2026‐09 - Arrow-air/project-caribou GitHub Wiki

September 22, 2026

Caribou Call — Recording · transcript from the Google recording (HearHear failed to join this call; only a short fragment was captured)

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

• PT 1.5 Layout (Julius)

  • Julius presented the PT 1.5 layout. It replaces PT1's welded steel inner frame with a sandwich layout: carbon fiber tubes (all 60 mm) and end adapters that he plans to 3D print in aluminum. The goal is fast assembly. PT1's frame took a long time to weld and needed welding jigs, and its aluminum motor-beam connectors were welded too.
  • Weight goal: save about 40 kg compared to PT1. The inner part was around 10 kg last time Julius checked; he still has to confirm this.
  • Electronics stay the same as PT1.
  • Payload: two DJI agricultural drone tanks. They are only available as STL, but buying them also gets you the pumps and electronics. Julius sees the two tank bays as universal, interchangeable cargo mounts. Spray tanks and granule/pellet spreading tanks already give two payloads to test. The tanks sit on a small adapter in the middle of the sandwich structure, like the angled square tube on the DJI drone. The batteries mount at the edge so the original XT15 cables can be used without routing power wires through the frame.
  • The overall shape keeps open the option of later adapting it to carry a person.
  • The landing skids will be bent tubes. Bending had mixed results on Feather PT1, probably because of missing tools; there is now a simple bending machine. If bending fails, joints are the fallback.
  • The centre currently has no landing-gear support; another tube will be added there. KBM suggested two smaller pairs of landing-gear structures instead of one large pair. Julius: it could also be split in the middle and held together with two walls.
  • Julius also plans to use cheap clamps (about $8 each) for the round tubes, to avoid drilling into them. KBM: a CNC version could be made in-house if needed.
  • Julius wants to start building in November. Thomas said he can start ordering components once the design is frozen.
  • Julius opened a forum thread for PT 1.5 feedback. KBM will copy his earlier PT2 design suggestions into it.

• Scope: PT 1.5, not PT2

  • KBM asked whether the layout should use vertical space (stacked levels) instead of horizontal. His reasons: larger cargo is hard to fit in the current layout, and the airframe may later need to become thicker or two-layer. He also suggested concentrating the batteries and moving the tanks to the outer side.
  • Julius will keep the current layout for now. For large cargo he suggested longer landing legs with a large container underneath. KBM's ideas are noted for later.
  • Julius on scope: "It should be a PT 1.5 and not a completely PT2." He doesn't want to spend too much time re-engineering. The size, layout and motor angles stay as they are.

• Square vs Round Tubes (Alperen)

  • Alperen suggested square tubes. Most of the load on the tubes is bending, and square sections handle bending better.
  • Square motor arms would also make manufacturing easier and avoid the motor-tilt problem Quiver had. Alperen uses square tubes on Spearhead and says the centre junction connections got better.
  • With square tubes, the printed junction connectors could be replaced by aluminum plates, which could also save weight.
  • The catch: off-the-shelf folding mechanisms for square tubes don't seem to exist. The current folding connectors come with clamps for round tubes.
  • Julius agrees square tubes would have advantages but hasn't decided. It's his call.

• Folding Motor-Beam Connectors: CNC and PEEK (KBM)

  • PT 1.5 keeps the foldable motor beams. KBM wants to make Arrow's own folding connectors by CNC machining, When asked why, since the current parts work, he said the off-the-shelf parts limit design freedom.
  • Julius is open to it as long as the parts are interchangeable with the current ones, including folding direction. He also pointed out the current connectors come from a single Chinese supplier. A CNC part could later be made locally, for example by a manufacturing node.
  • KBM has a friend whose factory will do CNC machining for Arrow at a private price. They will quote small batches from STL/STEP files. Thomas suggested bringing them on as a manufacturing node and connecting them to the Arrow store.
  • Material: KBM recommends PEEK because the material is cheaper than aluminum and could offset the machining cost. Julius prefers to start with aluminum, then get a PEEK quote for the same design to compare. PEEK needs slightly thicker walls and larger diameters. Plan: quote both.
  • KBM asked whether the locking lever could be replaced by a large screw. Julius: the lever is useful because it also presses the two halves together axially, but a screw along the axis would probably work too. KBM prefers screws for cost and simplicity (field crews carry screwdrivers) and pointed to the DJI M300's coaxial screw-cap arms. Others preferred the current lever design. DJI Agras drones use a double-lever release, which Julius thinks is even more complicated.
  • KBM noted CNC parts would make any motor-beam angle possible. The beams could angle upward while the motors stay horizontal. Tilted motors were discussed: they let you fly forward without tilting the aircraft as much and keep front and rear battery drain balanced. Julius said you can balance the drain by flying forward and backward on field patterns. KBM has only seen uneven drain in extreme side wind or temperature.
  • KBM asked whether to focus on layout first or go straight to the CNC parts. Julius: go straight to CNC machining.

• Open Items

  • Julius: continue the PT 1.5 design; confirm the weight estimate; add the centre landing-gear support; aim to build in November.
  • Julius: decide round vs square tubes.
  • KBM: make a simple aluminum folding-connector version and get a CNC quote from his friend's factory, then a PEEK quote for comparison. The parts must be interchangeable with the current connectors, including folding direction.
  • KBM: post the earlier PT2 design suggestions in the PT 1.5 forum thread.
  • All: add feedback to the PT 1.5 forum thread.
  • Thomas: order components once the design is frozen.

• Open Questions / Risks

  • Round vs square tubes is still open. Switching later would also change the folding connectors, and no off-the-shelf square-tube folding parts are known.
  • KBM is on vacation from October 2 for about a week or more (still checking in), which overlaps the quoting work.
  • The ~40 kg weight saving is a goal. The current estimate hasn't been checked yet.

September 29, 2026

Caribou Call — HearHear transcript (Discord) plus Erick's full recording transcript (Discord), which filled the gaps from 254 failed HearHear segments

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

• Tube Adapters: 3D-Printed Aluminum → CNC (Julius)

  • Julius had a quote and order set up to 3D print one of the tube adapters in aluminum. The manufacturer said the tolerance would be around 1–3 mm, which he wasn't comfortable with, so he didn't go ahead.
  • He converted the part into a CNC-machinable design instead, using Claude and Build123d on his original design. He showed the iterations: his original, a first version that still wasn't nice to machine, and a V3 that was much better. The process also produced images showing which surfaces get machined with which tool, drawings marking the threads, and a sheet for the RFQ.
  • There were more iterations after V3. The parts are close to final, but not quite.
  • Julius will use the new frame parts for both his build and Thomas's. Thomas suggested ordering them in one batch.

• PCBs

  • Julius will set up the next PCB order (with pick-and-place) in Thomas's JLC account so the parts ship straight to Thomas and the data is already in his account. Thomas agreed.
  • Julius explained the pick-and-place ordering steps: you check all the parts again and click through each one to confirm its orientation.
  • Before ordering, Julius has one small adjustment left on each PCB.

• Thomas's Build Prep

  • Thomas has ordered batteries and is ordering the spot welder today.
  • He doesn't have many days at home before the meetup, so he asked what he can start on. Julius: print the battery PCB enclosures and the main PCB enclosure, straight from Fusion.
  • Julius has another order of foldable motor-beam connectors coming and will ship them to Thomas.
  • Thomas asked about carbon fiber tubes and what else he needs. Julius will put it in the sheet.

• Longshot Cells

  • For Longshot, Thomas mostly needs the cells. Julius used BAK 65E cells (posted in chat). Thomas will look for cells he can source and send options to Julius for approval. Julius is open to one or two alternatives.
  • KBM: BAK's factory is in inland China. 6.5 Ah cells in this size are the highest capacity available, but they cost about twice as much as 5.0 Ah cells. The other maker of 6.5 Ah cells he knows is FEB (Far East Battery). He described both makers as struggling.
  • KBM: 5.0 and 5.8 Ah cells share dimensions, but 6.5 Ah cells are slightly larger, mostly in length. If sizing causes problems, dropping to 5.8 Ah is an option.
  • Julius: cell dimensions vary a bit, so before printing a Longshot enclosure he test-fits a cell in a small printed piece. He's fine with 6.5 Ah for now, but 5.8 Ah would make a cheaper, less cutting-edge pack.
  • Julius planned another Longshot flight after the call if there was enough daylight, otherwise the next morning.

• Folding Connector Material (KBM)

  • KBM asked whether to use aluminum or something else for the CNC folding connectors, and whether steel was worth considering. Julius: stick with aluminum. Steel would be heavier and more expensive to machine.
  • KBM noted that in China material is usually the biggest cost, not the machining. Julius: labor may be cheaper in China, but there are still operating costs and tool wear. KBM thinks off-the-shelf parts are cheap mainly because of mass production. He'll keep other materials as a longer-term research topic.

• BMS Dev Kit (Julius)

  • Julius received the BMS hardware, a development board about the size of a palm. It's the planned BMS for Longshot. He'll use it for testing, then copy its schematic into Arrow's own design. Alperen is fine with using a dev board for the first prototype.

• Spearhead Battery (cross-project)

  • Alperen: an 18 × 18 cm square battery layout will work best for Spearhead. Julius will try to make it as small as possible, since his first measurement left some margin.
  • Erick caught Julius up on Friday's discussion: Spearhead will likely use two batteries of the same capacity (primary and secondary). Alperen expects that for PT2.
  • Alperen proposed moving the Spearhead PT2 target from November to December. Reasons: several PCBs still need to be designed and made (main, tail, battery connector, BMS), JLCPCB supply issues are likely, PT1/PT1.5 hasn't flown yet, and there isn't enough time between the Turkey meetup and the US meetup to design the PT2 airframe and for Thomas to laser-cut and build it. Erick agreed. He finished the tail PCB schematic and started the main PCB.
  • Erick suggested the US meetup could still help with parts of the Spearhead build (wing, fuselage, batteries), but Alperen said only whatever is designed by then can be built.
  • Alperen: "At least we will have Caribou [to] fly." Julius: it's mainly ordering parts. Alperen said the Caribou design looks easy to build. By contrast, Spearhead is a large aircraft that needs to weigh about 10 kg total, which makes its structure harder.

• Other

  • KBM's Fusion tokens ran out, blocking his FEA work. Julius will take care of it.
  • KBM tried having Codex build a simple concept CAD tool (place, drag and join tubes) for early airframe concepts, without much success so far. Julius suggested asking Vector for concept models in Build123d.
  • Meetup travel: the group discussed meeting in Houston and flying out to Marfa together. Thomas will be flying back from India.
  • Julius will also visit Turkey after the Germany meetup, until Friday the 16th. Erick is coming to Turkey but not Germany (a family wedding that Saturday).

• Open Items

  • Julius: finish the CNC tube adapters and send out the RFQ; order a batch for both builds.
  • Julius: make the last PCB adjustments, then place the PCB order in Thomas's JLC account.
  • Julius: add carbon fiber tubes and the rest of Thomas's parts to the sheet.
  • Julius: ship the foldable motor-beam connectors to Thomas.
  • Thomas: order the spot welder; print the battery PCB and main PCB enclosures from Fusion.
  • Thomas: find Longshot cells he can source and send options to Julius for approval.
  • Julius: refill KBM's Fusion tokens.
  • Julius: make the Spearhead battery layout (18 × 18 cm or smaller).
  • Julius: test the BMS dev board.

• Open Questions / Risks

  • Longshot cell choice: 6.5 Ah (BAK 65E, pricier, slightly longer, few makers) vs 5.8 Ah (cheaper, standard size).
  • Spearhead PT2 moving to December was proposed on this call. Thomas didn't respond on the transcript, so it still needs confirming.