Close
0%
0%

OE5XRX — Open-Source Modular Remote Ham Station

A modular, repairable remote station for amateur radio — open hardware, built almost entirely AI-assisted.

Similar projects worth following
0 followers
Modular, open-source remote stations for amateur radio — built to be easy to deploy, maintain, and repair in the field. Almost everything built AI-assisted, documented honestly.

What it is

OE5XRX is a young amateur radio club in Upper Austria building **modular remote stations** — full radio sites you operate over the network — and deploying them at good locations across Austria. The priority isn't spec-maxing. It's the unglamorous stuff that decides whether a remote site survives in the field: **easy to build, easy to maintain, easy to repair.**

Why modular

The best antenna locations are usually the worst to drive to. When something fails at a remote site, "fix it" can mean hours on the road — so a fault should mean **swapping one board, not recovering a monolith**, ideally done by any club member with no special gear. Everything is open source, so other clubs can replicate a proven site instead of reinventing it.

Architecture

Each function is its own board, validated on the bench before it goes into a station:

  • Bus board — the backbone that ties the modules together
  • Power board — supply and power management
  • CM4 carrier — Raspberry Pi CM4 as the compute and control core
  • FM transceiver — the radio front end
  • Device tester — a bring-up jig to validate each module on its own

Built AI-assisted — the honest part

What makes this personal for me: almost all of the software, infrastructure and tooling was built AI-assisted, in a tight loop with Claude Code — station control, deployment, even the docs site. The honest exception is the hardware: KiCad at this scale is still too much for that workflow, so the board design is done by hand. I document that boundary as I go — where AI genuinely carries real embedded work, and where it doesn't (yet). No hype, just the actual line.

Status

Early days, and that's the fun part. Modules are in development and bring-up, the documentation site is live (oe5xrx.org, in German), and the project has been featured in Austria's QSP magazine. We're still shaping the station concept, so ideas and input are genuinely welcome.

Get involved

Open source — contributions and questions welcome. If you're in Austria and want to help build or host a site, the club is open for members (details on oe5xrx.org). Follow along here to watch the modules go through bring-up one by one.

  • First status log: four modules verified at v1.0, FM transceiver off to fab

    Peter Bucheggeran hour ago 0 comments

    Quick first log to set the baseline for where the hardware stands.

    Four of the five modules are verified and now have clean v1.0 releases:

    Bus board — verified, v1.0
    Power board — verified, v1.0
    CM4 carrier — verified, v1.0
    Device tester — verified, v1.0

    The fifth, the FM transceiver, is the one still in motion — and the one with the best story. v0.2 (STM32F302) surfaced a few issues on the bench that warranted a respin:

    USB D+ pull-up tied straight to VCC. The external D+ pull-up was first forgotten, then hardwired to VCC — so D+ sits permanently high and the device can’t signal a disconnect. That kills DFU updates, which require a forced re-enumeration by the host: the pull-up has to be controllable (via GPIO, or the MCU’s internal software-controlled pull-up) to trigger a soft-disconnect. Hardwired, it can’t.
    USB composite device trouble under Zephyr.
    A few GPIOs misrouted to the FM module.

    v1.0 moves to an STM32U575 and addresses these. It just went out to JLCPCB for fabrication, with the missing parts on a Mouser order; once it’s back and populated, it goes through the same bench validation as the other four.

    On the tooling side: versioning and releases across the module repos are now handled by a release-automation script — part of the AI-assisted workflow that runs most of this project’s software and infra. The hardware itself is still hand-designed in KiCad.

    Next: FM-TRX back from fab → assemble → populate → validate. I’ll log how the respin holds up.

View project log

Enjoy this project?

Share

Discussions

Does this project spark your interest?

Become a member to follow this project and never miss any updates