You are on page 1of 14

Analysis and implementation of an adaptative PV based battery

oating charger
Nabil Karami
a,
, Nazih Moubayed
b
, Rachid Outbib
a
a
Laboratory of Sciences in Information and Systems (LSIS), Aix-Marseille University, Marseille, France
b
Department of Electrical and Electronics, Faculty of Engineering 1, Lebanese University, Tripoli, Lebanon
Received 3 August 2011; received in revised form 9 March 2012; accepted 7 May 2012
Available online 6 June 2012
Communicated by: Associate Editor Elias Stefanakos
Abstract
In a system composed of a photovoltaic (PV) cell, a converter and a resistive load, the Maximum Power Point Tracking (MPPT)
techniques are applied at the output of the PV panel and not at the level of the load. In this study, the considered load is a battery
at dierent States Of Charge (SOC) that is charged by the PV panel. The power consumed by the battery is related to its SOC. Conse-
quently, an empty battery consumes more current than a charged one. At full state of charge, the battery does not call for more energy
and thus it is not rewarding to extract more power from the PV panel.
Besides, in a stand-alone photovoltaic system, the size of the PV panel and the battery should be respected. Thus, the PV current at
dierent irradiances should be compatible with the charging current required to charge the battery at dierent SOC. A critical situation
occurs at high irradiance when the PV panel delivers a high current at Maximum Power Point (MPP) that exceeds the tolerated charging
current. The current reaches the top limit when the battery is totally empty, caused by the big dierence in potential between the con-
verter output and the battery voltages. In this case, the battery starts to gas when attempts are made to charge it faster than it can absorb
the energy. On the other hand, in a fully charged battery, the dierence in potential between the converter and the battery is zero. In this
case, there is no need to track the MPP.
In this study, we will focus on the load type and suggest new methods to reach the MPP depending on the load state. In the proposed
designs, the components of the stand-alone system are protected even if they are not well sized. In addition, we will focus on the devel-
opment of the PV array mathematical model. The results achieved with the system, as well as the experimental results of a laboratory
prototype, will be given.
2012 Elsevier Ltd. All rights reserved.
Keywords: Renewable energy; Solar panel; Photovoltaic cell; Battery; DC/DC converter; MPPT
1. Introduction
Photovoltaic energy has become one of the most prom-
ising sources of energy as it is a free and sustainable energy.
A PV array under constant uniform irradiance has a cur-
rentvoltage characteristic (IV curve), as shown in
Fig. 1. There is a unique point on the curve, called the max-
imum power point, at which the array operates with max-
imum eciency and produces maximum output power
(Tan et al., 1991; Hohm and Ropp, 2000). When a PV
array is directly connected to a load direct-coupled sys-
tem (Fig. 2a), the system operating point is at the intersec-
tion of the IV curve of the PV array and the load line. In
general, this operating point is not at the PV arrays MPP,
which is clearly seen in Fig. 1. That is, the MPP of the PV
generator is reached only in some moments throughout the
year. Thus, it is possible to talk about lost PV generator
utilization, therefore, in a direct-coupled system, the PV
0038-092X/$ - see front matter 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.solener.2012.05.009

Corresponding author.
E-mail addresses: nabil.karami@lsis.org (N. Karami), nmoubayed@
ieee.org (N. Moubayed), rachid.outbib@lsis.org (R. Outbib).
www.elsevier.com/locate/solener
Available online at www.sciencedirect.com
Solar Energy 86 (2012) 23832396
array should be oversized to ensure that the load power
requirements are complied with. This leads to an intolera-
bly expensive system (Salas et al., 2005).
To overcome this problem, a switch-mode power con-
verter, called a maximum power point tracker, is used to
maintain the PV arrays operating point at the MPP
(Fig. 2b). The MPPT does this by controlling the PV array
voltage (or current) independently of that of the load.
The MPPT of photovoltaic cells has been a research
topic over the past few years, with the aim of reaching
the maximum power point in minimal time and error
(Moubayed et al., 2008; Esram and Chapman, 2007). Sev-
eral algorithms for control of the switching converters
achieve MPP (Jain and Agarwal, 2007). Some of the widely
used schemes are the hill climbing methods (Bose et al.,
1985), the incremental conductance method (Hussein
et al., 1995), the ripple based method (Calais et al., 1998)
and the constant voltage method (Salameh et al., 1991).
These techniques dier in many aspects including simplic-
ity, convergence speed, hardware implementation, sensors
required, cost range of eectiveness and needs for parame-
terization. All these techniques focus on the maximum PV
power extraction regardless of the load types and natures.
This paper is organized as follows. The dierent types of
trackers are described in Section 2. The mathematical
model of a PV cell and the maximum power point tracker
are reviewed in Sections 3 and 4, respectively. An open and
a closed control loop for DC/DC buck converter designs
are available in Sections 5 and 6, respectively. Section 7
proposes a design using a microcontroller with three dier-
ent circuits and algorithms that manage the current and the
voltage depending on the battery SOC. The nal design is
called a PV based battery oating charger, which protects
the PV panels and increases the battery lifetime by control-
ling the converter output voltage with respect to the con-
sumed current, the available PV current and the
converter Mosfet tolerated current.
2. Type of trackers
2.1. Motorized solar cell tracking
Motorized tracking consists in implementing PV panels
on a mechanical rotating board with dierent control strat-
egies depending on the country and the day of the year.
There are two types of mechanical tracker: single-axis
trackers and dual-axis trackers.
Single-axis solar tracker: a mechanical design allows the
PV panel to move along the X axis and thus able to fol-
low the sun from dawn to nightfall.
Dual-axis solar tracker: This tracker is an electrome-
chanical device that has the photovoltaic modules xed
to its upper frame and which achieves maximum irradi-
ance on top of them. The entire structure moves from
East to West (azimuth tracking) and moves along a sec-
ond axis with a tilt over the horizontal plate position.
The panels are positioned in such a way that they are
always directed towards the sun, consequently improv-
ing their performance.
There are many algorithms used to control the tracking
motion of the PV panels whose benets vary depending on
the cost or the complexity of the system. In fact, some con-
trollers are time based and move the PV panel periodically
in the same direction of the sun. Other controllers imple-
ment the sun map location inside the tracker controller
memory. This technique is more precise but there are
energy losses due to the process of supplying the panel
motors on a cloudy day when searching for sun is worth-
less. Another method used to solve the problem of energy
loss in the panel motors is to use a small pilot cell to track
the sun and send the best position for the main panel. A
light sensor technique is also used to detect the sun irradi-
ance on the corners of the PV panel and then it is driven to
the position where the sensors detect a uniform irradiance.
0 0.5 1 1.5 2 2.5 3 3.5
0
1
2
3
4
5
6
7
8
Vmp
Imp
Voc
Isc
MPP
D
C
lo
a
d
lin
e
Operating
point
Voltage (V)
C
u
r
r
e
n
t

(
A
)
Current vs. Voltage curve
Fig. 1. Maximum power point location on the current versus voltage
curve.
Fig. 2. PV coupling methods: (a) direct coupled method and (b) coupling
with MPP tracker.
2384 N. Karami et al. / Solar Energy 86 (2012) 23832396
2.2. Electrical maximum power point tracking
A maximum power point tracker is an electronic system
that operates the photovoltaic modules in a manner that
allows the modules to produce all the power they are capa-
ble of. MPPT is not a mechanical tracking system that
physically moves the modules to be oriented directly to
the sun, but it is a fully electronic system called power con-
ditioning converter (switching converter), which varies the
electrical operating point of the modules, so that they are
able to deliver the maximum available power. The problem
considered by MPPT techniques is to nd automatically
the voltage V
MPP
and the current I
MPP
at which the PV
array should operate to obtain the maximum power output
P
MPP
for a given temperature and irradiance (Fig. 1).
An MPP tracker consists of two basic components
(Fig. 2b); a switch-mode converter and a controller with
tracking capability. The switch-mode converter is the core
of the entire supply. The converter allows energy at one
potential to be drawn, stored as magnetic energy in an
inductor, and then released at a dierent potential. The
goal of a switch-mode power supply is to provide a con-
stant output voltage or current.
When it is properly applied, a maximum power point
tracking control can prevent the collapse of the PV array
voltage under excessive load demand, particularly when
supplying a constant-power type of load. The control pro-
cess feedback signals, such as the array current and voltage,
determine a proper direction in which to move the operat-
ing point. These continuously updated set points uctuate
around the voltage corresponding to the array peak power
point. By adjusting the operating point of the array to the
point V
mp
shown in Fig. 1, the array output power is max-
imized, and the most ecient use of the solar array is
realized.
The Perturb-and-Observe (P&O) algorithm is by far the
most commonly used in commercial MPP trackers.
Although, there is as yet no consensus on which algorithm
is the best. In this algorithm, perturbations are periodi-
cally introduced into the control signal of the switching
converter, and the resulting eects are observed on the
PV output power. The P&O method is described in details
in Section 4.
3. Solar array mathematical model
The photovoltaic cell is represented as an equivalent cir-
cuit containing a current generator (modeling the conver-
sion of solar radiation to electric energy), a diode
(accounting for the physical properties of the semiconduc-
tor cells) and two resistors, a shunt resistor and a series
resistor (Fig. 3). The four variables involved in this model
are the two input variables, solar radiation E
G
(W/m
2
) and
ambient temperature Ta (C), and the two output terminal
variables, PV cell current I (A) and voltage V (V).
The characteristic equation of the PV cell model is
obtained by applying Kirchos current law to the equiva-
lent circuit shown in Fig. 3, where I and V are respectively
the terminal current and voltage of the model. The current
I is obtained by:
I I
Ph
I
d
I
sh
1
where I
ph
is the Photo current; I
d
is the Diode current; I
sh
is
the Shunt current.
The mathematical equation expressing the output cur-
rent of a single cell is the Shockely equation for an illumi-
nated pn junction and is given as (Al-Amoudi and Zhang,
2000; Premrudeepreechacham and Patanapirom, 2003):
I I
ph
I
0
e

qV IRs
AKT

1

V IR
s

R
sh
2
where I
ph
is the Photo current; I
0
is the Leakage or reverse
saturation current; q is the Electron charge =
1.602 10
19
C; V is the Solar cell voltage; A is the Ideality
factor 1.5 3; k is the Boltzman constant =
1.3806 10
23
JK
1
; R
s
is the Series cell resistance; R
sh
is
the Shunt cell resistance.
I
0
can be expressed in another form which depends on
the temperature of the solar cell as:
I
0
I
0r
T
T
r

3
e
qE
G
kA

1
Tr

1
T

3
where I
0r
is the I
0
at reference temperature T
r
; E
G
is the
Band gap energy; T
r
is the Reference tempera-
ture = 301.18 K; T is the Cell temperature.
I
ph
of Eq. (1) is a function of incident solar radiation and
cell temperature and is given as:
I
ph
I
src
k
i
T T
r

S
100
4
where I
scr
is the Short circuit current at T
r
; k
i
is the Short
circuit current temperature coecient; S is the Global solar
radiation in mW/cm
2
.
The mathematical equation expressing the output cur-
rent of n
p
cells is given as:
I 1
R
s
R
sh

n
p
I
ph
n
p
I
0
e
k
0

V
ns
IRs
1

V =n
s
R
sh
5
where n
p
is the Number of parallel cells; n
s
is the Number of
series cells; K
0
is the q/AkT.
The following equation can be applied to simulate the
characteristics of a solar array, where the given parameters
such as n
p
, n
s
, R
s
, R
sh
and I
scr
are known.
Fig. 3. PV electrical model.
N. Karami et al. / Solar Energy 86 (2012) 23832396 2385
P n
p
I
ph
V n
p
I
0
V e
K
0

V
ns
IRs
1

V
R
sh
V
n
s
IR
s

6
The temperature of a solar cell is expressed by:
T 3:12 0:25S 0:899T
a
1:3w
s
273 7
where T
a
is the Ambient temperature; w
s
is the Wind speed,
S is the Solar radiation.
Considering a constant wind speed, Fig. 4 shows the
variation of power and current at dierent irradiances
and temperatures. The current owing from the array is
directly related to the irradiance for which the array volt-
age remains almost constant. Voltage increases as panel
temperature decreases. Maximum power is extracted at
higher irradiance and lowest temperature.
4. Perturb and observe method
The P&O algorithm operates by periodically perturbing
(i.e. incrementing or decrementing) the array terminal volt-
age or current and comparing the PV output power with
that of the previous perturbed cycle. When the PV array
operating voltage changes and the power increases (dP/
dV
PV
> 0), the control system moves the PV array operat-
ing point in one direction, and vice versa. This scenario is
repeated in the next perturbation cycle (Esram and Chap-
man, 2007; Wong, 2008).
4.1. P&O algorithm
A common problem in P&O algorithms is that the array
terminal voltage is perturbed every MPPT cycle; therefore,
when the MPP is reached, the output power oscillates
around the maximum point, resulting in power losses in
the PV system. This is especially true in constant or slowly-
varying atmospheric conditions. The change in voltage is
done using a DC/DC converter (shown in Fig. 5), which
can be either a boost, a Buck or a Buck-Boost converter.
There are four dierent possible scenarios, shown as follow:
Case 1: V
present
> V
previous
)P
present
> P
previous
S=200W/m
2
S=400W/m
2
S=600W/m
2
S=800W/m
2
S=1000W/m
2
0

o
C
1
0

o
C 2
0

o
C
3
0

o
C
4
0

o
C
5
0

o
C
0

o
C
1
0

o
C
2
0
o
C
3
0

o
C
4
0

o
C
5
0
o
C
S=200W/m
2
S=400W/m
2
S=600W/m
2
S=800W/m
2
S=1000W/m
2
(a) (b)
(c) (d)
PV power at different irradiance
PV power at different temperature PV current at different temperature
PV current at different irradiance
Voltage (V)
Voltage (V)
Voltage (V) Voltage (V)
C
u
r
r
e
n
t

(
A
)
C
u
r
r
e
n
t

(
A
)
P
o
w
e
r

(
W
)
P
o
w
e
r

(
W
)

Fig. 4. Variation of current and power at dierent irradiances and temperatures. (a) Array current increases with irradiance. (b) Array power increases
with irradiance. (c) Array voltage decreases as temperature increases. (d) Array power decreases as temperature increases.
Fig. 5. PV system with MPPT.
2386 N. Karami et al. / Solar Energy 86 (2012) 23832396
DV > 0 )DP > 0
Case 2: V
present
< V
previous
)P
present
< P
previous
DV < 0 )DP < 0
Case 3: V
present
> V
previous
)P
present
< P
previous
DV > 0 )DP < 0
Case 4: V
present
< V
previous
)P
present
> P
previous
DV < 0 )DP > 0
At rst, the P&O algorithm of Fig. 6 starts by calculat-
ing the power value at 0 V. Then, the voltage is slightly
increased and a new power value is calculated. In fact,
the rst voltage increment provides a power escalation
since the rst assumption is at 0 V.
The innite loop of the tracker algorithm starts by com-
paring the new value of P with the previous one. The four
cases described above have to be considered in order to
decide on which direction the new voltage value will be.
The voltage step change is considered small (around
0.1 V) for a modest power variation. The right voltage step
value can be determined experimentally regarding the time
response and the power overshoot.
4.2. Perturb and Observe simulation results
The simulation results illustrated in Figs. 79 show the
growing voltage value as well as the current and power
variations. Voltage (in dashed blue) starts from zero and
increases by 0.1 V on every perturbation cycle. Current
Fig. 6. P&O algorithm.
N. Karami et al. / Solar Energy 86 (2012) 23832396 2387
(in solid green
1
) starts from I
SC
and decreases until the
power reaches the maximum. The implemented algorithm
should start inversely from the open circuit voltage V
oc
value to prevent the PV short circuit I
SC
on every startup.
Power ripples shown in Fig. 9 are the results of the 0.1
perturbation voltage.
In a direct coupled PV/battery mode (Fig. 10a), the
operating point is determined by the batterys potential.
Typically, this operating point is not the ideal operating
voltage at which a module is able to produce its maximum
available power. In addition, the P&O experimental steps
of the MPPT are not applicable when the load is a battery
that has to be charged by the PV panel. Given that the PV
panel voltage should be xed to 2.232.3 V per cell (Wong,
2008; Genesis Series Batteries Application Manual, 2005;
Yuasa All NP Series Batteries User Manual, 1999), that
is, 13.8 V for a battery of 12 V. Moreover, the internal
resistance of the battery is related to its state of charge.
Therefore, using Ohms law, the current needed to charge
the battery is xed by the battery itself and cannot be chan-
ged. For instance, if the battery is empty, it behaves like a
resistor with low impedance and absorbs signicant cur-
rent. Vice versa, when the battery is fully charged, it
behaves like a resistor with high impedance, and consumes
modest current. Therefore, using the MPPT is not always
rewarding as the load may not be able to absorb all the
generated power and there is no need to locate the maxi-
mum power point in similar kinds of PV couplings.
Many systems described in Pernia et al. (2009), Eakbur-
anawat (2006), Muhida (2003), Eftichios et al. (2003) use
the MPPT by assuming a xed battery voltage for simpli-
cation. If the input and the output powers of a DC/DC
converter are given equal (case of 100% converter e-
ciency), the relation can be described as:
Power
PV
Power
bat
8
Therefore,
V
PV
I
PV
V
bat
I
bat
9
Assuming a xed battery voltage, V
bat
is replaced by a
constant E
B
, and the above equation will be:
V
PV
I
PV
E
B
I
bat
10
Therefore, from Eq. (10), the PV power P
pv
depends
only on the battery current I
bat
. The PV power reaches
its maximum, P
pv,max
, only when battery current reaches
its maximum I
bat,max
.
In a step-down DC/DC converter, there is only one con-
trol parameter, namely the duty cycle D, of the power elec-
tronics switch. The duty cycle D of a DC/DC converter can
be expressed as (Muhida, 2003):
Fig. 7. Voltage and current variations searching for MPP.
Fig. 8. Power value swings around MPP.
Fig. 9. A zoom-in on the power oscillation around the MPP.
1
For interpretation of color in Figs. 110, 12, 14, 15 and 1722, the
reader is referred to the web version of this article.
2388 N. Karami et al. / Solar Energy 86 (2012) 23832396
D
V
bat
V
PV

I
PV
I
bat
; with 0 < D < 1 11
From Eq. (11), we can conclude that by maximizing the
battery current at the converter output, we can track
the maximum power point of the PV panel by controlling
the duty factor of the DCDC converter. The drawback
of this method lies in the assumption that the battery volt-
age remains relatively constant, which is not the genuine
situation at dierent states of charge.
This study takes into consideration variable battery
voltages at dierent states of charge. The new design
applies a feedback system totally dierent from the tradi-
tional MPPT that reads the PV cell output and not the load
input. The system is treated by means of two methods; the
rst one is based on a PID controller and the second one is
based on a microcontroller with current protection algo-
rithm. These two methods use the DC/DC converter to
reach the desired output voltage.
5. Direct coupling with a buck converter
This study considers coupling a PV panel with a battery
via a DC/DC converter, as shown in Fig. 10b. The PV
panel used is KD135GH-2PU from Kyocera, where V
MMP
and I
MPP
are respectively 17.7 V and 7.63 A at 1000 W/m
2
Fig. 10. Dierent PV/battery connection modes. (a) Direct PV/battery connection. (b) A controlled DC/DC converter is used to x the charging voltage
based on V
ref
. (c) A PID controller is used to control the PWM signal based on the dierence between V
ref
and the battery voltage V
bat
. (d) Smart
controller design where voltage and current are controlled based on V
ref
and I
ref
.
Table 1
Electrical characteristics of the PV panel with an irradiance level of 1 kW/
m
2
.
Symbol Quantity Value
P
MMP
Maximum power 135 W
V
MMP
Voltage at P
MMP
17.7 V
I
MMP
Current at I
MMP
7.63 A
I
SC
Short-circuit current 8.37 A
V
OV
Open-circuit voltage 22.1 V
T
SC
Temperature coecient of I
SC
0.00502 A/C
T
OC
Temperature coecient of V
OC
0.081 V/C
N. Karami et al. / Solar Energy 86 (2012) 23832396 2389
irradiance. The PV module main specications are shown
in Table 1.
The load is made of 6 OPzV series batteries, 2 V each. A
buck converter is used to step down the PV potential to the
battery limit. The buck converter, shown in Fig. 11, is com-
posed of a switching component (a Mosfet or an IGBT
transistor) and a low pass lter made of a coil and a capac-
itor (Salas et al., 2002; Anthony et al., 1994).
The expression of the buck average output voltage is
obtained as follows:
V
0;avg
DE 12
When the switch is ON, the voltage across the inductor
L is expressed by:
v
L
t L
di
L
dt
E v
0
13
When the output voltage V
0
remains steady at V
0,avg
, the
inductor current i
L
increases linearly during the ON period
of the switch. Then:
DI
E V
0;avg
L
DT
DE V
0;avg

fL
14
where f is the frequency (Hz), T the period (S), and L the
inductance (H).
During the ON period, the inductor current rises from
i
L,min
to i
L,max
, where:
i
L;min
i
L
0 i
L
T
V
0;avg
R

DI
2
15
and
i
L;max
i
L
DT
V
0;avg
R

DI
2
16
R represents the internal resistor of the connected batteries.
The capacitor current i
C
is expressed as follows:
i
C
t i
L
t i
R
t i
L
t
V
0;avg
R
17
i
C
0 i
C
T
DI
2
18
i
C
DT
DI
2
19
Since the current through the capacitor varies linearly,
the average charging current is going to be half of its peak
value of the triangular waveform. This peak is shown to be
DI/2. Hence:
I
C;avg

DI
4
20
Based on the average charging current, the change in
output voltage DV becomes:
DV
I
C;avg
T=2
C

DI T
8 C

DI
8 f C
21
The duty cycle of a DC buck converter that converts
from the PV panel voltage of 22.1 V to the series batteries
voltage of 13.8 V is given by output/input ratio which is
62%. For a 1 Amp peak-to-peak current output and at
25 kHz switching frequency, the inductor value is 328 lH.
A capacitor of 500 lF produces a 10 mV peak-to-peak out-
put ripple voltage.
The buck open loop transfer function is given by:
Gs
1
LC
S
2

S
RC

1
LC

x
2
n
S
2
2nx
n
S x
2
n
22
where the natural frequency x
n
1=

LC
p
and the damping
ratio n 1=2R

L=C
p
.
The open loop step response illustrated in Fig. 12 shows
a 13.8 V steady output, a 48% overshoot and 7 ms settling
time, which are not acceptable in a real application, espe-
cially for loads with sensitive electrical characteristics.
Therefore, the system should be controlled by an external
controller like a PID controller in order to reach an accept-
able overshoot (usually 5%) and a smaller settling time.
6. Direct coupled systems with PID output potential control
In a typical closed loop design criteria, the tolerated
overshoot is 5% and the settling time depends on the e-
ciency and the cost of the components implemented in Fig. 11. Buck converter design.
A
m
p
l
i
t
u
d
e

(
V
o
l
t
)
Time (sec)
Fig. 12. Voltage time response of the buck converter.
2390 N. Karami et al. / Solar Energy 86 (2012) 23832396
the design. For an ecient buck controller design, it is
decided to get the settling time for 5 ms. Fig. 10c shows
the block diagram of the charger system using a PID
controller.
The transfer function of the PID controller looks like
the following:
GcS
KdS
2
KpS Ki
S
23
K
p
is the proportional gain, K
i
the integral gain, and K
d
the
derivative gain.
The transfer function of the buck design (plant) is:
GS
6:09 10
6
S
2
1111S 6:09 10
6
24
The closed loop of the block diagram shown in Fig. 13
is:
TS
Gc G
1 Gc G
25
In order to get the right overshoot and settling time, the
desired time domain equation requires that the maximum
overshoot be 5% and the settling time 5 ms. Therefore:
n 0:6901 and x
n
1449 rad=s
Thus, the desired closed loop transfer function becomes:
GclS
2:1 10
6
S
2
2000S 2:1 10
6
26
Comparing the denominator of T(S) and G
cl
(S) yields:
K
i
1721; K
p
0:9841 and K
d
0:00096:
The output response of the system is shown in Fig. 14
where the design criteria are respected.
In the case of a battery load, the overshoot and the set-
tling time are not critical elements of the circuit since the
battery charging dynamic response behaves like a charging
capacitor and reduces the eect of fast overshoots. Also,
the time response of the controller is not critical if it is
between 5 ms and 5 s. Therefore, the implementation of a
PID controller in the circuit is not rewarding and it will
not bring any benets to the complete system.
7. Microcontroller based smart direct coupled system
In general, a battery can be represented as a resistive
load with dierent values depending on the state of charge.
Here a question may be asked: what happens when the bat-
tery is fully discharged? In this case, the battery behaves
like a very small resistance, and requires a big amount of
current. Based on the study of Calais et al. (2008), Quin-
tana et al. (2002), Goss et al. (2011), Yang (2006), Kawak-
ita et al. (2002), over-current can cause the heating of cells
and interconnections and indirectly aects the degradation
of the PV module. On the other hand, when the battery is
fully charged, the consumed current becomes zero. Thus,
there is no need to nd the MPP unless the battery is cou-
pled with another load, and this is not the case in this
study. In the case of a partially discharged battery, the
maximum power delivered by the PV panel may not be
compatible to the battery capacity and therefore the excess
of energy is turned into heat, which then causes the electro-
lyte to boil and evaporate.
In order to avoid the above mentioned problems, a spe-
cial controller is designed. The proposed system is able to
control the oating voltage dierence between the PV
and the battery in order to moderate the consumed current.
Moreover, it protects the global system (PV, converter and
battery) from the PV over-current and from the battery
overcharge. Therefore, the system can be considered an
adaptive PV based battery oating charger.
7.1. Controller design
Using an ascending voltage technique, the proposed
controller can charge the battery until it reaches 13.8 V.
Note that the current value depends on the dierence in
potential between the source and the load. This technique
is called oating charging (Wong, 2008; Niu, 2004) and it
is used to supply the battery with a proportional amount
of current and voltage depending on its state of charge.
Fig. 10d and Fig. 15 represent the block diagram of the
system. The desired voltage and the maximum allowed cur-
rent are supposed to be determined. The primary proposed
system reads the actual current and voltage and compares
them to the desired ones based on the algorithm proposed
in Fig. 16.
Fig. 13. Closed loop PID controller.
Time (sec)
A
m
p
l
i
t
u
d
e

(
V
o
l
t
)
Fig. 14. Open loop, desired, and closed loop response.
N. Karami et al. / Solar Energy 86 (2012) 23832396 2391
In fact, the role of the microcontroller is to read the
actual battery current and compare it to the maximum
allowed one (Max current in Fig. 15). If the dierence
eI is negative in the case of high battery consumption cur-
rent, the duty cycle of the PWM signal decreases. If not, the
battery voltage is read and compared to the reference volt-
age (Desired voltage in Fig. 15). The desired voltage is
limited to 13.8 V in our experimental case. If the battery
voltage is less than the desired one, the duty cycle is
increased providing a voltage raise across the battery.
When the actual battery voltage exceeds the desired limit,
the PWM duty cycle decreases. Besides, an equality match-
ing between the desired and the actual voltages maintains
the error eV to zero. Therefore, the duty cycle maintains
its old value. Having an adjustable desired voltage and cur-
rent, the system becomes a universal charger for all battery
sizes. The algorithm used is illustrated in Fig. 16.
As shown on the algorithmin Fig. 16, the microcontroller
increases and decreases the value of the duty cycle by one.
Thus, if the PWM signal generator is based on an 8-bit divi-
sion, the microcontroller needs 256 cycles to change the duty
cycle fromzero (0%) to its maximum(100%). Therefore, 256
loops of the proposed algorithm are needed to get the max-
imum value when starting from 0. In a well structured
Fig. 15. The Microcontroller compares the preset voltage and current values with the actual ones and drives the buck converter.
Fig. 16. Microcontroller program algorithm.
2392 N. Karami et al. / Solar Energy 86 (2012) 23832396
microcontroller program code, the loop can take about
50 ls for a controller running at 1 ls per instruction; conse-
quently, the response time for 256 loops is 13 ms, which is
good enough to supply battery loads.
7.2. Determination of the maximum allowed current
The weakness of the above design lies in its inability to
accommodate the PV current which is a function of irradi-
ance. PV panels are able to deliver extra current at higher
irradiance and the maximum tolerated current proposed
in Fig. 15 should be function of that irradiance in order
to grab as much PV power as possible. Therefore, three
enhanced designs are proposed. The rst one uses a pilot
PV cell, the second one takes around 90% of the PV cell
short circuit current and the third one uses a shunt current
sensor.
7.2.1. First method: Pilot cell
This method consists in placing a pilot cell next to the
main PV cell to measure the irradiance value and feed
the microcontroller with the appropriate value in order to
determine the I
max
(Fig. 17). The relation between the real
I
pv
and the I
pilot
can be experimentally extracted and saved
in a lookup table inside the microcontroller memory or
modeled as a formula.
7.2.2. Second method: PV short circuit
This method consists in disconnecting the PV cell from
the buck converter and shorting it in order to get the I
sc
using a current sensor. The I
max
can be between 78% and
92% of the I
sc
, as described in Esram and Chapman
(2007), Hart et al. (1984), Masoum et al. (2002). This
method causes the destruction of the PV cell by heating
up the internal bus-bar and wires connecting the series
and parallel cells.
Fig. 17. Design enhancment using pilot PV cell.
Fig. 18. Design enhencement using on-board current measurement.
Fig. 19. Controller board design.
N. Karami et al. / Solar Energy 86 (2012) 23832396 2393
7.2.3. Third method: PV resistor shunt
This method is similar to the previous one, except that a
small resistor is placed in series with the PV cell. The resis-
tor can be chosen so that the PV maximum current is at
90% of the PV short circuit current (Fig. 18). This method
is unworthy and needs a resistor and a relay to switch
between the buck and the resistor circuits. In addition,
the inconvenience of this method appears in the periodic
power disconnection at every current sample reading.
7.3. Pilot cell design description
7.3.1. Pilot cell design implementation
The design in Fig. 17 is implemented on two Printed Cir-
cuit Boards (PCBs). The rst one is the controller board
(Fig. 19) and the second one is the buck converter
(Fig. 20). The controller board is based on the microcon-
troller PIC16f874A from Microchip

, in which the algo-


rithm described in Fig. 15 is implemented. This
microcontroller has eight analog inputs, two built-in
PWM signal generators and a Universal Synchronous
Asynchronous Receiver Transmitter USART module.
Four analog inputs are used to read the desired and the
actual voltage and current. The buck converter is driven
by the PWM signal delivered based on the proposed algo-
rithm. On every program cycle, data are sent serially to a
Visual Basic program designed to plot the input and output
voltages and currents.
The buck converter is based on the IRFP064 HEXFET
Power Mosfet having a 0.009 X R
DS(on)
resistance, a 60 V
V
DSS
voltage, and 70 A I
D
current. The Mosfet is driven
by the IR2110chip from International Rectier

.
Fig. 20. Buck converter design.
Fig. 21. Visual Basic program interface.
2394 N. Karami et al. / Solar Energy 86 (2012) 23832396
The visual basic interface shown in Fig. 21 is designed to
read the PIC microcontroller data, plot them, and export
them to an excel sheet le.
7.3.2. Experimental results obtained on a pilot cell
The system is connected to a low state of charge battery
and is charged at two dierent irradiance levels. The irradi-
ance is sensed by a pilot PV cell on which the microcontrol-
ler can determine the maximum allowed battery current.
The reference voltage is xed to 13.8 V. At startup, the
PIC generates a PWM signal with 0% duty cycle and incre-
ments it until the battery voltage and current are under the
voltage and current criteria. The graph in Fig. 22 shows
that the battery voltage is 10 V and it increases with the
duty cycle. At point A the battery current reaches
the upper limit and the microcontroller holds the value of
the duty cycle to protect the PV panel from current surge.
It is obvious that the charging voltage is not reaching the
reference 13.8 V at this point. At point B the irradiance is
increased and a new current limit is proposed by the micro-
controller based on a lookup table implemented inside its
memory. The duty cycle is therefore increased and is held
when the battery voltage reaches 13.8 V. Between points
B and C, the battery current and the PWM graph
show a slight decrease because the current needed for the
battery decreases while charging. After a full charge, the
current decreases to around zero amperes while the voltage
remains uctuating around 13.8 V.
7.3.3. Eect of temperature and irradiance on the battery
charger
As mentioned before, the output voltage of the PV panel
used is 22 V which is higher than the battery voltage by
10 V. From the specications of the PV panel listed in
Table 1, an increase of 1C removes 0.08 V from the PV
panel terminal. At 50 C, the PV panel output voltage
decreases by 2 V to around 20 V, which is also greater than
the proposed charging voltage. Therefore, the system is not
aected by the temperature as far as the PV provided volt-
age is greater than 13.8 V (assuming a 100% DC/DC con-
verter eciency).
Moreover, the battery charging time depends mainly on
the charging current applied to it, and so to the irradiance
applied to the PV panel. The charging time is longer at low
irradiance without PV voltage collapse and is shorter at
high irradiance without surging over current.
The proposed design can be adapted to all PV and bat-
tery sizes, provided only that the open circuit PV voltage is
greater than the connected battery potential.
8. Conclusion
This study shows the dierence between the direct-cou-
pled and the indirect-coupled applications of a PV panel.
Also, it focuses on two closed loop techniques to supply
a xed voltage load. The maximum power point tracking
considers only the maximum benets of the solar cell and
is almost used to cross the output with the power lines so
that this power can be sold.
The second technique focuses on the load nature and
respects its potential and current regardless of the PVpower
and losses. The PID controller of the buck output voltage
shows a complicated study especially for various load nat-
ure and parameters. For example, a buck supplying a resis-
tive load is modeled in a dierent way than if it is supplying
a battery load. Batteries are modeled as resistive and capac-
itive load arranged in a complicated circuit and therefore
the pole placement of the PIDdepends on these values, con-
sequently on the state of charge of the battery.
The microcontroller based control technique computes
only the dierence between the desired and the actual load
voltage and tracks like a proportional controller with a
xed up/down value. There is no need to enhance the set-
tling time and the response time in a non-sensitive load like
batteries. Finally, the implemented system shows great and
faultless results and assures an adaptive oating battery
Fig. 22. Experimental charging result.
N. Karami et al. / Solar Energy 86 (2012) 23832396 2395
charger independently of the irradiance and temperature
factors.
References
Al-Amoudi, A., Zhang, L., 2000. Application of radial basis function
networks for solar-array modeling and maximum power-point predic-
tion. IEE Proc. Gener. Transm. Distrib. 147 (5), 310316.
Anthony, J., Stratakos, S., Sanders, R., Brodersen, R., 1994. A low-
voltage CMOS DCDC converter for a portable battery-operated
system. In: Proceedings of the IEEE Power Electronics Specialists
Conference.
Bose, B. et al., 1985. Microcontroller control of residential photovoltaic
power conditioning system. IEEE Trans. Ind. Appl. 21 (5), 11821191.
Calais, M. et al., 1998. A ripple-based maximum power point tracking
algorithm for a single-phase, grid-connected photovoltaic system.
Solar Energy 63 (5), 277282.
Calais, M., Wilmot, N., Ruscoe, A., Arteaga, O., et al., 2008. Over-current
protection in PV array installations, In: 3rd International Solar Energy
Society Conf. Asia Pacic Region (ISES-AP-08), Sydney.
Eakburanawat, J., 2006. Development of a thermoelectric battery charge
with microcontroller based maximum power point tracking technique.
Appl. Energy.
Eftichios, K., Kostas, K., Voulgaris, N., 2003. Development of a
microcontroller-based, photovoltaic maximum power point tracking
control system. IEEE Trans. Power Electron. 16 (1), 717724 (4654,
2001).
Esram, T., Chapman, P.L., 2007. Comparison of photovoltaic array
maximum power point tracking techniques. IEEE Trans. Energy
Convers. 22 (2), 439449.
Genesis Series Batteries Application Manual, Sixth ed., Publication No:
EN-GPLAM-002, 2005, pp. 1023.
Goss, B., Reading, C., Gottschalg, R., 2011. A review of overcurrent
protection methods for solar photovoltaic DC circuits. In: 5th
Photovoltaic Science Application and Technology (PVSAT-5) Con-
ference and Exhibition, Edinburgh, Scotland: PVSAT.
Hart, G.W., Branz, H.M., Cox, C.H., 1984. Experimental tests of open
loop maximum-power-point tracking techniques. Solar Cells 13, 185
195.
Hohm, D., Ropp, M., 2000. Comparative study of maximum power point
tracking algorithms using an experimental, programmable, maximum
power point tracking test bed. In: Photovoltaic Specialists Conference,
2000. Conference Record of the Twenty-Eighth IEEE. pp. 16991702.
Hussein, K. et al., 1995. Maximum photovoltaic power tracking: an
algorithm for rapidly changing atmosphere conditions. IEE Proc. G
142, 5964.
Jain, S., Agarwal, V., 2007. New current control based MPPT technique
for single stage grid connected PV system. Energy Convers. Manage.
48, 625644.
Kawakita, S., Imaizumi, M., Yamaguchi, M., Kushiya, K., Ohshima, T.,
Ito, H., Matsuda, S., 2002. In situ measurement of degradation of
Cu(In, Ga)Se2 thin lm solar cells during electron and proton
irradiations. In: Photovoltaic Specialists Conference, Conference
Record of the Twenty-Ninth IEEE, vol. no., 1924 May 2002, pp.
978981.
Masoum, M.A.S., Dehbonei, H., Fuchs, E.F., 2002. Theoretical and
experimental analyses of photovoltaic systems with voltage and
current based maximum power-point tracking. IEEE Trans. Energy
Convers. 17, 514522.
Moubayed, N., El Ali, A., Outbib, R., 2008. Comparison between
dierent control methods of a solar energy conversion system. In: 33rd
IEEE Photovoltaic Specialists Conference, 1116 May 2008, San
Diego, California, USA.
Muhida, R., 2003. A maximum power point tracking for photovoltaic
SPE system using a maximum current controller. Solar Energy Mater.
Solar Cells.
Niu, J., 2004. Comparative studies of self-discharge by potential decay and
oat current measurements at C double layer capacitor and battery
electrodes. J. Power Sources.
Pernia, A.M., Arias, J., Prieto, M.J., Martinez, J.A., 2009. A modular
strategy for isolated photovoltaic systems based on microcontroller.
Renew. Energy 34 (7), 18251832.
Premrudeepreechacham, S., Patanapirom, N., 2003. Solar-array modeling
and maximum power point tracking using neural networks. In: IEEE
Bologna PowerTech Conference, Italy.
Quintana, M.A., King, D., McMahon, T.J., Osterwald, C.R., 2002.
Commonly Observed Degradation in Field-Aged Photovoltaic Mod-
ules, Proc. 29th IEEE PVSEC, New Orleans, USA.
Salameh, Z. et al., 1991. Step-down maximum power point tracker for
photovoltaic systems. Solar Energy 46 (5), 279282.
Salas, V., Manzanas, M.J., La zaro, A., Barrado, A., Olas, E., 2002. The
control strategies for photovoltaic regulators applied to stand-alone
systems. IEEE Ind. Electron. Soc.
Salas, V., Olas, E., Lazaro, A., Barrado, A., 2005. New algorithm using
only one variable measurement applied to a maximum power point
tracker. Solar Energy Mater. Solar Cells 14, 675684.
Tan, C.W., Green, T.C., Hernandez, C.A., 1991. An improved maximum
power point tracking algorithm with current-mode control for
photovoltaic applications. In: Proceedings of 1991 IEEE International
Conference on Power Electronics and Drive Systems, Kuala Lumpur,
Malaysia, November 28December 1, 1991.
Wong, Y.S., 2008. Charge regimes for valve regulated lead-acid batteries:
performance overview inclusive of temperature compensation. J.
Power Sources.
Yang, H., 2006. Inuence of the charge regulator strategy on state of
charge and lifetime of-VRLA battery in household photovoltaic
systems. Solar Energy.
Yuasa All NP Series Batteries User Manual, 1999, pp. 1324.
2396 N. Karami et al. / Solar Energy 86 (2012) 23832396

You might also like