Dependence of solar panel performance on exposure to cosmic radiation

__MATHJAX_DOLLAR__ Daniel Blomdahl, Jorin Graham, and Zach Halma

Launch: Whitworth Spring 2016

With an increasing amount of technology being sent into near space, much of which is powered by solar panels, it is important to learn how the environment of near space effects solar panel performance. We chose to research the effect of cosmic radiation on polycrystalline solar panels. We did this by comparing voltage from the solar panel to radiation counts from an in-pod geiger counter. We also added two thermocouples to isolate voltage vs radiation from other variables in solar panel performance. Unfortunately, due to several circuit malfunctions, we did not receive useful data. We conclude by suggesting ways to improve similar research in the future.

Background
An increasing amount of technology is currently being sent into near space, much of which is powered by solar panels. It will thus be important to know whether cosmic radiation in near space will effect the performance of solar panels. Our project will focus on determining the effect of cosmic rays on polycrystalline solar panels. We will accomplish this by measuring the level of atmospheric radiation and the effect of this radiation on the immediate and permanent voltage output of the solar panels.

Radiation in Earth’s atmosphere takes several forms and originates from space, from the ground, and from particles in the atmosphere. However, for the purposes of this project, we are chiefly concerned with cosmic rays. Cosmic rays were discovered in 1912 by Victor Franz Hess. A form of high energy particle radiation, they are primarily constituted of protons and alpha particles. There are two main types of cosmic rays: primary and secondary. Primary cosmic rays are those particles which originate in space and enter the earth’s atmosphere. Secondary cosmic rays occur when primary cosmic rays interact with molecules in the atmosphere creating showers of lighter particles. As a result of the conversion of primary cosmic waves to secondary cosmic rays, the peak occurrence of cosmic rays takes place at about 15 km. The number of occurrences of cosmic rays as well as other forms of atmospheric radiation can be counted using Geiger counters. We are assuming for this experiment that most of the counts received from the Geiger counter will be due to cosmic radiation and not to other forms of radiation in the atmosphere. Other forms of radiation will create noise in the data, however, they will not significantly alter the data. We are making this assumption for the following reasons: (1) ground radiation decreases with altitude and should have no or little effect on the data once the pod is in flight; (2) the Geiger counter will be shielded by the shell of the pod and as such should pick up mostly high energy radiation such as cosmic rays; (3) cosmic radiation increases significantly in the atmosphere. In addition, we believe that cosmic rays will have a greater effect on solar panels as compared to other forms of radiation because: (1) they are high energy; (2) previous research by the International Business Machines Corporation has shown that cosmic rays can harm electronics devices.

In general, radiation can have several different effects on electronic devices, but only one, a single event effect, will be applicable to our experiment. A single event effect, which may or may not be destructive, occurs suddenly and quickly when a charged proton enters an electronic system. Since the other types, total ionizing dose and displacement damage deal with radiation build-up over time, they are not applicable to our experiment as the solar panels will only be exposed to high levels of radiation for a small period time.

We will measure the radiation levels using a Geiger counter. As such, we will be unable to measure the intensity of the radiation but only the number of radiation counts. In addition, there is a period after each count called "dead time" during which the Geiger counter is unable to pick up additional counts. Thus, the Geiger counter will be unable to count occurrences that are too close together.

Besides measuring the level of radiation, our research will also require the measurement of the temperature of the solar panels during flight. Data from a past Whitworth Near Space project (Evaluation of solar panel performance in Near Space), reveals that the output of polycrystalline solar panels are positively and linearly related to decreasing temperature. Using this data as a model, we will look for abnormal changes in the voltage output of solar panels during the flight.

We will also measure the voltage output of the solar panels in relatively controlled laboratory conditions before and after the flight. The data gathered in the first test will act as a baseline to which we can compare the voltage output after the flight and see if any changes in the ability of the solar panels to produce voltage has occurred. Our hypothesis is that cosmic rays will cause permanent damage to polycrystalline solar panels resulting in a decreased ability to produce voltage.

Mechanical Design
The primary components of the pod were the four solar panels microcontroller with breadboard, and the Geiger counter. We also housed a second microcontroller and breadboard system for another group's testing of pod motion throughout the flight. We attached the solar panels to the top of the pod, one on each corner, by mounting them on cardboard using epoxy and attaching the cardboard mounts to the pod using duct tape. We placed the Geiger counter inside the box, with the detector pointing upwards to minimize radiation readings from the ground. The temperature of the solar panels was measured using thermocouples. Thermocouples were used instead of thermometers because it gave us the ability to read the temperature on the surface of a solar panel instead of the atmospheric temperature. We used two thermocouples to measure the temperature of two of the solar panels on opposite corners. We ran them through holes in the lid, placed them in contact with the solar panels and secured them using electrical tape.

Besides the geiger counter, all of the other components were secured with zipties and bolts or to baseplates which were directly bolted to the walls of the pod. A challenge was attaching the breadboard and microcontroller to the baseplates in a way that both components could be reused again after the experiment was over. That meant we could not use epoxy, so we used small bolts to attach the microcontroller to the wooden baseplate through small ports on the corners that were designed for securing the microcontroller to another object. To attach the breadboard, we put an orange plastic covering (a remnant of an initial 3D-printed part that we tried to use to attach the solar panels to the pod) on top of a portion of the breadboard. A screw was then used to attach the plastic housing to the baseplate, creating a clamp that held the breadboard securely. We finally added a half-inch-tall piece of wood between the breadboard and the baseplate so that the ports of the microcontroller and the breadboard could be flush with each other. This greatly helped in the strength of the wires staying secure between the two components.

A slight design problem we came across was the inserting and taking out of the SD card from its port in the FRDM-K64F once the circuitry was attached in the bottom of our pod. This was solved by cutting a hole into the pod close to where the SD port was situated. This allowed the SD card to easily be taken out with needle nose pliers for calibration and testing. For the final launch, we covered the hole with duct tape to protect the inside of the pod.

Circuit
The 4 solar panels were all soldered together in series and their voltage then divided using a 100K ohm potentiometer, so that the voltage could be at a manageable level for the FRDM-K64F. The thermocouples were connected to the FRDM-K64F together as slaves in digital inputs, and the microcontroller served as the master and directed which thermocouple to collect data. Each thermocouple was also connected to a 0.1 mF capacitor to ensure more precise data collection. The breadboard system and the microcontroller were powered by a 9 V battery pack consisting of 4 AA batteries. The voltage reading from the solar panels was wired to an analog input. The data was saved on a microSD card. To transfer the data to a computer for analysis, the microSD card was ejected from the microcontroller and inserted into a computer.

The geiger counter was independent system with its own microcontroller and power source. It was powered by 3 AAA batteries, and the microcontroller saved the data directly into its own memory. The data was then accessed through the microUSB port.

Code
Coding for the electronic data collection relied on public published files for the MAX31855 thermocouples using the mbed website. The public .cpp and .h files were accessed from the created main.cpp file. The main.cpp consisted of while and if loops to continually collect data from the thermocouples and the solar panels. The code begins with a function mountSDCard to make sure that the SD card is mounted. The function openDataFile is then called, which starts the data collection. Because the timer of the FRDM-K64F has a limit of a little over 30 minutes, the data collection is a while loop that resets the time until there has been seven 30 minute cycles, or 3.5 hours. The temperature from both thermocouples and the voltage from the solar panel is collected every 0.5 seconds. Once the 3.5 hour time has been completed, the closeDataFile function is called, and the SD card is safe to be ejected form the microcontroller.

The data collection is automatically started when there is a power source (In this case, when the battery was turned on). As a visual check to make sure the data collection has begun, a red LED will turn on for 0.5 seconds, followed immediately by a green LED for 0.5 seconds. After the data collection session has finished, a green LED will turn on continually to show that it is safe for the SD card to be ejected.

Link to final code.

Validation and Calibration
Solar panels: we validated the solar panels by ensuring that they produced a voltage when exposed to light and that this voltage changed when the intensity of the light source changed. We used two different light sources for this: indoor lighting and sunlight. However, we did not check that the voltage reading we were receiving corresponded to the expected voltage readings and as such to the actual voltage being produced. This assumption eventually led to incorrect data from the experiment.

Geiger counter: we validated and calibrated the Geiger counter using two sources of radiation for which we knew the expected readings. The first source was background radiation. For this source, we simply allowed the Geiger counter to sit in a normal environment without known sources of extra radiation for an extended period of time and then examined the counts received to ensure that they where as expected. We expected readings of 25$$\pm$$5 counts per minute with a 95% confidence interval and received readings in this range. Our second source was a plastic disc containing strontium. For this test, we placed the Geiger counter and strontium disk in close proximity. We then placed led bricks around the strontium disks so as to minimize radiation exposure. We expected and received readings between 100 and 300 counts per minute. Because we received the expected readings at two different levels of radiation, we concluded that the Geiger counter was a reliable data collector for the purposes of our experiment.

Thermocouples: we had two thermocouples which we calibrated separately. We did so by placing them in a beaker filled with well mixed ice-water and heating it to boiling using a UL® type 1900 hot plate 374B model HP-A1915B (ID# 002099). We calibrated them to a Fluke 52 thermometer (ID# 002116) by allowing the Fluke 52 to measure the temperature of the water in which we had placed the thermal couples and treating the temperature measured by the Fluke 52 as the real temperature. We found that the thermocouples gave somewhat different temperatures than the Fluke 52 but that they had an essentially linear relationship to the temperature measured by the Fluke 52. However, we forgot to record the temperature difference prior to connecting them to our pod (the difference between measured temperature and real temperature was less than 2 degrees Celsius for both thermocouples). Originally, we planned to recalibrate the thermocouples afterward. However, because our circuitry for our solar panels malfunctioned, we did not receive useful data from the solar panels and as such the temperature of the solar panels was no longer needed.

Data and Analysis
We performed our pre-flight and post-flight testing in the basement of the Eric Johnston Science Center at Whitworth University, on a table just outside of room 127. We chose this location because it limited the number of uncontrolled variables we would need to deal with for our post flight testing. Firstly, this area was not near any windows so the temperature would not be effected by having an open window and as such would be similar for both tests. Secondly, by performing it late at night, we did not receive significant light from outside or from the rooms near the table, so the light exposure was similar for both trials. This testing allowed us to compare the output of the solar panels before and after the flight and see if any significant change had occurred. It is important to note that the voltage divider was replaced between these two tests as we had discovered that it was originally set up incorrectly. The new voltage divider was adjusted so as to have the same effect on the voltage output as the previous divider. However, our data revealed that the voltage measured in the post-flight test was much higher than that measured before the flight. This difference is almost certainly a result of the fact that the voltage divider had originally been set up incorrectly and does not correspond to actual voltage output.

During the flight, the circuit malfunctioned, so we did not receive voltage data from our solar panels. However, we were still able to measure radiation and altitude.

Note that the sudden jumps in radiation counts per minute seen in figures 9 and 11 probably do not correspond to an actual increase in solar ray activity but instead to sudden violent jolts to the pod. The radiation spike is too narrow to be due to an increase in solar ray activity above a certain altitude and too wide to correspond to a single primary incident of radiation. If the increased radiation counts was due to altitude, we would expect the increase to be less sudden and for a symmetrical increase to occur during the ascent. If the increase in radiation counts was due to a single cosmic ray event, we would expect the increase to last for a much shorter period of time.

Note that the sudden jumps in radiation counts per minute seen in figures 9 and 11 probably do not correspond to an actual increase in solar ray activity but instead to sudden violent jolts to the pod. The radiation spike is too narrow to be due to an increase in solar ray activity above a certain altitude and too wide to correspond to a single primary incident of radiation. If the increased radiation counts was due to altitude, we would expect the increase to be less sudden and for a symmetrical increase to occur during the ascent. If the increase in radiation counts was due to a single cosmic ray event, we would expect the increase to last for a much shorter period of time.

Conclusion
Unfortunately, we were unable to draw conclusions as to whether cosmic rays effect the performance of polycrystalline solar panels for several reasons: (1) The voltage divider was incorrectly set up during the pre-flight test and flight. This caused the voltage data collected during the pre-flight test to be random and not reflect the actual voltage produced. (2) The circuitry malfunctioned during the flight causing us to receive a constant voltage reading during the flight. (3) The programming limited the data recorded to two decimal places, limiting the precision of the data we did receive. The limited precision was not corrected for the post-flight test, but the voltage divider and the circuitry functioned as it should have.

In order to improve future experiments, several steps should be taken: (1) More rigorous testing should be taken to be sure that the voltage being measured is indicative of the voltage being produced. We had assumed that during the testing that the changes in voltage reading was caused be the change in voltage produced while in reality the readings we were receiving were random. (2) The code needs to be edited so as to allow higher precision in the data recorded.