Experimental Modelling of Wireless Power

MUHAMMAD ALI QAISER for DR.
MIHAI ROTARU University of Southampton ECS
UNIVERSITY OF SOUTHAMPTON
Faculty of Physical Sciences and Engineering
School of Electronics and Computer Science
Experimental modelling and simulation of

wireless power transfer systems
by
Muhammad Ali Qaiser

3rd October, 2014
http://www.eee.hku.hk/tgif/25nov2011.html
Project Supervisor: Dr. Mihai Dragos Rotaru ……………………………..
Second Examiner: Dr. Adam Prugel-Bennett ……………………………...
A dissertation submitted in partial fulfilment of the degree of

MSc Energy & Sustainability with Electrical Power Engineering
© University of Southampton 2014, all rights reserved
P a g e 1 | 60
EXPERIMENTAL MODELLING AND SIMULATION OF WIRELESS POWER TRANSFER SYSTEMS MSc Thesis 2014
UNIVERSITY OF SOUTHAMPTON
Faculty of Physical Sciences and Engineering
School of Electronics and Computer Science
Experimental modelling and simulation of

wireless power transfer systems
by
Muhammad Ali Qaiser

3rd October, 2014
Project Supervisor: Dr. Mihai Dragos Rotaru ……………………………..
Second Examiner: Dr. Adam Prugel-Bennett ……………………………...
A dissertation submitted in partial fulfilment of the degree of

MSc Energy & Sustainability with Electrical Power Engineering
© University of Southampton 2014, all rights reserved
P a g e 2 | 60
MUHAMMAD ALI QAISER for DR. MIHAI ROTARU University of Southampton ECS
ABSTRACT
This project aims to refine the work previously done in academic and industrial circles, for transfer of electrical
energy wirelessly through magnetic induction. The concept derives from Tesla coil invented over a century ago, with
principles of mutual field induction well established through operation of transformers.
The dissertation is based on the excitation of magnetic structure in LCR (series or parallel) resonating
arrangement, which allows power pickup at mid-range distances (i.e. d is few times coil diameter, but much smaller
than wavelength of energy) and prevents omnidirectional wastage. Several variables that affect this power transfer
are investigated namely axial separation, planar separation, coil geometry, magnetic shielding, resonating
arrangement and impedance matching.
The above parameters are tested primarily through a laboratory setup involving bifiliar copper pancake coils
with aluminium foil as shield, with quantitative analysis performed on the resulting dataset. This is done with the
aim of comparing the validity of results by first principles against circuit network theory. For further exploration, a
full-wave simulation of the resonant coils is executed in a commercial software to plot scattering parameters. These
results are checked against an equivalent circuit solver for extraction of coupling factors, thus leading to a AC model
of an otherwise 3D problem.
Experimentation is developed throughout to achieve results for the following goal statements: 1) numerical
laboratory study for computation of resistance and inductance of the resonator coils, 2) computed extraction of
equivalent impedances and S21 parameter for series and parallel LCR arrangements, 3) comparison of extracted
scattering parameter against AC circuit and full-wave simulations of two-coil setup, 4) investigation of effect of
shielding (e.g. for passenger safety) on resonant coupling, and the development of an equivalent circuit model for
laboratory verification, and 5) exploration of factors affecting load matching in real, lossy environment.
Findings of the project indicate that load matching can be employed to improve power transfer when shielding
and axial separation effects come into play. Industrially, this may be achieved using an automatically controlled
variable impedance or electronic phase angle management, which can cancel the reactive part and match the resistive
component of the load looking towards the source. This could also compensate for inductances introduced by external
shields; as such, experimental and mathematical analyses of shielding are conducted for a wireless power transfer
system.
The experimental approach of this project required a considerable number of days to gather laboratory data by
incremental change of variables and repeated manual recording. A large span of time was also needed to
accommodate the computational load of 3D simulator for dataset sweeps. On average, compromise between realistic
mesh detail and simplified conductor modelling led to around 30 hours per simulation sweep. The collected data and
circuit conclusions resulting from this work may be presented for consumption by the hybrid car industry, in order
to assist design and optimization of charging mat systems.
KEYWORDS
Resonant coupling, Wireless power, WPT, experimental modelling, full-wave simulation, pancake coils
P a g e 3 | 60
ACKNOWLEDGEMENTS
I dedicate this dissertation to both my loving parents, Saira Khayyam and Khayyam Qaiser, for letting me
explore my dreams and ambitions unfettered by tradition.
I dedicate this dissertation to my affectionate wife, Amina Javaid-Qaiser, who patiently encouraged me while
giving up time otherwise meant to be spent between us.
I am grateful to my supervisor, Dr. Mihai Dragos Rotaru, whose guidance and advanced research in the field of
wireless power was elementary in helping me complete this humble piece of work.
I thank the laboratory staff of Southampton-ECS department, who I bothered incessantly for assistance with
device fabrication.
Finally, I owe it to the Common Scholarship Commission, whose funding of my entire MSc study made it
possible for me to pursue my degree. It is indeed an honour.
I hope this work shall provide new, albeit minor, insights into the relatively nascent field of efficient wireless
power transmission. With it I also hope, that I have discovered a passion that shall last a lifetime, whether as an
amateur hobbyist or a professional.
In loving memory of my four grandparents who raised me into adulthood, but then all departed within a span
of four years…
P a g e 4 | 60
DOCUMENT STATISTICS
Abstract Words [477]
Content Words [15,190] (incl. footnotes, textboxes but excl. figures, tables, equations, bibliography, appendices)
Figures [55]
References [55]
Appendices [3]
Total Pages [60]
TABLE OF CONTENTS
Abstract....................................................................................................................................... 3
Chapter 1 Historical Proof of Concept ........................................................................................... 8
1.1 Pioneer Works of Nineteenth Century ........................................................................................8
1.2 Resurgence of Engineering Interest ............................................................................................9
1.3 Preceding Research Work ........................................................................................................ 10
1.4 Content Overview ................................................................................................................... 10
Chapter 2 Laboratory Setup ....................................................................................................... 11

2.1 Bifiliar Pancake Coil Fabrication ............................................................................................... 11
2.2 Introduction to Quantitative Spreadsheets............................................................................... 13
2.3 Parametric Measurement of Fabricated Coils ........................................................................... 14
2.3.1 Method A: Inductance Measurement by Voltage Split ................................................................................ 14
2.3.2 Method B: Inductance Measurement by Resonant Tank ............................................................................. 14
2.3.3 Method C: Inductance Calculation by Coil Geometry .................................................................................. 15
2.3.4 Method D: Inductance Measurement by Impedance................................................................................... 16
2.3.5 Critique for Choice of Method and Results of Measurements ..................................................................... 16
2.4 Establishment of Coupling Factors............................................................................................ 18

2.4.1 Axial Coupling between Identical Pancake Coils .......................................................................................... 18
2.4.2 Planar Coupling between Identical Pancake Coils ........................................................................................ 21
2.4.3 Axial Coupling between Non-Identically Sized Coils ..................................................................................... 23
Chapter 3 Discussion of Shielding ............................................................................................... 25

3.1 Construction and Gashing ........................................................................................................ 25
3.2 Experimental Testing of Shield ................................................................................................. 26
3.2.1 Test of Aluminium Shield as EM Barrier ....................................................................................................... 26
3.2.2 Analysis of Effect of Shield on Coupling Factor ............................................................................................ 27
3.2.3 Electrical Reduction of Shield Surface .......................................................................................................... 29
3.3 Mathematical Analysis of Conductive Shield ............................................................................ 29
Chapter 4 Discussion of Load Matching ...................................................................................... 33

4.1 Circuit Derivation .................................................................................................................... 33
4.2 Strong Coupling Regime ........................................................................................................... 34
P a g e 5 | 60
4.3 Derivation of Circuit Impedance ............................................................................................... 35

4.4 Derivation of Scattering Parameters......................................................................................... 38
4.5 Demonstration of Inductor V-Pickup Variance .......................................................................... 40
4.5.1 Explanation of scattering by V-pickup measurements ................................................................................. 40
4.5.2 Comparison of lossy environment against ideal scenario ............................................................................ 42
4.5.3 Frequency sweep data models ..................................................................................................................... 43
Chapter 5 Microwave and Circuit Simulation Environment ......................................................... 45

5.1 3D Setup Parameters ............................................................................................................... 45
5.1.1 Coil Dimensions ............................................................................................................................................ 45
5.1.2 LCR Values and Mesh Setup ......................................................................................................................... 46
5.2 Equivalent Circuit Design ......................................................................................................... 47

5.2.1 Circuit Setup and Scatter Calculation by Z .................................................................................................... 47
5.2.2 Extraction of Coupling Factors...................................................................................................................... 48
5.2.3 Power Measurements .................................................................................................................................. 49
5.2.4 Explanation of Double-Peak Effect ............................................................................................................... 49
5.2.5 Scatter Calculation by V ................................................................................................................................ 50
5.3 Impact of Shield ...................................................................................................................... 51

5.4 Quasi-static Approximation for Future Simulations................................................................... 52
Chapter 6 Conclusion ................................................................................................................. 53

Chapter 7 Bibliography .............................................................................................................. 54
Appendix 1: List of Equipment .................................................................................................... 58
Appendix 2: Modelling Worksheet Structure .............................................................................. 59
Appendix 3: Project Management .............................................................................................. 60
TABLE OF FIGURES
FIGURE 1: Nikola Tesla holding Wireless Light (1898), adapted from Electrical Experimenter (1919) [5] __________________ 8
FIGURE 2: Tesla's patent 649,621 for wireless transmission [8] ___________________________________________________ 8
FIGURE 3: Bifiliar Speaker Wire ___________________________________________________________________________ 11
FIGURE 4: Tesla's patent for EM coils (US. 512,340) ___________________________________________________________ 11
FIGURE 5: Construction setup for pancake coil fabrication ______________________________________________________ 12
FIGURE 6: Structure of bifiliar pancake coils and winding orientation of layers______________________________________ 12
FIGURE 7: Voltage-split circuit ____________________________________________________________________________ 14
FIGURE 8: Resonant tank circuit ___________________________________________________________________________ 14
FIGURE 9: Left to right – V responses at 30 kHz, 35 kHz (resonant) and 40 kHz______________________________________ 15
FIGURE 10: Multi-layer coil _______________________________________________________________________________ 15
FIGURE 11: Circuit setup and voltage triangle for I-V method ___________________________________________________ 16
FIGURE 12: MSO screenshot for I-V method _________________________________________________________________ 17
FIGURE 13: In-phase polarity of coil windings ________________________________________________________________ 18
FIGURE 14: Axial separation of pancake coils ________________________________________________________________ 19
FIGURE 15: Parasitic capacitance modelling _________________________________________________________________ 19
FIGURE 16: Coupling factors K with axial sweep for different circuit scenarios ______________________________________ 20
FIGURE 17: Planar separation ____________________________________________________________________________ 21
P a g e 6 | 60
FIGURE 18: Coupling factor K for planar separation ___________________________________________________________ 22

FIGURE 19: B-field of I-loop ______________________________________________________________________________ 22
FIGURE 20: Setup for Non-identical Coils ____________________________________________________________________ 23
FIGURE 21: Coupling investigation between non-identical ring coils ______________________________________________ 24
FIGURE 22: Investigation of effect of Al foil shield on WPT coupling ______________________________________________ 25
FIGURE 23: Simulation of shield lamination by gashing (half square on left sketch, full on right actual) __________________ 25
FIGURE 24: Circuit arrangement to check shield intervention____________________________________________________ 26
FIGURE 25: Effect of shield on wireless emf transfer and its explanation by eddy currents ____________________________ 27
FIGURE 26: Shield placement _____________________________________________________________________________ 27
FIGURE 27: K-curves for varying vicinity of shield (d) through axial separation of coils (x) _____________________________ 28
FIGURE 28: K-curve lifting by shield reduction ________________________________________________________________ 29
FIGURE 29: Eddy sectors _________________________________________________________________________________ 29
FIGURE 30: Circuit modelling of EM shield ___________________________________________________________________ 30
FIGURE 31: Trend of coil L and esr values with shield introduction________________________________________________ 32
FIGURE 32: Load matching theorem _______________________________________________________________________ 33
FIGURE 33: Power transfer peak __________________________________________________________________________ 34
FIGURE 34: Series Resonant Circuit for WPT _________________________________________________________________ 35
FIGURE 35: Parallel Resonant Circuit for WPT ________________________________________________________________ 37
FIGURE 36: Voltage gain for parallel LCR arrangement ________________________________________________________ 40
FIGURE 37: Contradictory forces in a parallel resonant WPT system ______________________________________________ 41
FIGURE 38: S21 frequency response ________________________________________________________________________ 41
FIGURE 39: Comparison of lossy (left) against ideal (right) circuit response ________________________________________ 42
FIGURE 40: V-gains for resonant and disturbed cases __________________________________________________________ 43
FIGURE 41: Impedance ratio for frequency sweep ____________________________________________________________ 43
FIGURE 42: Voltage pickup at receiver, for frequency sweep ____________________________________________________ 44
FIGURE 43: Coils V-gain for resonant frequency ______________________________________________________________ 44
FIGURE 44: Scatter behavior at resonant frequency ___________________________________________________________ 44
FIGURE 45: Inter-layer linkage structure, and dimensions of modelled coil _________________________________________ 45
FIGURE 46: Calculation of spiral length _____________________________________________________________________ 46
FIGURE 47: Lumped elements_____________________________________________________________________________ 46
FIGURE 48: Tetrahedral meshing properties used; example of a circular mesh on the right ____________________________ 46
FIGURE 49: Resonance by f-sweep _________________________________________________________________________ 47
FIGURE 50: Dual measurements for S21 solution of asymmetrical circuit __________________________________________ 48
FIGURE 51: Correspondence between measured and simulated coupling __________________________________________ 48
FIGURE 52: Shape correspondence between full-wave scatter (left) and AC circuit power (right) _______________________ 50
FIGURE 53: Voltage division across 2-port WPT ______________________________________________________________ 50
FIGURE 54: Concurrence of S21 curves between V-gain (left) and Z (right) methods _________________________________ 51
FIGURE 55: Effect of Al shielding in 3D EM model _____________________________________________________________ 51
P a g e 7 | 60
Chapter 1 HISTORICAL PROOF OF CONCEPT

This chapter presents previous work accomplished in the field of wireless electrical power transmission. It aims
therefore to establish the concept through a grounded-theory approach.
1.1 Pioneer Works of Nineteenth Century

The idea of power transmission through natural medium1 has been explored
since early days of commercial electricity. Its pioneer, Nikola Tesla, worked on
the concept as an emergent consequence of superiority in practical utility of his
AC supplies against Thomas Edison’s DC systems [1]. From refinement of his
works developed the field of small-signal radio propagation [2], even to the
extent that the Supreme Court of United States overturned in 1943 Guglielmo
Marconi’s prior claims to wireless communications, ruling in favour of Tesla’s
patent no. 645,576 [3].
Field induction in conventional transformers is a well-established

phenomenon. It is limited by inter-coil separation (d). Power transfer through
air-core is generally efficient only for short gaps (where d is not very large
compared to coil diameter) otherwise omnidirectional field wastage increases
dramatically [4]. Using ferro-magnetic soft core for field linkage allows FIGURE 1: Nikola Tesla holding
transformers to be classed as amongst the most efficient electrical machines. Wireless Light (1898), adapted from
Electrical Experimenter (1919) [5]
This but constrains the development of truly wireless power transfer system.
Tesla’s lifelong dream pursuit was to

accomplish large-scale power transfer. For this purpose, he held demonstrations
for glitterati from US political and intellectual elite2 at his Manhattan laboratories,
where he was able to light up vacuum bulbs in his hand without wires [5]. The
basis of his experimentation was the idea of inductive coupling through what is
now referred to as Tesla coil [6], where a large primary coil feeds electromagnetic
energy into an enclosed secondary coil of very high mesh and turns (essentially
an air-core transformer3). This arrangement stepped up secondary voltage to high
magnitude and pulse frequency, allowing spark discharges from a gap capacitor;
consequently, an electromagnetic field was transmitted such that it would
harmlessly graze the skin of Tesla’s human subjects instead of penetrating
organs4. This pilot concept appears in his patent for electrical lighting [7]; this is
later refined with themes specifically focussed on wireless power [8].
FIGURE 2: Tesla's patent 649,621
for wireless transmission [8] Tesla was aware of the inefficiency of energy transfer by weak inductive
coupling; in his writings he describes a quest for improvement through what can
in contemporary terms be understood as an increase of primary transmission power, minimization of flux linkage
losses and propagation amplification through impedance matching5. He states these three factors thus:
“…perfecting a method of transmission of electrical energy through the natural medium, led me to
recognize three important necessities: First, to develop a transmitter of great power; second, to perfect
1
As “air” is referred to in classic literature
2
Mark Twain is often mentioned as one of the interested audience
3
Analogous to a step-up autotransformer
4
“Skin Effect” in humans is anecdotal and controversial in terms of safety; the term is today applied to AC conductors
5
Later classified as maximum power transfer theorem, discussed in Section 4.1
P a g e 8 | 60
means for individualizing and isolating the energy transmitted; and, third, to ascertain the laws of
propagation of currents through the earth and the atmosphere.” [9]
Tesla’s pursuit of wireless electric transmission led him to investigate earth’s telluric currents, which result from
disturbances in magnetosphere due to thunderstorms and bombardment by ionized particles flowing from the sun6
[10]. Resulting low-frequency currents migrate towards equator from poles in a space between earth’s surface and a
negatively charged atmospheric layer (ionosphere). Tesla calculated an Eigen frequency of this space7 to be 11.3 Hz;
later work in 1952 by Winfried Otto Schumann led to the determination of fundamental resonance of 7.83 Hz [11].
The concept of utilizing system resonance to maximize wireless power transfer had therefore appeared in early
works of the nineteenth century. To globalize this phenomenon, Tesla sought to create an amplifying transmitter that
would create standing waves of electromagnetic potential in earth’s atmosphere through repeated low-frequency
injection pulsing; receivers all over the world would tap out power from this giant conduction system [12]. For this,
he built a demonstrative tower in Colorado Springs; the project caught J.P. Morgan’s interest who agreed to finance
a 60 m tall power tower8 at Shoreham, New York [13]. Due to several reasons including Marconi’s illegitimate
precedence to wireless invention [14] and differences of business model between with the sponsors, the project was
abandoned by its financiers.
It may be argued that the interest in this domain of wireless electricity was lost to a commercially more viable
wired transmission, not least because user consumption metering could not be tracked in case of global power
broadcast. This doubt is illustrated by Morgan’s comment:
"If anyone can draw on the power, where do we put the meter?" [15]
1.2 Resurgence of Engineering Interest

A renaissance of wireless electricity has been observed in the past two decades or so. The paradigm has shifted
from weak coupling constrained by separation distance to a strong coupling regime that circumvents flux leakage
utilizing inter-coil resonance. Some interest of application has centred on powering of biomedical implants using
stagger tuning for link performance in the style of wide-pass amplifier [16].
Several factors restrict the efficiency of a simple air-core transformer arrangement industrially. Generally,
transfer improves in direct proportion to winding cross-section (as coupling factor increases with flux linkage) and
number of turns (as magnetizing inductance rises), whether on primary or secondary side; it depends inversely on
inter-coil separation [17] and spatial misalignment of coils. Such considerations are constrained by cost and size
limitations in practice, hence casting doubt on the aptitude of wireless energy beyond near-field or short-range9
attempts. The goal being coupling maximization without using a ferrite core10, research direction therefore tends
towards minimization of flux leakage. For commercial robustness, ways need to be discovered that could allow
efficient power transfer whilst compensating for variation in separation, alignment and load matching.
Capitalizing on the use of LCR series tank resonance to transfer energy through evanescent11 electromagnetic
fields, high link efficiency through air was demonstrated by André Kurs et al. [18] at MIT for 60 W bulb at distances
up to 2 m, by a driving source of 9.9 MHz. This mechanism has spurred subsequent interest in EM field capture
between two resonantly synchronized magnetic coils that interact weakly with off-resonant objects in their
6
Known as solar winds
7
Termed Scumann’s Cavity
8
Referred to as the Wardenclyffe Tower, after the lawyer who owned the land
9
Near-field gap defined as 0.1λ of source and short-range gap as same order as coil diameter [55]
10
Hence air
11
Non-radiative
P a g e 9 | 60
environment, rather than earlier attempts at simple EM field loading through use of arbitrary capacitance. In terms
of circuit equivalence, this phenomenon may be explained as the appearance of local minimum for a tertiary
inductance (leakage) at resonant frequency, in comparison with the total mutual inductance of the coils [19].
1.3 Preceding Research Work

Design of inductively coupled resonant links has been studied by various approaches in the past, with a goal to
optimize the efficiency of power transmission. Electromagnetic simulation is a popular method for link modelling,
and while convenient in terms of hardware absence, is constrained by factors such as high calculation load and
idealizations. This research takes off from simulations performed by Rotaru et al. [20], which focussed on the relation
between power transfer efficiency and load matching, in order to overcome separation and passenger shielding
problems associated with hybrid vehicle charging. Their methodology involved full 3D field modelling of the WPT12
coils compared against coupling results from an equivalent circuit model. The coils constructed were staggered
double-layer flat (pancake) style, whose exact dimensions shall be discussed in final sections of this thesis where
further simulations are performed. Dr. Rotaru’s work shows that magnetic field intensity distribution between the
coils is strongest around resonant frequency of the system, with critical coupling factor13 less than 1. Additionally, it
was observed that separation leakage can be compensated by means of improved load matching, suggesting perhaps
a dynamic load variation mechanism.
1.4 Content Overview

This work follows a hybridized approach, primarily focussing on real laboratory experimentation accompanied
by quantitative spreadsheet modelling, explored partially in later sections with full-wave 3D simulation and steady-
state (AC) circuit element modelling. Attempt is made to explain the findings throughout by mathematical circuit
and port analyses. Indications are made for a potentially deeper research in the future where results need more
investigation. Previous research studies have given less attention to coil geometry, shape mismatch or lateral
misalignment [21], whereas this study shall also attempts to touch these aspects. Chapter 1 has already discussed the
background premise for research into resonant wireless power coupling. Chapter 2 will explain laboratory fabrication
and measurement techniques for evaluation of coil parameters and extraction of coupling factors under a variety of
external variables. Chapter 3 looks into the phenomenon of conductive shielding, and concludes with mathematical
resolution supported by laboratory measurements. The idea of impedance matching is investigated extensively
in Chapter 4, where port derivations for LCR series and parallel arrangements are conducted for follow-up lab
verification. Brief simulation work is performed in Chapter 5, using full-wave and circuit simulator tools. A summary
of findings is presented in Chapter 6, with any complementary information produced in Appendices.
12
Wireless Power Transfer
13
Critical coupling is the K-value where flux is transmitted most efficiently, without any reflections or leakage
P a g e 10 | 60
Chapter 2 LABORATORY SETUP

This section introduces laboratory setup and quantitative spreadsheets programmed in the course of
investigations performed to test the theory of wireless power transfer by first principles. Initially, fabrication
technique is explained with the aim of producing a WPT framework so that factors such as alignment and coupling
can be explored. Discussion then proceeds to establish an accurate method to determine inductance and equivalent
resistance of coils in laboratory conditions using standard electronic generator and measurement equipment. Finally,
a methodology for extraction of coupling factors is discussed, accompanied by theoretical explanation.
2.1 Bifiliar Pancake Coil Fabrication

After initial testing with simple rings, a final prototype set of coils was
constructed as pancakes, using bifiliar speaker wire of inner conductor diameter
1.5 mm [22]; each conductor being composed of 7 bundles : 27 strands per
bundle : 0.1 mm strand thickness14 as sketched in FIGURE 3.
The underlying design principle was based on Nikola Tesla’s patent for
electromagnetic coils [23] of FIGURE 4, which seeks to maximize parasitic
capacitance for neutralization of self-inductance. This is not the aim here; rather,
we utilize a pancake configuration because it permits a more evenly dispersed
field due to spatially distributed (as opposed to bunched) components of
resistance, inductance and capacitance. This is thought to improve the chances
of field pickup at receiving coil [24]. It is also geometrically more efficient in
that it can be fit into a relatively
thin-depth package, and is hence
space-efficient for charging mat
arrangements of hybrid vehicles.
Whilst the quality factor of FIGURE 3: Bifiliar Speaker Wire
traditional cylindrical coils is of
a better order [25] due to decreased parasitic capacitance with axial
separation of winding turns (i.e. pitch), and greater volume enclosed
by same number of turns15, such coils can obviously not be housed
in thin enclosures.
FIGURE 4: Tesla's patent for EM coils Some differences from Tesla’s original design are observed in
(US. 512,340) our fabricated coil. Firstly, the bifiliar wire is stood vertically, instead
of horizontal running as seen in the patent. Secondly, the coil is not double-layered all through its diameter; instead,
for the first 14 turns it is single-layered (uni-filiar), with the remaining 7 turns near the outer edge in double-storey
fashion (bifiliar). Overall, this becomes a 21+7 windings configuration set out in two layers. While the main reason
for this construction was to imitate the commercial arrangement already explored by Rotaru et al. [20], it also
improves field intensity without introducing too much self-capacitance or equivalent series resistance16, as well as
directing a stronger field presence near the outer edges, to reinforce the area where flux leakage is most probable.
14
Pro-Power audio cable 27/7/0.1 x 2, copper with total outer diameter 3.5 mm
15
Q = XL/R, with XL = 2πfL, where L depends on volume enclosed by turns whereas parasitic C depends on pitch
16
ESR or esr
P a g e 11 | 60
Laboratory fabrication was achieved by creating a spool from PVC pipe (φ = 5 cm) glued onto circular base cut
out of double-layered Corotherm plastic17. Another circular sheet cut-out was then used to press the arrangement by
manual pressure while winding the coil from the cable spindle. This pressure is meant to prevent the sprial from
unravelling (see FIGURE 5) during construction. A complete list of equipment can be consulted in APPENDIX 1.
Pressing cover to prevent

unwinding of coil during spooling 5 cm
PVC pipe stub as spool
Circular base cut-out

from Corotherm sheet Hole drilled to tuck
inner end of coil
FIGURE 5: Construction setup for pancake coil fabrication
Once wound, the product was taped and upper core of the bifiliar cord was excised from the centre (starting
point) up to the 14th turn; at this point it was then electrically jointed with the lower core on the outer extremity of
coil. In doing so, care had to be taken with regards to the winding orientation of the system as a whole. The two
layers of each coil were connected such that the sense of turning (i.e. clockwise or anticlockwise) from the centre of
the coil towards its edges on the lower layer was also maintained when the windings jointed with the upper layer (as
explained by FIGURE 6); otherwise partial cancellation of the inductance would occur instead of an increase. The
whole coil thus created had one terminal from the lower layer (21 turns throughout the full diameter) and the other
terminal from the upper layer in series (7 turns only towards the outside of the coil). Two such identical coils were
manufactured, one to act as transmitter and other as receiver of electromagnetic power.
Lower layer with 21 turns

(inductance Llower)
Lupper
Llower
19.5 cm
Upper layer with 7 outer

turns (inductance Lupper)
FIGURE 6: Structure of bifiliar pancake coils and winding orientation of layers
17
Selected for its ease of cutting due to material’s simultaneous thinness and rigidity
P a g e 12 | 60
2.2 Introduction to Quantitative Spreadsheets

Laboratory input was collected into, and quantitatively analysed using an extensively programmed set of
spreadsheets. The aim was to speed up the modelling, as well as provide a means to store the databank for later
manipulation in a possibly different manner. These worksheets, listed in TABLE 1, contain the entire dataset of this
project, and form a set of supporting files included separately with this dissertation.
TABLE 1: List of Worksheets for Quantitative Modelling 18

Tag Worksheet Purpose
A P-coils: L and Resr measurements
B B-coils: L and Resr measurements
C P-coils: extraction of axial coupling factors with distance
D B-coils: extraction of axial coupling factors with distance
E P-coils: extraction of planar coupling factors with distance
F P coils: investigation on effect of Al shield on K-curves, coil L and esr values
G P coils: measured for V-ratio (i.e. VL2/VL1), with load in parallel resonant LCR
H P coils: parallel LCR tested at RL = 50 Ω to see effect on ratio ZL/Zout and V-gain
I P coils: series LCR tested to demonstrate S21 does not coincide with load-
match unless esr = 0 or Rs is variable
J P coils: series LCR tested to demonstrate S21 extracted is maximum at resonant
frequency
K P coils: 2-port series LCR setup for testing impact of changing circuit
parameters (e.g. RL) on output curves (S, Z, V)
L1-4 Extracted results from Spreadsheet K into static output record, followed up
with additional experiments
M P coils: series LCR 3D plots for frequency sweep of impedance and coil voltage
ratios against axial separation
N P coils: 2-port parallel LCR setup for testing impact of changing circuit
parameters (e.g. RL) on output curves (S, Z, V)
O P coils: : series LCR circuit frequency swept to demonstrate S21 peak at
resonant f
Sample screenshots of the modelling worksheets are provided in APPENDIX 2. All laboratory results and
graphical trends discussed in this thesis, especially Chapter 2 to Chapter 4, are based on them.
TABLE 2: Major input/output categories of spreadsheets19

Main Inputs (measurements) Main Outputs (calculations)
Vin primary side voltage mag(Zin) impedance referred to source
Vout load side voltage mag(Zout) impedance referred to primary
VL1 transmitter coil voltage mag(S21) scattering parameter
VL2 receiver coil voltage Sload apparent load dissipation
Δ phase difference SS apparent source power
F frequency (LCR setups in Ratios various V, Z and power ratios
this project resonant to 10 from above
kHz)
18
P-coils refer to pancake type, B-coils refer to bunched ring type
19
10 kHz resonance selected for accuracy with available capacitors; investigation principles remain same for any f
P a g e 13 | 60
2.3 Parametric Measurement of Fabricated Coils

Once the coils were constructed, their inductance and series resistance were established by comparing results
from a variety of available methods. One of the measurement methods was then chosen by the author for use in
further testing, on basis of accuracy and ease. The various methods and factors affecting final choice shall now be
discussed here.
For all experimentation, advanced function generator and mixed-signal oscilloscope20 equipment was used;
wherever voltage was measured, its rms value averaged over samples21 was considered. Oscilloscope’s configuration
for measurement acquisition was set to high resolution type, to avoid noisy components in signal average, and hence
concentrate only on the quality portion of the input. Statistical averaging (such as rms) was set to 40 input samples
per window. Circuit equipment like inductors, capacitors and resistors were schematized using standard breadboard.
Each measurement method was tested by repeating the experiment with varied inputs (such as values of external
resistance or input frequency) and then taking the mean.
2.3.1 Method A: Inductance Measurement by Voltage Split Rref

This method relies on equalizing voltage division in RL series A B
arrangement [26]. The circuit is set up as shown in FIGURE 7, and voltage
measurements are probed at points A and B with respect to common ground Vs L
(G). It is important that the value of external resistance Rref be chosen to be
much larger than esr of coil. For our purpose, 100 Ω suffices, as LCR meter G
shows the range of coil’s resistance to flicker between 0.3 and 1.2 Ω22.
FIGURE 7: Voltage-split circuit
Sine waveform is applied as VS across the circuit. Signal amplitude is 4 V;
however, this does not affect the outcome as ratio of voltage division is determined by impedance magnitudes of
resistor and inductor. Inductor’s impedance XL depends on frequency; manual sweep is performed until VA = 2.VB
by rms. Time and error in sweeping to half-value are practical considerations to be born in mind. Nevertheless, at
that instant, it follows:
EQUATION-SET 1
2 2
( )
= 1 2 , giving = 1 2 , so therefore 2 2 2
= 14 ( . )
( + ) 2 + 2 2 +
2 2 2
which evaluates to 3 = , finally yielding = where = 2� ( . )
3
2.3.2 Method B: Inductance Measurement by Resonant Tank Rext

This method uses frequency sweep to detect maximum voltage
response across a parallel LC configuration [27]; the tank is extended by
Vs C L V
adding in series an external resistor (again chosen as earlier at a large
value compared to inductor esr) as shown in FIGURE 8. The resonant
frequency (fres) is sought by manual sweep; at this point the LC portion
becomes a large impedance, and hence voltage across is (VLC) will tend to
FIGURE 8: Resonant tank circuit
a maximum (refer to FIGURE 9). Any frequency slightly either above or
20
Tektronix AFG2021 Arbitrary Function Generator and Tektronix MSO-3034 Mixed Phosphor Oscilloscope
21
The statistical mean of certain number of samples, configurable in MSO
22
LCR meter is unable to accurately register esr for a short length of wire due to small value
P a g e 14 | 60
below the Eigen point will decrease V response. This method therefore, consumes time and visual judgement of the
researcher to reach the correct value. At resonance, reactive components equalize in impedance (hence circulating
the EM energy, and appearing nearly as an open circuit to the source). It follows then that:
EQUATION-SET 2
�� = �� and so � = � , giving = .
�
FIGURE 9: Left to right – V responses at 30 kHz, 35 kHz (resonant) and 40 kHz
2.3.3 Method C: Inductance Calculation by Coil Geometry

Here, a theoretical calculation is performed according to established principles of
EM physics, to arrive at value for inductance by geometric parameters. Interestingly, the depth
radius
formula yields results comparable to experimental values. According to ARRL
Handbook [28], the value for inductance of multi-layer, single-row coil is given by:
EQUATION-SET 3
×
= with units of length in m .
. . ℎ
Since we have a double-layered coil, we calculate the L values separately and add FIGURE 10: Multi-layer coil
them up because windings of two layers assist each other’s inductance in series. The
parameters used for this calculation are as follows (noting inner hole radius from
FIGURE 5 to be 2.5 cm):
TABLE 3: Dimensions of pancake coil

Turns Radius (cm) Depth (cm) L (mH)
Lower 21 (19.5)/2 = 9.75 9.75 – 2.50 = 7.25 0.1050
Upper
Layer 7 (19.5)/2 = 9.75 (7.25)/3 = 2.42 0.0175
Layer Total 0.1180
P a g e 15 | 60
2.3.4 Method D: Inductance Measurement by Impedance Rref

A B
Refer to Spreadsheet A
I-V method is employed to calculate the impedance of device under test, 50 Ω Resr
in this case an inductor [29]. As shown in FIGURE 11, the source generator
has internal series impedance of 50 Ω so the selected value for external Rref
Vs L
should be comparable for good signal resolution of voltages probed at points
A and B with respect to common ground (G).; several ohmic combinations are G
tried and it is seen that the order of 100 Ω allows good resolution of measured
signals.
VA + VB VB
In addition to VA and VB, the phase angle between the probed voltages is θ
also measured from the oscilloscope as (VA – VB), as seen in FIGURE 11. VA
θ
Now across the inductor (including its esr), it can be observed that the VA - VB -VB
values for voltage and current (dependent ohmically on voltage drop across
FIGURE 11: Circuit setup and voltage
Rref) are given by: triangle for I-V method
EQUATION-SET 4
(� − � ) � � .
� = � � = , = � = ( . )
� −�
Now from the law of cosines for triangles, FIGURE 11 implies:

2 2
� −� = � + � − 2. � � . cos ( . )
− � .
The impedance angle α can be derived as [ �
. ]. Using the voltage triangle in the figure above,
�
we therefore deduce that:
−1
�= − − � . � −� . ( . )
Finally, in rectangular coordinates we know independently that � = + � ; by comparing this against

the measured values of mag(ZL) and α, we arrive at results:
= � . cos � .
� sin �
= � .
This method carries the advantage that it does not require sweep time or judgement by the student, and the
average values of VA, VB and are read off instantaneously for plugging into a spreadsheet of equations.
2.3.5 Critique for Choice of Method and Results of Measurements

Refer to Spreadsheet B
The experimentation was carried out initially on bunched ring coils using 0.6 mm breadboard wire, for a
preliminary assessment of various measurement methods discussed above, and a comparative critique between them
leading to a preferred choice for Method D. It can be seen from the foregoing elaboration that reliable, replicable and
quick experimentation requires a method which depends less on individual judgement or adjustment of conditions,
P a g e 16 | 60
and more on instantaneous readings. Further investigations were then based on Method D alone. Complete data
record is available in supporting spreadsheets discussed in Section 2.2.
FIGURE 12: MSO screenshot for I-V method
A summary of outcomes for bunched type coils (labelled B) and pancake type coils (labelled P) is presented in
TABLE 4. It can be seen that the results from various methodologies compare well with each other, and also with
LCR readings; the latter is not considered very accurate as it has a factor of minimum resolution that affects the
displayed value as a rounded number.
TABLE 4: Method outcomes for different coil types

Units Coil B1 Coil B2 Coil B3 Coil P1 Coil P2
Coil type - Bunched ring Bunched ring Bunched ring Bi pancake Bi pancake
Wire type - Enamel. Cu Insul. Cu Insul. Cu Insul. Cu Insul. Cu
Wire gauge SWG 17 SWG - - - -
Wire φ mm 1.422 0.6 0.6 27 x 7 x 0.1 27 x 7 x 0.1
Conductor CSA mm2 1.588 0.283 0.283 1.5 1.5
Coil φ cm 11.5 6.5 10.5 19.5 19.5
Coil depth cm 1.0 0.64 0.70 0.70 0.70
Turns (N) - 10 10 11 21 + 7 21 + 7
Coil loop CSA m2 0.0104 0.0033 0.0087 0.0299 0.0299
L: LCR meter mH 0.0200 0.0100 0.0250 0.1030 0.1020
L: Method A mH 0.0205 0.0119 - - -
L: Method B mH 0.0208 0.0113 - - -
L: Method C mH 0.0220 0.0122 0.0253 - -
L: Method D mH 0.0215 0.0110 0.0268 0.1034 0.1022
Resr: Method D Ω 0.0962 0.1262 0.4640 0.4619 0.3601
P a g e 17 | 60
2.4 Establishment of Coupling Factors

Once the individual coil inductance values have been measured, an important step forward is the investigation
of coupling between coils as axial separation is increased. The experimentation will be repeated by changing certain
other variables, such as mismatch in geometric sizes of the two coils (transmitter and receiver) or introduction of
shield in vicinity.
To assess the coupling factor K between coils, mutual winding orientation was first determined. This was
checked by simply applying a sinusoidal voltage on the transmitting coil and observing its phase difference with
receiver voltage. If the waveforms are anti-phase (180o difference), polarity of winding terminals is adjusted so that
they are in phase as in FIGURE 13. The tandem of coils thus arranged shall be referred to as “assisted windings”.
FIGURE 13: In-phase polarity of coil windings
In order to extract K, the two coils are connected in series with polarity of assisted windings, and the total
inductance is measured through Method D discussed in Section 2.3.4. In this situation, a mutual inductance M
develops and adds to the self-inductance values of coils [30] in order to give:
EQUATION-SET 5
= + + and = ⁄√ . .
The two coils are therefore set up as a single inductor, and values of M and K are derived from total inductance
using these relationships.
2.4.1 Axial Coupling between Identical Pancake Coils

Refer to Spreadsheet C
To ensure that the movement is in one dimension, pancake coils are mounted on to a calibrated PVC pipe of
same diameter as their inner gap (5 cm), and separated 1 centimetre at a time (along x-axis). Effects of misalignment
by rotation or planar sweep (i.e. along y or z-axes) are thus avoided (FIGURE 14). The range tested is from 0 to 35
cm, and as the coil diameter is 19.5 cm, this implies mid-range separation (where the order of expanse is more than
coil diameter). The experiment was repeated using by slight variations, to give following scenarios:
1. Case A: Axial separation sweep at f = 10 kHz, without any resonant setup

2. Case B: Axial separation sweep at f = 30 kHz, without any resonant setup
3. Case C: Axial separation sweep at f = 10 kHz, with arbitrary C parallel to one side (150 nF)
4. Case D: Same as Case C, but with arbitrary C parallel to both sides (150 nF)
P a g e 18 | 60
FIGURE 14: Axial separation of pancake coils

Interesting results are observed, which are presented as curves in FIGURE 16 .
Cp
Firstly, it is observed that at 0 separation, when the coils are practically touching, the
coupling factor is still not 1. The author attempts to explain this by the following non-
ideal conditions that are experienced in a practical environment:
Resr L
1. Even at 0 separation, the conductor insulation intervenes to create a slight
gap; as well, even the slightest rotary movement of coils around the PVC
pipe due to loose grip is seen to affect the oscilloscope probe readings.
2. The maximum possible applied signal from function generator is 10 Vpp,

meaning rms23 value of 3.53 V. It is seen from Section 2.3.5 that ESR of
coils is not 0 as in theoretical ideal, and the measurement circuit itself
contains series external resistance of the order of 100 Ω. This computes
to a very small driving current through the inductance even if we ignore FIGURE 15: Parasitic
the impedance of the coils (around 0.35 A)24. Such a small current will capacitance modelling
induce a weak field, leading to a decreased pickup for mutual inductance.
An attempt to retrofit data curve was done for Case-A using Matlab curve-fitting tool with the aim of a close
regression; results were as follows:
− .
1. Exponential decay = . with regression factor R2 = 0.95
2. Fifth order polynomial regression with R2 = 0.99 leading to expression

= − − + − − . + . − . + .
Secondly, it is observed that coupling decreases with application of a higher frequency signal, when compared
to base case of 10 kHz. Since no resonance is considered in this situation, this has to be explained otherwise, and as
such needs further investigation, which was not performed under the scope of this dissertation. While it would not
be correct to predict as a pattern that wireless transmission decreases in general with increase of system frequency
in the absence of a resonant arrangement, a possible explanation for this phenomenon could nevertheless be devised
as follows. An inverse relation exists between capacitive reactance and frequency; this is equally applicable to the
23
Vrms = Vpp ∕ 2 ∕ √2
24
I = V/R = 3.53/100
P a g e 19 | 60
parasitic-capacitance (Cp) of the inductor, which is modelled as a component self-paralleled to the inductor [31] as
illustrated in FIGURE 15.
1.2000
1.0000
0.8000
Couping factor K
0.6000
0.4000
0.2000
0.0000
- 5 10 15 20 25 30 35 40
-0.2000
Axial separation (cm)
Case A Case B Case C Case D
FIGURE 16: Coupling factors K with axial sweep for different circuit scenarios
Using circuit theorem for parallel equivalent impedance, and ignoring Resr for simplicity, we get:
EQUATION-SET 6
�� . �� ( ⁄ )
= ⁄ = = . .
�� + �� ( + ) −
The gradient balance of the equivalent impedance is hence determined by reciprocal of term on the right hand
side, i.e. − . Near resonant frequency, ωL cancels out 1/ωCp , leading to a large impedance; this is the
peak point of the impedance graph in FIGURE 15. This decreases the current fed into the conductor from a voltage
controlled source. By Ampere’s law, a smaller current shall produce a weaker magnetic field, and hence smaller
transformer linkage over air-core [32]:
EQUATION-SET 7
∮ �. =� � .
where H is magnetic field

Ienc is the enclosed current
dl is path differential
At what point the inductive and capacitive reactance terms equalize depends on the interaction of frequency
value with L and C magnitude. In general, as ω = 2πf, we get resonant frequency f0:
P a g e 20 | 60
EQUATION-SET 8
= so that = .
� .
According to FIGURE 16, switching to 10 kHz from 30 kHz in Case B, causes a fall in inductive coupling as
measured through application of a voltage-controlled signal source. From the foregoing, it may thus be postulated
that Zeq rises and so according to FIGURE 15, the system is in the rising impedance linear zone where capacitance
does not yet begin to limit the total impedance by inverse proportion to the frequency.
In order to check this hypothesis, an attempt was made to measure the Cp for inductor coil P1, using frequency
sweep technique that was earlier employed in Method B of Section 2.3.2; this time no external capacitor was used
however, and the inductor coil was considered an inherent LC tank. Again for simplicity, we shall ignore Resr in the
following quantification, as its small value does not affect much the series addition to inductance. A maximum
response of measured voltage was seen around 4.5 MHz (f0); using EQUATION 8.1 with L = 0.1034 mH from
Section 2.3.5 to solve for capacitance then yields Cp value of 15.31 pF.
We plug this set of L and Cp values back into EQUATION 6.1 to yield:
for 10 kHz, Zeq = -j 6.49 Ω

for 30 kHz , Zeq = -j 19.49 Ω
The factor increase of 3 follows the frequency increase ratio of 3 times (from 10 kHz to 30 kHz); the coils are
thus operating in the zone of direct proportion between the modulus of equivalent impedance and frequency from
FIGURE 15. A rise in impedance will then cause a lower driving current in the transmitting circuit, and hence a
weaker transmission flux. Since our testing operates in a frequency order much below the self-resonance of coils, we
can ignore the parasitic capacitance of the system from future analysis.
Finally, it is seen that adding successive capacitors in parallel (first on one and then on both sides, as in Case C
and Case D), albeit not resonant tuned and arbitrary in value, leads to improved coupling curves. An attempt to
explain this could be made by inquiring the resonant capacitance for pancake coils at f = 10 kHz using EQUATION
8.1 with inductance parameters from Section 2.3.5 as LP1 = 0.1034 mH and LP2 = 0.1022 mH. The following values
are obtained for resonance at 10 kHz:
for Coil P1, CP1.res = 2.45 μF

for Coil P2, CP2.res = 2.48 μF
Since parasitic capacitance of the coils is very small (of the order of pF), any external parallel addition of
significant capacitance (such as 150 nF used) will mean therefore that, the system tends to move towards resonant
state. An approach to resonance state in general tends to produce cancellation of inductive reactance brought about
by capacity addition, thus freeing up the circuit’s bandwidth for field transmission by increasing driven current (in
series arrangement) or voltage drop (in parallel arrangement) across the coils.
2.4.2 Planar Coupling between Identical Pancake Coils

Refer to Spreadsheet E
A different axis of coupling was explored by separating the

coils through a radial line; the axial distance for this was kept 0 (i.e.
coils touching). The same evaluation method for inductance and
FIGURE 17: Planar separation
P a g e 21 | 60
coupling measurement was used as discussed previously in Section 2.4.1, through assisted windings of two-coil
series.
The resulting curve is displayed in FIGURE 18; it shows a similar general decay as the axial case seen earlier;
however, the rate of decay is sharper, and an aberration is seen around 10 cm in that coupling dips below 0, staying
negative until around 20 cm. The diameter of the two coils is 19.5 cm each; therefore it may be stated that K values
become negative after about one radius (half diameter) of planar separation, and stay so until full diameter of
separation has been achieved.
1.0000
0.8000
0.6000
Coupling factor K
0.4000
0.2000
0.0000
0 5 10 15 20 25 30 35 40 45
-0.2000
Radial distance (cm)
FIGURE 18: Coupling factor K for planar separation
In order to explain this, the author looked at the magnetic flux pattern around a circular coil for simplified
analogy [33]. Combining Biot-Savart law with Pythagores’ theorm and
conducting a loop integral in FIGURE 19 gives:
EQUATION-SET 9
� � × �̂
= .
�
Since y-components cancel, we focus along x-axis to give:
� � .
= .
� +
Integrating around path for an undisturbed loop totals to
� �
= ∮ = → .
When there are two loops radially coupled instead, it can be deduced the
central B-field along x-axis in an assisted winding configuration will add up as
long as the coils are commonly centred (i.e. planar separation 0). As they are FIGURE 19: B-field of I-loop
pulled apart, the assisted field continues to play a role until one radius of
P a g e 22 | 60
separation i.e. the larger portion of mutual flux still flows in the same direction. However, after this displacement,
contradictory flux from left coil begins to cancels the inflowing field lines from right coil, in the area of overlap. It
may be theorized that during this time, the mutual inductance behaves opposite to EQUATION 5.1, and characterizes
itself instead as Ltot = L1 + L2 – 2M. This phenomenon becomes stronger initially as the coils are near 1-radius
separation, but the negativity consequently weakens in magnitude as the displacement brings the coil edges closer;
this is because not only do the opposite field lines start to equalize, but the reinforcing flux from farther edges is
unable to reach the overlap region. At one diameter of separation, the opposing fluxes cancel out perfectly; hence K
is 0 here. The above explanation can be visually ascertained in the curve of FIGURE 18.
Negative coupling indicates a reversal of field that decreases wireless transmission, as the flux is repelled. Axial
coupling regime therefore, offers a more robust configuration for WPT; in practical systems, this also means that any
radial misalignment due to rotation or displacement of coils should be prevented, and it should be ensured that they
are concentred along the same axis.
2.4.3 Axial Coupling between Non-Identically Sized Coils

Refer to Spreadsheet D
A variation on Section 2.4.1 was performed by testing bunched-ring coil (referred to as B1, B2 and B3 in TABLE
4). The experiment was again repeated using by slight variations, for comparison against FIGURE 16:
1. Case A: Axial separation sweep at f = 10 kHz, without any resonant setup, Coils B1and B2
2. Case B: Same as Case A, but with arbitrary C parallel to one side (1 μF)
3. Case C: Same as Case C, but with arbitrary C parallel to both sides (1 μF)
4. Case D: Same as Case A, but with coils B1 and B3
The results can be seen from curves in FIGURE 21. As before, the trend of
exponential decay can be discerned. As well as that, the addition of arbitrary
capacitance again increases M (and hence K) as discussed previously in
Section 2.4.1. It must be taken into account however, that addition of parallel
capacitance does not always guarantee improved coupling; a decrease or
negative K could result as well. This is determined by what part of the Zeq vs .f
graph the system is in (see FIGURE 15); addition of external capacitance may
shift the system’s resonance away from actual applied frequency, hence
adversely affecting coupling. What is different however, is that when coil sizes
are mismatched, such as for Case A (B1 φ = 11.5 cm and B2 φ = 6.5 cm from
TABLE 4), there appears to be a sharp dip in K-curve near 1 diameter separation
extent of the smaller coil. When however, ring coils of identical sizes are tested
for results as a control (Case D), the K-curve becomes smoother to bear
resemblance with the model in FIGURE 16. For a more rigorous solution to this FIGURE 20: Setup for Non-identical Coils
problem, and to indeed determine whether this pattern is exceptional or
characteristic of coils mismatched in size, a fuller investigation to this effect would have to be carried out. That is
not pursued in this dissertation, but a preliminary explanation is set out here in terms of magnetic flux capture
efficiency. When the receiver coil is much smaller than the transmitter, it could be theorized in non-ideal settings
(i.e. lossy environment, low amplitude flux, bunched ring type coil) that a lot of the magnetic field stays around the
outer edges of the transmitter. The receiver then has little chance of capturing the flux efficiently as soon as it moves
even a diameter away, leading to a visible dip in K-curve. More uniformly distributed coils, such as the pancakes
discussed earlier, and also coils identically sized with respect to each other, seem to enhance wireless flux capture.
P a g e 23 | 60
Curve fitting again yields expressions for Case D, which are comparable to Case A of Section 2.4.1.
− .
1. Exponential decay = . with regression factor R2 = 0.90
2. Fifth order polynomial regression with R2 = 0.99 leading to an expression

= − × − + . − . + . − . + .
0.6000
0.5000
Coupling factor K
0.4000
0.3000
0.2000
0.1000
0.0000
0 5 10 15 20 25
Case A Case B Case C Case D
FIGURE 21: Coupling investigation between non-identical ring coils
Conversely, when regression is attempted for Case A in this instance, the closest polynomial obtained is of a
third order, as the curve characteristics there are different. Even at 0 separation, the coupling achieved for Case A of
mismatch sized coils is only around 0.25 (y-intercept), which is far smaller than 0.80 seen earlier for pancake coils.
Again what this experimentation points out to in a preliminary manner is that for industrial systems such as hybrid
car charging mats, the coils should be identical in size and geometrically distributed (rather than ring-bunched) to
avoid flux leakage at close range.
P a g e 24 | 60
Chapter 3 DISCUSSION OF SHIELDING

Metallic shielding of humans from radiation has come into focus with the dawn of telecommunications era.
Protection from direct EM fields is a debated matter of medicinal study, with exposure limits being recommended
[34], and as such, shielding mechanisms may evolve in hybrid cars. For example, ICNIRP25 [35] recommends that
between frequency range of 100 kHz and 1 MHz (corresponding to full-wave model of Chapter 5), induced tissue
current density must be limited to f/500 mA m-2 rms (with f in Hz), or a specific absorption rate26 of 0.08 W kg-1.
The impact of external shield on pancake coils WPT27 system was therefore also investigated in the laboratory
(FIGURE 22). For this purpose, aluminium foil was utilized; being paramagnetic [36], it displays weak field domain
properties. More importantly, it is the material of car bodies, and thus its testing was expected to reveal practical
insights on the problem of WPT for hybrid car charging. Also possible is the presence of stray ferromagnetic items
in vicinity of wireless coils; understanding of this subject would therefore be beneficial for design activity.
FIGURE 22: Investigation of effect of Al foil shield on WPT coupling
3.1 Construction and Gashing

The shield was made by wrapping foil around a cardboard square of 22 x 22 cm (with a central hole, diameter
5 cm to allow pipe mounting). This ensured that the shield was slightly larger than the circular pancake coil CSA28.
Later, when the effect of lamination in shield for interrupting eddy currents was to be studied, gashes were created
in steps i.e. one half and then whole area (see FIGURE 23).
gashes
22 cm
FIGURE 23: Simulation of shield lamination by gashing (half square on left sketch, full on right actual)
25
International Commission on Non-Ionizing Radiation Protection
26
SAR
27
Wireless power transfer
28
Cross-sectional area
P a g e 25 | 60
3.2 Experimental Testing of Shield

This section discusses testing performed to understand effects of shielding on a wireless power transfer system.
3.2.1 Test of Aluminium Shield as EM Barrier

Refer to Spreadsheet F
To check the effectiveness of the constructed aluminium shield, the latter was inserted at an arbitrary point
between pancake coils, and received voltage (V2.rms) was measured while axial separation was increased. This
measurement was repeated without the shield present. Circuit configuration used was of a parallel resonant type (at
f = 10 kHz) as shown in FIGURE 24.
Qty Unit Tx(1) Rx(2)

Vs Vrms 2.83
(1) -
(2) Rs L1 L2
Rs Ω 50.0 - C2 V Rext.2
C1 Rext.1
C μF 2.450 2.477 Vs Resr.1 Resr.2
Rext Ω 1000 1000
L mH 103.4 102.2 Al
Zin Zout
Resr Ω 0.462 0.360
FIGURE 24: Circuit arrangement to check shield intervention
Although Rext.1 in above arrangement is not essential due to presence of Rs, it has been added to allow
subsequent adjustment of impedance for load matching. The latter concept shall be explored further in Chapter 4.
Results of transferred voltage pickup are illustrated in FIGURE 25. Two main features can be discerned:
1. Received voltage rises initially with increase in axial distance; this may be attributed the concept of
changing equivalent impedance seen by the source or load (Zin and Zout respective to primary and secondary
sides) that occurs with variation of coupling factor. This subject shall be taken up in Chapter 4.
2. Absolute value of voltage pickup at secondary falls visibly with insertion of shield; this trend is further
emphasized by ratio of voltages (with and without shield) plotted on secondary axis of FIGURE 25.
The foregoing preliminary investigation supports a conclusion that Aluminium is able to act as a magnetic shield
by causing flux absorption or diversion through it. The well-known transformer phenomenon of eddy currents may
be applied here, which follows directly from a simplification of Faraday’s law:
EQUATION-SET 10
∅
∮ �. = − .
where E is induced electromotive force

dl is path differential
δφ/δt is time variation of magnetic flux
negative sign is from Lenz’s law
P a g e 26 | 60
This means that a flux intercepted by conductive shield shall cause an emf and subsequent currents to be set up
in the surface such that they oppose the field creating it (Lenz’s law [37]). By right-hand thumb rule, a quasi-straight
field (as between two distant poles or coils) will be accompanied by circular current as shown in FIGURE 25. In this
manner, the shield will attempt to cancel wireless flux originating from transmitter and travelling towards receiver.
It should lead us to postulate that a shield therefore affects the value of coupling factor between the system coils, and
that premise shall be explored next.
3.5 0.40
3 0.35
Received voltage (V.rms)
Voltage ratio with shield

0.30
2.5
0.25 B.coils
2
0.20
1.5
0.15
1
0.10 –B.eddy
0.5 0.05 I.eddy
0 0.00
- 5 10 15 20 25 30 35 40 45
V2 (Rx.rms) V2 (Rx.rms) w/ Al (V2.Al)/V2
FIGURE 25: Effect of shield on wireless emf transfer and its explanation by eddy currents
3.2.2 Analysis of Effect of Shield on Coupling Factor

The placement of shield is now changed from Section 3.2.1, in that it is now d
taken outside the transfer zone, adjacent to Coil-2 (receiver). By then fixing the
shield at various distances, a coupling factor measurement for sweep of axial L1 L2
separation at f=10 kHz is carried out according to Method D of 2.3.4. Results are
presented in FIGURE 27, which illustrate the fact that as shield gets closer to Resr.1 Resr.2
vicinity of the coupled system, mutual inductance falls. It should be noted that for
purpose of measurement, simple induction of pancake coils is under consideration, x
and parallel or series resonance is not brought into play. Al
FIGURE 26: Shield placement
P a g e 27 | 60
It can be observed that the K-curves depreciate to lower levels for same axial separation (x) between coils, in
inverse proportion to the distance of shield (d) from receiver. This directly confirms our earlier presumption that a
conductive object present around WPT system has the ability to disrupt source magnetic field; the mechanism of it
can be elaborated upon by phenomena of diversion through lower reluctance and dissipation as heat of eddy currents
[38]. However, the net effect is wastage of original mutual inductance between pancake coils, and this can be thought
of as an increased flux leakage into the environment.
0.600-0.800
0.800 0.400-0.600
0.600 0.200-0.400
Coupling factor K
0.000-0.200
0.400
0.200
0.000
0
1 2
-0.200 3 4
5
6 No shield
-0.400 7
8 Al: 20 cm
9
10 Al: 10 cm
11
12 Al: 6 cm
13
14 Al: 3 cm
15
20 Al: 1 cm
25
30
35
FIGURE 27: K-curves for varying vicinity of shield (d) through axial separation of coils (x)
During measurement, at various instances, the arrangement was repeated with the shield connected to source
and oscilloscope common ground. No difference to the results was seen, and hence it may be suggested here that
grounding of a conductive object will not automatically destroy its magnetic interruptive capacity the way grounding
of electrical shields leads to equipotential.
Another interesting feature of the curves in FIGURE 27 is negative K-factor values for cases where the shield
is very close to receiver; it is not until 1 radius distance (curve of shield at 10 cm) that the K-curve rises into first
quadrant between K and x axes. Relating this with the discussion in Section 2.4.2, a negative coupling indicates flux
reversal or cancellation. While the significance of 1 radius distance from shield is not very clear with this
investigation, it may be stipulated that a shield can set up eddy currents in multiple sub-zones within the larger
surface, depending on defects scattering [39]. An electromagnetic resultant of the various eddy currents can then
determine the net annulation vector for the opposing magnetic field. In general, we can confirm that as shield-system
separation increases, this inductive cancellation weakens. The concept of breakup of surface into disconnected zones
shall be explored in next section.
P a g e 28 | 60
3.2.3 Electrical Reduction of Shield Surface

The impact of reducing eddy currents is further investigated by gashing the shield as shown in Section 3.1, first
on one half of the shield and then on the complete square. Coupling measurements are then repeated and results are
presented in FIGURE 28.
An improvement in coupling is seen

1.0000
as the shield is broken up successively
over its surface area, as manifested by 0.8000
slight curve lifting. The increments
Coupling factor K
0.6000
between intact, half and fully gashed
shield however, do not appear very 0.4000
significant. This could be due to the
0.2000
extent of breakage. It should be noted
that the gashes do not traverse the 0.0000
complete width of the square, hence - 5 10 15 20 25 30 35 40
-0.2000
leaving some margins near the edge to Axial separation (cm)
set up large radius swirling currents. The
electrical reduction could be made more No shield Al: 3 cm Half gashed Full gashed
effective in future by completely
destructive diagonals instead; this could FIGURE 28: K-curve lifting by shield reduction
be accomplished by holding the shield in
place using an outer frame mount so that it does not fall apart. Further quantitative analysis could be proposed to
compare mathematically and by experiment, the integrated total of split eddy currents against large current currents
of an intact shield (FIGURE 29).
Nevertheless, it is fairly established through experimentation that the flux wastage into shield is a product of
electrical flow.
3.3 Mathematical Analysis of Conductive Shield

An attempt is made in this section to quantify the conclusions from
experimental observations in the foregoing discussion. This shall be done by
reducing the impedance matrix of a three-port network into a two-port one.
First, the author tries to establish physically if the shield acts as an inductor. An
indicative experiment for establishing inductance value is performed according to
Original swirl
Method D of Section 2.3.4 at f = 10 kHz on the gashed shield. In a separate run, the
(intact shield)
mutual inductance between shield and one pancake coil is also determined using the
same I-V methodology. It is important that the frequency reference of source is kept Sector swirl
same (in this case 10 kHz) when making inductance measurements, as slight changes (gashed shield)
may occur with frequency according to explanation discussed in Section 2.4.3. FIGURE 29: Eddy sectors
The shield returns an average value of L = 1.622 μH and esr = 0.359 Ω, while its mutual inductance variation
with separation distance from second pancake coil (L2 = 102.2 μH) is presented below. When calculating mutual
inductance with shield, the equation for opposing windings scenario is used from Section 2.4.1 instead of assisted
windings; that is because of Lenz’s law which comes into effect when inducing eddy currents on a free surface,
unlike the constrained winding sense of pancake inductors (refer to Section 3.2.2).
P a g e 29 | 60
TABLE 5: Mutual inductance between perforated shield and receiver coil

x (cm) 1 2 3 4 5 6 7 8 9 10 20
M (μH) 15.52 10.84 7.87 5.72 4.99 2.74 1.94 1.48 1.13 0.90 0.21
Since the two pancake coils are practically identical, we shall suppose that at any given distance, mutual
inductance value is same between shield and either coil. In addition, for curves in FIGURE 27 where the shield is
very near the receiver (d), we can take it to be equidistant from both coils for large separations (x) such that x > > d.
An interesting point to note here is that TABLE 5 accounts for a fully perforated shield; we see curve lifting
effect in FIGURE 28, and can safely state that an undamaged shield would yield higher results for mutual inductance.
It is instructive to calculate some simple coupling comparisons for ratio of mutual to self-inductances between coils
and coil-shield scenario; these ratios are displayed below.
TABLE 6: Mutual to self-inductance ratios
M/L Ratios
L1 inductance of pancake Coil-1 µH 103.4
Ls inductance of gashed shield µH 1.62
x axial separation along x-axis cm 1 2 3 4 5 10 20
M1,2 mutual L between Coils 1-2 µH 69.41 57.86 45.81 36.56 30.51 12.11 2.84
(M1,2)/L1 M/L ratio - 0.67 0.56 0.44 0.35 0.30 0.12 0.03
M1,s mutual L between coil-shield µH 15.52 10.84 7.87 5.72 4.99 0.90 0.21
(M1,s)/L1 M/L ratio - 0.15 0.10 0.08 0.06 0.05 0.01 0.00
From TABLE 6, we see that although mutual induction between shield and a coil is smaller than that between
two pancake coils, it is not insignificant. At close vicinities in fact, the M/L ratio of shield-coil situation is only 5 to
6 times smaller than that of coil-coil arrangement. At longer distances however (about 1 coil radius i.e. 10 cm), the
mutual induction between shield and coil (M12) falls very rapidly, becoming insignificant compared to inter-coil M.
This conclusion is useful in that it should be ensured that any shielding is located closer to the passenger, but much
farther away from WPT inductors.
Having established the shield as a third inductor in the system, we now model a circuit based on instructive
theory by Lenaerts and Puers [40]. For this, the equivalent circuit is postulated as shown in in FIGURE 30, where L3
depicts the shield. According to our measurements above, we see that mutual induction exists between the coils (M12)
and between the coils and shield (M13 and M23). As eddy currents
swirl through the shield to dissipate heat, we take it to be a self-
shorted loop, with V3 = 0. We shall apply principles of transformer VTX M1,2 VRX
L 1 L2
equivalence on this circuit using the following rules of physics:
1. Voltage in any loop is the sum of voltages (by Resr.1 M1,3 M2,3 Resr.2
Ohm’s law IR) across that loop and reflected values
from other loops (Kirchoff’s voltage law [41])
L3
2. Impedance of inductor is jωL
Resr.3
Then applying KVL on the three voltage loops gives:
FIGURE 30: Circuit modelling of EM shield
EQUATION-SET 11
� � = �1 1 + 1 + �2 12 + �3 13 .
P a g e 30 | 60
� � = �1 12 + �2 2 + 2 + �3 23 .
�3 = 0 = �1 13 + �2 23 + �3 3 + 3 .
Rearranging the third loop to express I3 in terms of I1 and I2 yields:

�1 − �2 −
�3 = 13 ⁄( + 23 ⁄( .
3+ 3) 3 + 3)
This now allows us to eliminate I3 from the loop equations of coil circuits, remembering that j2 equals –1. For
example, for primary side, substitution results in:
2 2 2
13 13 23
� � = �1 1+ 1+ + �2 12 + .
3 + 3 3 + 3
A three-port impedance matrix is thus reduced to a two-port one:

2 2 2
13 13 23
1+ 1+ 12 +
� � 3+ 3 3+ 3 �1
= .
� � 2
13 23
2
23
2 �2
12 + 2 + 2 +
3 + 3 3 + 3
Now if there had been no tertiary shield circuit, the primary side loop would simply have been:
��= � + + � .
Comparing the last two statements demonstrates that both self and mutual inductances of a WPT circuit
arrangement is affected by introduction of a shield. By denoting the modified values with subscript new, we can
construct an equivalent equation as:
� � = � . + . + � . .
Finally, by rationalizing the denominator of complex fractions in the shield circuit loop, we can arrive at the
new values of inductances and resistance as follows:
2 2
1. = 1 − 13 3
⁄ 2 2 2 .
( 3 + 3 )
2 2
1. = 1 + 13 3
⁄ 2 2 2 ( . )
( 3 + 3 )
2
( + 3)
12 − ( ⁄
= 3
12. 13 23 ). [ 2 2 2 ] ( . )
( 3 + 3 )
A similar analysis would hold correct for secondary side loop parameters. With the introduction of an adjacent
conductive shield therefore, the modified electrical array becomes:
P a g e 31 | 60
2 2
3. 13
1 − 2 2
( + 2
3 3
2 2
3. 23
1. 2 − 2 2 2
( 3 + 3
2.
2 2
1. 3. 13
= 1 + 2 2
( . )
2. ( + 2
3 3
12. 2 2
3. 23
2+ 2 2 2
( 3 + 3
2
3 + 3
12 − 13 23 . 2 2
( + 2
3 3
From the foregoing analysis, the following quantitative trends can be predicted with the introduction of a
conductive shield into a WPT system:
1. Inductance values of the original coils decrease (from behaviour of L1.new and L2.new above)
2. Series resistance values of the original coils increase (from behaviour of R1.new and R2.new above)
3. Mutual inductance of the original coils decreases (from behaviour of M12.new above)
The third conclusion has already been

verified by experimental testing in Section 3.2.2, 120.00 6.00
which shows a drop in coupling factor as the
Coil inductance (μH)
100.00 5.00
shield is brought nearer the WPT system. To
Coil esr (Ω)

80.00 4.00
check the first two conclusions however, the
author measured inductance and esr of Coil-2 60.00 3.00
(originally 102.2 μH) using Method D of 40.00 2.00

Section 2.3.4 in presence of shield at various 20.00 1.00
distances.
- -
- 10 20 30 40
Results presented in FIGURE 31 confirm that
Shield distance (cm)
measured inductance falls while esr rises as shield
becomes more proximal (i.e. M23 increases from L2 Resr Expon. (Resr)
TABLE 5, hence causing a larger subtraction from
L2). As the shield moves farther away, the coil FIGURE 31: Trend of coil L and esr values with shield introduction
inductance approaches its original value.
The esr values are less accurate in measurement, and more prone to error; although they generally confirm the
numerical model prediction, they have been plotted with an exponential trend-line to indicate averaged pattern on
secondary axis. The source of error comes from reviewing the induction measurement technique (Method D of
Section 2.3.4). Reference resistor chosen (Rref ≥ 100 Ω) is much larger than esr of coil (0.36 – 0.46 Ω). This means
that the esr compares to less than 1% in terms of source voltage division against Rref (refer to FIGURE 11), thus
introducing a significant possibility of noise measurement. Although the trend for esr is confirmed and largely
justified by this approach, a technique more calibrated to esr accuracy could be proposed for further research of this
aspect of shield’s mathematical modelling.
P a g e 32 | 60
Chapter 4 DISCUSSION OF LOAD MATCHING

The theory of impedance matching emerges directly from voltage division principles of electrical circuit
analysis; as such, it shall be explored for WPT experimentation. Limitations such as fixed source impedance (function
generator 50 Ω) in our laboratory setup are expected to constrain the results; however, a mathematical analysis shall
also be conducted to explain results or discrepancies against theory.
4.1 Circuit Derivation

Let us consider a simple AC loop, with the goal of maximizing Zs
power transfer to output (load) resistor RL. The case is a premise for ZL VL
radio frequency antenna matching concept, known as maximum power
transfer theorem [42]. Using FIGURE 32 as a departure point, and Vs
applying voltage division:
EQUATION-SET 12
(using rms voltage, and vector notation for complex quantities) FIGURE 32: Load matching theorem
⃗⃗⃗⃗�
⃗⃗⃗� = � .
� .
⃗⃗⃗⃗ + ⃗⃗⃗⃗�
Initially, a simplification is done to replace complex impedance by a pure resistance scenario:

�
�� = � . .
+ �
Then by definition of power across resistor:
�� ⁄ �
�� = =� . .
� + �
Considering source resistor and voltage to be fixed, the goal is to maximize output power by adjusting load.
The power dissipation expression being of second order, there has to be a local turning point as we plot P L against
RL. As load resistance approaches 0, the numerator dominates to nullify power transfer (i.e. no dissipation across 0
load). Increasing RL should improve the dissipation; however, from the quadratic observation above, a peak would
occur beyond which, output would fall. This peak is located by quotient rule of differentiation, and equating it to 0:
′ −
= 2
.
� 0 − 2( + )
= � 2. =0, = .
( + )4
Therefore, the local maximum occurs where the load and source resistances equalize. Extending this to a general
case of complex impedance (such as of form Z = R + jX), we postulate that presence of a reactance will only circulate
electromagnetic power (component Q of apparent power S), preventing it from doing real work (P). In such a case,
the load impedance should thus be a conjugate of source impedance, so that reactive parts cancel out to leave pure
resistance; that is, for maximum power transfer:
ZS = R + jX = ZL* so that ZL = R – jx (12.6)
P a g e 33 | 60
Mathematically, we could retrieve the vector case of complex impedance to express power dissipation across
real component (RL) of load impedance (ZL) as:
�� ⁄ �
|
�� = =� . , + �| = + � + � + �� .
� | + �|
The fraction can be increased by reducing the denominator; one way

to doing this is to set XS = – XL; the remaining portion can be maximized
power transfer
by setting RS = RL as before. Resulting conjugate should then indicatively
provide a peak in the second order P L curve of FIGURE 33. Normalizing
the P L curve against peak power point leads to what is referred to as power
transfer coefficient τ = PL / Pmax [43]. Conjugation of the two impedances
at maximum power point cancels reactive load and equalizes real load,
impedance
hence reducing the previous equation to: ZS = ZL*
FIGURE 33: Power transfer peak
� � �
� =� . = � . = .
� � �
Then,
��
� = � = .
| + �|
By developing source power, we also find an expression for the efficiency of power transfer as:
� 2⁄ �2
� = = ( . )
| + |
�
= � = | ( . )
+ |
If plotted against different factors of RL /RS , it will be observed that τ peaks at matched impedance (i.e. RL =
RS) but efficiency is not maximum here; in fact, when ZS = ZL* the denominator reduces to 2RL and so the efficiency
= η0% only. Efficiency will only approach 1 when ZL > > ZS, but here the power transfer is not very high i.e. little
power is developed by source due to high total impedance, but most of it is transferred. Therefore, line transmission
circuits follow the paradigm of maximum efficiency (so as not to waste power on generation side), but wireless
systems tend to utilize load matching to shuttle high absolute values towards the load side. This paves the way for
dynamic load adjustment (through variable series resistors and shunt reactors), in order to optimize link power
transfer when external variables such as distance, alignment or shielding can interfere detrimentally.
4.2 Strong Coupling Regime

This work concentrates on resonant circuits as mentioned in Section 2.4, where inductive part cancels the
capacitive reactance, thus leaving behind only the real load component on both primary and secondary sides.
Primarily, such an arrangement prevents flux leakage into environment or wasted circulation of reactive power Q in
either side’s circuit. In effect, the system creates a synchronized LC-tuned filter when both sides resonate at the same
frequency, channeling high amounts of power through the pass-band. This magnetic resonance technique allows
operation in medium range [44] (see TABLE 7), and is known as strong coupling regime due to evanescent waves.
A resonant setup also carries the advantage that load matching as discussed in Section 4.1 becomes more
straight-forward, since reactive component of circuit approaches 0.
P a g e 34 | 60
TABLE 7: Summary of wireless transfer categorization by distance [45]

Field: Near Mid Far
Range: x ≤ 0.1λ 0.1 < x ≤ λ x> λ
Short x≤ φ Loose inductive or Capacitive
Medium x ≈ few φ Magnetic resonance Light EM wave or Laser
Long x >> φ EM wave EM wave
where x is transfer distance, φ is device/coil diameter, λ is wavelength29
Non-resonant coupling cannot perform well in distances approaching a coil diameter; electromagnetic energy
is largely wasted as copper losses in source oil or flux leakage into environment [46]. The resonant technique is more
efficient for such medium range due to reasons mentioned above.
4.3 Derivation of Circuit Impedance

To explore the concepts of load-matching and power transfer, we need to examine the behavior of circuit
impedance as coupling is varied (e.g. by increasing separation between coils). In this section, we proceed to analyze
series and parallel cases.
Series Resonant Circuit

Refer to Spreadsheet K
A series arrangement is shown in FIGURE 34, along with initial resonant parameters for experimental setup. R1
and R2 on either side are taken as sum of respective Resr and Rext values (the latter allowing variation as the former
is fixed by inductor’s characteristics).
Port-1 Port-2
Qty Unit Tx(1) Rx(2) C1 M1,2 C2
Vs Vrms 2.83 -
Rs Ω 50.0 - I1 L1 L2 – I2
RS I2
Resr Ω 0.462 0.360 RL
V2
Rext Ω 0 0
V1 Resr1 Resr2
R Resr + Rext 0.462 0.360
C μF 2.450 2.477
L mH 103.4 102.2 Rext1 Rext2
Zin Zout
FIGURE 34: Series Resonant Circuit for WPT
The arrangement may be visualized as a black-box with input (1) and output (2) ports on either side terminated
with RS and RL respectively; this analogy shall be useful later when determining scatter parameters. In the above
circuit, V1 is the source voltage whereas V2 is the developed voltage across receiver coil as a result of WPT. We now
attempt to construct impedance matrix using Ohm’s law with KVL (as earlier in Section 3.3) for the system using
the well-known principle of impedance addition for series (noting that XL = jωL and XC= /jωC ):
EQUATION-SET 13
�1 = �1 + 1 + 1 + 1 + �2 1,2 ( . )
1
�2 = �1 1,2 + �2 2 + 2 ( . )
29
The relation C = f λ holds for electrical AC source e.g. for 10 MHz, λ = 30 m
P a g e 35 | 60
However, V2 can also be defined in the opposite sub-section of the secondary circuit as:
� =−� + � .
Equating the last two expressions (for V2) and rearranging to yield I2 in terms of I1:
− �1 12 = �2 1 + + 2 + 2 ( . )
2
− �1 12
�2 = ( . )
1 + + +
2 2
2
By inserting I2 back into the port equation for V1, we arrive at:
2
12
�1 = �1 1 + 1+ 1+ + ( . )
1 1 + + +
2 2
2
�
= � = � + = � + .
�
It is very instructive to note that Z1 (impedance looking out from source) is simply the series total of the primary
side added to series admittance of secondary adjusted by factor of ωM1,2)2. This then is taken as a generalized rule
to ease the analysis when calculating other referred impedances. Then, impedance bridges marked Zin (voltage
developed across source and RS) and Zout (voltage developed across load), we obtain:
2
12
1 + 1 +1 +
1 + 2 + 2 +1
2
= 2 ( . )
12
2 + 2 +1 +
2 + 1 + 1 +1
1
Further, M can be expressed in terms of the coupling factor K discussed repeatedly in previous sections:
= √ .
Load matching can now be attempted during experimentation by plotting Zout against RL to see where they
equalize as coils are separated. As all other parameters in impedance equation are fixed, this matching will be a
function of variation of K (hence M) with distance; since Zout remains purely resistive in a resonant arrangement
(reactive cancellation on both sides of transformer), load matching becomes straight-forward as:
= � = + .
+
If no external resistances are added (Rext of FIGURE 34), then R1 and R2 are very small (esr) compared to RS;
ignoring them yields simplified load-match condition:
√ �.
� = , = , � = .
P a g e 36 | 60
This last result could serve as a useful method to approximate the required output resistance for load-matching;
the M (or K) curve could be pre-programmed into an IC, to electronically command variable impedance adjustment
to track the maximum power transfer of a series resonant WPT system. Alternatively, if values of RL and RS are
known, the mutual inductance can be directly approximated at a visual maximum power point (seen by watt-meter
reading, brightness of light or speed of motor as the load scenario may be).
Parallel Resonant Circuit

Refer to Spreadsheet N
A similar impedance analysis can now be readily conducted for parallel resonance (FIGURE 35).
Port-1 Port-2
M1,2
RS C1 L1 L2 C2
RL
V1 Resr1
Resr2
Rext1 Rext2
Zin Zout
FIGURE 35: Parallel Resonant Circuit for WPT
On each side, the capacitor is in parallel with the inductor coil, which in turn is in series with its esr. We apply
the circuit rule of parallel impedances (ignoring Rext as 0 for general case):
EQUATION-SET 14
1 1 1 1 1 1 + 1 + + 1 + 1
= + + = ( . )
1 1+ 1 ( 1 + 1)
1
Although esr value is negligible compared to inductive impedance at higher frequencies under question (of
order of kHz), it is this small value that keeps the LC arrangement from resonating to infinity (through reactive
circulation in parallel). Therefore, its series effect will keep the 1/Zeq value LC tank from becoming 0, hence allowing
numerical analysis. The part of the primary that forms portion of Zin can be expressed as:
− +
= + = .
�′ + +
A similar reasoning can be applied to get Zsec and Zsec’. Then by applying the rule of impedance addition and
factoring by ωM12)2 as proven previously, Zin is simply:
� = �′ + ⁄ .
�
And analogously for Zout to yield:

2
1 + 1 12 [ ( 2) + 2 + 2 + 2 + 2 ( 2 )]
+
1− 2
1 1 + 1 1 ( 2 + 2 )( 2)
= 2 ( . )
2 + 2 12 [ ( 1) + 1 + 1 + 1 + 1 ( 1 )]
+
1− 2
2 2 + 2 2 ( 1 + 1 )( 1)
P a g e 37 | 60
As is evident, this case is more involved than series one; equating Zout against RL for load matching does not
yield any more a simplified expression. This situation was monitored in practice by setting up Spreadsheet N to chart
the changes in Zout with distance (hence with variation in K or M).
Series or Parallel: Which Circuit?

A question that arises at this point is which LCR arrangement to prefer for power transfer application. This can
be resolved by looking at impedance derivations of EQUATIONS 13.8 and 14.3. The series application nullifies
reactive impedance to 0 (i.e. capacitance cancels reactance to give LC = 0), leading to a high current at resonance
limited only by resistive components. In parallel arrangement however, the reactive impedance becomes infinite (i.e.
1/LC = 0), appearing as an open circuit, with total voltage drop prevented only by resistive part of circuit. This means
therefore, that series LCR resonators can be used for applications where high power transfer is required (such as
vehicle charging mats), because a large current produces a large magnetic flux by Ampere’s law EQUATION 7.1.
The parallel setup can be used where power has to be limited due to sensitivity of use, such as biomedical implants.
4.4 Derivation of Scattering Parameters

S-parameters of a port network indicate the degree of reflected return or transmission of AC wave signal; this
is analogous to optical behaviour [47]. The S-parameters correspond to the square root of electrical power that was
discussed above, and are important in characterizing attenuation of circuit models for wireless power networks. We
are interested in S21, which represents transfer coefficient from source (Coil-1) to load side (Coil-2). In evaluating
thus, the impedance calculations from Section 4.3 shall be utilized.
Modifying the elaboration of two-port network parameters by Pozar [48], we get:
EQUATION-SET 15
= , ℎ ∆ = + + − .
∆
Subscript numbers here refer to impedance matrix of first port with respect to the second e.g. Z11 means primary
referred to primary, but Z21 means secondary referred to primary. The original theory by Pozar is for transmission
lines, where Z0 is noted singularly (without further subscripts) throughout the expression as characteristic impedance
(looking into the line from one end, and often matched by terminating resistance for maximal transmission). For the
WPT case therefore, the student has modified this concept of Z0 to that of terminating resistance for each port
considered separately in the various parts of the fractional expression for S21 above. For example, at port 1, Z01
indicates the terminating impedance of RS and at port 2, Z02 indicates the terminating impedance of RL.
For the transmitted impedance between two sides i.e. Z012 the student has decided to evolve the analogy of
geometric mean from mutual inductance in that M12 = jω L1L2) . Therefore Z012 = RSRL).
Series Resonant Circuit

By retrospection on EQUATIONS 13.1 and 13.5, and their comparison with EQUATION 15.1, we extract
impedance matrix for the series circuit of FIGURE 34, from point of view of two-port theory:
1 + +
�1 1 1+ 12 �1
1
= ( . )
�2 1 + + + �2
12 2 2
2
�1 11 + 01 12 �1
= ( . )
�2 21 22 + 02 �2
We then consolidate the referred impedances in table below:

P a g e 38 | 60
TABLE 8: Consolidation of S21 by Z parameters for series setup

Impedance Meaning Expression
Z11 Primary impedance, wrt primary 1/jωC1 + jωL1 + R1
Z22 Secondary impedance, wrt secondary 1/jωC2 + jωL2 + R2
Z21 = Z12 Transmitted impedance jωM112
between secondary and primary
Z01 Terminal impedance, primary RS
Z02 Terminal impedance, secondary RL
Z021 Terminal impedance, mutual RSRL)
Therefore,
2 12 √
21 = ( . )
1 + 1 2 2
1+ 1+ + 2+ 2+ + 12
1 2
Parallel Resonant Circuit

Again, by retrospection on EQUATIONS 14.1 and 14.3, and their comparison with EQUATION 15.1, we extract
impedance matrix for the parallel circuit of FIGURE 35, from point of view of two-port theory as done for series
circuit. For simplification, Resr has been ignored:
1
+ 12
�1 1− 2
1 1 �1
= ( . )
�2 2 �2
12 +
1− 2
2 2
�1 11 + 01 12 �1
= ( . )
�2 21 22 + 02 �2
We again consolidate the referred impedances in table below:
TABLE 9: Consolidation of S21 by Z parameters for parallel setup
Impedance Meaning Expression

Z11 Primary impedance, wrt primary jωL1 /(1 – ω2C1L1)
Z22 Secondary impedance, wrt secondary jωL2 /(1 – ω2C2L2)
Z21 = Z12 Transmitted impedance jωM112
between secondary and primary
Z01 Terminal impedance, primary RS
Z02 Terminal impedance, secondary RL
Z021 Terminal impedance, mutual RSRL)
Therefore,
2 12 √
21 = ( . )
2
( 1
+ )( 2
+ )+ 2
1− 2
1 1 1− 2
2 2
12
P a g e 39 | 60
4.5 Demonstration of Inductor V-Pickup Variance

From Section 4.3and 4.4, it is clear that the coupling factor K (and hence M) plays a central role in changing the
impedance and scattering parameter S21 as the coils are moved apart; the remaining variables of these expressions
are fairly constant. We have extracted the K-curves from our earlier investigations in Section 2.4.1, which decay
inversely with distance. The presence of M term in both numerator and denominator of S term means that it will
influence the latter in a non-linear fashion, possibly even causing local minimum or maximum points.
4.5.1 Explanation of scattering by V-pickup measurements

Refer to Spreadsheets G, H, J
It is mentioned earlier in Section 4.4 that S21 is square root of electrical power transferred from Coil-1 to 2. By
looking at the basic relation between power and voltage (P α V2), we can then stipulate that S21 would generally
follow the trend of V-pickup (GV-L2/L1); the latter should be used with caution as it is not strictly true because the
value of S21 de-factors power that is reflected back. The author measured voltage pickup on the wireless coils in a
parallel resonant arrangement (FIGURE 36). Some main observations are:
1. Voltage gain peaks at some non-zero distance (about 2.5 cm) for resonant circuit; this may be attributed
to the concept of load matching (where real loads equalize and reactance cancels) by changing
equivalent impedance as seen by the source (Zout = RL). For non-resonant circuits, such a phenomenon
does not occur.
2. When resonance is disturbed (i.e. C added is less or more than resonant value), the V-pickup tends to
be less than the resonant circuit at same axial separation, but more than non-C circuit.
A small source of experimental error must be mentioned here. The calculated C values for resonance at f = 10
kHz may be off-mark when set up due to component tolerances. Indeed by sweeping, it is seen that the maximum
response tends to occur around 10.11 kHz and so the operating frequency is set accordingly. Another interesting
aside is that the input and output voltage are not in phase despite resonant cancellation; this can be explained by
presence of impedance jωM . To decrease this reactive contribution, a re-test is done with large RL (1 kΩ) such that
RL > > ZM. It is then seen that phase difference decreases.
Qty Unit Tx(1) Rx(2)

Vs Vrms 2.83 -
Rs Ω 50.0 -
RL Ω - 50
Resr Ω 0.462 0.360
C μF 2.450 2.477
L mH 103.4 102.2
Note: The circuit follows the

setup of FIGURE 35; source Z
is 50 Ω and MSO input Z is 1
MΩ (i.e. high impedance
measurement). For disturbance
readings, capacitor bank of
receiver side is changed by
orders of μF or nF (increase or
decrease at different points)
FIGURE 36: Voltage gain for parallel LCR arrangement
P a g e 40 | 60
To further test the dynamics behind the resonant voltage pickup curve above, the author plotted ratio of ZL/Zout
using parameters tabulated in FIGURE 36 and expressions from EQUATION 14.4. From the results in FIGURE 37,
contradictory forces can be observed. Although impedance never matches to 1 (for reasons discussed in
Section 4.5.2), it does plateau out, whereas K curve falls with distance (see exponential regression in Section 2.4.1).
Crossover occurs around 2.5 cm, which is also the maximum V-ratio point in FIGURE 36.
0.44 0.900
0.44 0.800
0.44 0.700
Coupling factor K
Impedance ratio
0.600
0.43
0.500
0.43
0.400
0.43
0.300
0.43 0.200
0.43 0.100
0.43 0.000
0 5 10 15 20 25 30 35
Axis Title
ZL / Zout K
FIGURE 37: Contradictory forces in a parallel resonant WPT system
In fact when using the circuit parameters of FIGURE 36, a red LED30 is connected to secondary coil of WPT
system across RL (with primary fed at 2.83 V rms) and distance varied; it is seen brightest at around 2.5 cm separation.
The LED can be thought of as a simple voltage sensing device. By adjusting RL to very low (or 0) values, a situation
was achieved where the LED does not light up as the coils are touching; conversely, it lights up only after some
separation occurs. This is a non-intuitive result, explained by cross-over maximization discussed above, and if further
illustrated by delayed peaking of S21 in FIGURE 39.
0.80
Finally, the correctness of S21 parameter
0.70
calculated for series WPT circuit in
0.60
EQUATION 15.4 is tested with various
0.50
frequencies inserted into the mathematical
S21
0.40
model, where LC parameters equalize at 10
0.30
kHz. The S21 curve for resonant case (f = 10
0.20
kHz) is clearly raised above other frequencies
0.10
in FIGURE 38, thus confirming indirectly the
0.00
S-parameter calculations. 0 5 10 15 20 25 30 35
f 10 kHz f 10.5 kHz f 9.5 kHz f 8 kHz f 12 kHz
FIGURE 38: S21 frequency response
30
Forward-voltage of 2V
P a g e 41 | 60
4.5.2 Comparison of lossy environment against ideal scenario

Refer to Spreadsheets I, K, L
From FIGURE 37, we see that load-matching (RL/Zout = 1) is not always achieved in practical laboratory
environment, as the graph begins to tail off to a plateau below 1. A major reason for this is the fixed source resistance
(RS = 50 Ω) and esr values of coils; this constrains load matching to a narrow range of external R additions.
The student explored this concept further by setting up test Spreadsheets I, K, L for series LCR arrangement, to
allow user to check various combinations of circuit parameters. Representative summary of quantitative experiments
is displayed in FIGURE 39. When actual esr coil values are included in analysis, the maxima of S21 and voltage
pickup curves do not coincide with impedance-match point (left graph); neither does S21 ever achieve the maximum
value of 1. In this instance, a value of RL = 0.5 Ω was connected to secondary for matching at around x = 4 cm. This
deviation from the ideal prediction (that power transfer peak should coincide with load-match point) can be explained
by comparing RL (0.5 Ω) with receiver coil’s esr value (0.46 Ω). Since the two are comparable, we conclude that a
large part of power transferred at resonant, impedance-match point is actually wasted in esr of Coil-2.
Trying various permutations with Spreadsheet K, the author has observed that due to the limitation of RS being
fixed at 50 Ω, circuit parameters of our 10 kHz series LCR circuit compute Zout of smaller orders only (< 1 Ω); trying
to match it with esr will only increase copper losses in the coils.
However, when esr of both coils are removed by simulating 0 values, (right graph), the Z-match point coincides
well with peaks on S21 and V-gain curves; here S21 approaches a value of 1 at its peak, which is its maximum
possibility, unlike the earlier real scenario where it does not attain 1. This is an interesting finding, which implies a
conclusion that for useful load matching31, variable control should exist for RS as well as for RL and intervening
circuit resistors on each side (Rext) of FIGURE 34, because other parameters such as coil inductance and esr cannot
be changed. Indeed, the student observes that changing RS in calculative model allows a less constrained load-
matching with greater chances of S21 peaking.
2.500 18.00 7.00

16.00 6.00
2.000 14.00
5.00
12.00
V, S ratio
1.500 4.00
Z ratio
10.00
8.00 3.00
1.000
6.00
Z-match point 2.00
0.500 4.00
2.00 1.00
0.000 0.00 0.00

0 10 20 30 40 0 10 20 30 40
Axial separation (cm) Axial separation (cm)
Zload/Zout mag(S21) VL2/VL1 ZL/Zout mag(S21) VL2/VL1
FIGURE 39: Comparison of lossy (left) against ideal (right) circuit response
Again, for RL = 0.5 Ω and real esr parameters, the researcher attempted to verify importance of resonance in
achieving high V-gain; laboratory results are graphed in FIGURE 40. It can clearly be observed that V-pickup ratio
31
Useful in the sense that load matching also allows max S21
P a g e 42 | 60
between coils as well as that between load and source terminals is larger for resonant LC scenario; these gains fall
from their respective curves when the C-bank on receiver side is disturbed (represented by dots).
0.060 2.500
0.050
2.000
0.040
1.500
Vload / VS
+1μF
VL2 / VL1
0.030
+470 nF
1.000
0.020
0.500
0.010
-1μF -1μF -7 nF
0.000 0.000
0 5 10 15 20 25 30 35
Vload/VS Dis.Vload/VS VL2/VL1 Dis.VL2/VL1
FIGURE 40: V-gains for resonant and disturbed cases
4.5.3 Frequency sweep data models

Refer to Spreadsheet M, O
The final part of experiments looked at LCR series WPT response to frequency sweep, where load matching
(ZL/Zout = 1) is achieved at coils separation x = 4 cm with RL = 50 Ω and Rext = 0 (of FIGURE 34).
1.400
1.200
1.000
0.800
0.600 1.200-1.400
Zload/Zout
0.400 1.000-1.200
0.800-1.000
0.200
0.600-0.800
0.000
0.400-0.600
0 1
2 3 0.200-0.400
4 5
6 7 0.000-0.200
8 9
10
11
12
13 15
14 11
15 10
20 9
25 5
3
30
FIGURE 41: Impedance ratio for frequency sweep

P a g e 43 | 60
FIGURE 42: Voltage pickup at receiver, for frequency sweep
The superiority of resonance is evident from 2.000

FIGURE 41. This was further demonstrated by
1.500
measuring receiver voltage pickup for the same
VL2/VL1
circuit; the consequent surface curve is plotted in

1.000
FIGURE 42. We observe significant elevation for
the resonant frequency of 10 kHz. In addition, 0.500
measuring the coils V-pickup ratio (VL2/VL1) for f
= 10 kHz reveals a peak at axial separation x = 4 0.000
cm (FIGURE 43); this corresponds to the load 0 10 20 30 40
match point as mentioned above. Axial separation (cm)
FIGURE 43: Coils V-gain for resonant frequency
1.2000 250
Before embarking upon simulated
1.0000 200
environment in Chapter 5, a frequency sweep of
phase(S21) / deg
0.8000 150
mag (S21)
S21 was performed to display its peaking effect at

resonance, using spreadsheet calculations. 0.6000 100
90o
Mimicking real esr (orange curve in FIGURE 44) 0.4000 50
lowers S21 due to decreased quality factor32 (as
0.2000 0
resistive dissipation increases), in line with our
previous findings in Section 4.5.2. It can also be 0.0000 -50
seen that impedance angle of S21 (plotted on - 5,000 10,000 15,000
secondary axis) is 90o at resonance; this result will Frequency (Hz)
be useful in explaining frequency split
|S21|, esr = 0 |S |, esr ≠ δ(S21), esr = 0
phenomenon of Section 5.2.4.
FIGURE 44: Scatter behavior at resonant frequency
32
Q-factor is ratio of coil impedances as (ωL)/R, and it decreases with rising esr
P a g e 44 | 60
Chapter 5 MICROWAVE AND CIRCUIT SIMULATION ENVIRONMENT

The final portion of investigative work briefly attempts to simulate series resonant, 2-coil system in 3D full-
wave studio and its comparison with equivalent AC circuit. The solutions were obtained for frequency domain,
primarily for power scattering parameters S21 (power ratio load to source, refer to EQUATION 15.4) and S11 (power
reflected back to source). The main motivation here was to see how circuit extraction compares with EM wave
model, and how in general virtualization concurs with real laboratory testing.
5.1 3D Setup Parameters

An overview of the EM simulation environment is provided here, to enable the reader to replicate results if
required, and to define the framework employed. Software used is listed in APPENDIX 1.
5.1.1 Coil Dimensions

A coil pre-fabricated in structure spatial software33, was imported by the author into Microwave Studio using
.SAT method. This coil was the departure point for earlier investigations on hybrid cars performed by Dr. Rotaru
[20], and was created by him physically as well as virtually. Geometry is similar to two-layered flat pancake produced
in Section 2.1 with 21 lower-layer and 7 outer upper-layer turns; a notable difference however is that each of the
turns in lower layer first connects to its counterpart in upper layer, where the latter joins the next turn in lower layer
(unlike our lab fabrication of FIGURE 6 where the two layers are joined in series once). FIGURE 45 shows this turn-
by-turn linkage, along with properties of the corresponding real metal coil.
Dimension Symbol Value

Coil outer φ34 D 300 mm
s
Coil inner φ d 150 mm
D
Conductor element - Litz Cu
Conductor φ35 - 2.6 mm d
Wire gauge - AWG 33 Lower
layer
Wire strands - 210
Inter-turn gap36 s 3.6 mm Upper
N-upper N1 7 turns layer
N-lower N2 21 turns Interlinkages
FIGURE 45: Inter-layer linkage structure, and dimensions of modelled coil
The length of the wire used in constructing the coil can be calculated by polar integration [49], or by average of
circumferences as:
EQUATION-SET 16
2
= 2 + ( ) ( . )
= 2� ( + ) 2 ( . )
33
ACIS 3D Spatial Modeling Engine
34
Diameter
35
Measured by point-picking in CST
36
Calculated as [(300 – 150) / 2] / 21 = 3.57 i.e. subtraction of inner radius from outer radius, divided by no. of turns
P a g e 45 | 60
EQUATION 16.2 is a simplified estimate of the integral; it considers a mean radius (as average or inner r and
outer R radii of coil) to evaluate the circumference, which is then multiplied by number of turns. Employing this
gives a total length of 20.892 m (FIGURE 46).
Portion Length Calculation Result (m) 75 + (14/21)x(150 – 75)

l1 (lower layer) 2π x 21 x (75+ 150)/2 14.844
l2 (upper layer) 2π x 7 x (75 + 2/3 x 75+ 150)/2 6.048 150 mm
l tot (total) l1 + l2 20.892
FIGURE 46: Calculation of spiral length 75 mm
5.1.2 LCR Values and Mesh Setup

While skin-effect is negligible for frequencies in the order below 1 MHz [50], we consult AC resistance OEM
specifications [51] for accuracy. With 210 strands (FIGURE 45) and single bundle operation, we obtain a unit value
of 1.0591 Ω / 103 ft, or 3.53 x 10-3 Ω / m37. Applying this unit value to the length of coil calculated in FIGURE 46
yields coil R.esr = 0.0738 Ω.
Ignoring parasitic capacitance as for preceding laboratory work, we use two

C1
capacitances of 22.6 nF, in series with the coil to produce C = 11.3 nF per coil38.
port
An excitation port of 50 Ω impedance (similar to RS of Section 4.3), expending
1 W, is fed into the 3D coil. This completes the LCR arrangement on primary side.
The receiver coil is then produced by using a Transform > Translate > Copy C2
command on copper components bunched together, setting a variable translation
FIGURE 47: Lumped elements
vector so that various inter-coil axial separations can be studied.
The whole system is immersed in vacuum (using Normal, Open type boundary conditions, with extent reaching
far beyond coil dimensions e.g. 1000 mm on xyz axes). This WPT construct is first resolved as tetrahedral meshing;
adjustments to local and global properties are made by trial and error to obtain a problem size of around 250,000
tetrahedrons (see FIGURE 48). This provides a compromise between quality refinement (average 0.6 out of 1) and
computational load; larger mesh sizes were seen to make the frequency solution inviable on a 16 GB-RAM machine.
Global properties
Material based refinement Smooth density transitions (80% factor)
4 steps per wavelength; min 10 steps High surface smoothing (60% factor)
No lossless or thin-wire assumptions Surface method: general
Curve refinement ratio 0.05 Volume method: Delaunay
Local properties
Maximum step width 15 Considered for simulation and bounding
FIGURE 48: Tetrahedral meshing properties used; example of a circular mesh on the right
The system was swept by a frequency domain solver accurate to 2nd order, using general purpose method (not
fast-S), with mesh accuracy 10-4 and S-parameters normalized to 50 Ω, for frequency range 0 to 1 MHz; adaptive
mesh refinement feature was switched off to assist quicker convergence. At these settings, the total meshing and
37
1000 ft = 300 m
38
Reduction rule of capacitance in series, as 1/jωC = 1/jωC1 + 1/jωC2 leaving 1/C = 1/C1 + 1/C2
P a g e 46 | 60
frequency sweep solution took around 30 hours per WPT system. S11 and S21 parameter curves were obtained as a
result for various configurations listed in TABLE 10.
TABLE 10: Varied configurations simulated for full-wave model

Variable Symbol/Unit Values
Axial Separation d / mm d = { 25, 50, 75, 100, 150, 200 }
Shield from Receiver h / mm d = 100, h = { 0, 25, 50, 75, 100, 200 }
Transmitter Rotation a / deg d = 75, a = { 10, 20 }
FIGURE 49 shows the curve of S21 for axial separation d = 200 mm (to avoid
short-range split discussed in Section 5.2.4); it shows a peak at 101 kHz (resonant
point). From EQUATION 8.1, we obtain thus:
= × =
� × . × −9
and L = 220 μH
peak S21 at
Further, at the same frequency, full-wave model shows a sharp reversal of 0.101 MHz
phase plot, similar to FIGURE 44 of lab experiments; this again confirms
resonance point (around which the system behaves as capacitive or inductive on
either side, with opposing phasors). LCR parameters for the two identical coils
in our system thus established, we proceed to circuit equivalence. FIGURE 49: Resonance by f-sweep
5.2 Equivalent Circuit Design

A series resonant circuit was set up in Simplorer, with LCR parameters adjusted to laboratory conditions (values
from Section 4.3) or 3D model (values from Section 5.1.2). Sweep was set accordingly; for example, resonance
around 10 kHz is expected for laboratory circuit, so solver was set for linear range 1 to 30 kHz in increments of 1
kHz. On the other hand, to resolve full-wave system, range 1 to 500 kHz was used since resonance occurs around
100 kHz. Also, as Cu used in 3D model has very high conductivity (5.96 x 107 S/m), esr values may be ignored, or
instead included for finer accuracy.
5.2.1 Circuit Setup and Scatter Calculation by Z

To obtain parameter S21 from AC solver, we extend the two-port analogy developed earlier in Section 4.4. Now
Zxy in a multi-port network is defined as the impedance looking into Port-x when only Port-y is loaded; alternatively,
it is the ratio of voltage output at Port-x to current driven in Port-y with all ports except y open [48]. We arrive at
formulation thus:
EQUATION-SET 17
�
= | → ℎ = ≠ .
For example in the two-port system of FIGURE 34, Z11 would correlate voltage and current on Port-1 while
Port-2 was open; similarly Z21 would measure voltage across Port-2 and current on Port-1 with Port-2 open.
For symmetrical circuits (where secondary mirrors primary in terms of LCR values), Z21 = Z12 and Z11 = Z22
naturally, and thus only half measurements need be taken in AC solver; however, for some experiments conducted
during course of Chapter 4, RL ≠ RS. A dual circuit (parts A and B) was therefore simulated as shown in FIGURE 50
for independent evaluation of all four Z-parameters to calculate S21 through EQUATION 15.1. Note that K-value is
kept same in both portions of the set up.
P a g e 47 | 60
Part A: To measure Z11 and Z21
Part B: To measure Z22 and Z21

Port-1 Port-2
FIGURE 50: Dual measurements for S21 solution of asymmetrical circuit
5.2.2 Extraction of Coupling Factors

1
The circuit of FIGURE 50 corresponds to
Coupling factor K
laboratory setup of Spreadsheet L2 (RL = 0.6 Ω). 0.8
Adjusting the K values in AC solver by trial and error 0.6

to equalize the magnitude of S21 at 10 kHz (resonant 0.4
point) against experimental values for various axial
0.2
separations, the author extracted corresponding
coupling factors. Results from the solver are plotted 0
against K-curve measured in lab (Section 2.4.1). As 0 5 10 15
FIGURE 51 shows, there is close correspondence. Axial separation (cm)

The slight difference could be explained by flux
Extracted K Measured K
leakage error in real environment (which is not the
case for simulated circuit), causing a lower power
FIGURE 51: Correspondence between measured and
transfer at same coupling levels. simulated coupling
Further attempts to compare extracted K-curves from full-wave and laboratory work by normalizing the axial
distance to coil diameter, yielded no significant results. Extended research work is proposed in this regard, to find a
way to meaningfully compare coupling values between WPT systems of different dimensions.
P a g e 48 | 60
5.2.3 Power Measurements

As discussed in Section 4.4, scatter corresponds to square root of power transmission or reflection between ports
of a circuit. So while S21 indicates degree of power transmission from source to load, S11 would derive from power
reflected back to source [52]; the author thus considers the latter as the difference between total source power and
the fraction absorbed by load. With this approach, we proceed to derive power components39:
EQUATION-SET 18
� = �∠ � �∠− � = �� ∠ �−�
⃗⃗⃗ ⃗⃗⃗∗
=� where is angle from reference .
In the above equation, conjugate of vector I (denoted by I*) is necessitated so that phase difference between
current and voltage (and not phase addition) results from polar multiplication. Real power is then simply the cosine
component of apparent power (S), or dot product of above vectors:
⃗⃗⃗ . ⃗⃗⃗∗ = ��
�=� = � where denotes real component .
The above phasor resolution uses peak vector values (magnitude); to reduce this to actual sinusoidal scenario,
we next utilize the √2 relationship between peak and rms values to give for load:
⃗⃗⃗⃗� ⃗⃗⃗⃗⃗
� �
∗
�
�� = ( )= ( �⃗⃗⃗⃗⃗� ⃗⃗⃗⃗⃗
� )
∗
where � = .
If negative of P L above is taken, the resulting curve inverts to correspond to in shape and orientation to results
of S21 from full-wave solver (FIGURE 52). Now, the power reflected back to source becomes:
� = � − �� = ( ⃗⃗⃗⃗⃗� ⃗⃗⃗⃗⃗
� ∗
� )−
⃗⃗⃗⃗⃗� ⃗⃗⃗⃗⃗
( � � )
∗
.
To obtain scatter coefficients, we would have to normalize40 these quantities to total power base P S, as scattering
reflects a simple ratio of output to input:
� �
∝ √ ⁄� and ∝√ � � .
These equations were implemented in circuit solver using functions Re, Conjg, Sqrt and algebraic operators.
Frequency sweep results in resonance points at 10 kHz and 106 kHz for laboratory and full-wave circuits
respectively; these values are very close to those seen earlier.
5.2.4 Explanation of Double-Peak Effect

It was observed at simulations for shorter ranges in 3D model (or high K in AC solver), that power transfer
suffers a frequency split around resonant point, which itself is at a local trough (see results in FIGURE 52) so that
resonance diverges from Eigen frequency ω0. This problem is not seen for longer distances (as in FIGURE 49).
Although this phenomenon is well-known, much theoretical analysis to explain it in terms of circuit parameters has
not been done [53]. The author has attempted to assess the cause by looking at earlier analyses of impedance in
EQUATION 13.8, scatter in EQUATION 15.4 and behaviour of S21 in FIGURE 44. At resonance, impedance
magnitude is minimized (as reactive component nullifies, hence S21 peaks) but impedance angle swings to 90o; the
resulting large current promotes internal losses in of esr and RS, and little power is transferred to loads due to reactive
circulation. Above and below the resonant point, this behaviour reverses and power is shuttled to load more
efficiently. An alternative reasoning is that the system is over-coupled at close range; the receiver acts as a source
and reflects excess power back to origin. This causes a trough in load power curve, but a crest in source power (see
FIGURE 52). When separation is such that total transmission to receiver occurs (i.e. S21 = 1, S11 = 0), critical coupling
39
Scattering parameter S is not to be confused by apparent electrical power S
40
Normalization of power curves against source to obtain S-parameter removes frequency-splitting behavior
P a g e 49 | 60
is achieved. Beyond this point, only single resonant peak is seen; it can be seen from EQUATION 15.4 that in such
scenarios, a diminishing mutual inductance causes drastic fall in power transfer (i.e. referred impedance becomes
negligible); the transfer behaviour here is thus simplified.
P_S (rfl)
FIGURE 52: Shape correspondence between full-wave scatter (left) and AC circuit power (right)
5.2.5 Scatter Calculation by V

A classic definition of scatter parameters is by travelling voltage waves between ports [52]; for instance, S21 is
ratio of voltage totally absorbed by RL (0 reflection) to that incident on Port-1. Here thus, the concept is similar to
voltage gain. It is important that voltage be considered as directional quantity here; it is not simple magnitude reading.
From FIGURE 53, we observe that the above ratio is VRL/VZin. This can be rationalized by voltage division in
terms of impedance, provided there is no reflection from load:
RS Zin
EQUATION-SET 19
�
� � = �. . LCR
+ �
series RL VRL
VZin
VS WPT
�
� � = �. .
+ �
Port-1 Port-2
��
ℎ = .
��
FIGURE 53: Voltage division across 2-port WPT
For symmetrical network, at resonance when all reactance nullifies, the ratio in EQUATION 20.3 maximizes
because Zin ≈ RL (slight differences due to coils’ esr). The EQUATION 20.2 would thus evaluate to � � = . �
since Zin = RL = RS. Substituting this in our original definition of S21, we get:
��
= = .
�� . � �
We generalize x as a factor that depends on the maximum value of ratio � � � measured by frequency sweep.
By comparison, the adjusted expression for S21 then becomes:
P a g e 50 | 60
��
= × .
− max mag � � � �
EQUATION 20.5 was implemented in Simplorer, and the results obtained for S21 compare correctly with those
by Z-method from Section 5.2.1 (see FIGURE 54) with same LCR and K inputs.
FIGURE 54: Concurrence of S21 curves between V-gain (left) and Z (right) methods
5.3 Impact of Shield

To investigate the effect of conductive shield in
1
full-wave model, aluminium sheet of dimensions 400 x
0.9
400 x 5 mm (larger than coil diameter) was set up at
0.8
variable distance from receiver coil. The arrangement is 0.7
similar to that of laboratory in Section 3.2. Resulting f-
mag(S21)
0.6
sweep S21 data, extracted as ASCII CSV files, is plotted 0.5
in FIGURE 55. We see a fall in power transfer with 0.4 ω0 shift
increasing shield proximity, indirectly signifying a 0.3
decrease of coil coupling. This is in agreement with 0.2
experimental results obtained earlier. 0.1
0
A further simulation was conducted with Fe instead 0 50 100 150 200 250
of Al; no differences were seen in results, lending weight Frequency (kHz)
to our inference in Section 3.2.3 that a shield disrupts
WPT system by electric conduction of eddy currents, no Al 0 mm 25 mm 50 mm
acting as flux leakage sink (third inductor), and not by
virtue of its own magnetic properties. FIGURE 55: Effect of Al shielding in 3D EM model
Another feature seen clearly in FIGURE 55 is the forward shift of resonance point as shield is brought closer;
further research in this regard is required. An implication could be offered, in that the third inductor affects referred
impedance (magnitude and angle) seen by source in a way similar to that of frequency shifting phenomenon discussed
in Section 5.2.4.
P a g e 51 | 60
5.4 Quasi-static Approximation for Future Simulations

To speed up computation in future models, it may be useful to approximate the full-wave problem from a
theoretical physics perspective, by looking at Maxwell’s equations. For situation where EM waves can traverse the
system’s dimension (L) in a duration much smaller than period of the wave itself41, time derivative aspects of
Maxwell’s equations may be safely neglected [54]. In our low-frequency and short-distance case of free space,
magnetic waves would traverse a separation distance of even 10 m in 33 ns, which is miniscule compared to τ = 10
μs for f = 100 kHz 42, or:
EQUATION-SET 20
≪ , where is speed of EM waves in free space .
In effect, we ignore time retardation of measurement with respect to wave’s arrival, for scenarios when WPT
dimensions are far smaller than wavelength of the AC source used. For low frequencies of orders of kHz (λ = 3 km
for 100 kHz), this allows treatment of magnetic fields as quasi-static (i.e. equilibrium that does not necessitate
knowledge of previous system state at the observation instant). Because the problem changes at a rate much slower
than speed of wave, and copper circuit has high conductivity (low-loss), the magnetic field (H) of inductors is taken
to depend only on AC source current; contribution of electric field (due to moving charges and thus displacement
current term of Ampere-Maxwell law) is ignored to yield:
∇ ×�=� + � � → ∇ ×� = , as �=� � .
According to Haus and Melcher [54], the error ratio of magnetoquasistatic approximations is � , which
-5
in our case works out to a small value of 1.1 x 10 (at L = 10 m, f = 100 kHz). For continued research on the case
presented in this thesis, it is therefore suggested that the full-wave simulation be reduced to a quasi-static finite
element model for decreased computation load.
41
Considered as instantaneous propagation
42
Using c = 3 x 108 m/s in free space, and f = 1/τ
P a g e 52 | 60
Chapter 6 CONCLUSION
In this work, the researcher investigated the behavior of resonant wireless power transfer systems using a
combination of hardware and software methodologies. Experimental results were processed by quantitative
modelling and mathematical derivations, while simulations were by full-wave and circuit analyses. The dissertation
forms an analytical extension of grounded theory, especially the work undertaken by Dr. Mihai Rotaru for hybrid
cars. Initially, bifiliar pancake coils were constructed to measure a range of parameters. It was found from a
comparison of techniques, that the impedance (I-V) method for inductance measurement is the most suitable for
repeated readings in terms of accuracy and speed. Coupling factors for sweep of axial distance were observed to
improve with addition of capacitance, and its relation to reactive impedance nullification at resonance was pointed
out. Whereas earlier research works do not extensively focus on misalignment, this work also investigated the effect
of radial misalignment between coils. It was proven that planar coupling is inferior to axial one; the reasoning was
derived from Biot-Savart field behaviour.
We then proceeded to analyze effects of metallic shied in vicinity of receiver coil; power transfer was observed
to degrade severely at close proximity as K-factor measurements fell. The hypothesis of eddy current interference
was confirmed by lifting of K-curves through gashing. Mathematically, the shield was modelled as an inductor using
3-port network theory, and predicted trends of coil parameters were confirmed through experimentation (M falls,
esr increases with shield proximity).
Maximum power transfer theorem was also investigated; two-port derivations for LCR series and parallel
arrangements were performed for impedance and scattering parameters, and experimental values extracted. Surface
curves for load ratio and voltage pickup were constructed via frequency sweep. Load-matching and resonant
frequency were seen to maximize S21 in general; however, where source resistance was fixed but load resistance
approached the order of esr value of coils for load-matching, it was seen that power transfer may not peak where
RL/Zout = 1. This conclusion was expounded by cross-over between rising Z-ratio and falling K curves with distance.
For commercial solutions this means that there must be a method of adjusting source impedance if meaningful
maximization of S21 is to be achieved.
Final part of this dissertation examined the possibility of using 3D (full-wave) and AC (circuit) models for
approaching the WPT problem. Correspondence between these two methods, and the earlier experimental analysis,
was established by comparing distance sweeps of coupling factors, shield intervention and derivation of S21 by
electrical power. Frequency splitting phenomenon was confirmed at close ranges, and an attempt to explain it was
made by considering impedance magnitude and phase at natural frequency.
While this work has attempted to explore the notion of wireless power transfer in both a theoretical (circuit and
port derivations) and practical (laboratory and simulation work) manner, some interesting findings have been
identified throughout as potential areas meriting deeper research in the future. These include the effect of mismatched
coil size on mutual inductance, the relation between source frequency and coupling, establishing the possibility that
off-resonant circuit arrangement can allow a greater power transfer magnitude but lower ratio of power transfer and
determining a non-interfering distance of shield from WPT system in terms of order of coil diameter.
P a g e 53 | 60
Chapter 7 BIBLIOGRAPHY
[1] A. Waser, "Nikola Tesla's Wireless Systems," 29 Jul 2000. [Online]. Available: http://www.andre-
waser.ch/Publications/NikolaTeslasWirelessSystems.pdf. [Accessed 01 Jul 2014].
[2] N. Tesla, "Apparatus for utilizing Effects transmitted through Natural Media". United States Patent
Office Patent 685,956, 05 Nov 1901.
[3] Marconi Wireless Telegraph Co. of America V. United States at Supreme Court of USA, 1943.
[4] K. Fotopoulou and B. W. Flynn, "Wireless Power Transfer in Loosely Coupled Links: Coil
Misalignment Model," IEEE Transactions on Magnetics, vol. 47, no. 2, pp. 416-429, 2011.
[5] H. Gernsback, "The Tesla Wireless Light," Electrical Experimenter, no. 1006, pp. 696-698, Feb
1919.
[6] G. L. Johnson, "Building the world's largest Tesla coil - History and theory," in Annual North
American Power Symposium (twenty-second), Auburn, 1990.
[7] N. Tesla, "System of Electric Lighting". United States Patent Office Patent 454,622, 23 Jun 1891.
[8] N. Tesla, "Apparatus for transmission of electrical energy". United States Patent Office Patent
649,621, 15 Mar 1900.
[9] N. Tesla, "The Transmission of Electrical Energy without Wires," Electrical World and Engineer,
05 Mar 1904.
[10] S. Kangliang and W. Zezhong, "Calculation of Telluric Current in Geomagnetic Induced Problems
Using the Finite Element Method," in Asia-Pacific Power and Energy Engineering Conference
(APPEEC), Wuhan, 2011.
[11] K. Schlegel and M. Füllekrug, "50 Jahre Schumann-Resonanzen," Physik in unserer Zeit, vol. 33,
no. 6, pp. 256-61, 2002.
[12] A. Marinic and V. Popović, Colorado Spring Notes 1899-1900, Beograd: Angriff Press, 1978.
[13] I. Hunt and W. W. Draper, Lightning in his hand: the life story of Nikola Tesla, Denver: Sage Books,
1964.
[14] P. Brenner, "Tesla against Marconi," in IEEE Eurocon Congress, Sanct Petersburg, 2009.
[15] L. Dr. Vujovic, "Tesla Biography: Nikola Tesla, the Genius who lit the World," Tesla Memorial
Society of New York, 10 Jul 1998. [Online]. Available: http://www.teslasociety.com/biography.htm.
[Accessed 12 Jun 2014].
P a g e 54 | 60
[16] N. Donaldson, "Comments on “Efficient Transdermal Links with Coupling-Insensitive Gain”," IEEE
Transactions on Biomedical Engineering, vol. 35, no. 4, pp. 280-81, 1988.
[17] C.-G. Kim, D.-H. Seo, J.-S. You, J.-H. Park and B.-H. Cho, "Design of a Contactless Battery Charger
for Cellular Phone," 15th Annual IEEE APEC (Applied Power Electronics Conference and Exposition),
vol. 2, pp. 769-773, 2000.
[18] A. Kurs, A. Karalis, R. Moffatt, J. D. Joannopoulos, P. Fisher and M. Soljačić, "Wireless Power
Transfer via Strongly Coupled Magnetic Resonances," Science, vol. 317, no. 5834, pp. 83-86, 2007.
[19] C. Femhdez, O. Garcia, R. Prieto, J. A. Cobos and J. Uc, "Overview of different alternatives for the
contact-less transmission of energy," 28th Annual Conference of IEEE IECON (Industrial Electronics
Society), vol. 2, pp. 1318-23, 2002.
[20] M. D. Rotaru, R. Tanzania, R. Ayoob, T. Y. Kheng and J. K. Sykulski, "Numerical and experimental
study of the effects of load and distance variation on wireless power transfer systems using magnetically
coupled resonators," in 9th IET International Conference on Computation in Electromagnetics (CEM),
London, 2014.
[21] F. C. Flack, E. D. James and D. M. Schalpp, "Mutual inductance of air-cored coils: Effect on design
of radio-frequency coupled implants," Medical & Biological Engineering & Computation, vol. 9, p. 79–
85, 1971.
[22] Pro-Power, "Farnell Data Sheet," 16 Jun 2012. [Online]. Available:

http://www.farnell.com/datasheets/1662623.pdf. [Accessed 10 Jul 2014].
[23] N. Tesla, "Coil for Electro Magnets". United States Patents Office Patent 0,512,340, 06 Jan 1894.
[24] U.-M. Jow and M. Ghovanloo, "Design and Optimization of Printed Spiral Coils for Efficient
Transcutaneous Inductive Power Transmission," IEEE Transactions on Biomedical Circuits and System,
vol. 1, no. 3, pp. 193-202, Sep 2007.
[25] M. Nagel, M. Hochlehnert and T. Leibfried, "Quality-factor considerations for single layer solenoid
reactors," in 15th International Symposium on High Voltage Engineering, Ljubljana, 2007.
[26] J. G. Webster, H. Eren and M. Szyper, "Chapter 30: Inductance Measurement," in Measurement,
Instrumentation, and Sensors Handbook: Electromagnetic, Optical, Radiation, Chemical, and Biomedical
Measurement, Boca Raton, CRC Press LLC, 2014, pp. 30-1 to 30-14.
[27] F. Alferink, "Measuring self-inductance & ESR," Meettechniek, 22 Jan 2014. [Online]. Available:
http://meettechniek.info/passive/inductance.html. [Accessed 15 Jul 2014].
[28] ARRL, The ARRL Handbook for Radio Communications, Newington: The American Radio Relay
League, Inc., 2013.
P a g e 55 | 60
[29] Tektronix, "Capacitance and Inductance Measurements Using an Oscilloscope and a Function
Generator," 2012. [Online]. Available: http://uk.tek.com/document/application-note/capacitance-and-
inductance-measurements-using-oscilloscope-and-function-ge. [Accessed 20 Jul 2014].
[30] IET Labs Inc., "The Measurement of Mutual Inductance," The General Radio Experimenter, pp. 1-
8, Jan 1937.
[31] T. C. Neugebauer and D. J. Perreault, "Parasitic Capacitance Cancellation in Filter Inductors," in

35th Annual IEEE Power Electronics Specialists Conference, Aachen, 2004.
[32] E. R. Williams and P. T. Olsen, "A Method to Measure Magnetic Fields Accurately Using Ampere's
Law," IEEE Transactions on Instrumentation and Measurement, vol. 27, no. 467-469, pp. 192-200, Feb
1978.
[33] C. H. Colwell, "Magnetic Field Along the Axis of a Current Loop," Physics LAB, 1997. [Online].
Available:
http://dev.physicslab.org/Document.aspx?doctype=3&filename=Magnetism_BiotSavartLaw2.xml.
[Accessed 17 Jun 2014].
[34] EPA, "Radiation Protection Basics," US Environmental Protection Agency , 06 Jul 2012. [Online].
Available: http://www.epa.gov/radiation/understand/protection_basics.html. [Accessed 15 Jul 2014].
[35] ICNIRP Guidelines, "Guidelines for limiting exposure to time-varying electric, magnetic, and
electromagnetic fields," Health Physics, vol. 74, no. 4, p. 494‐522, 1998.
[36] C. G. Jeans, R. J. Cruise and C. F. Landy, "Static and dynamic force characteristics of paramagnetic
shields in linear synchronous motors," in IEEE Africon, Cape Town, 1999.
[37] C. R. Nave, "Faraday's Law," Georgia State University, 2005. [Online]. Available:
http://hyperphysics.phy-astr.gsu.edu/hbase/electric/farlaw.html. [Accessed 10 Jun 2014].
[38] M. Yilai, L. Li, J. Kaiwen and Z. Xulin, "The Application of Magnetic Shielding Effect in Drill Pipe
Magnetic Leakage Flux Testing," in IEEE 5th International Conference on Measuring Technology and
Mechatronics Automation (ICMTMA), Hong Kong, 2013.
[39] V. Thomas, P.-Y. Joubert and E. Vourc'h, "A novel modeling of surface breaking defects for eddy
current quantitative imaging," in IEEE Sensors Applications Symposium (SAS), Limerick, 2010.
[40] B. Lenaerts and R. Puers, "Section 3.3 - Tertiary Circuits," in Omnidirectional Inductive Powering
for Biomedical Implants, Belgium, Springer Science , 2009, pp. 55-65.
[41] P. R. Clayton, "Section 1.6 - Kirchoff's Voltage Law," in Fundamentals of electric circuit analysis,
New York, John Wiley & Sons, 2001, pp. 18-27.
[42] D. K. Paul and P. Gardner, "Maximum Power Transfer Theorem: A Simplified," International
Journal of Electrical Engineering Education, vol. 35, pp. 271-273, 1998.
P a g e 56 | 60
[43] K. V. Rao, P. V. Nikitin and S. F. Lam, "Impedance matching concepts in RFID transponder design,"
in IEEE 4th Workshop on Automatic Identification Advanced Technologies, Buffalo, 2005.
[44] F. Bien and J. Kim, "Electric Field Coupling Technique of Wireless Power Transfer for Electric
Vehicles," in IEEE Tencon, Shaanxi, 2013.
[45] K. Siddabattula, "Why Not A Wire? The case for wireless power," Texas Instruments, Wireless
Consortium, [Online]. Available:
http://www.wirelesspowerconsortium.com/data/downloadables/1/2/1/1/why-not-a-wire-the-case-for-
wireless-power.pdf. [Accessed 09 Jun 2014].
[46] C. P. Steinmetz, Elementary Lectures on Electric Discharges, Waves, and Impulses, and Other
Transients, New York: McGraw-Hill Book Company, 1914.
[47] TDK Japan, "SEAT Tutorial - Appendix 1: S-parameter Basics," 2010. [Online]. Available:
http://product.tdk.com/en/technicalsupport/seat/tutorial.html. [Accessed 06 Aug 2014].
[48] D. M. Pozar, "The Transmission (ABCD) Matrix," in Microwave Engineering (4th Edition),
Amherst, John Wiley & Sons, Inc., 2012, pp. 188-194.
[49] B. Simmons, "Arc Length of a Curve," Mathwords, 28 Jul 2014. [Online]. Available:
http://www.mathwords.com/a/arc_length_of_a_curve.htm. [Accessed 21 Aug 2014].
[50] J. C. Whitaker, "Section 1.3.1.4 - Skin Effect," in AC Power Systems Handbook (3rd Ed), Boca
Raton, CRC Press, Taylor & Francis Group, 2007, p. 10.
[51] HSM Wire Inc., "AC DC Resistance Calculator," HSM Wire International, 2013. [Online].
Available: http://www.hsmwire.com/AC_DC_Resistance_Calculation.php. [Accessed 03 Sep 2014].
[52] T. Bluhm, "S-Parameters - Characteristics of Passive Components," BC Components, 20 Nov 2001.

[Online]. Available: http://www.ieee.li/pdf/essay/s_parameters_of_passive_components.pdf. [Accessed
08 Sep 2014].
[53] Y. Zhang, Z. Zhao and K. Chen, "Frequency splitting analysis of magnetically-coupled resonant
wireless power transfer," in IEEE Energy Conversion Congress and Exposition (ECCE), Denver, 2013.
[54] H. A. Haus and J. R. Melcher, "Electromagnetic Fields and Energy," in Chapter 3 - Introduction to
Electroquasistatics and Magnetoquasistatics, Englewood Cliffs, Prentice-Hall, 1989.
[55] U. S. Inan and A. S. Inan, "Electromagnetic Radiation and Elementary Antennas (Section 6)," in
Electromagnetic Waves, New Jersey, Prentice Hall, 2000, pp. 476-499.
P a g e 57 | 60
APPENDIX 1: LIST OF EQUIPMENT
Hardware


Arbitrary Function Generator: Tektronix AFG-2021

Oscilloscope: Tektronix MSO-3034 Mixed Phosphor
Bifiliar speaker Cu wire: Pro-Power audio cable 27/7/0.1 x 2, single conductor φ 1.5 mm, total outer φ

3.5 mm
PVC Pipe: Schedule 40, white, nominal size 2”, outer φ 2.375”


Plastic Sheet: Corotherm 10 mm clear, twin-wall
Capacitors: Assorted bead and ceramic types (non-electrolytic), of order of nF and μF
 Breadboard: AD-102 with φ 0.6 mm connecting wires of assorted colours


LCR Meter: Precision Gold WG-020

Electrical Toolkit: Wire stripper, pliers, blade (to scrape off wire enamel), solder station
LED: red, forward-voltage 2 V
Software


Quantitative Setup: Microsoft Excel v. 2013

EM Simulator: CST Suite Microwave Studio v. 2010

Circuit Analyser: Ansoft Simplorer v. 9

Curve-fitting Toolkit: MATLAB v. R2014a
Diagramming: Microsoft Visio v. 2007
Supporting Files of this Dissertation



Quantitative Spreadsheets formulated in Excel

Full-wave Models built in CST MW

Circuit Simulation analyzed in Simplorer

Images List drawn in Visio

Coil Structure pre-fabricated in 3D engine

MSO Screenshots downloaded from Tektronix oscilloscope

Key Findings constructed in PowerPoint
Demonstration Video filmed as MP4
P a g e 58 | 60
APPENDIX 2: MODELLING WORKSHEET STRUCTURE

Sample extract for reader familiarity
The MS Excel workbook that accompanies this project, contains a first worksheet with instructions; these must
be read before use to understand overall structure. A sample worksheet screenshot is shown below.
Scrolling through f values

causes a change in beady green
curve to demonstrate highest S21 at
resonant frequency
The formulation in various cells can be seen by clicking and viewing formula bar at top, or by pressing the <
Ctrl + ` > key combination to display all formulae simultaneously as demonstrated in extract below.
P a g e 59 | 60
APPENDIX 3: PROJECT MANAGEMENT

Listing of schedule and research milestones
Task ID Week # 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Phase I: Initialization
1: Background literature review
2: Software tutorials and setup of 3D and AC solvers
3: Writing-up of dissertation
Phase II: Laboaratory Work
4: Research on L-measurement techniques
5: Coil fabrication and spreadsheet framework setup
6: Baseline measurements and K- curves
7: Investigation of load-matching and S21
8: Experiments with aluminium shield
9: Consolidation and quantitative processing of lab results
Phase III: Mathematical Derivations
10: 2-port Z and S calculations for parallel, series LCR
11: 3-port Z calculations for shield scenario
12: Review of Maxwell's equations for MQS approach
Phase IV: Simulation Work
13: Data sweeps of full-wave model for changing variables
14: Running of AC solver for K- factor extraction
15: Comparison of results from 3 methods and conclusions
Milestones Deliverables
A: Demonstrations for examiners X
B: Production of first thesis draft X
C: Production of final thesis draft and supporting files X
D: Dissertation hand-in X
P a g e 60 | 60

Experimental Modelling of Wireless Power

Uploaded by

Copyright:

Available Formats

Experimental Modelling of Wireless Power

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Experimental Modelling of Wireless Power

Uploaded by

Copyright:

Available Formats

MUHAMMAD ALI QAISER for DR.

MIHAI ROTARU University of Southampton ECS

Experimental modelling and simulation of

Muhammad Ali Qaiser

Project Supervisor: Dr. Mihai Dragos Rotaru ……………………………..

Second Examiner: Dr. Adam Prugel-Bennett ……………………………...

A dissertation submitted in partial fulfilment of the degree of

© University of Southampton 2014, all rights reserved

Experimental modelling and simulation of

Muhammad Ali Qaiser

Project Supervisor: Dr. Mihai Dragos Rotaru ……………………………..

Second Examiner: Dr. Adam Prugel-Bennett ……………………………...

A dissertation submitted in partial fulfilment of the degree of

© University of Southampton 2014, all rights reserved

Chapter 2 Laboratory Setup ....................................................................................................... 11

2.4 Establishment of Coupling Factors............................................................................................ 18

Chapter 3 Discussion of Shielding ............................................................................................... 25

3.3 Mathematical Analysis of Conductive Shield ............................................................................ 29

Chapter 4 Discussion of Load Matching ...................................................................................... 33

4.3 Derivation of Circuit Impedance ............................................................................................... 35

Chapter 5 Microwave and Circuit Simulation Environment ......................................................... 45

5.2 Equivalent Circuit Design ......................................................................................................... 47

5.3 Impact of Shield ...................................................................................................................... 51

Chapter 6 Conclusion ................................................................................................................. 53

FIGURE 18: Coupling factor K for planar separation ___________________________________________________________ 22

Chapter 1 HISTORICAL PROOF OF CONCEPT

1.1 Pioneer Works of Nineteenth Century

Field induction in conventional transformers is a well-established

Tesla’s lifelong dream pursuit was to

1.2 Resurgence of Engineering Interest

1.3 Preceding Research Work

1.4 Content Overview

Chapter 2 LABORATORY SETUP

2.1 Bifiliar Pancake Coil Fabrication

Pressing cover to prevent

PVC pipe stub as spool

Circular base cut-out

FIGURE 5: Construction setup for pancake coil fabrication

Lower layer with 21 turns

Upper layer with 7 outer

FIGURE 6: Structure of bifiliar pancake coils and winding orientation of layers

2.2 Introduction to Quantitative Spreadsheets

TABLE 1: List of Worksheets for Quantitative Modelling 18

TABLE 2: Major input/output categories of spreadsheets19

2.3 Parametric Measurement of Fabricated Coils

2.3.1 Method A: Inductance Measurement by Voltage Split Rref

2.3.2 Method B: Inductance Measurement by Resonant Tank Rext

FIGURE 9: Left to right – V responses at 30 kHz, 35 kHz (resonant) and 40 kHz

2.3.3 Method C: Inductance Calculation by Coil Geometry

TABLE 3: Dimensions of pancake coil

2.3.4 Method D: Inductance Measurement by Impedance Rref

Now from the law of cosines for triangles, FIGURE 11 implies:

Finally, in rectangular coordinates we know independently that � = + � ; by comparing this against

2.3.5 Critique for Choice of Method and Results of Measurements

FIGURE 12: MSO screenshot for I-V method

TABLE 4: Method outcomes for different coil types

2.4 Establishment of Coupling Factors

FIGURE 13: In-phase polarity of coil windings

2.4.1 Axial Coupling between Identical Pancake Coils