You are on page 1of 14

Available online at www.sciencedirect.

com

ScienceDirect
Solar Energy 123 (2016) 102115
www.elsevier.com/locate/solener

Design and implementation of a dynamic FPAA based


photovoltaic emulator
Marco Balato, Luigi Costanzo , Daniele Gallo, Carmine Landi, Mario Luiso,
Massimo Vitelli
Department of Industrial and Information Engineering, Second University of Naples, Italy
Received 2 September 2015; received in revised form 27 October 2015; accepted 3 November 2015

Communicated by: Associate Editor Mario Medina

Abstract
In this paper, a new Photovoltaic (PV) emulator is presented and discussed. Its main feature is represented by the use of a Field Programmable Analog Array (FPAA) on which the desired current vs. voltage (IV) PV characteristic can be implemented. The FPAA provides a suitable analog time varying reference signal for the output current control of a proper DC/DC converter whose output port
emulates the PV IV curve. The proposed emulator allows to track time varying irradiance values and therefore it allows also to emulate
typical scenarios of automotive applications or involving fast time varying weather conditions (e.g. the ones which usually occur in tropical locations). Additional, not less important advantages of the proposed solution are the following ones: (1) no numerical interpolations
and no storage of big amount of data in memory are required; (2) the FPAA is characterized by a great ease of reconguration and
programming with respect to FPGA or DSP based implementations; (3) no DAC or ADC converters are needed; (4) not only uniform
but also mismatching operating conditions can be easily emulated; (5) power sources dierent from PV sources can be easily emulated by
using the same architecture. The presented experimental results allow to conrm the validity of the proposed FPAA based architecture.
2015 Elsevier Ltd. All rights reserved.

Keywords: Photovoltaic emulator; FPAA; DC/DC converter control

1. Introduction
In the scientic literature of the last ten years, a huge
amount of papers has been published on Maximum Power
Point Tracking (MPPT) algorithms Liu et al., 2014;
Miyatake et al., 2011; Sera et al., 2013; Reisi et al., 2013;
Al Nabulsi and Dhaouadi, 2012 and on advanced static
or dynamic PV architectures (e.g. distributed MPPT applications, array reconguration strategies, etc.) (Qin et al.,
2014; Balato and Vitelli, 2014; Feng et al., 2014; La
Manna et al., 2014; Spagnuolo et al., 2015).
Corresponding author.

E-mail address: luigi.costanzo@unina2.it (L. Costanzo).


http://dx.doi.org/10.1016/j.solener.2015.11.006
0038-092X/ 2015 Elsevier Ltd. All rights reserved.

Whichever the MPPT algorithm or the PV architecture


or the reconguration technique to analyze and test, the
execution of the experimental test activities by using real
PV modules is nearly always unpractical or even impossible. In fact, there are many limitations that have to be
taken into account when using real PV modules. First of
all, the availability of enough solar radiation in correspondence of the planned experimental activity cannot be
always guaranteed because, of course, the external climatic
conditions are not under control. Secondly, the repeatability of the experimental tests cannot be ensured, once again
because of the variability of weather conditions. Moreover,
a huge variety of solar panels exist. It would be prohibitively expensive to purchase each type of panel and test

M. Balato et al. / Solar Energy 123 (2016) 102115

it individually if needed, for example for eciency comparison purposes, by the specic study to be carried out.
Moreover, the cost, the physical dimensions of the PV
array and the limited space on the laboratory roof for
the installation, may represent a hard problem for Universities or for small/medium Companies research laboratories. In addition, the outdoor installation of the
measurement station is not easy, especially in presence of
unfriendly environmental conditions. Last, but not least,
if tests concerning mismatching conditions are needed, in
order for example to examine the eect on the overall eciency of PV sources with portion of cells which are
cracked, damaged or shaded, such tests should be destructive or might cause, for example in case of shading, the
accelerated aging of the PV source under test. In fact, it
is well known that mismatching operating conditions
unavoidably lead to nonuniform aging and to damaging
of the PV cells (Report IEA-PVPS T13-01, 2014).
The need of PV Emulators (PVEs) is the natural consequence of the above drawbacks. A PVE allows to reproduce the current vs. voltage (IV) characteristic of a
given PV eld by means of a suitable controllable power
source. A non exhaustive list of commercial PVEs which
are available on the market is the following one: Keysight
(Agilent) E4360 Modular Solar Array Simulators (http://
literature.cdn.keysight.com/litweb/pdf/5989-8485EN.pdf),
Magna-Power Photovoltaic Power Prole Emulation
(PPPE)
http://www.magna-power.com/les/datasheet/
pppe/datasheet_pppe_2.0.pdf, Newdoll Enterprices PhotoVoltaic Emulator (http://www.accuratesolarpower.com/
PVE_User_Guide_V1.5.pd).
The main drawback which prevents the widespread use
of commercial PVEs in nearly all the University laboratories or in small/medium Companies research laboratories
is represented by their quite high cost. In fact, it is worth
noting that in many cases, PV arrays with a relatively large
number of PV modules need to be tested and therefore as
many PVEs are needed.
For such reasons, many papers in the literature are
devoted to the development of low cost PVEs (Sanchis
et al., 2007; Lu and Nguyen, 2012; Rana and Patel, 2013;
Ickilli et al., 2012; Zeng et al., 2002; Midtgard, 2007;
Mukerjee and Dasgupta, 2007; Martin-Segura et al.,
2007; Chavarria et al., 2014; Gadelovits et al., 2014;
Balakishan and Sandeep, 2014; Woojoo et al., 2011; Di
Piazza and Vitale, 2010; Nagayoshi, 2004; Mellit et al.,
2010; Dolan et al., 2011; Tornez-Xavier et al., 2013;
Schoeld et al., 2011; Roncero-Clemente et al., 2013). In
most papers, the PVE consists of a DC/DC power converter which is controlled by means of a FPGA or a
DSP-based unit. More specically, the PVEs can be
grouped in three main families. The rst and largest one
collects static PVEs, that is emulators which are only able
to operate with a xed IV curve which must be set before
the beginning of any test and that cannot be changed
during the test itself. If dierent operating conditions, as
concerns irradiance and temperature values, need to be

103

analyzed, then the system must be turned o and a new


IV curve can be set (Sanchis et al., 2007; Lu and
Nguyen, 2012; Rana and Patel, 2013; Ickilli et al., 2012;
Zeng et al., 2002; Midtgard, 2007; Mukerjee and
Dasgupta, 2007; Martin-Segura et al., 2007).
The second family collects instead semi-dynamic PVEs,
that is emulators which are able to switch, during the test,
from a starting IV curve to other dierent IV curves
which have been a priori evaluated and stored in the memory (Chavarria et al., 2014; Gadelovits et al., 2014;
Balakishan and Sandeep, 2014; Woojoo et al., 2011; Di
Piazza and Vitale, 2010; Nagayoshi, 2004; Mellit et al.,
2010; Dolan et al., 2011; Tornez-Xavier et al., 2013). Of
course, only a limited number of curves can be stored
and, due to such a limitation, only relatively simple tests
can be carried out. A few PVEs are indeed able to evaluate
the actual IV curve in real time, during the test, but a not
negligible time is required to carry out the necessary calculations; so that, in any case, the actual, continuous tracking
of rapid uctuations of the irradiance is not possible. This
may be acceptable in some cases (e.g. emulation of stationary PV applications on buildings in case of slowly timevarying weather conditions) but it is not acceptable at all
during tests on mobile applications such as the automotive
ones which can be characterized by irradiance rate of variations as high as 40 kW/m2 s (Stracke et al., 2014) or when
fast time-varying weather conditions (e.g. the ones usually
characterizing tropical locations) need to be accounted for
(Ye et al., 2013).
The last family of PVEs is composed by fully dynamic
devices. They are able to continuously track, during the
tests, time-varying irradiance behaviors. In fact, such PVEs
are characterized by the presence of an analog input port
where a time-varying voltage signal, which is proportional
to the irradiance, can be applied. The input voltage signal
representing the irradiation can be characterized by a quite
high speed of variation since the only limitation is represented by the bandwidth of the control circuitry of the
DCDC power supply. Such a bandwidth is in turn limited
by the switching frequency and by stability considerations
(Erikson and Maksimovic). As concerns the ambient temperature, which, together with the irradiance value and
the position of the operating point dictate the PV module
operating temperature, due to its relatively slow variations,
no particular fast PVE dynamic capabilities are instead
required. Therefore, except that in very rare, particular
cases concerning tests carried out with PVEs and characterized by very long durations, the modules temperature can
be nearly always considered constant due to their relatively
large thermal inertia.
It is worth noting that, to the best of the authors knowledge, only very few papers dealing with PVEs belonging to
the third family have been presented in the literature
(Schoeld et al., 2011; Roncero-Clemente et al., 2013). In
(Schoeld et al., 2011) the implementation of a very low
power (up to 4 W) analog circuitry which is able to approximate (single diode model) the IV characteristic of a PV

104

M. Balato et al. / Solar Energy 123 (2016) 102115

cell is discussed. Such an analog circuitry is not useful for


experimental activities requiring higher powers (over
4 W). In principle, the above circuitry could be used in
order to provide the reference control signal for a proper
power stage of a high power emulator. But, since well
dened analog components, such as diodes, operational
ampliers (OP-AMP), and bipolar junction transistors
(BJT), are adopted, there is little or no exibility at all as
concerns the nal shape and characteristics of the desired
IV curve. That is, little or no tuning capabilities can be
obtained. Therefore it is nearly impossible to test performances associated to PV systems adopting dierent PV
modules. Moreover it is not possible to obtain IV curves
which exhibit the classical shape deformations appearing
when mismatching conditions occur.
Another example of PV array emulator belonging to the
third family is presented in Roncero-Clemente et al. (2013).
The power conversion stage is represented by a three-phase
synchronous rectier (Semikron SEMISTACK SKS 230F
B8CI 190 V12). The control strategy is implemented in a
PC which is also equipped with data acquisition cards; Hall
eect sensors measure the DC-link current and the corresponding currents drawn from the grid. Such an emulator
is specically devoted to the testing of commercial PV
inverters at quite high power levels.
In this paper, a new emulator belonging to the third
family is proposed. Its main feature is represented by the
use of a Field Programmable Analog Array (FPAA)
http://www.anadigm.com/an231e04.asp on which the
desired IV characteristic is implemented. In particular,
the FPAA has an input port where the analog voltage signal Sin(t), which is proportional to the time-varying solar
irradiance, is applied. In turn, an output port provides a
suitable analog time varying reference signal ipan_ref(t) for
the output current control of a proper DC/DC converter.
The output port of such a converter emulates the PV IV
curve. The main advantages of the proposed solution are:
Both static and dynamic irradiance conditions can be
emulated; in particular it is possible to emulate any type
of dynamic irradiance condition; this is highly desirable
when testing MPPT techniques that should be able to
properly work under time varying environmental
conditions.
The signal ipan_ref is obtained without numerical interpolations and without the need of the storage of big
amount of data in memory in order to emulate timevarying climatic conditions.
As discussed in Selow et al. (2009), the FPAA is characterized by a great ease of reconguration with respect to
FPGA or DSP based implementations. A very user
friendly graphical interface is provided by the manufacturer and allows to take advantage of the various available FPAA analog building blocks (adders, multipliers,
integrators, PIDs, look up table, lters, etc.). Therefore
the designer programming is strongly simplied.

Not only uniform but also mismatching operating conditions can be easily emulated.
Power sources dierent from PV sources can be easily
emulated by using the same architecture; it is only necessary to program the desired IV characteristic on the
FPAA.
Since both FPAA output and input signals are analog,
no DAC or ADC converters are needed. Therefore all
the drawbacks associated to the handling of digital signals are avoided.
The paper is organized as follows: Section 2 describes
the programming of the FPAA to generate the reference
signal and contains some preliminary results. Section 3
reports the description and the experimental results
obtained by using a Commercial Power Supply based
PVE which emulates a Solar World SW225 PV module
(http://www.solarworld-usa.com//media/www/les/datasheets/sunmodule-poly/sunmodule-solar-panel-225-polyds.pdf). Section 4 reports the description and the
experimental results of a Buck DCDC converter based
PVE which emulates the same PV module. Section 5
reports additional experimental results concerning the
emulation of PV sources which operate in mismatching
conditions. In Section 6, the conclusions are reported.
2. Generation of the reference signal ipan_ref(t)
2.1. FPAA programming
The block diagram of the proposed PVE architecture is
shown in Fig. 1. It consists of a power stage (DC/DC converter) whose output current is regulated by means of a
proper controller. The FPAA input signals are: ipan(t)
(PV output current), vpan(t) (PV output voltage) and Sin(t)
(time varying signal which is proportional to the desired
irradiance). The FPAA output signal is ipan_ref(t). The input
signal of the controller is represented by the error signal
ipan_ref(t)ipan(t). Once the desired PV module type has been
selected, a set of 7 parameters must be provided to the
FPAA programming tool. Such a set, as indicated in

ipan(t)

DC-DC
POWER SUPPLY

vpan(t)

CURRENT
CONTROLLER

vpan(t)

L
O
A
D

ipan(t)

Sin(t)

FPAA
ipan_ref(t)

{ IscSTC, VocSTC, IMPP, VMPP, Tmodule, V, I}

Fig. 1. Block diagram of the proposed architecture.

M. Balato et al. / Solar Energy 123 (2016) 102115

Fig. 1, is composed by the following parameters which are


typically provided by photovoltaic modules manufacturers:
STC short circuit current (IscSTC), STC open circuit voltage
(VocSTC), STC Maximum Power Point voltage and current
(VMPP and IMPP), voltage and current temperature
coecients aV ([%/K]) and aI ([%/K]). Without loss of
generality, in the following we will refer to Solar
World SW225 PV modules whose characteristics are
reported in Table 1 (http://www.solarworld-usa.com//media/www/les/datasheets/sunmodule-poly/sunmodule-solar-panel-225-poly-ds.pdf).
The single diode circuital model of a PV module and its
typical IV curve are shown in Fig. 2 (Amrouche et al.,
2012). The values of the circuital components appearing
in Fig. 2 can be found on the basis of the PV module characteristics and of the values of ambient temperature Tamb
and solar irradiance S (Kadri et al., 2012; Eichker, 2003).
Eqs. (1)(5) rule the working of the circuit of Fig. 2:
ipan t iph t  id t  iRp t
i
St h
aI
 T module  T STC
 1
iph t I scSTC 
S
100
 v tSTC 
d
id t I sat  e V T  1
vpan t Rs ipan t
Rp

3 1 EgSTC Eg 
T module
k  T STC T module
I satSTC
e
T STC

1
2
3

iRp t

I sat

where S is the irradiance in W/m2, Tmodule is the module


temperature in K (in the following it has been assumed
Tmodule = 313 K), Isat is the diode reverse bias saturation
current, VT is the thermal voltage, IsatSTC is the STC diode
reverse bias saturation current, vd is the voltage across the
diode, k is the Boltzmann constant, EgSTC is the STC band
gap in eV, Eg is the band gap at Tmodule in eV, TSTC = 298 K,
SSTC = 1000 W/m2. In the sequel, with the symbol Voc
we will indicate the open circuit voltage at Tmodule given
by the following expression (Eichker, 2003):
h
i
aV
 T module  298
V oc V ocSTC  1
6
100
In order to emulate the operation of the PV module, the
above equations need to be implemented on the FPAA.
The adopted FPAA is the AN231E04 by ANADIGM
(http://www.anadigm.com/an231e04.asp). Such a device

Table 1
Solar world SW225 PV module characteristics in STC
(1000 W/m2, 25 C, AM 1.5).
VocSTC
IscSTC
VMPP
IMPP
aI
aV

36.8 V
8.17 A
29.5 V
7.63 A
0.034%/K
0.34%/K

105

can be programmed by means of the software ANADIGM


DESIGNER 2 (http://www.anadigm.com/anadigmdesigner2.
asp). A schematic block diagram of the architecture which
has been implemented in the FPAA is shown in Fig 3.
As it is evident from Fig. 3, only some gains, adders and
a Look Up Table (LUT) are necessary. Such blocks can
be easily obtained by using the Congurable Analog
Blocks (CABs) composed by sets of switched capacitors,
OP-AMPs, Conguration SRAMs, etc. which are available
in any FPAA. The CABs can be easily arranged by means
of the user programming interface. In the following, the
meaning and the role of the various blocks of Fig. 3 will
be explained. Preliminarily, it is worth noting that the
signals which are marked in Fig. 3 with an asterisk symbol
simply are scaled versions of the original signals. As an
example, v*pan(t) is obtained from the original voltage
vpan(t) by means of a proper scaling. Such a scaling is necessary because of the limited range of the FPAA signals. In
particular, the allowed range of the input and output signals is [0, 3]V. Instead, internally, the FPAA can work only
with signals belonging to the range [3, 3]V. Indeed, for
simplicity, for the internal signals (i*r (t), i*ph(t), i*d(t), i*pan(t))
the same scaling of the input and output signals has been
considered so that all the signals have been conned in
the range [0, 2.8]V. 2.8 V has been chosen, instead of 3 V,
in order to ensure the right working of the FPAA, with
enough margin with respect to the actual internal upper
limit. Such a choice leads to a negligible accuracy reduction, in fact, as it will be evident in the following, the
obtained results are very accurate for the considered
application. The only internal signal for which the whole
range [3, 3]V needs to be considered during the scaling
process is v*d(t), that is the scaled version of vd(t). The
reason of such a need will be explained in the following.
The scaled signals are dened as:
vpan t
V MAX
ipan t
ipan t 2:8 
I MAX
St
S  t 2:8 
S STC
v
d t
9
vd t 12 
V OC
id t
id t 2:8 
I MAX
iR t
iRp t 2:8  p
I MAX
i
ph t
iph t 2:8 
I MAX
ipan ref t
ipan ref t 2:8 
I MAX

vpan t 2:8 

7a
7b
7c
7d
7e
7f
7g
7h

where VMAX = 48 V and IMAX = 15 A are the maximum


allowed values of the power stage output voltage and current. The LUT is used in order to get the IV characteristic

106

M. Balato et al. / Solar Energy 123 (2016) 102115


ipan
ipan

Rs

iRp

id

iph

Rp

vd

vpan

(a)

(b)

Vpan

Fig. 2. Single diode circuit model of a PV module (a) and corresponding IV curve (b).

vpan*(t)

vpan(t)

G1

2.8/VMAX

ipan(t)

ipan*(t)

vd*(t)

G2

2.8/IMAX

Look Up
Table

id*(t)

G3

+
+

G4
*

ipan_ref*(t)
Filter

S (t)

S(t)

iRp*(t)

2.8/S STC

G5

iph (t)

FPAA
Fig. 3. Schematic block diagram of the architecture which has been implemented in the FPAA.

of the diode of Fig. 2. The LUT provides a specied output


voltage in response to the value of the input voltage. The
FPAAs LUT can implement a user specied voltage transfer function with 256 quantization steps. Such a transfer
function is provided to the FPAA by uploading a proper
.csv le where a single column containing 256 samples
of the output variable (in our case i*d) are recorded. Such
samples represent the output values in correspondence of
256 values, taken in ascending order, of the input signal
(in our case v*d) belonging to the range [3, 3]V. The range
of the input signal of the LUT is xed and cannot be changed by the user. The IV diode characteristic depends on
the PV module temperature. In the proposed emulator,
during the generation of the .csv le provided to the
LUT, such a temperature is considered as a xed parameter
(in particular Tmodule = 313 K has been assumed). Of
course, as discussed in Section 1, such an assumption does
not represent any loss of generality. In fact it is worth noting that, although the PV module temperature has a great
inuence on the shape and on the values of the IV characteristic, in all realistic scenarios, PV modules are characterized by a more or less large thermal inertia so that their
temperature can be considered constant in nearly all short
or medium duration experimental tests. This is the reason
why nearly all the PV emulators presented in the literature
assume a xed PV module operating temperature. What is
really important in a PVE is the capability of taking into
account dynamic variations of the irradiance rather than
of the ambient temperature. In any case, if a dierent operating temperature needs to be considered, the LUT can be
easily and quickly reloaded by means of a new .csv le.

Of course, in case of experimental activities characterized by longer durations, also the module temperature
needs to be handled by the FPAA as a time-varying signal.
In such a case, the number of CABs which are available in
a single FPAA are not enough and two FPAAs are indeed
necessary. But, we explicitly remark here, that the increase
of complexity and of hardware resources required in order
to take into account time-varying temperatures is not at all
necessary in nearly all realistic cases. Therefore, without
any further specication, all the results shown in the following sections refer to a constant module temperature
equal to 313 K.
It is worth noting that, as shown in Fig. 3, the irradiance
S(t) is instead entered in the FPAA as a time varying voltage signal. As shown in the following, the FPAA based
control architecture is fully able to track realistic irradiance
temporal variations.
The voltage vd(t) across the diode can assume only values belonging to the interval [0, Voc]. In order to improve
the accuracy, by lowering the quantization step, it has been
assumed that id(t) = 0 for vd(t)I[0, Voc/2] and id(t) > 0 for
vd(t)I[Voc/2, Voc]. Therefore it is necessary to store in the
LUT only the values of id(t) for vd(t)I[Voc/2, Voc] since
the LUT will automatically output a value equal to zero
for vd(t)I[0, Voc/2]. In this way, the LUT resolution DVd
changes from the value reported in Eq. (8) to the value
reported in Eq. (9):
V OC
if vd t 2 0; V OC  ! DV d
256


V OC
V OC =2 V OC

; V OC ! DV d
if vd t 2
256
2
512

8
9

M. Balato et al. / Solar Energy 123 (2016) 102115

The above assumption means that we are considering


equal to zero
 V ocall the
 diode currents lower than
2V T
id tjV oc I sat  e  1 6:85  105 A. Such a thresh-

107

The model shown in Fig. 3 has been implemented on the


FPAA by loading, by means of the software ANADIGM
DESIGNER 2, the conguration shown in Fig. 5.

old is reasonable for practical purposes since, in the


considered case, at vd(t) = Voc/2 the photocurrent iph(t) is
nearly equal to 8.17 A, current iRp = 7.4 s_ 102 A, current
iRs = 8.096 A and therefore id tjV oc can be considered neg2
ligible with respect to all other currents owing in the circuit of Fig. 2.
In order to better clarify the meaning of the above
assumption, the output of the LUT is shown in Fig. 4 in
correspondence of an input voltage belonging to the range
[3, 3] V. In particular, Fig. 4(a) has been obtained by
using Eq. (10a) while Fig. 4(b) has been obtained by using
Eq. (11a). It is evident that, by using Eq. (11a), a considerably greater accuracy can be obtained.
vd t
 3 2 3; 3V
V OC
8


< 8vd t 2 0; V OC
! vd t 2 3; 0V
2
: 8v t 2 V OC ; V ! v t 2 0; 3V
d
OC
d
2

if vd t 6 

vd t
 9 2 9; 3V
V OC
8


< 8vd t 2 0; V OC
! vd t 2 9; 3V
2
: 8v t 2 V OC ; V ! v t 2 3; 3V
d
OC
d
2
if vd t 12 

10a
10b
11a
11b

On the basis of the above considerations, the gains of


the scheme in Fig. 3 assume the following values:
G1

12  V MAX
2:8  V oc

12a

G2

12  Rs  I MAX
2:8  V oc

12b

G3

V MAX
Rp  I MAX

12c

Rs
G4
Rp
h
i I
aI
scSTC
G5 1
 T module  T STC 
100
I MAX

12d
12e

2.2. Preliminary test results


A preliminary test activity has been carried out on the
FPAA. The results are represented by the oscilloscope
screenshots of Fig. 6. In particular, a periodic (frequency
100 Hz) ramp signal growing from 0 V to 2.15 V (corresponding to a PV module ramp voltage vpan(t) from 0 to
VOC) has been applied at the FPAA input v*pan(t) of
Fig. 5. The output signal i*pan_ref(t) has been connected to
the FPAA input port i*pan(t). Moreover, in Fig. 6(a), the
input signal S*(t) is equal to 2.8 V (corresponding to S(t)
= 1000 W/m2), while in Fig. 6(b) the input signal S*(t) is
equal to 1.12 V (corresponding to S(t) = 400 W/m2). In
Fig. 6, the input ramp v*pan(t) and the corresponding output
signal i*pan_ref(t) of the FPAA are reported in the two considered irradiance conditions. The obtained results are very
accurate as conrmed by Fig. 7 where the comparison
among the emulated curves (non scaled versions) and the
theoretical curves are shown. It can be observed that emulated and theoretical curves are nearly overlapped, thus
conrming the goodness of the characteristics provided
by the FPAA. The main motivation behind the preliminary
test activity which has been discussed in this section is to
check the results which could be obtained in an ideal
PVE characterized by a perfect current control in the
DC/DC converter representing its power stage. In fact,
the FPAA output signal i*pan_ref(t) has been directly fed
back to the FPAA input port i*pan(t). In a real application
instead, the actual input signal at the input port i*pan(t),
rather than i*pan_ref(t), would be the scaled version of the
actual current sensed at the output of the power stage.
But, in case of perfect control, such a current would just
be perfectly coincident with i*pan_ref(t).
3. Experimental results obtained by means of a commercial
power supply
In order to test the potentialities of the FPAA based PV
emulator, the system shown in Fig. 1 has been assembled in
the laboratory (Fig. 8). In particular, the set composed by

Fig. 4. Comparison among the theoretical diode characteristic and the LUT output characteristic obtained by using Eq. (10a) (a) and Eq. (11a) (b).

108

M. Balato et al. / Solar Energy 123 (2016) 102115

vpan*(t)

ipan*(t)

S*(t)

ipan_ref*(t)
Fig. 5. Architecture of the Congurable Analog Blocks (CABs) of the FPAA.

Fig. 6. IV curves provided by the FPAA and displayed by the oscilloscope, (a) S = 1000 W/m2, Tmodule = 313 K and (b) S = 400 W/m2, Tmodule = 313 K.

Fig. 7. Comparison among emulated and theoretical curves, (a) S = 1000 W/m2, Tmodule = 313 K; (b) S = 400 W/m2, Tmodule = 313 K.

the DC/DC converter and the current controller has been


obtained by means of a commercial power supply. The
adopted power supply is a Kepco BOP 36-12M, a device
able to work in all four quadrants of the current voltage
plane. It is a linear power supply with two bipolar control
channels (voltage or current mode), selectable and individually controllable either by front panel controls, or by

remote signals. The positive and negative current or


voltage limit points can be manually set or remotely
programmed simultaneously or individually. The BOP
can act as either a source or a sink (http://www.
kepcopower.com/support/opmanls.htm#bop). It is worth
noting that, any other equivalent controllable power
supply, which in case is available in the laboratory, can

M. Balato et al. / Solar Energy 123 (2016) 102115

109

Fig. 8. Commercial power supply based PV emulator system.

be used. Of course, a PV emulator architecture based on


the use of a commercial power supply is not cheap, but it
allows a consistent saving of time and eorts. Therefore
we explicitly remark here that, before showing the experimental results obtained by using a homemade Buck DC/
DC converter, it is useful also to preliminarily show that
the use of a FPAA grants great exibility and great simplicity of implementation whichever the IV characteristic to
be emulated. It is enough to obtain by the FPAA, after a
few, easy and fast programming steps, a proper analog output reference signal which can be used to control the power
supply.
The picture of the whole system which has been assembled in the laboratory is shown in Fig. 8. In such a system,
two commercial power supplies have indeed been used. The
rst one (POWER SUPPLY PS1, a Kepco BOP 36-12M)
represents the power stage of the PVE and it is used as a
voltage controlled current generator. The second one
(POWER SUPPLY PS2, a Kepco BOP 100-10MG
(http://www.kepcopower.com/support/opmanls.htm#bop)
is instead used as a controlled load in order to scan the
whole IV characteristic of the PV module to emulate.
In addition, in Fig. 8 it is also possible to observe the
presence of a Teledyne Lecroy oscilloscope for the visualization of the emulated PV characteristic, a National
Instruments data acquisition board for uploading the
obtained experimental data in Labview, a signal generator
which is used to reproduce dierent dynamic irradiance
conditions, a LEM current transducer and an OPAMP
based FPAA-Power Supply interface (which is used to
adapt the allowed ranges of the FPAA output and of the
PS1 input) with their DC voltage supply.
The power supply PS2 has been used to apply a periodic
ramp voltage at the output of PS1. Such a voltage grows

from 0 V to VOC with a frequency of 1 Hz. It has been veried that the adoption of scan frequencies higher than a
few Hz unavoidably leads to the distortion of the obtained
current waveform. This is of course a drawback associated
to the more or less limited bandwidth of PS1, since it has
been previously shown in Section 2 that the FPAA is
instead fully able to correctly work at scan frequencies up
to 100 Hz.
The oscilloscope screenshots of Fig. 9 show the scaled
scan voltage v*pan(t) (periodic ramp provided by PS2 and
characterized by a repetition frequency equal to 1 Hz)
and the corresponding scaled sensed output current i*pan(t)
of the PVE (PS1). In particular, Fig. 9(a) refers to a static
irradiance level S(t) = 1000 W/m2, while in Fig. 9(b) it is S
(t) = 400 W/m2. In practice, the results shown in Fig. 9 are
the waveforms which have been obtained in the power
stage and which directly correspond to the FPAA signals
shown in Fig. 6. The obtained results are very accurate
as conrmed by Fig. 10 where the comparison among the
unscaled emulated curves (PS1 output) and the theoretical
curves are shown. Also in this case it can be observed that
the emulated and the theoretical curves are nearly superimposed, thus conrming the goodness of the proposed system. It is worth noting that, in nearly all the papers
presented in the literature and dealing with PV emulators,
usually, to the best of the authors knowledge, the comparison among theoretical results and experimental results is
carried out by showing the position, in the IV plane, of
a more or less limited number of experimental points with
respect to the continuous theoretical IV characteristic. In
the case shown in Fig. 10 instead, not only the theoretical
IV curve but also the experimental IV curve is
continuous. This conrms the excellent dynamic behavior
of the proposed system. In other words, the proposed

110

M. Balato et al. / Solar Energy 123 (2016) 102115

Fig. 9. IV curves emulated by the power supply PS1 and displayed by the oscilloscope, (a) S = 1000 W/m2, Tmodule = 313 K, (b) S = 400 W/m2,
Tmodule = 313 K).

Fig. 10. Comparison among emulated and theoretical IV curves, (a) S = 1000 W/m2, Tmodule = 313 K, (b) S = 400 W/m2, Tmodule = 313 K.

system is very fast so that it is not necessary to wait for


more or less relatively long transients when changing from
a steady-state operating point to another steady-state operating point and the proposed system is able to track a relatively fast time-varying PV module ramp voltage signal
(up to 1 Hz).
Also the case of time varying irradiance levels has been
of course examined, in order to assess the dynamic capabilities of the proposed emulator. The results of the corresponding experimental tests are shown in Fig. 11. In
particular, the results shown in Fig. 11(a) refer to an irradiance characterized by a periodic (frequency 0.1 Hz) ramp
like time domain behavior which leads to an irradiance
change from 200 W/m2 to 1000 W/m2 in 10 s (constant
irradiance rate of change equal to 80 W/(m2 s)). The results
shown in Fig. 11(b) refer instead to an irradiance characterized by a periodic (frequency 0.1 Hz) square wave signal
with a lower level equal to 400 W/m2 and an higher level
equal to 1000 W/m2. The oscilloscope screenshots of
Fig. 11 show both the scaled input irradiance signal S*(t)
(blue curve1) and the PV emulator output current i*pan(t)
(green curve) which is obtained while the PV emulator output voltage is periodically swept (frequency 0.5 Hz) by
means of PS2 between 0 and Voc.

1
For interpretation of color in Fig. 11, the reader is referred to the web
version of this article.

In both cases the obtained results are in perfect agreement with the expected theoretical results; the comparison
among experimental and theoretical curves has not been
reported here for the sake of brevity.
4. Experimental results obtained by means of a Buck DC/DC
converter
The results which are reported in this section refer to a
power stage of the PVE which has been obtained by means
of a current controlled Buck DC/DC converter (which has
been built in the laboratory) rather than by a commercial
power supply. The resulting PVE system is of course
cheaper, with respect to the adoption of a commercial
power supply (Table 2), but of course more time and additional eorts have been required in order to design and
build both the power stage and the control circuitry of
the Buck converter. The block diagram of the Buck based
PVE is shown in Fig. 12. The objective of the PWM control
circuitry is of course to get an output current i*pan(t) which
correctly tracks the reference signal i*pan_ref(t) generated by
the FPAA device.
The design of the PI compensators transfer function Gc
and a preliminary analysis of the scheme of Fig. 12 are
reported in Barra et al. (2014). The experimental setup is
shown in Fig. 13. It can be observed that two FPAAs have
been used. The rst one (FPAA PV Reference, FPAA1) is
used to generate the reference signal i*pan_ref(t) by adopting

M. Balato et al. / Solar Energy 123 (2016) 102115

111

Fig. 11. Constant irradiance rate of change equal to 80 W/(m2 s) (a), periodic square wave irradiance with a lower level equal to 200 W/m2 and an higher
level equal to 1000 W/m2 (b).

Table 2
Costs of the main commercial photovoltaic emulators and of the proposed emulator.
Model

Cost per unit

Total cost

FPAA PVE based on


Buck DC/DC converter
FPAA PVE based on
Kepco BOP 36-12M
Keysight (Agilent) E4360 modular solar array simulators (http://literature.cdn.keysight.com/litweb/pdf/
5989-8485EN.pdf)
Magna-Power (PPPE) (http://www.magna-power.com/les/datasheet/pppe/datasheet_pppe_2.0.pdf)
Emulation software for Magna-Power DC power supply

11 (FPAA)
100 (Buck)
11 (FPAA)
3200 (Kepco)
10,000

111

441 (PPPE software)


3000 (Magna-power DC
power supply)
7000

Chroma ATE 62150H programmable DC power supply for solar array simulation (http://www.
mhzelectronics.com/ebay/manuals/chroma_62000h-series_power_supply_datasheet.pdf.)

3211
10,000
3441
7000

ipan (t)
L

Vg

v pan (t)

Current
sensor

Load
Voltage
sensor

v pan (t)
Transistor
gate driver

ipan * (t)

FPAA
Compensator

(t)

v c(t)
PWM

Gc

e(t)

ipan _ref (t)


-

Fig. 12. PWM current control of the Buck PVE.

the scheme of Fig. 5. The second one (FPAA compensator,


FPAA2) is instead used to implement the PI compensator
Gc and the PWM block in order to generate the duty cycle
signal d(t) driving the MOSFET of the Buck. The FPAA2
is programmed by adopting the scheme shown in Fig. 14.
Of course the PI compensator and the PWM block could
also be obtained by using discrete analog components
rather than the FPAA2. The choice to use the FPAA2
has been once again dictated by the possibility to get the
desired control blocks in a very easy way, after only some
trivial and rapid FPAA programming steps. The results
shown in Fig. 15 (white curves) have been obtained by
uploading in the Labview environment the experimental

results obtained by setting at the output port of the Buck


converter, by means of PS2, a ramp voltage growing from
0 to Voc in 1 s. The black curves of Fig. 15 represents
instead the theoretical results. In particular, Fig. 15(a)
refers to a static irradiance level S(t) = 1000 W/m2, while
in Fig. 15(b) it is S(t) = 600 W/m2. A very good matching
among experimental and theoretical curves is evident.
5. Experimental results concerning mismatching operating
conditions
The adoption of a FPAA based PVE control circuitry
makes quite easy also the emulation of IV characteristics

112

M. Balato et al. / Solar Energy 123 (2016) 102115

Fig. 13. Buck DCDC converter based PV Emulator.

Fig. 14. Implementation of the PI compensator Gc and of the PWM block on the FPAA2.

Fig. 15. Experimental and theoretical IV curves; S = 1000 W/m2, Tmodule = 313 K (a); S = 600 W/m2, Tmodule = 313 K (b).

M. Balato et al. / Solar Energy 123 (2016) 102115

associated to mismatching conditions. As an example, with


the architecture which has been proposed in this paper, it is
very easy to emulate a PV array composed by N PV modules connected in parallel and working in mismatching conditions. It is enough to use the instantaneous sum of the N
reference signals provided by the N dierent FPAAs as a
reference signal for the output current of a unique PVE
power stage. In other words, no change of the power stage
is required in order to emulate N PV modules connected in
parallel, under mismatching conditions. This is of course
true provided that the voltage and/or the current ratings
of the power stage are enough high for the applications
under study, that is, provided that the power stage is able
to emulate the same N PV modules connected in parallel,
under uniform operating conditions. Instead, in order to
emulate a PV array composed by N PV modules connected
in series and working in mismatching conditions, it is necessary to use N FPAAs programmed with a dual implementation. In fact, while the implementation which has
been proposed in this paper is based on the relation
ipan_ref = f(vPAN, iPAN), the dual implementation, useful
in order to emulate a PV array composed by N series connected PV modules, must be based on the relation
vpan_ref = g(vPAN, iPAN), where vpan_ref is the reference signal for the output voltage of the PVE power stage. Such
an implementation is of course not discussed in this paper
for the sake of brevity and also because of the fact that it
can be easily derived from the original implementation by
means of obvious dual principles and considerations.
With such a dual implementation, it is very easy to emulate a PV array composed by N PV modules connected in
series and working in mismatching conditions. It is enough
to use the instantaneous sum of the N reference signals provided by the N dierent FPAAs as a reference signal for the
output voltage of a unique PVE power stage.
In conclusion, the emulation of N PV modules under
non-uniform operating conditions does not imply any substantial additional practical diculty. In both cases of parallel connection of N PV modules under mismatching
conditions or of series connection of N PV modules under

113

mismatching conditions an analog adder is additionally


required. It is needed just in order to sum the signals which
are provided by the N FPAAs and which represent the N
PV currents, in the case of parallel connection, or the N
PV voltages, in the case of series connection. Obviously,
a proper scaling of the signals is also needed in order to
comply with the input range of the control circuitry of
the power stage.
In this section experimental results concerning the emulation of non-uniform operating conditions are reported.
In particular, two test cases have been considered with reference to a couple of PV modules which operate in mismatching conditions. One PV module (module A)
operates at an irradiance level SA and at a module
temperature TA. The other PV module (module B) operates
at an irradiance level SB and at a module temperature TB.
SA = 1000 W/m2, TA = 363 K, SB = 500 W/m2, TB =
303 K. In Fig. 16 the oscilloscope screenshots containing
the output signals i*pan_ref(t) of the two FPAAs emulating
the characteristics of modules A and B are reported
together with the input ramp v*pan(t) used to scan such
characteristics. The rst test which has been performed
concerns the parallel connection of such two modules. In
Fig. 17(a) the oscilloscope screenshot with the reference
signal i*pan_ref_tot(t) (sum of the two current reference signals
which are provided by the two FPAAs) together with the
ramp voltage v*pan(t) used to carry out the scan is reported.
The second test case concerns the series connection of the
two modules. In Fig. 17(b) the oscilloscope screenshot with
the output signal v*pan_ref_tot(t) (sum of the two voltage reference signals which are provided by the two FPAAs)
together with the ramp signal i*pan(t) used to carry out the
scan is reported. As written above, i*pan_ref_tot(t) and
v*pan_ref_tot(t) can be respectively used as reference signals
for the output current (in the case of the parallel connection) or for the output voltage (in the case of the series connection) of the PVE power stage. The experimental IV
curves (Fig. 17) are in very good agreement with the
corresponding theoretical curves. The comparison is not
reported here for the sake of brevity.

Fig. 16. Oscilloscope screenshots with the IV curves provided by the two FPAAs, (a) module A, SA = 1000 W/m2, TA = 363 K, (b) module B,
SB = 500 W/m2, TB = 303 K.

114

M. Balato et al. / Solar Energy 123 (2016) 102115

Fig. 17. Oscilloscope screenshots, (a) IV curve of the parallel connection of modules A and B, (b) VI curve of the series connection of modules A and B.

6. Conclusions
In this paper, a new fully dynamic PVE has been proposed. Its main feature is represented by the use of a FPAA
on which the desired IV characteristic can be implemented. The main advantages of the proposed solution
are the following ones:
both static and dynamic irradiance conditions can be
emulated;
no numerical interpolations and no storage of big
amount of data in memory are required;
the FPAA is characterized by a great ease of reconguration and programming with respect to FPGA or DSP
based implementations;
no DAC or ADC converters are needed;
not only uniform but also mismatching operating conditions can be easily emulated;
power sources dierent from PV sources can be easily
emulated by using the same architecture.
The obtained experimental results allow to assess the
validity of the proposed FPAA based PVE architecture.
A further advantage of the proposed PVE is represented
by its quite low cost. For the sake of completeness, in
Table 2 the costs of the main commercial devices and of
the proposed PVE are reported.
References
Al Nabulsi, A., Dhaouadi, R., 2012. Eciency optimization of a DSPbased standalone PV system using fuzzy logic and dual-MPPT control.
IEEE Trans. Industr. Inf. 8 (3), 573584.
Amrouche, B., Guessoum, A., Belhamel, M., 2012. A simple behavioural
model for solar module electric characteristics based on the rst order
system step response for MPPT study and comparison. Appl. Energy
91 (1), 395404.
Balakishan, C., Sandeep, B., 2014. Development of a Microcontroller
Based PV emulator with current controlled DC/DC buck converter.
Int. J. Renew. Energy Res. 4 (4), 10491055, November.
Balato, M., Vitelli, M., 2014. Optimization of distributed maximum power
point tracking PV applications: the scan of the power vs. voltage input
characteristic of the inverter. Int. J. Electr. Power Energy Syst. 60,
334346.

Barra, F., Balato, M., Costanzo, L., Gallo, D., Landi, C., Luiso, M.,
Vitelli, M., 2014. Dynamic and Recongurable Photovoltaic Emulator
Based on FPAA. In: 20th IMEKO TC4 International Symposium and
18th International Workshop on ADC Modelling and Testing
Research on Electric and Electronic Measurement for the Economic
Upturn, Benevento, Italy, September 1517, 2014.
Chavarria, J., Biel, D., Guinjoan, F., Poveda, A., Masana, F., Alarcon, E.,
2014. FPGA-based design of a step-up photovoltaic array emulator for
the test of PV grid-connected inverters. In: 2014 IEEE 23rd International Symposium on Industrial Electronics (ISIE), pp. 485490.
Di Piazza, M.C., Vitale, G., 2010. Photovoltaic eld emulation including
dynamic and partial shadow conditions. Appl. Energy 87 (3), 814823,
March.
Dolan, D.S.L., Durago, J., Tauk, 2011. Development of a photovoltaic
panel emulator using Labview. In: 2011 37th IEEE Photovoltaic
Specialists Conference (PVSC), pp. 17951800.
Eichker, U., 2003. Solar Technologies for Building. WILEY, ISBN: 0-47148637-X.
Erikson, R.W., Maksimovic, D., Fundamentals of Power Electronic,
second ed., Kluwer Academic Publishers, ISBN: 0-7923-7270-0.
Feng, W., Xinke, W., Lee, F.C., Zijian, W., Pengju Kong, K.,
Fang, Z., 2014. Analysis of unied output MPPT control in
subpanel PV converter system. IEEE Trans. Power Electron. 29
(3), 12751284.
Gadelovits, S., Sitbon, M., Kuperman, A., 2014. Rapid prototyping of a
low-cost solar array simulator using an o-the-shelf DC power supply.
IEEE Trans. Power Electron. 29 (10), 52785284.
http://literature.cdn.keysight.com/litweb/pdf/5989-8485EN.pdf.
http://www.accuratesolarpower.com/PVE_User_Guide_V1.5.pd.
http://www.anadigm.com/an231e04.asp.
http://www.anadigm.com/anadigmdesigner2.asp.
http://www.kepcopower.com/support/opmanls.htm#bop.
http://www.magna-power.com/les/datasheet/pppe/datasheet_pppe_2.0.
pdf.
http://www.mhzelectronics.com/ebay/manuals/chroma_62000h-series_
power_supply_datasheet.pdf.
http://www.solarworld-usa.com/~/media/www/les/datasheets/sunmodulepoly/sunmodule-solar-panel-225-poly-ds.pdf.
Ickilli, D., Can, H., Parlak, K.S., 2012. Development of a FPGA-based
photovoltaic panel emulator based on a DC/DC converter. In: 2012
38th IEEE Photovoltaic Specialists Conference (PVSC). pp. 1417
1421.
Kadri, R., Andrei, H., Gaubert, J., Ivanovici, T., Champenois, G., Andrei,
P., 2012. Modeling of the photovoltaic cell circuit parameters for
optimum connection model and real-time emulator with partial
shadow conditions. Elsevier Energy 42, 5767.
La Manna, D., Li Vigni, V., Riva Sanseverino, E., Di Dio, V., Romano,
P., 2014. Recongurable electrical interconnection strategies for
photovoltaic arrays: a review. Renew. Sustain. Energy Rev. 33, 412
426.

M. Balato et al. / Solar Energy 123 (2016) 102115


Liu, Y., Li, M., Ji, X., Luo, X., Wang, M., Zhang, Y., 2014. A
comparative study of the maximum power point tracking methods for
PV systems. Energy Convers. Manage. 85, 809816, September.
Lu, D.C., Nguyen, Q.N., 2012. A photovoltaic panel emulator using a
buck-boost DC/DC converter and a low cost micro-controller. Sol.
Energy 86, 14771484.
Martin-Segura, G., Lopez-Mestre, J., Teixido-Casas, M., Sudria-Andreu,
A., 2007. Development of a photovoltaic array emulator system based
on a full-bridge structure. In: 2007 9th International Conference on
Electrical Power Quality and Utilisation (EPQU), pp. 16.
Mellit, A., Mekki, H., Messai, A., Salhi, H., 2010. FPGA-based
implementation of an intelligent simulator for stand-alone photovoltaic system. Expert Syst. Appl. 37 (8), 60366051.
Midtgard, O.M., 2007. A simple photovoltaic simulator for testing of
power electronics. In: 2007 European Conference on Power Electronics
and Applications, pp. 110.
Miyatake, M., Veerachary, M., Toriumi, F., Fujii, N., Ko, H., 2011.
Maximum power point tracking of multiple photovoltaic arrays: a
PSO approach. IEEE Trans. Aerosp. Electron. Syst. 47 (1), 367380.
Mukerjee, A.K., Dasgupta, N., 2007. DC power supply used as photovoltaic simulator for testing MPPT algorithms. Renewable Energy 32
(4), 587592.
Nagayoshi, H., 2004. IV curve simulation by multimodule simulator
using IV magnier circuit. Sol. Energy Mater. Sol. Cells 82, 159167.
Qin, Shibin, Cady, S.T., Dominguez-Garcia, A.D., Pilawa-Podgurski, R.
C.N., 2014. A distributed approach to maximum power point tracking
for photovoltaic submodule dierential power processing. IEEE Trans.
Power Electron. 30 (4), 20242040.
Rana, A.V., Patel, H.H., 2013. Current controlled buck converter based
photovoltaic emulator. J. Indust. Intell. Inf. 1 (2), 9196.
Reisi, A.R., Moradi, M.H., Jamasb, S., 2013. Classication and comparison of maximum power point tracking techniques for photovoltaic
system: a review. Renew. Sustain. Energy Rev. 19, 433443, March.
Report IEA-PVPS T13-01, 2014. Review of Failures of Photovoltaic
Modules.
Roncero-Clemente, C., Romero-Cadaval, E., Minambres, V.M., GuerreroMartinez, M.A., Gallardo-Lozano, J., 2013. PV array emulator for
testing commercial PV inverters. Elektonika IR Elektotechnika 19
(10).

115

Sanchis, P., Lopez, J., Ursua, A., Gubia, E., Marroyo, L., 2007. On the
testing, characterization, and evaluation of PV inverters and dynamic
MPPT performance under real varying operating conditions. Progr.
Photovoltaics: Res. Appl. 15, 541556.
Schoeld, D.M.K., Foster, M.P., Stone, D.A., 2011. Low-cost solar
emulator for evaluation of maximum power point tracking methods.
Electron. Lett. 47 (3), 208209.
Selow, R., Lopes, H.S., Lima, C., A comparison of FPGA and FPAA
technologies for a signal processing application. In: 2009 International
Conference on Field Programmable Logic and Applications (FPL), pp.
230235, August 31 2009September 2 2009.
Sera, D., Mathe, L., Kerekes, T., Spataru, S.V., Teodorescu, R., 2013. On
the perturb-and-observe and incremental conductance MPPT methods
for PV systems. IEEE J. Photovoltaics 3 (3), 10701078, July.
Spagnuolo, G., Petrone, G., Lehman, B., Ramos Paja, C.A., Zhao, Ye,
Orozco Gutierrez, M.L., 2015. Control of photovoltaic arrays:
dynamical reconguration for ghting mismatched conditions and
meeting load requests. IEEE Ind. Electron. Mag. 9 (1), 6276.
Stracke, S., Wunderlich, R., Heinen, S., 2014. An all digital speed adaptive
maximum power point tracker for automotive photovoltaic applications. In: 3rd International Conference on Renewable Energy Research
and Applications (ICRERA), Milwakuee, USA, October 1922, 2014,
pp. 5560.
Tornez-Xavier, G.M., Gomez-Castaneda, F., Moreno-Cadenas, J.A.,
Flores-Nava, L.M., 2013. FGPA development and implementation
of a solar panel emulator. In: 2013 10th International Conference on
Electrical Engineering, Computing Science and Automatic Control
(CCE), pp. 467472, September 30 2013October 4 2013.
Woojoo, L., Younghyun, K., Yanzhi, W., Naehyuck, C., Pedram, M.,
Soohee, H., 2011. Versatile high-delity photovoltaic module emulation system. In: 2011 International Symposium on Low Power
Electronics and Design (ISLPED). pp. 9196.
Ye, J.Y., Ding, K., Reindl, T., Aberle, A.G., 2013. Outdoor PV module
performance under uctuating irradiance conditions in tropical
climates. Energy Proc. 33, 238247.
Zeng, Q., Song, P., Chang, L., 2002. A photovoltaic simulator based on
DC chopper. In: IEEE CCECE 2002: Canadian Conference on
Electrical and Computer Engineering, vol. 1, pp. 257261.

You might also like