I made a graphic display for my dorm during MIT's Residence Exploration (REX) period during summer 2023! We built a three story fort and I worked on the lighting for it. Here's a preview what the final project looked like:
 |
| The letters hanging from the 3rd floor of the fort |
It took several months of planning and building to get here, starting with a rough, back-of-the-napkin type sketch in April (below). I planned out the approximate arc for the letters and calculated an average viewing distance to get a sense of how big they should be to still be readable. I found a size range from 11-14" to be reasonable, and I ended up making them 11.5" in diameter so they could be cut out of standard 12"x24" sheets of acrylic.
At this point I also chose LED type and color. For the letters I went with solid-color silicone covered strips (red, orange, and yellow) to look like a neon sign and the lights around the edges of each circle were controllable RGBW LEDs. Lastly, I planned out the electrical connections on each letter to avoid splicing together LED strips as much as possible. As neat as the silicone covered strips look, they are a bit annoying to scrape the plastic off and solder together.
 |
| Initial planning sketch |
The next step was to design the control boards and send them off to be printed. I used KiCAD and chose parts I knew I had (like using ethernet cables for power/data between panels because MIT vending machines dispense two free ethernet cables a week).
 |
Cannonball control board schematic
 | Cannonball control board PCB layout
|
|
After waiting a few weeks for the boards to come in, I started on the assembly. I used a hot plate to solder on surface mount components for the first time and was pleasantly surprised how nice and fast it went compare to soldering things individually. However, I still had to solder in all 330 pins from the ethernet ports and screw terminals for the LED strips by hand. I made 15 of these boards to have some backups (and ended up using two of those), although the project only needed 11. I covered the electrical connection with hot glue and used clear tape to connect the LED strips and acrylic panels. In total, soldering and assembly took approximately 30 hours over the course of a few very long days.
 |
| Soldering surface mount components |
 |
| Splicing the silicone covered LED strips |
With everything built, it was time to test. The most common problem I ran into was with the older RGWB strips, as they were crowdsourced instead of bought new. For about half of them I had to splice in sections to replace broken LEDs. The other problem I found was a poor pull-up resistor connection on one of the boards caused the strip to not display green, but that was also quickly fixed with resoldering. The last fix I made was adding a bypass capacitor on the back of each board between power and ground on the ethernet connection. I discovered when chaining together all 11 boards the current surges would cause the boards to malfunction, but adding the capacitor smoothed them out nicely.
 |
| Adding a bypass capacitor on the back. It's not pretty but it worked :D |
 |
| First test of the lightning bolt |
After assembly I moved on to programming the letters. I used project WLED and an ESP32 to program some set colors for the LEDs. I used several of the WLED open-source presets (Chase, Rainbow, something called "Fairytale glitter"), and also wrote some of my own. The fort had a central router/Wifi network to which we connected all of the lights. For a while I tried programming the ESP32 directly in C++ to send a request to the cannonball IP address and cycle between presets at the touch of a button, but I found the response was too slow. In the end, I just connected my phone to the Wifi network and went to the cannonball lights IP address manually to change the display.
 |
| Programming the letters |
The final step in the process was to rig these up onto the third floor of our fort. I got four friends and two 20ft ladders and managed to get them hanging in an approximate (a little wonky looking) arc. In hindsight, I really should have just tried to hang them on one wire instead of trying to make an arc with two because things got messy and tangled quite quickly. There are a few other things along the way during this project that I would do differently in the future, like adding in a bypass capacitor into the design and getting more people to help me with assembly, but overall I enjoyed working on it and I'm happy with the final result!
Here's a video of the project:
