Available online at www.sciencedirect.com

www.elsevier.com/locate/solener

Solar Energy 86 (2012) 1477–1484

A photovoltaic panel emulator using a buck-boost DC/DCconverter and a low cost micro-controller

Dylan D.C. Lu ⇑, Quang Ngoc Nguyen

School of Electrical and Information Engineering, The University of Sydney, NSW 2006, Australia

Available online 7 March 2012

Received 5 September 2011; received in revised form 8 February 2012; accepted 9 February 2012Available online 7 March 2012

Communicated by: Associate Editor Nicola Romeo

Abstract

In order to facilitate the design and testing of photovoltaic (PV) power systems, a PV emulator which models the electrical charac-teristic of a PV panel or array is needed. Among different approaches to modeling PV characteristic, namely the I–V curve, curve-fitting isa popular approach. Even though a single high-order polynomial equation may accurately represent the I–V curve, the process of der-ivation and implementation is rather complex. This paper hence proposes the use of piecewise linear approach which is easier to deriveand implement in a low-cost micro-controller. A two-switch buck-boost DC/DC converter is selected as the PV emulator and is analyzed.Experimental results on a hardware prototype of the proposed PV emulator are reported to show the effectiveness of the approach.Crown Copyright � 2012 Published by Elsevier Ltd. All rights reserved.

Keywords: DC/DC converter; Photovoltaic; Micro-controller; Emulator

1. Introduction

The demand of photovoltaic (PV) power system installa-tion has been increased over the past decade due to techno-logical improvement, better environmental awareness,lowered system costs, governmental initiatives, risingelectricity bills, etc. While these installed PV systems andproducts are operating properly, there are still ongoingissues to be investigated and solved. For example, reliabilityof PV power systems (Petrone et al., 2008), PV power gener-ation analysis (Ishaque et al., 2011; Paraskevadaki andPapathanassiou, 2011) and electricity network performance(van der Borg and Jansen, 2003) due to partial shading,development of power electronics interfaces (Marsh, 2011,2010), etc. All these research and development activitiesrequire a stable, repeatable and variable PV source for

0038-092X/$ - see front matter Crown Copyright � 2012 Published by Elsevie

doi:10.1016/j.solener.2012.02.008

⇑ Corresponding author. Tel.: +61 2 9351 3496; fax: +61 2 9351 3847.E-mail address: dylan.lu@sydney.edu.au (D.D.C. Lu).

design and testing. Hence there is a need of a PV generatoremulator.

The main task for a PV generator emulator is to repro-duce the I–V curve of a practical PV panel. There are differ-ent approaches to performing this task. In Nagayoshi(2004), a p–n photodiode is used and a DC power amplifierincreases the power level to match with that of a PV panel.However, this approach requires a light source and associ-ated circuit to reproduce the I–V curves of a PV panel. Infact, a power electronics converter can mimic the I–V curveaccurately with only a DC input voltage source (Mukerjeeand Dasgupta, 2007). In Khouzam and Hoffman (1996), aAC/DC buck converter is used as the PV emulator to emu-late a PV cell circuit model. However this approach requiresthe knowledge of the values of the parameters which areusually difficult to obtain. In fact, to model a PV panel,one may use the data available from the datasheet of thePV panel manufacturer and derive an analytical model torepresent the I–V curves (Ortiz-Rivera and Peng, 2005).

r Ltd. All rights reserved.

1478 D.D.C. Lu, Q.N. Nguyen / Solar Energy 86 (2012) 1477–1484

Look-up table and curve fitting are two popularapproaches to implementing I–V curves of a PV panel bythe power electronics converters. Look-up table wouldrequire a large memory storage of the micro-controller aslarge amount of panel data is stored if many I–V curvesat different conditions and with high accuracy are imple-mented. Hence to implement look-up table in a low costmicro-controller which has limited memory space is usuallydifficult. Curve-fitting approach in general uses one ormore polynomial equations to model an I–V curve andneeds a digital controller with fast computational speedto find the solution. While this method requires less mem-ory space, the non-linearity of the I–V curve requires theequations to be of higher orders which may increase thecomputational time substantially. A powerful DSP control-ler is usually needed to produce very fast and accurateresults (Zhang and Zhao, 2010). Also the derivation pro-cess of the polynomial equations for different conditionssuch as insolation and temperature is rather troublesome.In order to use curve-fitting efficiently on a low-costmicro-controller, this paper introduces a PV emulatorusing multiple simple linear equations to mimic an I–Vcurve of the PV panel. This approach reduces computa-tional time while maintaining sufficient accuracy and canbe implemented in a low-cost 8-bit micro-controller. Thepaper is organized as follows: Section 2 describes the circuitand operation principle of the proposed PV emulator. Sec-tion 3 reports the experimental results of the emulatorwhich models a BP Solar SX-10 PV panel. Section 4 dis-cusses the limitations of the emulator and followed bythe conclusions in Section 5.

2. Description of the PV emulator

2.1. System overview

The PV emulator, as shown in Fig. 1, consists of a DCinput source, Vin, a DC/DC converter for shaping theoutput I–V curves of the PV panel, a micro-controller forsensing the output voltage vpv and current ipv, calculationand sending duty cycle command, and a gate driver for

Fig. 1. Block diagram of the PV emulator.

amplifying the incoming duty cycle command suitable fordriving the power transistor (MOSFET in this case). Theoutput load RL is modeled as a variable resistor to repre-sent an equivalent resistance of a maximum power pointtracker (MPPT).

2.2. Mathematical modeling of a PV panel

Apart from measuring an actual PV panel, one can alsouse an analytical model to represent the data in the data-sheet from the manufacturer to obtain the I–V curves ofa specific PV panel. In Ortiz-Rivera and Peng (2005), theauthors have generated an analytical model for a PV panelwhich is adopted in this paper:

IðV Þ ¼ a � Imax � si

� 1� expV

bða � cþ 1� cÞðV max þ sV Þ� 1

b

� �� �ð1Þ

where a is the percentage of effective intensity of the light, b

is the characteristic I–V curve constant, c is the shading lin-ear factor, si is the rate of change with the temperature forthe current (A/�C), sV is the rate of change with the temper-ature for the voltage (V/�C) and Imax is the ideal maximumcurrent (when V = �1 at STC).

For this paper, a PV panel from BP Solar (Model: SX-10) is modeled. Assuming no shading and using a = 1 andothers values provided by the datasheet (BP Solar PC SX-10 data sheet, 2003), a numerical expression of this PVpanel can be found:

IðV Þ ¼ 0:65

1� e�1=b1� V

b� 21� 1

b

� �ð2Þ

Using the maximum power point condition at 16.8 V and0.59 A, the value of b can be calculated by (2) as0.085. At 25 �C, (2) can be further simplified to:

IðV Þ ¼ 0:65½1� eðV =1:785�11:7647Þ� ð3Þ

Similarly at 75 �C one can get:

IðV Þ ¼ 0:6711½1� eðV =1:445�11:7647Þ� ð4Þ

Fig. 2 shows the MATLAB plot of the I–V characteristiccurves of SX-10 PV panel Eqs. (3) and (4).

2.3. Two-line and multiple-line fitting approaches

To generate N number of fitting lines, N + 1 points fromthe curve need to be selected. To begin with, a two-lineapproach as shown in Fig. 3 is discussed. The two ends pointsfrom the curve are the open-circuit voltage (21 V, 0 A) andshort-circuit current (0 V, 0.65 A). The third point is selected(16.8 V, 0.62 A) as the maximum power point (MPP) of thecurve where the two lines converge. Therefore the two equa-tions which represent the two lines are expressed as

IðV Þ ¼ 0:65� 0:004V ð5ÞIðV Þ ¼ 2:94� 0:14V ð6Þ

Fig. 2. I–V characteristics of SX-10 at 25 �C and 75 �C.

Fig. 3. Two-line curve fitting approach.

Fig. 4. Five-line curve fitting approach.

D.D.C. Lu, Q.N. Nguyen / Solar Energy 86 (2012) 1477–1484 1479

To improve the accuracy of the curve-fitting method, fivelines as shown in Fig. 4 are used. Similarly, with six pointsselected, the five equations which represent the five lines aregiven by

IðV Þ ¼ 2:94� 9:2143e�4V ½for V ¼ 0–14� ð7ÞIðV Þ ¼ 0:8233� 0:0133V ½for V ¼ 14–16� ð8ÞIðV Þ ¼ 1:2633� 0:0408V ½for V ¼ 16–18� ð9ÞIðV Þ ¼ 2:1651� 0:0909V ½for V ¼ 18–19� ð10ÞIðV Þ ¼ 4:599� 0:2190V ½for V ¼ 19–21� ð11Þ

2.4. DC/DC converter

Among the basic converters, buck and buck-boostconverters are able to be implemented as the DC/DC

converter for the PV emulator. The design considerationis that the converter is able to sweep through the entirevoltage range of the PV panel. For a buck converter, aDC input voltage which is higher than the panel is requiredas it is a step-down converter. The buck-boost converter ismore flexible as it can perform both step-up and step-downfunctions. The boost converter which is a step-up con-verter, however, can only operate when the input voltageis lower than the output voltage hence it cannot reachdown to 0 V and cannot be used in this case.

3. Experimental setup and results

3.1. Design considerations and hardware description

To verify the proposed PV emulator for the BP SX-10model, a hardware prototype is built and tested. The sche-matic of the PV emulator circuit is shown in Fig. 5. The rea-son to select a two-switch buck-boost converter for thisimplementation is twofold. Firstly it can work with lowerinput voltage (12–15 V) that reduces the power loss whenthe input voltage is used to step down further for themicro-controller circuit (5 V). If a buck converter is used,the input voltage has to be higher than 21 V which is theopen-circuit voltage of the PV panel. Secondly, the outputvoltage is non-inverting as compared to the single-switchbuck-boost converter with inverting output. The advantageis that the output ground will be the same for the inputground and ground of other equipment which is connectedto the output of the PV emulator, e.g. a MPPT switchingconverter. Another advantage is that the use of either split-ting power supply or opto-coupler and associated circuit fornegative voltage feedback sensing, which added complexityand slowed down the response of the system, is eliminated.

The two-switch buck-boost converter consists of twopower MOSFETs (Q1 and Q2), two diodes (D1 and D2),an inductor (L1) and an output capacitor (C1), as shown

Fig. 5. Schematic of the proposed PV emulator based on a buck-boost converter.

1480 D.D.C. Lu, Q.N. Nguyen / Solar Energy 86 (2012) 1477–1484

in Fig. 5. The two power transistors Q1 and Q2 share thesame driving pulses from the Gate Driver block. Whenboth transistors are closed, the inductor is charged by theinput voltage. When the transistors are turned off, theinductor is discharged via two diodes (D1 and D2) to out-put. By controlling the duty cycle of the transistors as aresult of the linear equations calculation as stated in Sec-tion 2.2, the I–V curve of the PV panel is implemented.

For the micro-controller, a 8-bit PICAXE AXE-08Mchip is selected. It is a modified version of MicrochipPIC12F68 model. It has three ADC inputs and a PWMoutput pin which is easily configured by a single-line com-mand. The chip runs at 4 MHz and therefore it can operateeasily at 50 kHz switching frequency for the converter withsufficient accuracy and speed.

The inductor (L1) is chosen to operate in continuousconduction mode (CCM) as it produces less conductionloss. A ±20% maximum current ripple in the inductor isselected. At 50 kHz switching frequency, the minimuminductance Lmin to meet such requirement is at 15 V inputand 21 V output (open-circuit voltage)

Lmin ¼V in � Dfs � Di

¼ 15 � 0:58

50000 � 0:4 ¼ 435 lH ð12Þ

where Vin is the input voltage, D is the duty cycle of thepower switches, fs is the switching frequency, and Di isthe maximum current ripple.

In order to provide at least 12 V and floating gate drivefor the MOSFET, a high-side driver IR2117 is used. Sincethe PICAXE chip operates at 5 V but the IR2117 driver

requires at least 9.6 V input, a simple transistor inverteras a level lifting circuit is implemented. This small transis-tor (2N7000) requires only 2.5 V to drive. When a highpulse from PICAXE chip is generated, the transistor isturned on and pulled the output to ground. When a lowpulse is generated, the transistor is turned off and outputis risen up to Vcc (12–15 V in this design).

The programming for PICAXE is done on a free pro-gramming editor provided by PICAXE. Once the programis written it can be downloaded to the PICAXE AXE-08Mmicro-controller chip via a USB cable (AXE027) or RS-232cable (AXE028) connecting the computer and the chip(Download Socket in Fig. 5). This cable is not a normalUSB or RS-232 connector as it contains some electronicparts and it is pre-programmed. The programming lan-guage for PICAXE is similar to the BASIC language.The program flow chart is shown in Fig. 6. The programstarts with defining the symbols for voltage and currentmeasurements and for counters. Then it outputs a smallduty cycle to start the buck-boost converter. Once the con-verter operates, the program can take readings from thevoltage and current (V_sense and I_sense in Fig. 5) todetermine the operating point for the PV panel the con-verter mimics. Note that we are using fractional openvoltage MPPT algorithm so only a voltage reading isneeded to operate the converter to the desired operatingpoint. The current measurement is only used for overload-ing protection. The PICAXE-08M chip can contain maxi-mum of 8 linear equations so we have used 5 linearequations for 75 �C and 3 linear equations for 25 �C. The

Fig. 6. Program flow chart for implementation of 5-line for 75 �C and 3-line for 25 �C of the BP SX-10 PV panel.

Fig. 7. Output voltage ripple of the PV emulator (Y-axis: 200 mV/div;time scale: 10 ls/div).

Fig. 8. Comparison between experimental and theoretical results on two-line approach.

D.D.C. Lu, Q.N. Nguyen / Solar Energy 86 (2012) 1477–1484 1481

selection of what temperature to use is by providing a highor low signal to available ADC input of the PICAXE (Pin4). Here we selected signal high for 75 �C and low for25 �C. After the temperature is selected, the program willuse the voltage reading to locate the nearest linear equationto find out the operating point. A new duty cycle is thendetermined and produced from the PWM pin of the PIC-AXE and the converter duty cycle is updated. After thatthe program goes back to the voltage and current checkingprocess and repeats the procedure. The full program codefor implementing the five-line approach is shown in Appen-dix A.

3.2. Results

Fig. 7 shows the output voltage ripple of the PV emula-tor is less than 450 mV at 50 kHz switching frequency. The“On” and “Off” labels in the figure indicate the turn-oninstant and turn-off instant of the power transistors Q1

and Q2 for each switching period. The duration of “On”

period indicates the duty cycle of the transistors. And fromthe waveform it can be observed that the converter is oper-ating in continuous conduction mode (CCM) as the ripplehas only two stages, confirming our design in Section 3.1and Eq. (12). Note that the voltage spikes during theturn-on and turn-off instants and appeared beyond the hor-izontal cursors are due to the pick-up of electromagnetic

noise from the voltage probe. Minimizing the ground loopof the voltage probe will greatly reduce the pick-up. Fig. 8shows the measured results on the prototype using the two-line approach. The results are very close to the two operat-ing lines. Fig. 9 shows further results on the five-lineapproach at both 25 �C and 75 �C. Due to the memorylimitation of the micro-controller, only eight lines can beimplemented in a single program. Therefore Fig. 9 shows5 lines for 25 �C and 3 lines 75 �C. Nevertheless, the mea-sured results are closely matched with the theoreticaldesigned curves.

As shown in Fig. 10, the PV emulator reaches a maxi-mum efficiency of around 80% near maximum power pointvoltage for both temperature settings. The low efficiencyoccurs at lower voltages because the output power of thepower converter is small and the switching losses of powertransistors and diodes are dominant. When PV emulator

Fig. 9. Comparison between experimental and theoretical results on five-line approach and different temperature settings.

Fig. 10. Measured efficiency of the PV emulator at different temperaturesettings.

1482 D.D.C. Lu, Q.N. Nguyen / Solar Energy 86 (2012) 1477–1484

voltage increases and is moving towards the maximumpower point, the output power of the converter alsoincreases. Since the rate of switching losses only increasesslightly as compared to the rate of increase of outputpower, there is less switching losses proportionally to theoverall input power and the efficiency improves as outputpower increases as a result.

In order to demonstrate the usefulness of the PV emula-tor. A system is set up, as shown in Fig. 11, in which a buckconverter as a maximum power point tracker is connectedto the output of the PV emulator. The tracker is first testedusing the real PV panel outdoor. Fractional open circuitvoltage technique (Esram and Chapman, 2007; Ahmad,2010) is used as the MPPT algorithm in this case. Frac-tional open circuit voltage technique measures the open cir-cuit voltage of the PV panel at the start-up process. Andthe MPP is approximated by multiplying the measured

open voltage with a factor, usually between 0.7 and 0.78.The converter will alter the duty cycle in order to adjustthe panel voltage to be equal to this MPP value. Thismethod is simple as no input current of the PV panel isneeded. The limitation of this MPPT approach is that theMPP voltage is only an approximate value. Using the sameMPPT code, different operating conditions are tested toconfirm that the tracker can adapt to the change of theenvironment. Fig. 12 shows the capability of the trackerto reach the maximum power points at different tempera-tures and insolations.

The tracker has been tested with the PV panel emulator.The tracker has successfully tracked the MPP voltages at16.9 V for 25 �C and 13.7 V for 75 �C respectively whichare very close to the theoretical values. The results demon-strated that the performance of the PV panel emulatorreacts identically to the real PV panel which it models.

4. Discussions

Using multiple straight lines to model an I–V curve ofthe PV panel is fast and straight-forward. The low costPICAXE-08 M chip has four basic mathematical functions:addition, subtraction, multiplication and division, and theyare well suited for this implementation. To improve theaccuracy of modeling further, however, exponential expres-sions can be used and a more powerful micro-controllerneeded to be used. Also, as mentioned in Section 3.2, withlimited program memory (800 lines memory) of this micro-controller only eight lines can be implemented. But thePICAXE family has higher end micro-controller to imple-ment more number of lines, such as 40 � 2 with 3200 linesmemory.

The PICAXE-08 M chip has a pin dedicated to PWMgeneration. By using the PWMOUT function in theprogram, frequency and duty cycle are set using a singlecommand line. The resolution of the duty cycle increaseswith decreasing switching frequency. For instance, at50 kHz switching frequency, there are 80 steps. While at40 kHz, the steps increase to 100. Larger steps of duty cycleenable the converter to operate with smaller fluctuationwhen duty cycle has to be altered to adjust the output con-tinuously due to change of input or output condition, pro-vided the inductance and output capacitance haveincreased to maintain the same ripple current and ripplevoltage with decreasing switching frequency.

The power losses of the PV emulator are mainly due toconduction loss and switching loss. Conduction loss can bereduced by using better devices with smaller internal resis-tance. This includes smaller turn-on resistance for powertransistor and diode, and less core loss and less copper(winding) loss for the inductor. Switching loss can bereduced by improving the slew rate of turn-off and turn-on instances of the power transistor and of the powerdiodes. For instance, we can use Silicon-Carbide (SiC)instead of Silicon (Si) diodes. SiC diodes have negligible

Fig. 11. System setup for proposed PV emulator and a maximum power point tracker.

Fig. 12. Measured results on a maximum power point tracker using abuck converter.

D.D.C. Lu, Q.N. Nguyen / Solar Energy 86 (2012) 1477–1484 1483

reverse-recovery current which usually causes additionalswitching loss (Spiazzi et al., 2003).

5. Conclusions

This paper presents the design and implementation of aPV emulator based on a two-switch buck-boost DC/DCconverter and a low cost 8-bit micro-controller. By usingmultiple straight lines approach, the PV emulator canmimic a PV panel with acceptable accuracy. The PV emu-lator has been tested using resistive loads as well as a max-imum power point tracker. Experimental results have

demonstrated the effectiveness and usefulness of the PVemulator.

Appendix A. This appendix shows the original PICAXEprogram code for 5-line approach for implementing the I–V curve of BP SX-10 PV panel at 25 �C and 3-line appro-ach at 75 �C.

symbol v = b1 ’define symbolssymbol i = b2symbol dc = b0symbol t = b3let dc = 75 ’set duty cycle for low start voltagemain:

pause 100gosub changedutygoto check

check:pause 100readadc 1, v ’ read voltage into v = 1/5 voltagevaluereadadc 4, i ’ read current into i = current valueif v > 250 then goto overload ’voltage of pin1 >= 5 Vif i > 51 then goto overload ’current flow >= 1 Alet t = i/5 ’i = i/5 because of voltage dividerlet v = v -t ’calculate load voltage = V-2iif pin3 = 1 then ’pin 3 = 1 (75 �C); pin 3 = 0 (25 �C)

if v < 123 then ’0 to 12 Vlet v = v/150let v = 71/2 – v ’equation I = 0.67–0.0017*V

elseif v > = 123 and v < = 153 then ’12 to 15 Vlet v = v*3/14let v = 64 – v ’equation I = 1.238–0.049*V

1484 D.D.C. Lu, Q.N. Nguyen / Solar Energy 86 (2012) 1477–1484

elseif v > 153 and v < 182 then ’15 to 17 Vlet v = v*5/let v = 219 – v ’equation I = 4.283–0.252*V

elselet dc = dc + 1 min 28goto changeduty

endifelse if v < 143 then ’0 to 14 V

let v = v/200let v = 35 – v ’equation I = 0.65–9 e-4*V

elseif v > = 143 and v < = 163 then ’14 to 16 Vlet v = v/15let v = 45 – v ’equation I = 0.8233–0.0133*V

elseif v > 163 and v < 184 then ’16 to 18 Vlet v = v/5let v = 66 – v ’equation I = 1.2633–0.04*V

elseif v > = 184 and v < 194 then ’18 to 19 Vlet v = 5*vlet v = v/11let v = 113 – v ’equation I = 2.1651–0.0909*V

elseif v > = 194 and v < 224 then ’19 to 21 Vlet v = 65*vlet v = v/61let v = 238 – v ’equation I = 4.599–0.2190*V

elselet dc = dc + 1 min 28goto changeduty

endifendifif i < v then

let dc = dc-1 min 28goto changeduty

elseif i > v thenlet dc = dc + 1 max 79goto changeduty

elsegoto check

endifgoto check

changeduty:pwmout 2,19,dcgoto check

overload:let dc = 75gosub changedutygoto main

References

Ahmad, J., 2010. A fractional open circuit voltage based maximum powerpoint tracker for photovoltaic arrays. In: 2010 2nd InternationalConference on Software Technology and Engineering (ICSTE), vol. 1,pp. V1-247–V1-250. doi:doi:10.1109/ICSTE.2010.5608868.

BP Solar PC SX-10 data sheet, 2003. <http://www.solaris.com.au/documents/BPSX05_10DataSheet.pdf>.

Esram, T., Chapman, P., 2007. Comparison of photovoltaic arraymaximum power point tracking techniques. IEEE Transactions onEnergy Conversion 22 (2), 439–449. doi:10.1109/TEC.2006.874230.

Ishaque, K., Salam, Z., Syafaruddin, 2011. A comprehensive matlabSimulink PV system simulator with partial shading capability based ontwo-diode model. Solar Energy 85 (9), 2217–2227. doi:10.1016/j.solener.2011.06.008.

Khouzam, K., Hoffman, K., 1996. Real-time simulation of photovoltaicmodules. Solar Energy 56 (6), 521–526. doi:10.1016/0038-092X(96)00008-4.

Kim, H., Kim, J., Kim, H., Lee, K., Kim, J., Yoo, D., Shin, D., 2010. Ahigh efficiency photovoltaic module integrated converter with theasymmetrical half-bridge flyback converter. Solar Energy 84 (8), 1376–1381. doi:10.1016/j.solener.2010.04.019.

Marsh, G., 2011. Partner in power: Part two: Whilst the micro-inverterrevolution looks set to spread, central and string inverters remain themainstream. Renewable Energy Focus 12 (3). doi:10.1016/S1755-0084(11)70062-X, 38–40, 42.

Mukerjee, A., Dasgupta, N., 2007. DC power supply used as photovoltaicsimulator for testing MPPT algorithms. Renewable Energy 32 (4), 587–592.

Nagayoshi, H., 2004. I–V curve simulation by multi-module simulatorusing I–V magnifier circuit. Solar Energy Materials and Solar Cells 82(1–2), 159–167. doi:10.1016/j.solmat.2004.01.014 (cANCUN 2003).

Ortiz-Rivera, E., Peng, F., 2005. Analytical model for a photovoltaicmodule using the electrical characteristics provided by the manufac-turer data sheet. In: Proceedings of the IEEE Power ElectronicsSpecialists Conference, 2005, pp. 2087–2091.

Paraskevadaki, E.V., Papathanassiou, S.A., 2011. Evaluation of MPPvoltage and power of mc-Si PV modules in partial shading conditions.IEEE Transactions on Energy Conversion 26 (3), 923–932.doi:10.1109/TEC.2011.2126021.

Petrone, G., Spagnuolo, G., Teodorescu, R., Veerachary, M., Vitelli, M.,2008. Reliability issues in photovoltaic power processing systems.IEEE Transactions on Industrial Electronics 55 (7), 2569–2580.doi:10.1109/TIE.2008.924016.

Spiazzi, G., Buso, S., Citron, M., Corradin, M., Pierobon, R., 2003.Performance evaluation of a Schottky SiC power diode in a boost PFCapplication. IEEE Transactions on Power Electronics 18 (6), 1249–1253. doi:10.1109/TPEL.2003.818821.

van der Borg, N., Jansen, M., 2003. Energy loss due to shading in a BIPVapplication. In: Proceedings of 3rd World Conference on PhotovoltaicEnergy Conversion, vol. 3, pp. 2220 –2222.

Zhang, H., Zhao, Y., 2010. Research on a novel digital photovoltaic arraysimulator. In: Proceedings of the IEEE International Conference onIntelligent Computation Technology and Automation, pp. 1077–1080.