MASTER'S THESIS

Optimization of Wireless Power

Oskar Rnnbck
2013

Master of Science in Engineering Technology

Engineering Physics and Electrical Engineering

Lule University of Technology

Deptartment of Computer Science, Electrical and Space Engineering

Optimization of Wireless Power

Oskar R
onnb
ack
Lule
a University of Technology
Dept. of Computer Science, Electrical and Space Engineering

December 5, 2013

A BSTRACT
Today, the limit of wireless devices lays in the way they are powered. Imagine a device
that doesnt need a charger or even a battery, which instead gets the power wirelessly
over the air. To make such a device possible the transfer distance of currently known
systems have to be increased. That will be the aim of this thesis, to investigate how
to increase the transfer distance of a wireless power system, WPS, purposed to charge
low power electronic devices. In order for the system to be usable certain design limits
are set to restricts the size of the coils, flat spiral coils with diameter < 90mm and wire
diameter < 2mm, and thereby also narrowing the scope of the thesis.
This thesis starts with a presentation of the theoretical framework behind wireless
power, including techniques for modeling a complete system. The framework is then
broken down to its basic components which generates expressions with geometrical and
material properties as variables. These expressions are implemented in Matlab creating a simulator, which finds optimal values of geometrical and material properties that
maximizes the transfer distance.
The simulator is set up and ran for each system, 2, 3 and 4 coils, this because each
system behaves differently and all have some desirable properties. The findings are
implemented in Comsol which provides verification and illustrates the electromagnetic
fields that are generated. The results from Comsol and Matlab are similar and shows
that a 2-coil system can transfer power with 40% efficiency over a distance of 150mm.
While 3- and 4-coil systems significantly improve the transfer distance and can transfer
power with the same efficiency over a distance of 350mm.
As a last step were WPSs built using the findings from the simulations. The coils were
made according to the optimal parameters and capacitors were added to tune them to
the same resonance frequency. An E-class amplifier was designed and built to excite the
transmitting coil in the real system. The measurements made are the power delivered
to the amplifier and the power delivered to the load. From that the efficiency of the
complete system can be calculated. The measurements made in this thesis dont hold
up to the simulations in the sense of transfer distance. The main reasons for that is that
the amplifier is included in the measured PTE and not in the simulations and that the
coils are not perfectly built or tuned.
iii

P REFACE
This thesis work were conducted as the last part of the Master Programme in Engineering
Physics and Electrical Engineering at Lulea University of Technology, LTU.
During my project work course I first came into contact with wireless power and I
thought is was a fascinating technology. Seeing the possibilities for wireless power it is
clear that it will play a huge part in the future of electronics. I was not aware of any
research in this area in Sweden, therefore I made the thesis work as a project on my own
initiative which I carried out at LTU.
I would like to thank Kalevi Hyyppa for his understanding and guidance and my family
for always supporting me.

Oskar Ronnback

C ONTENTS
Chapter 1 Introduction
1.1 Wireless power today . . . . . . . . . . . . .
1.2 Benefits of wireless power systems . . . . . .
1.2.1 Environmental . . . . . . . . . . . .
1.2.2 Social . . . . . . . . . . . . . . . . .
1.3 Baseline . . . . . . . . . . . . . . . . . . . .
1.4 Delimitations . . . . . . . . . . . . . . . . .
1.5 Goal . . . . . . . . . . . . . . . . . . . . . .
1.6 Outline . . . . . . . . . . . . . . . . . . . . .
1.7 Frequently used variables and abbreviations

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

Chapter 2 Theory
2.1 Resistance . . . . . . . . . . . . . . . . . . . . .
2.1.1 Resistance in a wire . . . . . . . . . . . .
2.1.2 Litz wire resistance . . . . . . . . . . . .
2.2 Inductance . . . . . . . . . . . . . . . . . . . . .
2.2.1 Self inductance . . . . . . . . . . . . . .
2.2.2 Inductance of pancake coil . . . . . . . .
2.2.3 Inductor quality factor . . . . . . . . . .
2.2.4 Mutual inductance . . . . . . . . . . . .
2.2.5 Coupling coefficient . . . . . . . . . . . .
2.3 Induction . . . . . . . . . . . . . . . . . . . . .
2.4 Resonance . . . . . . . . . . . . . . . . . . . . .
2.4.1 Electrical Resonance . . . . . . . . . . .
2.5 Wireless power using magnetic resonance . . . .
2.6 Coupled Mode Theory . . . . . . . . . . . . . .
2.6.1 Lossy model with source excitation . . .
2.6.2 Model of lossy 2-coil coupled system . .
2.6.3 Wireless Power Transfer Efficiency . . .
2.6.4 WPT expanded to 3- and 4-coil systems
2.7 Reflected Load Theory . . . . . . . . . . . . . .
2.7.1 Expanded for m-coil systems . . . . . . .
2.8 Unified theory . . . . . . . . . . . . . . . . . . .

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

1
1
2
2
2
3
3
4
4
4

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

7
7
7
8
8
8
9
9
9
10
10
11
11
12
12
13
13
14
14
15
15
16

Chapter 3 Simulations
3.1 Matlab simulations . . . . . . . .
3.1.1 Finding the optimal PTE
3.1.2 Simulation results . . . . .
2-coil system . . . . . . .
3-coil system . . . . . . .
4-coil system . . . . . . .
3.2 Comsol . . . . . . . . . . . . . . .
3.2.1 Simulation setup . . . . .
3.2.2 Simulation results . . . . .
2-coil system . . . . . . .
3-coil system . . . . . . .
4-coil system . . . . . . .
Chapter 4 Electronic design
4.1 Tuning of the coils . . . . .
4.2 Source . . . . . . . . . . . .
4.3 E-class amplifier . . . . . . .
4.4 Simulation . . . . . . . . . .
4.4.1 Component selection
4.4.2 PSpice . . . . . . . .

.
.
.
.
.
.

.
.
.
.
.
.

Chapter 5 Real testing

5.1 Measurement setup . . . . . . .
5.2 Test procedure . . . . . . . . .
5.3 Measurements on 2-coil system
5.4 Measurement on 3-coil system .
5.5 Measurements on 4-coil system
Chapter 6 Discussion
6.1 Simulations . . . .
6.2 Real tests . . . . .
6.3 Conclusions . . . .
6.4 Health issues . . .
6.5 Future work . . . .

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

viii

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

17
17
18
19
20
21
22
23
23
24
24
26
28

.
.
.
.
.
.

31
31
31
32
33
33
34

.
.
.
.
.

37
37
37
38
39
40

.
.
.
.
.

41
41
42
42
43
43

C HAPTER 1
Introduction
Wireless power is an old concept, Nikola Tesla experimented with it in the late 1800s.
He was considering it as an alternative to building the electric grid. History tells us
that wireless power were never realized at a consumer level and the concept was almost
forgotten. Induction stoves and transformers transfers power wirelessly and have been
around for some time but they all work over negligible distances. In 2007 scientists at
MIT issued a press release describing how to transfer power wirelessly using magnetic
resonance and presented results of transfer distances up to a couple of meters [1] . Since
then interest in this technology have boomed and it is easy to see why. Wireless power
could be used in a wide range of applications stretching from mobile devices cell phones,
tablets, laptops, sensors, medical implants to electric cars, trains and buses. Estimations
indicate that wireless power could be a billion dollar industry within the next 10 years.

1.1

Wireless power today

Today, six years since that press release hardly any products have hit the market. At least
not using magnetic resonance or who can transfer the power over a significant distance.
The scientists at MIT that was behind this technology started a company, WiTricity,
to commercialize their discovery. Since then most of their work are kept secret and
protected by hundreds of patents. Their primary targets are OEMs that can embed
their technology directly into their products. But no such products have been released
yet. They have four own products, all using magnetic resonance [2]. Three of them are
low power development kits aimed to showcase the technology for developers. The fourth
one is a high power system for charging electrical cars.
The products that are starting to pop up are charge pads/mats most of them uses the
qi standard which is created by the Wireless Power Consortium [3], WPC. The WPC
consists of over 140 members including industry leaders in mobile phones, batteries and
consumer electronics. Their qi standard is made for low power wireless charging, <5W,
1

Introduction

and specifies coil geometries, frequencies, communication, control and electric sources.
The standard enables some design freedoms and is said to work with both direct induction
and magnetic resonance. Most products today uses the first technique and the maximum
transfer distance for a qi product today is 4cm. The WPC are working on a standard for
medium power < 120W, but the specification for that is not made public yet. Their goal
is to make worldwide standards for wireless power which is compatible for all devices,
similar to Wi-Fi.
Many of the large companies are doing their own research in this area e.g. Apple, Qualcomm, Duracell and Texas Instrument. But most is kept secret and the only available
products are a few development kits and short distance charge pads/mats.
There are research going on in a wide range of other applications also, from medical
implants, consumer electronics to electrical cars and electric roads [4]. The medical
implant research focuses on low power transfer using small coils. A 2-coil solution using
direct induction, similar to a transformer, has been present for some time. Research
for implementing a 4-coil system which is much more efficient and can work over longer
distances have been made [5].

1.2
1.2.1

Benefits of wireless power systems

Environmental

One can argue that wireless chargers are not environmental friendly because they have
lower efficiency than regular chargers i.e. will consume more power while charging. Be
that as it may, wireless chargers can be made with high efficiency similar to regular
chargers. But the biggest benefit will come if a global medium-range-wireless standard
is implemented. A standard that enables charing of all mobile devices, phone, tablet,
laptop, sensors etc, using the same charger. Then all devices dont need to have an own
charger, which in turn will save a lot of resources and energy. Another benefit is in
the battery area, sensors, remotes etc could run without batteries and phones, tablets
etc could be fitted with smaller ones because they would be charged in many places.
Decreasing the need for batteries will have a big effect on the environment because of the
hazardous materials they are made of.

1.2.2

Social

In todays society a lot of people, especially young, carry their chargers with them most
of the time. This is because the battery of todays phones dont last the entire day. If
universal medium-range-wireless systems are developed. There could be charge zones
everywhere e.g in cars, coffee shops, class rooms etc. and thereby eliminate this need.
Similar advantages to WiFi could be achieved and there would be a truly wireless society.

1.3. Baseline

1.3

Baseline

Published work in this area utilizes different technologies, coils sizes and load resistance
making a baseline for the transfer distance a bit hard to set. The qi standard have reached
a maximum of 4cm but it is not specified if direct induction or magnetic resonance is used
for that case [3]. Other research have proven that adding 1-2 resonating coils /repeaters
could significantly improve the transfer distance and power can be transfered up to two
times the coil radius with reasonable efficiency [1], [6], [7], [8]. WiTricity [2] claims to be
able to transfer the power up to a couple of feet but no values or coil setups are published.
A common conception is that it is possible to transfer power a distance a couple of times
longer than the coil diameter. A good baseline would therefore be a transfer distance
twice the coil diameter, which is limited to 90mm in this thesis, thus 180mm.

1.4

Delimitations

A system that charges mobile devices has to be somewhat small in size, no one will use
a bulky system, preferably the receiver is small enough to embed in the product. Mobile
devices are primary flat and therefore the coils have to be flat as well, flat spiral coils
were then a natural choice. The coil and wire diameters dont have an obvious limit
and were chosen to be 90mm and 2mm, which seemed reasonable. Litz wire were also
simulated but is was only available with diameter of 0.78mm.
The frequency of the AC current that drives the system is chosen to 2MHz and kept
constant. This because the gate driver ,which drives the E-class amplifier in the real
tests, doesnt support frequencies higher than 2MHz and the amplifier has to be designed
around a specific frequency. The load resistance is chosen to 10 which is a ballpark value
of the resistance in mobile devices. Table 1.1 summarizes all the design limits.
Coil diameter
Coil type
Wire type
Magnet wire diameter
Litz wire diameter
Load Resistance
Frequency

< 90mm
Pancake, flat spiral
Litz/magnet
< 2mm
0.78mm
10
2MHz

Table 1.1: Design limits

1.5

Introduction

Goal

The goal with this thesis is to find a way to optimize the transfer distance and by doing
that design a system that can transfer power longer than twice the coil diameter.

1.6

Outline

The outline of this thesis will be as follows, starting in Ch. 2 with a presentation of the
basic theories of resistance, inductance, induction, resonance as well as techniques for
modeling WPSs. In Ch. 3 this framework is implemented in Matlab [9], in an attempt
to find optimal parameters for increasing the transfer distance for WPSs. The system,
with optimal parameters, is then also simulated in Comsol [10], to verify the results and
provide data on the magnetic and electric fields. Ch. 4 describes the electrical circuits
used for the real tests described in Ch. 5 and the last chapter summarizes the work and
discusses the results.

1.7

Frequently used variables and abbreviations

WPC: Wireless Power Consortium.

WPS: Wireless Power System.
PTE: Power Transfer Efficiency.
L1 : The transmitting coil inductance in a 4-coil system.
L2 : The transmitting coil inductance in 2- and 3-coil systems / resonator coil inductance
in a 4-coil system.
L3 : The receiving coil inductance in a 2-coil system / resonator coil inductance in 3- and
4-coil systems.
L4 : The receiving coil inductance in 3- and 4-coil systems.
C1,2,3,4 : The capacitance to make each coil resonate at the desired frequency.
r1,2,3,4 : Coil radius.
rw1,w2,w3,w4 : Wire radius.

1.7. Frequently used variables and abbreviations

n1,2,3,4 : Number of turns in the coil.
di,j : Distance between coil i and j, when the coils are parallel.

C HAPTER 2
Theory

This chapter describes the theoretical framework used in this thesis. Starting with basic
electrical and electromagnetic definitions, moving on to wireless power and techniques to
model wireless power systems.

2.1

Resistance

Resistance is defined as the opposition to pass current through a conductor. Losses will
always be present when a current moves through a conductor. The power dissipated by
the resistor will mainly be in the form of heat and is given by
Pdiss = V I.

2.1.1

(2.1)

Resistance in a wire

The resistance of a wire for DC or low frequencies is given by the resistivity of the material
, the cross section area A and the length of the wire l
Rdc =

l
.
A

(2.2)

When the frequency increases, the current distribution in the wire changes. It goes from
uniformly distributed to concentrated along the surface of the conductor.

The second chapter

The skin depth is a measure on how far this change has

gone and is defined as the distance from the outer surface to
where the current density has fallen to 37 % of its value at the
surface, Fig. 2.1.
r
2
=
.
(2.3)

Where is the angular frequency and is the absolute magFigure 2.1: Skin depth
netic permeability of the conductor. The resistance for high
frequency currents therefore has a more complex expression [11]


Ber(q)Bei0 (q) Bei(q)Ber0 (q)

Rac =
/m
(2.4)
Ber0 (q)2 + Bei0 (q)2
2rw
where

2rw
,
(2.5)
q=

rw is the wire radius and Ber and Bei is the real and imaginary part of the Bessel
function.

2.1.2

Litz wire resistance

Litz wires are designed for reducing the skin effect i.e reducing
the HF resistance. It consists of multiple small strands, isolated from each other and braided in a specific pattern, Fig.
2.2. The multiple strands will give a larger surface area at
high frequencies and the braiding pattern reduces the proximity effect between the strands. Calculations of the resistance
gets very complex and no available method for it was found.
An approximative method based on measurements and table
values, developed by a manufacturer, is described in [12].

2.2

Figure 2.2: Litz wire

Inductance

Inductance is a property of a conductor, the electromagnetic definition of inductance L

is the ratio of magnetic flux linkage to the current I
L=

2.2.1

.
I

(2.6)

Self inductance

In electronics, inductors make use of the principle described by Eq. 2.6. A changing
current flows through the windings of an inductor, creating a changing magnetic field.

2.2. Inductance

Each winding of the inductor captures this flux and produces an induced voltage, back
EMF, which is why it is called self inductance. According to Faradays law Eq. 2.16
the induced voltage will oppose the change in flux which gives inductors the property
of resisting changes in current. The value of the inductance L is purely defined by its
material and geometrical properties. For a circular wire coil it is approximated by [11]



k2
2
L = n 0 (2r rw ) 1
K (k) E (k)
(2.7)
2
where

s
k=

4r(r rw )
,
(2r rw )2

(2.8)

n is the number if turns, r is the radius of the coil, rw is the wire radius, 0 is the
permeability and K and E are the complete elliptic integrals.

2.2.2

Inductance of pancake coil

The inductance of pancake coils, single layer flat spiral coils, differ from Eq. 2.7. A better approximation
of the inductance in pancake coils is given by [13]
L=

r 2 n2
,
8r + 11w

(2.9)

where r is the radius to the half winding width in

inches, w is the width of the windings in inches and
n is the number of turns, Fig 2.3. The approximation
is good for frequencies below 30MHz and is correct
within 5% if w > 0.2r.

2.2.3

Inductor quality factor

Figure 2.3: Illustration of how r

and w are defined for a pancake
coil

Quality factor is a measure on how ideal an inductor is and is defined by

2f L
(2.10)
R
where f is the frequency, L is the inductance and R is the wire resistance. An ideal
inductor have no resistive loss resulting in an infinite quality factor. But all real inductors
have at least some wire resistance.
Q=

2.2.4

Mutual inductance

Similar to self inductance, two inductors carrying currents I1 and I2 in close proximity
interacts magnetically. Both inductors induces a voltage in each other, this is defined by

10

The second chapter

[14]
12
21
=
,
(2.11)
I2
I1
where ij is the flux linkage form i to j. The equality M12 = M21 can be proven, by
energy concepts, for linear mediums surrounding the inductors e.g. air. The mutual
inductance can be calculated from the geometrical and material properties, similar to
the self inductance, but the main parameter is the distance between the inductors, must
be parallel and perfectly aligned, as M12 1/d312 . For two circular single turn wire coils,
with radius r1 and r2 , the mutual inductance is given by



r
2
2
r1 r2
M12 =
K() E() ;
=2
(2.12)
r1 r2 1

2
(r1 + r2 )2 + d212
M12 = M21 =

Z
K() =
0

/2

p
;
1 2 sin2 ()

/2

E() =

p
1 2 sin2 ()d

(2.13)

For coils with multiple turns the mutual inductance becomes

Mtot =

N1 X
N2
X

Mij .

(2.14)

i=1 j=1

2.2.5

Coupling coefficient

From self and mutual inductance can a coupling coefficient k be derived. It is a measure
on how much two coils interact, 1 being totally interacted and 0 no interaction
M12
.
k=
L1 L2

2.3

(2.15)

Induction

Induction is the creation of voltage, electromotive force, when a conductor is placed

inside a time varying magnetic field. If the conductor forms a closed circuit then this is
expressed by Faradays law [14]
d
Vemf = N
,
(2.16)
dt
where N is the number of turns in the circuit and is the flux through each turn. The
negative sign indicates that the induced voltage acts in such a way to oppose the flux
producing it. Eq. 2.16 can be written in terms of the magnetic field B
Z
d
Vemf = N
B dS,
(2.17)
dt S
where S is the surface area of the closed circuit. A transformer works according to this
concept; a time varying current is applied to the primary windings causing a time varying

2.4. Resonance

11

magnetic field, the secondary windings placed inside this magnetic field will have an
induced voltage. When an electrical load is applied to the secondary windings, current
will flow. In order to make this transfer as efficiently as possible, all of the magnetic
field created has to flow through the secondary windings making the distances of power
transfer negligible.

2.4

Resonance

Resonance occurs in many areas of physics and describes the tendency of a system to
oscillate with larger amplitude at some specific frequencies. The response of a resonant
system depends highly on the physical parameters of the system. The intensity of a
lightly damped linear oscillating system can often be approximated with the formula [15]
I()

2
2
2

( ) +

2
2

(2.18)

where is the resonance frequency and is a parameter depending on the damping.

2.4.1

Electrical Resonance

In electrical systems resonance occurs at a particular frequency where the imaginary

parts of the systems impedance cancels out. An example of this is the series RLC circuit,
Fig. 2.4.
R
Vs

Figure 2.4: Series RLC-circuit

The impedance is given by

Z(j) = R +

1 2 LC
.
jC

(2.19)

At the frequency = 1/ LC the imaginary parts cancels and the system starts to
resonate.

12

2.5

The second chapter

Wireless power using magnetic resonance

To transfer power with magnetic resonance capacitors are added to the transmitting
and receiving circuits according to the circuit diagram in Fig. 2.6. These capacitors
tunes the circuits to achieve resonance. When the oscillating source, Vs , excites the
transmitter circuit energy is stored in the transmitter. The transmitter is coupled to the
receiver by their mutual inductance, an analogy for the energy transfer can be made using
two pendulums connected by a spring [8]. In the case of the pendulums, the spring is
equivalent to the mutual inductance or coupling between the coils. The stiffness/coupling
determines how much energy is transfered in each cylce, the rate of the energy transfer.
The efficiency is not affected by the spring/coupling, it is only defined by the losses in
the system, friction/winding resistance, etc. That is the major difference between direct
induction and magnetic resonance. When you extract work from the receiver you add
constrains to the system, the amount of power transfered to the receiver must be enough
to drive the load else the magnitude of the oscillation will decrease. This gives the system
a region where there is an equilibrium and beyond that region the system can not drive
the load at maximum efficiency and the magnitude decays. A more in depth analysis is
found in [1].

2.6

Coupled Mode Theory

Coupled mode theory, CMT, is an analytic tool for systems involving interacting oscillations and leads to solutions for oscillating and propagating waves [16]. Therefore it is
used to model wireless power system that uses magnetic resonance.
Considering an ideal LC circuit, Fig. 2.5, two coupled first order differential equations
can be stated
i
v
L

Figure 2.5: Parallel LC-circuit

di
,
(2.20)
dt
dv
i = C .
(2.21)
dt
These equations, 2.20 and 2.21, can be combined to a second order differential equation
v=L

d2 v
+ 2 v = 0,
dt2

(2.22)

2.6. Coupled Mode Theory

13

where the resonant frequency is = 1/ LC. Using coupled mode theory a complex
amplitude is defined as
!
r
r
C
L
v(t) j
i(t) ,
(2.23)
a(t) =
2
C
where the energy stored in the circuit is then given by |a|2 . Eq. 2.22 can then be stated
as one first order differential equation
da(t)
= ja(t).
dt

2.6.1

(2.24)

Lossy model with source excitation

To be able to consider a real system, the model Eq. 2.24 has to be expanded to account
for losses and source excitation. The losses are represented by which is the rate of
decay and the excitation by Fs , where |s|2 is the input power. The model then becomes
da(t)
= (j + ) a(t) + Fs (t).
dt

2.6.2

(2.25)

Model of lossy 2-coil coupled system

Rs

C2

C3

k23
L2

L3

Vs

RL
R2

R3

Figure 2.6: Circuit diagram of 2-coil system

A resonant circuit is added as a load to the system described in Eq. 2.25, Fig 2.6, the
transmitting circuit is denoted by the index 2 and receiving by 3. The two circuits are
coupling to each other by the term k and the load resistor is taken into account by L .
The complete system can now be described by
da2 (t)
= (j2 + 2 ) a2 (t) + Fs (t) + jka3 (t),
dt

(2.26)

da3 (t)
= (j3 + 3 + L ) a3 (t) + jka2 (t),
dt

(2.27)

14

The second chapter

2.6.3

Wireless Power Transfer Efficiency

In order to calculate the efficiency of the power transfer, the theory of energy conservation
is applied. If the radiated power in the near field is neglected the following statement
can be made [7]
PS = P2 + P3 + P L ,
(2.28)
where the average power in each circuit, coil and capacitor, is
Pi = 2i |ai |2

(2.29)

PL = 2L |aL |2 .

(2.30)

and in the load

With Eq. 2.28 and 2.34 the efficiency can be stated as
2coil =

PL
=
Ps

1
1+

3
L


1+

2 3
2
K23

1+

L
3

2  ,

(2.31)

where K23 is the coupling rate.

2.6.4

WPT expanded to 3- and 4-coil systems

Rs

C2

k23

k34

L2
Vs

C4

L3 L4
C3

RL

R2

R3 R4

Figure 2.7: Circuit diagram of 3-coil system

Rs

C1

k12

k23

L1 L2
Vs

k34
L3 L4

C2 C3
R1 R2

C4

RL
R3 R4

Figure 2.8: Circuit diagram of 4-coil system

The efficiency can be significantly improved at larger distances by using more than two
coils [1], [6], [7],[8]. The law of energy conservation can be stated for an arbitrary number

2.7. Reflected Load Theory

15

of coils
PS =

m
X

Pi + PL .

(2.32)

i=1

Form Eq. 2.32 an expression of the efficiency for a m-coil system can be derived in a
similar way as for the 2-coil system
mcoil =

PL
=
Ps

m + L +

L
Pm1
i=1

 2 .
 
i  AAmi 

(2.33)

In the 3-coil system an extra load circuit is added for impedance matching, Fig 2.7. From
Eq. 2.33 the 3-coil efficiency given by
3coil =

2.7

2
2 [K34

K23 K34 L
.


2
2
2
(4 + L )
K34
3 (4 + L )2 + K23
+ 3 (4 + L )] + K23
2

(2.34)

Reflected Load Theory

Reflected load theory, RLT, is widely used by electrical engineers to analyze transformers
but can also be applied to WPSs. The theory is based on that the current in the primary
coil is dependent on the load in the secondary coil. The load that is reflected to the
primary coil is not the same value as the load present in the secondary coil. It can be
shown that the
is found when both coils is tuned to the same resonance
highest PTE

frequency 1/ L2 c2 = 1/ L3 c3 . At resonance the reflected load will appear as function

of the mutual induction between the coils [7]
2
Rref = k23
L2 QL3

(2.35)

where k23 is the coupling coefficient Eq. 2.15, QL3 = Q3 QL /(Q3 + QL ) and QL =
RL /L3 . The power applied to the primary coil will then be divided between R2 and
Rref . The power transfered to the second coil will be divided between the load and the
coil resistance. From this the PTE, from source to load, can be derived
2 =

2.7.1

2
k23
Q2 Q3L
Q3L

.
2
1 + k23
Q2 Q3L QL

(2.36)

Expanded for m-coil systems

Assuming that the coupling between non-neighboring coils is negligible, the partial i,i+1
can be stated
2
ki,i+1
Qi Q(i+1)L
i,i+1 =
.
(2.37)
2
1 + ki,i+1
Qi Q(i+1)L
The PTE for the full system with m coils is achieved by
m =

m1
Y
i=1

i,i+1

Q3L
.
QL

(2.38)

16

2.8

The second chapter

Unified theory

It can be shown that both CMT and RLT will result in the same steady state equations
for the PTE [7], [17]. The rate of decay in CMT can be expressed as i = /2Qi and
the coupling rate as Kij = kij /2. Substituting this in Eq. 2.33 will lead to the same
expression as for the RLT Eq. 2.38. The PTE of a 3-coil system, Fig. 2.7, can be
expressed by setting m = 3 in Eq. 2.38
2 2
Q
k23
k34 Q2 Q3 Q4L
i 4L
3 = h
2
QL
2
2
2
Q3 Q4L )
Q2 Q3 (1 + k23
Q3 Q4L ) + k23
(1 + k23

(2.39)

and The PTE of a 4-coil system, Fig. 2.8, by setting m = 4 in Eq. 2.38
4 =

2
2
2
Q4L
(k12
Q1 Q2 ) (k23
Q2 Q3 ) (k34
Q3 Q4L )

(2.40)
2
2
2
2
2
[(1 + k12 Q1 Q2 ) (1 + k34 Q3 Q4L ) + k23 Q2 Q3 ] [1 + k23 Q2 Q3 + k34 Q3 Q4L ] QL

C HAPTER 3
Simulations
Simulations of the PTE are made both in Matlab [9] and in Comsol Multiphysics [10].
Matlab is a numerical computing environment, ideal for implementing analytical expressions and combining them into a simulator. Comsol is a finite element analysis, solver
and simulation software. It can combine multiple physics into the same solution, making
it very useful for this kind of problem.
The framework presented in Ch. 2 is implemented in Matlab creating a simulator, which
is designed to find optimal geometrical parameters for increasing the transfer distance.
The results are then implemented in Comsol to verify the model and to get a picture of
the electromagnetic fields.

3.1

Matlab simulations

The simulator is built around the PTE as a function of the transfer distance, Eq. 2.40,
2.39 and 2.36. But the PTE is also dependent on the frequency f , the load RL and for each
coil the quality factor Qi and the coupling coefficient kij . The task for the simulations is
then to find an optimal set of these parameters, which gives as long transfer distance as
possible and on the same time keeps the efficiency at reasonable levels. These parameters
can be broken down further, Eq. 2.10 shows that the quality factor is dependent on f ,
the coil inductance Li and the coil resistance Ri . The coupling coefficient kij given by Eq.
2.15 depends on coil inductances Li and Lj and the distance between the coils dij . The
coil inductance and resistance, Eq. 2.9 and 2.4, can be broken down to its geometrical
and material properties, where the parameters are coil type, coil radius, wire radius, wire
type, frequency and material.
Design limits was set in order to keep the system small, use materials that are reasonable
priced and available and narrow the scope of the simulations. To keep the system small
and have the possibility to embed it in a mobile device, the coil radius was limited to
45mm. All mobile devices today are flat in some sense therefore the flat coil type, pancake

17

18

The third chapter

coil, is a natural choice. The thickness of the coil is then only dependent on the wire
radius, which in turn is limited to 1mm for magnet wires and for litz wires there are only
one radius, 0.39mm, that is considered. All wires are made of copper, other wires like
silver plated or super conductive could make a huge difference but their unavailability
and price excludes them as an option in this thesis. Due to limitations in the gate drivers
and the design of the amplifier, for the real tests, the frequency is set at a constant 2MHz.
Most mobile devices charges at 5V with 0.5-1A, making 10 a good choice for the load
resistance. All the limits are summarizes in Table 3.1.
Coil radius
Coil type
Wire type
Magnet wire diameter
Litz wire diameter
Wire material
Load Resistance
Frequency

<45mm
Pancake
Litz/magnet
<2mm
0.78mm
Cu
10
2MHz

Table 3.1: Design limits

Collecting all parameters and using the limits in Table 3.1 gives the following set of
parameters for optimization, for each coil: radius, wire radius and number of turns.
Systems with three or four coils will have additional parameters in form of distances
between coils. These parameters are input to the simulator and it calculates all possible
combinations to find the optimal set. The simulations are made with an ideal source and
with added source resistance. The added resistance will compensate for non ideal source
properties.

3.1.1

Finding the optimal PTE

To setup the simulator to find an optimal PTE, a definition on what an optimal PTE
is has to be made first. This is not as straight forward as you might think, by varying
all the parameters very different PTE curves can be achieved, Figures 3.1a - 3.1c shows
examples of how theses curves can look.
It becomes clear that a compromise between a curve that has a short but high plateau,
curves with a maximum shifted from the origin and curves with a long and low plateau
have to be made. If the PTE is integrated over the transfer distance you will get a
measure of the total efficiency over the distance. That is a good starting point, but
because curves with high and short plateaus have much higher values on short distances
there will not be a good compromise just by looking at the biggest total efficiency e.g.
Fig 3.1a, high and short, has a bigger total efficiency than Fig 3.1b, low and wide, but

3.1. Matlab simulations

19

100

100

90

90

80

80

70

70

60

60

60

50

40

PTE [%]

PTE [%]

PTE [%]

50

50

40

40

30

30

20

20

10

10

30

20

10

0.1

0.2

0.3

0.4

0.5
d23 [m]

0.6

0.7

0.8

0.9

(a) High and short plateau

0.1

0.2

0.3

0.4

0.5
d23 [m]

0.6

0.7

0.8

0.9

(b) Low and wide plateau

0.1

0.2

0.3

0.4

0.5
d23 [m]

0.6

0.7

0.8

0.9

(c) Shifted maximum

Figure 3.1: Examples of PTE curves

on the same time has a shorter transfer distance at 40% efficiency. To make it better the
integrating interval is narrowed, the best compromise is found with an interval of 0.2-1m.
With the definition of the optimal PTE, the simulator calculates this value for all
possible combinations of the parameters. The combination with the highest value will
be the optimal set.

3.1.2

Simulation results

There are two cases for all systems, coils made of magnet- and litz-wire. The Matlab
simulator calculates all possible combinations of the parameters to find an optimal set.
Figures 3.2 - 3.7 shows the optimal PTE for each case and system, for three different
source resistances.

20

The third chapter

2-coil system
100
Rs = 0

Design choices
Coil type Pancake
Wire type Cu, singel strand
Frequency 2MHz
Optimal parameters
r2
45mm
r3
45mm
rw2
1mm
rw3
1mm
n2
20
n3
1

Rs = 1

90

Rs = 10
80

70

PTE [%]

60

50

40

30

20

10

0.1

0.2

0.3

0.4

0.5
d23 [m]

0.6

0.7

0.8

0.9

Figure 3.2: PTE for the magnet wire 2-coil system

with different source resistances

Table 3.2: Parameters used for

the magnet wire 2-coil system

100
Rs = 0

Design choices
Coil type Pancake
Wire type Cu, Litz
Frequency 2MHz
rw2
0.39mm
rw3
0.39mm
Optimal parameters
r2
45mm
r3
45mm
n2
20
n3
1

Rs = 1

90

R = 10
s

80

70

PTE [%]

60

50

40

30

20

10

0.1

0.2

0.3

0.4

0.5
d23 [m]

0.6

0.7

0.8

0.9

Figure 3.3: PTE for the litz wire 2-coil system with
different source resistances

Table 3.3: Parameters used for

the litz wire 2-coil system

3.1. Matlab simulations

21

3-coil system

80
Rs = 0
Rs = 1

70

Rs = 10

Design choices
Coil type Pancake
Wire type Cu, singel strand
Frequency 2MHz
Optimal parameters
d34
31mm
r2 , r3 , r4
45mm
rw2 , rw3
1mm
rw4
0.4mm
n2 , n3 , n4 20

60

PTE [%]

50

40

30

20

10

0.1

0.2

0.3

0.4

0.5
d23 [m]

0.6

0.7

0.8

0.9

Figure 3.4: PTE for the magnet wire 3-coil system

with different source resistances

Table 3.4: Parameters used for

the magnet wire 3-coil system

80
Rs = 0

Design choices
Coil type Pancake
Wire type Cu, Litz
Frequency 2MHz
rw2 , rw3
0.39mm
Optimal parameters
d34
31mm
r2 , r3
45mm
r4
33mm
n2 , n3
20
n4
14

Rs = 1

70

Rs = 10

60

PTE [%]

50

40

30

20

10

0.1

0.2

0.3

0.4

0.5
d23 [m]

0.6

0.7

0.8

0.9

Figure 3.5: PTE for the litz wire 3-coil system with
different source resistances

Table 3.5: Parameters used for

the litz wire 3-coil system

22

The third chapter

4-coil system
80
Rs = 0

Design choices
Coil type
Pancake
Wire type
Cu, singel strand
Frequency
2MHz
Optimal parameters
d12
1mm
d34
31mm
r1 , r2 , r3 , r4
45mm
rw1 , rw2 , rw3 1mm
rw4
0.4mm
n1 , n2 , n3 , n4 20

Rs = 1

70

Rs = 10

60

PTE [%]

50

40

30

20

10

0.1

0.2

0.3

0.4

0.5
d23 [m]

0.6

0.7

0.8

0.9

Figure 3.6: PTE for the magnet wire 4-coil system

with different source resistances

Design choices
Coil type
Pancake
Wire type
Cu, Litz
Frequency
2MHz
rw1 , rw2 , rw2 , rw2 0.39mm
Optimal parameters
d12
1mm
d34
31mm
r1 , r2 , r3
45mm
r4
33mm
n1 , n2 , n3
20
n4
14

80
Rs = 0
Rs = 1

70

Rs = 10

60

PTE [%]

50

40

30

20

10

0.1

0.2

0.3

0.4

0.5
d23 [m]

0.6

0.7

0.8

0.9

Table 3.6: Parameters used for

the magnet wire 4-coil system

Figure 3.7: PTE for the litz wire 4-coil system with
different source resistances

Table 3.7: Parameters used for

the litz wire 4-coil system

3.2. Comsol

3.2

23

Comsol

To make simulations in Comsol you have to have a CAD model of the system, it can
be in 2D or 3D. Comsol provides and internal CAD software, it also enables geometrical
properties as parameters that can be swept. 2D models are simple and quite fast to simulate, which works well for some systems. 3D models require more boundary conditions
and larger mesh areas, which makes them more complex therefore also computationally
heavy. As a middle ground, there is 2D-axisymetrical. It will revolve a 2D solution
around a symmetry axis creating a 3D model. This will let you have the benefits of both
models, but the system have to have a symmetry axis.

3.2.1

Simulation setup

The model of the power transfer system has symmetry along the separation axis and can
therefore be modeled as 2D-axisymetrical.
The system is drawn up in 2D, the coils are made of rectangles with rounded corners.
When revolved they make discs with a hole in the center. This will emulate a pancake
coil made of wire, wound in a spiral pattern. The Comsol simulations are set up to use
Magnetic Fields for the coils and the energy transfer. The discs are set to multi-turn coil
domains, which makes them behave as they are made of wire. The coils are set to be
made of copper and the surroundings of air. The mesh is set up using boundary layers
for the coil, this is because the coils will suffer from skin effect and therefore most of
the current will be on the surface and it is good with small mesh to not lose accuracy.
The surroundings will have a fine triangular mesh. The rest of the system, source, load
and resonance capacitors are simulated as Electrical circuits. The components are added
and connected by their node numbers. The optimal parameters for the systems, found
in Matlab, are implemented. The resonator capacitors have to be tuned separately for
each coil, to achieve resonance everywhere. The distance between the coils are then being
swept and the efficiency is plotted as power received/power sent, the data is then exported
and plotted in Matlab. The magnetic and electric field plots are made by revolving the
2D model and then looking at the fields in a plane.

24

3.2.2

The third chapter

Simulation results

2-coil system
Simulations are made with the optimal parameters for the 2-coil system, Table 3.2.
Figure 3.8 shows an overview of the 2-coil setup. The disc in the top right corner is the
transmitter and the one in the bottom left corner is the receiver. The electrical circuits
used for the simulation is shown in Fig. 3.9, where L2 and L3 are the coils from Fig. 3.8.

Figure 3.8: Representative 3D picture of the 2-coil system

Rs

C2

C3

k23
L2

L3

R2

R3

Vs

RL

Figure 3.9: The electrical circuits simulated in Comsol, where L2,3 are the coils in Fig. 3.8

Figure 3.10 shows the PTE as a function of the distance between the transmitter and
receiver. Figures 3.11a and 3.11b shows the electric and magnetic field when d23 is 80mm.

3.2. Comsol

25
100

90

80

70

PTE [%]

60

50

40

30

20

10

0.1

0.2

0.3

0.4

0.5
d23 [m]

0.6

0.7

0.8

0.9

Figure 3.10: PTE for the 2-coil system simulated in Comsol

(a) Electric filed

(b) Magnetic filed

Figure 3.11: Comsol simulations of the 2-coil system with a transfer distance, d23 , of 80mm

26

The third chapter

3-coil system
Simulations are made with the optimal parameters for the 3-coil system, Table 3.4. Figure
3.12 shows an overview of the coils in the setup. The disc in the top right corner is the
transmitter, the middle disc is the resonator and the one in the bottom left corner is the
receiver. The electrical circuits used for the simulation is shown in Fig. 3.13, where L2 ,
L3 and L4 are the coils from Fig. 3.12.

Figure 3.12: Representative 3D picture of the 3-coil system

Rs

C2

k23
L2

Vs

k34
L3 L4

C3
R2

C4

RL
R3 R4

Figure 3.13: The electrical circuits simulated in Comsol, where L2,3,4 are the coils in Fig. 3.12

Figure 3.14 shows the PTE as a function of the distance between the transmitter and
the resonating coil. Figures 3.15a and 3.15b shows the electric and magnetic field when
d23 is 80mm, the transmitter is the top coil.

3.2. Comsol

27
100

90

80

70

PTE [%]

60

50

40

30

20

10

0.1

0.2

0.3

0.4

0.5
d23 [m]

0.6

0.7

0.8

0.9

Figure 3.14: PTE of 3-coil system simulated in Comsol

(a) Electric field

(b) Magnetic field

Figure 3.15: Comsol simulations of the 3-coil system with a transfer distance, d23 , of 80mm

28

The third chapter

4-coil system
Simulations are made with the optimal parameters for the 4-coil system, Table 3.6. Figure
3.16 shows an overview of the coils in the setup. The disc in the top right corner is the
transmitter, the disc next to it and the one in the middle are resonators and the one in
the bottom left corner is the receiver. The electrical circuits used for the simulation is
shown in Figure 3.17, where L1 , L2 , L3 and L4 are the coils from Fig. 3.16.

Figure 3.16: Representative 3D picture of the 4-coil system

Rs

C1

k12

k23

L1 L2
Vs

k34
L3 L4

C2 C3
R1 R2

C4

RL
R3 R4

Figure 3.17: The electrical circuits simulated in Comsol, where L1,2,3,4 are the coils in Fig. 3.16

Figure 3.18 shows the PTE as a function of the distance between both resonating coils.
Figures 3.19a and 3.19b shows the electric and magnetic field when d23 is 80mm, the

3.2. Comsol

29

transmitter is the top coil.

100

90

80

70

PTE [%]

60

50

40

30

20

10

0.1

0.2

0.3

0.4

0.5
d23 [m]

0.6

0.7

0.8

0.9

Figure 3.18: PTE of 4-coil system simulated in Comsol

30

The third chapter

(a) Electric filed

(b) Magnetic filed

Figure 3.19: Comsol simulations of the 4-coil system with a transfer distance, d23 , of 80mm

C HAPTER 4
Electronic design

This chapter describes the design of the electrical circuits used for the real tests and
states the components that are used.

4.1

Tuning of the coils

A coil is primary inductive but the there are always wire resistance and parasitic capacitance present. The impedance of the coil will look like
Z(s) = Ls + Rw +

1
Cp s

(4.1)

At a certain frequency the inductance and capacitance will cancel each other and the coil
is defined only by the wire resistance. That is the resonance frequency and all coils in
a WPS must operate at the same resonance frequency to have efficient power transfer.
The resonance frequency is given by
1
=
.
LC

(4.2)

To be able to tune this frequency a capacitor is added either in parallel or in series with
the inductance, Eq. 4.2 is true for both cases. The added capacitor is usually much larger
than the parasitic capacitance and therefore the parasitic capacitance does not need to
be included in Eq. 4.2.

4.2

Source

The most important properties of the source is low output resistance and a good sine
output. This is because the resistance will be directly added to the wire resistance of the
primary coil, significantly decreasing the quality factor, hence the efficiency. It also has
31

32

The fourth chapter

to be able to handle high frequencies and have high output power. There are a variety
of circuits that might be suitable for this: a power oscillator, high-frequency amplifier
or a half/full bridge. A power oscillator has the advantage that it will run at resonance
frequency without any tuning. But reliable and stable oscillators is hard to make. A
bridge circuit has an external clock that sets the operating frequency which adds one
more degree of tuning but the circuit is very stable. The downside is that it will have
issues at frequencies above 1MHz. The RF amplifier also requires an external clock signal
to set the frequency but it can be made very effective and made to work with very high
frequencies. The E-class amp is therefore the choice in this thesis.

4.3

E-class amplifier

The E-class is a switch mode amplifier designed for very high power efficiency. It is
non linear in the sense that the output amplitude does not corresponds to the input
amplitude. In order to adjust the output amplitude the supply voltage can be adjusted.
It is primary made for high frequency applications and is commonly used to create carrier
waves for radio transmissions. Fig. 4.1 shows an ideal E-class circuit.
+Vcc
Lc
L
T1

Cs

RL

Figure 4.1: E-class amplifier

It consists of a RF choke Lc , a switch T1 , a shunt capacitor (which includes the transistor

capacitance) Cs , a load network L-C and a load RL . The switch is operated at the desired
output frequency, for maximum efficiency a 50% duty cycle is used. The load network
gives the required phase shift to prevent high current and high voltage at the switch
,transistor, and acts as a open circuit for the first harmonic, passing a sine wave to the
load. There are some conditions that should be fulfilled in order to make it as efficient
as possible, when the switch goes from open to closed the voltage over the switch, VT 1 is
suppose to behave as
VT 1 = 0,
(4.3)
dVT 1
= 0.
(4.4)
dt

4.4. Simulation

33

To fulfill these conditions the following expressions have been derived [18]. The power
delivered to the load is given by


0.451759 0.402444
(Vcc Vo )2
0.576801 1.001245

,
(4.5)
RL =
P
QL
Q2L
where Vcc is the supply voltage, Vo the transistor saturation voltage (0 for FET transistors), RL is the load, and QL is the loaded quality factor. The shunt capacitor is then
given by


1
0.91424 1.03175
0.6

0.99866 +
Cs =

+
,
(4.6)
2
2
QL
QL
2f RL 4 + 1 2
(2f )2 Lc
where Lc is the choke inductor and the load network



1
1
1.01468
0.2
C=
1.00121 +

2f RL QL 0.104823
QL 1.7879
(2f )2 Lc

(4.7)

and

QL RL
.
(4.8)
2f
The design choices left to do is to specify the supply voltage Vcc , the output power P
and the quality factor QL .
L=

4.4
4.4.1

Simulation
Component selection

Setting Vcc to 5V , RL to 1, Lc to 100L and QL to 10 and using Eq. 4.5-4.8 gives

the component values displayed in Table 4.1. The switch is replaced by an MOSFET
transistor, IRFB5615PBF, which has a low Rdson and small gate charge in order to be
efficient and easy to drive. A gate driver is used to drive the transistor, to make it as
close to an ideal switch as possible. The capacitors are polyester and for the resonators a
variable capacitor is used for easier tuning. All the components used are shown in Table
4.1.
Transistor
Gate driver
C
Cs
L
Load resistance

IRFB5615PBF
lm5104
6.6nF
11.5nF
1.1uH
1.37

Table 4.1: Design limits

34

4.4.2

The fourth chapter

PSpice

With all components selected the E-class amplifier is implemented in PSpice [19]. In
PSpice the transient behavior of the circuit is studied. What is important is that the
voltage and current behaves similar to Eq. 4.3 and 4.4 and that the output is a nice
sine wave. Sokal [18] describes in his report a way to fine tune the component values
to get the best transistor current and voltage curves as possible. After the tuning the
component values differ slightly from Table 4.1 and because the transistor capacitance
is parallel with the shunt capacitor is it removed from the shunt value. The load RL is
replaced by a series RLC circuit, representing the transmitting coil and capacitance. The
resistor is the coil resistance and at the resonance frequency the inductor and capacitor
will cancel out leaving only the resistance. The 100M resistor is only added to make
simulations possible and will not affect the circuit. Figure 4.2 shows the schematic of the
complete and tuned circuit. The transistor current and voltage along with the output
voltage are shown in Fig. 4.3.

0
50F
F

000uH
000

000
00

00u

50u0u00
50u0u0
00u0u0
00u0u00n
0Fu0u00n
F7u0u0uF77u
F70u0uF77u

50u

000
200

99

FuFn

2u0

0u0900u

0un00uun

200
00u0n

n
F0FFFF0FFFF

000
u0Fuuu9u

0n0
000g7g
0u9
0uuu0

Figure 4.2: PSpice schematic of the E-class amplifier

4.4. Simulation

35

Transistor voltage [V]

Transistor current [A]

5
4
3
2
1
0
1.066

1.067

1.068

1.069

1.07

1.071

1.072

1.073

1.074

1.075
4

x 10
200

Voltage [V]

100

100

200
1.06

1.062

1.064

1.066

1.068

1.07
Time [s]

1.072

1.074

1.076

1.078

1.08
4

x 10

Figure 4.3: PSpice simulation of the voltage over the transistor, the current through the transistor and the output voltage

C HAPTER 5
Real testing

This chapter contains the real tests of the WPS. The results from the simulations in Ch.
3 are tested to see how they hold up in reality.

5.1

Measurement setup

The E-class power amplifier designed in Ch. 4 is built and serves as the source for all
measurements. The supply voltage is taken from a voltage cube with a current limitation
of 2A and the clock signal used to trigger the gate driver comes form a 5V square
wave generator. The coils are made according to the optimal parameters found in the
simulations, Ch. 3, and capacitors are added and tuned to make all coils resonate at
the desired 2MHz. The transmitting circuit is connected to the E-class amplifier and the
receiving to a 10 resistor.

5.2

Test procedure

The transmitter creates a large magnetic field which makes it hard to make accurate
measurements on the transmitter. The oscilloscope probe and ground clamp creates a
wire loop in which there are induced voltage. Therefore the power delivered from the
voltage cube will be measured instead. On the receiver it is possible to use a probe
because the voltage over the load resistance is much bigger than the induced voltage in
the probe. The PTE for the real test is then (power received by load)/(power delivered
from voltage cube). The PTE will then differ from the simulations because of the losses
in the amplifier. The measurements are related to the distance between the coils as in
the simulations and are made every other centimeter to enable good plotting.

37

38

5.3

The fifth chapter

Measurements on 2-coil system

The E-class amplifier is connected to the transmitter circuit which is tuned to resonate
at 2MHz and a load resistor, 10, is connected to the receiver circuit. Fig. 5.1 show the
measurement setup and Fig. 5.2 shows the measured PTE.

Figure 5.1: Measurement setup of the 2-coil system

18

16

14

PTE [%]

12

10

0.05

0.1

0.15

0.2

0.25

0.3

d23 [m]

Figure 5.2: PTE measurements of the whole 2-coil system

0.35

5.4. Measurement on 3-coil system

5.4

39

Measurement on 3-coil system

The E-class amplifier is connected to the transmitter circuit which is tuned to resonate
at 2MHz. A load resistor, 10, is connected to the receiver circuit and a resonator circuit
is added between the transmitter and receiver at a distance of 31mm from the receiver.
Fig 5.3 show the measurement setup and Fig. 5.4 shows the measured PTE.

Figure 5.3: Measurement setup of the 3-coil system

40

35

30

PTE [%]

25

20

15

10

0.05

0.1

0.15

0.2

0.25

0.3

d23 [m]

Figure 5.4: PTE measurements of the whole 3-coil system

0.35

40

5.5

The fifth chapter

Measurements on 4-coil system

The E-class amplifier is connected to the transmitter circuit which is tuned to resonate
at 2MHz. A load resistor, 10, is connected to the receiver circuit and two resonator
coils are added between the transmitter and receiver. One resonator is placed 1mm from
the transmitter and the other 31mm from the receiver. Fig 5.5 show the measurement
setup and Fig. 5.6 shows the measured PTE.

Figure 5.5: Measurement setup of the 4-coil system

0.25

0.2

PTE [%]

0.15

0.1

0.05

0.05

0.1

0.15

0.2
d

23

0.25

0.3

[m]

Figure 5.6: PTE measurements of the whole 4-coil system

0.35

C HAPTER 6
Discussion
This chapter summarizes the work done in this thesis and discusses issues and future
work in this area. The key parts are as follows
The WPS models for 2, 3 and 4 coils are extended with the theory of the individual
components, implemented as Matlab models with actual material and geometrical
properties as input parameters.
An optimization technique is defined and a simulator is created that finds the
optimal set of input parameters in Matlab which results in the best PTE curve.
The simulator have found optimal setups for 2, 3 and 4 coil systems and the validity
of theses systems are backed up by FEM simulations of the electromagnetic fields
made in Comsol Multiphysics

6.1

Simulations

Choosing design limits that makes the system reasonable sized and using these in the
Matlab simulator for the different models gives a set of optimized parameters for each
of the three cases. A Comsol model, for each of the cases are made and the PTE and
electromagnetic fields are looked at. Comparing the PTEs from Matlab and Comsol
shows fairly good resemblance for all cases. Some deviations were to be expected due
to the fact that the system models in Matlab and some component expressions were
approximations. But the similarity of the curves and transfer distances gives the Matlab
simulations validity.
From the electromagnetic simulations it is evident that adding one or two resonating
coils really helps to couple the magnetic and electric fields to the receiver and transmitter
(4-coil system). It is due to that the resonator coils works as impedance matching systems
for the source and load resistances. The electromagnetic field plots also show that both
fields rapidly goes to zero in the free space.
41

42

The sixth chapter

The simulations with litz wire are very limited due to the lack of analytical formulas
or approximations. The only available way to model it with geometrical properties is
by tabled values from the manufacturers, therefore there are problems to sweep the wire
radius. But the simulations using 0.39mm litz wire gives PTE curves that only have
<20% shorter transfer distance than models using magnet wire with variable radius (01mm). That indicates that litz wire probably is beneficial and a proper model would be
of interest.
The extended models of the WPSs, implemented in Matlab, could be useful in a couple
of ways, an existing magnetic resonance system could be simulated just by having the
geometrical and material properties. The simulator can then be customized to suit
different needs e.g. an existing 3- or 4-coil system could be made better without changing
the coils only by finding optimal distance for the resonator coils. Or as presented in this
thesis, an optimal system can be created from the ground just by setting the design
limits.

6.2

Real tests

The measurements will always differ from the simulations because the PTE of the measurements is over the complete system including the amplifier, while the PTE of the
simulations is only between the transmitter and receiver coils.
From the figures 5.2, 5.4 and 5.6 it is evident that the measurements dont hold up
to the simulations. What can be seen in the 2- and 3-coil cases is that even though the
transfer distance doesnt hold up with the simulations the shape of the curves is similar.
The transfer distance of the 3 coil case is, as in the simulations, significantly longer than
with only two coils. This proves that adding at least one resonator coil will improve the
transfer distance.
The 4-coil case is definitely the worst with only 0.2 % efficiency at the most. But that
is not a surprise, each coil adds degrees of freedoms and uncertainties to the system.
All coils have to be carefully tuned because of the mutual inductance, the resonance
frequency of one coil depends on the other coils especially the closest one. To manually
tune four coil therefore becomes quite a difficult task, which might be a reason for the
poor results. Another reason is that the coils are not perfectly flat which makes it hard to
align the resonator coil precisely 1mm from the transmitter, thus the impedance matching
might be off.

6.3

Conclusions

The Matlab simulator gives a way to optimize the PTE for a system given the design
limits. Systems suitable for portable devices are designed which in the simulations can
transfer power up to four times the coil diameter. In that sense the goal of the thesis is

43
met but the real system can not confirm the results from the simulations.
Looking at both the simulations and the real tests it is obvious that adding a resonator
coil close to the receiver significantly improves the transfer distance. The Matlab simulation also show that adding a second resonator coil, close to the transmitter, decreases
the dependence on the source resistance enabling similar curves as with no resistance.
This could not be simulated using my Comsol model because the PTE doesnt include
the source resistance.

6.4

Health issues

A question that comes to mind when hearing about wireless power is: Is it safe? No study
regarding safety and health was made in this thesis. However, IEEE [20] and ICNIPR
[21] have set up guidelines for exposure to RF electromagnetic fields. They both conclude
that there is no established evidence that RF electromagnetic fields causes cancer, but
there is evidence that they can cause heating in body tissue and stimulate muscle and
nerve tissue. Both IEEE and ICNIPR concludes that even the most sensitive tissue is not
adversely effected when the whole body average SAR level is less than 4 [W/kg], which
corresponds to a maximum rise of 1 C in body temperature. But they both recommend
using a safety factor and therefore sets the limit at 0.08 [W/kg] for the general public.
WiTricity have made a study [22] investigating the safety of their WPT system. They
simulate SAR levels in a human body, for both their high and low power (3kW/5W)
systems. The study concludes that it is safe and that the SAR levels are below the limit
set for the general public, for both cases.
The thing you cant predict or simulate are the longterm effects of exposure RF electromagnetic fields. But with current knowledge this technology can be regarded as safe
to use in consumer products.

6.5

Future work

The Matlab simulator could be improved by adding a model that supports litz wires with
variable diameter, because of the beneficial frequency properties. Then complete litz as
well as mixed systems could be evaluated.
To confirm the Matlab model it would be a good idea to make better real tests with
more professionally made coils, exact tuning as well as doing the measurements with a
network analyzer.

R EFERENCES
[1] A. Karalis, J. Joannopoulos, and M. Soljacic, Efficient wireless non-radiative midrange energy transfer, 2007.
[2] http://www.witricity.com.
[3] http://www.wirelesspowerconsortium.com.
[4] http://olev.kaist.ac.kr/en/.
[5] A. K. RamRakhyani, S. Mirabbasi, and M. Chiao, Design and optimization of
resonance-based efficient wireless power delivery systems for biomedical implants,
2011.
[6] B. Wang, K. H. Teo, T. Nishino, W. Yerazunis, J. Barnwell and J. Zhang, Experiments on wireless power transfer with metamaterials, 2011.
[7] M. Kiani and M. Ghovanloo, The circuit theory behind coupled mode magnetic
resonance based wireless power transmission, 2012.
[8] A. P. Sample, D. A. Meyer and J. R. Smith, Analysis, experimental results, and
range adaptation of magnetically coupled resonators for wireless power transfer,
2011.
[9] MATLAB, version 8.1.0 (R2013a). Natick, Massachusetts: The MathWorks Inc.,
2013.
[10] COMSOL Multiphysics, version 4.3b. COMSOL AB, 2012.
[11] S. Ramo, J. R. Whinnery and T. Van Duzer, Field and Waves in Communication
Electronics. John Wiley & sons, 3rd ed., 1993.
[12] http://www.litzwire.com/nepdfs/Litz_Design_PDFs.pdf.
[13] H. A. Wheeler, Simple inductance formulas for radio coils, pp. 13981400, 1928.
45

46
[14] M. N. O. Sadiku, Elements of Electromagnetics. New York: Oxford University Press,
4th ed., 2007.
[15] A. E. Siegman, Lasers. University Science Books, 1986.
[16] H. A. Haus and W. Huang, Coupled-mode theory, 1991.
[17] E. Bou, E. Alarcon and J. Gutierrez, A comparison of analytical models for resonant
inductive coupling wireless power transfer, 2012.
[18] N. Sokal, Class-e rf power amplifiers, 2001.
[19] OrCAD Capture, version 16.6, PSpice Plugin. Cadence Design Systems, Inc., 2012.
[20] IEEE Std. C95.1-2005, IEEE Standard for Safety Levels with Respect to Human
Exposure to Radio Frequency Electromagnetic Fields, 3 kHz to 300 GHz,
[21] ICNIRP Guidelines, International Commission on Non-Ionizing Radiation Protection, Health Physics, 74, no. 4, Guidelines for limiting exposure to time-varying
electric, magnetic and electromagnetic fields (up to 300 ghz), pp. 494522, 1998.
[22] M. Keller, Highly Resonant Wireless Power Transfer: Safe, Efficient, and over Distance. WiTricity Corp., 2013.