See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/326284372

Design Guidelines to Avoid Bifurcation in a Series-Series Compensated

Inductive Power Transfer System

Article  in  IEEE Transactions on Industrial Electronics · July 2018

DOI: 10.1109/TIE.2018.2851953

CITATIONS READS
51 657

2 authors, including:

Kunwar Aditya
Vitesco Technologies
21 PUBLICATIONS   338 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Kunwar Aditya on 02 January 2019.

The user has requested enhancement of the downloaded file.

 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 66, NO. 5, MAY 2019 3973

Design Guidelines to Avoid Bifurcation in a

Series–Series Compensated Inductive
Power Transfer System
Kunwar Aditya , Member, IEEE, and Sheldon S. Williamson , Senior Member, IEEE

Abstract—Resonant inductive power transfer (RIPT) is

gaining in popularity for wireless charging applications of
future electric transportation. A fundamental impediment to
the efficient operation of an RIPT system is the existence
of bifurcation phenomenon in a doubly tuned circuit. Bifur-
cation can be avoided by either selecting complicated con-
trol strategies or by proper design strategies. Many control Fig. 1. Equivalent circuit of an SS-RIPT system.
strategies have been presented in the literature for avoid-
ing bifurcation. However, systematic design strategies for
avoiding bifurcation are still unknown. This paper presents
a simplified, and easy to follow set of design guidelines for
filter to mitigate the grid-side harmonics, followed by an ac–
series–series compensated RIPT (SS-RIPT) systems. The dc power factor correction (PFC) preregulator to draw power
design guidelines avoid bifurcation by calculating the pa- at unity power factor and generate regulated dc-link voltage
rameters for a given load profile in a systematic and easy to for the dc–dc converter stage. Only difference is that the dc–dc
follow approach. Following the design guidelines, a 3.6-kW converter stage in an RIPT-based charger is formed by loosely
setup has been fabricated as an example and verified using
a finite element analysis as well as experimental testing.
coupled inductors bound on separate cores separated by a large
Selection of a resonant frequency and output voltage for airgap instead of tightly coupled inductors bound on the same
3.6 kW was influenced by the availability of testing equip- core in a plug-in charger. While presence of an airgap allows
ment available in the lab. However, the design guidelines the vehicle-side component to move freely with respect to the
are generalized, and by following it, an SS-RIPT system of way-side component, it is also responsible for a large leakage
any frequency and output voltage rating, applicable for EV
charging, can be designed.
flux and hence poor efficiency of the system.
To compensate the effects of poor coupling, due to large leak-
Index Terms—Battery chargers, energy storage, electric age flux, both way-side and vehicle-side coils are compensated
vehicles, inductive power transfer, bifurcation. using capacitors. Depending upon the connection of capacitors
I. INTRODUCTION numerous resonant topologies are possible with each topology
having its own merits and demerits. Most common of these
RESONANT inductive power transfer (RIPT) based
A charger provides a safer charging method for electric ve-
hicles (EVs) as compared to traditional plug-in charger. In ad-
topologies are series–series (SS), series–parallel (SP), parallel–
series (PS), and parallel–parallel (PP) resonant topology [3].
Advantages and disadvantages of each topology have been cov-
dition to safety, range extension, opportunity charging, battery ered in numerous publications and will be not discussed here.
volume reduction, and reduced maintenance requirements are This paper will essentially focus on a series–series RIPT (SS-
some of the other perks of having power transferred to the vehi- RIPT) system shown in Fig. 1.
cle wirelessly [1], [2]. An RIPT-based EV charger has the same In Fig. 1, symbols have their usual meaning with subscript “P”
basic layout as a traditional plug-in charger. Input is connected and “S” signifying primary-side (way-side) and secondary-side
to the grid supply through an electromagnetic interference (vehicle-side) components, respectively. As seen from Fig. 2,
SS-RIPT system is a double-tuned circuit due to the presence
Manuscript received July 14, 2017; revised October 25, 2017, January of two resonant networks, one on each side of the primary and
13, 2018, and May 7, 2018; accepted June 11, 2018. Date of publication secondary coils. In such a double tuned circuit, although system
July 9, 2018; date of current version December 28, 2018. This work
was supported by NSERC, Canada. (Corresponding author: Sheldon S. is tuned to resonant frequency ωo , which is also the zero-phase
Williamson.) angle (ZPA) frequency, it is very common that this resonant fre-
K. Aditya is with the BorgWarner Waterloo, Inc., Kitchener, ON N2G quency is split into multiple resonant frequencies due to sudden
4X8, Canada (e-mail: [email protected]).
S. S. Williamson is with the Electrical Engineering Department, Uni- change in some parameters. This phenomenon where more than
versity of Ontario Institute of Technology, Oshawa, ON L1H 7K4, Canada one ZPA frequency or resonant frequencies exist is known as
(e-mail: [email protected]). bifurcation in the RIPT system and the parameter which causes
Color versions of one or more of the figures in this paper are available
online at http://ieeexplore.ieee.org. this phenomenon to occur is known as the critical parameter. In
Digital Object Identifier 10.1109/TIE.2018.2851953 SS-RIPT system, bifurcation is onset by increase in coupling
0278-0046 © 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications standards/publications/rights/index.html for more information.
3974 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 66, NO. 5, MAY 2019

Equations (1)–(5) have been derived in [4]–[7]. From (1)–(5),

the appearance of bifurcation depends upon the coupling and
load resistance.
The frequency of interest in the RIPT system usually lies
between 20–200 kHz, therefore MOSFET switches become the
obvious choice for primary side inverter in the RIPT system.
As turn-on losses dominate in MOSFET switches, zero volt-
age switching (ZVS) is preferred to minimize switching losses.
To ensure ZVS, it is a requisite that effective impedance seen
by inverter should be inductive irrespective of change in load
and coupling. However, due to a sudden change in load and/or
coupling, load might change from inductive to capacitive, this
causes the inverter to lose ZVS. Loss of ZVS results in high
current stress, even shoot-though at turn-on transition because
of poor body diode reverse recovery of opposite switch.
To regulate the output voltage and current, a dc–dc converter is
introduced between rectifier and load. However, for stationary
charging of EV, it is beneficial to avoid the additional dc–dc
converter stage and instead use primary side control as it leads to
reduce component count and hence reduce losses in the system.
Fig. 2. Effect of increasing coupling in an SS-RIPT system for ideal- All the primary side control strategies for an RIPT system can
tuning and constant input voltage V P and constant load R L . (a) Voltage- be broadly classified into three types.
gain versus frequency plot. (b) Input phase-angle versus frequency plot. 1) Fixed frequency control: Fixed-frequency controller is
tuned to switching frequency, which is slightly deviated from
coefficient above a critical value kc . For more clarity on the the resonant frequency fo to accommodate soft-switching [8].
subject, consider Fig. 2. This is the simplest of all control strategies. However, due to the
From Fig. 2(a) and (b), one can observe that as k increases onset of bifurcation by a sudden change in load and/or coupling,
above kc , voltage gain decreases and resonant frequency fo the input impedance can become capacitive instead of inductive
splits into three frequencies (fo , fH , and fL ). Here, ideal- and vice versa; therefore, soft-switching will be lost as shown
tuning means primary and secondary are tuned to same res- in Fig. 2(a). Additionally, reduction of voltage gain due to the
onant frequency. Values of fL and fH are given by (1) and (2), bifurcation [see Fig. 2(b)] requires a higher current to deliver
respectively the same amount of power to the load. This results in increased
copper losses in the coils and conduction losses in the switches.
⎡   −2 2   ⎤0.5 2) Variable Frequency Control: In this type of control scheme,
2 − Q−2 − Q − 2 − 4 1 − k 2
fL = ⎣ ⎦ fo
S S the duty cycle of the pulses driving the inverter switches is kept
2 (1 − k 2 ) constant at 50% and the switching frequency is varied to regu-
(1) late the output power. To guarantee the soft-switching operation,
operating frequency is selected higher than all the three ZPA fre-
⎡   −2 2   ⎤0.5 quencies in the inductive region. Such a large deviation from the
2 − Q−2 + QS − 2 − 4 1 − k 2
=⎣ ⎦ fo .
S
fH true resonant frequency will cause a large circulating current in
2 (1 − k 2 ) the resonant tank and therefore will cause a drop in the over-
(2) all efficiency of the system due to large conduction losses in
switches as well as copper losses in the coils.
In (1) and (2), fo is the true resonant frequency and QS is 3) Dual Control or Self-Sustained Oscillation Control:
the secondary quality factor of SS-RIPT link and is given by (3) Switching frequency is not externally imposed like the variable
and (4), respectively frequency control, rather, it is free to oscillate at a frequency
1 1 dependent on both the RIPT parameter and phase shift neces-
fo = √ =√ (3) sary for soft switching. For this purpose, a phase-locked-loop
LP CP LS CS
controller is used to track the ZPA angle between primary volt-
(2πfo ) LS 1 age and current [9]. However, as derived earlier, an SS-RIPT
QS = = . (4)
RS + RL 2πfo CS (RS + RL ) system can have three ZPA frequencies. Therefore, due to bifur-
cation, it may uncontrollably switch between the three available
Value of critical coupling kc beyond which bifurcation occurs ZPA frequencies. This result in a loss of frequency stability and
is given by (4) controllability [5].

As above-mentioned control strategies could not avoid bi-
1 1
kc = 1− . (5) furcation issue, therefore perplexed control methods have been
QS 4Q2s
proposed in the literature. In [10] and [11], instead of tracking
ADITYA AND WILLIAMSON: AVOID BIFURCATION IN A SERIES–SERIES COMPENSATED INDUCTIVE POWER TRANSFER SYSTEM 3975

The magnitude of vP , i.e., VP is equal to the mains voltage.

This assumption is valid since there is usually a PFC stage
between mains supply and full bridge inverter. This PFC stage
can adjust the dc-link voltage, Vdc , of full bridge inverter to give
VP equal to the mains voltage. Since VP has been assumed equal
to the grid voltage VG , the RMS value of primary current can
Fig. 3. Wireless charger employing the SS-RIPT link.
be given by
Pin
the ZPA between primary voltage and current, the ZPA between IP rm s = . (8)
VP rm s
primary voltage and secondary current has been tracked. The
ZPA frequency between primary voltage and secondary current Once the value of the primary and the secondary currents
depends upon the instantaneous parameters of the RIPT system, are known, the value of the mutual inductance for the desired
but is unique for a set of these parameters, thus avoiding any amount of output power can be derived by applying the KVL
instability issues caused by bifurcation. This control requires equation on the secondary side of the SS-RIPT link at resonance
complex sensor design and shielding to eliminate electromag- ωo = √ L 1 C
s S

netic interference. In [12], Chen et al. defined the low-frequency

|jωo M iP | = RL |iS | . (9)
limit in the controller to ensure that system operates in the
high-frequency region inductive region. Since RIPT system is The secondary inductance is calculated from the secondary
designed around true resonant frequency, therefore, a large devi- quality factor QS . The value of QS is usually selected between
ation from the resonant frequency increases the voltage stress on 2 to 10 as a higher value can make the system difficult to tune,
compensation capacitor and lowers the overall efficiency [13]. and a lower value will generate harmonics in the current and
In this paper, authors have formulated a design methodology, the voltage waveform [15]. Equation (10) gives the value of the
following which the RIPT system can be designed to operate required secondary inductance
without bifurcation for the entire load and coupling variations.
QS RL
Formulated design steps calculate the electrical parameters of LS = . (10)
SS-RPT system considering the critical value of coupling co- ωo
efficient. Since the system is designed to operate without bi- The value of the coupling-coefficient k can be obtained using
furcation, therefore a complicated control strategy is no longer the equation of the critical coupling derived in Section II in this
required to avoid bifurcation. Following the proposed design paper. To avoid the bifurcation, value of k should be selected to
steps, a 3.6-kW SS-RIPT based charger has been fabricated in be less than kc , that is
the lab. For the control of the charger, well-known fixed fre-
quency control tuned for ZVS has been used. Charger operated 1 1
k< 1− . (11)
with ZVS for entire load and coupling variations. QS 4Q2S

Equation (11) will decide the minimum airgap between the

II. CALCULATION OF ELECTRICAL PARAMETERS FOR
primary and the secondary coils. It also tells that there is a
BIFURCATION FREE OPERATION
maximum value of the coupling, exceeding which will let the
Fig. 3 shows the block diagram and components of an SS- system into the bifurcating mode. Once the value of k is decided,
RIPT based wireless charger. The aim is to calculate the param- the primary inductance can be calculated using
eters of the SS-RIPT link (shown in the red dotted box) for a
M2
given load (battery pack). LP = . (12)
The primary side of the SS-IPT link is fed from the voltage- LS k 2
sourced full-bridge inverter, which converts the dc voltage Vdc Following the above design steps, parameters for a 3.6-kW
into high-frequency ac voltage vP . On the secondary side, a output were calculated. A majority of commercially available
rectifier with a capacitive filter Cf converts the high-frequency battery voltages range between 300–600 V [16]. The battery
voltage into dc voltage Vo and dc current Io required by the modules for electric transportation and mobility systems are
load, Ro . Then, ac equivalent resistance seen by secondary is formed by connecting 4.2-V Li-Ion cells in appropriate series
given by [14] and parallel connections. For example, in the Nissan Leaf, each
module consists of four cells in a 2-series, 2-parallel formation;
8 Vo2
RL = . (6) 48 such modules are connected in series to form a 403.2 V
π 2 Po battery pack [17]. However, due to the rating of dc electronic
The secondary voltage vs is a square wave due to the capaci- load available in the lab, output voltage of 168 V (4.2 × 40 cells)
tive output filter. The rms value of the fundamental component was considered. Moreover, the resonant frequency was restricted
of secondary current iS can be given by to 40 kHz, due to the limitation of DSP module and switching
elements available in the lab. The value of secondary quality
√ Vo
IS rm s = 2 2 . (7) factor QS is 4.0. A value of 4.0 for QS gives critical coupling
πRL kc equal to 0.248. To eliminate bifurcation, nominal coupling k
3976 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 66, NO. 5, MAY 2019

TABLE I
IDEAL ELECTRICAL AND CIRCUIT PARAMETERS OF 3.6-KW CHARGER

Fig. 5. Physical representation of Archimedean spiral. (a) 3-D view. (b)

Cross-sectional view.

Therefore, for this paper, circular charging pad was considered

for the static charging application.
In [21], Wheeler presented several formulas for planar spiral
inductors, which were intended for the discrete inductors. Fig. 5
shows the representation of an Archimedean spiral. From Fig. 5,
it is obvious that Dout is the outer diameter; Din is inner diam-
eter; T is the spacing between turns; and w is the diameter of
the wire used for making the coil. Equation (12) gives the orig-
inal Wheeler expression for calculating the inductance of spiral
coils [21]
a2 N 2
L= . (13)
8a + 11c

With the help of Fig. 5, c and a can be given by (14) and (15),
Fig. 4. Efficiency versus k plot for SS-RIPT link. respectively
dout − din
c= (14)
is assumed to be 0.2 for bifurcation free operation. Table I gives 2
the calculated (ideal) parameters for 3.6-kW SS-RIPT link. dout − c dout + din
It should be noted that although design guidelines limit the a= = . (15)
2 4
coupling to a maximum value of 0.248, it has negligible impact
on the efficiency of the system. To make this point, the efficiency Putting (14) and (15) in (13), a modified and more usable
of SS-RIPT link versus k has been plotted in Fig. 4, for fixed version of the Wheeler formula can be derived as
primary voltage. For coupling variation of 0.1 to 1, efficiency N 2 (Dout + Din )2
varies from 98.6% to 99.7%. Besides this, targeting an airgap L= . (16)
8 (15Dout − 7Din ) 2.54
of 15 ∼ 20 cm, the coupling of 0.2 is very typical for an RIPT
system. This is evident from [2], [18], and [19] in which different Using arithmetic progression, relation between Dout and Din
chargers have been presented to operate efficiently for coupling can be given by
values of 0.15 ∼ 0.25. This is an important difference between
Dout = Din + 2w + (T + w) (2N − 1) . (17)
a loosely coupled system such as the SS-RIPT system and a
closely coupled system such as power transformers in which it Using (16) and (17), all the geometric characteristics of circu-
is desirable to have a coupling as high as possible. lar coils can be derived from the calculated value of inductance.
It has been established in the literature that the fundamental
III. FABRICATION OF CIRCULAR CHARGING PAD height of the flux path in an Archimedean spiral is about 1/4 of
the coil diameter [22]. Therefore, it gives a rough idea about the
A. Calculation of Geometric Parameters of Primary and
how large the primary coil should be for desired airgap. Target-
Secondary Coil ing a nominal airgap of 10 ∼ 16 cm “Dout ” was selected to be
In the literature, different types of planar charging pads have 47 cm. Interturn spacing “T” was kept the minimum possible,
been proposed such as circular pads, double-D pad, double-D since it has been established in the literature that an increase
quadrature pads, bipolar pads, and H-shaped pads [3]. Among in turn spacing decreases mutual inductance [22]. Since T, w
these, circular charging pads are nonpolarized pads and have the (4.6 mm), and Dout are fixed, the inner diameter “Din ” and
same tolerance to misalignment in all directions and are there- number of turns “N” are adjusted using (16) and (17) to achieve
fore easier to operate. They are nondirectional, i.e., a vehicle can inductance value as close as possible to the ideal value cal-
approach them from any direction, which ensures flexibility of culated in Table I. For the coils, 8 AWG 5 × 5 × 42/32 with
use by the driver. Due to aforesaid advantages, the circular ge- single Nylon served was ordered from the New England Wire
ometry is widely used for static EV charging applications [20]. Technologies. Fig. 6 shows the assembled coil.
ADITYA AND WILLIAMSON: AVOID BIFURCATION IN A SERIES–SERIES COMPENSATED INDUCTIVE POWER TRANSFER SYSTEM 3977

Fig. 6. Fabricated coils. (a) Primary coil. (b) Secondary coil.

TABLE II
ACTUAL PARAMETER OF FABRICATED COILS

Fig. 8. (a) 3-D (1/2) model of final charging pad created in JMAG.
(b) Dimensions of final charging pad created.

Fig. 7. Reference figure for designing the shielding.

The coils were fabricated manually and therefore some dis-

crepancies in geometric parameters crept in during the fabrica-
tion process. Therefore, there is some difference between the
actual parameters of the assembled coils and ideal parameter Fig. 9. Photograph of charging pad fabricated in lab. (a) Primary coil
with ferrite spoke. (b) Secondary with ferrite spoke.
values. Table II gives the actual geometric parameters and self-
inductance values achieved.
TABLE III
The Keysight E4980AL precision LCR meter was used to ELECTRICAL PARAMETERS OF CHARGING PADS AT 16 CM AIRGAP
achieve mutual inductance M of value 40.14 μH by manually
adjusting the airgap. An airgap of 16 cm was found to give an
M value close to 40.14 μH.

B. Design of Shielding
It is known that adding an aluminum sheet with ferrite spokes
to the back of an Archimedean spiral is very effective in min-
imizing the leakage flux. However, it was found that due to
the addition of shielding, the original values of the inductances
was used. Fig. 8 shows the final charging pad, modeled in JMAG.
of the coils get changed. To keep the original value of self-
Fig. 9 shows the picture of the primary pad and secondary
inductance for the primary and secondary coils, several itera-
pad. Table III gives the measured electrical parameters of the
tions were performed in JMAG Designer 14.1 [23]. In iterations,
primary and secondary pads with aluminum shield at 16 cm of
two variables were adjusted: angle θ between two ferrite spokes
separation.
and inner diameter Din of the coils as shown in Fig. 7. The outer
diameter, as well as turn spacing, were not varied.
IV. PIECEWISE-LINEAR MODEL OF LI-ION BATTERY PACK
Ferrites spokes were modeled based on I core, of material
PC95 and dimension 55.9 × 19.3 × 6.6 mm, available from the A Li-Ion battery pack used in EVs is required for testing the
MH&W International Corporation. An aluminum sheet of 1 mm designed charging pad. However, the actual battery pack was
3978 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 66, NO. 5, MAY 2019

TABLE IV
CIRCUIT PARAMETERS FOR SIMULATION AND EXPERIMENTAL RESULTS

Fig. 10. Piecewise-linear model of Li-Ion battery charging profile.

not available in the lab. Therefore, dc electronic load was used

to emulate battery charging profile. For a Li-Ion battery, the
charging process usually consists of two main stages: constant
current (CC) charging and constant voltage (CV) charging stage.
The SS-RIPT system was designed for the output voltage of
168 V. Therefore, a charging voltage of 168 V was considered
for the CV mode. A battery pack of capacity 6.1 kWh was
considered. For 19 A of current in CC mode, the charge rate
(C rate) of the battery pack can be calculated by
168 × 19
C= = 0.523. (18)
6100
Usually charging is terminated when the current drops to 2%
of the rated current [24]. Based on these information, the charg-
ing parameters of the designed battery pack can be calculated
by extrapolating the values at different stages on the charging Fig. 11. Experimental setup.
profile of the single Li-Ion cell. The piecewise-linear charging
profile of 6.1 kWh battery pack obtained is plotted as shown in
Fig. 10.
Chroma System Solutions was used as the load. XR Series
V. SIMULATION AND EXPERIMENTAL RESULTS power supplies from Magna-Power Electronics were used as
the programmable dc power supply. The inverter was built using
To avoid hard-switching of the inverter, the experimental full-bridge MOSFET power module (APTM120H29FG) from
setup was tuned for ZVS. For this purpose, secondary coil was Microsemi. For the rectifier, low loss fast recovery diode module
tuned to switching frequency, which is greater than resonant fre- from IXYS (DSEI2X101-06A) was used. For the primary and
quency and the primary coil was tuned to the resonant frequency; secondary compensations, capacitors with very low ESR Metal-
this will make the equivalent impedance of SS-IPT system to be lized Polypropylene film type from Cornell Dubilier Electronics
inductive and ZVS can be achieved. Since the primary and sec- have been selected. Fig. 11 shows a picture of the experimental
ondary coils are now tuned to different frequencies, this type of setup. Various simulation and experimental cases are discussed
tuning will be referred to as “ZVS-tuning” in this paper. Due to below to verify the design guidelines.
the practical availability of capacitors, the switching frequency
was selected to be 41.426 kHz and the primary was tuned to
A. Verification of Calculations for Electrical Parameters
the resonant frequency of 39.031 kHz. The minimum value of
Vdc required for 3.6-kW output power, at the nominal load resis- To verify the calculated parameters of the 3.6 kW design, the
tance of 7.84 Ω and ZVS tuning, was calculated to be 337.13 V. hardware was tested in open loop for the nominal load resistance
Therefore, a round figure of 340 V was used. Table IV shows of 7.84 Ω and airgap of 16 cm. Results obtained are shown in
the final parameters of the SS-RIPT system for simulation and Fig. 12.
experimental measurements. In the experiment, an output current of 21 A, and output
For the experimental results, the closed-loop voltage con- voltage of 169 V was obtained. Peak value of secondary cur-
troller was implemented in real-time using dSPACE Con- rent is 34 A (24 A RMS) and peak value of primary current
trolDesk 5.3. LEM sensor LV20-P was used to sense the is 20 A (14 A RMS). Results obtained are very close to the
output voltage and LA 55-P was used to sense the cur- analytical calculations. This, in turn, validates the designed
rent. A 63800-series programmable ac/dc electronic load from parameters.
ADITYA AND WILLIAMSON: AVOID BIFURCATION IN A SERIES–SERIES COMPENSATED INDUCTIVE POWER TRANSFER SYSTEM 3979

Fig. 12. Experimental result for V dc = 340 V, R o = 7.84 Ω, and airgap

= 16 cm.

Fig. 14. Effect of shielding on leakage field. (a) Flux density in coils
without shielding for 3.6 kW, 16 cm airgap, and 3.6-kW output. (b) Flux
density in coils with shielding 3.6 kW, 16 cm airgap, and 3.6-kW output.

Moreover, one can observe that the phase angle is positive for
all the frequencies equal and above the resonant frequency of
primary coil 39.031 kHz. Therefore, ZVS tuning gives wider
inductive region as compared to “ideal tuning.” The existence
Fig. 13. Input phase angle versus frequency for entire load variation. of inductive region on either side of switching frequency ensures
(a) Ideal tuning. (b) ZVS tuning. ZVS operation even if the system parameters deviates over time
due to thermal and ageing effects.
B. Verification of Bifurcation Free Design in Fabricated
C. Effect of Shielding on Leakage Field
Charging Pads and Effect of ZVS Tuning
To show the impact of shielding on leakage field, FEA result
Input phase angle of experimental setup was plotted against
for coreless pad (see Fig. 6) and shielded pad (see Fig. 9) were
frequency, for “ideal tuning” as well as “ZVS tuning” and has
simulated in JMAG Designer for 40 kHz, 16 cm airgap, and
been shown in Fig. 13(a) and (b), respectively. Fig. 13 was
3.6 kW output. Results obtained for flux density distributions in
plotted for minimum and the maximum value of load resistance,
both cases are shown in Fig. 14.
obtained from Fig. 10, for different value of k.
Fig. 15 shows the direction of the instantaneous current in
In Fig. 14, the shaded portion represents the inductive region,
primary coil and the eddy current induced in the aluminum
i.e., ZVS region. From Fig. 13(a), one can observe that for
shield when the supply phase is 45° and 135°. One can observe
“ideal tuning” condition, the input phase angle versus frequency
that the eddy current direction in aluminum sheet is opposite
plot crosses zero degrees line only once for the entire charging
to that of the secondary current for both the 45° phase, i.e.,
profile if coupling coefficient is below the critical value kc .
when the primary current is rising and 135° phase, i.e., when
Due to misalignments, the coupling can only decrease, therefore
the primary current is falling. Moreover, in Fig 14(b), it can
bifurcation free operation is ensured. There is a slight possibility
be observed that maximum flux density is 77.4 mT, which is
of an increase in coupling due to the decrease in tire pressure
significantly lower than the saturation flux density (380 mT ∼
of EVs, however, the margin between kc = 0.248 and k = 0.2
530 mT) of PC95 core material [25].
will ensure that the phase angle curve maintains the single zero-
phase crossing.
D. Safety Regulations
As compared to the “ideal tuned,” the “ZVS tuned” system
will maintain ZVS in inverter switches for all values of the A car is usually made up of steel, which is an excellent shield-
coupling coefficient even greater than critical coupling value. ing material [26], [27]. Therefore, people sitting inside the car
3980 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 66, NO. 5, MAY 2019

Fig. 17. Efficiency versus load plot for different value of misalignment
and fixed airgap of 16 cm.

Fig. 18. Switching scheme and voltage waveform of ACM control.

Fig. 15. Current density vector plot in charging pad when supply phase
7.84 Ω resistance and 16 cm of airgap. Waveform for rated
is 45° and 135°. load condition is shown in Fig. 12. In addition, dc–dc efficiency
for entire charging profile was calculated by implementing a
closed-loop controller for CC and CV mode of battery charger.
Controller design has been presented in [30] by Aditya and
Williamson and due to the length of the paper, it will be not
be discussed here. Fig. 17 shows the efficiency of the designed
charger for the entire charging period calculated for 0, 5, and
10 cm of misalignment, with a constant airgap of 16 cm.
From Fig. 17, one can observe that the efficiency of designed
Fig. 16. Position of the leakage flux measurement point (P) demon- charger remains above 85% for entire charging profile in perfect
strated with the sketch of the top view of an RIPT-powered EV.
alignment condition as well as 5 cm of misalignment. Efficiency
at 10 cm of misalignment falls to 84% at the end of charging
are shielded from the leakage flux. However, people standing profile. The efficiency figure does not include the front-end
next to the car are at the risk of exposure to leakage magnetic power factor and correction stage, but with recent publications
field. Therefore, as shown in Fig. 16, if the width of vehicle is [31], it can be shown that the efficiency of this stage can reach
1.6 m and secondary pad is installed such that it is centered un- as high as 98%. Factoring this component, the grid to battery
der the vehicle body, the user will be exposed to the flux density efficiency of 89% is expected from the designed charger at rated
levels occurring at and beyond the vehicle chassis, i.e., at point load condition.
P [28]. It should be noted that the charger was designed for output
From Fig. 14(b), one can see that magnitude of flux density voltage of 168 V. Therefore, efficiency is expected to improve
at coordinates (X = 800 mm, Y = 0.04 mm, Z = 81.9 mm) is when design guidelines is applied for 400 V output as less
8.14 μT. This value is within the safety guidelines recommended current in secondary will be required to supply the 3.6 kW of
by ICNIRP 2010 and IEEE Std. C95.1-2005 [26], [29]. output power.

E. Efficiency Analysis F. Verification of ZVS in Inverter Switches

DC–dc efficiency of the designed charger was calculated to For the voltage and current regulation asymmetrical clamped
be 91.2% for the rated load condition, i.e., 168 V output at mode (ACM) control (see Fig. 18) operating at fixed switching
ADITYA AND WILLIAMSON: AVOID BIFURCATION IN A SERIES–SERIES COMPENSATED INDUCTIVE POWER TRANSFER SYSTEM 3981

Fig. 20. Verification of ZVS in an inverter for V o = 168 V and R o =

Fig. 19. Verification of ZVS in inverter switches for charging under per- 7.84 Ω at different misalignment values. (a) Misalignment = 5 cm.
fect aligned condition. (a) R o = 6.315 Ω, Io = 19 A. (b) R o = 8.842 Ω, (b) Misalignment = 10 cm. (c) Misalignment = 15 cm.
Io = 19 A. (c) R o = 23.13 Ω, V o = 168 V.

output. For this purpose, the reference voltage was kept 168 V
frequency has been used [5], [32]. In Fig. 18, VA B is inverter at 7.84 Ω and alignment between the primary and the secondary
output. coils was varied at a fixed airgap of 16 cm. The results obtained
The polarity of current at four switching instances for 5, 10, and 15 cm of misalignment are shown in Fig. 20. In
(t0 , t1 , t2 , t3 ) can be observed to confirm ZVS in switches. Fig. 20, waveform satisfies the condition for achieving ZVS for
These instances are shown in Fig. 18 and the polarity of the cur- all the three cases.
rent at these instances are defined by (19) for achieving ZVS [33]
iP (t0 ) < 0 For S1 VI. CONCLUSION AND FUTURE WORK
iP (t1 ) > 0 For S3 This paper presents a unique, simplified, and easy to follow
design guidelines for the SS-RIPT system. The design guide-
iP (t2 ) > 0 For S2 lines avoid the bifurcation by calculating the parameters for a
iP (t3 ) < 0 For S4. (19) given load profile in a systematic and easy to follow approach.
A step-by-step procedure for calculating circuit parameters of
Since it is not possible to verify ZVS for each load condition, the SS-RIPT system, fabrication of coil, and design of shielding
only the following three load conditions were considered. has been presented. Following the design guidelines, a 3.6 kW
1) Beginning of the CC mode: In this mode, load resistance setup has been fabricated as an example and verified using finite
is 6.315 Ω and output power is 2.28 kW, which is 63.3% element analysis as well as experimental testing.
of rated power of 3.6 kW. This paper essentially focuses on providing a proof of con-
2) Beginning of the CV mode (or end of the CC mode): In cept rather than an actual product. Therefore, the hardware was
this case, load resistance is 8.842 Ω and output power is designed for the frequency and voltage level depending on the
3.192 kW (88.6% of rated power). equipment ratings (load, power supply, switches), available in
3) End of the CV mode: In this case, load resistance is the lab at the time. However, following the design guidelines, a
23.13 Ω and the output power is 1.22 kW (33.8% of rated bifurcation free charger could be designed for any given fre-
power). quency and output voltage level applicable to EV charging
The experimental results for the above three load conditions applications.
were plotted in MATLAB to determine the polarity of current
and are shown in Fig. 19.
REFERENCES
In Fig. 19, one can observe that the polarity of the current for
each load profile fulfils the condition given by (19). Therefore, [1] T. Fujita, T. Yasuda, and H. Akagi, “A dynamic wireless power transfer
system applicable to a stationary system,” IEEE Trans. Ind. Appl., vol. 53,
it can be said that the designed charger will operate under ZVS no. 4, pp. 3748–3757, Jul./Aug. 2017.
condition for the entire charging profile. The results shown in [2] J. M. Miller and A. Daga, “Elements of wireless power transfer essential
Fig. 19 were obtained at perfect alignment and fixed airgap of to high power charging of heavy duty vehicles,” IEEE Trans. Transport.
Electrific., vol. 1, no. 1, pp. 26–39, Jun. 2015.
16 cm, i.e., mutual inductance was constant. [3] D. Patil, M. Mcdonough, J. Miller, B. Fahimi, and P. T. Balsara, “Wireless
It is also necessary to verify the ZVS operation at different power transfer for vehicular applications: Overview and challenges,” IEEE
coupling coefficients during the charging process for 3.6-kW Trans. Transport. Electrific., vol. 4, no. 1, pp. 3–37, Mar. 2018.
3982 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 66, NO. 5, MAY 2019

[4] K. Aditya, “Design and implementation of an inductive power transfer [29] IEEE Standard for Safety Levels With Respect to Human Exposure to Radio
system or wireless charging of future electric transportation,” Ph.D. dis- Frequency Electromagnetic Fields, 3 kHz to 300 GHz, IEEE C95.1-2005,
sertation, Univ. Ontario Inst. Technol., Oshawa, ON, Canada, 2016. Apr. 2006.
[5] C. S. Wang, G. A. Covic, and O. H. Stielau, “Power transfer capability and [30] K. Aditya and S. Williamson, “Linearization and control of series–series
bifurcation phenomena of loosely coupled inductive power transfer sys- compensated inductive power transfer system based on extended describ-
tems,” IEEE Trans. Ind. Electron., vol. 51, no. 1, pp. 148–157, Feb. 2004. ing function concept,” Energies, vol. 9, vol. 9, no. 11, pp. 962–978,
[6] W. Q. Niu, J. X. Chu, W. Gu, and A. D. Shen, “Exact analysis of frequency Nov. 2016.
splitting phenomena of contactless power transfer systems,” IEEE Trans. [31] F. Musavi, W. Eberle, and W. G. Dunford, “A high-performance single-
Circuits Syst. I: Reg. Papers, vol. 60, no. 6, pp. 1670–1677, Jun. 2013. phase bridgeless interleaved PFC converter for plug-in hybrid elec-
[7] M. Košı́k, R. Fajtl, and J. Lettl, “Analysis of bifurcation in two-coil induc- tric vehicle battery chargers,” IEEE Trans. Ind. Appl., vol. 47, no. 4,
tive power transfer,” in Proc. IEEE 18th Workshop Control Model. Power pp. 1833–1843, Jul./Aug. 2011.
Electron., Stanford, CA, USA, 2017, pp. 1–8. [32] J. M. Burdı́o, L. A. Barragán, F. Monterde, D. Navarro, and J. Acero,
[8] S. Bai, Z. Pantic, and S. Lukic, “A comparison study of control strategies “Asymmetrical voltage-cancellation control for full-bridge series resonant
for ZVS resonant converters,” in Proc. Annu. Conf. IEEE Ind. Electron. inverters,” IEEE Trans. Power Electron., vol. 19, no. 2, pp. 461–469,
Soc., Glendale, AZ, USA, 2010, pp. 256–262. Mar. 2004.
[9] R. Bosshard, J. W. Kolar, and B. Wunsch, “Control method for Induc- [33] L. A. Barragán, J. M. Burdı́o, J. I. Artigas, D. Navarro, J. Acero, and
tive Power Transfer with high partialload efficiency and resonance track- D. Puyal, “Efficiency optimization in ZVS series resonant inverters with
ing,” in Proc. IEEE Int. Power Electron. Conf., Hiroshima, Japan, 2014, asymmetrical voltage-cancellation control,” IEEE Trans. Power Electron.,
pp. 2167–2174. vol. 20, no. 5, pp. 1036–1044, Sep. 2005.
[10] E. Gati, G. Kampitsis, and S. Manias, “Variable frequency controller
for inductive power transfer in dynamic conditions,” IEEE Trans. Power
Electron., vol. 32, no. 2, pp. 1684–1696, Feb. 2017.
[11] C. Zheng, “Loosely coupled transformer and tuning network design for
high-efficiency inductive power transfer systems,” Ph.D. dissertation, Vir-
ginia Polytechnic Inst. State Univ., Blacksburg, VA, USA, 2015.
[12] R. Chen et al., “Analysis and parameters optimization of a contactless Kunwar Aditya (S’14–M’17) received the B.E.
IPT system for EV charger,” in Proc. IEEE Appl. Power Electron. Conf. degree from Bharati Vidyapeeth University,
Expo., Fort Worth, TX, USA, 2014, pp. 1654–1661. Pune, India; the M.Tech. degree from Indian
[13] L. Xu, Q. Chen, X. Ren, S. C. Wong, and C. K. Tse, “Self-oscillating Institute of Technology (Banaras Hindu Univer-
resonant converter with contactless power transfer and integrated cur- sity), Varanasi, India; and the Ph.D. degree from
rent sensing transformer,” IEEE Trans. Power Electron., vol. 32, no. 6, the University of Ontario Institute of Technology,
pp. 4839–4851, Jun. 2017. Oshawa, ON, Canada, in 2009, 2012, and 2016,
[14] V. Vorperian, “Simplified analysis of PWM converters using model of respectively, all in electrical engineering.
PWM switch. Continuous conduction mode,” IEEE Trans. Aerosp. Elec- From July 2017 to November 2017, he was
tron. Syst., vol. 26, no. 3, pp. 490–496, May 1990. a Research Fellow with Rolls-Royce@NTU Cor-
[15] G. Covic and J. Boys, “Modern trends in inductive power transfer for porate Lab, Nanyang Technological University,
transportation applications,” IEEE J. Emerg. Sel. Topics Power Electron., Singapore. Since December 2017, he has been working as a Hard-
vol. 1, no. 1, pp. 28–41, Mar. 2013. ware Engineer with BorgWarner Waterloo, Inc., Kitchener, ON, Canada.
[16] P. Miller, “Automotive lithium-ion batteries,” Johnson Matthey Technol. His research interests include power electronics and control applied to
Rev., vol. 59, no. 1, pp. 4–13, 2015. wireless charging, electrical energy storage systems, and transportation
[17] Cell, Module, and Pack for EV application. [Online]. Available: electrification.
http://www.eco-aesc-lb.com/en/product/liion_ev/. Dr. Aditya was awarded the IIT (BHU) Gold Medal, for securing high-
[18] M. Budhia, J. T. Boys, G. A. Covic, and C. Y. Huang, “Development of a est academic standing among all the electrical engineering students of
single-sided flux magnetic coupler for electric vehicle IPT charging sys- 2012 cohort. He was also awarded the FEAS Outstanding Thesis Award
tems,” IEEE Trans. Ind. Electron., vol. 60, no. 1, pp. 318–328, Jan. 2013. for his PhD work at UOIT, Oshawa, ON, Canada.
[19] H. H. Wu, A. Gilchrist, K. D. Sealy, and D. Bronson, “A high efficiency
5 kW inductive charger for EVs using dual side control,” IEEE Trans. Ind.
Informat., vol. 8, no. 3, pp. 585–595, Aug. 2012.
[20] B. Esteban, M. Ahmed, and N. C. Kar, “A comparative study of power
supply architectures in wireless EV charging systems,” IEEE Trans. Power
Electron., vol. 30, no. 11, pp. 6408–6422, Nov. 2015. Sheldon S. Williamson (S’01–M’06–M’13)
[21] H. A. Wheeler, “Simple inductance formulas for radio coils,” Proc. Inst. received the Bachelor’s of Engineering (B.E.)
Radio Eng., vol. 16, no. 10, pp. 1398–1400, Oct. 1928. degree in electrical engineering with high dis-
[22] J. M. Miller, C. P. White, O. C. Onar, and P. M. Ryan, “Grid side regulation tinction from the University of Mumbai, Mumbai,
of wireless power charging of plug-in electric vehicles,” in Proc. IEEE India, in 1999, the Master’s of Science (M.S.)
Energy Convers. Congr. Expo., Raleigh, NC, USA, 2012, pp. 261–268. degree and the Doctor of Philosophy (Ph.D.) de-
[23] JSOL, JMAG-Designer-Introducing JMAG-Designer Ver.14.1: JMAG, gree (with Hons.) both from the Illinois Institute
2017. of Technology, Chicago, IL, USA, in 2002 and
[24] F. Marra, G. Y. Yang, C. Træholt, E. Larsen, C. N. Rasmussen, and S. You, 2006, both in electrical engineering and special-
“Demand profile study of battery electric vehicle under different charging izing in automotive power electronics and motor
options,” in Proc. IEEE Power Energy Soc. General Meet., San Diego, drives, at the Grainger Power Electronics and
CA, USA, 2012, pp. 1–7. Motor Drives Laboratory.
[25] TDK Ferrites Datasheet. [Online]. Available: https://product.tdk.com/ From June 2006 to May 2011, he held a Tenure-track Assistant Profes-
info/en/catalog/datasheets/ferrite_mn-zn_material_characteristics_en.pdf. sor position with the Department of Electrical and Computer Engineering,
[26] S. Li and C. C. Mi, “Wireless power transfer for electric vehicle applica- Concordia University, Montreal, Canada. He also held a tenured Asso-
tions,” IEEE J. Emerg. Sel. Topics Power Electron., vol. 3, no. 1, pp. 4–17, ciate Professor position with Concordia University from June 2011 to
Mar. 2015. June 2014. He is currently a Professor with the Smart Transportation
[27] J. H. O. C. M. Ibrahim, L. Pichon, L. Bernard, and A. Razek, “Advanced Electrification and Energy Research (STEER) Group, within the Depart-
modeling of a 2-kW series–series resonating inductive charger for real ment of Electrical, Computer, and Software Engineering, University of
electric vehicle,” IEEE Trans. Veh. Technol., vol. 64, no. 2, pp. 421–430, Ontario-Institute of Technology (UOIT), Oshawa, ON, Canada. He also
Feb. 2015. holds the prestigious NSERC Canada Research Chair position in Elec-
[28] A. Zaheer, H. Hao, G. A. Covic, and D. Kacprzak, “Investigation of tric Energy Storage Systems for Transportation Electrification. His main
multiple decoupled coil primary pad topologies in lumped IPT systems research interests include advanced power electronics and motor drives
for interoperable electric vehicle charging,” IEEE Trans. Power Electron., for transportation electrification, electric energy storage systems, and
vol. 30, no. 4, pp. 1937–1955, Apr. 2015. electric propulsion.

View publication stats