Senior Design II ECE 4902 Spring 2018 Solar Mobius Final Report Team 1821 Members: James Fisher (CMPE) David Pettibone (EE) George Oppong (EE) Advisor Professor Ali Bazzi Sponsor University of Connecticut School of Engineering
Background & Existing Design The Mobius Solaris statue was built in 1995 and designed by artist Robert Perless. Located in front of the Castleman building, this statue features a stainless steel base connected to five polycarbonate prisms rested on a stepper motor. The stepper motor is wired to the base using slip rings, allowing the prisms to rotate 360 degrees while keeping the base itself stationary. The intention of this artwork was originally to track the sun and reflect a rainbow onto the main entrance of Castleman, but that was never achieved due to various problems. The existing electronics located in the base of the statue have only one purpose, which is to rotate the prisms at a constant rate of 360 degrees every 24 hours. A block diagram showing the current setup is shown below in Figure 1. Figure 1: Current state of the Mobius Solaris Statue. The controller is a simple circuit that handles the logic telling the stepper motor driver when to turn, and how many steps. The stepper motor is currently set to rotate at a rate of 8,000 steps per full revolution.
Project Objective Our design project was split into two separate tasks. The first task was designing a solar tracking mechanism to acquire the position of the sun and, using this information, reflect a rainbow above the main entrance of the Castleman building. This is a task that we planned to accomplish using a simple circuit with two branches, each consisting of a photoresistor in series with a static resistor. Using this circuit, we would obtain angles of the sun and correlate the angle of the Mobius to reflect a rainbow at certain times throughout the day. The second task given to us was that of creating a website specifically for the Mobius. This website was only supposed to feature very general information about the Mobius and the solar tracker circuit, such as the pulse counts of the stepper motors, and could possibly be expanded upon in future projects. Light Sensor Design Our proposed design for our solar tracker was to mount a light sensing circuit onto a separate stepper motor. This motor will spin until the two voltage readings from the photoresistors in the circuit are equal, at which point we will know the angle of the sun with respect to a reference point we manually set. A circuit diagram and its physical representation can be seen below. Figure 2: Light Sensor Design physical implementation (left) and circuit diagram (right)
This circuit works by using an opaque plate so that the two photoresistors see different levels of light in most cases. This will allow us to see which side of the circuit the light is coming from and rotate accordingly until the two photoresistors have the same reading, which will mean that the opaque plate is lined up with the sun. This will then be sent to the Mobius driver and the statue will rotate according to where the sun is. Full Design Our full design is comprised of our light sensor design described above along with the Raspberry Pi, stepper motors, and drivers. Our design functions by reading the pulses sent from the light sensor to determine the relative angle of the sun, and relaying this information through the Pi in order to move the Mobius accordingly when it is possible for a rainbow to appear on the front of the Castleman building. The Raspberry Pi also accounts for the times of day when it is not feasible to display a rainbow. During the night or when light levels reach an acceptable low amount, the Raspberry Pi will have the stepper motor of the light sensor and the stepper motor of the Mobius return to reference. This will allow it to save some energy, and is also convenient as the majority of angles where a rainbow is possible is when the Mobius is at reference. An outline of our design, had it been implemented, can be seen below in Figure 3.
Figure 3: Block Diagram outlining the system with the Mobius In order to implement our design into the statue, we had to do some experiments to determine when exactly it was possible to obtain a rainbow, as well as determine the most optimal way to control the Mobius. Through experimentation at different frequencies, we were able to experimentally determine that the Mobius is best controlled at a resolution of 8000 steps per revolution. We also determined, at least during the months of March and April, that it is only possible to display a rainbow on the center of Castleman at the times of between about 9:00am and 11:30am. This rainbow occurs when the statue is set to our reference angle, which is directed east. We also observed one more data point where the statue is oriented 90 degrees clockwise from our reference. This time was observed to be at 11:23am, though the time will definitely change throughout the year. It was determined that the goal could not be achieved in the afternoon due to the rainbow only appearing on the far right side of the building instead of centered over the door as requested. This is in part due to the placement of the statue not being aligned with the center of the Castleman building.
Figure 4: Prototype circuit for our full design with a placeholder driver and stepper motor to replicate the Mobius The circuit seen above in Figure 4 is our completed prototype circuit. This circuit is a replica of the system seen in Figure 3, except instead of being connected to the Mobius, our light sensor is connected to a separate driver and stepper motor. However, both of these drivers would receive the same inputs, so the design itself would act the same. This circuit accurately tracks light and responds accordingly in a laboratory setting, which implies that the circuit would work if implemented in the Mobius. It was unfortunately not implemented for a variety of reasons, with the largest reason being an overheating issue due to a bad choice in the stepper motor driver connected to the light sensor. We had also planned on using a PCB in our design, but the designed pcb ended up not working as it was supposed to. Through debugging, we were able to figure out the underlying problem as to why the PCB was not working. The first problem was the fact that the ADC in the circuit was not receiving the power it needed to operate but rather it was
only receiving half of its needed power. The second problem was that the 5V provided by the light sensor s driver board was interconnected with the 5V provided by the pi, this led to the pi coming to a sudden shutdown whenever the grounds of these two components were connected to each other. These two problems were later corrected in the pcb design but unfortunately there was not enough time to order a new pcb and have it implemented in our overall circuit design. With this in mind the team chose to build our circuit on vector boards instead. A picture of our PCB, with the driver and ADC attached, can be seen in Figure 5. Figure 5: Designed PCB Software/Raspberry Pi Code The Raspberry pi can only read digital voltage signals so all the inputs going to it are first passed through an analog to digital converter. With this, the Raspberry pi can determine which of the two branches of our solar tracker has a greater voltage. Based off the branch with the greater voltage, it will set the direction pin on its GPIO to high or low, with low corresponding to clockwise direction, and high being counterclockwise. After each voltage reading, it will send a pulse to the sensor s stepper motor in the direction of whichever side had a greater voltage.
Doing so at a rate of one pulse about every half second will allow for the sensor to always be able to adjust its angle and point its fin at the sun. Once the voltage reading for both sides of the sensor is equal within a small threshold value, it stops sending pulses, only sending more once one side has a greater voltage again. This will keep the fin on our solar sensor always pointed at the sun. The Raspberry pi is able to keep track of the angle of the sensor s stepper motor by incrementing a count value for each pulse sent. Knowing the count for the number of pulses sent to the motor allows us to keep track of the angle because we know the resolution of the motor and how many pulses have been sent. This count value correlates to a value on the look-up table for a specific angle of the sun where it knows a rainbow can be reflected onto the building. Based off the data found in our experimentation, when the count of the sensor s stepper motor is in range 45-55 the pi will know it is time to rotate the statue s motor by 90 degrees, which is 2000 steps in the clockwise direction. Doing so will put a rainbow right on the entrance of Castleman. The software also checks if the voltage readings on both photo resistors are both below a certain threshold, indicating that it is night time or a cloudy day where it is no longer possible for a rainbow to be reflected. At this point the Raspberry pi will pulse the solar sensor s stepper motor back to reference in the opposite direction it was moving to avoid the wires in the sensor to tangle up. Then it will calculate the shortest path to pulse the statue s stepper motor back to reference, and bring it back using that direction it found to be the shortest. The software is also mindful to make sure that the count value of the sensor s stepper motor cannot go past the maximum resolution of the motor before correcting its angle back to reference to ensure the wires attached to the sensor cannot get tangled or twisted. Lastly, the software implements a homing function that will ensure both motors are at reference every time the software first starts running. To accomplish this, we need to position a
very bright LED at our reference angle for the solar sensor that gets turned on at boot up, and also place a reed switch inside the statue at our reference angle. We would also need to attach a magnet to the shaft of the statue that will close the reed switch once the statue is positioned at reference. This would cause the solar sensor to pulse until it faced that LED and reported equal voltage levels across its photoresistors. After that, we would have the statue pulse until the magnet lined up with the reed switch, indicating that both statues are now at reference. After this homing phase is complete the code simply changes from the homing state to solar tracking state, where it shuts off the bright LED and begins to track the sun as normal. For our demo day circuit, we implemented a regular LED in our circuit to demonstrate the functionality of our homing function for the solar sensor. We show that the LED turns on at boot and the solar sensor turns until both sensors read an equal voltage, indicating it is facing the light; then it resets the counter and the reference is set. Next it turns off the LED and tracks the sun as usual. Since we do not have any reed switches and couldn t get any in time for demo day, we weren t able to show how the magnet would work with the reed switch to set the statue to reference, however the logic is the same for setting both motors to reference, and we can demonstrate that logic works with the solar sensor. On demo day we can show how our solar sensor tracks a flashlight instead of the sun since demo day is indoors. This way we can demonstrate the full functionality of our code and how it would be implemented with the solar sensor and the statue. Website The software generates and updates HTML code which is used to display our webpage on Uconn guest. We were assigned the static IP address: 10.66.59.9 to our Raspberry pi s MAC address, and used Apache to host a webpage on that IP address. The software updates the
webpage every 5 seconds with the current count values of the two stepper motors used in our design, as well as displaying some pictures of the statues and rainbows it can reflect. A photo of our designed website can be seen below in Figure 6. Figure 6: Current version of the Mobius website Components The Solar Mobius is rotated with a stepper motor as stepper motors are very inexpensive and can provide very high torque at low speeds. This motor has a total of 8-leads which allows for different types of configurations including; series, unipolar, or parallel, which allows for different number of application. This motor has a NEMA size 34 frame size, which has a high torque that goes up to 17,000 oz-in. At full step the motor rotates at 1.8 degrees per step which means it takes exactly 200 steps to make a full rotation.
Figure 7: Mobius Stepper Motor The driver being used to control the stepper motor is an M880 microstepping driver, which can be used to control both the pulse and direction of the motor. This driver is suitable for driving both 2-phase and 4-phase motors. It also has 14 selectable resolutions that goes up to 51,200rev/step. The driver has a total of 8 dip switches first three dip switches (SW1, 2, 3), are used to control current running through the driver. A 4-5V is considered a high pulse whilst a 0-0.5V is considered a low pulse. Specifications: High performance, low cost Supply voltage up to +80VDC Output current up to 7.8A Inaudible 20KHz chopping frequency TTL compatible and optically isolated input Automatic idle-current reduction
Figure 8: Mobius Motor Driver Since the light sensor is independent of the statue, we are going to need a different motor to control the rotation of the light sensors. Figure 9 show a picture of our chosen motor which is a 42MM Hybrid Stepping Motor. This motor operates at 12V and 350mA with a total of 200 steps for a complete revolution at full step. And with the Big Easy Driver ROB-128 as shown in Figure 9 we will be able to control the steps of the motor from full step all the way down to 1/16th of a step. This driver operates at 2-35V at 2A, it also has the capability of providing 5VDC which can be used to power the microcontroller. Specifications: Table 1: Stepper Motor Specifications
Figure 9: Light Sensor Stepper Motor Figure 10: Light Sensor Motor Driver For a microcontroller we chose to go with the Raspberry Pi 3 Model B mainly because of its internet capabilities. It has a built in wifi as well as an ethernet port which gives us to ways to connect to the web as the second phase of the project is to make a website for the Solar Mobius which has various general information about the system. The Pi is powered via micro-usb, requires a power supply of 5V at 2.5A for optimal operation, and has a total of 40 GPIO pins which is more than enough for this project.
Figure 11: Raspberry Pi 3 Model B Budget While our budget did not include any specific maximum value, we still are attempting to finish our project using as few funds as possible. An overview of our budget is shown below. 1. 2 AC-DC Power supply ($77.86) 2. 2 ADC ($19.90) 3. 2 Stepper Drivers ($ 39.90) 4. 2 Stepper Motors ($ 28.00) 5. 2 Raspberry Pi ($70.00) 6. PCB (~$50.00) 7. Miscellaneous Materials (~$50.00) Total= ~$335 We purchased two of everything we need so that we would be able to do testing in the lab as well as outside on the Mobius itself. Some costs that were eliminated were having to purchase a new stepper motor and driver that matched those in the Mobius. These would be priced at around 500 dollars if we were to purchase both of them, so we were able to acquire similar
stepper motors and drivers that we would be able to test in the lab and then simply transfer our working circuit to the Mobius. Overall, our expected budget for the project is just under 350 dollars, which is reasonable considering the materials we required to complete the design. The miscellaneous materials consist of many smaller items that we needed, including items that we did not end up using. Some of these items include an acrylic dome and aluminum plate that we were planning on mounting to the Mobius Solaris statue to house our design. Possible Future Improvements There are a few improvements that could be applied to our design in order to make it a more complete circuit and allow for seamless integration into the Mobius. The first, and the most important improvement that must be made, is to select a new driver for the light sensor stepper motor. This was the most important factor that led to us not being able to implement our system into the Mobius this semester. The problem with the driver we have selected is that it does not limit the current enough, supplying almost three amps to the stepper motor that is supposed to run on 300 milliamps. The motor still runs correctly, but gets very hot very quickly. The second improvement to our design, which we had attempted to do, was create a PCB that acts as an interconnect for all of the circuit components. This would allow our design to be much more compact, as we had planned to have the Raspberry Pi, driver, and analog-digital converter mounted on this PCB. Finally, the website created is very basic, and could definitely be made more sophisticated and aesthetically pleasing. Ideally, this website would be able to incorporate CGI scripts so that a user can enter a password and update the lookup table of possible angles as they are discovered, as we suspect that there will be more possible angles that allow for a rainbow in the late spring and summer months.
Conclusion Our assignment was rather interesting in that the functionality of the statue was not ideal, there was no math behind the creation of the statue, and the weather did not cooperate with us a majority of the time we attempted to test the statue. In large part due to these issues that were out of our control, we were able to determine that there must be more experimentation required in order to find more angles that can reflect a rainbow onto the front of the Castleman building. We were unfortunate in that the statue only had two possible angles in which a rainbow is displayed, at least during the time of the year where we were testing. The largest limiting factor to this, in our opinion, is that the statue was placed about 15 feet to the right, offset from the center of the building. Due to this, it is not possible to reflect a rainbow onto the center of the building throughout the entire afternoon when the sun is in the western part of the sky. However, we were still able to design and put together a working prototype circuit that accomplishes the required goals that we were tasked with. Minus implementation, we did successfully create a circuit that can track the sun, report the angle of the sun, and relay that information so that the Mobius reacts accordingly. We also have a website being hosted through the Raspberry Pi, which is very basic, but is a good foundation to be improved upon in the future. By those standards, our design was completed, but there are definitely improvements that need to be made before our design could be implemented.