Contactless Power Charger for Light Electric Vehicles Featuring Active Load Matching

Abstract

Contactless power transfer technology is gaining increasing attention in city transportation applications because of its high mobility and flexibility in charging and its commensurate power level with conductive power transfer method. In this study, an inductively coupled contactless charging system for a 48 V light electric vehicle is proposed. Although this study does not focus on system efficiency, the generic problems in an inductively coupled contactless power transfer system without ferromagnetic structure are discussed. An active load matching method is also proposed to control the power transfer on the receiving side through a load matching converter. Small signal modeling and linear control technology are applied to the load matching converter for port voltage regulation, which effectively controls the power flow into the load. A prototype is built, and experiments are conducted to reveal the intrinsic characteristics of a series-series resonant inductive power charger in terms of frequency, air gap length, power flow control, coil misalignment, and efficiency issues.

I. INTRODUCTION

Unlike fast-developing, highly penetrated, and high-mobility information technology, power transmission systems still rely on conductive energy transfer. Extensive designs with contactless power transfer (CPT) systems [1], including industrial conveyance systems and vehicle battery-charging systems, exist, and their system efficiency reaches up to 90% [2]-[8].

The air gap length in inductively coupled CPT applications has been restricted to the close proximity range in contrast to transmitter/receiver coil size. References [5]-[7] proposed inductive CPT applications from low power electronics to heavy industry. References [8]-[11] evaluated the inductive CPT in low-voltage (LV) applications, such as electric vehicles (EVs), in terms of cost, efficiency, and converter topologies. In the aforementioned applications, the converter switching frequency is limited below hundreds of kilohertz for loss control purpose. Ferrite structures are widely used in the literature to facilitate energy transfer and to protect against magnetic flux irradiation [2]-[13]. Reference [12] studied the alignment issues in inductively coupled CPT application by passively compensating for leakage inductance. Reference [13] eliminated misalignment problems through a carefully designed guide rail structure.

Strongly coupled magnetic resonance offers omnidirectional transmission distance that is longer than that of the inductively coupled CPT [4]. Given that the coil quality factor dominates the spatial energy transfer, the coil design and associated passive components selection becomes crucial for such CPT system. References [14]-[22] discussed the properties and performance improvement of such systems through detailed modeling, frequency optimization, coil design, and impedance matching. However, high-efficiency power electronic converters that operate in the megahertz range remain a great challenge for magnetic resonance CPT applications.

Inductive coupled and magnetic resonance CPT technologies and microwave and electric field coupling methods are also discussed in the literature for specialized applications [23], [24].

Conventional isolated sensing methods, such as Hall effect and opto-coupling, are impractical because of spatial separation between the sending and the receiving coils. The load feedback control of a CPT can be implemented through several ways. References [25]-[27] used active bridge topology at both sides. Both primary side and secondary side regulations were implemented by applying phase shift control, in which a well-designed phase lock loop is essential for power control. References [28] and [29] applied the frequency modulation method to regulate load power. The load power information is fed back via a wireless communication channel from the receiving side. Reference [29] also proposed a charger controller for load power regulation through load current feedback. Reference [30] proposed a chopper circuit to modulate the load to control the power flow, in which the chopper switching period and the resonant frequency was matched precisely for the load power transfer.

Electric propulsion is becoming increasingly popular in city transportation systems. Although high-power applications, such as electric city buses and electric cars, are still not prevalent, medium-low power level applications, such as electric bikes, are well developed and affordable. E-bikes are usually equipped with a 36 or 48V lead–acid battery pack with a brushless DC motor as the driving unit. A series-excited brushed DC motor supplied with a 48 or 72V battery pack is more popular in medium power vehicles, such as service vehicles and electric forklifts. Contactless charger offers an automatic charging solution to the large number of EVs, which are considered a possible candidate for the imminent city transportation infrastructure.

An inductively coupled contactless power charger for a 48V EV is proposed in this study. The main power stages of the systems, including a high-frequency inverter, resonant network without ferrite, full-bridge rectifier, and load matching converter, are evaluated and designed. Although this study does not focus on system efficiency, an active load matching method is proposed to control the contactless power flow on the receiving side. The prototype is built with a digital signal controller (DSC) as the control unit. Tests are performed to quantify and reveal the intrinsic characteristics of a series–series resonant inductive power charger in terms of frequency, air gap length, power flow control, coil misalignment, and efficiency issues.

 

II. CPT CHARGER SYSTEM STRUCTURE

Charging infrastructures in cities are usually constructed beneath sheds. Thus, a solar power roof mount is considered the main input to the charger system. Therefore, the charger system is designed with an LV input. The proposed system structure, which consists of an inverter, resonant network, rectifier, and load matching converter, is illustrated in Fig. 1. A full-bridge inverter (FBI) is used to interface the LV DC input to utilize the LV input fully, as shown in Fig. 1(b). A medium-frequency square wave is obtained with the FBI to drive a sending-side series resonant network through L1 and C1. On the receiving side, the receiving coil L2 and a series connected capacitor C2 form the secondary resonant tank, which is connected to the AC terminal of a full-bridge rectifier with large output capacitance. A ferrite structure is not considered to achieve a compact design in a highly vibrational driving condition. An impedance matching converter, which is a boost converter in this case, is connected between the rectifier and battery load for charging power conditioning.

Fig. 1.Contactless power charger: (a) system structure and (b) circuit schematic.

 

III. ANALYSIS OF INDUCTIVE POWER TRANSFER

A. Analysis of Series Resonant Power Transfer

A conventional voltage-source square-wave inverter is chosen as the sending-end converter, such that the LV solar input can be fully utilized and transferred by the downstream resonant network. The series resonant structure is chosen over the parallel structure because of its good current limiting nature as a series-connected variable impedance, where the resonance frequency fr is given by

where L1,2 and C1,2 are the series capacitance and coil inductance (including leakage), respectively.

The series–series (SS) resonant topology is shown in Fig. 2, where the self-inductance from the sending and receiving coils is analyzed in a T-equivalent circuit. M is the mutual inductance, whereas R1 and R2 are the equivalent series resistance (ESR) for the coil and resonant capacitor. A simple AC circuit analysis is performed, and the ideal efficiency can be obtained as in Equ. (2) through the relationship between the primary current I1 and secondary current I2 in condition (3).

Fig. 2.SS resonant topology.

As can be observed in the expression, the energy transfer efficiency is closely related to the coupling factor and the quality factor of the resonant network. The maximum efficiency can be obtained by Equ. (4) if the squares of operation frequency is sufficiently larger than the term R1·(RL+R2)/M2.

The maximum efficiency in the reduced expression (4) depends on the receiving side parasitic parameters and load. Load resistance is usually considerably larger than the secondary parasitic resistance. Thus, the maximum theoretical efficiency is high. We can also infer that load variation does not substantially affect the efficiency in a medium-low CPT system.

B. Load Matching Converter Design

As a DC–DC converter with continuous input power, the boost converter is particularly useful in load-shaping applications, such as photovoltaic power conditioning and power factor correction. In these applications, the boost converter, as an active load, can change the steady-state impedance, such that the source can detect different loads during the power transfer.

As discussed, the power control function is required in a battery charger application. Thus, the concept of loss-free resistor is adopted, as indicated in Fig. 3. This resistor is a boost converter with S1 and S2 switching synchronously. The boost converter can shape the port variable such that the port impedance can be adjusted to the desired value. The power flow into the battery can be controlled properly by adjusting the impedance of the load matching converter.

Fig. 3.Load matching with a power converter.

As an SS resonant topology is inherently a current link topology in a small-signal sense. The uncontrolled rectifier with a large output capacitor is used as a voltage sink to interface the current link topology. The rectifier shows voltage source characteristics to the downstream converter because of the large output capacitance. Therefore, regulating the port voltage v1 in Fig. 3 with a current–sink circuit helps regulate the power transferred from the sending end. Unlike the control methods in [28] and [30], port voltage is a direct state variable to the boost converter circuit, which can be adjusted continuously without jeopardizing the trajectories of the other state variables, particularly the inductor L current.

Fig. 4.Schematic of a load-matching converter.

The average switch model is used to obtain the transfer function of the load-matching converter. The input resistance Rl is obtained by the rated power of 200 W, and Ro is the battery ESR at 80% state of charge. The transfer function of the duty cycle to the input voltage is determined by applying small signal perturbations, as in Equ. (5). This transfer function is a third-order system with a zero on the numerator. The negative sign indicates that the input voltage decreases with the positive perturbation of the duty cycle, which can be ensured by flipping the control logic in the feedback loop.

in which

A double-pole single-zero PI controller is proposed, as in Equ. (6). The position of the feedback and the reference is flipped in the digital implementation because of the negative sign in the Gvc1d. Hence, this position is neglected in the loop analysis. The bode plot for the compensated system open loop transfer function Tloop, plant Gvc1d, and voltage controller Cv is plotted, as illustrated in Fig. 5. Given that the charging occurs during the parking state, no dynamic misalignment or air gap variation is assumed. Therefore, the power to be transferred to the battery is not necessarily regulated tightly. Based on this criterion, the load matching controller is designed conservatively, thereby resulting in a cross-over frequency of 2 Hz and a phase margin of 91.5°, as indicated in Fig. 5.

Fig. 5.Bode plot for port voltage regulation loop.

 

IV. EXPERIMENT RESULTS AND ANALYSIS

A. Prototype Setup and Frequency Test

The hardware prototype is implemented, and a 16-bit dsPIC controller with a 40 MHz clock is used as the controller. An adjustable bench power supply with 30 V and 20 A is selected as the input source to prove the principle. The load is a lead–acid battery pack rated at 48 V and 100 Ah. The major system parameters are listed in Table I.

TABLE ISYSTEM PARAMETERS

Fig. 6 shows the prototype and the test bed. The experiments are conducted under different scenarios.

Fig. 6.Hardware prototype and test bed.

Frequency sweep and transient test are performed first to ensure the frequency characteristics and the feasibility of the hardware design. Fig. 7 shows the start-up transient of 200 W power transfer when the full-bridge converter works at 32 kHz. The waveform indicates that the system can achieve smoothed output power transfer without oscillated voltage or current in the steady state. The resonant inverter is designed to achieve a stable operation on the circuit level by operating at the slightly inductive region. Fig. 8 shows the phase delay between the inverter output voltage and current waveform on the sending end. The coil current lags slightly behind the inverter voltage, which is suitable for series resonant operation.Frequency sweep and transient test are performed first to ensure the frequency characteristics and the feasibility of the hardware design. Fig. 7 shows the start-up transient of 200 W power transfer when the full-bridge converter works at 32 kHz. The waveform indicates that the system can achieve smoothed output power transfer without oscillated voltage or current in the steady state. The resonant inverter is designed to achieve a stable operation on the circuit level by operating at the slightly inductive region. Fig. 8 shows the phase delay between the inverter output voltage and current waveform on the sending end. The coil current lags slightly behind the inverter voltage, which is suitable for series resonant operation.

Fig. 7.Startup transient under an open-loop load-matching control.

Fig. 8.Steady-state performance at tuned frequency; v1 - inverter output voltage (10V/div), i1 - sending coil current (10A/div), iBatt - battery charging current (5 A/div).

B. Active Load Matching Test and Analysis

The active load matching is verified by fixing the system input voltage and by varying the load-matching converter input voltage control reference Vdc*. The air gap length is set at 70 mm. The sending and receiving coil voltages v1 and v2, input source current is, and battery charging current iBatt are measured at different rectifier output voltages vdc, as shown in Fig. 9. The voltage glitch across the coil reflects the inverter output voltage, which is experimentally verified with KVL. Figs. 9(a) to 9(e) illustrate that the rectifier output voltage vdc is regulated to different levels from 13 V to 21 V, respectively. As vdc increases, both input current and charging current increase; the charging power also increases. The effective charging power control via the proposed active load matching method is demonstrated.

Fig. 9.Active load matching control with variable rectifier output voltage, 2 us/div; is - dc input current (5A/div); iBatt - battery charging current (2A/div); v1 - sending coil voltage (100 V/div); v2 - receiving coil voltage (50 V/div): (a) vdc = 13 V, (b) vdc = 14 V, (c) vdc = 16 V, (d) vdc = 19 V, and (e) vdc = 21 V.

Fig. 10 shows the collective experimental results with active load matching control at three different air gap lengths; these results indicate the susceptibility of the transferred power to the port voltage vdc. Both input and output powers increase linearly with the load port voltage vdc in Fig. 10(a), while the system efficiency remains nearly constant during the load match for different power levels. Thus, the conclusions in Section III A are confirmed.

Fig. 10.CPT via active load matching at different air gap lengths: (a) input and output powers and (b) system throughout efficiency.

C. Sensitivity Analysis

A series of sensitivity tests is performed. As confirmed in Section IIIA, load port voltage dominates the transferred power. The influences from air gap and coil misalignment are studied in the following test. The charging power is maintained at 200 W to ensure fair comparisons.

The coil size is relatively compact for a light EV (e-bike), which is only 170 mm × 250 mm. No high magnetic permeability structure exists at either side to assist the coupling. Therefore, the designed maximum air gap length is within the half short side length of the coil. In the air gap sensitivity test, the air gap length is swept from 30 mm to 75 mm. As can be observed from the data in Fig. 11, the efficiency drops as the air gap increases. Distance plays an important role in an inductively coupled CPT because the power transfer is closely related to the linkage between the two coils. The test results also indicate that the maximum air gap length can be increased if the sending/receiving coil size is increased.

Fig. 11.System efficiency at 200 W output power with different air gap distances.

The misalignment tests are carried out under horizontal and vertical misalignment conditions. As indicated in Fig. 12, both Δd and Δl are swept from 0 to 70 mm. The maximum misalignment is also visually noticeable to vehicle drivers during vehicle parking and can be corrected by the driver easily.

Fig. 12.Horizontal and vertical misalignment condition setup.

Fig. 13 shows the misalignment test results under different air gap lengths. The output power is fixed at 200 W in the tests. Horizontal misalignment tests indicate approximately 10% efficiency drop throughout the misalignment range, whereas only a 5% drop exists in the vertical misalignment conditions.

Fig. 13.Efficiency measurement with coil misalignment: (a) horizontal misalignment and (b) vertical misalignment.

The tilting angles between the sending and receiving coils vary under the full alignment condition, as shown in Fig. 14. The efficiency under 200 W load power condition undergoes a 6% drop when the tilting angle changes from 0 to 15 degrees.

Fig. 14.Tilting condition setup.

Fig. 15.Efficiency variation with tilting angle.

The misalignment at different directions reflects changes in the coil overlapping area. Thus, the power transfer efficiency is affected by the effective coil overlapping area. The performance of the charger stays within an acceptable range.

The measurement of the sending and receiving coil voltages in Fig. 9 does not reflect a clear relationship between these two voltages because of the voltage deformation caused by power electronic circuits. Therefore, tests are performed to investigate the coupling by varying both input source voltage and load matching port voltage. The coil current is sinusoidal in the resonant state. Thus, a comparison between the sending coil current I1 and receiving coil current I2 is performed. The relationship among the current transfer ration I1/I2, input source voltage, and load matching port voltage is obtained in Fig. 16. The “coupling” k is typically used in analyzing the inductive coupled CPT. Thus, the data obtained from the experimental work does not show explicit coupling behavior by observing the current transfer ratio. This result is due to an extra current shunt path formed by the mutual inductance M in the canonical T-model, which is not negligible compared with a traditional power transformer. Therefore, applying linear magnetic circuit analysis in static performance prediction in a CPT system with power electronic conditioners is challenging. However, the flow of power can be controlled in a close loop, which renders predictable power transfer behaviors, through the proposed active load matching method.

Fig. 16.Current transfer ratio plot at 70 mm air gap.

 

V. CONCLUSION

An inductive coupled contactless charger for a 48 V light EV is proposed in this paper. The main power stages of the system, including the high-frequency inverter, resonant network, full-bridge rectifier, and load matching converter (the boost converter), are evaluated and designed. Although this study does not focus on the system efficiency, an active load matching method is proposed to control CPT. The prototype is built with a dsPIC DSC. The tests are conducted in several meaningful scenarios to illustrate the multifold characteristics of the proposed contactless power charger without ferrite structure support. Experimental verification includes the frequency characteristics of SS resonant topology, the sensitivity of some important system parameters, and the power flow control method. On the basis of the experimental results, the following conclusions can be obtained: