A Battery Charger with Maximum Power Point Tracking Function

Document Sample
A Battery Charger with Maximum Power Point Tracking Function Powered By Docstoc

        A Battery Charger with Maximum Power Point
        Tracking Function for Low-Power Photovoltaic
                    System Applications
                        Chun-Wei Lin                                                                    Jian-Min Wang
             Dept. of Electrical Engineering,                                                 Dept. of Vehicle Engineering,
         Chung-Yuan Christian University, Taiwan                                           National Formosa University, Taiwan
    Huang-Jen Chiu, Yu-Kang Lo, Ting-Peng Lee,
     Qing Su Chen, Wen Long Yu, Jian-Xing Lee                                                             Frank Shih
             Dept. of Electronic Engineering,                                                       Macroblock, Inc., Taiwan
  National Taiwan University of Science and Technology

Abstract—A battery charger with MPPT function for low-power                  discussed. Moreover, an auxiliary energy storage device such
PV system applications is presented in this study. For effective             as lead-acid battery is necessary to improve the dynamic and
miniaturization, the battery charger is designed with high                   steady-state performance for the stand-alone PV system. The
frequency operation. Some current-sensing techniques are                     lead-acid battery has low cost and high capacity features and
studied, and their MPPT implementation is compared. A pulse                  is widely used in various applications such as uninterruptible
charging method is also designed to prolong battery lifetime. The
                                                                             power system (UPS), automotive power system and telecom
operation principles and design considerations of the proposed
PV charger are analyzed and discussed in detail. A laboratory                power supply, among others. However, poor energy density
prototype is implemented and tested to verify the feasibility of the         characteristics, charging time and lifetime encumber
proposed scheme. Experimental results show that high MPPT                    commercial applications. Some charging methods such as
accuracy and conversion efficiency can be simultaneously                     constant voltage (CV) charging, constant current (CC)
achieved under high frequency operation.                                     charging, CC/CV two-phase charging and pulse charging are
   Keywords- Battery Charger, Low-Power PV System, Pulse                     commonly used for various battery charging systems. The CV
Charging, MPPT Accuracy, Conversion Efficiency                               and CC charging methods have the simplest control
                                                                             algorithms but shorten the battery lifetime due to their over-
                       I.    INTRODUCTION                                    current or overcharging problems. In this paper, performance
       In recent years, the growing concern over                             comparisons between a conventional CC/CV two-phase
environmental issues has brought about great interest and                    charging and a novel pulse charging methods will be studied
remarkable investments in photovoltaic (PV) technology.                      for lead-acid battery in low-power PV system applications.
Compared with other renewable energy sources, PV systems
                                                                                              II.    SYSTEM DESCRIPTION
have numerous advantages such as clean electricity
generation, high reliability, little maintenance and no audible                     A block diagram of the studied PV charger is shown in
noise due to the absence of moving parts [1-3]. Stand-alone                  Figure 1. The MPP tracking can be achieved by connecting a
PV systems are commonly used in low-power applications                       DC/DC converter between the PV array and battery stack.
such as LED lighting. The output power of PV arrays is                       MPPT algorithm for PV array and battery charging scheme are
always changing with solar irradiation conditions and                        both implemented in a DSP controller. PV array voltage and
atmospheric temperature. An MPPT control is usually applied                  current are sensed by the DSP controller to calculate the PV
to extract maximum power from the PV arrays. Hall-effect                     array power and directly control the power converter duty
sensors are commonly used to sense PV array current for                      cycle. The battery voltage is also fed back to predict the state
MPPT implementation. However, a hall-effect sensor with                      of charge (SOC) of the battery in determining its appropriate
high bandwidth feature is too expensive and huge for low-                    charging mode.
power PV system applications. Meanwhile, a current-sensing
resistor method has the merits of high linear quality, low cost
and simple configuration. However, the power dissipation in
current-sensing resistor brings about a thermal problem that is
undesirable in miniaturizing a PV system. In this paper, a
novel current-sensing method which uses the drain-source on-
resistance Rds(on) of the power MOSFET is studied for MPPT
implementation. Some performance comparisons among
various current sensing techniques using Hall-effect sensor,
current-sensing resistor and drain-source on-resistance Rds(on)
                                                                             Figure 1 Studied battery charger with MPPT function for low-power PV
for low-power PV system applications are analyzed and                        system applications

        A current-sensing technique by using a Hall-effect                 addition of rest period can relax the electrolyte reaction and
sensor that is commonly used for MPPT implementation in                    prolong the battery lifetime.
high-power PV system applications. The Hall-effect sensor
has the advantages of high accuracy and high noise immunity.
However, it is too expensive and huge for low-power
applications. Another current-sensing technique by using a
simple resistor Rsense has both high linear quality and low cost.
Power dissipation on current-sensing resistor causes a thermal
problem that is undesirable for miniaturizing a PV system in
low-power applications. In this paper, a current-sensing
technique is studied by using the drain-source on-resistance
Rds(on) of power MOSFET. It is more suitable for low-power
PV system applications because there is no extra-resistive
sensing loss so that the PV system can be miniaturized                                                      (a)
        Some parameters including conversion efficiency ηeff,
MPPT accuracy ηmppt and overall efficiency ηoverall are
regularly used to evaluate the electrical performance of the
DC/DC converter with MPPT function as follows:
 η eff = o                                          (1)
 η mppt =                                           (2)
η overall = η eff × η mppt                         (3)
where Po and Pin denote the output power and input power of
the DC/DC converter while Ppv and Ppv,m represent the actual                                                (b)
output power and maximum output power of the PV array.                     Figure 2 a) Conventional two-phase charging and b) the proposed pulse
         Figure 2(a) shows the conventional CC/CV two-phase                charging methods.
charging scheme. At the beginning of charging process, a CC
                                                                                           III.   DESIGN CONSIDERATIONS
charging scheme with MPPT function is adopted to avoid the
occurrence of over-current problems due to low battery                              The aim of this paper is to study the battery charger for
voltage. As the battery voltage rises to a preset voltage Vb,cv,           low-power PV system applications. The design considerations
the MPPT function is disabled and a CV charging scheme is                  for battery charging schemes, MPPT algorithm, power circuit
used to prevent overcharging problem. The CC/CV two-phase                  and PV current sensing techniques will be described and
charging method has some drawbacks as follows:                             discussed in detail as follows:
1) Battery current sensing is necessary to implement CC                    A. Battery Charging Design
    charging. The charging performance depends on the                               Figure 3 shows the flowchart of the studied pulse
    accuracy and reliability of the used current sensor.                   charging method. The battery voltage Vb is noted to determine
2) During CV charging period, MPPT function must be                        the charging mode while the PV array voltage Vpv and current
    disabled to avoid overcharging problems.                               Ipv are noted to implement MPP tracking as follows:
3) As shown in Figure 2(a), the continued electrolyte reaction             1) When the battery voltage Vb is lower than the preset level
    during the charging process brings about shortening the                     Vb,set, the pulse charging control flag (PC) is reset to “0”
    battery lifetime.                                                           and the pulse charging function is disabled. During this
         In this paper, a new pulse-charging method for low-                    charging mode, MPPT algorithm is carried out to extract
power PV system applications is studied, as shown in Figure                     the maximum power output of the PV array.
2(b). When the battery voltage stays less than the preset level            2) When the battery voltage Vb is higher than Vb,set, the
Vb,set, the MPPT charging proceeds. When the battery voltage                    control flag PC is set to “1” and the pulse charging
rises to Vb,set, a pulse charging current is sent to battery. MPPT              function is enabled. The charging period Tp and rest period
control remains during the charging period Tp while                             Tr of battery current is determined by calculating
electrolyte reaction of battery is relaxed during the rest period.              Equations (4) and (5).
The battery voltage Vb is sensed to determine the pulse width                                       ⎡ (V [k ] − Vb [k − 1])         ⎤
                                                                           T p [k ] = T p [k − 1] − ⎢ b                     × 0.9Tc ⎥ (4)
of battery charging current. When the battery voltage Vb
reaches Vb,f, a pulse current with a fixed narrow width is sent
                                                                                                    (             )
                                                                                                    ⎢ Vb, f − Vb,set
                                                                                                    ⎣                               ⎥
to float-charge the battery. For the studied pulse-charging                Tr = Tc − T p                                         (5)
method, no other sensor for battery current is needed. The

where Tp[k] and Tp[k-1] are the charging period at the                      algorithm is shown in Figure 4. The PV array voltage V[k]
sampling times k and k-1, respectively. Vb,f denotes the preset             and array current I[k] are sampled and used to calculate the
float-charging voltage and Vb,set is the preset voltage level for           instantaneous PV array power P[k]. The power obtained in the
pulse charging. Vb[k] and Vb[k-1] represent the sampling                    next iteration P[k+1] is compared with the power calculated in
battery voltage at times k and k-1.                                         the previous iteration P[k]. In addition, the PV array voltage
3) When the charging period counter P_count is less than the                V[k+1] at the next iteration is compared with the previous PV
    calculated charging period Tp, it will accumulate to                    array voltage V[k] to determine whether the converter duty
    continue the MPP tracking. As the P_count counts to Tp, the             cycle should be increased or decreased to reach the MPP.
    rest period counter R_count begins accumulating. The PWM
    signal is disabled till the counter R_count equals Tr. When
    the reset period is ended, the counters P_count and R_count are
    both reset. The charging period Tp and rest period Tr of
    battery current are then recalculated.
4) As long as the battery voltage Vb is higher than the preset
    float-charging voltage Vb,f, the pulse width of charging
    current is set as the minimum pulse width Tp_min. During
    the charging period, MPP tracking is remained for float-
    charging the battery. This period will be terminated as the
    battery voltage reaches the preset overcharge level Voc.

                                                                                      Figure 4 Flowchart of the adopted MPPT algorithm
                                                                            C. Power Circuit Design
                                                                                   The PV voltage in the practical implementation is
                                                                             between 7V to 17V for typical climate conditions and the
                                                                             nominal battery voltage is equal to 24V to allow for the
                                                                             power circuit in boost operation. The switching-frequency is
                                                                             designed as equal to 300kHz for circuit miniaturization. The
                                                                             continuous current conduction mode (CCM) operation is
                                                                             considered in the selection of the power inductor L as
                                                                                  Vpv × D
                                                                            L=                                              (6)
                                                                                 ΔIpv × fs
                                                                            where △Ipv denotes the PV current ripple that is usually equal
                                                                            to 10% of rated current. The output filter capacitor C can be
         Figure 3 Flowchart of the studied pulse charging method
                                                                            determined by Equation (7) as follows:
       As mentioned above, the pulse width of charging
current supplied by the studied PV charger is determined by                       Io × D
                                                                             C=                                              (7)
the battery voltage. MPP tracking is retained during the                         ΔVo × fs
charging period to raise the utility of PV array. The electrolyte           where △Vo is the voltage ripple which is usually equal to 1%
reaction can be relaxed during the reset period to prolong the              of the nominal battery voltage. In this paper, the Texas
battery lifetime.                                                           Instruments DSP chip TMS320F2808 which has 16-bit PWM,
B. MPPT Algorithm                                                           12-bit ADC and 100 MIPS is used to realize the digital
       As mentioned in Section I, an MPPT algorithm that                    controller of the studied PV charger. Figure 5 shows the DSP
provides high accuracy tracking is needed. For the low-power                control diagram. The internal PWM counter is used to
PV system studied in this paper, a Perturb and Observe (P&O)                generate PWM signal. The PV array current is sampled by
method which has some important advantages such as                          using a PWM interruption. The duty cycle of the PWM signal
simplicity and good performance is adopted. The control

is determined according to the battery charging control and the           Vds(on) can be used to represent the PV array current Ipv as
MPPT algorithm with ADC sampling average.                                 follows:
                                                                            V ds(on) = I pv Rds(on)                    (11)
                                                                                 It is obvious that the MPPT performance will be
                                                                          strongly related to the variations of the drain-source on-
                                                                          resistance Rds(on) shown in Figure 7. The characteristic
                                                                          consideration of Rds(on) is important for the power MOSFET
                                                                          selection. There is no extra-resistive sensing loss for Rds(on)
                                                                          current-sensing technique so that the PV system can be
                    Figure 5 DSP control diagram
                                                                          miniaturized significantly and suitable for low-power
D. Current-Sensing Design                                                 applications.
       Figure 6(a) shows the current-sensing circuit by using a
Hall-effect sensor. The Hall-effect sensor has a current
attenuation ratio Ai and can convert the PV array current Ipv to
a voltage signal Vs_pv by using a simple resistor Rp as follows:
         I pv
V s_pv =        Rp                                 (8)
       The voltage signal Vs_pv is then sent to ADC pin of the
DSP chip via a voltage follower to avoid loading effect. A
current-limiting resistor Rs and a 3.3V zener diode Zc are
added to protect the ADC pin of DSP chip. The current-
sensing technique with Hall-effect sensor has high noise
immunity, but it is too expensive for the studied low-power
applications. The bandwidth of a commercial Hall-effect
sensor is usually not high enough such that the implemented
MPPT accuracy may deteriorate under high-frequency
operation. Figure 6(b) shows the current-sensing circuit by                                                      (b)
using a current sensing resistor. The PV array current Ipv can
be converted to a negative voltage signal Vs_pv by a simple
resistor Rsense as follows:
V s_pv = − I pv Rsense                             (9)
        The voltage signal Vs_pv is then sent to ADC pin of the           Figure 6 a) Hall-effect sensor b) current-sensing resistor and c) Rds(on) current-
DSP chip via an inverting amplifier (INV Amp) with a                      sensing circuits
negative voltage gain (-R2/R1). It can be observed from
Figure 6(b) that a sensed error for PV array voltage Vpv may
occur due to the addition of a current-sensing resistor Rsense as
V pv,sen = V pv − I pv Rsense                     (10)
where Vpv,sen and Vpv are the sensed and actual PV array
voltages. For example, by using a current-sensing resistor
Rsense of 0.22Ω, the MPPT accuracy will deteriorate about 2%
under an operation condition with PV array voltage Vpv of 7V                              Figure 7 Drain-source on-resistance variations
and PV array power Ppv of 4W. The sensed voltage error
problem can be solved as long as the calculation for the                              IV.     EXPERIMENTAL VERIFICATIONS
instantaneous PV array power is compensated with the                              A laboratory prototype for the studied PV charger is
considerations of PV array current. The current-sensing                   designed and tested to verify the feasibility of the proposed
resistor technique possesses both high linear quality and low             schemes. The maximum power of the prototype circuit is
cost. However, power dissipation on current-sensing resistor              25W. The switching frequency is designed at 300kHz for
brings about the thermal problem that is undesirable in                   circuit miniaturization. Based on the design equations (6) and
miniaturizing a PV system in low-power applications. In this              (7), the power choke L is the 47μH SMD inductor and the
paper, a new current-sensing circuit is studied by using the              output filter capacitor C is 10μF. Two 12V/7Ah batteries are
drain-source on-resistance Rds(on) of power MOSFET as shown               series-connected to serve as the load of the designed prototype
in Figure 6(c). The drain-source voltage Vds(on) is sensed                circuit. The overcharge voltage Voc, the float-charging voltage
during on-state via a sampling transistor Qs. A 3.3V zener                Vb_f, and the pulse charging voltage for Vb_set are set at 27.6V,
diode Zc is added to protect the ADC pin of DSP chip. The                 27.3V and 26.4V, respectively in accordance with the battery

manufacturer’s suggestions. Thee current sensing techniques                     voltage mentioned in Section III worsen MPPT accuracy.
shown in Figure 8 are tested and compared for MPPT tracking                     Therefore, the studied current-sensing technique, which uses
of PV array. A commercial Hall-effect sensor LA 100-P with a                    the drain-source on-resistance Rds(on) of the power MOSFET
current attenuation ratio of 2000:1 is used in experiments for                  will be the best choice considering the system cost, circuit
current-sensing circuit as shown in Figure 6(a). LA 100-P is                    miniaturization and overall efficiency.
very popular for industrial applications due to its high noise
immunity even though its bandwidth is only 200kHz. For
another current-sensing circuit shown in Figure 6(b), a 0.22Ω
resistor Rsense is used to sense the PV array current. The
voltage Vs_pv is amplified by an inverting amplifier with a
voltage gain of -10 before it is sent to the ADC pin of DSP
chip. For the third current-sensing circuit shown in Figure
6(c), the PV array current is sensed directly by using the
power MOSFET with Rds(on) of 0.22Ω. Figure 8(a) shows the
measured waveforms for MPP tracking by using the drain-                                                            (a)
source on-resistance Rds(on) under 7V PV array voltage (at
maximum power point of 4.08W) while Figure 8(b)
demonstrates the measured MPPT waveforms under 17V PV
array voltage (at maximum power point of 24.08W). It can be
observed that the actual tracked powers are 3.88W and 23.6W,


                                                                                Figure 9 Measured a) conversion efficiency b) MPPT accuracy and c) overall
                                                                                       Figure 10 shows the measured waveforms in the
                                                                                studied pulse-current charging scheme. During the charging
                                                                                period, the MPPT operation is retained in the studied PV
                                                                                charger. The measured accuracy is approximately 97.63%.
                                                                                The MPPT function is disabled and the electrolyte reaction of
                                                                                battery stack can be relaxed during the reset period. The
                                                                                instantaneous PV power and duty cycle of the power switch at
                                                                                the end of the previous charging period are recorded and
                                                                                applied to the beginning of the next charging period. Thus, a
                                  (b)                                           high MPPT accuracy can be maintained for the studied pulse-
Figure 8 MPPT waveforms by using Rds(on) under a) Vpv=7V and b) Vpv=17V
                                                                                current charging scheme.
       Figure 9 shows the measured conversion efficiency ηeff,
MPPT accuracy ηmppt, and overall efficiency ηoverall by using
the studied three current-sensing techniques. Highest
conversion efficiency and MPPT accuracy features can be
achieved by using the Hall-effect sensor. However, the Hall-
effect sensors are usually too expensive and huge so they are
not suitable for low-power PV system applications. By using
current-sensing resistor, the overall efficiency becomes the
worst for the prototype circuit. The extra power loss and
thermal dissipation on the current-sensing resistor result to
lower conversion efficiency while a sensed error for PV array

         Figure 10 Measured pulse-current charging waveforms             implementation of a DC/DC module directly mounted on a
                                                                         single solar panel in low power applications.
                      V.     CONCLUSION
       In this paper, a novel battery charger for low-power PV                  The authors would like to acknowledge the financial
system applications is studied. The operation principles and             support of the Macroblock, Inc. and National Science Council
design considerations for the studied PV charger were                    of Taiwan through grant number NSC 96-2628-E-011-115-
analyzed and discussed in detail. A laboratory prototype was             MY3.
designed and tested at a switching frequency of 300kHz to                                         REFERENCE
verify the feasibility of the studied scheme. Three current-             [1] B. Sahan, A. N. Vergara, N. Henze, and A. Engler, P. A.
sensing techniques namely those that use the Hall-effect                     Zacharias, “Single-Stage PV Module Integrated Converter
sensor, the current-sensing resistor Rsense and the drain-source             Based on a Low-Power Current-Source Inverter,” IEEE Trans.
on-resistance Rds(on) of power MOSFET were tested and                        Industrial Electronics, vol. 55, no. 7, pp. 2602-2609, July. 2008.
compared for MPPT implementation. For the considerations                 [2] Mohibullah, Imdadullah, and I. Ashraf, “Estimation of CO2
                                                                             Mitigation Potential through Renewable Energy Generation,”
of system cost, circuit miniaturization and overall efficiency,
                                                                             IEEE PECon’06, pp.24-29, Nov. 2006.
the studied Rds(on) current-sensing technique has good                   [3] D. Sera, R. Teodorescu, J. Hantschel, and M. Knoll, “Optimized
performance. A pulse-current charging method with MPPT                       Maximum Power Point Tracker for Fast-Changing
function was also studied and tested. The electrolyte reaction               Environmental Conditions,” IEEE Trans. Industrial Electronics,
of battery stack can be relaxed to prolong the lifetime without              vol. 55, no. 7, pp. 2629-2637, July. 2008.
the use of extra battery current sensor. The studied PV charger
with MPPT function is especially suitable for the