Dependence of solar panel performance on light spectrum and temperature

Research Group: Jeriah Bankson, Nathan Bartol, and Tom Dale

Launch: Whitworth Spring 2016



This experiment focused on researching how the intensity of red, green, blue and ultraviolet light wavelengths and temperature affect the voltage output of a mono crystalline photovoltaic. The experiment would record the amounts of certain wavelengths for red, green, blue and uv light and voltage with respect to time. This would be to discern how the quantity of each variable increased or decreased the efficiency of the solar panel, which would be seen as either higher or lower voltage output. We hypothesized that there would be a linear relationship where voltage would increase with light intensity. Also that voltage would increase as temperature decreased since a frigid environment allows metal to conduct energy more effectively and allow electronics function better. We designed a 16.5cm x 12.7cm x 9.4cm Styrofoam pod with eight mono crystalline solar panels and two spark fun sensors to record both rgb and uv light intensity. There were also temperature sensors on the outside and inside of the pod to determine how voltage changed with variable temperature. In analysis it was found that increased light intensity increases voltage output. Also that higher wavelengths of light make the solar panel more efficient at converting light into voltage because the band gap can better absorb lower energies; lower wavelengths of light with higher energies decrease the ability of the panel to create voltage because most of the energy is lost due to heat and limited band gap receptivity. Temperature was not shown to affect the voltage output as much as expected because the pod did not record data high enough in the cold stratosphere. This experiment was successful in the knowledge retrieved. We were able to determine how certain factors influence photovoltaic functionality and learn how to overcome environmental adversity and human error to accomplish a meaningful and valuable research project.

Background
The 21st century has been a generation of scientific innovation, with interest directed in finding sustainable sources of energy. Currently scientists are trying to replace fossil fuels, because of its limited nature and the detrimental effects it has on the environment. The goal is to rely on a more abundant, powerful and cleaner source of energy-the sun. However being able to efficiently convert and store the sun's energy is complicated, which is why the study of how to improve photo voltaic cells is a growing and important field of study. The data collected from this experiment will show at what height the effect of different light wavelengths, UV light and temperature, will make the solar cell generate the maximum voltage.

The solar cell converts the energy from the sun into voltage through the photoelectric effect. This occurs when the photons from the sun's light are absorbed, which separates the electrons from the atoms inside the cell. When the electrons break away, work is done as the electrons cross the conduction band to the valance band within the cell. This work done by the electrons releases energy into the solar cell which the solar cell converts to a specific voltage. The limiting factor to the cell's ability to convert the sun's energy to voltage its absorption of light. The solar cell has a specific spectral response (the wavelength of light that can be detected)and absorption depth (how far wavelengths can penetrate without loosing intensity). Different absorption depths and spectral responses are dependent on the different materials used to make the solar cell. Another limiting factor to the photovoltaics generating voltage is the cell's ability to absorb certain wavelengths of light. The spectrum of light is made up of different wavelengths; photovoltaics can only use infrared, ultraviolet, and visible light for energy transfer. The infrared spectrum has wavelengths of 700nm to 1mm, the ultraviolet spectrum has wavelengths of 10nm to 400nm, the visible spectrum has wavelengths of 400-700nm. The longer wavelengths of visible light have the lowest amounts of energy but are the most easily absorbed by a solar cell. The shorter wavelengths of visible light have the highest amounts of energy. In order for voltage to be created, work has to be done by the electrons that have absorbed light energy. However the size of the solar cell's band gap, the space between the conduction and valence band, determines what wavelength of light can be absorbed. Therefore to maximize the voltage generated by the photo voltaic, it is important to know the size of the band gap that will absorb wavelengths with enough energy to make the electrons do work, but not too high of energy where most it is lost to heat. Another important factor that scientists include in determining the efficiency of solar cells, is temperature. Electronics are made of metal and a lesser amount of heat enables the metal to condense;the fluidity of metal and its ability to conduct electricity increases when the temperature decreases. Scientists avoid hot places for the placement of solar cells because the cell's efficiency decreases. The optimum placement for solar cells is a cold place with a high exposure to intense light.

Objective
This experiment seeks to answer the question of to what extent does temperature, light wavelength, and light intensity effect the productivity of photovoltaics in near space. Mono-crystalline solar cells will be tested not to necessarily determine how to make them more efficient, but show the factors that affect the cell's efficiency. In this experiment, temperature readings and the wavelength intensities of red, green, blue and UV light will be taken to find at what point the intensity of light creates the maximum voltage without impeding the electrical benefits of the cold temperature in the stratosphere. The temperature and light intensity are known to directly affect the production of electricity by solar cells. As temperature decreases and light intensity increases photovoltaics produce more energy. The lower stratosphere is much colder than the upper portion making the electric productivity due to temperature most likely greatest from 10 - 20 km. The light intensity however should steadily increase as altitude increases. Due to these conditions a 20km altitude will most likely be the opportune height for photovoltaic production. The change in light wavelength will also be tested, however color is spectrum is believed to have very little effect on the panels in comparison to the other tested conditions. The predicted max efficiency of the solar panels is at a 20km altitude.

Mechanical Design
The eight mono-crystalline photo voltaic cells are attached to a Styrofoam pod that measures 16.5cm x 12.7cm x 9.4cm and has a thin aluminum paper covering. The eight photo voltaic cells are soldered together on a single wire, 5.1 cm apart. Two holes were drilled into the left and right side of each face of the pod, where the connecting wire enters the inside of the pod. For each of the four faces, two solar cells in sequence are attached with epoxy to 3.8cm x 5.1cm pieces of cardboard. The piece of cardboard is then adhered to the face of the pod with epoxy, and then secured with duct tape. The ML8511 Spark Fun UV sensor breakout and the ISL29125 Spark Fun RGB light sensor are adhered to a 2cm x 1cm piece of cardboard that is epoxied to the 16.5 cm sides of the pod; a hole is also drilled behind each sensor for the wires to plug into the microchip on the inside. One temperature sensor is connected to the circuit board inside the pod, a second temperature sensor is attached in close proximity to one of the solar cells through a hole drilled above the solar panels secured by duct tape. The wires connecting the solar cells are connected to the bread board with specific resistors to ensure the cells do not receive a terminal voltage that would short out the cells. The micro controller is fastened to the bottom of a board, with four holes drilled into each corner of the micro controller. We used screws and nuts to secure the controller to the board. To stabilize the bread board, we drilled four holes on each corner of the bread board and used washers on the corners to prevent any movement during the flight. The same method was used to attach the battery except with one screw and washer. The board that has the attached components is secured to the inside of the pod by two zip ties through holes drilled into the bottom of the pod. This will minimize any shifting inside the pod that could damage the circuitry during the ascent and decent of the trip.

Electrical Design
The microcontroller is connected to the digital RGB sensor, two analog temperature readers, one analog UV sensor, and the single chain of eight analog solar panels. The temperature sensors, RBG sensor, and uv sensor were connected directly to the microcontroller while the solar panel chain needed a voltage divider of equivalent resistance of _____ as to not short the digital in.

WARNING: The TMP36 temperature sensors must be used with diligence. Switching the small device 180 degrees will short the analog in pin on the micro controller as well as an adjacent pin. The surge of power that comes from the reversed sensor will fry the data for two analog inputs and create an extremely hot TMP36 sensor. Regrettably this happened in our group and a new micro controller had to be ordered.

Analog Inputs

 * UV sensor = PTB2
 * Temperature sensor (outside, TMP36) = PTB3,
 * Temperature sensor (inside, TMP36) = PTC10
 * Solar Panels = PTB11

Digital Input

 * RGB sensor = SCL, SDA

Circuit
Total current in the circuit: 130 mA

Li-Fes2 battery capacity: 2700mAh-3400mAh


 * To determine if the battery would power the system for the minimum required time, 4 hours, we calculated

Capacity of Current over time/total current =time the battery can power the system =20.7-26.2 hrs

Code
For data collection code was written in the C language to be executed by the microcontroller in order to write information in an onboard SD card. The code is structured using a source file that calls on a timer four partnering files to get RGB, UV, Solar Panel, and temperature data and then saves all information to an SD save file. The main function starts with declarations of LED and timer variables. The onboard LED lights are set to red= on and blue= off at the start, once the timer starts red is turned off, later when the code successfully is completed the blue light will turn off. These LED indicators are useful in the testing and design process to find any potential errors but are not necessary for launch. After variables are initialized the timer starts and confirmation of this is printed to the screen. Next the SD card is verified to be properly mounted by calling a function in the SDSave.cpp and will return an error/ stop program if the SD card is not properly mounted. An array for RGB sensor data is created because three separate pieces of data must be recorded: rgb[0]= red, rgb[1]= green, and rgb[2]= blue. The file to record data is then initialized and a while loop is started to record all sensor data. The while loop repeats every .5 seconds and gets data from the four sensor types. With a half second refresh rate our data can accurately show minor fluctuations in information from the sensors. Once the loop restarts the data from each getter function is passed both to a screen print for testing/calibration in the lab before hand, as well as saved to the local SD card. After four hours of flight data the loop ends, the data file is safely closed, and the blue LED is turned on. Four hours is sufficient time for recording all required data because information is only needed at each height on the ascent. After our successful flight we realized a major bug in our code that was unfortunately not caught sooner. If the code is not refreshed then the timer will reach slightly over 30 minutes and stop data collection. We knew this and coded a loop that would refresh every 30 minutes and start saving new data as a timer + 30 minutes. We tested our code multiple times and could see that the code now read from .5 seconds to 14400 seconds (4 hours). Unfortunately we failed to realize that while the first 30 minutes were taking data .5 seconds apart, the last 3 1/2 hours were taking data 30 minutes apart. Our graph ended up being very imprecise because of this miscalculation. The error in our code came from the section shown below. In our data collection if statement, it was said that if total time is greater than last time then find data. Unfortunatley lastTime goes from 0-30 minutes and total time goes from 0-4 hours. THis means that for the first 30 minutes data is recorded, and then data only records at the end of every 30 minute reset loop. This bug made major damage to our experiment but because we are focused more on how the sensor data interacts with each other rather than how it interacts over time, we can still draw some accurate conclusions.

Link to Final Code: https://developer.mbed.org/users/tdale19/code/FinalProject/

Validation and Calibration for Temperature sensor
The TMP36 temperature readers measure at points inside and outside of the box. They were plugged into ports on the micro-controller and were validated on 3/21/16 by showing a spike in temperature when brought near body heat. When a warm hand came very near a TMP the data rose by some amount. Data was then calibrated using a already working thermometer, beaker filled with ice, and a beaker of water on a hot plate. By comparing data from the TMP36 to the data collected by the working thermometer at different temperatures the correct conversion factor was found. The TMP36 output (in voltage) is converted to proper temperature (in Celsius) by a factor of 100. After adding this conversion factor into our code, the TMP36's and the test thermometer measured the same data within +/- 1 degree. Temperature was measured at 10 (c), 23 (c), and 44 (c). Using these three data points our Temperature reading system was verified to be accurate.

Validation and Calibration for Solar Panels
to establish validation of the 4.0V 20mA monocrystalline Solar Cell. a 4.0V 20mA monocrystalline Solar Cell was attached to a volt meter that was set to measurements in direct current units of voltage while light was shown onto it if there was a measurement of voltage visible on the volt meter then the 4.0V 20mA monocrystalline Solar Cell was functioning as expected. the solar panel was then covered to provide a situation without light if the solar cell reads an extremely low or zero voltage then we can safely assume the solar panel is functioning completely, low voltages are acceptable because electronic noise present in our environment. then these conditions were replicated for all eight of the solar cells. Calibration required a more precise approach. considering we are measuring light transformed into electrical energy it required a controlled light source, for this I used a green LED, any color LED can be used. the input to the LED was controlled using an AC to DC step-down transformer, there were small amounts of sway on the voltage output of the transformer but that was recorded. place the LED securely in a stand. Next, the 4.0V 20mA monocrystalline Solar Cell is placed in a stand directly in front of the LED. between the two stands place a third stand that holds the absorbance lenses, this sand should be as close to the solar cell as you can get it. then proceed to change filters and record the voltage of the 4.0V 20mA monocrystalline Solar Cell with the diffrent filters in place.



Validation and Calibration for RGB sensor
Verification: 3/17/16 To establish on a basic level that the Spark Fun RGB sensor was functioning, the RGB sensor was plugged into a circuit with the micro controller. The computer program Putty read the quantity of Red, Green and Blue light frequencies with time, and every 20 seconds the sensor was covered with a box and jacket to get readings of zero. Also, a light was shinned on the sensor to get maximum readings. The RGB readings reached zero with the absence of light and reached maximum values with the artificial light, indicating the sensor could correctly indicate changes and quantities of visible light.

Calibration: 3/17/16 To validate that the RGB sensor can indicate changes in the RGB frequencies and do so accurately, a calibration experiment was created. Below there is a list of materials included in the experiment and the data from the experiment. To begin a circuit was created where the RGB sensor was connected to the micro controller in the SCL and SDA inputs. The led light was plugged into ground and the voltage wire was placed in series with a 1 ohm resistor and then to the 3.3V on the micro controller. Female wires were used to connect the sensor to the FRDM-K64F so that the sensor could be held upright on a screwed in stand in front of the led light. Also another mounted stand was used to hold the led light about 5 cm in front of the RGB sensor. In between the light and sensor there is a stand that holds light filters which are to be used to create a controlled variable in this calibration. In the experiment I used a red, green and blue led light to check the sensor's receptivity to changes in each color frequency. For each led light, 5-6 trials were performed, starting with no light filter and then optical filters that increased in absorption with each trial. This was to indicate that the sensor reads the decrease in the specific light frequency correlating to the color of the led used. The experiment was completed in a dark room, and I placed a cardboard box and black cloth over the system to block out any excess light. Each trial lasted 60 seconds and then data was collected for the average amount of Red, Green and Blue light. In each test, the sensor reading for the amount of light frequency that corresponded to the led light, decreased as the filter absorbency increased. This shows that the sensor is able to accurately discern changes in RGB frequencies. Our team will be able to use the data collected by the RGB sensor during flight to reliably determine the relationship between voltage output of solar panels and quantity of Red, Green and Blue light frequencies. \begin{equation} V_\mathrm{LED} = \frac{1240\ \mathrm{V \cdot nm}}{\lambda} \end{equation}








 * The calibration process that I completed did not reveal a linear relationship where light intensity increased as the transmission of the filter increased. The reason this happened was that in covering the light and sensor system for each trial, the led light was tilted in its stand. This extra variable made the data non-linear. However, the graph below shows that the data points of the green light intensity changed in consistent increments with the red light intensity. Thus, the problem was not with the RGB sensor being able to read data, but the data it recorded was affected by human error. The sensor was still able to create a consistent change in intensity with increasing light transmission, just not in a linear trend. To fix this for next time, tape could have been used to hold the led light in place to decrease the tilting problem.

Validation and Calibration for UV sensor
Validation and Calibration: 3/31/16

Since it is difficult to test the UV sensor's ability on a basic level, the calibration experiment for the UV Spark Fun sensor will be sufficient to asses if the sensor is able and accurate at reading changes in UV light. Below there is a table of the materials and UV sensor readings during the experiment. In the experiment the same set up is required for the UV sensor as the RGB sensor calibration experiment. I made sure to wear plastic protective goggles during the experiment because UV light is harmful to the bare retina. The same method was utilized by using stands that were screwed into the table to hold the UV sensor and UV led light. Varying light filters were used to determine if the UV sensor was systematic in having decreasing UV light readings with increasing absorbency in the filters. To set up the circuit, female wires were connected the sensor to the analog input (PTB2) and plugged into the UV led light into ground and 3.3V on the micro controller. The UV light has a lower wavelength and required a higher voltage to illuminate the led light (V=1240V*nm/wavelength). So a 100 ohm resistor was used and plugged the light into the 5V input on the FRDM-K64F. There were 6 completed trials, starting with no filter and then a different filter with increased absorbency for the rest of the trials. There was a base line of about 0.2-0.3 mW/cm^2 for the amount of light detected, which I verified by unplugging the UV led light. However there was a steady decrease in the quantity of UV light validating that the UV sensor is accurate in detecting changes in UV light. The sensor will be able to provide trustable data during flight that will help answer the question analyzing the dependency of solar panel voltage and varying UV light.



Data and Analysis


Data shows that although the temperature on the outside of the pod rose with height, the temperature on the inside of the pod remained relatively constant. The relationship between Panel voltage and Temperature were shown to be very random, this suggests that temperature has a low effect on the efficiency when used within the medium temperatures of an eastern Washington spring. The panel voltage vs UV intensity shows that as the intensity of ultraviolet rays rises so does panel voltage, this is accurate because solar panels convert UV into energy well. The color spectrum data graphs show that red light has a strong correlation with panel voltage however green and especially blue don't have much effect on the photovoltaic production of energy.The panel voltage is shown to increase with total light, this is again because of the way solar panels produce light into voltage.

Conclusion


A past Whitworth near space research group completed an experiment to analyze at what height both poly and mono crystalline solar panels were most efficient. At the end of their experiment they posed a question to further researchers asking if there were specific factors that increased the panels' efficiency more so than other variables. This lab sought to answer this question by showing the impact temperature and the light intensity of different wavelengths had on solar panels. Our data is somewhat incomplete because we had a coding problem that only recorded data during the first half an hour and one data point every half hour. The data only recorded the sensors' readings while the pod was still on the ground and a few minutes into flight. The data does not show the relationship between variable intensity of light at high altitudes and panel voltage, however it still is able to answer part of the research question. Our original question asked how the visible light spectrum, UV light and temperature affect the voltage output of photovoltaics. The data collected during this experiment is able confirm our hypothesis that there is a correlation between intensity of light and higher voltage, but we are unable to confirm our hypothesis that temperature has an impact because there was no data recorded at higher and colder atmosphere.

From the graph that shows voltage changing with red light intensity, it can be deduced that with a higher intensity of light with 620-750 nm wavelength (red light), the solar panels created more voltage. Since red light has a lower energy, the band gap in the panel can absorb the light easier. Since the solar panel can absorb most of that specific light, the greater amount of red light allows more of the light to be converted into voltage without any loss due to heat- with increasing wavelength, the efficiency of the photovoltaic increases. In contrast, the graphs of blue and green light show that there is not a large correlation between high intensities of light with wavelengths of 450–495nm (blue light) or 495–570nm (green light) and increased voltage. The data shows the same response with UV light. This was to be expected since green, blue, and ultraviolet light have higher energies, and the efficiency of converting the light into voltage decreases. To convert the majority of light with higher energy into voltage, the panel needs to be able to absorb the light with shorter wavelengths. But the second law of thermodynamics requires that the energy absorbed must equal the work done by the panel plus an amount of emitted energy. The panel looses energy though heat or if light frequencies are too large for the conduction and valence bands to absorb. The monocrystalline panels need energy to do work across the band gap and create voltage, so if most of the energy is un-absorbed or lost due to heat, there shouldn't be a high voltage output. This process was reflected in our data. On the graph with combined red, green and blue light intensity, it is shown that voltage increases linearly with higher amounts of light. We have very high readings of light intensity even though the pod was on the ground for the majority of the time. This can be attributed to the reflection of light off of the earth through shinny objects. Although it is unknown if there is a spike of red, green or blue wavelengths higher in the stratosphere, the original question can be answered. Higher intensity of light increases voltage, but the efficiency of the panel to create voltage depends on light with a higher wavelengths and lower energy. The other factor that could have impacted voltage was temperature. In theory it is known that colder temperature allows electronics to function more efficiently. However in this experiment the pod did not record voltage output at the height were colder temperature is able to make an impact. For this reason the data shows that voltage output was not impacted by temperature, because the pod was collecting data on the ground where it was warm and not cold.

The areas that needed improvement for a future experiment would be to double check the reading of the timer. We needed to check that the time went up in increments of 0.5 sec. all the way up to the final amount of time. Instead we saw that the timer went for the full three hours and did not catch that 8 of the last time points jumped about a thousand seconds each. Another thing that was improved but at a cost, was the knowledge of how to connect temperature sensors. The loss of a micro controller was the price for future knowledge on how to better test temperature sensors. Our code was very important in this experiment and we learned that it is more prudent to create one line of code, test and then add to it. We had to restart once after a couple days of coding, to create multiple files of different code to effectively test the RGB, UV and temperature sensors. If we could repeat this experiment, we would also run one more test on the RGB sensor to see at what maximum light intensity the sensor could function. This was not a problem in our experiment or analysis but there was time where the readings of red, green and blue light which was inconvenient and possible preventable. Further research that could be done would be to more accurately measure one variable and see the actual impact temperature has on voltage output at given heights. To do this, this experiment could be repeated but record data at higher altitudes where temperature is a bigger factor. Also, the ability of solar cells to absorb the majority of light energy varies with the size of the band gap. It would be interesting to use solar cells made out of different materials to see which material is the most efficient at absorbing the most light.