A High-Frequency Resonant Converter Based on the

Class Φ2 Inverter for Wireless Power Transfer

Jungwon Choi, Wei Liang, Luke Raymond, Juan Rivas

Department of Electrical Engineering
Stanford University
Stanford, California, 94305
Email: [email protected]

Abstract—This paper presents the design and implementation In this paper, a push-pull class Φ2 inverter for inductive
of a high frequency resonant converter based on the Class Φ2 wireless power transfer at 13.56 MHz is designed and evalu-
inverter for inductive power transfer. MHz frequency operation ated. Section II provides details regarding the configuration
can allow for higher power density than conventional switching of a single inverter, the transmitter and receiver coils, and
frequencies. The converter is based on the Class Φ2 inverter, push-pull version of the inverter. Section III presents details
reducing the voltage stress across the switch compared to other
regarding the simulation and experimental results of a single
resonant topologies. In order to increase power while reducing
input current ripple, a push-pull Class Φ2 inverter was imple- inverter and the push-pull inverter with coils. Section IV
mented to drive resonant coils for wireless power transfer (WPT). concludes the paper.
Specially, we demonstrated 219 W, 160 V push-pull inverter at
13.56 MHz. This paper provides experimental and simulated II. C IRCUIT CONFIGURATION
results of the WPT system investigated.
A. Class Φ2 inverter
I. I NTRODUCTION The Class E inverter [8] of Fig. 1 is a well known resonant
topology capable of operating at frequencies reaching into the
Wireless power transfer (WPT) has attracted recent atten- 10s of MHz. This topology provides zero voltage switching
tion as a power deliver method for numerous mobile, medical (ZVS) and has a single ground-referenced active switch (Q1 ).
applications, and electric vehicle charging systems [1]–[3]. While the class E topology has many merits as an rf inverter,
More specifically, WPT through inductive resonant coupling it achieves ZVS but imposes a large voltage stress across the
has been analyzed and demonstrated as a method of delivering active switch. The peak switch voltage can reach in excess of
power to a load with relatively high efficiency [4], [5]. Various 3.6 times the input voltage. Moreover, this circuit is normally
frequency ranges can be used for WPT. However, this paper designed with a relatively large input inductance (Lchoke in
implements a 13.56 MHz frequency WPT system in order to Figure 1), but designs with smaller inductance values are
study the trade-offs in design, size and feasibility of WPT possible [9]. This input “choke” tends to slow the transient
at high frequencies. Operating at frequencies above 10 MHz response of the system.
allows for a significant reduction in the values of passive
energy storage components [6]. This reduction allows for all
magnetic components in our design to be implemented without
Lchoke LS CS
using ferro-magnetic cores.
Q1 + +
The WPT system presented herein consists of class Φ2 VIN +
− vds (t) RL vload (t)
+
power amplifier driving a tuned set of transmitting and receiv- vgs (t) CP -
ing coils. We use resonant power amplifier because it enables -
-
operation at switching frequencies above 10 Mhz. Moreover, a
Φ2 inverter allows for a higher input voltages for a given switch Fig. 1: Class E inverter. Peak drain voltage reaches ≥ 3.6 × VIN . This
topology is widely used in rf and high frequency applications.
rating. The Φ2 inverter operates at reduced device voltage
stresses compared to other commonly used single-switch high
frequency resonant topologies. A topology that overcomes many of the shortfalls of the
class E is the class Φ2 inverter of [10]. This circuit topology
The design presented herein is a push-pull version of the is shown in Fig. 2. It has an additional circuit branch formed
Φ2 inverter [7] which allows for an increase in the output by LM R and CM R in parallel with the active switch. These
power delivered to the load while exhibiting other promising components are tuned to provide a low impedance at the 2nd
attributes such as lower input current ripple. Each side of the of the switching frequency. Following the procedure outlined
push-pull inverter operates from a single input voltage source in [10], the drain impedance under biased conditions is tuned
and shares a common ground, which reduces the complexity to produce a waveform with a substantial reduction in peak
of the gate drive. A push-pull implementation can potentially voltages compared with the class E of Fig. 1. The resulting
lead to consolidation of passive components which promises trapezoidal voltage waveform at the drain has a peak of
increased power density over a single inverter design. approximately twice the input voltage. This allows the use

U.S. Government work not protected by U.S. copyright

of a MOSFET with lower voltage rating and hence better n:1
conduction, and switching characteristics for a given input Ll
Cl
CM LM
voltage. Moreover, as described in [10], the interaction of LF
and CP with the remaining resonant components of the circuit
and the load make the impedance at fundamental and third
harmonics to attain the values that make the MOSFET voltage Fig. 3: Model of transmitting and receiving coil. LM and LR presents
approximate a trapezoid. It is important to notice that in the magnetizing inductance and leakage inductance. n is the the effective turns
Φ2 of Fig. 2, the inductance LF place an active role in shaping ratio between the two coils. Ll and n is changed according to the variation of
distance between coils. CM and Cl are selected to be resonant at 13.56 MHz.
the drain voltage, and has a significantly smaller value than in
most class E designs. Smaller input inductance value results in
a faster transient response but at the expense of a significantly
larger input current ripple. Evidently, the larger input current
ripple of a Φ2 design is not an issue when two converters are
operated in push-pull due to the current interleaving effect.

LF LS CS
LMR +
Q1 +
VIN +
− vds (t) RL vload (t)
+
vgs (t) CP -
-
CMR -
Fig. 4: Variation of leakage inductance LM and effective ratio n with coil
Fig. 2: Class Φ2 inverter. Peak drain voltage is reduced to ∼ 2 × Vin
separation. In our coils, the self-inductance of transmitter coil is L1 =0.403 uH,
while the self-inductance of receiver coil is L2 =16.3 µH. As coil separation
B. Resonant coupling increases, LM increases rapidly. So does, n, the effective turns ration of the
structure.
Figure 3 shows the model of the coupled coils connected
to the capacitors used to resonate the magnetizing and leak-
age inductance of the structure. Specifically, in this figure modest increase in complexity. Moreover, as described in [7]
LM represents the magnetizing inductance of the cantilever several components can be consolidated to improve power
model [11], while Ll is the leakage. Here, n models the the density. The gates for each MOSFET in the push-pull inverter
effective turn ratio. We used the cantilever transformer model are driven with out-of-phase signals to operate in alternation.
throughout our design because it allows us to easily incorporate In Fig. 6, coils are placed between the inverter and load to
part or the totality of the leakage and magnetizing inductance
into the design of the Φ2 inverter. In this model, when the
distance and/or position between the transmitter and receiver
coils varies, Ll and n change simultaneously, and this change
can be substantial. The variation in the parameter of the model
depends, among other things, on the number of turns of each
coil, its diameter etc. Moreover, due to the limited number of
low-loss magnetic materials able to operate in this frequency
range, our design does not use sheets of magnetic material
behind the transmitter and receiver coils used to steer the
magnetic flux and improve coupling.
If the number of turns on the primary coil is less than
the secondary coil, the effective ratio increases with distance.
This effective ratio is a significant factor when designing the
inverter. Therefore, if the distance changes significantly, the Fig. 5: Push-pull class Φ2 inverter. A single inverter is tuned first for the load
and another inverter is duplicated. MOSFET gate signals are phase shifted
inverter performance is adversely affected. In highly resonant 180◦ to operate inverters alternatively.
converters active re-tuning may be difficult or impractical.
Frequency variation can help mitigate de-tuning due to coil
form an air core transformer central to the WPT effort. The
parameter variation, but the benefits of this approach is limited
high Q resonance of the circuit containing the coils results in
as the frequency range in which the Class Φ2 inverter operates
an ac output with low harmonic content.
efficiently is narrow.

C. Push-pull inverter with resonant coil

III. E XPERIMENTAL V ERIFICATION
In this paper a push-pull class Φ2 inverter was designed to
deliver 175 W to a 50 Ω load, when VIN = 180V. Figure 5 The WPT system implemented here was designed to deliver
shows a schematic of the Φ2 push-pull inverter formed by 250 W of power through a 4 cm air-gap with a switching
combining two single ended inverters [12]. This configuration frequency of 13.56 MHz frequency. The design was simulated
leverages the tuning efforts of one converter, and applies it in and implemented experimentally to verify the feasibility of this
a way that increases output power and power density with a approach.
Fig. 6: Push-pull class Φ2 with receiving and transmitting coil. The primary
coil is connected to the output of the inverter. The load is attached to the
secondary coil. The impedance seen by inverter should be resistive to transfer
maximum power. The load coils form a transformer that also provides isolation Fig. 8: Measured drain and output voltage of the inverter under WPT
between the inverter and load. operation. Here Vin = 180 V and RL = 50 Ω, LM R =300 nH, LF =390 nH,
Ls =900 nH, CM R =114 pF, Cp =185 pF, Cs =1 nF, L1 =748 nH, L2 = 3.1 µH,
CM =184 pF, Cl =44 pF.
A. Single ended Φ2 Inverter
We started by designing and simulating a single ended Φ2
WPT circuit in LTspice. Our simulation incorporated relevant
parasitic parameters extracted by measurement on the PCB
used during the implementation. Moreover, our library models
pay careful attention to model the non-linear Coss and Cgd as
function of the applied drain voltage.
Figure 7 shows the simulated vds (t) and vload (t) of a single
ended Φ2 inverter of Fig. 2.

Fig. 9: Transmitting and receiving coil implementation. Here L1 =748 nH,

L2 =3.1 µH, CM =185 pF, Cl =44 pF and R=10 Ω.

while the 5 turn receiver has a self inductance of 3 µH. LM ,

Ll and n, the parameters of the cantilever model are obtained
from open and short impedance measurements as described
in [11]. LM is 0.748 µH and Ll is 14.75 µH. These values
are used to determine the value of capacitances CM and Cl
that resonate LM and LS at the switching frequency. For these
Fig. 7: Simulated drain voltage for a single inverter. Here fs = 13.56 MHz, values, the inductances of the cantilever model resonate at fs =
VIN = 180 V and RL = 50 Ω. Here LF =390 nH, LM R =300 nH, 13.56 MHz with CM = 185 pF and Cl = 44 pF connected to
CM R = 114 pF, LS =900 nH and Cs =1 nF. each coil as shown in Fig. 3. CM was connected to the primary
coil in parallel to resonate the The effective turn ratio of the
The simulation and implementation of the class Φ2 inverter coil arrangement is n =2.24 when the separation of the coils
driving a 50 Ω resistive load was designed for use with a is 4 cm. A 10 Ω load resistor is connected to the receiver coil.
500 V Microsemi ARF475FL. This MOSFET comes in a dual This way, when properly tuned, the equivalent resistance at the
rf package with low package inductance. In the preliminary fundamental seen at input of the transmitter coil is ≈ 50 Ω.
single ended implementation of the Φ2 inverter only one side of
ARF475FL was utilized. We proceeded to implement and test The output power and efficiency as function of input
the complementary single ended inverter that would eventually voltage is depicted in Fig. 10. Increasing input voltage leads
form a push-pull. We made sure both single ended converters to higher power and efficiency. Also, the variation of power
performed at comparable output power levels and present and efficiency with coil separation are plotted in Fig. 11.
similar vds (t) waveforms. Figure 8 shows the measured vds The transmitter and receiver coils were tuned at a 4 cm
and vload (t) when VIN = 180 V. In this operating condition, distance. Therefore, this is the point in the figure of peak output
the maximum vds (t) is ≈ 360 V. Notice the good agreement power and efficiency.
between experiment and simulation.

B. WPT with single ended Φ2 driving stage. C. WPT with a Φ2 Push-pull converter.
Figure 9 shows the two coils used for the WPT described Figure 12 shows the assembled PCB of the push-pull
here. The 2 turn transmitter coil has a inductance of 748 nH Φ2 inverters described here. Table I list the values of the
 TABLE I: Components values. Here L1 is the self inductance of the
180   75   transmitter coil while L2 is the self inductance of the receiver.
Parameter Value Units
170   74  
Output  power   Lf 390 nH
160   73  
Output  Power(W)  

LM R 300 nH
Efficiency  

Efficiency(%)  
Ls 900 nH
150   72  
CM R 114 pF
140   71   Cp 185 pF
Cs 1 nF
130   70  
L1 748 nH
120   69   L2 3.1 µH
110   68   CM 184 pF
Cl 44 pF
100   67  
M OSFET ARF475FL
150   160   170   180  
Input  Voltage(V)  

Fig. 10: Measured power and efficiency vs. VIN . Airgap is 4 cm. Efficiency with the appropriate phase difference. Figure 14 shows the
varies slightly between 68 % and 71 % in the 150 V ≤ VIN ≤ 180 V range. experimental measurement of the two drain voltages. Notice
that they have similar shape but are shifted by 180◦ . The single
260   80  
ended Φ2 inverters were designed for a 50 Ω load. When
connecting two out-of-phase inverters in a push-pull, the load
220   70   can be connected differentially between the output rails of the
Output  Power(W)  

Efficiency(%)  
inverters. To keep the respective drain impedance of each stage
unchanged the load impedance of the push-pull configuration
180   60  
must double. Based on the effective turn ratio of our trans-
mitting and receiving coils, a 20 Ω resistance connected to
Output  power  
140   50   the receiving end will have an equivalent impedance of 100 Ω
Efficiency   at the output of the push-pull. Figure 15 shows the output
voltage waveform across the 20 Ω load of the push-pull when
100   40  
2   3   4   5   VIN = 160 V. The output power under this operating condition
Distance(cm)   was 219 W with an efficiency of 71 %. The output power was
measured under varying voltage and distance as depicted in
Fig. 11: Measured power and efficiency vs. distance. Here, Vin = 170 V , Fig. 16 and 17. For push-pull inverter, the coil was tuned for
fs = 13.56 MHz. As coil separation increases, power delivery decreases. a separation of 2.5 cm.
Efficiency is maximum at 4 cm, the nominal coil separation.
Future work will include improvements in the layout and
components of the Φ2 inverters that form our push-pull design. reduce the inductance between the outputs of the two single-
ended stages and the transmitter coil. We also plan to study
the effect of low-loss, low-permeability materials to steer the
magnetic flux and improve coupling between the coils and
increase their separation.

Fig. 12: Photograph of assembled PCB with all components of the Φ2 push- Fig. 13: Set up used to measure converter performance. An rf PA is used
pull inverter. Note the gate drive tri-filar transformer next to the MOSFET. to provide power to the respective gates via the tri-filar transformer. During
testing, the converter and coils were placed at enough distance from metallic
structures that could affect coupling.
The PCB was laid out paying close attention to symmetry
to ensure parasitics of the two single ended converters were
the same. Out-phased gate signals, were obtained by using IV. C ONCLUSION
a tri-filar wound air-core transformer (shown on the right of
Fig. 12). Applying the output of a rf power amplifier to one This paper presents the design and implementation of a
of the windings, we obtained two equal isolated gating signal, wireless power transfer system based on the class Φ2 inverter.
that can be connected to drive the gates of the two MOSFETs High frequency operation provides high power density by
 230   80  

200   70  

Output  Power(W)  
170   60  

Efficiency(%)  
140   50  

110   40  

80   Output  power   30  

50   20  
Efficiency  
20   10  
2   2.5   3   3.5   4   4.5   5  
Distance(cm)  
Fig. 14: Measured drain voltage of push-pull inverter under WPT operation.
Here VIN = 160 V and RL = 20 Ω, fs = 13.56 MHz, LF =390 nH,
Fig. 17: Measured power and efficiency of the Φ2 push-pull inverter vs. coil
LM R =300 nH, Ls =900 nH, CM R =114 pF, Cp =185 pF, Cs =1 nF, L1 =748 nH,
separation. Here Vin = 160 V and RL = 20 Ω.
L2 = 3.1 µH, CM =184 pF, Cl =44 pF. The two drain voltage have same
waveform.
R EFERENCES
[1] W. Chen, R. A. Chinga, S. Yoshida, J. Lin, C. Chen, and W. Lo, “A
25.6 w 13.56 mhz wireless power transfer system with a 94amplifier,” in
Microwave Symposium Digest (MTT), 2012 IEEE MTT-S International,
June 2012, pp. 1–3.
[2] S. Kim, J. Ho, and A. Poon, “Wireless power transfer to miniature
implants: Transmitter optimization,” Antennas and Propagation, IEEE
Transactions on, vol. 60, no. 10, pp. 4838–4845, Oct 2012.
[3] A. Sample, D. Meyer, and J. Smith, “Analysis, experimental results,
and range adaptation of magnetically coupled resonators for wireless
power transfer,” Industrial Electronics, IEEE Transactions on, vol. 58,
no. 2, pp. 544–554, Feb 2011.
Fig. 15: Measured output voltage of push-pull inverter under WPT operation.
Here VIN = 160 V and RL = 20 Ω, LM R =300 nH, Ls =900 nH, [4] A. Kurs, A. Karalis, R. Moffatt, J. D. Joannopoulos, P. Fisher, and
CM R =114 pF, Cp =185 pF, Cs =185 pF, L1 =748 nH, L2 = 3.1 µH, M. Soljai, “Wireless power transfer via strongly coupled magnetic
CM =184 pF, CL =44 pF. Power is 219 W with 71 % efficiency. resonances,” Science, vol. 317, no. 5834, pp. 83–86, 2007. [Online].
Available: http://www.sciencemag.org/content/317/5834/83.abstract
[5] G. Covic and J. Boys, “Inductive power transfer,” Proceedings of the
reducing passive component size. The power amplifier for IEEE, vol. 101, no. 6, pp. 1276–1289, June 2013.
WTP is implemented based on class Φ2 inverter in order
[6] D. J. Perreault, J. Hu, J. M. Rivas, Y. Han, O. Leitermann, R. Pilawa-
to achieve low voltage stress and fast transient response. To Podgurski, A. Sagneri, and C. R. Sullivan, “Opportunities and chal-
enhance the concept of class Φ2 inverter, a push-pull inverter lenges in very high frequency power conversion,” in Proc. Twenty-
was designed to deliver more power to the load with little Fourth Annual IEEE Applied Power Electronics Conf. and Exposition
additional tuning effort. A 219 W 160 V Φ2 push-pull inverter (APEC), 2009.
switching at 13.56 MHz was implemented and experimentally [7] J. S. Glaser and J. M. Rivas, “A 500 W push-pull dc-dc power converter
tested. The efficiency is about 71 %. Efficiency increases are with a 30 MHz switching frequency,” in Proc. Twenty-Fifth Annual
IEEE Applied Power Electronics Conf. and Exposition (APEC), 2010,
expected through the development of different gating methods pp. 654–661.
and improved coil modeling and design. [8] N. Sokal and A. Sokal, “Class E-a new class of high-efficiency tuned
single-ended switching power amplifiers,” IEEE Journal of Solid-State
Circuits, vol. 10, pp. 168–176, 1975.
260   74  
[9] M. Iwadare, S. Mori, and K. Ikeda, “Even harmonic
Output  power  
250   73   resonant class-E tuned power amplifier without RF choke,”
Efficiency   Electronics and Communications in Japan (Part I: Communications),
240   72  
Output  Power(W)  

vol. 79, no. 1, pp. 23–30, 1996. [Online]. Available:

Efficiency(%)  

230   71   http://dx.doi.org/10.1002/ecja.4410790103
220   70   [10] J. M. Rivas, Y. Han, O. Leitermann, A. D. Sagneri, and D. J. Perreault,
“A high-frequency resonant inverter topology with low-voltage stress,”
210   69   IEEE Transactions on Power Electronics, vol. 23, no. 4, pp. 1759–1771,
200   68   2008.
190   67   [11] R. Erickson and D. Maksimovic, “A multiple-winding magnetics model
having directly measurable parameters,” in Power Electronics Special-
180   66   ists Conference, 1998. PESC 98 Record. 29th Annual IEEE, vol. 2, May
150   160   170   180   1998, pp. 1472–1478 vol.2.
Input  Voltage(V)   [12] J. Glaser and J. Rivas, “A 500 W push-pull dc-dc power converter with a
30 MHz switching frequency,” in Applied Power Electronics Conference
Fig. 16: Measured output voltage in the push-pull inverter. Here, coil sepa- and Exposition (APEC), 2010 Twenty-Fifth Annual IEEE, Feb 2010, pp.
ration is 2.5 cm. Power delivery increases with input voltage. The efficiency 654–661.
varies slightly within this input voltage range.