Ronin: early airflow, working display controls, and a harness to check

October 5 follow-up: the native menu and saved profiles are now implemented. Windows app launches have been checked; the launch-or-focus correction still awaits physical acceptance. The text below records the earlier increment.

Ronin is still changing rapidly. If you want to build one now, contact me first so we can check the CAD revision, parts list and assembly steps together. The enclosure has not been fabricated yet.

This update adds an early airflow study and the first physically tested firmware/companion controls. It also records what is known and what still needs checking about wiring the two Rainbow on a Matchstick boards.

Airflow worth investigating

The complete CFD Revision C report and video are now available. The Ronin project page presents the Voron commercial first, the calm-audio cinematic commercial second and the full CFD animation third.

CFD Revision C considers the solid outer caps and vertical channels of the PCB heatsinks, the route behind the backbone to the rear VRAM heatsink, the exposed APU cooler opening, and the top and rear cover vents. The current calculation shows net outward flow at both cover-vent groups, with some small inward patches.

The CFD is rough and needs updates to the model before it can get closer to accurate. Fin dimensions include estimates. The current mesh does not resolve individual fins, their roughly 2 mm channels, the smallest vents or the narrow rear passage well enough. The solve has not met its convergence targets. The optional P12 Slim is not modeled, and the final wiring and filter arrangement still need adding.

The diagrams help review air routing. Their annotated numbers are not validated case airflow, and the colors are not temperatures. The APU analysis samples a projected external section rather than isolating cooler flow. The study's animation replays a frozen flow field. It does not predict chip temperatures or establish a safe power setting or overclock.

More accurate geometry, a finer mesh and physical load testing in the assembled enclosure come next.

The companion can now control the display

The first paired firmware and companion increment is installed on the physical ESP32-P4 panel. Tested from my Windows laptop, it sends CPU, memory and disk readings and offers four working controls: Sleep display, Wake display, Mute sound and Unmute sound. Windows CPU temperature remains available through LibreHardwareMonitor.

The app asks the panel which controls it supports and waits for command acknowledgements. Sleep changes the display scene; it does not suspend the computer. Mute affects scene audio, not a voice-companion session. Existing SD animations and audio remain installed.

Run the app on the computer you want to monitor, and close the old resource helper before connecting: only one program can own the serial port. The new companion is a local prototype without a finished installer or startup service. Shared Python/Qt source includes Linux support, but hardware testing on Linux will wait for the BC250 to be reassembled.

LED-scene, touch-profile and agent-setting editors currently save local configuration only. Physical RGB control, host wake, touch-triggered scripts and the voice/agent portal are still work ahead.

Before making the Matchstick cable

The two ten-LED Rainbow on a Matchstick boards by Blamm are in the CAD, but are not wired to the display. The intended data path is display to board one DIN, then board one DOUT to board two DIN, with the appropriate 5 V supply and common ground.

The published V1.1 design uses three-contact JST XH footprints at 2.50 mm pitch. Input and output footprints face opposite ways. A cable made by matching cavity numbers alone could connect the wrong nets. The actual fitted board revisions, mating views and continuity need checking first.

The display data pin and mating connectors are still candidates. The actual LED data-voltage requirements and power budget also need confirming; the presence of a 5 V display header does not establish enough supply capacity for both boards. Any external LED supply must avoid backfeeding the USB-powered display.

A verified wiring table and low-brightness tests will turn this review into a usable harness procedure. Testing starts with one board before adding the second. Cable lengths, mounting and strain relief remain part of that work.

The project case study now reflects these changes and retains the films and original stills. The downloadable Rev0.3 manual remains the current draft. Companion usage and verified harness instructions are queued for a later revision, once they are ready to build from.