Lumo / Elektra — CAN bus HAT

Context

At TeroLabs we kept ending up in projects that needed to read data off a vehicle. The OBD-II dongles on the market were either too expensive, too proprietary, or both. The simpler answer was a HAT — a board that sits on top of a Raspberry Pi, talks to the car’s CAN bus through the standard OBD-II port, and hands the decoded data to whatever stack the project needed next.

Internally it went by two names — Lumo and Elektra — depending on whichever branding the studio was running with at the time. Same board.

What I did

• Schematic and PCB in Eagle CAD. CAN transceiver, the right protection on the OBD-II side because cars are electrically noisy and occasionally unkind to anything you plug into them, the Pi-HAT pinout, and an EEPROM so the Pi could identify the board the way the HAT spec wants it to.
• Enclosure in Fusion 360. Snap-fit, prints on a hobby printer, big enough for the Pi underneath, small enough that it sits flat under a dashboard without becoming a problem during a long drive.

Where it ended up

Lumo became the sensor layer for a driver-behaviour analysis pipeline: read CAN frames, decode the parameters we cared about (speed, throttle, brake, RPM, steering inputs where available), label them, and feed the stream into the analytics that came after. The interesting work downstream wasn’t mine, but every datapoint that fed it came through this board.

It also got used as plain telemetry hardware for a couple of internal TeroLabs projects that needed vehicle data and didn’t have anything better to use.

Status

The design files are public in two places. The official TeroLabs repo holds the canonical version, though I no longer work there and can’t guarantee it stays online — if you’re reading this and the link is dead, my personal mirror should still be up. If you’re trying to do something similar and want the design files, get in touch.