Wall-mounted smart thermostat with OLED behind tinted acrylic

Nov 02, 2026 Last reply: 2 days ago 3 Replies
Jump to replies (3)

The detail that caught me off guard: the laser-cut acrylic faceplate was embedded partway through printing the enclosure, so the plastic forms around it in one go. That is a tidy trick I had not come across on a thermostat build before, and it explains how the screen and the housing end up looking so seamless together.



The path to the finished unit is a proper progression. Designing Good Ideas started on breadboards, moved to perfboards, and only then laid out custom PCBs in KiCad, with board outlines drawn in Onshape and brought in as DXFs. The boards from PCBWay were hand-soldered, and the housings came off a Prusa Core One+. An early prototype lived in a cardboard box with a couple of breadboards, which makes the final black enclosure with gray knobs feel well earned.



Several choices stand out. The OLED sits behind tinted acrylic so only the lit pixels show through, with no backlight glow washing over the panel. The PhotoMOS relays can be driven directly from the ESP32 and handle 24V AC, which seems like a neat fit for HVAC control. A dedicated power circuit converts the 24V AC from the HVAC system down to 5V DC, and the measured output came in at 4.9V. Firmware is C++ with the Matter library, and a QR code generator handles Matter pairing.



One open thread: the maker plans unspecified improvements before releasing everything as open source, and will look into whether PhotoMOS relays have unusual failure modes. Also worth noting, encoder readings jumped until interrupts via digitalPinToInterrupt fixed the sampling.



Project by Designing Good Ideas. Watch it on their YouTube channel:

formatting link


Going with PhotoMOS relays looks like the smartest call in the whole build to me. Being able to drive them straight from the ESP32 pins keeps the control side simple, and they handle the 24V AC side fine. The tradeoff is that solid state switching behaves differently from mechanical relays, which is probably why the failure mode question is still open.

How did the mid-print acrylic embedding actually work in practice? Was the print paused at a known layer height so the faceplate could be dropped in, or does the design have a recess that the acrylic slots into while the nozzle keeps going? I keep picturing the alignment problem and would love to know how it was handled.

Since the maker already mentioned adding sanity checks, I would suggest folding them in before the open source release rather than after. A periodic check that the temperature sensor is still answering and that commanded modes match reported state would catch a failing thermometer early. It would also make the first public version more trustworthy for anyone wiring it to real HVAC equipment.

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required