Last updated on

door-opener 2

Avatar for Eric Park Eric Park

Longtime readers of our blog will know that one (or two?) of the earliest projects we shared in a previous blog post are passports and door-opener. If you’ve missed it, you should definitely go and check out how they work, but today we’ll be focusing on door-opener.

We spent the past year designing, building, and deploying the new door-opener that would solve some of the problems that started to manifest with the original, while adding new quality-of-life improvements for the organizers setting up Hack Night.

I shared parts of how door-opener-v2 came together during my BENTO talk, but because the goal of BENTO was to condense everything down to within 15 minutes, a lot of material was left on the cutting room floor. So here’s a more detailed look at how door-opener-v2 came together, starting all the way back at the beginning.

Entropy, and the cruel passage of time

door-opener was initially built by Hazel Roeder, and her design had been working flawlessly, up until Fall 2025. Unfortunately, while the design itself would’ve lasted a lot longer, the materials for the suction cups, as well as some of the internal components like the fan bearings, were starting to wear out.

We decided to prioritize a more permanent solution for this problem when people reported that they couldn’t get to Hack Night anymore, because the modules kept losing their grip and falling off of the surfaces.

So on September 2025, a new channel on our Discord was created, and I’ll try my best to summarize the important milestones we hit up to the present.

Idea-sharing and trimming

We first had to decide what the second revision of the project would do better compared to the first one. Some of the ideas proposed were:

Most of the ideas did eventually make it to the final product, but we haven’t even gotten to the first prototype yet. For that, we needed to order some parts.

After looking through the potential ideas and features list, we decided that a new enclosure for door-opener would be needed after all, and a parts list that looked suspiciously similar to the first revision was ordered, with the addition of a camera. Since we were redesigning the enclosure anyway, we also decided to use a different part for the screen, which did complicate the design and build process quite a bit. (More on that later.)

Branching off a sub-project

While we worked on the new revision for door-opener, we also focused on the idea where the servo that sat on top of the ADA button would be made more independent compared to how it was before. Previously, an organizer had to run a microUSB cable between the servo module and door-opener, and depending on how it was plugged in, the servo would fail to be initialized and require a power cycle on the Pi’s side. Unfortunately, this wasn’t immediately apparent to the organizer setting things up, and because they had several other tasks to complete in order to get Hack Night ready, attendees would often show up to a door-opener that wouldn’t actually open the door.

So the spin-off project got the name ada-pusher, and we took the opportunity to swap out the servo with a linear actuator for better reliabilty. For the connection back to door-opener, we decided to use Bluetooth Low Energy, or BLE for short. Wi-Fi was briefly considered but was decided against because it might become a security risk for Bechtel, the building that we were hosting Hack Night at.

Updating the codebase

While waiting for the hardware parts to arrive, I also started work on updating the repository and the dependencies required. For example, the repository in the beginning did not compile successfully on Windows, until the NFC library was updated to properly work with CMake. Since the repository hadn’t been worked on since Hazel graduated, I focused mostly on making sure the dependencies were up-to-date.

Any visual changes would have to wait until the new prototype was built, but by compiling and transferring the updated build to the first revision of door-opener and making sure nothing had changed or broken, I could be reasonably sure that I was working on solid ground.

Prototype one

At the end of November, once all the parts had arrived from online marketplaces, I quickly connected together the loose parts on my desk and loaded up the same build from revision one. The visuals were messed up, as expected, but at least we had somewhere to start from!

First prototype booting and showing the opener app

And since I had all the parts sitting in front of me, I took out the calipers and started to take measurements. After sketching out a rough idea of what I wanted, I opened Autodesk Inventor for the first time and made a rudimentary enclosure that would hold the parts together like a test bench:

First prototype enclosure design

And that wrapped up Fall 2025, and the rest of the semester was quite busy. After winter break, things picked back up again during the Spring 2026 semester.

What do we name the thing that pushes the ADA button? Some sort of ADA pusher?

Several things happened in parallel as we raced to get this project complete. We finally had the proper linear actuator that had the specifications we needed, shipped all the way from China. It needed 12V of power to operate, and we had an ESP32 board to interface via BLE and send signals to the linear actuator.

At first, I thought that I would only need one USB-PD trigger board and one step-down converter, to change the voltage from 12V to 5V for the ESP32. The trigger board would negotiate 12V from the power source and cleanly power everything.

Unfortunately, if you are familiar with the USB consortium, they actually made the 12V Power Delivery profile optional, which meant a lot of chargers flat-out do not support it. The justification was lack of use: phones usually topped up at around 9V, while laptops and tablets usually went up to 15V and 20V. Thus, the 9V profile was out of the question, even though it would’ve been perfect for our use case.

We couldn’t rely on specific chargers supporting it, because there was a chance that the charger we were using broke and the replacement wouldn’t have the optional profile. And if the charger didn’t support it, rather than risk frying the downstream device, the negotiation automatically shifted down to the next lower voltage profile, which was either 9V or 5V.

Thus, the new plan was to use two step-down converters and bring the 15V voltage negotiated by the trigger board down, to 12V and 5V, respectively.

After waiting a couple more days for all the parts to arrive, the first prototype of ada-pusher also materialized on the dining table:

ada-pusher first prototype

Through this test, we confirmed that the linear actuator module worked, and that the ESP32 board could properly send the signals when prompted.

Then came the topic of mounting the ada-pusher. We came up with several design ideas, including utilizing the wall panel gaps just above the ada-pusher:

ADA pusher mount idea 1

ADA pusher mount idea 2

Ultimately, we decided not to move forward with these designs. For the first design, the plastic perimeter around the button was thin and brittle, and we were worried that the contraption hanging off of it could snap it. For the second one, the work to model it in 3D was honestly intimidating to me, as I didn’t have that much experience with CAD software.

So the lazy but surprisingly effective (?) solution: the original had two suction cups. What if we have four of them?

But before we could even start designing how that would look like, we first had to measure the distances and the dimensions of the actual ADA button. So I went on over to Bechtel with my calipers and spent half an hour measuring it.

Bechtel ADA button with dimensions sketched on above

Sorry, Bechtel, now the world knows how big your ADA buttons are

Since I had such a terrible time dealing with Autodesk licensing at this point, I switched on over to FreeCAD, which worked just as well, given that my project was simple in scale.

ADA pusher CAD design on FreeCAD

ADA pusher initial print fit on button

For the electronics part, Bechtel had an electronics lab with good soldering equipment. However, the circuit for ada-pusher required two step-downs and a big driver board for the linear actuator, and since we didn’t have time for a custom PCB, everything was hastily put together on a perfboard.

Bechtel's electronics lab

ADA pusher perfboard circuit

Meanwhile, on the software side, everything was written atop Rust. The ESP32 platform had good enough support for Rust, although you did have to install a custom toolchain for Xtensa-based boards, which was what I was using. But the nimble crate made BLE connections quite easy, and before long I had a working contraption sitting on my desk!

First working prototype of ada-pusher

On the door-opener side, the btleplug crate (yes, it really is named that) reached out to ada-pusher whenever someone tapped their valid passport. Once the integration tests worked, the entire caboodle was put in for real-world testing at Hack Night 6.18.

Unfortunately, after an hour or two, someone sent us a photo of the mount failing. Adding insult to injury, because I forgot to put in fillets, one of the legs cleanly sheared off:

The broken ada-pusher

Thankfully, adding the fillets was enough to give the next print structural integrity, and with that, the ada-pusher spinoff project was complete!

Back to the new door-opener

My focus went back to completing door-opener-v2, which also required its fair share of revisions to the CAD. Even though Purdue has plenty of 3D printers scattered around campus, trying to find one that isn’t in use was difficult. At some point, Bechtel’s printers were constantly in use, so I had to go over to the 3D Printing Club and use their printers to print and test the final few revisions.

Most of the revisions had to do with tolerances. 3D prints expand on a heated bed and shrink when cool down. 3D printers have different tolerances and prints differ from one another, even with the same Gcode. This made it difficult to tune in the dimensions exactly, such that all the parts fit together with the screws tightly holding them together.

door-opener-v2 print front

door-opener-v2 print back

One of the failed prints being used to test the overall fit of components

The screw hole of the door-opener-v2 not lining up with the display module

This happened over, and over, and over again. 3D printing is cursed

Meanwhile, the software also needed some patches. Besides integrating with ada-pusher, it had to be adapted to work with the new LCD module of door-opener-v2. The first revision had hardcoded values for the square LCD module, which meant that elements were either stretched out, squashed in, or even off-screen entirely on the new rectangular display.

door-opener app UI changes with the text overflowing

This wasn’t the actual message, but a test for more text overflowing the dialog box

But after the design revisions and the software patches, door-opener-v2 was finally completed in April and put into service at Hack Night 6.25. For context, we cut it so close that the Hack Night in question was one day before the BENTO talk.

The software saw a ton of improvements and new features, which were shown off at BENTO, but here are some of the major ones again:

After BENTO, one last modification was required when we realized that the Pi and power circuit was overheating. Thus, the back plate was modified for a small 40 mm fan to sit in:

door-opener-v2 backplate with fan cavity and holes

And on the software side, one feature that unfortunately didn’t make it in time for testing was the camera module. The idea was that organizers could take a photo through the camera to see who was in front of the door, then issue the door open command remotely. Most of the client-side code was implemented relatively quickly thanks to the nokhwa crate, but the semester ended before we could start working on the server-side and integration testing.

Retrospective and the future

So that was, in a way, two hardware projects in a trenchcoat. I’m quite happy with the results and the improvements made to the original revision, and the valuable lessons learned working on physical hardware.

With the documentation on the door-opener repository, anybody should be able to contribute and add new features to door-opener. That’s exactly how I got started with Purdue Hackers back in 2024, contributing patches to things that I found each Hack Night while having a great time with the community. Hopefully, you’ll see door-opener when you visit one of our Hack Nights and think of something fun, silly, or cool to add to it!

Also, many thanks to the other organizers and various community members in helping with the new revision of door-opener! I’m quite hopeful and excited to see where the project goes next semester.