Multimedia

Receiver Simulation with Location Calibration

Video 1: Receiver Simulating Randomly Chosen Data

Video 1 shows the receiver electronics simulating randomly chosen data from golf strikes.  The LCD displays the maximum force and the location of the strike.  The MATLAB program on the laptop shows the visual representation of the golf strike characteristics in a 3D plot, visually displaying location and force.  Four trials of simulation are shown in the video.  This video helps to support the overall accuracy of the location calibration analysis and confirms that Matlab and Arduino can communicate to each other in order to together display the golf strike characteristics.  

Driver Club Final Design

The driver club design shown in Figure 33 encompasses the driver handle, driver shaft, and driver head. The sensing strategy on the acrylic driver face was secured to the driver head. The main DAQ components were encased inside the driver head while other electronic components such as the wireless transceiver and battery were encased inside the driver handle. Cables connecting the components ran inside along the shaft. 

DAQ System Final Design

The final version of the analog components printed circuit board is featured in Figure 31. It includes a more compact layout on the front of the board to provide more room on the bottom layer for a more continuous ground plane. Grid-like power lines were routed as per IPC-2221a power layout recommendations. The break in the ground plane is down the middle of the board to run the signal lines, but ground strapping is included to provide ample return pathways for the signals. Ground strapping is done perpendicular to the signal runs. At the bottom of the board are two power in headers, bypass capacitance for the 5 V and 3.3 V power inputs, pads for an optional voltage divider, buffer amplifier,  programmable gain amplifier, and 16 to 1 multiplexer, all functioning in a similar fashion to the initial design. The digital board operates in exactly the same function as the initial design, with headers for the digital components in the system and a place to mount the Teensy 4.0 microcontroller.

DAQ System Initial Design

A major component of the Smart Clubs design is the data acquisition system (DAQ) in which analog signals from the PVDF piezoelectric sensors are converted to digital data that can be analyzed to understand the strike characteristics. Figure 28 shows the first iteration of the analog components board, which houses the signal conditioning and multiplexing circuitry used to sample all 16 sensors. Figure 29 shows the digital printed circuit board, which houses the Teensy 4.0 microcontroller, external memory, radio frequency transceiver inputs and outputs, and accelerometer inputs and outputs. The digital board also includes headers for writing to the analog components that require digital input, including the multiplexer and programmable gain amplifier. Figure 30 is a 3D CAD model of the PCBs in their complete stack up as they would appear in the potting mold. The voltage regulator on the top of the stack is used to convert 5 V power from the handle into 3.3 V power that is required by the digital components.

Driver Head Final Design

The final design for the driver head, shown in Figures 26 and 27, was adapted from the initial design, shown in Figures 20 and 21. The notable change was the addition of a larger offset between the edges of the acrylic face and the inner lip of the driver head. The distance between the back of the acrylic face and the entrance to the hollowed region of the driver face was also increased. These adjustments were made to better accommodate the overhang and bend radius of the PVDF sensors from the back of the acrylic driver face.

Electronic Component Encapsulation

The bilayer encapsulation method, shown in Figure 24, was done to mitigate the adverse effects of high shock on the printed circuit board (PCB) inside the driver head. The bilayer encapsulation method features an interior layer of hard potting compound (epoxy) that encapsulates the PCB. The interior layer of hard compound serves to provide rigidity to the PCB to prevent deflections that could damage the component. Then, an exterior layer of soft potting compound is applied to surround the hard potting compound. The soft compound (silicone) serves as a shock absorber to the shock induced by the impact event of the clubface to golf ball. Figure 25 shows the potting mold used to contain the electronic components that require potting. The dual height potting mold places the PCB stack to the rear of the mold. This allows for the majority of the mass of the entire driver head to be placed to the rear which improves the moment of inertia of the club, thus improving swing performance. While not shown in Figure 25, the PCB stack will be encapsulated using the bilayer method since the components in the stack are sensitive to the harsh mechanical stresses imparted from a golf swing.

Receiver Enclosure

The receiver enclosure, shown in Figure 22, was comprised of the Arduino Mega, NRF24L01 wireless transceiver, and a liquid crystal display (LCD) as shown in Figure 23. The enclosure was electrically connected by the Arduino Mega to a laptop that had MATLAB running. While the user practices their driver swings, the wireless transceiver receives the data from the transceiver in the driver handle. The wireless transceiver connected to the laptop will then send the data to MATLAB for processing and calculating the golf strike characteristics like location of strike and magnitude of the strike force. The enclosure acts as a neat assembly of electronic components so the user can just plug it into their laptop and get results without having to deal with wires getting tangled and making sure everything is properly connected before usage.  

Driver Head Initial Design

The driver head design, shown in Figure 20, consisted of a 3D printed driver head with the acrylic driver face containing the sensing strategy secured to it. Inside the driver head was the electronic component encasement embedded mold compartment, shown in Figure 21. The key components of the DAQ system were encased in this compartment. The design brought most of the mass closer to the bottom and rear of the driver and ensured that the center of gravity remained in line with the desirable center of gravity in today's drivers. The compartment was also removable, allowing for a less confined space for potting since it could be done externally from the driver head.

Driver Handle Design

The driver handle, shown in Figures 18 and 19, was designed to house various electronic components and to create a more complete product. The handle was connected to the driver head with a hollow TaylorMade Fujikura Speeder 65 Regular Flex shaft. The wires connecting components from the driver head to the handle ran inside the shaft. The design decision to move some electronic components to the handle freed space inside the driver head for the key components of the DAQ system. Placing the battery in the handle as opposed to the driver head addressed a safety concern regarding battery explosion from high impact experienced by the driver head. 

Sensor Layout Adhered to Acrylic Plate with Epoxy

Figures 16 and 17 show the sensor layout encased in and adhered to the acrylic golf face with insulative epoxy adhesive. A foreseeable issue was the epoxy getting in the holes that were going to be used to secure the acrylic face to the driver head. This happened with one of the four holes. The team solved the issue by drilling out the epoxy in that specific hole.

Sensing Strategy Layout Final Design

The final design for the sensor layout, shown in Figures 14 and 15, was adapted from the initial design shown in Figures 10 and 11. The initial design incorporated a force sensor that encompassed the entire area of the acrylic plate. However, the fabrication of this shape was difficult to achieve especially cutting out the holes. Additionally, the use of metal fasteners in these holes posed a risk of shorting the exposed edges of the PVDF sensor near the holes. For these reasons, the force sensor design was adapted to account for better fabrication and adhesion. Lastly, the overhang of the PVDF strips on the non-connection side was trimmed off to prevent loose ends on the inside of the driver head.

4 kSa/s Sampling Rate Test

The purpose of this test was to prove whether 4 kSa/s would be an acceptable sampling rate capable of identifying the voltage peak associated with golf ball impact on the club face. As seen in Figure 12, the PVDF sensor was adhered to a club face using painter's tape and extra long wires attached along the shaft connected to the DAQ externally. Reuben swung the golf club over a patch of artificial grass, striking the golf ball which was sitting on a tee, sending the golf ball soaring into a safety net. The LabVIEW output auto adjusted the amplitude as it was changed as seen in Figures 13a-13c. Before the swing was initiated, there was noise with an amplitude of 0.004 V seen in Figure 13a. During the swing but before golf ball impact, there was a small spike in amplitude of 0.09 V due to the club rubbing along the artificial grass, shown in Figure 13b. The impact strike caused an amplitude strike of 1.3 V which can be seen in Figure 13c. Since the impact peak was clearly identifiable, it was proven that 4 kSa/s would be an acceptable sampling rate based on initial analysis, but a higher sampling rate would allow for a finer reconstruction of the signal.

Sensing Strategy Layout Initial Design

The initial design for the sensor layout on the acrylic golf face consisted of the large PVDF sensor for impact force quantification and two layers of the PVDF sensor strips for impact location detection as shown in Figures 10 and 11. The first layer of nine PVDF strips was in the vertical direction while the second layer of six PVDF strips was in the horizontal direction. The sensor layout was encased and adhered to the acrylic golf face with LOCTITE EA E-20HP epoxy adhesive. 

PVDF Sensor Fabrication

The piezo film sheet purchased from Measurement Specialties Inc. was 110 µm thick, 280 mm long, and 203 mm wide. The sheet was cut into 3 mm wide strips using a Silhouette CAMEO 3 Cutting Machine. Six strips were cut to a length of 130 mm, and nine strips were cut to a length of 70 mm. The larger force sensor was cut to a width of 65 mm and a length of 130 mm. This sensor was then cut by hand to match the shape of the laser cut acrylic golf face. 24 American Wire Gauge multistrand wire was soldered to conductive adhesive copper tape. The copper tape was applied to an insulative layer of polyimide tape on either side of the PVDF sensor to prevent a short between the two sides of the PVDF material. 

A connection between the copper tape and PVDF was formed by applying Ted Pella fast drying silver paste. The connection point was reinforced with LOCTITE EA E-00CL insulative epoxy adhesive. A protective layer of polyimide tape was also applied to each side of the remaining exposed PVDF to prevent scratches and sensor damage from repeated handling and testing. The polyimide tape also provided an electronically insulative layer between the PVDF sensor and the substrate they were adhered to in order to ensure that the PVDF sensor was not shorted by nearby components or the body of the golf club. The layers involved in the PVDF fabrication process and the completed sensor strip can be seen in Figures 8 and 9.

Data Transmission and Reception

This test aimed to verify the compatibility between the NRF24L01 wireless transceiver and the Teensy 4.0 microcontroller for transmitting data, the NRF24L01 wireless transceiver and the Arduino Mega for receiving data, and the Arduina Mega and Matlab for analyzing data, all three of which were accomplished successfully. The components involved in this test are shown in Figures 6 and 7.

2x2 Sensor Layout Proof of Concept Test

To attempt to improve resolution from the previous 1x1 sensor layout test, a similar test using a 2x2 sensor layout was done. In theory, at the four sensor intersection coordinates from Figure 5 also indicated in red in Figure 3, the output of the two intersecting sensors should have the highest voltage outputs of the four sensors. However, the results from these tests conducted over three days did not confirm our expectations. Sensors A and B consistently read higher than sensors C and D regardless of the impact location. These results emphasized that no two sensors are the same, not only due to material imperfections but also due to the delicate by hand fabrication process. The steps planned to resolve this were to improve geometric similarity by using an automated cutter, to reduce damping and provide sensor protection with polyimide tape, and to reinforce sensor connections with insulative epoxy adhesive to prevent degradation of the connection point. Additionally, this test stressed the importance of calibrating each individual sensor to obtain a sensitivity constant for each so that all obtained data can be normalized and more accurately compared.

1x1 Sensor Layout Risk Reduction Test

A main part of the team's Risk Reduction efforts involved the initial fabrication and testing of PVDF sensor strips. As shown in Figure 1 and 2, the two sensor strips were arranged in a layered, overlapping pattern. A golf ball was dropped from roughly 1 ft above each testing location, labeled in the figures. The hypothesis was that both sensors would output a high voltage when impacted at their intersection, and as the impact location moved further from the sensor intersection, the voltage readings would decrease relative to the sensor intersection point. The hypothesis was confirmed through repeated testing using the procedure described above over multiple days. At location 2, both sensors would output a high voltage. As the impact moved further from the center along the horizontal sensor, either at location 1 or location 3, the output from the horizontal sensor would remain high while the output from the vertical sensor decreased relative to its initial reading. The limitation this test proved was the difficulty in perceiving whether the impact moved left or right of the sensor since the readings were similar for both instances. The intent moving forward was to increase the number of sensors to improve resolution and to also provide more data for more accurate impact location detection.