2026‐08

August 4, 2026

Caribou Call — Recording · HearHear transcript

Attendees: far1no, errrks.eth, thomasg, kbmollysuh, sleety.eth

• PT1 Fully Assembled — Ground Testing Went Well

  • Caribou is fully assembled and last week's testing went well with no major problems. It is powerful and notably quiet — a small DJI hovering above it was louder; the props run at a deeper, less intrusive tone.
  • Motor RPM stayed in the same range despite the batteries sitting at different charge levels.
  • Two small issues: the Cube node's Ethernet adapter intermittently crashes (suspected cause: mounting too close to the antenna), and battery 6's telemetry displayed zero in Mission Planner's data window without any unhealthy-battery notification — logs to be checked to see whether it's a display issue.
  • One small CBC PCB fix is wanted before anyone else orders boards; issue still to be filed.

• First Flight Moves Into Phase 2 — Done Legally

  • Julius will file for a German authority approval allowing testing at the same location for up to two years. Stated processing time is a few weeks, faster where danger is low — likely applicable (private, secured, sparsely populated field).
  • Motivation stated plainly: the repo has attracted too many eyes (~75) to fly informally. Thomas agreed — more organized than the Feather PT1 days.
  • Pre-flight tests that need no approval continue meanwhile: folding motor-arm connectors attached this week, another ground test with them fitted, and a battery rundown test from 100% to cutoff.

• Phase 1 → Phase 2

  • The July progress report is published. The Phase 2 document posts shortly for feedback, followed by a new Snapshot proposal with adjusted phases. Thomas remains eager to build a Texas prototype in Phase 2.
  • A light website-docs status pass happens at Phase 1 close (full documentation remains later-phase scope).
  • Remaining Phase 1 funds stay in the Caribou multisig until the Phase 2 Snapshot, usable for wrap-up and documentation.
  • The design goes into the repo: Fusion → STEP export (Thomas offered AI-assisted help converting toward build123d). Signal cables are absent from the CAD; acceptable so long as the structure is there.
  • Issues cleanup: most stale or finished issues closed.

• PEEK Machined Parts (KBM)

  • Proposal: CNC-machined PEEK (printed PEEK considered unreliable; PEEK stock is cheap) starting with the folding motor-beam mechanism, later possibly landing gear and beam joints — and eventually Quiver's center beam, top plate or attachment interface.
  • KBM's recalled figures (unverified, from a supplier exhibition and test data): CF-filled PEEK approaches aviation aluminum alloy strength; pure PEEK is roughly 30–35% of aluminum's strength at about 20% of the weight. Advantage is weight; trade-off is larger part sizes unless fiber-filled.
  • Agreed path: Julius models the purchased folding beam connector → KBM adjusts → KBM gets CNC quotes (China-first on cost) → compare against the bought aluminum part and metal 3D printing.
  • IP note: the model is measured from a purchased Chinese part. Practical enforcement in Chinese UAV parts is near-nil, and Thomas sees a well-made machined adapter as a potential Arrow-shop product — the IP question deserves a proper look before that step.

• Open Items

  • Julius: post the Phase 2 document for feedback, then the Snapshot proposal.
  • Julius: file the German site-approval application.
  • Julius: check battery-6 telemetry logs; investigate the Ethernet-adapter crashes.
  • Julius: folding motor-arm connectors on, ground test with them, battery rundown test.
  • Julius: file the CBC PCB fix as an issue; export STEP to the repo; light website-docs status update.
  • Julius: model the folding beam connector for KBM.
  • KBM: adjust the model, source PEEK CNC quotes, compare costs.

• Open Questions / Risks

  • The approval timeline (weeks) now gates the first flight; no fallback was discussed.
  • Battery-6 telemetry and the Ethernet crashes are open reliability items going into approval.
  • PEEK strength figures need verification against datasheets before designing to them; the copy-part IP question needs a real answer before any shop product.