AC Voltage Sensorless Control of Battery Charger System in Electric Vehicle Applications Cong-Long Nguyen
Hong-Hee Lee
School of Electrical Engineering University of Ulsan Ulsan, South Korea
[email protected]
School of Electrical Engineering University of Ulsan Ulsan, South Korea
[email protected]
Abstract—This paper presents an effective control scheme for an electric vehicle battery charger where a symmetrical bridgeless power factor-corrected converter and a buck converter are cascaded. Both converters have been popular in industries because of their high efficiency, low cost, and compact size, hence combining these two converters makes the overall battery charging system strongly efficient. Moreover, this charger topology can operate under universal input voltage condition and attain a desired battery voltage and charging current without any ripple. In order to reduce the system cost, an estimating ac input voltage technique is introduced, which especially does not require the converter component information and therefore it is robust to the circuit parameters variation. Additionally, by adopting a duty ratio feed-forward path in current control loop, a unity input power factor and zero input current harmonic are achieved. The feasibility and practical value of the proposed approach are verified by simulation and experimental results. Keywords-Electric vehicles (EV); battery charger; buck converter; symmetrical bridgeless PFC converter; duty ratio feedforward; total harmonic distortion (THD); power factor (PF);
I.
INTRODUCTION
In recent years, increasing concern of environment and energy resources has made the use of electric vehicles (EVs) obviously necessary and ubiquitous in nature. Due to the exhaust emissions and the poor energy conversion efficiency, the conventional vehicles powered by internal combustion engines are blamed as a major source of the current global warming and energy crisis [1]. In addition, the demand of using vehicles in the world is increased dramatically in the past tenyear, and this demand seems to be continued in next decades [2]. Therefore, to reduce disadvantageous effects of the vehicle, government agencies and organizations have been developing stringent standards for fuel consumption together with emissions in vehicular products. These standards urge the automotive makers moving to new vehicle generation, and EVs become one of the most promising candidates to replace the current petroleum-based vehicular system [3]. In EVs, power electronics including power conversion unit, traction motor part, and battery charger module play a critical role to decide whether the EVs are popular in the market [4]. The biggest limitation of the EVs now is located in energy storage device, i.e. battery. Factors such as short driving range,
978-1-4673-4584-2/12/$31.00 ©2012 IEEE
515
long charging time, and high initial cost confine batterypowered EVs to only small vehicular applications [5]. To overcome this restriction, new concept of EVs named plug-in hybrid electric vehicles (PHEVs) is introduced [2]-[5]. In PHEVs, the battery can be charged from a grid at home or at charging station so that the vehicles are able to increase their driving range. In order to charge the battery of PHEVs from the grid, a high power charger circuit capable of AC-DC conversion with power factor correction and a DC-DC converter to adjust the charging current and voltage are required and generally shown in Fig. 1 [6]. In this battery charging topology, the first converter is a front-end power factor correction (PFC) converter, which is used to control input power factor (PF) at unity, to regulate the AC current being a pure sinusoidal waveform, and to keep a DC voltage in the DC-link at a desired value. Usually, three sensors including the DC voltage, AC input voltage and AC input current sensors are required to control this PFC converter. Whereas, the second converter needs to detect the battery charging current and voltage to execute a desired charging algorithm [7]. The requirements result in a high cost and complicated charging system. In this paper, by cascading a symmetrical bridgeless power factor corrected converter (SBPFC) and a buck converter, a new battery charger topology is suggested. Both of these converters have been popular in industries because of their advantageous features such as high efficiency, low cost and compact size, hence combining two converters makes the proposed charging system strongly efficient. Inherently, the buck converter is a low-pass filter and the SBPFC belongs to the boost type converter so that the proposed charger is able to charge batteries with any nominal charging voltage and obtain a desired charging current without any ripple. To control the charger topology and reduce the system cost, an estimating AC input voltage technique is introduced, which especially does not require the converter parameters and therefore it is robust to the converter parameters variation. Additionally, by adopting a duty ratio feed-forward path in current control loop, a near unity input power factor and an input sinusoidal current are achieved. The feasibility and practical value of the proposed approach are verified by simulation and experimental results with 110V/60Hz ac line input and 1.5kW-72V DC output of the battery charging system.
IPEC 2012
Figure 1. An overview of the battery charger for PHEVs.
II.
Figure 2. The proposed battery charger.
DESCRIPTION OF THE PROPOSED CHARGER TOPOLOGY
A. Proposed Charger Topology Fig. 2 shows the configuration of the proposed battery charger, which contains a front-end PFC converter and a DCDC converter. From the point of view of input and output voltage waveform, the charger is an AC-DC converter that converts an ac voltage of the grid to a dc voltage of the battery. To utilize the grid at a maximum active power, a unity input power factor of the battery charger must be achieved, so that the front-end PFC is adopted. Under unity input power factor condition, the grid current will be in phase with the grid voltage and, hence, the dc output voltage of the front-end PFC oscillates twice of the grid frequency because of the imbalance between the alternating input power and DC output power. As this output voltage ripple of the front-end PFC is significantly high to apply to the battery, the second stage DC-DC converter is necessary for safe operation of the battery. Owing to intrinsic advantages over the conventional PFC rectifier boost converter [8], the SBPFC is the best appropriate topology for high efficiency, compactness, and low cost [9]. Therefore, the SBPFC is suggested to take a role of the frontend converter in the battery charger. However, it is noted that the dc output voltage of this converter is higher than the grid peak voltage, thus, to make the charger be capable of operating under universal input voltage circumstance, a simple but efficient buck converter is selected for the second stage converter position. Overall scheme of the proposed battery charger is shown in Fig. 2. B. The SBPFC Operating Principle In order to control the proposed battery charger effectively, principal operation of the front-end PFC converter should be defined. Fig. 3 illustrates the distinct operation modes of the converter. At any given instant, only two switches are on state in the power flow path. When the input AC voltage is positive, inductor L0, switch M0 and body diode of M1 constitute the conventional boost converter circuit (Fig. 3a & b). To decrease the conduction losses, switch M1 should be on during this time interval. Fig. 3(a) shows the current flow when the input AC voltage is positive and the switch M0 is closed. Input current flows through switch M0 and backs through switch M1. At the same time, the output capacitor C0 discharges and supplies current to the load. Fig. 3(b) shows the situation when input AC voltage is positive and the switch M0 is open. Current flows through diode D0, the capacitor and load, and backs through switch M1. Similarly, when the input AC voltage is negative, inductor L0, switch M1 and diode D1 constitute the PFC circuit (Fig. 3c & d), and switch M0 should be kept on during this negative AC voltage interval.
(a)
(b)
(c)
(d)
Figure 3. Principal operation of the SBPFC. In (a) and (b), the grid voltage is positive; M0 is the switching device; M1 is always turned on. In (c) and (d), the grid voltage alternates to be negative state then M0 is always kept on and M1 becomes the switching component.
C. Modeling of SBPFC In order to model the behaviors of the SBPFC, some assumptions are initially made: 1) the converter elements are ideal and thus, lossless. 2) The switching frequency f S is infinite as compared with the grid frequency. 3) The output capacitor C0 is large enough to maintain the output dc voltage vDC constant. 4) The input power factor is unity, hence, if the grid voltage is defined as v AC (t ) = VP sin(ωt ) then the grid current will be iAC (t ) = I P sin(ωt ) and the input power is
516
pI (t ) = v AC (t )iAC (t ) =
VP I P VP I P − cos(2ω t ) . 2 2
(1)
Based on the above assumptions and the instantaneous power conservation law, the following equation is achieved pI (t ) = pO (t ) = vDC (t )iDC (t )
(2)
where pO (t ) is the instantaneous output power and iDC (t ) is output current of the converter. Due to the bulk output capacitor C0, the output voltage vDC is supposed to be its averaged value VDC . Therefore, the output current iDC can be derived as following iDC (t ) =
pI (t ) VP I P VP I P = − cos(2ωt ) = I DC + idc (t ) VDC 2VDC 2VDC
(3)
where I DC and idc are DC component and ripple component of the output current, respectively. They are defined I DC = idc (t ) = −
VP I P 2VDC
VP I P cos(2ωt ) . 2VDC
(4) (5)
The ripple component idc is just a current through the output capacitor C0, which leads to variation of the output voltage. If vdc denotes this voltage variation then it can be calculated as vdc (t ) =
VP I P 1 idc (τ )dτ = − sin(2ωt ) . ³ C0 4ω C0VDC
(6)
Considering that the converter is operating between kth and (k+1)th switching cycle within time interval from tk to tk +1 . When the switch M0 (M1) turns on or off, the grid current can be described as (7) and (8), respectively.
L0 L0
di AC = v AC when tk ≤ t < tk + d k Ts dt
(7)
di AC = v AC − vDC sgn(v AC ) when tk + d k Ts ≤ t < tk +1 (8) dt
where d k is duty ratio of the switch at kth switching cycle and TS is the switching period or TS = 1 f S . By using the stateaveraging approach, the above two equations can be combined to obtain a relationship between the duty ratio, the grid current, the output and grid voltage of the SBPFC as following
L0
III.
d iAC dt
= v AC − (1 − d )vDC .
(9)
Figure 4. Block diagram of the proposed control scheme for the charging system.
As shown in Fig. 4, the proposed control scheme implemented in a digital signal processing (DSP) microcontroller does not require AC grid voltage sensor. For the purpose of digital computation, required analog control variables including the grid current iAC , the output voltage of the SBPFC vDC , the battery voltage vB and the charging current iB must be converted into digital quantities by using analog-to-digital converter (ADC) functions of the DSP. A. PFC Algorithm Controlling the SBPFC is turning on and off the switch M0 (M1) at a proper duty ratio d in order to achieve the unity input power factor and the desired dc output voltage. By detecting the output voltage vDC and comparing with its reference, a ratio K between grid current and grid voltage is determined. Then the reference of the grid current is constructed by multiplying the ratio K with the estimated grid voltage. From (6), the output voltage always consists of the ripple component with double frequency of the grid, which leads to reducing the performance of the outer proportional-integral (PI) controller. In order to overcome this issue, a second order notch-filter owning a center frequency doubles that of the grid is adopted to remove this output voltage variation. Based on the output voltage and the duty ratio of the switch M0 (M1) at previous switching cycle, the AC grid voltage is estimated
v eAC = (1 − d F )vDC
(10)
THE PROPOSED CONTROL SCHEME
This section focuses on control of the charging system. As discussed earlier, the charging system comprises two cascaded converters that are the SBPFC used to regulate input power factor correction and the buck converter used to manage the charging side. In order to control this system effectively, the proposed control shown in Fig. 4 contains two independent parts. The first part named PFC-Algorithm takes the role of controlling the SBPFC and the other called ChargingAlgorithm adopt to control the buck converter.
517
where v eAC denotes the estimated AC grid voltage and d F is obtained from the duty ratio d by using the one-cycle delay operator z −1 as illustrated in Fig. 4. From (9), the duty ratio can be derived
d =1− with
v AC vDC
+
1 L0 d iAC = d DF + d PI vDC dt
(11)
d PI =
1 L0 d i AC , vDC dt
d DF = 1 −
and
v AC vDC
.
(12)
(13) Driver
Power Stage
According to (11), the duty ratio d contains two control variables, which are d PI part as outcome of a PI controller and d DF element supplied through a duty ratio feed-forward path. As defined in (12), the PI controller loop is just to compensate the voltage drop in the inductor instead of compensating for | v AC | entirely. This leads to the grid current iAC tracking its reference with smaller error. Actually, output of the PI controller is very small compared with unity, hence, from (11) and (12), the duty ratio feed-forward path can be considered as the switch turn-on time ratio at previous switching cycle
d DF = d − d PI ≈ d F .
DSP
(14)
As a result, the estimated AC grid voltage expressed in (10) is demonstrated through combining (13) and (14). B. Charging Algorithm The charging algorithm named constant-current constantvoltage (CC-CV) method is used to charge lithium-ion, leadacid or some other batteries, which are vulnerable to damage if their upper voltage limit is exceeded [10]. The charging strategy is implemented in the charging system by two PI controllers. The outer one regulates the battery voltage at a desired value, and its output is limited at the charging current rate I C . Meanwhile, the inner controller controls the charging current by creating a duty cycle for the buck switch M2. This control signal is also limited to ensure the battery voltage is not over its upper limit [11].
Battery Bank Fig. 5. Hardware setup of the battery charging system. TABLE I.
KEY COMPONENTS USED IN THE CHARGING SYSTEM
Devices
Part # / Value
MOSTFET: M0, M1, M2
IRFP460
Diode: D0, D1, D2
RURG8060
Capacitor: C0, C1
840 uF
Inductor: L0, L1
1.05 mH
Driver
HCPL-3120
Voltage Transducer
LEM LV 25-P
Current Transducer
LEM HX 30-P
TABLE II.
PARAMETER VALUES OF THE BATTERY BANK
Battery Model
NEWMAX PNC 12500
Nominal Voltage
12V
Typical Capacity
50Ah Max. Current
16A
Max. Voltage
14.4 ~ 15.0V
Internal Resistance
48 mΩ
Charging Current
16A
Cut-Off Current
2A
Max. Voltage
86V
Charge Condition
Battery Bank
IV.
SIMULATIONS AND EXPERIMENTAL RESULTS
Construction
Simulations and experiments are carried out in order to verify the effectiveness of the proposed control scheme. The suggested battery charger is setup and used to charge a 72V50Ah seal lead-acid battery bank from an 110V/60Hz AC grid. The simulation is performed through using PSIM software, and the overall charging system as shown in Fig. 5 is developed in laboratory with a high performance DSP (TMS320F28335 by Texas Instruments). Table I lists the semiconductors and power components used in the 1.5 kW experimental prototypes. In both simulation and experiment, the battery bank has specifications as shown in Table II. Furthermore, all MOSFETs in the system are controlled and switched at 40Khz. The commanded value of the SBPFC output DC voltage is set at 170V. In simulation results, Fig. 6 shows steady-state performance of the system during charging the battery with 12A. In Fig. 6a, three signals including the grid current, the real and estimated grid voltage are plotted out. It is clear to recognize that the grid voltage and current have no phase displacement with the unity
518
(12[V]x6=72[V])
input power factor. In addition, the grid current is a pure sinusoidal signal that prevents the battery charging system contaminating the grid. Another benefit of the proposed control scheme is no ripple in both battery voltage and charging current as illustrated in Fig. 6b. Actually, the estimated grid voltage lags its real value but the lagged-time is very small ( 20 μ s ). Therefore, this error of the estimating method does not significantly affect to the system performances. In order to test dynamic response of the proposed control scheme, the commanded charging current value is alternated from 8A to 12A. The simulation results in this case are shown in Fig. 7 providing features: the grid current keeps a sinusoidal waveform and near unity input power factor even though the load current is changed (Fig.7a); the charging current tracks its reference simultaneously without any ripple in the battery voltage (Fig. 7b).
v eAC
v AC
i AC
veAC
v AC
0
10ms / div
50V / div
15 A / div
5ms / div
(a). The real grid voltage, the estimated grid voltage and the grid current.
75V / div
Figure 8. Experimental results: Testing the method of grid voltage estimation.
v AC
Battery voltage 50V / div
0
Charging Current 0
i AC
0
12 A / div 10ms / div (b). The battery voltage and the charging current.
10ms / div
50V / div
15 A / div
Figure 6. Simulation results: The battery charging system at steady state time.
v AC
Battery voltage
i AC 0
0
50V / div Charging Current
20ms / div
50V / div
0
15 A / div
12 A / div
(a). The grid voltage and the grid current. Figure 9. Experimental results: Steady-state performances of the proposed control scheme with 12A charging current.
Battery voltage
0
v AC
50V / div
0
Charging Current 8 A / div
10ms / div 50V / div
20ms / div (b). The battery voltage and charging current.
FFT of vAC
Figure 7. Simulation results: Testing dynamic response of the charging system.
100 Hz / div 5V / div
In experiment, the system is also tested at both steady state and transient state. In Fig. 8, the grid voltage and its estimated value are shown. As seen in this figure, the estimating method is able to detect the grid voltage exactly. In Fig. 9, the grid voltage, the grid current, the battery voltage and the charging current are illustrated when the system charges with 12A to the battery bank. In order to demonstrate that the charging system operates at the near unity input power factor and injects zero current harmonic to the grid, FFT of the grid voltage and current are illustrated in Fig. 10. It is recognized that the grid current is not distorted so much (THD=2.37%) and in phase with the fundamental component of the grid voltage (PF=0.994).
519
i AC 10ms / div 15 A / div FFT of iAC 100 Hz / div 3 A / div
Figure 10. Experimental results: FFT of the grid voltage and the grid current.
v AC
A
1s / div
30V / div
B
i AC
input current; 2) no ripple in both the battery charging current and the battery voltage; 3) fast dynamic response; 4) no requirement of AC voltage sensor; 5) no need of circuit parameters during estimating the grid voltage. Simulation and experimental results are shown to prove the value of the proposed control scheme.
20 A / div
ACKNOWLEDGMENT
Battery voltage
This work was partly supported by the NRF grant funded by the Korea government (MEST) (No. 2010-0025483) and the Network-based Automation Research Center (NARC) funded by the Ministry of Knowledge Economy.
50V / div Charging Current
4 A / div
REFERENCES (a)
A
v AC
10ms / div
i AC
[1]
B
50V / div
15 A / div (b)
Fig. 11. Experimental results: (a) Testing dynamic response of the proposed controls scheme. (b) Zooming in at A and B regions within (a) where charging current command is changed.
In order to test dynamic response of the proposed control scheme, the charging current reference is alternated between 8A and 12A within 6s and the system performances in this case are illustrated in Fig. 11. At the times of altering the charging current, the battery voltage is kept without any variation (Fig. 11a) and the grid current is regulated to be in phase with the grid voltage (Fig. 11b) are demonstrated clearly good dynamic performances of the proposed control scheme. V.
CONCLUSION
In this paper, an AC sensorless control scheme is proposed for a universal input battery charger where two efficient converters including a symmetrical bridgeless power factorcorrected converter and a buck converter are cascaded. The proposed control scheme provides following advantageous features: 1) a near unity input power factor and a sinusoidal
520
A. Emadi, Y. J. Lee, and K. Rajashekara, “Power Electronics and Motor Drives in Electric, Hybrid Electric, and Plug-In Hybrid Electric Vehicles,” IEEE Trans. Ind. Electron, Vol. 55, No. 6, pp. 2237–2245, Jun. 2008. [2] C. C. Chan, ‘‘The state of the art of electric, hybrid, and fuel cell vehicles,’’ Proc. IEEE, vol. 95, no. 4, pp. 704---718, Apr. 2007. [3] C . C. Chan, A. Bouscayrol, K. Chen, ‘‘Electric, Hybrid, and Fuel-Cell Vehicles: Architectures and Modeling, ’’ IEEE Trans. Veh. Technol., vol. 59, no. 2, pp. 468–476, Feb. 2010. [4] A. Emadi, Y. J. Lee, K. Rajashekara, “Power Electronics and Motor Drives in Electric, Hybrid Electric, and Plug-In Hybrid Electric Vehicles, ” IEEE Trans. Ind. Electron., vol. 55, no. 6, pp. 2237–2245, Jun. 2008. [5] A. K. Yadav, P. Gaur, A. K. Jha, J. R. P. Gupta, and A. P. Mittal, “Optimal Speed Control of Hybrid Electric Vehcles,” Journal of Power Electronic, vol. 11, No. 4, pp. 393-400, July 2011. [6] B. Bilgin, A. Emadi, and M. Krishnamurthy, “Design Considerations for a Universal Input Battery Charger Circuit for PHEV Applications,” Industrial Electronics (ISIE), 2010 IEEE International Symposium on, pp. 3407 –3412, July 2010. [7] Y B. Bilgin, E. Dal Santo, and M. Krishnamurthy, “Universal Input Battery Charger Circuit for PHEV Applications with Simplified Controller,” Applied Power Electronics Conference and Exposition (APEC) 2011, PP. 815 -820 , March 2011. [8] B. Singh, B. N. Singh, and A. Chandra, “A review of single-phase improved power quality AC-DC converters,” IEEE Trans. Ind. Electron.,vol. 50, no. 5, pp. 962–981, Oct. 2003. [9] R. Martinez and P. N. Enjeti, “A high-performance single-phase rectifier with input power factor correction,” IEEE Trans. Power Electron., vol.11, pp. 311–317, July 1996. [10] Michael G. Egan, Dara L. O’Sullivan, John G. Hayes, Michael J.Willers, and Christopher P. Henze, “Power-Factor-Corrected Single-Stage Inductive Charger for Electric Vehicle Batteries,” IEEE Transactionon Industrial Electronics, Vol. 54, No. 2, pp.1217-1226, Apr 2007. [11] K. Y. Kim, S. H. Park, S. K. Lee, T. K. Lee, and C. Y. Won, “Battery charging system for PHEV and EV using single phase AC/DC PWM buck converter,” 2010 IEEE Vehicle Power and Propulsion Conference, pp.1217-1226, Sept. 2010.