PH.D.

QUALIFYING EXAM-II (DEPTH EXAM), MARCH 2009 1

A Report on Wireless Power Transfer via Strongly
Coupled Magnetic Resonances
Saroj Rout, ECE Department, Tufts University, Medford, MA-02155
AbstractThis report is for satisfying the requirement of
Ph.D. Qualifying Exam-II (Depth Exam), 2009. It pertains
to the work by Andre Kurs, et al. [1], on wireless power
transfer using strongly coupled magnetic resonances. This
work uses self-resonant coils in a strongly coupled regime.
They claim to transfer 60W of non-radiative power with
~40% effciency over a distance up to 8 times the radius of
the coils which is beyond the regime of near eld induction.
This report aims at understanding and validating the
claims in the paper and also providing a well reasoned
theory on why this work is pioneering.
Index TermsWireless energy, Wireless power, Strong
coupling
I. INTRODUCTION
S
INCE the early 20th century, starting with the work
of Nicola Tesla [2], there has been a push to trans-
port power wirelessly, preferably over a long distance.
However, there has been essentially two fundamental
modes of transport, each of which pose their own
limitations for any meaningful wireless power transport
circuit. One of the modes being the near eld induction
which is limited to very near distance transport which
further limited by large elds and bulky equipments.
The other mode being radiative transfer, which can be
transported over a long distance but with large loss
if omnidirectional and for unidirectional mode, it is
limited by the requirement of uninterrupted line-of-sight,
very large antenna gain and some sophisticated control
mechanism.
The work by Andre Kurs, et al. [1], have attempted
at breaking the limitation of these two fundamental
modes of transportation. They use self-resonant coils in a
strongly coupled regime to transfer power over distances
which is beyond the near eld regime and efciencies
upto 40% which is very difcult to achieve of radiative
transfer.
Before we analyze and understand the authors
1
work,
lets try to understand the presently available mecha-
nisms and their limitation to transfer power wirelessly.
E-mail: [email protected]
1
From now on authors refer to the authors of [1]
Figure 1. Basic principle of magnetic induction
A. Near-Field Inductive Power Transfer
Figure-1 shows a basic setup for magnetic induction.
The transmit coil is excited with an AC source (V
t
=
V
0
sin(t)) which results in a current I
t
of the same
angular frequency. Lets assume that the frequency is
low enough such that Biot-Savart law can be used to
determine the magnetic ux at the receiving coil which
is given by [11],
B(r) =

0
4
_
I
t
r
r
2
dl

(1)
On the receiving coil, the general law for the electric
eld associated with the changing magnetic eld is given
by Faradays Law, rst written in the differential form
by Maxwell as
E =
B
t
(2)
Using Stokes theorem, this law can be represented in
its integral form as
_

E ds =
_
S
(E) n da =
_
S
B
t
n da (3)
where, as usual, is any closed curve and S is any
surface bounded by it. In this example, the closed curve
is the wire loop and S is the normal surface enclosed
2 PH.D. QUALIFYING EXAM-II (DEPTH EXAM), MARCH 2009
by the wire loop. Then the time derivative can be taken
outside the integral and we have
_

E ds =

t
_
S
B n da (4)
For the receiving coil the rst loop integral is voltage
induced in the coil and the second term is the rate change
of ux through the coil. Then it can be written as
V
r
=

t
(flux through coil) (5)
This is the basic principle behind transferring power
wirelessly between the coils. The major drawback of
this technique can seen in Eq-(1) where B 1/r
2
and
therefore the magnetic eld strength drops off rapidly
with distance. To counter the limitation, usual imple-
mentation are with large coil sizes and many turns to
increase the eld strength at the source. But this results
in poor efciency and high eld strengths near the source
coil which may pose health hazard. Also note, in most
applications, each of the coils are made resonant to the
transmitting frequency to optimize power transfer. Usual
distance of transmission is in the order of the radius of
the coil.
Despite this limitation, this method of power transfer
is used in various applications such as wireless power in
consumer electronics gadgets [3], robotics [4], industrial
motors [5], very widely used in biomedical implants
[6][7], and various other similar application.
But this still is not a viable solution for high power
transfer over a longer distance.
B. Far-Field Radiative Power Transfer
Coulombs law for electric eld and Biot-Savarts law
for magnetic eld are for stationary charges or currents
and both of them are proportional to 1/r
2
or higher.
But for accelerating charges or time-varying current
there is a radiative term generally referred to as the
electromagnetic wave. This is energy is separated from
the source and can travel innitely long in vacuum or
loss-less media.
If you imagine a gigantic spherical shell, out at radius
r , and a source at the center, the total power passing
out through this surface is the integral of the Poynting
vector(S) [11]:
P(r) =
_
S da =
1

o
_
(EB) da (6)
The power radiated is the limit of this quantity as
r :
Figure 2. Dipole radiator
P
rad

lim
r
P(r) (7)
This power that is transported out to innity, never
comes back. Now, the area of the sphere is 4r
2
, so
for radiation to occur the Poynting vector must decrease
(at large r) no faster than 1/r
2
or else the term inside
the integrand in Eq.(6) will vanish. Therefore, the eld
from Coulombs law and Biot-Savarts law are not going
to contribute to radiation. So static sources do not
radiate. But if we look at the time-dependent Jemenkos
equations [11]:
E(r, t) =
1
4
o
_
_
(r

, t
r
)
r
2
r +
(r

, t
r
)
cr
r

J(r

, t
r
)
c
2
r
_
d

(8)
and for the magnetic eld
B(r, t) =

o
4
_
_

J(r

, t
r
)
r
2
+

J(r

, t
r
)
cr
_
rd

(9)
Without going in to the detail of the equations, it
can be seen that both in E(r, t) and B(r, t), the terms
involving and

J are proportional to 1/r which are the
radiative components.
Now consider a dipole as shown in Figure-2. Lets
assume the two charges on the dipole are moving along
z-axis such that
q(t) = q
o
cos(t) (10)
The result is an oscillating electric dipole:
p(t) = p
o
cos(t) (11)
where p
o
q
o
d is the maximum value of the dipole
moment. On taking the retarded potential of the dipole
and solving for E(r, t) and B(r, t), the time-averaged
Poynting vector can be shown to be [11]:
SAROJ ROUT, ECE DEPT., TUFTS UNIVERSITY, MEDFORD, MA 3
Figure 3. Radiated pattern for a dipole. Cross-section along x-y
plane
S =
_

o
p
2
o

4
32
2
c
_
sin
2

r
2
r (12)
As expected the energy decays proportional to 1/r
2
which means when we integrate the power around in-
nite sphere surrounding the dipole, we will get a nite
power which is the radiated quantity.
Now if you plot the energy function in Eq-(12), its
a donut shape with the z-axis going through the middle
of the donut ie. its got azimuthal symmetry. A cross-
section along the x-y plane will look like as shown in
Figure-3.
So, although the energy is radiated out from a dipole,
its spread in a volume almost consisting of the whole
3-D space. If you have an antenna to receive power,
you lose energy rapidly as you go farther away from
the source. In order to successfully transport power over
a long distance, you need to create radiators which have
enormous directionality or in other words very high
gains which are practically unrealizable. Also, once you
achieve very high gain in order to direct the power to
the source only, you need some sophisticated tracking
mechanism to keep the transmitter and receiver aligned.
Furthermore, for lower frequencies, typically where you
want to transfer high power, the antenna sizes become
prohibitively large.
In spite of the limitations, there are applications such
as embedded sensors in automotive, where the power is
transferred in a radiative fashion [8].
C. Coupled Resonators
In this work [1], the authors use the coupled-mode
theory (CMT) [9], to break away from the fundamental
limitation posed by the last two methods of power
transfer wirelessly. Although CMT was developed to un-
derstand mode coupling in optics, the theory is generally
applicable to any physical system which exhibit similar
L
1
C
1
C
2
L
2
C
c
+
-
+
-
V
1
V
2
Figure 4. Two LC resonators coupled by Cc capacitor
resonant coupling. The authors have ingeniously used
the theory to apply it to power transfer in magnetic
resonant coupled system.
A typical resonant coupled system is shown in Figure-
4. The two resonators when isolated have their respective
resonant frequencies as
1
= 1/

L
1
C
1
and
2
=
1/

L
2
C
2
. Each of the isolated resonators are governed
by the following differential equation:
a
1
(t) = (i
1

1
)a
1
(t)
a
2
(t) = (i
2

2
)a
2
(t) (13)
where the state variable a
m
(t) V
m
e
jmt
. In the next
section, this basic system will be analyzed to show power
transfer in a regime that does not fall into either of the
ones explained previously.
II. MODELING OF COUPLED RESONATORS
A. Analysis of Split Resonant Frequencies in Coupled
Resonators
Starting with with Eq. (1) in [1], we can write two sets
of differential equation for resonator 1 and 2 respectively
as
a
1
(t) = (i
1

1
)a
1
(t) + i
21
a
2
(t)
a
2
(t) = (i
2

2
)a
2
(t) + i
12
a
1
(t) (14)
where the indices denote the two different resonators.
The variables a
m
(t) are dened so that the energy con-
tained in object m is |a
m
(t)|
2
,
m
is resonant frequency
of that isolated resonator, and m is its intrinsic energy
loss (e.g., due to absorption, ohmic loss as heat or
radiated losses). And each resonator is governed by the
following phasor
a
m(t)
= exp(i
m
t
m
t) (15)
and
12
=
21
= (say) are the coupling coefcients
between the resonators.
4 PH.D. QUALIFYING EXAM-II (DEPTH EXAM), MARCH 2009
If the resonators are identical ie.
1
=
2
=
0
and

1
=
2
=
0
then the differential equation in (14) can
be written as
a
1
(t) = (i
0

0
)a
1
(t) + ia
2
(t)
a
2
(t) = (i
0

0
)a
2
(t) + ia
1
(t) (16)
which can be expresses in matrix form as
_
a
1
a
2
_
=
_
i
0

0
i
i i
0

0
_ _
a
1
a
2
_
(17)
and note a
1
(t), a
2
(t) have been replaced with a
1
, a
2
for brevity. In order to nd the solution of Eq-(17) we
can write in vector format as
a = A a (18)
where boldface variables refer to vectors. Let us now
guess the solution to be of the form
a = x e
t
(19)
a = xe
t
= Axe
t
(20)
Ax = x (21)
The solution to Eq-(21) are the eigenvalues of A ie.
det[AI] = 0. On solving for the eigenvalues we get

1
= i(
0
+ )
0
(22)

2
= i(
0
)
0
As seen from the eigenvalues, when two identical
resonators are coupled their frequencies split by the
factor
= 2 (23)
and the decay factor remains the same as that of the
isolated ones ie.
0
.
B. Power Efciency
Based on the phasor denition in Eq-(15), we can
write the energy of the system as:
E = |a(t)|
2
= e
2t
|e
it
|
2
(24)
Now, the power can be derived by taking the time
derivative of energy as:
P(t) =
d
dt
|a(t)|
2
= 2 |a(t)|
2
(25)
L R
Figure 5. Series RL network
Denoting the source, receiver and load with subscript
S,D and W respectively. we can write the power ef-
ciency as
=
P
load
P
total
=

W
|a
D
|
2

S
|a
S
|
2
+ (
D
+
W
)|a
D
|
2
(26)
Using the relation |a
S
|
2

2
= (
D
+
W
)
2
|a
D
|
2
and
doing some algebraic manipulation, we get the Eq-(2) in
[1] ie.
=
W
D

2
SD
__
1 +
W
D
_

2
SD
_
+
_
1 +
W
D
_
2
(27)
The efciency is maximized when
W
/
D
= [1 +
_

2
/
S

D
]. Resonance plays an essential role in power
transfer mechanism, as the efciency is improved ap-
proximately by
2
/
2
D
relative to the case of inductively
coupled non-resonant objects.
C. Q of the Inductor
A real inductor always has some loss associated with
it due to nite conductance of the wire. The loss can be
represented as a series resistance as shown in Figure-5
For a system under sinusoidal excitation, perhaps the
most fundamental denition of Q ([10]) is
Q
energy stored
average power dissipated
(28)
If we excite the series RL circuit by a sinusoidal
current source (I = I
o
sin(t)), using the denition in
Eq-(28), it can be shown that the Q of the inductor is
Q =
L
R
(29)
where is the excitation frequency, L is the self-
inductance and R is the resistance of the wire. Typically
the resistance of the wire is calculated assuming all the
current ows uniformly in the cross-section of the wire.
But for low-frequency AC current, it mostly ows on the
surface (skin) of the conductor as shown in Figure-6
How deep into the surface does the current ow,
known as the skin depth(), depends on the material
properties and the the excitation frequency. It can be
derived using the Maxwells equation in its differential
SAROJ ROUT, ECE DEPT., TUFTS UNIVERSITY, MEDFORD, MA 5
Figure 6. Current owing on the surface (skin) of the conductor
form ([11]) or in its integral form ([10]) and it can be
shown to have the following expression:
=

(30)
where and are the permeability and conductivity
of the wire respectively. Using the physical nature of the
problem as shown in Figure-6, we can write the effective
resistance (R
e
) as
R
e
=
l
A

l
2r
(31)
where r is the radius of the wire. Substituting Eq-(30)
in Eq-(31) and then using the expanded expression of
R
e
in Eq-(29) we get the Q of the inductor as
Q = L
_
2r
l
_

(32)
As seen from this expression, for a given frequency Q
is proportional to

.
The authors claim that their measured Q 950 which
is lower than the calculated Q 2500 is due to copper-
oxide formation on the surface of the inductor.
If copper-oxide is a conductor with much lower con-
ductivity (), then according to Eq-(32) the Q will be
lower. But on the other hand, if copper-oxide behaves
more like dielectric, then it should not have much effect
on the Q, since none of the surface current will ow in
the dielectric. This needs to be veried.
Also, the assumption of the uniform skin depth around
the conductor is not always true [10]. It depends on the
uniformity of the magnetic eld around the conductor.
If its not uniform, then you will also see a lower Q.
III. RESULTS
As seen in Figure-7, the experimental setup has a
source A, which is a single copper loop of radius
25cm that is part of the driving circuit, which outputs
a sine wave with frequency 9.9 MHz. S and D are
respectively the source and the device coils. B is a loop
of wire attached to the load (light bulb). The various
s
Figure 7. Schematic of the experimental setup.
Figure 8. Comparison of experimental and theoretical values for the
parameter /as a function of the wireless power transfer distance.
The Theory values were obtained by using the theoretical and
experimentally measured . The shaded area represents the spread in
the theoretical / due to 5% uncertainty in Q.
represent direct couplings between the objects indicated
by the arrows. The angle between coil D and the loop
A is adjusted to ensure that their direct coupling is
zero. Coils S and D are aligned co-axially. The direct
couplings between B and A and between S and B are
negligible. The parameters for the two identical helical
coils built for the experimental validation are h = 20 cm,
a = 3 mm, r = 30 cm and n = 5.25. Both coils are
made of copper.
The coupling coefcient was found experimentally
by placing the two self resonant coils a distance D apart
and measuring the split in the resonant frequency as
predicted in Eq-(23) (ie. = 2). All the results in
this work has been done with the coils aligned co-axially.
A. Measurement of the Efciency
The maximum theoretical efciency depends only on
the parameter /

L
S
L
D
= / which is greater than
1 even for D = 2.4m (8 times the radius of the coils,
clearly outside the near eld regime) as can be seen in
Figure-8.
The driving circuit is a standard Colpitts oscillator
whose inductive element consists of a single loop of
6 PH.D. QUALIFYING EXAM-II (DEPTH EXAM), MARCH 2009
Figure 9. Comparison of experimental and the theoretical efcien-
cies as functions of the wireless power transfer distance. The shaded
area represents the theoretical prediction for maximum efciency and
is obtained by inserting the theoretical values from Fig.8 into Eq.(27)
with W/D =
_
1 + (
2
/
2
). the black squares are the maximum
efciency obtained from Eq.(27) and the experimental values of /
from Fig.8 . The red dots present the directly measured efciency, as
described in the text.
copper wire 25 cm in radius (Figure-7) which inductively
couples to the source coil. The load is a calibrated light
bulb and is attached to its own loop which is inductively
coupled to its device coil. The driver and load being
inductively coupled to their respective coil is matter of
convenience. In principle, they can be integrated to the
coils itself. One of the convenience: the parameter
W
/
can be adjusted by varying the distance between the light
bulb and the device so that it matches its optimal value.
The efciency of the power transfer is calculated by
measuring the current at the midpoint of each coil I
S
and
I
D
. Then the power dissipated in each coil is calculated
from P
S,D
= L|I
S,D
|
2
and obtain the efciency from
= P
W
/ (P
S
+ P
D
+ P
W
). A comparison of measured
and theoretical values of efciency is shown in Figure-9.
Note, the efciency reported does not take into account
all the losses from the outlet to the load(bulb). Some of
the losses being that in the Colpitt oscillator, coupling
between the coil and devices, etc.
The authors measured the wall-to-load efciency of
15% at a distance of 2m where the reported number in
Figure-9 is 40 50%.
B. IEEE Safety Standard Limit
The authors work are operated at roughly 10 MHz .
Using the IEEE standard ([12] Table-9), the maximum
permissible exposure (MPE) at 10 MHz for general
public is shown in Table-I
Field Type MPE
Erms 82 V/m
Hrms 1.6 A/m
Table I
PERMISSIBLE ELECTRIC(Erms) AND MAGNETIC(Hrms) FIELD BY
IEEE SAFETY STANDARD FOR GENERAL PUBLIC AT 10 MHZ
SINUSOIDAL FREQUENCY.
The reported elds at half the distance between the
coil (1m) and near the coil (20cm) is shown in Table-II
Field Type Midway(1m) Near(20cm)
Erms 210 V/m 1400 V/m
Hrms 1 A/m 8 A/m
Table II
MEASURED STRENGTH OF ELECTRIC AND MAGNETIC FIELD AT
HALF-WAY DISTANCE (1M) AND NEAR THE COIL(20CM)
As seen from the result, they violate the electric eld
strength by almost 1-2 orders of magnitude. They pass
the magnetic eld strength requirement when midway
between the coils but not near it.
The authors propose to use a capacitively loaded
single-turn loop design [13], which has the advantage
of conning nearly all of the electric elds between the
plates of the capacitors and thus reduce it outside of the
apparatus. This needs to be veried.
IV. CONCLUSION
As seen from Figure-9, the authors have demonstrated
that they can transmit power wirelessly at an efciency
of 40 50% at a distance of 2m which is about 8 times
the radius of the coil. This distance is clearly beyond
the regime of the near-eld induction as described in
the introduction of this document and is not a radiative
source. Therefore, the authors have demonstrated a novel
method of efciently transferring power over a wireless
link.
Since the publication of this work, although there
has been little or no work directly associated with the
authors work, there has been work associated with the
same principle to transfer power in other domains. L.
Bian, et al. [14] have shown strong power coupling be-
tween resonance-type magneto-electric (ME) transducer
with high quality factor. Mark J. Bloemer, et al. [15] and
M. Tsang, et al. [16] have shown the use of materials
in so called superlens which are resonant with incident
wave, it enhances the evanescent component of waves.
Y. Kim and H. Ling [17] show the coupled-mode theory
improves power transfer efciency in resonant radiative
antennas. T. Aoki, et al. ([18]) show strong coupling
SAROJ ROUT, ECE DEPT., TUFTS UNIVERSITY, MEDFORD, MA 7
between individual caesium atoms and the elds of a
high-quality toroidal microresonator.
As seen from Figure-9, the efciency falls of rapidly
after a distance of 2.5m. Although this is clearly beyond
the near-eld regime, it needs more research to make it
work for longer distance to make it a viable high power
transport mechanism.
The authors state in their concluding remarks about the
high sensitivity of efciency to the detuning f/f
o
of
any one of the resonators. This will pose a big challenge
in implementation of high power transport systems.
Other electrical systems, not associated with the one
for power transfer, with their resonant frequency same
as that of the power transfer system may end up trans-
ferring power to the undesired device creating hazardous
situations such as in pacemakers, etc.
As shown in Section-III-B, the present implementation
of the system exceeds the IEEE safety standard for gen-
eral public exposure to electromagnetic eld exposure.
REFERENCES
[1] Andre Kurs, Aristeidis Karalis, Robert Moffatt, J. D.
Joannopoulos, Peter Fisher, Wireless Power Transfer via
Strongly Coupled Magnetic Resonances, Science, Vol-317,
p:83-86, July-2007.
[2] N. Tesla, U.S. Patent 1,119,732 (1914)
[3] J. M. Fernandez, et al., U.S. patent 6,184,651 (2001)
[4] A. Esser, et al., A New Approach to Power Supplies for
Robots, IEEE Tran. Ind. Appl. Vol-27, p-872 (1991)
[5] J. Hirai, et al., Integral Motor with Driver and Wireless Trans-
mission of Power and Information for Autonomous Subspindle
Drive, IEEE Trans. Power Electron. vol-15, 21 (2000)
[6] T. Akin, et al., A wireless implantable multichannel digital
neural recording system for a micromachined sieve electrode,
IEEE J. Solid-State Circuits, vol-33, no. 1, pp. 109-119, Jan
1998
[7] W. J. Heetderks, RF power of milimeter and submilimeter-
sized neural prosthetic implants, IEEE Trans. Biomed. Eng.,
vol. 35, no. 5, pp. 323-327, May 1988.
[8] T. A. Vanderelli, et al. , U.S. patent 7,027,311 (2006)
[9] H. Haus, Waves and Fields in Optoelectronics, Prentice-Hall,
Englewood Cliffs, NJ, 1984
[10] T. H. Lee, The Design of CMOS Radio-Frequency Integrated
Circuits, Cambridge, 2nd Ed.,2004
[11] David J. Grifths, Introduction to Electrodynamics, Pearson
(3rd Ed.), 1999
[12] IEEE Std C95.1-2005 IEEE Standard for Safety Levels with
Respect to Human Exposure to Radio Frequency
[13] Aristeidis Karalis, et al., Efcient wireless non-radiative mid-
range energy transfer, Annals of Physics (323), pp:34-48
(2008)
[14] Leixiang Bian, et al., Magnetoelectric transducer with high
quality factor for wireless power receiving, Sensors and Actu-
ators A, vol-150, p.207-211 (2009)
[15] Mark J. Bloemer, et al.,Energy considerations for a superlens
based on metal/dielectric multilayers, Optics Express, Vol.16,
No.23, p.19342 (2008)
[16] Mankei Tsang, et al., Theory of resonantly enhanced near-eld
imaging, Optics Express, Vol.15, No.19, p.11960 (2007)
[17] Y. Kim and H. Ling, Investigation of coupled mode behaviour
of electrically small meander antennas, Electronics Letters,
Vol.43, No.23 (Nov 2007)
[18] T. Aoki, et al., Observation of strong coupling between one
atom and a monolithic microresonator, Nature, vol-443, p.671,
(Oct. 2006)