Author: mgavarre

        This past week, I focused on finalizing the capacitive touch design and getting a reliable system in place. Through testing, I realized that the original grid approach was fundamentally flawed due to how the MPR121 sensor operates. Each time the sensor initializes, it sets a default baseline capacitance, which is significantly influenced by any materials in contact with the copper foil tape. Because of this, stacking multiple layers caused major inconsistencies in readings, especially on the bottom layer.
To address this, I shifted the design by offsetting the row and column layers and introduced rubber bands for the top layer. The idea was that the flexibility of the rubber bands would allow touch pressure to bridge the gap between rows and columns without keeping them in constant contact. However, the rubber bands proved to be too insulating for a finger to trigger a reliable capacitance change.
To solve this, I wrapped each of the 81 intersection points with small patches of copper foil tape. This modification made a significant difference and resolved the detection issue, enabling consistent touch registration.
At this stage, the system only needs some threshold tuning and full integration. My upcoming goals include finalizing the enclosure to house all components cleanly and completing integration with the software side. While there is still work to be done to ensure the system is fully functional, accurate, and cohesive, I’m confident that we are on track to finish before the final demo.

This week, I focused on developing a fully functional 9×9 capacitive touch grid. I laser-cut and assembled a new version of the grid, incorporating several design improvements for better usability and stability. While maintaining the 9-inch by 9-inch active area, I added extra border space to separate the wiring and relieve pressure on the connections, which had previously caused reliability issues. The extended edge also allows for clamping, making the overall structure more rigid.
For the top layer, I experimented with 1/16-inch acrylic instead of cardstock to take advantage of its rigidity. Although it initially showed some promise, the material ultimately required excessive pressure to register touches, resulting in inconsistent readings. I plan to revert to a thinner, more responsive material and iterate from there.
There was a minor setback this week: one of the MPR121 ICs was shorted during testing. I’ve ordered replacements and will rebuild the grid once they arrive, continuing to refine the material and construction for optimal performance.

This past week was primarily focused on preparing for and presenting at the interim demos. I made a few last-minute adjustments and ensured the capacitive touch grids were functioning as intended. Due to Carnival Week, I wasn’t able to make significant visual or hands-on progress, but I did spend time planning the next steps.

Specifically, I began evaluating materials for the next iteration of the 9×9 grid and started designing how all the components of the project will come together in a cohesive system. I also ordered acrylic sheets for laser cutting to begin prototyping an enclosure. Some initial design work has been completed for the enclosure, though it will require further iteration.

Goals for this upcoming week:

• Finalize and test a properly functioning 9×9 capacitive touch grid with consistent and accurate touches.

• Continue iterating on the overall cohesive design of the project.

• Review and possibly incorporate feedback received during the interim demo into the next design phase.

This week, I focused on building the 9×9 grid, which consists of a 9-inch by 9-inch piece of wood layered with copper foil tape and cardstock. I was able to connect both MPR121 and get them to read values accurately when separated. However, I didn’t initially account for the warping of the cardstock, which resulted in a non-flush integration of the materials. This caused inconsistent sensitivity and readings on the lower layer.

To address this for the interim demo, I decided to keep both grids as separate entities. In response to the warping issue, I also designed a potential enclosure to house the components and ordered acrylic sheets, as acrylic is more resistant to warping than wood. Additionally, I am considering 3D printing clamps to better secure and flatten the layered materials.

With regards to code, I modularized it a little bit more and stored the rows and cols being read as variables that can be used for other parts of the code. I also made it so although I am using interrupts, in order to not oversaturate the number of inputs, it checks this interrupt every couple of seconds (adjustable) leading to a lot cleaner results.

I’ve made significant progress with the 3×3 grid, and the Raspberry Pi is now successfully reading both the rows and columns. While I am able to receive readings and see both inputs, I’m not entirely satisfied with the consistency and accuracy of the touchpad at this stage. Occasionally, the sensor may register an incorrect value due to noise or other factors. For example, when touching row 3, it could mistakenly register rows 0, 1, or 2 before eventually defaulting to row 3 after a brief delay.

In addition to working on the core functionality of the touchpad, I’ve also shifted from polling to using interrupts to capture input. This change has brought some noticeable improvements. Polling involves continuously checking for changes in the inputs, which can be inefficient and lead to slower responses, especially when there are multiple inputs being processed at once. By using interrupts, the system can react immediately when a change occurs, rather than having to constantly check. This approach reduces processing overhead and ensures that inputs are captured more quickly, leading to a more responsive system.

However, there are some trade-offs with using interrupts. While interrupts improve responsiveness, they also require careful handling to avoid issues like missing interrupts or overloading the system with too many triggers. Additionally, using interrupts can sometimes make the system more complex to manage, as it introduces additional code to handle interrupt requests and ensure that each one is processed correctly. Despite these challenges, the benefits of faster, more efficient input handling outweigh the downsides, and I’m confident that this approach will lead to a better overall user experience.

Moving forward, my next step will be to focus on fine-tuning the system and improving the accuracy as much as possible. This will involve further adjustments to the sensor calibration and possibly adding noise-reduction techniques to ensure the readings are more reliable. Once that’s addressed, I’ll shift my attention to incorporating these touch inputs into the Sudoku game system. This will involve using the touchpad to select points on the grid and integrating the keypad for grid selection and number input. By combining the touchpad and keypad, I aim to create a seamless and intuitive interface where users can easily navigate the grid and enter values to solve the puzzle.

There were a lot of set backs with regards to making progress on my part this week. Along with interviews that I needed to study for, I was having some issues getting the raspberry pi setup to begin testing the capacitive touch grid. Since I’ve never worked with these microcontrollers, I didn’t take into account extra components we may have needed to get it functioning correctly. After a couple of days I acquired the components and was able to get the raspberry pi working and begin looking at testing the small 3×3 capacitive grid I made. I was able to detect touches for the rows but not for the columns. This can most likely be attributed to how I connected the wires for the columns and may require me to remake the grid to get proper testing completed.  So moving forward, I will remake a grid with more accurate alignment along with proper wiring for the columns and attempt to get intersection handling completed as soon as possible so I can begin working on creating the larger 9×9 grid.