What does it do?
It monitors your plant’s soil moisture and temperature, logs to SD, hosts a local web dashboard, and lets you tap into your plant’s inner workings. Because yes, your plant IS judging you.
Who’s done what beforehand?
Plenty of plant‑monitors and IoT gadgets exist. But few (none?) combine custom PCB, NFC handshake, cardboard packaging, local web UI, and an AI‑human design loop. So: unique and willfully irreverent.
What sources did you use?
Datasheets, forum threads, the Fab Lab Puebla labs, your own trial‑and‑error sessions, and constant dialogue with ChatGPT‑4 (yes, the AI sidekick). The FabAcademy handbook too, obviously.
What did you design?
- Custom PCB layout (KiCad – ended breadboard betrayal)
- 3D‑printed enclosure & laser‑cut cardboard box
- Local web UI (HTML/CSS/JS) served from SD card
- NFC interaction flow (“handshake” your phone)
- Documentation process that documents the documenting—because meta.
What materials and components were used?
ESP32‑S3 XIAO board, capacitive soil moisture sensor, DS18B20 temperature sensor, microSD card + adapter, NTAG215 NFC tag, PLA filament for print, cardboard sheet, various connectors and headers.
Where did they come from?
Online suppliers (AliExpress, Digikey), FabLab inventory, local makers’ supply shops, your own “junk drawer” for random headers. The cardboard? Straight from the laser‑cut stack at the Fab Lab.
How much did they cost?
Rough estimate: ~$25‑30 USD in components. Time, debugging, and coffee consumption? Priceless (and undocumented in the budget).
What parts and systems were made?
- PCB system (power, sensors, SD card, board)
- Enclosure system (3D print + packaging)
- Firmware & web interface system (logging, dashboard, NFC trigger)
Every part of the chain: from “soil” to “dashboard.”
What tools and processes were used?
Fusion 360 (CAD), KiCad (electronics), Creality K2 3D printer, Glowforge laser cutter, Roland SRM‑20 for PCB milling, Arduino IDE / VS Code for code, and endless nights of debugging.
What questions were answered?
- Can an embedded system host a full dashboard locally without cloud? – Yes.
- Can I design hardware and software in a human‑AI loop and actually finish? – Yes.
- Do plants appreciate sarcasm? – Probably not. But the process sure did.
What worked? What didn’t?
Worked: Custom PCB, SD card logging, local dashboard, NFC interaction.
Didn’t: ESP32‑C6 attempt (library hell), WiFi antenna missing (so limited final demo), BH1750 & other extra sensors dropped for reliability.
How was it evaluated?
The device boots, logs data, hosts a UI accessible via phone, and engages in the “plant‑stab‑tap” interaction flow. That’s good enough for FabAcademy — inputs, outputs, board, and makerspace process.
What are the implications?
You don’t need a cloud or huge budget to build useful environmental sensors. A human + AI design loop works. And yes, next version will include voice alerts and self‑watering. (Probably.)