Infrared Thermography for the Detection and Characterization of Photovoltaic Defects: Comparison between Illumination and Dark Conditions

Newly installed renewable power capacity has been increasing incredibly in recent years. For example, in 2018, 181 GW were installed worldwide. In this scenario, in which photovoltaic (PV) energy plays a leading role, it is essential for main players involved in PV plants to be able to identify the failure modes in PV modules in order to reduce investment risk, to focus their maintenance efforts on preventing those failures and to improve longevity and performance of PV plants. Among the different systems for defects detection, conventional infrared thermography (IRT) is the fastest and least expensive technique. It can be applied in illumination and in dark conditions, both indoor and outdoor. These two methods can provide complementary results for the same kind of defects, which is analyzed and characterized in this research. Novel investigation in PV systems propose the use of a power inverter with bidirectional power flow capability for PV plants maintenance, which extremely facilitates the electroluminescence (EL) inspections, as well as the outdoor IRT in the fourth quadrant.


Introduction
Renewable energies are nowadays a completely mainstream element in the global electricity mix. Combined with energy efficiency, renewables are playing a decisive role in decreasing emissions in the energy sector and in end-use sectors. Renewable electricity has expanded, due to both transferable and safe technologies and practical policy frameworks. With 181 GW added, the newly connected renewable power capacity set new records worldwide in 2018, raising the total by more than 8% relative to 2017, leaded by solar photovoltaic (PV), with 100 GW of new solar PV capacity installed in 2018, adding a total fixed PV solar capacity of 505 GW [1,2]. In this scenario in which PV energy plays a leading role, failure recognition systems in photovoltaic plants are important to all actors involved in them. [3]. Ensuring the reliability of photovoltaic plants, as well as their durability, has been essential in recent decades. This is important to know the origin of the failures, as well as their traceability [4].
It is widely known that the cost of purchasing PV solar modules has drastically fallen in recent decades, and especially in recent years [5][6][7], which makes the idea of installing PV solar plants even more attractive. Almost all the efforts of the PV exploitation are focusing on carrying out inspections of their plants (maintenance) that allow them to maintain their efficiency (operation), with the clear objective of obtaining high productivity levels. Therefore, the idea of advanced operation (illuminated) IRT provide complementary results for the same kind of defects, which are analyzed and characterized in this research. If this power inverter with bidirectional power flow capability was not available, the modules should be disconnected from the inverter and connected to a power source to do the inspections. Therefore, it is a novel system in PV plants which extremely facilitates the EL inspection as well as the outdoor (dark) IRT inspection, as it will be proved in this paper. Indoor IRT can be done in operation of the modules if proper lamps [31] in a solar simulator are used and injecting current in forward direction to the terminals using a current source, which can be denominated indoor (illuminated). However, this inspection technique will not be used along this paper, as it is not the purpose of the study. The introduced terms should not be confused with which other authors term as indoor photovoltaics (IPVs) [31], indoor lighting conditions [32] or indoor solar cells [33], which implies the conversion of the indoor light energy into direct electricity.
The main objective of this research is to analyze the outdoor (dark) IRT inspection using a power inverter with bidirectional power flow capability as a new inspection technique in PV plants and to demonstrate the things in common and the differences between outdoor (illuminated) and outdoor (dark) IRT relating both with other inspection techniques, as I-V curves and EL. The paper begins with the introduction to the topic covered, followed by the section on materials and methods, which explains the operating points of a PV system based on its voltage and current and details the sensors, equipment and modules used. Below are the main results of the study, which lead to its analysis and discussion. Finally, the conclusions of the work are indicated.

I-V Curve and Its Quadrants
The working principle of a solar PV cell consists in absorbing the light and convert it into electricity. The I-V curve under illumination represents all the possible combination of current and voltage of the PV device under certain conditions of irradiance and temperatures. It represents the superposition of the dark diode current with the photogenerated current, due to illumination, and can be described with the well-known one exponential model, including the series and shunt resistances [34], as shown in Figure 1. Equation (1) shows this expression: where I L is the photogenerated current, I 0 is the diode saturation current, m is the diode ideality factor, Rs is the series resistance, Rsh is the shunt resistance, and v t is the thermal voltage (K T/e) with K as the Boltzmann constant, e as the electron charge, and T as the temperature in Kelvin. Figure 2 shows the typical I-V curve of a PV device both under illumination and in dark conditions, with their characteristic points in illumination: short circuit current, Isc, open circuit voltage Voc, and maximum power point Pmax, with its corresponding voltage Vmax, and current Imax. The criteria that has been followed in Figure 2 and in the rest of the paper is considering that when the PV device is generating current, both its current and voltage are positive (first quadrant).
Sensors 2020, 20, x FOR PEER REVIEW 3 of 16 (dark) and outdoor (illuminated) IRT provide complementary results for the same kind of defects, which are analyzed and characterized in this research. If this power inverter with bidirectional power flow capability was not available, the modules should be disconnected from the inverter and connected to a power source to do the inspections. Therefore, it is a novel system in PV plants which extremely facilitates the EL inspection as well as the outdoor (dark) IRT inspection, as it will be proved in this paper. Indoor IRT can be done in operation of the modules if proper lamps [31] in a solar simulator are used and injecting current in forward direction to the terminals using a current source, which can be denominated indoor (illuminated). However, this inspection technique will not be used along this paper, as it is not the purpose of the study. The introduced terms should not be confused with which other authors term as indoor photovoltaics (IPVs) [31], indoor lighting conditions [32] or indoor solar cells [33], which implies the conversion of the indoor light energy into direct electricity.
The main objective of this research is to analyze the outdoor (dark) IRT inspection using a power inverter with bidirectional power flow capability as a new inspection technique in PV plants and to demonstrate the things in common and the differences between outdoor (illuminated) and outdoor (dark) IRT relating both with other inspection techniques, as I-V curves and EL. The paper begins with the introduction to the topic covered, followed by the section on materials and methods, which explains the operating points of a PV system based on its voltage and current and details the sensors, equipment and modules used. Below are the main results of the study, which lead to its analysis and discussion. Finally, the conclusions of the work are indicated.

I-V Curve and Its Quadrants
The working principle of a solar PV cell consists in absorbing the light and convert it into electricity. The I-V curve under illumination represents all the possible combination of current and voltage of the PV device under certain conditions of irradiance and temperatures. It represents the superposition of the dark diode current with the photogenerated current, due to illumination, and can be described with the well-known one exponential model, including the series and shunt resistances [34], as shown in Figure 1. Equation (1) shows this expression: Where IL is the photogenerated current, I0 is the diode saturation current, m is the diode ideality factor, Rs is the series resistance, Rsh is the shunt resistance, and vt is the thermal voltage (K T/e) with V I Figure 1. One-diode model of a photovoltaic (PV) cell including the series and shunt resistances. The optimum operational point of a PV device is the maximum power point, which is the point searched by the PV inverters in power plants. Nevertheless, under certain conditions it is possible that some devices move to second quadrant, or operate in the fourth quadrant (dark I-V curve) for testing purposes.
Operation in the second quadrant usually happens in PV modules with defective cells or in case of partial shading. A PV module is an association of PV cells, typically serially connected. If all the module cells are identical and connected in series, under maximum power point conditions the working current will be the cell Imax current, and the module Vmax will be the sum of each cell voltage. If all the cells in the module are not identical, due to differences in production, defects or shading, then the defective cell with lower current will move its operation point to the second quadrant, trying to reach the string current. In these conditions, the cell will dissipate the power produced by the other cells in the association, causing an increase in temperature. To avoid a high power dissipation which could damage the cell due to overheating [35,36], PV modules are manufactured with bypass diodes in their connection box. Bypass diodes are connected in parallel to a number of cells serially connected, and have the function of limiting the number of cells that can dissipate power into a defective one and to offer an alternative path to current in case of severe defects or partial shading. The operation of the bypass diode produce a knee in the module I-V curve, whose position depends on the number and distribution of bypass diodes in the module, and the type of defects or shading [37,38]. Figure 3 illustrates the working principle of the bypass diode.    The optimum operational point of a PV device is the maximum power point, which is the point searched by the PV inverters in power plants. Nevertheless, under certain conditions it is possible that some devices move to second quadrant, or operate in the fourth quadrant (dark I-V curve) for testing purposes.
Operation in the second quadrant usually happens in PV modules with defective cells or in case of partial shading. A PV module is an association of PV cells, typically serially connected. If all the module cells are identical and connected in series, under maximum power point conditions the working current will be the cell Imax current, and the module Vmax will be the sum of each cell voltage. If all the cells in the module are not identical, due to differences in production, defects or shading, then the defective cell with lower current will move its operation point to the second quadrant, trying to reach the string current. In these conditions, the cell will dissipate the power produced by the other cells in the association, causing an increase in temperature. To avoid a high power dissipation which could damage the cell due to overheating [35,36], PV modules are manufactured with bypass diodes in their connection box. Bypass diodes are connected in parallel to a number of cells serially connected, and have the function of limiting the number of cells that can dissipate power into a defective one and to offer an alternative path to current in case of severe defects or partial shading. The operation of the bypass diode produce a knee in the module I-V curve, whose position depends on the number and distribution of bypass diodes in the module, and the type of defects or shading [37,38]. Figure 3 illustrates the working principle of the bypass diode. The optimum operational point of a PV device is the maximum power point, which is the point searched by the PV inverters in power plants. Nevertheless, under certain conditions it is possible that some devices move to second quadrant, or operate in the fourth quadrant (dark I-V curve) for testing purposes.
Operation in the second quadrant usually happens in PV modules with defective cells or in case of partial shading. A PV module is an association of PV cells, typically serially connected. If all the module cells are identical and connected in series, under maximum power point conditions the working current will be the cell Imax current, and the module Vmax will be the sum of each cell voltage. If all the cells in the module are not identical, due to differences in production, defects or shading, then the defective cell with lower current will move its operation point to the second quadrant, trying to reach the string current. In these conditions, the cell will dissipate the power produced by the other cells in the association, causing an increase in temperature. To avoid a high power dissipation which could damage the cell due to overheating [35,36], PV modules are manufactured with bypass diodes in their connection box. Bypass diodes are connected in parallel to a number of cells serially connected, and have the function of limiting the number of cells that can dissipate power into a defective one and to offer an alternative path to current in case of severe defects or partial shading. The operation of the bypass diode produce a knee in the module I-V curve, whose position depends on the number and distribution of bypass diodes in the module, and the type of defects or shading [37,38]. Figure 3 illustrates the working principle of the bypass diode.  Operation in the fourth quadrant in dark conditions is applied to perform electroluminescence imaging of the PV module [29]. In this case, the module is forward biased with a power supply, to achieve a current similar to the Isc of the module. Other working points, as 10% of Isc, are commonly selected to better distinguish certain types of failures [11,19].

Sensors
Indoor and outdoor tests have been conducted in the School of Forestry, Agronomic and Bioenergy Industry Engineering (EIFAB) of University of Valladolid, in Soria, Spain.
Indoor tests include indoor (dark) EL, which has been performed in controlled ambient conditions simultaneously than indoor (dark) IRT in the fourth quadrant. For these two tests, the University has a temperature and humidity-controlled chamber, which is shown in Figure 4.
Sensors 2020, 20, x FOR PEER REVIEW 5 of 16 Operation in the fourth quadrant in dark conditions is applied to perform electroluminescence imaging of the PV module [29]. In this case, the module is forward biased with a power supply, to achieve a current similar to the Isc of the module. Other working points, as 10% of Isc, are commonly selected to better distinguish certain types of failures [11,19].

Sensors
Indoor and outdoor tests have been conducted in the School of Forestry, Agronomic and Bioenergy Industry Engineering (EIFAB) of University of Valladolid, in Soria, Spain.
Indoor tests include indoor (dark) EL, which has been performed in controlled ambient conditions simultaneously than indoor (dark) IRT in the fourth quadrant. For these two tests, the University has a temperature and humidity-controlled chamber, which is shown in Figure 4. EL images have been captured with a pco.1300 camera with an exposure of 5000 ms and indoor IRT images with a Flir C2 system, with a 80 × 60 pixels resolution, a thermal sensitivity of 0.1 °C and an accuracy of ±2% o ±2°C. In EL test, when the forward voltage is connected to the cell, the internal barrier and electric field in the barrier region will decrease accordingly, which breaks the carrier balance. Then, electrons and holes radiate compound and spontaneously radiate photons outward [26]. In this case, the forward voltage has been applied with a laboratory source, adjusting the voltage to feed the short circuit current (Isc) individually to each tested module.
On the other hand, outdoor tests, including outdoor (illuminated) IRT in the first quadrant, outdoor (illuminated) IRT in the fourth quadrant and outdoor (dark) IRT have been carried out in the PV field of the Campus Duques de Soria of the University of Valladolid, which can be seen in Figure 5. The thermographic camera used outdoor is a Workswell Wiris Pro camera, with a 640 × 512 pixels resolution, a thermal sensitivity of 0.05 °C and an accuracy of ±2% o ±2°C. This camera has a frame rate of 30Hz, is calibrated to be used with two different lenses, 32° and 69°, and includes a Full HD RGB sensor with a resolution of 1920 × 1080 pixels and x10 zoom. In this case, the forward voltage for tests in the fourth quadrant has been applied with a bidirectional inverter, adjusting the voltage to feed the Isc to the string of modules. This novel device, which is the key of the research performed as it extremely facilitates the outdoor EL inspections, as well as the outdoor (dark) IRT inspection and the outdoor (illuminated) IRT in the fourth quadrant, is in-depth detailed in the following subsection. Solar irradiance has been taken from ATERSA 4-20mA Compen solar cell, which registers the irradiance level every minute. The ambient temperature has been taken from Advanticsys CO2, temperature and humidity sensor, which registers the temperature every hour. All this information is registered in the ADVANTICSYS PV plant data logger, model MPC374.
Regarding the software used for the images treatment and reporting, Imaje J [39] was used for EL images, Flir Tools [40] for indoor Flir C2 IRT images and Workswell CorePlayer [41] for the outdoor Wiris Pro IRT images. EL images have been captured with a pco.1300 camera with an exposure of 5000 ms and indoor IRT images with a Flir C2 system, with a 80 × 60 pixels resolution, a thermal sensitivity of 0.1 • C and an accuracy of ±2% o ±2 • C. In EL test, when the forward voltage is connected to the cell, the internal barrier and electric field in the barrier region will decrease accordingly, which breaks the carrier balance. Then, electrons and holes radiate compound and spontaneously radiate photons outward [26]. In this case, the forward voltage has been applied with a laboratory source, adjusting the voltage to feed the short circuit current (Isc) individually to each tested module.
On the other hand, outdoor tests, including outdoor (illuminated) IRT in the first quadrant, outdoor (illuminated) IRT in the fourth quadrant and outdoor (dark) IRT have been carried out in the PV field of the Campus Duques de Soria of the University of Valladolid, which can be seen in Figure 5. The thermographic camera used outdoor is a Workswell Wiris Pro camera, with a 640 × 512 pixels resolution, a thermal sensitivity of 0.05 • C and an accuracy of ±2% o ±2 • C. This camera has a frame rate of 30 Hz, is calibrated to be used with two different lenses, 32 • and 69 • , and includes a Full HD RGB sensor with a resolution of 1920 × 1080 pixels and ×10 zoom. In this case, the forward voltage for tests in the fourth quadrant has been applied with a bidirectional inverter, adjusting the voltage to feed the Isc to the string of modules. This novel device, which is the key of the research performed as it extremely facilitates the outdoor EL inspections, as well as the outdoor (dark) IRT inspection and the outdoor (illuminated) IRT in the fourth quadrant, is in-depth detailed in the following subsection.

Power Inverter with Bidirectional Power Flow Capability
A specific power inverter with bidirectional power flow capability was placed in the pilot-site for this study. The power inverter is a NPC I-type that has been recently developed to help in the maintenance of PV plants by means of EL image processing [19]. Indeed, the three-level neutral point clamped (NPC) configuration is one of the most extended topologies in PV inverters currently on the market. One of the main advantages of this approach is that bidirectional current flow can be useful to avoid the typical inspection procedure, in which the modules are disconnected from the inverter to be connected to a power source, saving a lot of effort and resources.
However, PV inverters are not usually prepared to start under low-irradiance conditions, and some minor hardware modifications are required in this sense [19]. A DC-link pre-charge system is required to perform inspections in low-irradiance environments. Nevertheless, the cost of the precharge system is negligible compared to the cost of a utility-scale PV inverter, and the high automatization level to perform on-site inspections. In addition, the changes do not affect the inverter's efficiency, so the modifications do not introduce any additional power losses.
Therefore, the proposed inverter is a novel system in PV plants which extremely facilitates and place value on the on-site outdoor image processing inspections, taking special interests for the following techniques: the EL inspection; the outdoor (dark) IRT inspection; and the outdoor (illuminated) IRT inspection.
Regarding the pilot-site of this study, a power inverter with a rated output power of 3 kW has been installed in the facilities of the solar plant. On the one hand, the output of the inverter has been connected to the 400 V three-phase grid. On the other hand, the DC input side can set the PV string voltage between 330 and 550 V. Note that during the EL test execution, this voltage range is enough to allow the current control vary between 10% and 100% of Isc.
The image capturing procedure is as follows: in standard working operation conditions, the power inverter works autonomously, taking energy from the panels and feeding it into the grid whenever there is enough solar irradiance. However, when an inspection is required, as Figure 6 shows, a computer connected to the inverter gives it the right orders.
The user can set the desired inspection setpoint in the computer, usually the current necessary to carry out the inspection, and afterwards, proceeds to send a command to the inverter. The system, from that instant, ignores the maximum power point tracking (MPPT) algorithm, and follows the reverse reference current ( ), by means of varying the DC side voltage ( ), until the desired current has been reached. Solar irradiance has been taken from ATERSA 4-20mA Compen solar cell, which registers the irradiance level every minute. The ambient temperature has been taken from Advanticsys CO 2 , temperature and humidity sensor, which registers the temperature every hour. All this information is registered in the ADVANTICSYS PV plant data logger, model MPC374.
Regarding the software used for the images treatment and reporting, Imaje J [39] was used for EL images, Flir Tools [40] for indoor Flir C2 IRT images and Workswell CorePlayer [41] for the outdoor Wiris Pro IRT images.

Power Inverter with Bidirectional Power Flow Capability
A specific power inverter with bidirectional power flow capability was placed in the pilot-site for this study. The power inverter is a NPC I-type that has been recently developed to help in the maintenance of PV plants by means of EL image processing [19]. Indeed, the three-level neutral point clamped (NPC) configuration is one of the most extended topologies in PV inverters currently on the market. One of the main advantages of this approach is that bidirectional current flow can be useful to avoid the typical inspection procedure, in which the modules are disconnected from the inverter to be connected to a power source, saving a lot of effort and resources.
However, PV inverters are not usually prepared to start under low-irradiance conditions, and some minor hardware modifications are required in this sense [19]. A DC-link pre-charge system is required to perform inspections in low-irradiance environments. Nevertheless, the cost of the pre-charge system is negligible compared to the cost of a utility-scale PV inverter, and the high automatization level to perform on-site inspections. In addition, the changes do not affect the inverter's efficiency, so the modifications do not introduce any additional power losses.
Therefore, the proposed inverter is a novel system in PV plants which extremely facilitates and place value on the on-site outdoor image processing inspections, taking special interests for the following techniques: the EL inspection; the outdoor (dark) IRT inspection; and the outdoor (illuminated) IRT inspection.
Regarding the pilot-site of this study, a power inverter with a rated output power of 3 kW has been installed in the facilities of the solar plant. On the one hand, the output of the inverter has been connected to the 400 V three-phase grid. On the other hand, the DC input side can set the PV string voltage between 330 and 550 V. Note that during the EL test execution, this voltage range is enough to allow the current control vary between 10% and 100% of Isc.
The image capturing procedure is as follows: in standard working operation conditions, the power inverter works autonomously, taking energy from the panels and feeding it into the grid whenever there is enough solar irradiance. However, when an inspection is required, as Figure 6 shows, a computer connected to the inverter gives it the right orders.

Tested Modules
Four mono-crystalline modules with different kinds of defects have been tested in the present research. The main characteristics of the modules are: 72 cells (6 × 12) with 3 bypass diodes (one each 24 cells), power (P) 175W, open circuit voltage (Voc) 44.35V, maximum power point (Vmp) 36.26V, Isc 5.45A and maximum power current (Imax) 4.83A. The modules selected for the study contain several defects, in order to be able to identify failures with the different techniques, and to be able to carry out the comparison. If the modules were new and they did not show defects, the assessment would not be possible. The following Figure 7 and Figure 8 present the RGB images and the I-V curves of the analyzed modules, respectively. I-V curves have been traced using an I-V tracer Solar IV HT 1500V at 800, 859, 839 and 792 W/m 2 for modules 1 to 4, respectively. The interpolation to standard test conditions (STC) has not been applied, as the significant deterioration of the modules could importantly modify the coefficients of variation of Isc (α), Voc (β) and P (γ) with the temperature or other interpolation equations used by the I-V tracer, not obtaining real results. In Figure 8, typical knees of the operation of the bypass diode can be observed in the I-V curves, in which the defects induce the bypass conduction of current, as in module 1, module 2 and module 4. The defects present in module 3 do not seem to cause the operation of the bypass diodes. However, in module 1, module 2 and module 3 there is a reduction in the Voc values with respect to module 4, which could indicate the presence of short-circuited cells according to reference [11], as will be further analyzed in the discussion. The user can set the desired inspection setpoint in the computer, usually the current necessary to carry out the inspection, and afterwards, proceeds to send a command to the inverter. The system, from that instant, ignores the maximum power point tracking (MPPT) algorithm, and follows the reverse reference current (I Re f ), by means of varying the DC side voltage (V DC ), until the desired current has been reached. The modules selected for the study contain several defects, in order to be able to identify failures with the different techniques, and to be able to carry out the comparison. If the modules were new and they did not show defects, the assessment would not be possible. The following Figures 7 and 8 present the RGB images and the I-V curves of the analyzed modules, respectively. I-V curves have been traced using an I-V tracer Solar IV HT 1500 V at 800, 859, 839 and 792 W/m 2 for modules 1 to 4, respectively. The interpolation to standard test conditions (STC) has not been applied, as the significant deterioration of the modules could importantly modify the coefficients of variation of Isc (α), Voc (β) and P (γ) with the temperature or other interpolation equations used by the I-V tracer, not obtaining real results. In Figure 8, typical knees of the operation of the bypass diode can be observed in the I-V curves, in which the defects induce the bypass conduction of current, as in module 1, module 2 and module 4. The defects present in module 3 do not seem to cause the operation of the bypass diodes. However, in module 1, module 2 and module 3 there is a reduction in the Voc values with respect to module 4, which could indicate the presence of short-circuited cells according to reference [11], as will be further analyzed in the discussion.

Tested Modules
knees of the operation of the bypass diode can be observed in the I-V curves, in which the defects induce the bypass conduction of current, as in module 1, module 2 and module 4. The defects present in module 3 do not seem to cause the operation of the bypass diodes. However, in module 1, module 2 and module 3 there is a reduction in the Voc values with respect to module 4, which could indicate the presence of short-circuited cells according to reference [11], as will be further analyzed in the discussion.

Results
This section provides a description of the results obtained in the experimental tests detailed in Materials and Methods, their interpretation, as well as the experimental conclusions that can be drawn.
Indoor measurements have been done at 25 °C, in the temperature-controlled chamber showed in Figure 4. Outdoor and indoor IRT images irradiation, ambient temperature and module temperature are presented in Table 1. The module temperature corresponds with the average temperature of the whole surface of the module. This information will be used in the results discussion in the following section. The current level could produce quantitative variations in the temperature of the defect. However, the authors consider that there are no qualitative variations regarding the identification of the defect.

Results
This section provides a description of the results obtained in the experimental tests detailed in Materials and Methods, their interpretation, as well as the experimental conclusions that can be drawn.
Indoor measurements have been done at 25 °C, in the temperature-controlled chamber showed in Figure 4. Outdoor and indoor IRT images irradiation, ambient temperature and module temperature are presented in Table 1. The module temperature corresponds with the average temperature of the whole surface of the module. This information will be used in the results discussion in the following section. The current level could produce quantitative variations in the temperature of the defect. However, the authors consider that there are no qualitative variations regarding the identification of the defect.

Results
This section provides a description of the results obtained in the experimental tests detailed in Materials and Methods, their interpretation, as well as the experimental conclusions that can be drawn.
Indoor measurements have been done at 25 • C, in the temperature-controlled chamber showed in Figure 4. Outdoor and indoor IRT images irradiation, ambient temperature and module temperature are presented in Table 1. The module temperature corresponds with the average temperature of the whole surface of the module. This information will be used in the results discussion in the following section. The current level could produce quantitative variations in the temperature of the defect. However, the authors consider that there are no qualitative variations regarding the identification of the defect. 3.1. Module 1 Figure 9 shows the experimental results obtained when measuring the photovoltaic module 1. The Figure 9 shows the images taken indoor and outdoor, both in darkness and in illumination. In IRT images, the temperature of the spots hotter than the average temperature of the module is indicated in red, while the temperature of spots cooler than the average of the module is indicated highlighted in blue.
Sensors 2020, 20, x FOR PEER REVIEW 9 of 16  Figure 9 shows the experimental results obtained when measuring the photovoltaic module 1. The Figure 9 shows the images taken indoor and outdoor, both in darkness and in illumination. In IRT images, the temperature of the spots hotter than the average temperature of the module is indicated in red, while the temperature of spots cooler than the average of the module is indicated highlighted in blue.

Indoor (dark) EL (a)
Outdoor (illuminated) IRT 4 th quadrant (e)   Figure 10 shows the experimental results obtained when measuring the photovoltaic module 2. The Figure 10 shows the images taken indoor and outdoor, both in darkness and in illumination. In IRT images, the temperature of the spots hotter than the average temperature of the module is indicated in Sensors 2020, 20, 4395 10 of 16 red, while the temperature of spots cooler than the average of the module is indicated highlighted in blue.

Module 2
Sensors 2020, 20, x FOR PEER REVIEW 10 of 16 Figure 10 shows the experimental results obtained when measuring the photovoltaic module 2. The Figure 10 shows the images taken indoor and outdoor, both in darkness and in illumination. In IRT images, the temperature of the spots hotter than the average temperature of the module is indicated in red, while the temperature of spots cooler than the average of the module is indicated highlighted in blue.

Indoor (dark) EL (a)
Indoor ( Figure 11 shows the experimental results obtained when measuring the photovoltaic module 3. The Figure 11 shows the images taken indoor and outdoor, both in darkness and in illumination. In IRT images, the temperature of the spots hotter than the average temperature of the module is indicated in red, while the temperature of spots cooler than the average of the module are indicated highlighted in blue.  Figure 11 shows the experimental results obtained when measuring the photovoltaic module 3. The Figure 11 shows the images taken indoor and outdoor, both in darkness and in illumination. In IRT images, the temperature of the spots hotter than the average temperature of the module is indicated in red, while the temperature of spots cooler than the average of the module are indicated highlighted in blue. Figure 12 shows the experimental results obtained when measuring the photovoltaic module 4. The Figure 12 shows the images taken indoor and outdoor, both in darkness and in illumination. In IRT images, the temperature of the spots hotter than the average temperature of the module is indicated in red, while the temperature of spots cooler than the average of the module is indicated highlighted in blue. Sensors 2020, 20, x FOR PEER REVIEW  Figure 12 shows the experimental results obtained when measuring the photovoltaic module 4. The Figure 12 shows the images taken indoor and outdoor, both in darkness and in illumination. In IRT images, the temperature of the spots hotter than the average temperature of the module is indicated in red, while the temperature of spots cooler than the average of the module is indicated highlighted in blue.

Indoor (dark) EL (a)
Indoor ( Figure 12 shows the experimental results obtained when measuring the photovoltaic module 4. The Figure 12 shows the images taken indoor and outdoor, both in darkness and in illumination. In IRT images, the temperature of the spots hotter than the average temperature of the module is indicated in red, while the temperature of spots cooler than the average of the module is indicated highlighted in blue.

Discussion
As seen in the results section, it is possible to do IRT in different situations. However, before getting into this analysis, it is necessary for a brief reminder regarding photovoltaics. A PV module has two ways of working, in illumination and in darkness. Therefore, it is not correct to speak of IRT (or EL) just outdoor or indoor, but rather it is correct to speak of: outdoor (illuminated), outdoor (dark), indoor (illuminated), indoor (dark). Results obtained in the different situations are discussed through this section. To denominate a specific cell within a module, it will be done identifying its row with a letter and the column with a number, starting with the upper left cell of the module as cell A1.

Illuminated IRT
The first subject which will be discussed is the analysis of outdoor (illuminated) IRT first quadrant and outdoor (illuminated) IRT fourth quadrant. IRT in both situations is performed in illumination but the effect on the PV module is different. When the module is in illumination and receiving the injected current (fourth quadrant) (e) figures from the bidirectional inverter, the protection diodes cannot bypass the current, because they work as an open circuit with this polarity, and all cells (including damaged cells) receive all the injected current. Therefore, the damaged cells can receive all the current injected into the module. Instead, in illumination and operating normally (in the first quadrant), (d) images, the bypass diodes will protect the sub-strings with defective cells, limiting the power dissipation. In the previous figures, from Figure 9 to Figure 12, this slight difference in IRT can be seen in (d) and (e).
The operation of a PV module in the fourth quadrant under illumination is a condition that would be difficult to achieve in the field. Nevertheless, with the aid of the bidirectional inverter and for testing purposes, the operation point of the modules was moved to this quadrant to investigate the differences in temperature with respect to the other conditions. As future work, it is proposed to apply all these techniques in other plants. This research group collaborates with a company that has inverters of these characteristics in a real plant in operation. However, the objective of this article is the comparison of the different IRT techniques (illumination and darkness). It was observed that in some of the cases, for instance, in Figure 11d,e, the faulty cells increase their temperature over the module 3 average temperature, due to power dissipation in both conditions, outdoor illuminated first and fourth quadrant. In Figure 11d, the difference of temperature between the two hotspots and the module average temperature is 2.31 • C and 5.41 • C, for the left and right hotspots respectively. In Figure 11e, these differences are 4.33 • C and 6.33 • C, respectively. This can be due to the difference in the irradiation conditions between both cases, as image in Figure 11d was taken with 680 W/m 2 and (e) with 917 W/m 2 . However, regarding Figure 12d,e, there is agreement among the spotted cells, however, they appear hotter than the module 4 average temperature in outdoor (illuminated) IRT in the first quadrant (d), and one hotter and the rest cooler in outdoor (illuminated) IRT in the fourth quadrant (e). A similar effect is found in module 2 ( Figure 10).

Dark IRT
Secondly, with respect to the dark IRT, the technique can be performed both indoor (dark) and outdoor (dark), as can be seen in images from Figure 9 to Figure 12b,c in Section 3, respectively. Both images are similar, and complementary to the image obtained in EL (a). This means that dark IRT can be used to identify faults in PV modules similar to those identified with the EL technique, as it will be further discussed in Section 4.4. As seen in the previous Figures 9-12, cold-spot failures can be identified, which have been defined by authors in [11], and are recognized as black cells in EL images (a). This fact can be observed in Figure 9a, which has two contiguous black cells, detected as cold spots in Figure 9b,c. Additionally, module 1 presents two cells with some inactive areas (C2 and E6) in EL (Figure 9a). One of those cells (cell C2) is detected as a cold spot, while the other one, cell E6, is shown as a hotspot in Figure 9b,c. The difference of temperature of the inactive cells (D9 and D10) with respect to the average of the module is −8.80 • C in indoor (dark) IRT (b) and −7.82 • C in outdoor (dark) IRT (b). The differences of temperature of the broken cells (C2 and E6) with respect to the average of the module are −5.0 • C and +6.4 • C, respectively, in indoor (dark) IRT (b), and −4.52 • C and 12.48 • C, respectively, in outdoor (dark) IRT (b). As seen, similar results are obtained in indoor and outdoor (dark) IRT.
In all the modules, outdoor (dark) injection of current to the PV modules has been possible thanks to a bidirectional inverter. This new device allows EL and IRT to be done in the dark, without the need to disconnect the PV modules, and without having to connect them to an external power source.

Comparison between Illuminated and Dark IRT
The working point in an illumination curve is different from the working point in dark conditions (in the dark curve), which is responsible of the different results between the illumination and dark IRT. This can be seen comparing Figure 9 to Figure 12 in Section 3 (b), (c) versus (d) and (e). As can be seen in the figures in Section 3, when comparing images in darkness (b) and (c) and images in illumination (d) and (e), the results are different. In dark images, the cold spots of the PV module are easily recognizable, as seen with the inactive cells in Section 4.2, while in the illuminated images, these spots do not appear, and hot spots are located. This effect can be seen in Figure 10b,c, in which the three inactive cells (D5, E5 and E6) are seen as cold spots in indoor and outdoor (dark) IRT. On the other hand, in the illuminated IRT images, Figure 10d,e, these cold spots are difficult to identify, while hot spot of damaged cell C9 is easily located. The images in darkness are equal comparing the images taken indoors and outdoors (b) and (c), respectively, since, in both cases, we work on the darkness curve as already detailed in Section 4.2. They could present some minor differences, since those outdoors could be affected by weather conditions (mainly temperature), which are introduced in Table 1.

Indoor (Dark) IRT and EL
It can be highlighted the similarity between the outdoor (dark) IRT images (equally for those indoors) with EL images. It is possible to affirm that outdoor (dark) IRT images are capable of detecting failures inherent in cells, and that they only appear in EL and do not appear in IRT in illumination, as seen in Section 4.3. Finally, it can be reviewed the cold-spots temperatures in (dark) IRT of completely inactive areas in the modules evaluated (cells D9 and D10 in module 1 (Figure 9a), cells D5, E5 and E6 in module 2 ( Figure 10a) and cells E9 and E10 in module 3 (Figure 11a), which could correspond with short-circuited cells according to reference [11]. For the temperatures comparison, will be used the indoor (dark) IRT information (b), as it has been captured simultaneously to EL images (a). The difference of temperatures of the inactive cells with respect to the average of the module are: −8.80 • C in module 1 Figure 9b, −11.9 • C in module 2 Figure 10b, and −9.9 • C in module 3 Figure 11b. Therefore, the cooling is similar in the three cold-spot evaluated.
Consequently, it is possible to conclude that the outdoor (dark) IRT images are similar to the indoor (dark) IRT images, and highly coincide with the information seen in EL images. In contrast, outdoor (dark) IRT images are different from outdoor (illuminated) IRT images.

Conclusions
In recent years, the EL technique is gaining much interest in the scientific world, and its application in installed plants is beginning to be carried out. Traditionally, the EL technique requires the disconnection of the PV module and its connection to an external source, or the supply of a complete string through the power supply. In this work, authors have presented a bidirectional inverter that is used to carry out EL without disconnection of modules. Taking advantage of this device, authors of this work have raised the possibility of doing outdoor (dark) IRT, as a complement to EL images. The injection of current from the bidirectional inverter to the PV string will allow EL and IRT to be carried out in the dark, without the need to disconnect the installed modules.
With respect to IRT, the work has served to present the differences and similarities of said technique according to its point of operation. An outstanding conclusion is that IRT in the dark produces similar results, whether done outdoors or indoors, and furthermore, its results provide information that is significantly similar to that obtained with EL. In this sense, the detection of cold spots is very accurate, and the results are easily recoverable.
On the other hand, the work has shown that the information obtained with IRT in darkness is very different from that obtained with IRT in illumination. Therefore, the work invites to perform IRT in darkness when outdoor (dark) EL is performed, in order to have complementary information to that available with IRT in illumination.
The authors will carry out these described techniques in production plants, but these plants must have a photovoltaic inverter like the one presented. The authors will also work on advanced measurements in photovoltaic plants, which complement the conclusions obtained through images (EL and IRT). Funding: This research was funded by the "Ministerio de Industria, Economía y Competitividad" grant number "RTC-2017-6712-3" with name "Desarrollo de herramientas Optimizadas de operaCión y manTenimientO pRedictivo de Plantas fotovoltaicas-DOCTOR-PV" and by the University of Valladolid 2018 pre-doctoral research contract from the budget application 180113-541A.2.01e691.

Conflicts of Interest:
The authors declare no conflict of interest.