Dependence of photovoltaic performance on light spectrum

Research Group: Jackson Bell, Nicholas McGill, and Utsal Shrestha

Launch: Whitworth Spring 2017

Background
Data nowadays show that with the increasing use of fossil fuels as energy sources, global warming is rising and natural habitats of various species are depleting. As a countermeasure, scientists and businesses are investing their time and money in renewable (and clean) sources of energy, the prime one being solar energy. This is because of the promising possibilities solar energy offers with minimal cost in the energy market.

How it works: The photovoltaic cell is based on the photoelectric effect. When photons from sunlight hit the solar panel, they eject electrons from the metal. This flow of charge is converted into a specific voltage by the solar cell. There is a minimum energy required for the photons to eject electrons, associated with the cutoff frequency ($$E = hf$$ where $$E$$ = energy of the wave, $$h$$ = planck's constant, $$f$$ = wave frequency). Any light below that frequency will not succeed in ejecting electrons. There is a minimum threshold but not a maximum one. But as it turns out, only a limited portion of the light spectrum can generate electricity. Better photovoltaic cells are being manufactured nowadays which can make use of the full spectrum of sunlight.



The Experiment: The experiment measures how the voltage generated by the cells vary with change in temperature, UV light intensity and visible light intensity. It seeks to understand the factors that affect the efficiency of the solar panels. These data set help understand the optimum points of the data sets where maximum voltage can be generated.

Mechanical Design
The eight monocrystalline solar panels are mounted on 3D-printed brackets which are epoxied onto the outside of the pod. There are two panels and each sidewall leaving space for straps in the middle. Two holes were drilled for each cell: 4 holes on each sidewall for the solar panels. Some holes remained from the previous group, so we tried to use them as much as possible. The TMP102 is mounted onto one corner of the pod. The UV sensors are mounted on opposite corners from each other to read the most accurately as the pod freely moves and spins. The RGB light sensors are also mounted on opposite corners from each other for the same reason. Each corner will have a sensor, and one of the corners will have two (UV/RGB and TMP102). All of our SparkFun sensors are soldered to wires that connect them to he mBed where our data is being recorded. The mBed is mounted inside the pod and securely fastened with solder onto our perf board which is then mounted inside the pod using zip ties. Similarly, our two battery packs are also secured inside of the pod using zip ties running both inside and outside of the pod.

Electrical Design
The main processor was the mbed LPC1768, which was powered by a single battery pack consisting of four (4) Energizer Ultimate Lithium AA batteries.

Construction
The electronics board was fabricated on a 15.4 x 5.2 cm piece of 15 mm thick perforated board. 22-guage multicolored jumper wires were used to connect points and make discerning the various connections from each other an easier task. 0.8 mm lead-based, flux core solder was used for all solder connections on the board and splices between component wires.

Code
We used the following public/open-source libraries for our code:
 * mbed
 * SDFileSystem
 * ISL29125
 * ExtendedTimer (Dr John Larkin, Whitworth University)

Basics: The system stores the data in a file in SD card. Data is stored as tab separated values, which makes future analysis easier and efficient. The system prints data every second so even during an ascent of 90 ft/min, important data points will not be missed. After 3 hours of running, the program will stop executing and start the alarm (which makes finding the pod easier).

Saving data sets: The data stored are:
 * Time(s)
 * Temperature(°C)
 * UV intensity reading (from 2 sensors)
 * RGB values (from 2 sensors)
 * Voltage generated

Precautions: If the program fails to mount the SD card or open the file, the LED lights signal the failure. Every 5 minutes, it closes the file and reopens it 10 times to make sure that it is appending new values. So the maximum time for data loss due to file handling failure is 5 minutes.

Link to the Final code

Validation and Calibration for SparkFun Digital Temperature Sensor Breakout
The temperature sensor, TMP102, measures the temperature on the outer surface of the pod. The sensor was validated on March 9, 2017 when we connected it to the micro-controller and the sensor’s readings increased as body heat was applied to the sensor. Using puTTy, the values rose as body heat was applied and then fell when body heat was removed.

The temperature sensor was then calibrated using various locations along with a FLUKE thermometer. First, the TMP102 sensor was set at "room temperature" with the FLUKE thermometer as close as possible to the TMP102; the temperature sensor read 23.69 degrees Celsius and the FLUKE thermometer read 23.7 degrees Celsius. At "room temperature", we had a deviation of 0.01 degrees Celsius. Next, the TMP102 sensor and FLUKE thermometer inside of a the physics' departments freezer; the temperature sensor read -0.75 degrees Celsius and the FLUKE thermometer read 0.7 degrees Celsius. In the freezer, we had a deviation of 0.05 degrees Celsius. Last, the TMP102 sensor was placed directly in the stream of a heat gun along with the FLUKE thermometer (as pictured); the temperature sensor read 48.44 degrees Celsius and the FLUKE thermometer read 49.8 degrees Celsius. Under the heat gun, we had a deviation of 1.36 degrees Celsius.

Given all of this data, we concluded that the TMP102 sensor performed well in room temperature and below. This worked perfect for us considering that our live testing environment was about room temperature on the ground and colder in the air.



Validation and Calibration for SparkFun RGB Light Sensor
In order to validate that our RGB sensors were working properly, we turned them on in the lab and used shading to observe the light readings rise and fall. This was performed using puTTy. When the sensor was covered the sensor read zero and when the sensor was open to the light within the lab the sensor read around 250 for red, 1000 for green, and 500 for blue.

For calibration, red and green LEDs were used along with six absorbency filters on the RGB sensor. The readings were consistent with what was expected. As the filter absorbency rose, the RGB sensor's reading declined. Our graph altered the filter absorbency value (OD) to $$10^{-\mathrm{OD}}$$ in order to show a linear trend between sensor values and filter absorbency values. This is why the graph reads transmission instead of simply OD. Since we did include two separate sensors on our pod and we were strenuous on time, we only took complete data on one of the sensors. For the other sensor, we lined it up in the exact same position as the completely tested sensor and read those values. The values were nearly identical, concluding that the two sensors were zeroed at about the same position. Overall, the trend in light was more important to us than knowing exactly how much light was available in to the pod/photovoltaic cells. Because of this, we were okay with knowing the sensors read similar numbers rather than calibrating them to read the exact same all the way through.

Validation and Calibration for SparkFun UV Sensor Breakout
Validation and calibration for the UV sensors was done in the exact same fashion as the RGB sensors.

Validation consisted of programing both the sensors and then running them on puTTy. On puTTy we were able to observe the values rise as more surrounding light hit the sensors and watch them fall as the sensors were covered. In doing this, we confirmed that the sensors were both working properly as well as the coding used to run them.

Calibration worked the exact same as the RGB sensors. We thoroughly tested one of the sensors using a series of absorbency filter along with a UV LED. After gathering the data for that sensor, we set both sensors in the exact same spot in the classroom. We observed nearly exact number which was once again enough for us to send them up and gather a trend of UV light data to compare to our trend of photovoltaic voltage output.

Reading of UV sensor
As the sensor was backed away from the UV light source, the corrected UV reading reduced with the relationship of 1/(distance)^2



Conclusion
Concluding our experiment is a bit bitter-sweet for us. Unfortunately we were not able to read data for the entirety of the flight, yet we were able to get enough data to see some promising trends. Our overall research question developed into "How will temperature, visible light intensity, and UV light intensity effect the efficiency of photovoltaic cells?" We did some quick initial research and come to a final hypothesis for the tree separate variables measured against the efficiency of photovoltaic cells: After graphing and analyzing our data from all of the sensors, we observed a few trends. First, in an attempt to support our first hypothesis, regarding temperature, we graphed temperature as the independent variable with overall photovoltaic as our dependent variable. Given that we did not collect data the whole flight, there was only one obvious drop in temperature. As the temperature was at its lowest, the photovoltaic voltage reached its peak. This appears to be at the end of a linear trend moving down in temperature and up in photovoltaic voltage. Next, we constructed four separate graphs to test hypothesis number two, regarding visible light intensity. Unfortunately, this was another pitfall to our data. Our visible light sensors were not able to continue collecting data once they had been maxed out. Furthermore, we still successfully observed trends supporting our hypothesis. In all three of the separated (red, green, blue) graphs compared to photovoltaic voltage, the trend was light intensity increasing while photovoltaic voltage also increased. This trend obviously remained the same for our graph of total light intensity and photovoltaic voltage. Last, we created one more graph to compare the average UV light intensity with the photovoltaic voltage. It is possible that the lack of data effected this third hypothesis the most, yet there were no trends shown in our favor. Based off of further research and the brief dat we were able to collect, our third hypothesis has been proven incorrect. UV light appears to have no correlation to the overall output of photovoltaic voltage. For future experiments, we would like to create a more sound system to ensure a full collection of data. Also, we would say to ditch the idea of UV light intensity effecting photovoltaic cells because the sensors took up too much time that could have been spent testing and calibrating in other areas. This could have possibly given us the extra time we needed to fix the bug that rejected the collection of data past the hour or so we received data for. Another huge improvement that we could make and should have made is the use of more capable RGB sensors. The fact that we maxed out the sensors less than an hour into the flight shows that a full flight of data possibly wouldn't have been much better. We strongly suggest purchasing RGB sensors with a much higher maximum interpretation of data and also testing the RGB sensors at extreme light intensities before flight.
 * When the temperature of a photovoltaic cells is reduced, the efficiency of that cell will be increased due to the cells ability to more effectively capture the excited photons energy.
 * When the intensity of surrounding visible light is increased, the overall voltage output from the photovoltaic cell increase as well.
 * When the intensity of surrounding UV light is increased, the overall voltage output from the photovoltaic cells will increase as well.

[Please refer to the Data and Analysis section for all references to data]

References/Works Cited
https://developer.mbed.org/platforms/mbed-LPC1768/

http://butane.chem.uiuc.edu/pshapley/GenChem2/A2/1.html

http://lasp.colorado.edu/~bagenal/3720/CLASS5/5Spectroscopy.html

http://www.livescience.com/41995-how-do-solar-panels-work.html

http://www.newswise.com/articles/altitude-increases-sunburn-risk

http://www.nss.org/settlement/ssp/

http://www.wrh.noaa.gov/fgz/science/uv.php?wfo=fgz

https://en.wikipedia.org/wiki/Electromagnetic_spectrum

https://en.wikipedia.org/wiki/Solar_cell

https://en.wikipedia.org/wiki/Ultraviolet

https://science.nasa.gov/science-news/science-at-nasa/2002/solarcells/

https://www.nas.nasa.gov/About/Education/Ozone/radiation.html