Building the enclosure.

I took what I built in November and pushed it further by shrinking everything down to one plant. I built a fully sealed 6x6x9 inch enclosure around a Neofinetia falcata, chosen partly because I like the plant and partly because I have another one. Knowing there was a backup made it easier to experiment without being overly cautious.

A lot of time went into the physical side of the system. I learned more about PCBs, soldering, and wiring, all of which were new to me for the most part. I spent days just figuring out airflow, how to move air gently, how to exhaust it without leaks, and how not to overwhelm the environment I was trying to control.

I also spent time wiring and testing sensors, learning how they behave in practice instead of on spec sheets. Temperature, humidity, and light data were logged locally on a Jetson Orin Nano and mirrored to Google Sheets so I could check in on the system without babysitting it. Once things were stable, I ran simple block tests to see how fans and actuators actually changed the enclosure, which helped turn a lot of assumptions into something measurable.

By the end of the month, the enclosure worked and the data looked clean, but the biggest takeaway was not technical. Spending so much time fine tuning an environment made it clear that this problem has already been solved in many ways. More importantly, the plant itself was still one step removed from the system.

Moving forward, I want to focus on what data points I can extract directly from the plant. Instead of inferring plant health through environmental proxies, I am interested in whether bioelectrical signals can be translated into useful information and used as a more direct input for understanding and responding to a plant’s needs.