Project Caribou — July 2026 Progress Report
1. Summary
Phase 1 ended with July. The aircraft is built and ground-tested. The first flight moves to an later date under a proper legal test setup.
Done in July:
- Assembled PT1 into a complete aircraft: All powertrains, batteries, avionics are working.
- Wrote and validated the CBC _PCB firmware (v0.1.0 → v0.3.4).
- Passed all powered ground tests, including a full motor test of all six drivetrains on July 31.
- Built a verified first-flight parameter set and a step-by-step flight checklist.
- Received the commissioned PT2 frame design study. Community review is running now.
- Secured a test site suitable for tethered flight operations.
Moved outside of the Phase 1 timeframe:
- The first hover flight. We prioritized a professional setup — secured area, tether anchor, SORA 2.5 application — over a quick hop.
- The Snapshot proposal for the next project phase. A separate forum post covers the Phase 1 review and the Phase 2 plan.
First weigh-in: PT1 is 182 kg dry, about 40 kg over target. Weight reduction is now the top PT2 design driver.
2. Team
July core contributors:
| Contributor | Role | July allocation |
|---|---|---|
| Julius | Project lead | 30 h/week |
| KBM | CAD / PCB | 20 h/week |
| Alex | Avionics / Caribou Hub | 15 h/week |
| Erick | Electronics review / support | 10 h/week |
An external freelancer delivered the PT2 frame design study (see Budget). Vector (the Arrow assistant) wrote the CBC firmware and the flight-controller parameter set with Julius.
3. Progress
CBC bring-up and firmware
The battery-connector boards ordered in June arrived. We brought them up and wrote their firmware from scratch (PR #54). The firmware controls the power path: precharge, latch arming, and kill-trigger monitoring. It runs both CAN buses and both temperature sensors.
Battery telemetry was July’s biggest unlock:
- We captured live CAN traffic from two Tattu packs and reverse-engineered their protocol.
- The CBC now decodes the full 76-byte telemetry: voltage, current, temperature, state of charge, health, cycles, serial number, and all 18 cell voltages.
- Each CBC republishes its pack as a standard DroneCAN battery. ArduPilot sees six ordinary battery monitors — no custom flight-controller code.
- Packs identify themselves automatically via dynamic node allocation. We validated this on real hardware with multiple packs.
- Each board carries NODE_ID and BATT_ID parameters, editable over CAN from the DroneCAN GUI tool. Field configuration needs no USB cable.
Flight controller and system integration
- Created the standard Caribou parameter set (PR #55), forked from Quiver. Hexa frame, DroneCAN ESCs primary, PWM backup, 18S voltage configuration. We verified the motor-to-connector mapping against the CMAIN silkscreen.
- Finalized the CAN architecture: 500 kbps drone bus for the long harness runs, defined termination topology, single-point frame grounding with floating HV islands.
- Telemetry reuses Quiver’s ports: the SIYI HM30 and A8 camera connect identically, so the Quiver ground-station setup carries over.
Ground testing
Mid-July system tests proved the battery chain end to end: packs → CBC → DroneCAN → ArduPilot battery monitors.
July 31: full motor test, three runs. The longest ran 2 minutes sustained at 35% throttle. One run took all six motors to 45% together. Results:
- All six drivetrains matched within ~1% RPM. Per-motor power matched within a fraction of a percent.
- ESC temperatures stayed between 35 and 43 °C. No ESC errors.
- Zero CAN errors across the 20-minute session.
- The RTK GPS pair held a fixed moving-baseline heading: 32 satellites, GPS yaw active.
- We captured extensive photo and video material during the tests.
We then audited the complete dataflash log and all 1,178 parameters. The audit produced a ready-to-load first-flight parameter file:
- Thrust curve fitted from the X15 bench data.
- Battery failsafes unified and enabled on all six packs.
- LiHV voltage range corrected.
- Vibration filter reconfigured to track actual motor RPM.
- Conservative first-flight tuning values derived from geometry and inertia.
First-flight preparation
- Wrote a step-by-step checklist and mission plan following the ArduPilot Methodic Configurator process. The first flight runs four phases: spool check, light-on-gear, first hover, altitude hold with data collection. Each phase has go/no-go criteria.
- Secured a test site for tethered operation. We will file a SORA 2.5 application for Caribou at that site. Once granted, it allows official test flights for 12–24 months.
- Next hardware step: install the foldable motor-arm connectors on PT1. The rigid arms overhang the trailer for road transport. The install also gives early feedback on folding arms for PT2.
PT2 design work — ongoing
PT2 frame work moved into concept comparison. Three concepts are on the table:
- Nils’s commissioned design study: a modular, foldable steel frame (folds to 2.30 × 1.00 × 0.55 m, bare frame ~40 kg). Delivered and posted for review.
- KBM’s design suggestion (separate thread).
- Julius’s own concept: in progress, to be published separately.
The point of having three: they give the DAO concrete discussion material to converge on the PT2 frame direction. Review runs in the design threads (links in Sources).
The weigh-in sharpened the goal: PT1 is 182 kg dry vs. the ~142 kg target. At the motors’ continuous rating, every kilogram of structure costs one kilogram of payload.
4. Outlook
Phase 1 ended with July. Until the DAO approves the next phase, nothing below is official project agenda. Item 1 is the formal step; items 2–7 are recommendations and tasks I will pursue in the meantime.
- Create the Snapshot proposal for the next Caribou phase (see the separate Phase 2 discussion post).
- Install the foldable motor-arm connectors for road transport.
- File the SORA 2.5 application for the test site.
- Close the remaining pre-flight items: pack-voltage matching, joystick control-chain checks
- Fly PT1: first hover per the prepared checklist.
- Start the tuning campaign: verify the vibration filter from hover data, then raise gains stepwise.
- Continue PT2 studies: frame concept review and weight-reduction planning.
5. Budget
July contributor streams (July 1 – August 1, executed from the Caribou Safe on August 3 in a single batch):
| Contributor | Stream amount |
|---|---|
| Julius | 9,300.00 USDC |
| KBM | 3,968.00 USDC |
| Alex | 2,976.00 USDC |
| Erick | 2,645.00 USDC |
| Total streams | 18,889.00 USDC |
July reimbursements (executed from the Caribou Safe on August 3):
| Contributor | Item | Amount |
|---|---|---|
| Julius | Material (12 line items, see below) | 3,505.00 USDC |
| Julius | Freelancer: PT2 frame design study | 1,730.00 USDC |
| Total reimbursements | 5,235.00 USDC |
Largest material items: the CBC PCB assembly order at JLC ($1,273 — carried over from June as planned), sheet-metal battery holders ($701), and shipping/customs for PCB and component orders ($967). The rest: position LEDs and thermal pads ($160), PETG-CF filament ($145), workshop consumables ($125), CBC heatsinks ($81), threaded inserts ($53).
The Safe received 24,312.74 USDC from the treasury on August 1, replenishing the June allocation. After the July payouts the balance stands at 5,876.33 USDC. It reconciles exactly with the transfer history.
Total July allocation: 24,124.00 USDC (streams + reimbursements).
6. Key Sources
- Repository: GitHub - Arrow-air/project-caribou: Project Caribou documentation, designs, and project artifacts. · GitHub
- Issues: Issues · Arrow-air/project-caribou · GitHub
- Pull requests: Pull requests · Arrow-air/project-caribou · GitHub
- CBC firmware: PR #54 — flight-controller parameters: PR #55
- July meeting notes: 2026‐07 · Arrow-air/project-caribou Wiki · GitHub
- PT2 frame concept thread: Discord
- PT2 design suggestion thread (KBM): Discord
- Caribou Safe: Safe{Wallet} – Transaction history
