You are on page 1of 6

Nonlinear Modeling and Feedback Control of WPT

System via Magnetic Resonant Coupling Considering


Continuous Dynamic Tuning
Cheng Chen, Hong Zhou, Qijun Deng and Xin Luo
Department of Automation, Wuhan University
Wuhan, 430072, China
E-mail: dqj@whu.edu.cn

Abstract—In a wireless power transfer (WPT) system via in [7], where the impedance can be automatically
magnetic resonant coupling, the detuning will greatly reduce the reconfigured. Ren et al. [4] depict a synchronous-tuning
transmission power and efficiency. In this paper, a nonlinear method using a vacuum variable capacitance. A saturable DC-
tuning model and feedback controllers of the system considering feed inductor [8] and a transistor-controlled capacitance [3] are
the continuous dynamic tuning are developed to eliminate the used to provide continuously adjustable impedance. All of the
disturbance from the detuning. The phase difference between the methods mentioned above can maintain maximum efficiency
primary and secondary side is employed to predicate the and power by applying different iterative algorithms. But the
resonant state. The model is linearized around the operating dynamic analysis of the frequency-tuned or impedance-tuned
point by small-signal analysis methodology. Based on the
system has not been extensively studied, with previous work
dynamics of the linearized model, a PI controller for each side of
mainly concentrated on the steady-state models and
the system is designed. It enables the system to simultaneously
tuning at the primary and secondary side using a parallel
experimental verification.
digitally tunable capacitance. The closed-loop nonlinear state- The dynamics of the tuning model are necessary for the
space model is built and the numerical simulations verify the fast design of the feedback controller, and the phase and gain
dynamic tuning ability of the resonant tank to cope with the margin of the system should be properly designed to guarantee
detuning disturbance. the robustness and steadiness. In [9], a direct phase control
approach based on the Second-Order Generalized Integrator
Keywords—nonlinear tuning model; phase difference; feedback
Phase-Locked Loop (SOGI-PLL) is proposed to track the
control; wireless power transfer.
resonant frequency and the controller is designed by the
dynamic characteristics of the SOGI-PLL. Li et al. [10], [11]
I. INTRODUCTION present a nonlinear model to describe the coupled resonances
Magnetic resonant wireless power transfer system is by using the slowly varying amplitudes and phases of coupled
preferred in many daily and industrial applications due to modes rather than resonant currents and voltages. It is able to
delivering electrical energy cordlessly. As a kind of special analyze the dynamics of the phase around the resonant state.
power supply, the robustness and steadiness for power and To eliminate the disturbance from the detuning, this paper
efficiency of the system are an important issues of concern [1- presents a design procedure for the tuning controller of the
3]. Generally, the inherent parameters of the resonant tank and WPT system. In the design process, the dynamic
the operating frequency may dynamically drift away because of characteristics of the tuning model have been taken into
the temperature, moisture, ageing of the resonant tanks or consideration. The parallel digitally tunable capacitance is
ambient interference [4]. Furthermore, the detuning will greatly chosen as the tuning actuator and the PI controller is used to
reduce the transmission efficiency and power. In the remove the steady state error while maintaining the closed-loop
preliminary studies [3-9], various tuning technologies have system.
been investigated to maintain the maximum efficiency and
power against parameter variations. In [5], the tuning The paper is organized as follows: Section II analyzes the
technologies are divided into two types, frequency-tuned phase difference between the primary and secondary side under
method and impedance-tuned method. In order to compensate different resonant states and the minimal zero voltage
for changes of the detuning, the operating frequency of the switching (ZVS) phase. Section III focuses on the nonlinear
frequency-tuned system is variable and controllable. In modeling and linearizing. In section IV, the simulation model
contrast, the impedance-tuned system employs variable and circuit are set up to verify the proposed tuning controller
capacitance/ inductance to track the optimum impedance- and the experimental results are provided.
matching point in the case of varying parameters and operating
frequency. A serial or parallel capacitance matrix is introduced II. PROPOSED TOPOLOGY AND PHASE ANALYSIS

‹,((( 
Fig.1. Topology circuit of the analysed wireless charging system.

A. Analysis of Resonances and Phase Difference


A series–series compensated topology of the WPT system,
as shown in Fig. 1, consists of a DC input voltage, a full-bridge
inverter, the primary and secondary resonant tanks, a full-
bridge rectifier and load. By sensing the primary and secondary
currents, the phase difference and a PI controller are used to
maintaining the closed-loop system. The primary and
secondary sides consist of the inductance L1, L2, the series-
resonance capacitance C1, C2, the parallel digitally tunable
capacitance C1v, C2v, and the equivalent series resistance R1, R2
respectively. i1, i2, v1 and v2 represent the current across the
coils and the voltage of the series-resonance capacitance. RL is Fig.2. The equivalent circuit diagram of WPT system.
the load resistance and M is the mutual inductance between two
coils. where Zsreflect is the equivalent impedance that reflected by the
secondary side. The primary and secondary currents can be
Assuming that: 1) the angular switching frequency of the expressed as
full-bridge inverter is Ȧs, 2) the voltage and current waves of
the both side are sinusoid, 3) the filter capacitance of the full-
bridge rectifier is big enough. The fundamental components of vab Z 2
the input voltage at primary side of the resonant tank and the i1 = (4)
(ω s M )
2
+ Z1 Z 2
equivalent resistance of the rectifier and load can be simplified
as jωs Mi1
i2 = − (5)
Z2

vab = VI cos (ωs t + θ s ) (1) It is assumed that the phase difference between the current
i1 and equivalent fundamental voltage vab is ǻij, and ǻȖ is the
Rcd = 8 RL / π 2 (2) phase difference between the currents i2 and i1. According to
(4), when the operating frequency Ȧs and mutual inductance M
are fixed, the phase difference ǻȖ is only determined by Z2 and
where VI is the DC voltage source, șs is the initial phase of the
ǻij is related to Z1 and Z2. Moreover, only if C2+C2v=1/Ȧ2L2,
fundamental voltage vab.
the secondary resonant tank of the system achieves resonance
Combination of (1) and (2) leads to the equivalent circuit and the ǻȖ is constant by ʌ/2. Only when C1+C1v
shown in Figure 2. According to this equivalent circuit, the =1/Ȧ2L1=C2+C2v =1/Ȧ2L2, the both side of the system are
impedances of the primary and secondary resonant tanks are simultaneously at resonance state and the current i1 and voltage
expressed as following: vab have the same phase. Therefore, changing the value of ǻij
and ǻȖ by the adjustment of the parallel digitally tunable
capacitance enables the WPT system to maintain the resonant
Z1 = R1 + jωs L1 + 1/ jωs ( C1 + C1v ) + Z sreflect state.
Z sreflect = (ωs M ) / Z 2
2
(3)
B. Minimum ZVS Phase angle
Z 2 = R2 + Rcd + jωs L2 + 1/ jωs ( C2 + C2v ) The Zero voltage switching (ZVS) condition of the inverter
circuit may be interrupted by the detuning. In order to maintain


the ZVS operation, a minimum angle [12] between the current ­d m

i1 and equivalent fundamental voltage vab is needed to make the ° xi = fi ( x ) + ¦ gi ( x ) ui


resonant tank equivalent to an inductive load. The minimum ® dt i =1 (9)
angle ȗmin is defined as ° y = h ( x)
¯ i i

where, x= [a1㸪ș1㸪a2㸪ș2]T and y1= θ s − x2 , y2= x4 − x2 .


§ 2ωs CossVI ·
ζ min ≥ arccos ¨1 − ¸ (6) B. Model Linearization
© Im ¹
This nonlinear tuning model, also known as the large-signal
model, includes the frequency-tuned and inductance-tuned
where Coss is the output capacitor of the MOSFET, I m is the model. And it also represents the envelope of the phase
effective value of the current across the MOSFET and VI is the difference under both steady-state and transient conditions. In
DC input voltage. other words, the large-signal model contains both the operating
point and the small-signal model. Thus, each variables can be
III. DYNAMIC ANALYSIS OF THE TUNING SYSTEM written as the sum of operation point and Small-signal
Based on the amplitude and phase model and taking into
account the phase difference between the primary and
secondary side around the resonant state, a dynamic tuning x = xo + Δx (10)
model is established in this section.
u = uo + Δu (11)
A. Model Development
The following dynamic equations describe the behavior of In this paper, the parallel digitally tunable capacitance is
the equivalent circuit in Fig. 2: chosen as the tuning actuator. By substituting (10) into (9), the
nonlinear capacitance-tuned model can be linearized using the
Taylor’s series expansion and written in the state-space form
­ di1 (t ) di2 (t )
°vab = v1 + i1 R1 + L1 dt − M dt
°
°i = C + C dv1 (t ) ­d
°° 1 ( 1 1v )
° Δx = A ⋅ Δx + B ⋅ Δu
dt ® dt (12)
® (7) °¯ y = C ⋅ Δx
° M 1 = v + i ( R + R ) + L di2 (t )
di (t )
° dt
2 2 2 cd 2
dt
° dv ( t ) where ǻx=[a1o, ș1o 㸪 a2o 㸪 ș2o ]T, ǻu=[ǻC1v 㸪 ǻC2v]T, and
°i2 = ( C2 + C2 v ) 2 y1= θ s − Δx2 , y2= Δx4 − Δx2 .
°̄ dt
Parameters of main components are listed in Table I, where
According to [10], [11], when the resonances are loosely the ESRs of the coils are measured at the frequency 149.69kHz,
coupled, the voltages and currents of WPT system can be the value of mutual inductance is calculated using Ansoft
represented as Maxwell 15 when the power transfer distance is 100 cm.
IRF840 MOSFET is selected to construct the switching leg of
the inverter and the output capacitance can be achieved by
­ 2 querying the instructions. The initial value of the tunable
°in = ⋅ an cos(ωs t + θ n ) capacitances is 0.5nF.
° Ln
® (8) The operation point is calculated by setting the derivative
°v = 2
° n ⋅ an sin(ωs t + θ n ) terms in (9) to zero, and is given as
¯ ( C n + Cnv )

where an and șn(n=1,2) are defined so that the energy contained x0 = [ 0.239 −0.089 0.077 1.483]
T
(13)
in the primary and secondary resonant tank is a2n and the phase
angle of the current in, respectively. The nonlinear time-variant
model derived from equations (7) and (8) can be averaged by Substitution of xo and the circuit parameters into the state
taking the average values during a switching period. Because space-model yields the matrices (A B C) as given in (11)
the resonant state can be fixed by keeping the phase difference
ǻijm and ǻȖm to be constant, the nonlinear turning model can ª −309.5 26.1 −4134.4 8.686 º
be derived as follows: « −457.1 −309.5 −472.8 −1330.5 »»
A=« (14)
« 4133.8 −0.7 −1.3 × 104 −2.8 »
« 4 »
¬ 35.4 1.3 × 10 472.7 −1.3 × 10 4 ¼


ª 0 45.4 º ªG ( s ) º
« −1.3 × 105 −4.4 × 10−6 »
'
G pij ( s ) = « p11 »
G p 22 ( s ) »¼
(19)
B=« » (15) «¬
« −141.1 0 »
« −5 5 »
¬ −4.2 × 10 −1.3 × 10 ¼ where Gp11(s) and Gp22(s) is open-loop transfer function from u1
to y1 and from u2 to y2,
ª 0 −1 0 0 º
C=« » (16)
¬ 0 1 0 −1¼
1.3 × 105 s 3 + 3.5 × 109 s 2 + 2.6 × 1013 s + 3.6 × 1016
G p11 ( s ) = (20)
s 4 + 2.6 × 104 s 3 + 5.5 ×1011 s 2 + 4.5 ×1014
TABLE I. MAIN PARAMETERS OF WPT SYSTEM
1.3 × 105 s 3 + 2 × 109 s 2 + 5.7 ×1012 s + 4.7 ×1015
G p 22 ( s ) = (21)
Parameter Description Value s 4 + 2.6 ×104 s 3 + 5.5 × 1011 s 2 + 4.5 × 1014
The primary side resonant
L1 323.1ȝH
inductance
The primary side resonant
From the analysis of root locus and frequency-domain, the
C1
capacitance
3nF zero-poles, amplitude and phase characteristics of the closed-
loop system can be used to predict both time-domain transient
R1 The primary side ESR 0.2ȍ
and steady-state system performances. It can be seen from (20)
The secondary side resonant and (21), the root locus diagrams that the zero-poles composed
L2 323ȝH
inductance
by three pairs of dipoles and a dominant pole are all located in
The secondary side resonant
C2
capacitance
3nF the left s-plane, and each loop of the system indicate a first-
order inertia frequency characteristic. Thus the conclusions are
R2 The secondary side ESR 0.2ȍ drawn as following: 1) the transfer functions are closed-loop
M Mutual inductance 2.84ȝH stable, 2) the overshoot mainly depends on the damping
decrement of the dominant pole, 3) the settling times are
VI Input DC voltage 20V
6.1×10 -4 s and 7.8×10-5 s, 4) the loops can be simplified to a
fs
Switching frequency of the
149.69kHz first-order inertia process. To stabilize the system, a
inverter proportional gain much smaller than unity is necessary. This
RL Load resistance 10ȍ results in a PI controller of the following form:
Initial phase of the input AC
șs 0.01rad
voltage
§ 1 ·
C oss MOSFET output capacitance 350pF
Gc ( s ) = K c ¨1 + ¸ (22)
The range of the primary side © Tis ¹
C 1v 0.01~1nF
tunable capacitance
The range of the secondary side
C 2v 0.01~1nF where K c and T i are the proportional coefficient and the
tunable capacitance
integral time respectively. After several adjustments, the final
C. Input/output Characteristics and Controller Design parameter sets are selected to be: K c1 =0.4, Ti1=2.5×10-5,
From (12)-(16), it can be seen that the state-space model K c2 =0.2, T i2 =1.7×10 -5 . Assuming that the range of the
obtained is a linear process with couplings between each loops, digitally tunable capacitance is wide enough, comparing to the
where the output equations of y1 and y2 contain all the control disturbance from the detuning, the steady-state error can be
inputs u1 and u2. The relative gain Ȝij can be expressed as removed by the PI controllers.

IV. SIMULATION VARIFICATION


λij = Kij / K ij' (17) Fig. 4 shows the nonlinear tuning model of WPT system
shown in Fig. 1 is simulated by MATLAB Simulink with the
where Kij and K’ij are the open-loop gain and closed-loop gain, parameters listed in Table I to verify the capability of the
respectively. And the relative gain matrix is calculated by tuning controllers in this section. There are two PI controllers
combining (14)-(18) as are used to remove the steady state error while maintaining the
closed-loop system. The minimum angle ȗmin and constant ʌ/2,
as the set points, are used to achieve ZVS condition and the
ª0.9999999972 2.8 × 10−9 º maximum efficiency and power, respectively. To validate the
ȁ= « −9 » (18) capability of the controllers, the different detuning cases are
¬ 2.8 × 10 0.9999999972 ¼ listed in Table II and emulated by step changing the parameters
of the resonant capacitances C1 and C2, respectively. From the
The matrix is diagonal dominance so that the couplings perspective of the phase in both side, the detuning cases caused
between each loop are too weak to be considered. The transfer by the inductance and the resonant frequency are similar to the
function matrix of the multi input multi output (MIMO) model cases in Table II. And the dynamic performance indexes of the
can be simplified to cases are arranged in table III.


Fig.4. The nonlinear tuning model of WPT system with PI controller.

TABLE II. THE DETUNING CASES

Case The step changes of the parameters The step time


I C 1 : 3nF ĺ2.98nF 0.01s

II C 2 : 3nF ĺ2.94nF 0.01s

Fig.7. The tunable capacitances C1v and C2v of the case I.

is apparent that, without controller, the transmission power


drops from 145W to 65W and the value of the phase difference
ǻij changes from -0.09 rad to 1 rad. On the contrary, the PI
controller can remove the error from the disturbance and keep
the transmission power and phase difference stabilized after
Fig.5. The transmission power of the case I. two oscillations in 1.5ms. Fig. 7 shows the tunable capacitance
C1v in the primary side changes from 0.5nF to 0.52nF when the
detuning appeared at the meantime.

TABLE III. DYNAMIC PERFORMANCE INDEX

CASE Parameter Description Value


ı The overshoot of ǻij 30%

ts The regulation time of ǻij 1.5ms


I
ı The overshoot of ǻȖ 5.8%

ts The regulation time of ǻȖ 1.5ms

ı The overshoot of ǻij 30%

ts The regulation time of ǻij 1.5ms


Fig.6. The phase difference ǻij and ǻȖ of the case I. II
ı The overshoot of ǻȖ 2.1%

ts The regulation time of ǻȖ 1.5ms


The response curves of the power, phase differences and
tunable capacitances under the case I are shown in Fig. 5-7. It


V. CONCLUSION
This paper focuses on the dynamic analysis and feedback
controller design of a capacitance-tuned WPT system with two
digitally tunable capacitances. Based on the amplitude and
phase model and taking into account the phase difference
between the primary and secondary side around the resonant
state, a nonlinear capacitance-tuned model is proposed to
analyze the dynamics of the feedback control loops. After that
the nonlinear tuning model of WPT system is established by
using MATLAB Simulink. The experiments of the detuning
case I and II show that, under the control, the capacitance-
tuned WPT system is able to achieve the ZVS condition, and
Fig.7. The transmission efficiency of the case II. keep a steady transmission efficiency and power against the
tuning of the primary and secondary resonant tanks.

REFERENCES
[1] M. R. Basar, M. Y. Ahmad, J. Cho, and F. Ibrahim, “Stable and High-
Efficiency Wireless Power Transfer System for Robotic Capsule Using a
Modified Helmholtz Coil,” IEEE Transactions on Industrial Electronics,
vol. 64, pp. 1113-1122, 2017
[2] J. Lee, K. Lee et al. “Stability Improvement of Transmission Efficiency
Based on a Relay Resonator in a Wireless Power Transfer System,”
IEEE Transactions on Power Electronics, vol. 32, pp. 3297-3300, 2017.
[3] J. Tian and A. P. Hu, “A DC-Voltage-Controlled Variable Capacitor for
Stabilizing the ZVS Frequency of a Resonant Converter for Wireless
Power Transfer,” IEEE Transactions on Power Electronics, vol. 32, pp.
2312-2318, 2017.
[4] Ren, Jieshuai, et al. “Tuning of mid-range wireless power transfer
Fig.8. The phase difference ǻij and ǻȖ of the case II.
system based on delay-iteration method.” IET Power Electronics, vol. 9,
pp. 1563-1570, 2016.
[5] Heebl, Jason D, et al. “Comprehensive Analysis and Measurement of
Frequency-Tuned and Impedance-Tuned Wireless Non-Radiative
Power-Transfer Systems.” IEEE Antennas & Propagation Magazine,
vol. 56, pp. 44-60, 2014.
[6] A. P. Sample, D. A. Meyer and J. R. Smith, “Analysis, Experimental
Results, and Range Adaptation of Magnetically Coupled Resonators for
Wireless Power Transfer,” IEEE Transactions on Industrial Electronics,
vol. 58, pp. 544-554, 2011.
[7] Y. Lim, H. Tang, S. Lim, and J. Park, “An Adaptive Impedance-
Matching Network Based on a Novel Capacitor Matrix for Wireless
Power Transfer,” IEEE Transactions on Power Electronics, vol. 29, pp.
4403-4413, 2014.
[8] S. Aldhaher, P. C. Luk, A. Bati, et al. “Wireless Power Transfer Using
Class E Inverter With Saturable DC-Feed Inductor,” IEEE Transactions
Fig.9. The tunable capacitances C1v and C2v of the case II. on Industry Applications, vol. 50, pp. 2710-2718, 2014.
[9] P. Tan, H. He and X. Gao, “A Frequency-Tracking Method Based on a
SOGI-PLL for Wireless Power Transfer Systems to Assure Operation in
In the case II, the efficiency, phase differences and tunable the Resonant State,” Journal of Power Electronics, vol. 16, pp. 1056-
capacitances response to the disturbances are shown in Fig. 7-9. 1066, 2016.
Fig. 7 depicts the PI controllers are able to make the efficiency [10] H. Li, K. Wang, L. Huang, W. Chen, and X. Yang, “Dynamic Modeling
to stabilize after two period of adjustment in 1.5ms. It can be Based on Coupled Modes for Wireless Power Transfer Systems,” IEEE
seen in Fig. 8 that the phase difference ǻij and ǻȖ are also Transactions on Power Electronics, vol. 30, pp. 6245-6253, 2015.
quickly stabilized at the set points (ȗmin and ʌ/2). Fig. 9 depicts [11] H. Li, J. Li, L. Huang, et al. "A novel dynamic modeling method for
the tunable capacitance C2v in the primary side changes from wireless power transfer systems." Applied Power Electronics
Conference and Exposition IEEE, 2015:2740-2743.
0.5nF to 0.56nF due to the controller action.
[12] M. K. Kazimierczuk and D. Czarkowski, “Phase-controlled resonant
inverters”, chapter 6 in Resonant Power Converters, 2nd ed. Wiley –
IEEE process, 2011.



You might also like