromagnetic coupling, in Proc. 23rd International Conference on Electronics, Control and Instrumentation, Nov 1997, vol. 2, pp. 792 797.

PAGE 171

171 28. G. Gwon, D. Park, S. Choi, S. Han, Wireless charger decreased in variation of charging efficiency, International patent application, PCT/KR2006/001706, 4th May 2006. 29. Z. N. Low, R. A. Chinga, R. Tseng, J. Lin, Design and Analys is of a Loosely Coupled Planar Wireless Power Transfer System using Magnetic Induction, IEEE Transactions on Industrial Electronics vol. 5 6 pp. 1 801 1 812, May 2009 30. Minsik Ahn, ChangHo Lee, Laskar, J., CMOS High Power SPDT Switch using Multigate Structure, in Proc. IEEE International Symposium on Circuits and Systems 25th 27th May 2007, pp. 3283 3286. 31. M. Yu, R. J. Ward, G. M. Hegazi, High power RF switch MMICs development in GaN on -Si HFET technology, in Proc IEEE Radio and Wireless Symposium 22nd 24th Jan 2008, pp. 855 585. 32. United States Code of Federal Regulations (CFR) Chapter 47, Part 18, Section 307, 1st October 2001. 33. United States Code of Federal Regulations (CFR) Chapter 47, Part 1 5 Section 109, 10th July 2008.

PAGE 172

172 BIOGRAPHICAL SKETCH Zhen Ning Low was bor n ed in Singapore. He attended the Nanyang Technological University in Singapore earning the bachelors degree in electrical and electronic engineering in 2005 under the accelerated bachelors programme He worked as an intern at the Institute for Infocomm Research for half a year developing Zigbee wireless sensor networks. During his undergraduate, he joined Positioning and Wireless Technology Centre as an undergraduate research working on GPS and UWB position location systems and authored two conference papers and two journal papers. After obtaining his bachelors degree in February 2005, he joined the Institute for Infocomm Research as a Research Engineer to continue his work on Zigbee sensor networks and UWB position location s ystem. In August 2006, he came to the University of Florida to pursue his Ph.D. in electrical engineering. For his Ph.D. research he worked under the guidance of Dr Jenshan Lin in the Radio Frequency Circuits and Systems group working on wireless power tra nsfer systems. He has 18 publications in technical journals and c onferences and 6 patent applications.