This action might not be possible to undo. Are you sure you want to continue?

# 4344 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 56, NO.

11, NOVEMBER 2009

Modeling and Control of PV Charger

System With SEPIC Converter

S. J. Chiang, Hsin-Jang Shieh, Member, IEEE, and Ming-Chieh Chen

Abstract—The photovoltaic (PV) stand-alone system requires a

battery charger for energy storage. This paper presents the mod-

eling and controller design of the PV charger system implemented

with the single-ended primary inductance converter (SEPIC). The

designed SEPIC employs the peak-current-mode control with the

current command generated from the input PV voltage regulating

loop, where the voltage command is determined by both the PV

module maximum power point tracking (MPPT) control loop and

the battery charging loop. The control objective is to balance the

power ﬂow from the PV module to the battery and the load such

that the PV power is utilized effectively and the battery is charged

with three charging stages. This paper gives a detailed modeling

of the SEPIC with the PV module input and peak-current-mode

control ﬁrst. Accordingly, the PV voltage controller, as well as the

adaptive MPPT controller, is designed. An 80-W prototype system

is built. The effectiveness of the proposed methods is proved with

some simulation and experimental results.

Index Terms—Maximum power point tracking (MPPT), power

balance control, single-ended primary inductance converter

(SEPIC), stand-alone.

I. INTRODUCTION

S

OLAR POWER is more and more attractive due to the

severer environmental protection regulation and the pre-

dictable shortage of conventional energy sources [1], [2]. As a

result, many research works have addressed the development of

solar power system in recent years. Many types of photovoltaic

(PV) power conversion systems have been developed including

the grid-connected system for reducing the power from the

utility [3]–[6] and the stand-alone system for providing the

load power without the utility [6]–[8]. The stand-alone system

requires battery for energy storage to supply the load power dur-

ing the period without or shortage of solar power. Because the

P−V characteristic of the PV module is varied with the insola-

tion level as well as the temperature [9], [10], if the peak power

voltage of the PV module does not match with the battery volt-

age, the energy conversion efﬁciency of the PV module will be

reduced using the direct connection of the PV module and the

battery. Therefore, a battery charger is required to track the peak

Manuscript received October 31, 2007; revised August 11, 2008. First

published September 23, 2008; current version published October 9, 2009.

This work was supported by the National Science Council, Taiwan under Grant

NSC95-2221-E-239-048.

S. J. Chiang is with the Department of Electrical Engineering, National

United University, Miaoli 360, Taiwan (e-mail: sjchiang@nuu.edu.tw).

H.-J. Shieh is with the Department of Electrical Engineering, National Dong

Hwa University, Hualien, Taiwan (e-mail: hjshieh@mail.ndhu.edu.tw).

M.-C. Chen is with the FSP Technology Inc., Taoyuan, Taiwan (e-mail:

m9323003@gmail.com).

Color versions of one or more of the ﬁgures in this paper are available online

at http://ieeexplore.ieee.org.

Digital Object Identiﬁer 10.1109/TIE.2008.2005144

power of the PV module in all operation conditions [11]. In

addition, the battery charging needs control for achieving high

state of charge (SOC) and, consequently, longer lifetime of the

battery [12].

This paper explores the charger system implemented with

the single-ended primary inductance converter (SEPIC) [13].

Although the boost converter usually has higher efﬁciency than

the SEPIC, however, it is only applicable for cases where the

battery voltage is higher than the PV module voltage. The

buck–boost feature of the SEPIC widens the applicable PV

voltage and thus increases the adopted PV module ﬂexibility.

The comparison of various buck–boost converters from differ-

ent points of view is shown in Table I. Among these converters,

although the SEPIC is not the best from the views of efﬁciency

and cost, it still has the merits of noninverting polarity, easy-to-

drive switch, and low input-current pulsating for high-precise

MPPT that makes its integral characteristics suitable for the

low-power PV charger system. This paper will investigate the

SEPIC with the PV module input and the peak-current-mode

control that was seldom presented in previous studies. The

small-signal model of such a SEPIC will be derived, and upon

which, the PV voltage controller and the MPPT controller will

be designed.

Another important issue of the PV charger system is the

power balance control that was faced commonly for multiple-

source system, such as fuel-cell hybrid vehicle [14], hybrid

PV and fuel-cell power system [15], hybrid wind, solar, and

distributed-generation power system[16], and so on. For the PV

charger system, MPPT and battery charging must be coopera-

tive, and the load demand must be considered simultaneously

such that the PV power can be utilized effectively and the

battery is suitably charged. Jiang and Dougal [17] treat this

issue as a multiobjective problem. It classiﬁes the system into

various states based on the operating conditions of the PV

module, the battery, and the load. By judging the state and

setting the related control goal, the power will be balanced to

satisfy the MPPT control and battery charging requirement.

However, the multiobjective control algorithm requires sense

of many states and sophisticated state judgment, and thus, needs

software programming. This paper will present a power balance

control method that achieves the same functions as that in [17]

but with only a simple control circuit. In addition, the proposed

power balance control can also be extended to the system with

more sources like those shown in [14]–[16]. Furthermore, all

power sources can be utilized with the preset priority based on

their importance to the system.

An 80-W prototype charger system implemented with some

analog circuits and Matlab real-time control is designed and

0278-0046/$26.00 © 2009 IEEE

Authorized licensed use limited to: Mepco Schlenk Engineering College. Downloaded on October 10, 2009 at 04:23 from IEEE Xplore. Restrictions apply.

CHIANG et al.: MODELING AND CONTROL OF PV CHARGER SYSTEM WITH SEPIC CONVERTER 4345

TABLE I

COMPARISON OF VARIOUS BUCK–BOOST CONVERTERS

Fig. 1. Circuit conﬁguration of the proposed PV charger.

built. The effectiveness of the proposed system is proved with

some PSIM [18] simulation and experimental results.

II. CIRCUIT CONFIGURATION AND POWER BALANCE

CONTROL OF THE CHARGER

The circuit conﬁguration of the proposed PV charger is

shown in Fig. 1. The SEPIC converter employs the peak-

current-mode control with an outer PV voltage regulating loop,

where the voltage command (V

∗

p

) is generated by combining

the MPPT control loop and the battery charging loop. The com-

bination of MPPT and charging control is for instantaneously

balancing the system power to charge the battery with three

stages, namely, constant-current, constant-voltage, and ﬂoating-

charge stages.

Based on the PV power generation, the battery SOC, and

the load condition, the designed controller shown in Fig. 1

can operate the system in three operation modes, as shown

in Fig. 2. Fig. 2(a) shows the discharging mode wherein the

available maximum PV power is less than the load power. The

insufﬁcient power will be automatically supplied by the dis-

charge of the battery. Fig. 2(b) shows the partial charging mode

wherein the available maximum PV power is larger than the

load power and the excessive power will charge the battery, but

the charging current is still less than the preset charge current

command (I

∗

b

). In the aforementioned two modes, because the

battery current (I

b

) in Fig. 1 cannot reach its current command

(I

∗

b

), the signal generated by the battery current controller that

is a proportional and integral (PI) controller will go positive

and be limited to be zero. It results that the voltage command

(V

∗

p

) is determined completely by the MPPT controller, and

thus, the PV module is operated in the MPPT point, as shown

in Fig. 2(a) and (b). As the available peak power of the PV

module is larger than the battery charging and load requirement,

the battery current in Fig. 1 will reach its command (I

∗

b

),

and the signal generated by the battery current controller will

go negative and will now add voltage to increase the voltage

command generated by the MPPT controller. As a result,

Authorized licensed use limited to: Mepco Schlenk Engineering College. Downloaded on October 10, 2009 at 04:23 from IEEE Xplore. Restrictions apply.

4346 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 56, NO. 11, NOVEMBER 2009

Fig. 2. Operation modes of the charger system. (a) Discharging mode.

(b) Partial charging mode. (c) Charging mode.

the PV module will discard the MPPT because the voltage

command (V

∗

p

) is shifted to a higher level than the MPPT

voltage, and ﬁnally, the generated PV power will balance the

load and charging requirement in the charging mode shown in

Fig. 2(c).

The charging controller is a PI controller. The limiter behind

the charging controller will set the charging current command

(I

∗

b

) to be a maximum charging current level as the battery

voltage (V

b

) has not reached its maximum charged voltage

command (V

∗

b

). In this case, and if the power condition is

sufﬁcient as Fig. 2(c), the system will operate in the constant-

current charge stage. As the battery voltage approximately

reaches the voltage command (V

∗

b

), the limiter will enter the

linear region, and the charging current command (I

∗

b

) will

reduce. This stage is called the constant-voltage charge stage.

Finally, as the battery voltage reaches the voltage command

(V

∗

b

) and the limiter output (I

∗

b

) is reduced to be approxi-

mately zero, the battery is in the ﬂoating-charge stage, i.e., the

fully SOC.

Through the proposed control arrangement, the PV module

will operate at MPPT, and the MPPT is discarded only when

the available PV power is larger than the total power for battery

charge and supplying the load. This is the most effective way

for utilizing the PV power. In addition, the battery will not

be overcharged and will stay at a high SOC voltage level if

the PV power is enough. On the other hand, to guarantee no

overdischarge of the battery as the PV power is insufﬁcient, a

load switch controlled with a hysteretic comparator is used in

Fig. 1 to disconnect the load as the battery voltage is lower than

a low SOC level and reconnect the load if the battery voltage is

larger than a safety level (V

∗

bL

) higher than the low SOC level.

The proposed power balance control method shown in Fig. 1

can be easily extended to the system with more sources. Fig. 3

shows the extension that two PVcharge circuits charge the same

battery, and the related operation modes are shown in Fig. 4. By

passing the signal (V

L

) generated by the positive limiter to the

negative limiter, the V

L

signal can be distributed to be two as

V

L1

and V

L2

. These two signals are then used to correct the

signals generated by both MPPT controllers to produce the PV

voltage commands for both chargers as that done in Fig. 1. If

the positive limiter’s output is zero, then both MPPT controllers

will not be affected by the charge controller, and both PV

modules will be at MPPT. The operation modes shown in

Fig. 4(a) and (b) both belong to this case. If the positive limiter’s

output is negative and is not limited by the negative limiter, V

L2

will be zero, and V

L1

will shift the MPPT of PV module 1.

Therefore, PV module 2 is at MPPT, but PV module 1

is not, as shown in Fig. 4(c). If the positive limiter’s output is

negative and limited by the negative limiter, V

L2

will not be

zero. V

L1

and V

L2

will shift the MPPT of both PV modules.

If the level of V

L1

exactly shifts PV module 1 to be the open-

circuit voltage, then the total power is only contributed by PV

module 2, as shown in Fig. 4(d). The power balance control

method shown in Fig. 3 can also be applied to the system with

more sources just by distributing the signal V

L

. It allows the

sources with different power characteristics. In addition, the

power source utilization can be assigned with priority, as done

in Fig. 3, where PV module 2 will be utilized ﬁrst and then PV

module 1. This feature is not proposed by all previous studies

[14]–[17].

III. MODELING AND CONTROLLER

DESIGN OF CHARGER

SEPIC is a buck–boost-derived converter that possesses a

right-half-plane (RHP) zero in the continuous conduction mode

(CCM), even with the peak-current-mode control in the output-

voltage regulation mode [19]. Following the averaged switch

modeling technique presented in [19], the control-to-output

transfer function of the SEPIC in CCM with output-voltage

regulation can be derived from Fig. 1 to be the following form:

˜

V

B

˜

D

¸

¸

¸

¸

¸

˜

V

P

=

˜

V

Cs

=0

=

V

P

−V

B

D

2

_

1 −s

DL

D

R

L

_

_

1 +s

L

D

2

R

L

+s

2

LC

o

D

2

_ (1)

where D is the duty ratio of the switch, with D

= 1 −D.

Here, for deriving (1) from Fig. 1, the battery is changed to be

an output-ﬁlter capacitor C

o

, the input voltage V

p

is changed

to be a constant-voltage source, and the inductors are set to

be L

1

= L

2

= L for simpliﬁcation. Equation (1) possesses

an RHP zero located at D

**RL/(DL). Opposite to output-
**

voltage regulation as the conventional converter, the proposed

PV charger regulates the input voltage. The CCM operation is

preferred here for reducing the input-current ripple and reduced

the switch-current stress. As will be seen in the following,

opposite to the output-voltage capacitor charged by the diode

Authorized licensed use limited to: Mepco Schlenk Engineering College. Downloaded on October 10, 2009 at 04:23 from IEEE Xplore. Restrictions apply.

CHIANG et al.: MODELING AND CONTROL OF PV CHARGER SYSTEM WITH SEPIC CONVERTER 4347

Fig. 3. Proposed power balance control method applied for two PV charger systems.

current of the output regulation structure, there is no RHP zero

problemof the proposed PVcharger because the input capacitor

current is charge controlled with the input inductor current,

and the input is a PV current source for the input regulation

structure. In the following, the analysis and design are based on

the CCM operation mode.

A. Design of Power Circuit

Referring to Fig. 1, the voltage transfer ratio of the SEPIC in

CCM is [20]

M =

V

B

V

p

=

D

1 −D

(2)

where D is the duty ratio of the switch. If the input inductor is

designed based on the idea that the SEPIC is operated in CCM

within the prescribed power range, then the input inductor L

1

will satisfy the following relation:

Δi

L1,pp

= 2I

p,min

=

2P

pv,min

V

p

=

V

p

L

1

D

max

T

s

(3)

where Δi

L1,pp

is the peak-to-peak value of the input inductor

current. P

pv,min

is the prescribed minimum PV power. I

p,min

is

the PV module current corresponding to P

pv,min

. The inductor

L

2

is designed to couple with L

1

as a ﬂyback transformer to

reduce the volume. The turn ratio is one (L

1

= L

2

) to make

both inductors enter into CCM simultaneously. Substituting (2)

into (3) results in

L

1

= L

2

=

V

2

B

2M(M + 1)P

pv,min

f

s

. (4)

The input-voltage ripple is caused by the charge and dis-

charge of the capacitor C

s

, as well as the ripple caused by

the equivalent series resistance (ESR) of the capacitor. The

electrolytic capacitor having nonnegligible ESR is adopted

here, and the input-voltage ripple is caused mainly by it at the

adopted switching frequency (40 kHz). As a result, the input

capacitor is selected based on the ESR value (R

e

) and the

voltage ripple demand (ΔV

p

) as [20]

R

e

=

ΔV

p

Δi

L1,pp

=

ΔV

p

2I

p,min

. (5)

Once ESR is determined, the capacitance (C

p

) can be de-

signed from the datasheet of the capacitor. The voltage ripple is

speciﬁed with the precision of the MPPT. A larger ripple will

render a larger ﬂuctuation around the MPPT operation point.

As for the coupling capacitor C

s

, it has to ﬂow the maximum

input current pulse, and a small ESR is required for reducing

the loss. In this paper, it is selected to be the same as the input

capacitor C

p

.

It should be noted that if the switching frequency of the

converter is high enough such that the capacitance of the

input capacitor can be reduced signiﬁcantly, the tantalum and

the ceramic capacitors with lower ESR can be adopted cost

effectively. The design of capacitor can then be changed to be

with charge and discharge of the capacitor but not with the ESR.

B. Modeling and Control of SEPIC

The equivalent circuit of the PV cell is shown in Fig. 5

[21], where R

sh

and R

s

are the intrinsic shunt and series

resistances of the cell, respectively. Usually, the value of R

sh

is very large, and that of R

s

is very small; hence, they may

be neglected to simplify the analysis. PV cells are grouped in

larger units called PV modules which are further interconnected

Authorized licensed use limited to: Mepco Schlenk Engineering College. Downloaded on October 10, 2009 at 04:23 from IEEE Xplore. Restrictions apply.

4348 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 56, NO. 11, NOVEMBER 2009

Fig. 4. Operation modes of two PV charger systems. (a) Discharging mode.

(b) Partial charging mode. (c) Charging mode with PV module 1 not at MPPT.

(d) Charging mode with two PV modules not at MPPT.

in a parallel–series conﬁguration to form PV arrays. The PV

array mathematical model can be represented as follows [22]:

I

p

= n

p

I

ph

−n

p

I

sat

_

exp

_

q

KAT

V

p

n

s

_

−1

_

(6)

where I

p

is the PV module array output current (in amperes),

V

p

is the PV array output voltage, n

s

is the number of modules

connected in series, n

p

is the number of modules connected

in parallel, q is the charge of an electron, K is Boltzmann’s

constant, A is the p-n junction ideality factor, T is the cell

temperature (in kelvins), and I

sat

is the cell reverse saturation

current.

The peak-current-mode control with input PV voltage reg-

ulation of the SEPIC is shown in Fig. 6(a), where the PV

module is represented as a current source modeled by (6), with

n

s

= n

p

= 1. The PV input power is

P

pv

= V I

p

. (7)

Fig. 5. Equivalent circuit of the PV cell.

Fig. 6. Feedback control of the SEPIC.

Under a ﬁxed insolation condition, a small perturbation of (7)

can be found as

¯

P

pv

=

¯

V I

p

+

¯

I

p

V. (8)

Equation (8) can be rearranged to represent the small-signal

expression of the PV module current

¯

I

p

=

−I

p

V

¯

V +

¯

P

pv

V

. (9)

One can obtain the input capacitor current from Fig. 6(a) as

C

p

dV

dt

= I

p

−I

L1

. (10)

The small-signal expression of (10) is

C

p

d

¯

V

dt

=

¯

I

p

−

¯

I

L1

. (11)

Substituting (9) into (11) yields

C

p

d

¯

V

dt

=

−I

p

V

¯

V +

¯

P

pv

V

−

¯

I

L1

. (12)

The transfer function of the inductor current I

L1

to the voltage

V can be found from (12) as

V (s)

I

L1

(s)

¸

¸

¸

¸

˜

P

pv

=0

= −

R

p

1 +sC

p

R

p

, R

p

=

V

I

p

≈

V

p

I

p

(13)

where R

p

is the equivalent load resistor seen by the PV module.

The relation between V and V

p

can be found by considering the

Authorized licensed use limited to: Mepco Schlenk Engineering College. Downloaded on October 10, 2009 at 04:23 from IEEE Xplore. Restrictions apply.

CHIANG et al.: MODELING AND CONTROL OF PV CHARGER SYSTEM WITH SEPIC CONVERTER 4349

ESR of the capacitor as

¯

V +R

e

C

p

d

¯

V

dt

=

¯

V

p

. (14)

The transfer function fromI

L1

to V

p

can be found by combining

(13) and the transfer function of (14) as

H(s) =

V

p

(s)

I

L1

(s)

¸

¸

¸

¸

¯ P

pv

=0

= −

R

p

(1 +sC

p

R

e

)

1 +sC

p

R

p

. (15)

Equation (15) is a ﬁrst-order system that includes a pole and

a zero. The pole is determined by the capacitor C

p

and the

equivalent load resistor R

p

. As compared with that in (1), the

zero here is determined by C

p

and the ESR value R

e

that is a

left-half-plane (LHP) zero. The UC3846 control IC is adopted

for realizing the peak-current-mode control. The switch current

is sensed with the resistor R

s

. A low-pass ﬁlter formed by

R

f

−C

f

is added to reduce the leading-edge spike caused by

the discharge of the internal capacitor of the switch and the

diode recovery current. The slope compensation signal is added

to the sensed current to prevent the subharmonic oscillation, as

the duty cycle is larger than 0.5. The slope compensation signal

is provided by the ramp signal coupled with a capacitor C

d

.

The level of compensation is adjusted with the ratio of resistors

R

c1

and R

c2

. The compensated current signal is ampliﬁed three

times and then compared with the voltage error signal V

ea

of

the voltage controller to reset the ﬂip-ﬂop which is set pulse by

pulse by the clock signal v

CLK

and to ﬁnally determine the duty

ratio of the switch. Because the peak current of the inductor L

1

is the same as that of the switch, the transfer function from V

ea

to I

L1

can be found as

K =

I

L1

(s)

V

ea

(s)

=

1

3R

s

_

1 +

m

a

m

1

_ (16)

where m

1

and m

a

are the slopes of the sensed current and

the compensated signal, respectively. For assuring a stable

current loop under any PV voltage, m

a

/m

1

≥ 0.5 is chosen.

Combining (15) and (16), the transfer function of the voltage

loop can be derived as

v

p

(s)

V

ea

(s)

= P(s) = k

v

kH(s) = −

k

v

R

e

3R

s

_

1+

m

a

m

1

_

_

s +

1

C

p

R

e

_

s +

1

C

p

R

p

(17)

where v

p

is the sensed PV voltage (V

p

). The voltage sensing

factor is k

v

. The negative gain in (17) desires a positive-

feedback control of the input voltage. The low-pass ﬁlter, as

in the following, is adopted as the compensator of the voltage

controller

G

v

(s) =

V

c

V

o

=

1

R

1

C

v

s +

1

C

v

R

2

. (18)

Its pole is designed to cancel the zero of (17), i.e., R

2

C

v

=

R

e

C

p

. Considering that the gain and the pole of (17) will drift

Fig. 7. Adopted incremental conductance MPPT control algorithm.

with the PV voltage, the gain R

2

/R

1

of the error ampliﬁer

should be designed to satisfy the stability and provide a suitable

bandwidth of the voltage loop.

C. MPPT Controller Design

The proposed MPPT controller tracks the peak power of the

PV module based on the power–voltage (P−V

p

) characteristic

shown in Fig. 7 and the incremental conductance algorithm

[22], [23]. In the positive-slope region (dP/dV

p

> 0), the op-

eration voltage is increased. On the other hand, in the negative-

slope region (dP/dV

p

< 0), the operation voltage is decreased.

The peak power point starting from any operating point will

be ﬁnally reached through a few steps of voltage adjustment.

In this paper, for increasing the tracking speed as well as the

precision, the voltage adjusting speed is slope dependent. In the

positive-slope region, the adjusting speed will be slower than

the negative-slope region because the positive slope is smaller

than the negative slope in amplitude. The adjusting speed will

be slowed down near the peak power point. This is effective

to prevent the tracking oscillation near the peak power point

to increase the MPPT precision. The aforementioned MPPT

control algorithm implemented with the Matlab real-time con-

trol (the details will be described in Section IV) is shown in

Fig. 8. For avoiding error adjustment caused by the voltage and

current ripples, the voltage and power are processed using a

fourth moving-average ﬁlter before the slope calculation. The

slope ΔP/ΔV

p

is calculated and then multiplied by a gain

(Gain 1) and limited by a limiter (Limiter 1) to ﬁnd the direction

and set the voltage change in each step. The direction value is

accumulated with a memory and then scaled with Gain 2 to ﬁt

the required voltage range. To make the MPPT start from the

open-circuit voltage, the Init voltage nearing the open-circuit

voltage is adopted to set the initial voltage command before the

start of the converter. The voltage command is ﬁnally limited

by Limiter 2 to limit the attainable PV voltage.

IV. IMPLEMENTATION AND VERIFICATION

An 80-W PV charger system is designed and built and tested

with a 75-W Siemens SP75 PV module in this paper. The

speciﬁcation of the SEPIC is listed as follows:

V

p

=10 ∼ 22 V V

B

= 12 V f

s

= 40 kHz

k

v

=0.1 R

s

= 0.05 V

B

= 12 V

P

pv,min

=10 W I

p,min

= 0.5 A ΔV

p

= 0.2 V. (19)

Authorized licensed use limited to: Mepco Schlenk Engineering College. Downloaded on October 10, 2009 at 04:23 from IEEE Xplore. Restrictions apply.

4350 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 56, NO. 11, NOVEMBER 2009

Fig. 8. Proposed MPPT control algorithm implemented with the Matlab real-time control.

The PV voltage may be lower or higher than the battery voltage,

and the duty ratio range can be calculated with (2) as

M =

_

V

B

V

p

_

max

=

12

10

=

D

max

1 −D

max

, D

max

= 0.545.

(20)

The inductor value is then calculated with (4) as L

1

= L

2

=

68 μF. With the speciﬁcation of ΔV

p

and (5), the required R

e

=

0.2 Ω. Accordingly, the capacitor is found as

C

p

=

40 ×10

−6

R

e

= 200 μF (21)

where 40 ×10

−6

can be determined from the capacitor

datasheet. The typical value is 30−80 ×10

−6

. The real value

of C

p

is 220 μF in this paper. The coupling capacitor C

s

has

the same value as the C

p

.

The voltage controller in (18) is designed with (17). How-

ever, the parameters in (17) are varied with the operating point

of the PV module. For considering this factor, Fig. 9 shows ﬁve

P−V curves of the SP75 module corresponding to the cases of

PV peak powers of 75, 60, 40, 30, and 10 W, respectively. Three

lines on the right side of each curve are chosen to calculate the

equivalent output resistor value (R

p

) of the PV module. The

resistor value is determined based on the slope of the V −I

curve. The voltage controller in (18) is then designed based

on (17) and these resistor values. The MPPT in 75 W (R

p1

)

is adopted as the nominal case. The design procedure is as

follows: 1) Set the crossover frequency (f

c

) to be one-tenth

of the switching frequency (f

s

); 2) ﬁnd the value R

2

/R

1

=

1/|KH(f

c

)|; 3) assign R

1

and then calculate R

2

with the ratio

R

2

/R

1

; and 4) set the pole to be R

2

C

v

= C

p

R

e

and then

ﬁnd C

v

. The designed result of G

v

is shown in Fig. 10, in

which the bode plots of P(s) in (17) with various values of R

p

and the closed-loop response (G

v

P) of the nominal case are

also provided. The closed-loop response of the nominal case

Fig. 9. Five P−V curves corresponding to the cases of PV peak power of 75,

60, 40, 30, and 10 W, respectively.

satisﬁes the aforementioned design. It can be examined that the

closed-loop response for various cases is also stable.

The whole system is veriﬁed with the PSIM simulation ﬁrst

before implementation. It is constructed based on the aforeasaid

design. For verifying the circuit modeling, the open-loop re-

sponse of the voltage loop in the aforementioned nominal case

is ac swept with the PSIM simulation, as shown in Fig. 11(a), in

which the PVmodule is modeled as a voltage-controlled current

source possessing the prescribed SP75 module V −I curve. The

current command is set to satisfy the 75-W input. Fig. 11(b)

shows the sweep of the voltage loop that is closed further.

The sweep results are very close to the theoretic results shown

in Fig. 10, conﬁrming the previous modeling and controller

design.

The implementation of the whole system in the developing

stage is shown in Fig. 12. The MPPT controller of the SEPIC

charger is implemented with the Matlab real-time control on

PC; other circuits of the SEPIC are implemented with the

Authorized licensed use limited to: Mepco Schlenk Engineering College. Downloaded on October 10, 2009 at 04:23 from IEEE Xplore. Restrictions apply.

CHIANG et al.: MODELING AND CONTROL OF PV CHARGER SYSTEM WITH SEPIC CONVERTER 4351

Fig. 10. Open- and closed-loop responses of the voltage.

Fig. 11. AC sweep of the voltage loop. (a) Open loop. (b) Closed loop.

analog circuit. Fig. 13 shows the measured MPPT response at

50 W. The waveforms of the PVmodule voltage and current and

the P−V trace prove that the response of the proposed MPPT

controller is fast. The precision of the MPPT controller is found

to be up to 99.2% even at low PV power.

The measured waveforms for demonstrating the power bal-

ance feature are shown in Fig. 14. Fig. 14(a) shows that the load

Fig. 12. Implementation of the whole system.

Fig. 13. MPPT response of the charge.

power (84 W) is larger than the available PV power (MPPT,

60 W), the insufﬁcient power is supplied by the battery (24 W),

and the PV module is operated at MPPT. Fig. 14(b) shows that

the load power is small and that the PV module discards the

MPPT (20 W) to keep the battery charged in constant current

Authorized licensed use limited to: Mepco Schlenk Engineering College. Downloaded on October 10, 2009 at 04:23 from IEEE Xplore. Restrictions apply.

4352 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 56, NO. 11, NOVEMBER 2009

Fig. 14. Measured waveforms for demonstrating the power balance feature.

(a) Load power is larger than the available PV power. (b) Sum of the load power

and the charge power is less than the peak power of the PV module.

(0.6 A). These all demonstrate that the proposed power balance

control is effective.

V. CONCLUSION

This paper has presented a PV charger implemented with the

SEPIC converter. The system has been proved to be effective in

the MPPT and power balance control. The proposed modeling

method of the converter with the PV module input and peak-

current-mode control, the adaptive MPPT control method, as

well as the power balance control method can also be applied to

the charger with other types of converter. The MPPT controller

was implemented with the Matlab real-time control in this

paper, and it will be changed to be implemented with the mi-

croprocessor or DSP and integrated with the voltage controller

and PWM to make the system more practical in the future.

REFERENCES

[1] M. Rogol, S. Doi, and A. Wilkinson, “Sun screen: Investment opportu-

nities in solar power,” Solar Power Sector Outlook, Jul. 7, 2004, CLSA

Asia-Paciﬁc Markets.

[2] German Advisory Council on GlobalChange, 2003.

[3] S. B. Kjaer, J. K. Pedersen, and F. Blaabjerg, “A review of single-phase

grid-connected inverters for photovoltaic modules,” IEEE Trans. Ind.

Appl., vol. 41, no. 5, pp. 1292–1306, Sep./Oct. 2005.

[4] R. A. Mastromauro, M. Liserre, and A. Dell’Aquila, “Study of the effects

of inductor nonlinear behavior on the performance of current controllers

for single-phase PV grid converters,” IEEE Trans. Ind. Electron., vol. 55,

no. 5, pp. 2043–2052, May 2008.

[5] Y. K. Lo, T. P. Lee, and K. H. Wu, “Grid-connected photovoltaic system

with power factor correction,” IEEE Trans. Ind. Electron., vol. 55, no. 5,

pp. 2224–2227, May 2008.

[6] H. Koizumi, T. Mizuno, T. Kaito, Y. Noda, N. Goshima, M. Kawasaki,

K. Nagasaka, and K. Kurokawa, “A novel microcontroller for grid-

connected photovoltaic systems,” IEEE Trans. Ind. Electron., vol. 53,

no. 6, pp. 1889–1897, Dec. 2006.

[7] W. Xiao, N. Ozog, and W. G. Dunford, “Topology study of photovoltaic

interface for maximum power point tracking,” IEEE Trans. Ind. Electron.,

vol. 54, no. 3, pp. 1696–1704, Jun. 2007.

[8] R. J. Wai, W. H. Wang, and C. Y. Lin, “High-performance stand-alone

photovoltaic generation system,” IEEE Trans. Ind. Electron., vol. 55,

no. 1, pp. 240–250, Jan. 2008.

[9] F. Liu, S. Duan, F. Liu, B. Liu, and Y. Kang, “A variable step size INC

MPPT method for PV systems,” IEEE Trans. Ind. Electron., vol. 55, no. 7,

pp. 2622–2628, Jul. 2008.

[10] D. Sera, R. Teodorescu, J. Hantschel, and M. Knoll, “Optimized max-

imum power point tracker for fast-changing environmental conditions,”

IEEE Trans. Ind. Electron., vol. 55, no. 7, pp. 2629–2637, Jul. 2008.

[11] J. H. R. Enslin and D. B. Snyman, “Combined low-cost, high-efﬁcient

inverter, peak power tracker and regulator for PV applications,” IEEE

Trans. Power Electron., vol. 6, no. 1, pp. 73–82, Jan. 1991.

[12] E. Koutroulis and K. Kalaitzakis, “Novel battery charging regulation sys-

tem for photovoltaic applications,” Proc. Inst. Elect. Eng.—Elect. Power

Appl., vol. 151, no. 2, pp. 191–197, Mar. 2004.

[13] D. Adar, G. Rahav, and S. Ben-Yaakov, “A uniﬁed behavioral average

model of SEPIC converters with coupled inductors,” in Proc. IEEE PESC,

1997, pp. 441–446.

[14] H. Yun, Z. Zhong, Z. Sun, and G. Wan, “Research on power balance

strategy of fuel cell vehicle powertrain,” in Proc. IEEE ICVES, 2006,

pp. 388–393.

[15] Z. Jiang, “Power management of hybrid photovoltaic—Fuel cell power

systems,” in Proc. IEEE Power Eng. Soc. Gen. Meeting, 2006, pp. 1–6.

[16] M. H. Nehrir, C. Wang, and S. R. Guda, “Alternative energy distributed

generation: Need for multi-source operation,” in Proc. 38th NAPS, 2006,

pp. 547–551.

[17] Z. Jiang and R. A. Dougal, “Multiobjective MPPT/charging controller

for standalone PV power systems under different insolation and load

conditions,” in Conf. Rec. IEEE IAS Annu. Meeting, 2004, pp. 1154–1160.

[18] PSIM simulation software, Powersim Inc.

[19] R. W. Erickson, Fundamental of Power Electronics. Norwell, MA:

Kluwer, 1997.

[20] A. I. Pressman, Switching Power Supply Design. New York:

Mcgraw-Hill.

[21] S. E. Mineiro, Jr., S. Daher, F. L. M. Antunes, and C. M. T. Cruz, “Photo-

voltaic system for supply public illumination in electrical energy demand

peak,” in Proc. IEEE APEC, 2004, pp. 1501–1506.

[22] L. Wu, Z. Zhao, and J. Liu, “A single-stage three-phase grid-connected

photovoltaic system with modiﬁed MPPT method and reactive power

compensation,” IEEE Trans. Energy Convers., vol. 22, no. 4, pp. 881–

886, Dec. 2007.

[23] E. Koutroulis, K. Kalaitzakis, and N. C. Voulgaris, “Development of a

microcontroller-based, photovoltaic maximum power point tracking con-

trol system,” IEEE Trans. Power Electron., vol. 16, no. 1, pp. 46–54,

Jan. 2001.

S. J. Chiang was born in Tainan, Taiwan in

1965. He received the B.S. and Ph.D. degrees

in electrical engineering from National Tsing Hua

University, Hsinchu, Taiwan, in 1987 and 1994,

respectively.

From 1995 to 2000, he was an Associate Professor

with the Department of Electrical Engineering, Na-

tional United University, Miaoli, Taiwan, where he

has been a Professor since 2001. His current research

interests include power electronics, motor drives, and

control systems.

Authorized licensed use limited to: Mepco Schlenk Engineering College. Downloaded on October 10, 2009 at 04:23 from IEEE Xplore. Restrictions apply.

CHIANG et al.: MODELING AND CONTROL OF PV CHARGER SYSTEM WITH SEPIC CONVERTER 4353

Hsin-Jang Shieh (M’02) received the B.S. and Ph.D.

degrees in electrical engineering from National Cen-

tral University, Chung-Li, Taiwan, in 1992 and 1997,

respectively.

From 1997 to 2002, he was with the Mechani-

cal Industry Research Laboratories, Industrial Tech-

nology Research Institute, Hsinchu, Taiwan, as a

Researcher. Since August 2002, he has been with

the Department of Electrical Engineering, National

Dong Hwa University, Hualien, Taiwan, and is cur-

rently an Associate Professor. His research interests

include drive and control of piezoelectric mechanisms, power converters,

photovoltaic systems, and control theory applications.

Dr. Shieh is a member of the Taiwan Power Electronics Association and the

Chinese Automatic Control Society of Taiwan. He was the recipient of the 2007

Taiwain Power Electronics Conference Best Paper Award.

Ming-Chieh Chen was born in Taichung, Taiwan in

1980. He received the B.S. degree in electrical en-

gineering from National United University, Miaoli,

Taiwan, in 2004 and the M.S. degree in electrical

engineering from National Dong Hwa University,

Hualien, Taiwan, in 2006.

He is currently an Electronic Engineer with FSP

Technology Inc., Taoyuan, Taiwan. His research in-

terest includes power electronics.

Authorized licensed use limited to: Mepco Schlenk Engineering College. Downloaded on October 10, 2009 at 04:23 from IEEE Xplore. Restrictions apply.

where the voltage command (Vp∗ ) is generated by combining the MPPT control loop and the battery charging loop. because the battery current (Ib ) in Fig. It results that the voltage command (Vp∗ ) is determined completely by the MPPT controller. and ﬂoatingcharge stages. 2(a) shows the discharging mode wherein the available maximum PV power is less than the load power. 2009 at 04:23 from IEEE Xplore. and the signal generated by the battery current controller will go negative and will now add voltage to increase the voltage command generated by the MPPT controller. 2(b) shows the partial charging mode wherein the available maximum PV power is larger than the load power and the excessive power will charge the battery. but the charging current is still less than the preset charge current ∗ command (Ib ). 1. The effectiveness of the proposed system is proved with some PSIM [18] simulation and experimental results.CHIANG et al. As a result. 1 cannot reach its current command ∗ (Ib ). namely. 1 will reach its command (Ib ). constant-current. The combination of MPPT and charging control is for instantaneously balancing the system power to charge the battery with three stages. Restrictions apply. the designed controller shown in Fig. constant-voltage. Authorized licensed use limited to: Mepco Schlenk Engineering College. the signal generated by the battery current controller that is a proportional and integral (PI) controller will go positive and be limited to be zero. Fig. as shown in Fig. the battery SOC. As the available peak power of the PV module is larger than the battery charging and load requirement. . The SEPIC converter employs the peakcurrent-mode control with an outer PV voltage regulating loop. 2(a) and (b). 1. C IRCUIT C ONFIGURATION AND P OWER B ALANCE C ONTROL OF THE C HARGER The circuit conﬁguration of the proposed PV charger is shown in Fig. Circuit conﬁguration of the proposed PV charger.: MODELING AND CONTROL OF PV CHARGER SYSTEM WITH SEPIC CONVERTER 4345 TABLE I COMPARISON OF VARIOUS BUCK–BOOST CONVERTERS Fig. In the aforementioned two modes. 1 can operate the system in three operation modes. built. II. Based on the PV power generation. ∗ the battery current in Fig. the PV module is operated in the MPPT point. Fig. and thus. as shown in Fig. 2. The insufﬁcient power will be automatically supplied by the discharge of the battery. Downloaded on October 10. and the load condition.

but PV module 1 is not. Opposite to outputvoltage regulation as the conventional converter. In addition. and ﬁnally. the generated PV power will balance the load and charging requirement in the charging mode shown in Fig. opposite to the output-voltage capacitor charged by the diode Authorized licensed use limited to: Mepco Schlenk Engineering College. Here. 1 to be the following form: ˜ VB ˜ D DL 1 − s D RL VP − VB L D2 o 1 + s D 2 RL + s2 LC2 D = ˜ ˜ VP =VCs =0 (1) where D is the duty ratio of the switch. 4. Restrictions apply. and the inductors are set to be L1 = L2 = L for simpliﬁcation.e. the battery is changed to be an output-ﬁlter capacitor Co . The power balance control method shown in Fig. Following the averaged switch modeling technique presented in [19]. 1. and both PV modules will be at MPPT.4346 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS. It allows the sources with different power characteristics. with D = 1 − D. Operation modes of the charger system. If the positive limiter’s output is negative and is not limited by the negative limiter. The limiter behind the charging controller will set the charging current command ∗ (Ib ) to be a maximum charging current level as the battery voltage (Vb ) has not reached its maximum charged voltage command (Vb∗ ). 2(c). As the battery voltage approximately reaches the voltage command (Vb∗ ). a load switch controlled with a hysteretic comparator is used in Fig. to guarantee no overdischarge of the battery as the PV power is insufﬁcient. the proposed PV charger regulates the input voltage. the limiter will enter the ∗ linear region. the PV module will discard the MPPT because the voltage command (Vp∗ ) is shifted to a higher level than the MPPT voltage. 1. Equation (1) possesses an RHP zero located at D RL/(DL). VL2 will not be zero. By passing the signal (VL ) generated by the positive limiter to the negative limiter. If the positive limiter’s output is negative and limited by the negative limiter. In addition. Finally. i. 2. 3 can also be applied to the system with more sources just by distributing the signal VL . 1 to disconnect the load as the battery voltage is lower than a low SOC level and reconnect the load if the battery voltage is ∗ larger than a safety level (VbL ) higher than the low SOC level. NOVEMBER 2009 Fig. The CCM operation is preferred here for reducing the input-current ripple and reduced the switch-current stress. and if the power condition is sufﬁcient as Fig. The operation modes shown in Fig. In this case. and the charging current command (Ib ) will reduce. the fully SOC. and the related operation modes are shown in Fig. 3. PV module 2 is at MPPT. 4(d). the control-to-output transfer function of the SEPIC in CCM with output-voltage regulation can be derived from Fig. VOL. the PV module will operate at MPPT. Through the proposed control arrangement. 1 can be easily extended to the system with more sources. even with the peak-current-mode control in the outputvoltage regulation mode [19]. 2(c). M ODELING AND C ONTROLLER D ESIGN OF C HARGER SEPIC is a buck–boost-derived converter that possesses a right-half-plane (RHP) zero in the continuous conduction mode (CCM). If the level of VL1 exactly shifts PV module 1 to be the opencircuit voltage. the power source utilization can be assigned with priority. (a) Discharging mode. This stage is called the constant-voltage charge stage. NO. then both MPPT controllers will not be affected by the charge controller. the system will operate in the constantcurrent charge stage. as done in Fig. and the MPPT is discarded only when the available PV power is larger than the total power for battery charge and supplying the load. the VL signal can be distributed to be two as VL1 and VL2 . and VL1 will shift the MPPT of PV module 1. This is the most effective way for utilizing the PV power. As will be seen in the following. If the positive limiter’s output is zero. Downloaded on October 10. for deriving (1) from Fig. Fig. the input voltage Vp is changed to be a constant-voltage source. 56. where PV module 2 will be utilized ﬁrst and then PV module 1. then the total power is only contributed by PV module 2. as shown in Fig. The charging controller is a PI controller. 2009 at 04:23 from IEEE Xplore. These two signals are then used to correct the signals generated by both MPPT controllers to produce the PV voltage commands for both chargers as that done in Fig. . VL2 will be zero. Therefore. 11. the battery will not be overcharged and will stay at a high SOC voltage level if the PV power is enough. (c) Charging mode. 3 shows the extension that two PV charge circuits charge the same battery. as the battery voltage reaches the voltage command ∗ (Vb∗ ) and the limiter output (Ib ) is reduced to be approximately zero. The proposed power balance control method shown in Fig. as shown in Fig. III. 4(a) and (b) both belong to this case. On the other hand. VL1 and VL2 will shift the MPPT of both PV modules. (b) Partial charging mode. 4(c).. This feature is not proposed by all previous studies [14]–[17]. the battery is in the ﬂoating-charge stage.

If the input inductor is designed based on the idea that the SEPIC is operated in CCM within the prescribed power range. as well as the ripple caused by Authorized licensed use limited to: Mepco Schlenk Engineering College. 2009 at 04:23 from IEEE Xplore. (4) The input-voltage ripple is caused by the charge and discharge of the capacitor Cs . and that of Rs is very small. It should be noted that if the switching frequency of the converter is high enough such that the capacitance of the input capacitor can be reduced signiﬁcantly. Ppv. the input capacitor is selected based on the ESR value (Re ) and the voltage ripple demand (ΔVp ) as [20] Re = ΔVp ΔVp = . The electrolytic capacitor having nonnegligible ESR is adopted here. Ip. Modeling and Control of SEPIC The equivalent circuit of the PV cell is shown in Fig.min . The voltage ripple is speciﬁed with the precision of the MPPT. . A larger ripple will render a larger ﬂuctuation around the MPPT operation point. The turn ratio is one (L1 = L2 ) to make both inductors enter into CCM simultaneously. the voltage transfer ratio of the SEPIC in CCM is [20] D VB = M= Vp 1−D (2) the equivalent series resistance (ESR) of the capacitor. ΔiL1. then the input inductor L1 will satisfy the following relation: ΔiL1.: MODELING AND CONTROL OF PV CHARGER SYSTEM WITH SEPIC CONVERTER 4347 Fig. Design of Power Circuit Referring to Fig. it is selected to be the same as the input capacitor Cp .min = 2Ppv. the tantalum and the ceramic capacitors with lower ESR can be adopted cost effectively.min (5) where D is the duty ratio of the switch. and a small ESR is required for reducing the loss. hence. A. the analysis and design are based on the CCM operation mode. there is no RHP zero problem of the proposed PV charger because the input capacitor current is charge controlled with the input inductor current. they may be neglected to simplify the analysis. As a result. the value of Rsh is very large.min fs . In this paper. Restrictions apply.pp 2Ip. 3.CHIANG et al.pp is the peak-to-peak value of the input inductor current. Downloaded on October 10. current of the output regulation structure. respectively. Substituting (2) into (3) results in L1 = L2 = 2 VB Once ESR is determined. 5 [21]. 1.min is the prescribed minimum PV power. it has to ﬂow the maximum input current pulse. and the input is a PV current source for the input regulation structure. Proposed power balance control method applied for two PV charger systems. where Rsh and Rs are the intrinsic shunt and series resistances of the cell.pp = 2Ip. B. the capacitance (Cp ) can be designed from the datasheet of the capacitor. As for the coupling capacitor Cs .min Vp = Dmax Ts Vp L1 (3) where ΔiL1. The design of capacitor can then be changed to be with charge and discharge of the capacitor but not with the ESR. The inductor L2 is designed to couple with L1 as a ﬂyback transformer to reduce the volume.min is the PV module current corresponding to Ppv. Usually. In the following. PV cells are grouped in larger units called PV modules which are further interconnected 2M (M + 1)Ppv. and the input-voltage ripple is caused mainly by it at the adopted switching frequency (40 kHz).

4348 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS. (d) Charging mode with two PV modules not at MPPT. (a) Discharging mode. (c) Charging mode with PV module 1 not at MPPT. The PV input power is Ppv = V Ip . (7) Substituting (9) into (11) yields Cp Ppv −Ip dV = V + − IL1 . 6(a) as Cp dV = Ip − IL1 . . K is Boltzmann’s constant. The PV array mathematical model can be represented as follows [22]: Ip = np Iph − np Isat exp q Vp KAT ns −1 (6) The small-signal expression of (10) is Cp dV = Ip − IL1 . 6(a). 56. ns is the number of modules connected in series. 11. VOL. Equivalent circuit of the PV cell. np is the number of modules connected in parallel. The peak-current-mode control with input PV voltage regulation of the SEPIC is shown in Fig. a small perturbation of (7) can be found as Ppv = V Ip + Ip V. dt (10) in a parallel–series conﬁguration to form PV arrays. V V (9) One can obtain the input capacitor current from Fig. The relation between V and Vp can be found by considering the Authorized licensed use limited to: Mepco Schlenk Engineering College. Ppv −Ip V + . NO. 2009 at 04:23 from IEEE Xplore. (8) Equation (8) can be rearranged to represent the small-signal expression of the PV module current Ip = Fig. Under a ﬁxed insolation condition. q is the charge of an electron. Operation modes of two PV charger systems. Feedback control of the SEPIC. dt (11) where Ip is the PV module array output current (in amperes). Restrictions apply. A is the p-n junction ideality factor. dt V V (12) The transfer function of the inductor current IL1 to the voltage V can be found from (12) as V (s) IL1 (s) =− Rp . Fig. with ns = np = 1. where the PV module is represented as a current source modeled by (6). 6. T is the cell temperature (in kelvins). 5. 4. (b) Partial charging mode. 1 + sCp Rp Rp = V Vp ≈ Ip Ip (13) ˜ Ppv =0 where Rp is the equivalent load resistor seen by the PV module. Downloaded on October 10. NOVEMBER 2009 Fig. Vp is the PV array output voltage. and Isat is the cell reverse saturation current.

Combining (15) and (16). The low-pass ﬁlter. in the negativeslope region (dP/dVp < 0). The compensated current signal is ampliﬁed three times and then compared with the voltage error signal Vea of the voltage controller to reset the ﬂip-ﬂop which is set pulse by pulse by the clock signal vCLK and to ﬁnally determine the duty ratio of the switch.min = 10 W Rs = 0. Authorized licensed use limited to: Mepco Schlenk Engineering College. The slope compensation signal is added to the sensed current to prevent the subharmonic oscillation. MPPT Controller Design The proposed MPPT controller tracks the peak power of the PV module based on the power–voltage (P −Vp ) characteristic shown in Fig.5. The negative gain in (17) desires a positivefeedback control of the input voltage. . the zero here is determined by Cp and the ESR value Re that is a left-half-plane (LHP) zero.e. IV. The direction value is accumulated with a memory and then scaled with Gain 2 to ﬁt the required voltage range. is adopted as the compensator of the voltage controller Gv (s) = Vc = R1 Cv . A low-pass ﬁlter formed by Rf −Cf is added to reduce the leading-edge spike caused by the discharge of the internal capacitor of the switch and the diode recovery current. the operation voltage is decreased. i. As compared with that in (1). In the positive-slope region (dP/dVp > 0).5 A ΔVp = 0. 7 and the incremental conductance algorithm [22]. In the positive-slope region.: MODELING AND CONTROL OF PV CHARGER SYSTEM WITH SEPIC CONVERTER 4349 ESR of the capacitor as V + Re Cp dV = Vp . Because the peak current of the inductor L1 is the same as that of the switch. respectively. the Init voltage nearing the open-circuit voltage is adopted to set the initial voltage command before the start of the converter. The switch current is sensed with the resistor Rs . 8. Adopted incremental conductance MPPT control algorithm.. the gain R2 /R1 of the error ampliﬁer should be designed to satisfy the stability and provide a suitable bandwidth of the voltage loop. Considering that the gain and the pole of (17) will drift VB = 12 V (19) Ip. 2009 at 04:23 from IEEE Xplore. For avoiding error adjustment caused by the voltage and current ripples. The adjusting speed will be slowed down near the peak power point. The speciﬁcation of the SEPIC is listed as follows: Vp = 10 ∼ 22 V VB = 12 V kv = 0. For assuring a stable current loop under any PV voltage. 1 + sCp Rp (15) Fig. This is effective to prevent the tracking oscillation near the peak power point to increase the MPPT precision. ma /m1 ≥ 0.5 is chosen. 7. Equation (15) is a ﬁrst-order system that includes a pole and a zero. the adjusting speed will be slower than the negative-slope region because the positive slope is smaller than the negative slope in amplitude. The pole is determined by the capacitor Cp and the equivalent load resistor Rp . Restrictions apply. I MPLEMENTATION AND V ERIFICATION An 80-W PV charger system is designed and built and tested with a 75-W Siemens SP75 PV module in this paper. The voltage command is ﬁnally limited by Limiter 2 to limit the attainable PV voltage. the transfer function of the voltage loop can be derived as 3Rs vp (s) = P (s) = kv kH(s) = − Vea (s) kv Re 1+ ma m 1 s+ 1 Cp Rp 1 Cp Re s+ (17) where vp is the sensed PV voltage (Vp ). as in the following.05 fs = 40 kHz ma m1 where m1 and ma are the slopes of the sensed current and the compensated signal. the operation voltage is increased. dt (14) The transfer function from IL1 to Vp can be found by combining (13) and the transfer function of (14) as H(s) = Vp (s) IL1 (s) =− Ppv =0 Rp (1 + sCp Re ) .2 V. To make the MPPT start from the open-circuit voltage. The slope ΔP/ΔVp is calculated and then multiplied by a gain (Gain 1) and limited by a limiter (Limiter 1) to ﬁnd the direction and set the voltage change in each step.1 Ppv. On the other hand. the voltage and power are processed using a fourth moving-average ﬁlter before the slope calculation. for increasing the tracking speed as well as the precision. R2 Cv = Re Cp . The voltage sensing factor is kv . In this paper. Vo s + Cv1R2 1 (18) Its pole is designed to cancel the zero of (17). The slope compensation signal is provided by the ramp signal coupled with a capacitor Cd . The UC3846 control IC is adopted for realizing the peak-current-mode control. the transfer function from Vea to IL1 can be found as K= IL1 (s) 1 = Vea (s) 3Rs 1 + (16) with the PV voltage.CHIANG et al. C. [23]. Downloaded on October 10. The peak power point starting from any operating point will be ﬁnally reached through a few steps of voltage adjustment. as the duty cycle is larger than 0.min = 0. The level of compensation is adjusted with the ratio of resistors Rc1 and Rc2 . The aforementioned MPPT control algorithm implemented with the Matlab real-time control (the details will be described in Section IV) is shown in Fig. the voltage adjusting speed is slope dependent.

The coupling capacitor Cs has the same value as the Cp . The voltage controller in (18) is then designed based on (17) and these resistor values.4350 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS. respectively. It is constructed based on the aforeasaid design. 11(b) shows the sweep of the voltage loop that is closed further. With the speciﬁcation of ΔVp and (5). The real value of Cp is 220 μF in this paper. The whole system is veriﬁed with the PSIM simulation ﬁrst before implementation.545.2 Ω. 60. However. 11. satisﬁes the aforementioned design. The resistor value is determined based on the slope of the V −I curve. in which the bode plots of P (s) in (17) with various values of Rp and the closed-loop response (Gv P ) of the nominal case are also provided. 10. The closed-loop response of the nominal case Fig. 40. 30. NO. Fig. VOL. respectively. Accordingly. 8. The implementation of the whole system in the developing stage is shown in Fig. the open-loop response of the voltage loop in the aforementioned nominal case is ac swept with the PSIM simulation. For considering this factor. 9 shows ﬁve P −V curves of the SP75 module corresponding to the cases of PV peak powers of 75. 3) assign R1 and then calculate R2 with the ratio R2 /R1 . and 4) set the pole to be R2 Cv = Cp Re and then ﬁnd Cv . The PV voltage may be lower or higher than the battery voltage. It can be examined that the closed-loop response for various cases is also stable. in which the PV module is modeled as a voltage-controlled current source possessing the prescribed SP75 module V −I curve. The current command is set to satisfy the 75-W input. The design procedure is as follows: 1) Set the crossover frequency (fc ) to be one-tenth of the switching frequency (fs ). 10 1 − Dmax Dmax = 0. 56. The designed result of Gv is shown in Fig. 30. and 10 W. 40. Three lines on the right side of each curve are chosen to calculate the equivalent output resistor value (Rp ) of the PV module. 2009 at 04:23 from IEEE Xplore. conﬁrming the previous modeling and controller design. 9. The voltage controller in (18) is designed with (17). the parameters in (17) are varied with the operating point of the PV module. 10. For verifying the circuit modeling. the required Re = 0. and 10 W. 11(a). as shown in Fig. Proposed MPPT control algorithm implemented with the Matlab real-time control. The typical value is 30−80 × 10−6 . The MPPT in 75 W (Rp1 ) is adopted as the nominal case. 2) ﬁnd the value R2 /R1 = 1/|KH(fc )|. 12. the capacitor is found as Cp = 40 × 10−6 = 200 μF Re (21) where 40 × 10−6 can be determined from the capacitor datasheet. (20) The inductor value is then calculated with (4) as L1 = L2 = 68 μF. and the duty ratio range can be calculated with (2) as M= VB Vp = max Dmax 12 = . Five P −V curves corresponding to the cases of PV peak power of 75. Downloaded on October 10. The MPPT controller of the SEPIC charger is implemented with the Matlab real-time control on PC. other circuits of the SEPIC are implemented with the Authorized licensed use limited to: Mepco Schlenk Engineering College. 60. Restrictions apply. . The sweep results are very close to the theoretic results shown in Fig. NOVEMBER 2009 Fig. Fig.

analog circuit. 14(b) shows that the load power is small and that the PV module discards the MPPT (20 W) to keep the battery charged in constant current Authorized licensed use limited to: Mepco Schlenk Engineering College. (a) Open loop. 12. The precision of the MPPT controller is found to be up to 99. the insufﬁcient power is supplied by the battery (24 W). . Fig. Open. The measured waveforms for demonstrating the power balance feature are shown in Fig. The waveforms of the PV module voltage and current and the P −V trace prove that the response of the proposed MPPT controller is fast. power (84 W) is larger than the available PV power (MPPT. AC sweep of the voltage loop. and the PV module is operated at MPPT. Fig. 14. Downloaded on October 10. Fig. Fig. 60 W). Implementation of the whole system. Restrictions apply.: MODELING AND CONTROL OF PV CHARGER SYSTEM WITH SEPIC CONVERTER 4351 Fig. 13. 11. Fig. (b) Closed loop. MPPT response of the charge.2% even at low PV power. 2009 at 04:23 from IEEE Xplore.and closed-loop responses of the voltage. 14(a) shows that the load Fig. 13 shows the measured MPPT response at 50 W. 10.CHIANG et al.

S. vol.” IEEE Trans. [11] J. Teodorescu. “Photovoltaic system for supply public illumination in electrical energy demand peak. pp. [6] H. motor drives. Ozog. Rec. [15] Z. Blaabjerg. IEEE PESC.” Proc.. 2007.” in Proc. Yun. no. “A variable step size INC MPPT method for PV systems. G.” Solar Power Sector Outlook. “Power management of hybrid photovoltaic—Fuel cell power systems. E. C. pp. Power Electron. [4] R. B. Power Appl.. Dougal. 4. Ind. pp. pp. Kalaitzakis. [8] R. 22. Switching Power Supply Design.6 A). and C. M. “Alternative energy distributed generation: Need for multi-source operation. He received the B. (0. vol. and control systems. 55. Lin. vol. pp. F. no. Kalaitzakis. H. Electron. [9] F. and C. R. 1–6. Meeting. 2009 at 04:23 from IEEE Xplore. 7. Sera. P.” in Proc. vol. G./Oct. 55. no. [2] German Advisory Council on GlobalChange. 2003. pp. Liu. vol. Ind. J. Kurokawa. Pedersen.. “A uniﬁed behavioral average model of SEPIC converters with coupled inductors. “Development of a microcontroller-based. no. Jan. From 1995 to 2000. Electron. 1. no. B. “Study of the effects of inductor nonlinear behavior on the performance of current controllers for single-phase PV grid converters. and W. vol. Z. 6. [17] Z.. R. 3. C. [13] D. high-efﬁcient inverter. F. Koutroulis. 11. Miaoli. and K. 881– 886. Electron.. [16] M. Sep. Mastromauro. “Multiobjective MPPT/charging controller for standalone PV power systems under different insolation and load conditions. Ind. Electron. Fundamental of Power Electronics.” in Proc. 151.” IEEE Trans. 2004. Mar. “Novel battery charging regulation system for photovoltaic applications. 55. Taiwan. H. (a) Load power is larger than the available PV power. no. Gen. and J. 46–54. Restrictions apply. 2. Jun. 2004. Sun. Dunford. Electron. VOL.” in Proc. The system has been proved to be effective in the MPPT and power balance control.” IEEE Trans.4352 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS. The proposed modeling method of the converter with the PV module input and peakcurrent-mode control. Erickson.. Wang. Jr. Duan. [18] PSIM simulation software. 1997. Y. 2629–2637. 54. Jan. vol. Noda. no. B. 5. Wu. These all demonstrate that the proposed power balance control is effective. IEEE IAS Annu. Authorized licensed use limited to: Mepco Schlenk Engineering College. Ind. K. Jul. 547–551. Voulgaris. [10] D.” IEEE Trans. 14. “Optimized maximum power point tracker for fast-changing environmental conditions. Zhao.. Rogol. 1696–1704. where he has been a Professor since 2001. Xiao. 388–393. 2006. no. Energy Convers. Eng. Taiwan. 56. Jul. in 1987 and 1994.” IEEE Trans. Kjaer. pp. Goshima. 2622–2628. [3] S. Jul. J. pp. Inst. Dec. and S. Hsinchu. NOVEMBER 2009 Fig. 7.. [7] W. 2007. 5. Ind. “High-performance stand-alone photovoltaic generation system. pp. Lee. “A review of single-phase grid-connected inverters for photovoltaic modules. NO. Liu. K. Appl.” IEEE Trans. IEEE APEC. Ben-Yaakov. Nehrir. (b) Sum of the load power and the charge power is less than the peak power of the PV module. 1. J. 1. Elect. Liserre. Guda. 2008. R EFERENCES [1] M. “A novel microcontroller for gridconnected photovoltaic systems. Wu. photovoltaic maximum power point tracking control system. Ind. 55. N.S. vol. W. 1292–1306. Z. W.” IEEE Trans. degrees in electrical engineering from National Tsing Hua University. no. “Sun screen: Investment opportunities in solar power. and A. Y. 441–446. 16. 2001. 2224–2227. 2006. Kaito. [22] L. Taiwan in 1965. Dell’Aquila. 191–197. [5] Y. 6. Electron. 41. S. [21] S. May 2008. N. MA: Kluwer.” in Proc. S. Soc. His current research interests include power electronics. Norwell. Measured waveforms for demonstrating the power balance feature. K.” IEEE Trans. pp. and G. [12] E. M. respectively. 53. Cruz. J. IEEE Power Eng. [19] R.. Ind. and A. Enslin and D. C ONCLUSION This paper has presented a PV charger implemented with the SEPIC converter. pp. and F. 5. vol. H. Liu. R. as well as the power balance control method can also be applied to the charger with other types of converter. May 2008. 2043–2052. pp. pp. peak power tracker and regulator for PV applications. Power Electron. Rahav. he was an Associate Professor with the Department of Electrical Engineering. no. IEEE ICVES. Doi. 2005. I. “Research on power balance strategy of fuel cell vehicle powertrain. vol. Ind.—Elect.. New York: Mcgraw-Hill. and Ph. Knoll.” IEEE Trans. Lo. and N.” in Conf. 240–250. and S. 2008. Z. T.. 2004. Zhong. Wai. 38th NAPS. Dec. Wilkinson. 1889–1897. Snyman. K. Koutroulis and K. Jan. no. Downloaded on October 10. V. and it will be changed to be implemented with the microprocessor or DSP and integrated with the voltage controller and PWM to make the system more practical in the future. 1991. Wan. National United University.” IEEE Trans.D. Chiang was born in Tainan. vol. pp. and K. Mineiro. 1154–1160. Meeting. T. 1501–1506. no. M. Jiang and R. 73–82. Hantschel. S. pp. A. Adar.. pp. T.. vol. [23] E. [14] H. 55. A. Daher. “Grid-connected photovoltaic system with power factor correction. . and Y. “A single-stage three-phase grid-connected photovoltaic system with modiﬁed MPPT method and reactive power compensation. [20] A. 2006. T. Mizuno. The MPPT controller was implemented with the Matlab real-time control in this paper. 2008. pp. 7. 2004. CLSA Asia-Paciﬁc Markets. Kawasaki. Koizumi. Nagasaka. “Topology study of photovoltaic interface for maximum power point tracking. Jiang. Antunes. 2006. Kang. “Combined low-cost.” IEEE Trans. Electron. 1997. H. L. and M. Liu. Pressman. the adaptive MPPT control method. M. pp. Wang. Powersim Inc.

degrees in electrical engineering from National Central University. photovoltaic systems. Taoyuan. Restrictions apply. and control theory applications. Taiwan.CHIANG et al. Hsinchu. and is currently an Associate Professor. Shieh is a member of the Taiwan Power Electronics Association and the Chinese Automatic Control Society of Taiwan. degree in electrical engineering from National United University.S. degree in electrical engineering from National Dong Hwa University. His research interests include drive and control of piezoelectric mechanisms. and Ph. in 2006. . Miaoli. Hualien.S. His research interest includes power electronics.. he has been with the Department of Electrical Engineering.: MODELING AND CONTROL OF PV CHARGER SYSTEM WITH SEPIC CONVERTER 4353 Hsin-Jang Shieh (M’02) received the B. 2009 at 04:23 from IEEE Xplore. Taiwan. Taiwan in 1980. From 1997 to 2002. Taiwan. Downloaded on October 10. in 2004 and the M. National Dong Hwa University. power converters.D. Taiwan. Chung-Li. Ming-Chieh Chen was born in Taichung. He is currently an Electronic Engineer with FSP Technology Inc. Authorized licensed use limited to: Mepco Schlenk Engineering College. Hualien. He was the recipient of the 2007 Taiwain Power Electronics Conference Best Paper Award. he was with the Mechanical Industry Research Laboratories. Dr. as a Researcher. Since August 2002.S. in 1992 and 1997. He received the B. Taiwan. Industrial Technology Research Institute. respectively. Taiwan.