Parts List (basic FPV route, 6S)
These parts were checked on 2026-09-24 against the manufacturer’s recommended electronics list. The prices are point-in-time, before shipping and Idaho tax. The Amazon prices came from a browser session. GetFPV, the T-Motor store, BrotherHobby and EMAX international block automated reads, so their prices are absent.
Decisions assumed
Section titled “Decisions assumed”- Route: basic FPV with one RTK GPS, decided 2026-09-26. RTK lets the aircraft land on the hangar pad and geotag mapping images to centimetres. The airspeed sensor is included. For the first flights it is Ryan’s I2C MS4525DO on the LUX F765, whose one CAN port stays off for the first flights (Ryan, 2026-10-04); the DroneCAN DLVR-L10D (decided 2026-09-26) waits until CAN is on.
- Battery: 6S, decided 2026-09-25. The 4S table below is kept for reference only.
- Lift prop: 8-inch bi-blade. Thus the lift motors are 2807-class, 1050 to 1250KV.
- Pusher prop: a fixed APC 10x7E, the pusher version (LP10070EP), for now (Ryan, 2026-10-04), in place of the folding Aeronaut CAM Carbon Z 10x7 (decided 2026-09-27; its parts are in the optional table). It is the prop the propulsion model already uses, and not the list’s 12x8, which takes a 3115 640KV past its 60-second rating at full throttle on 6S. On the keel A landing gear the aircraft stands 168 mm below the pusher’s axis, so a blade at the bottom clears the ground by 41 mm (Propulsion). The pusher ESC’s brake stays off (freewheel).
- RC: ExpressLRS 2.4GHz. A RadioMaster radio with built-in ELRS is listed as optional.
- FPV: the Caddx Walksnail Avatar HD Moonlight kit and Walksnail goggles, chosen 2026-10-06 (Ryan). The Micro camera (19.6 x 19 x 24 mm, 2.1 mm M12 lens, F2.1) sits in the Camera Arm’s 20 mm cradle with no spacers. The cradle slot has no fore-aft stop, so mark the camera’s position or add a printed stop (th-blender, when the camera is in hand). The VTX board (34.5 x 34.5 x 15.3 mm, M2 holes on 20 x 20 and 25.5 x 25.5) goes on a holder that th-blender draws once th-telemetry has placed its two LHCP antennas on the nose ring. It records 4K on its own microSD card, with about 22 ms average latency. It runs from the power hub’s 12 V (row 9h) through the video switch (row 9c), at about 1.4 A (th-avionics), and its OSD comes over MSP DisplayPort on the LUX’s HD VTX port, SERIAL6 (th-avionics, 69aa35a). The goggles must be LHCP too, to match the kit’s antennas (th-telemetry). The earlier choice, the RunCam WiFiLink 2 with the Commonlands CIL292 lens (2026-09-28), and the DJI O4 Air Unit stay in the optional table. Only the kit and the goggles are listed below, not in the total.
- Mapping: MAPIR Survey3W RGB in a belly mount in the Fuselage Mid, added 2026-09-25. It is listed below, not in the total.
6S cart
Section titled “6S cart”| # | Item | Product | Qty | Unit | Line | Why | Link |
|---|---|---|---|---|---|---|---|
| 1 | Lift motors | EMAX ECO II 2807 1300KV: 47.6 g without its silicone leads, 18 AWG leads 200 mm long as shipped, 4 x M3 on a 19 mm circle, M5 prop shaft, 33.9 x 34 mm (EMAX spec sheet of 2020-09-26, from Ryan; was listed as 45 g with short leads). Ryan weighs one and measures the mount thread depth for the screw length | 4 | $25.99 | $103.96 | “EMAX ECO II all around” (Ryan, 2026-10-04), in place of the iFlight XING2 2809 1250KV. th-aeronautics (c49c8d9, on EMAX’s published 58 mΩ, ESTIMATE until the thrust stand), with the 3-blade F8041 that flies (row 3a): full throttle uncapped about 64 A a motor on a fresh pack (62.5 to 65.7), past EMAX’s 52 A and the Lucid 8S’s 50 A, so Q_M_SPIN_MAX drops to 0.82 until the stand (0.84 holds 49 A in the central case; th-avionics); thrust-to-weight about 3.6 to 3.8 at the cap; hover about 7.3 A a motor at 2.85 kg (6.7 to 7.8), about 29 A for all four. On the 2-blade (row 3) it was 54 A full throttle, held near 49 A by 0.95, and 6.8 A in hover (2d2917e). A long climb on the lift motors is still the case to watch (th-avionics): touch-check the motors after the first multi-minute hover. On the bench, check that the mount screws do not reach the windings (the XING2 needed M3x8) and that the F8041 hub fits the 5 mm shaft |
getFPV, EMAX USA |
| 2b | Lift ESC, 4-in-1 | TBS Lucid 60A 3-8S AM32 4-in-1 ESC, NDAA: 50 A x 4 continuous, 60 A x 4 (60 s), 85 A x 4 peak (10 s) (TBS manual, section 13.3), 3 to 8S, AM32 target AM32_TBS_8S_4IN1_F421 (firmware 2.18 or later), 50 x 42 mm, 17.6 g (22 g with brackets), no BEC, current sensor, bidirectional DShot and KISS telemetry; XT60, a 560 uF capacitor and brackets (Gorilla 39.5 x 16, 30.5 x 30.5, 20 x 20) in the box | 1 | $49.99 | $49.99 | The 8S model, bought in place of the 6S model this row named (34 x 44 mm, 13.8 g; Ryan, 2026-10-04); it runs 6S. CAUTION: it is 16 mm longer, so th-blender’s Lift ESC Cradle (4in1) and the starboard-wall position at (-43, 180, 9) must be checked again before g16 prints (th-blender, th-main). Flies first (Ryan, 2026-09-29; CLAUDE.md decisions): one board in the fuselage. The DroneCAN pod ESCs are the follow-on, below the total. The phase leads use the kit’s path, spar, slot and channel, 6 x 18 AWG per side, through a 10 mm exit hole (charge-node.md). TBS asks for good airflow with props over 7 inch; ours are 8 inch, so a hover heat check comes first. Its 8-pin lead must match the flight controller’s ESC #1 port (VBAT, GND, CUR, TLM, M1 to M4): check the silkscreen and meter pin 1 (th-avionics) |
getFPV, TBS |
| 2c | Motor wire extensions | 18 AWG silicone wire, 6 colours, 13.2 ft each | 1 | $28.79 | $28.79 | For row 2b: each motor’s three wires from the spar tip to the ESC, about 0.52 m on the starboard side and 0.65 m on the port side, about 7 m in all (ESTIMATE from the routes). The motors ship with 200 mm of 18 AWG lead: cut each to reach the spar tip only (about 40 mm, ESTIMATE until the dry fit) and make the joint there, outside the spar, because the bore has no room for a soldered joint (the leads fill it 90 to 98 %). Two colours per motor pair make the phase order easy to keep. Price from the listing, check it at checkout | eBay |
| 2d | Phase connectors | Amass MR30 male and female, 5 pairs | 1 | $8.99 | $8.99 | For row 2b: one MR30 per motor at the ESC end, so the ESC comes out without cutting the phase leads. Fit them after the leads are through the spar; the manual warns that a connector may not pass the 14 mm spar | Lumenier |
| 2e | 4-in-1 bulk capacitors | Rubycon 35PZJ560M10X15, 560 uF 35 V hybrid polymer aluminium (PZJ series), 10 mm diameter, 15 mm tall, 5.0 mm lead pitch, 11 mΩ, 4,000 h at 125 °C (DigiKey 1189-35PZJ560M10X15-ND) | 10 | $2.197 | $21.97 | Three in parallel at the Lucid’s battery pads, 1,680 uF (th-eda, 2026-10-14). TBS’s manual (Lucid Rev 1.15, section 13.5) asks for 1,000 uF or more within 1 cm of lead for 25 to 29 mm motors, and our 2807s are in that class. The bulk carries about 15 A rms of ripple at hover (th-eda, ESTIMATE): three hybrids are inside their rating, one 1,000 uF electrolytic would be about 5 times over. th-blender fits them in v2.4B, two ahead of the duct wall and one above the spar, about +3 g (ESTIMATE). Ten for the price break at 10; seven are spares. 35 V is enough on v2.4B (th-eda, 2026-10-14): the worst case on a full pack is about 31 V (ESTIMATE: 25.2 V, plus about 3 V of sustained regen into the pack, plus about 2.6 V if all four motors reverse 49 A at once into the 1,680 uF and about 150 nH of feed), under the 35 V rating, with the 40 V surge rating as margin. The earlier 50 V rule was for 300 uF at the end of a 300 mm feed. A 50 V hybrid of this size is only 220 uF, so three cannot meet the TBS minimum. No capacitor covers a pack that disconnects while the motors brake: keep the XT90 fully mated. Bench check on the thrust stand: scope the Lucid’s battery tabs (1 us sampling or faster; the INA228 is too slow) during braked spool-downs from the throttle cap on a full pack. Pass: the peak stays below 35 V. The 560 uF capacitor in the Lucid’s box is not used. Replaces three Panasonic EEU-FR1H101 (100 uF 50 V), which missed the TBS minimum (not ordered) | DigiKey |
| 2g | Bus capacitors, 470 uF | Nichicon UHE1H471MPD, 470 uF 50 V low-ESR electrolytic, 10 mm diameter, 31.5 mm tall (DigiKey 493-1613-ND) | 2 | $1.39 | $2.78 | One at the video unit’s 12 V input (from power hub A, row 9h, through the video switch, row 9c), and one across hub A’s BATT pads (row 9h; hub B, row 9b, left the build on 2026-10-15) against the brake spikes on a full 6S pack (their maximum input is 26 V). 50 V for margin on a bus that takes brake spikes (th-avionics). The Panasonic EEU-FR1H471 (12.5 x 20 mm, P14461-ND) is out of stock at DigiKey in every packaging (th-eda, 2026-10-08); if the 31.5 mm height does not fit beside a hub, the Nichicon UHE1V471MPD (35 V, 10 x 20 mm, 493-1582-ND, $0.94) is the fallback | DigiKey |
| 3 | Lift props | Gemfan F8041 folding 2-blade, set = 1 CW + 1 CCW | 4 sets | $5.99 | $23.96 | Bought to fly first; since 2026-10-08 the 3-blade (row 3a) flies (Ryan), so these are the comparison and spare sets | getFPV |
| 3a | Lift props, 3-blade | Gemfan F8041 folding 3-blade, black, set = 1 CW + 1 CCW | 4 sets | $6.49 | $25.96 | These fly: “we are using the 3 blade, folding gemfan props on lift motors” (Ryan, 2026-10-08, via th-blender), in place of the 2-blade (row 3); 2 sets fly, 2 spare. A 3-blade draws more current at the same throttle: about 64 A a motor uncapped, so Q_M_SPIN_MAX 0.82 until the stand, and 7.3 A in hover (th-aeronautics c49c8d9, ESTIMATE; row 1). Each prop is 11.6 g by Gemfan’s spec (Rotorama lists 13.2 g: weigh one), against 7.79 g for the 2-blade; the folded third blade costs about 0.5 % of cruise drag. Measure it on the thrust stand before it flies, capped at 0.82 first (docs/propulsion.md) |
getFPV |
| 4 | Pusher motor | EMAX ECO II 3115 640KV | 1 | $35.49 | $35.49 | Decided 2026-09-27 (Ryan); in stock again at getFPV (2026-10-04), so the 635KV search stopped. EMAX’s own drawing shows 4 x M3 on a 19 mm circle, so it bolts to the kit mount (a reseller’s “16x16” is wrong). 106 g, 6 to 8S, 63 A peak, 66 mΩ. Threaded M5 shaft, 17.5 mm above the bell, with a flange nut; the APC’s 5 mm bushing (row 6) fits it. On the APC 10x7E it turns about 13,000 RPM at full throttle, inside the prop’s 14,500 (th-aeronautics). Measure the current with the 10x7 on the thrust stand before flight | getFPV, EMAX USA |
| 5 | Pusher ESC | DYS Aria 70A 3-6S AM32 ESC: 70 A continuous, 80 A burst, no BEC, current sensor, 9.5 g, 21 x 28 x 6 mm | 1 | $49.49 | $49.49 | Ryan, 2026-10-04, in place of the Hobbywing Skywalker 60A V2; order the AM32 version, not the older BLHeli_32 page. Set it up in the AM32 configurator (esc-configurator.com) through a USB linker on the signal wire or the flight controller’s pass-through: brake off at zero throttle (the fixed APC freewheels), no low-voltage cutoff, PWM on output 7 (th-avionics, 58a861d) | getFPV |
| 6 | Pusher prop | APC 10x7E thin electric, pusher (LP10070EP, black), with 5 reducing bushings (3, 4, 5, 6, 8 mm) | 2 | $3.69 | $7.38 | Fixed prop for now (Ryan, 2026-10-04), the one the propulsion model uses; one is a spare. The 5 mm bushing fits the motor’s M5 shaft, and the motor’s flange nut clamps the hub. The pusher version suits a motor in its normal direction (th-aeronautics): on the bench, check that the air blows aft before the first flight. APC’s limit for this prop is 14,500 RPM (145,000 / diameter in inches): a 635KV motor on a full 6S pack (25.2 V) turns about 16,000 RPM with no load, so th-aeronautics confirms the loaded RPM | Innov8tive Designs |
| 7 | Elevon servos | Hitec MD65MG (PWM), 11.6 mm metal gear, contactless magnetic encoder, ball bearing | 2 | $57.99 | $115.98 | Decided 2026-09-27 (Ryan): a UAV-grade servo in place of the ES08MD II. Same body size (23.6 x 11.4 x 25.8 mm, 11.9 g), so it should fit the kit’s Servo Mount (th-blender to check). 1.8 to 2.2 kgf cm, 0.11 to 0.14 s/60°, 4.8 to 6.0 V. The command stays on PWM from the FC, so a CAN fault cannot take the elevons; our servo node powers it and reports current and voltage. It has no potentiometer to tap, so surface position comes from an angle sensor at the elevon hinge (th-eda). Hitec’s M25T spline: use the horns it comes with. The DroneCAN version (MD65MG-DroneCAN, $79.99) is the alternative | Hitec CS |
| 7a | Servo extension leads | Back Bay RC JR/Hitec servo extension, 60 cm, twisted 20 AWG (100 strands), with safety clips, pack of 5 | 1 | $11.99 | $11.99 | Fuselage C (th-avionics, th-blender 8765582, e0c1d05): the run from the Y lead to each servo is about 635 mm, 160 mm of it the MD65MG’s own lead (Hitec datasheet), so each extension is about 475 mm plus slack: 600 mm, cut to length. 20 AWG in place of 22 AWG: about 20 mV less drop at a stall, where hub B’s 5.0 V leaves only 40 to 90 mV over the servo’s 4.8 V minimum (4.84 to 4.89 V at a stall with the contacts counted, docs/electronics-notes.md, ESTIMATE); 22 AWG stays acceptable. JR style, to match the Y harness (row 9d). Two fitted, three spare. Each lead runs from the servo in the wing through the Wing Inner wire channel (manual, wing step 4) and the wing conduit into the bay. The joint sits inside the wing near the root, where you cannot reach it again: put heat shrink over it. Measure in the dry fit before you cut or join. Sand the servo plug’s square top edges if it does not enter the JR socket. Bench check: hold each elevon against its stop and measure 4.8 V or more at the servo plug |
Back Bay RC |
| 8 | Linkage set | Z pushrods + linkage stoppers | 1 | $9.98 | $9.98 | Control horns printed or bought separately | Amazon |
| 9 | Flight controller | Lumenier LUX F765 NDAA: STM32F765 (2 MB flash), one ICM42688 IMU, BMP280 barometer, no compass, 8 UARTs, 12 outputs, one CAN port, 30.5 x 30.5 mm M3 holes, 38 x 38 mm, 9 g; grommets and SH cables in the box | 1 | $89.99 | $89.99 | For the first flights (Ryan, 2026-10-04), in place of the NxtPX4v2 (in the optional table as the checked alternative); our own flight controller (two CAN buses, STM32H743, th-eda) follows. Its CAN port stays off for the first flights: it flies the 4-in-1 (rows 2b to 2g), the I2C airspeed sensor (row 11c) and no charge node. Outputs: the lift motors on 1 to 4 through the ESC #1 port, elevons 5 and 6, pusher 7, camera trigger 8; S1 is kept for the cargo-door servo (off in the first build) and S4 drives the CG slider’s servo (electronics-notes.md). It does not boot ArduPilot without a microSD card (row 9g). Parameters: params/board-luxf765.parm (th-avionics, 3c9f3f6). Its plate is the Avionics Plate (LUX F765) (th-blender) |
getFPV |
| 9e | Power module | Holybro PM02 V3 (12S), analog: voltage and current to the flight controller, 120 A current sensing, 60 A continuous on its PCB | 1 | $18.99 | $18.99 | Came with the 6C Mini. Its current signal goes to pin 3 of the LUX F765’s ESC #1 port (BATT_AMP_PERVLT 36.36, th-avionics). Re-lead it with 10 AWG and the XT90-S (rows 15a, 15b) |
Holybro |
| 9f | GPS and airspeed Y lead | JST-SH 1.0 6-pin housing and pre-crimped SH 1.0 wires, for the board end | 1 kit | about $8.00 | $8.00 | ESTIMATE, part to choose. The LUX F765’s only external I2C is on its GPS port, a 6-pin JST-SH 1.0: 1 5 V, 2 TX3, 3 RX3, 4 SCL, 5 SDA, 6 GND (th-avionics, 3c9f3f6). Make one Y lead: SH 6-pin at the GPS port, all six wires through to the Q43’s GH 1.25 8-pin (GH housing and crimps from the harness’s GH kit, H4), and a 4-wire branch (GND, 5 V, SCL, SDA) to the MS4525DO. For the branch, cut Ryan’s GH-to-GH jumper from the sensor kit in half and splice that half in, so the sensor keeps its own latching GH 1.25 4-pin plug (Ryan, 2026-10-04); its solder pads are the fallback. The compass (QMC5883P) and the MS4525DO (0x28) have different addresses, so they share the bus. CAUTION: before you connect the sensor, measure 5 V on the GPS port’s pin 1 and the resistance from the sensor board’s SDA and SCL to its own 5 V pin: pull-ups to 5 V there would drive 5 V into the H743’s I2C pins (electronics-notes.md). Splice by signal name, not by pin number: the GPS port, the Q43’s GH 8-pin and the sensor’s GH 4-pin number their pins differently. Read the sensor’s order off its silkscreen, then buzz the finished lead end to end before it touches the board (th-avionics) |
part to choose |
| 9g | microSD card | microSD card, 4 to 32 GB, FAT32, for ArduPilot’s logs; the LUX F765 does not boot ArduPilot without one (Lumenier’s manual) | 1 | about $8.00 | $8.00 | ESTIMATE, part to choose: any name-brand card | part to choose |
| 9h | Power hub A (video) | Matek Power Hub V2.1: 36 x 36 mm board on 30.5 mm holes, 6 g; input 7 to 26 V (2 to 6S); 12 V 2 A continuous (3 A for 10 s a minute) and 5 V 3 A continuous, synchronous buck, TVS at the input, an LED on each output; its ESC distribution pads stay unused | 1 | $9.68 | $9.68 | Hub A, chosen 2026-10-08 (Ryan), in place of the 12 V video BEC (a Matek BEC12S-PRO). Its BATT pads take the pack tap. Its 12 V feeds the Walksnail Moonlight (about 1.4 A) through the video switch (row 9c). Its 5 V (3 A) is free for loads that are not flight-critical, such as the CG slider’s servo and its XIAO RP2040; never the elevons, which have their own hub, hub B (row 9b; Ryan, 2026-10-08), and not the night lights (a strobe pulse is about 55 W, th-avionics). CAUTION: a full 6S pack is 25.2 V against its 26 V maximum, and the lift ESCs brake on stop, which sends short spikes back into the pack: fit a 470 uF low-ESR capacitor (row 2g) across its BATT pads, and check polarity before the first power-up. Use its 12 V and 5 V pads only, not the VTX and CAM pads (analog video). Tap it on the pack side of the PM02 V3. Place it like an ESC body for the compass: 100 mm or more from the Q43, its pack feed and its 12 V and 5 V leads as twisted pairs (th-telemetry). Needs a 36 x 36 mm place in the bay (th-blender) | Amazon |
| 9c | Video switch | Pololu Mini MOSFET Slide Switch with Reverse Voltage Protection, SV (#2811): 4.5 to 40 V, 2.0 A continuous at 55 °C, 0.6 g | 1 | $4.49 | $4.49 | In series between the power hub’s 12 V output (row 9h) and the Walksnail Moonlight. The LUX F765’s HD VTX port pin 1 (its 9 to 10 V, 2 A rail) stays unconnected: the Moonlight is too near that rail’s limit (electronics-notes.md). Its ON input is the S3 pad, GPIO 60, with RELAY2_DEFAULT 1 so the video is on at boot (th-avionics, board-luxf765.parm). Open or low means off, so a broken wire turns the video off; the slide switch in “on” overrides it for bench work. The Moonlight draws about 1.4 A at 12 V, inside the switch’s 2.0 A |
Pololu |
| 9d | Servo power Y lead (SUPERSEDED: do not order) | JR servo Y harness, 1 male to 2 female | 0 | about $7.00 | $0.00 | SUPERSEDED (2026-10-15): each servo now has its own BEC and supply run (row 9j), so no Y. Was: ESTIMATE, price and wire gauge not shown to scripts: check on the page. Make the supply pair (5 V and ground) 20 AWG: hub B gives 5.0 V against the MD65MG’s 4.8 V minimum, so keep the drop small (th-avionics). The flight controller’s outputs carry signal only, so the Y takes hub B’s 5 V (row 9b), with the bulk capacitor (row 9i) across it, to the power of both elevon servos; each servo’s signal wire, with a ground, goes to the LUX F765’s M5 and M6 pads (th-avionics). | Amazon |
| 9b | Power hub B (elevon servos) (SUPERSEDED: do not order) | Matek Power Hub V2.1, as row 9h: 36 x 36 mm on 30.5 mm holes, 6 g; input 7 to 26 V; 5 V 3 A continuous (its 12 V unused) | 0 | $9.68 | $0.00 | SUPERSEDED (Ryan, 2026-10-15): no place on Fuselage C keeps th-telemetry’s 10 mm coax rule for a 36 mm hub with its can (th-blender W48.2), so two Matek Micro BECs (row 9j) feed the servos. Was: hub B (Ryan, 2026-10-08, via th-blender; th-avionics eeea9fc), in place of the 5.2 V BEC12S-PRO (optional table). Its 5 V feeds the two elevon servos only, through the Y lead (row 9d) with a bulk capacitor (row 9i), never the flight controller: the LUX F765 keeps its own 5 V from the pack through the ESC connector’s VBAT, so one hub cannot take down both. Two MD65MG at stall draw about 2 A at 5 V, inside its 3 A. It gives 5.0 V against the MD65MG’s 4.8 V minimum, so the Y lead’s supply pair is 20 AWG. CAUTION: the same 26 V maximum as hub A against a full 6S at 25.2 V and the lift ESCs’ brake spikes: tap it on the pack side of the PM02 V3 with a 470 uF low-ESR capacitor (row 2g) across its BATT pads, and check polarity before the first power-up. On the starboard bay wall across from hub A (th-blender is placing it; the charge node’s bracket area, Y 158 to 212, may move); like an ESC body for the compass, 100 mm or more from the Q43, leads twisted (th-telemetry) | Amazon |
| 9i | Servo bulk capacitor | Panasonic EEU-FR1A102, 1000 uF 10 V FR low-ESR electrolytic, 10 x 17.5 mm (DigiKey P124221-ND) | 2 | $1.15 | $2.30 | One across each servo’s supply at the servo end of its run, on its own Micro BEC (row 9j, th-avionics 2026-10-15); was one at the hub B Y lead: it holds the rail up through servo stall peaks (th-avionics eeea9fc; part th-eda, 2026-10-08, 2,592 in stock) | DigiKey |
| 9j | Elevon servo BECs | Matek Micro BEC 6-30V to 5V/9V-ADJ (MBEC6S), 3-pack: 6 to 30 V in (2 to 8S), 5 V out (9 V with the tabs bridged), 1.5 A continuous (2.5 A for 5 s a minute), hiccup over-current protection, thermal shutdown, 10 x 14 x 3 mm, 1 g | 1 pack | $14.99 | $14.99 | Ryan, 2026-10-15: one per elevon servo, in place of power hub B (row 9b). One BEC on two servos is NOT enough: about 2 A at a stall would trip its hiccup protection and drop both elevons at once; with one each, a trip loses one elevon (th-avionics). Both clip into two slots in hub A’s pad: slide each in from aft with its leads soldered, and a sprung tab holds it; no screws (th-blender W53, 6611f51d). They are tapped together from the pack side of the PM02. Leave the 9 V tabs open on both. Each takes a 1000 uF at its servo’s end (row 9i) and a ceramic bank at its input (row 9k); a second servo-rail sense channel on the slider XIAO (th-eda to confirm a spare ADC pin; the divider parts are in rows 17g and 17k). CAUTION: 30 V maximum input, so bench check 5 applies at each input: the peak must stay under 29 V on a full pack; if it fails, the Matek BEC12S-PRO (5 A, 9 to 55 V) goes in instead (th-avionics’ first choice). One BEC spare. MyFPV’s price, out of stock there on 2026-10-15 | Amazon (Ryan’s), MyFPV specs |
| 9k | BEC input capacitors, 10 uF | TDK CNC6P1X7R1H106K250AE, 10 uF 50 V X7R 1210, soft termination, 2.7 mm maximum height (DigiKey 445-181655-1-ND); if out of stock, TDK C3225X7R1H106K250AC (445-C3225X7R1H106K250ACCT-ND, $0.647 at 10) | 6 | $0.853 | $5.12 | Two at each Micro BEC’s input pads (row 9j) with one 100 nF (row 9l), hand-soldered: there is no room for a 470 uF can there, and none is needed for energy since the Lucid’s 1,680 uF sits on the same bus (th-avionics, 2026-10-15). Four needed, two spare. Parts th-eda’s (power distribution v5, EDA 5b6cb9d); DigiKey price at 10 via Part-DB, 85 in stock on 2026-10-15 | DigiKey |
| 9l | BEC input capacitors, 100 nF | Murata GCM188R71H104KA57D, 100 nF 50 V X7R 0603 (DigiKey 490-4779-1-ND) | 10 | $0.114 | $1.14 | One at each Micro BEC’s input pads beside its two 10 uF (row 9k), hand-soldered. Two needed; ten because an 0603 is easy to lose (th-eda, EDA 5b6cb9d; DigiKey price at 10 via Part-DB) | DigiKey |
| 10 | RTK GNSS | Waveshare Q43 NEO-F9P GNSS Helical RTK Module (u-blox NEO-F9P-15B, L1/L5 helical antenna, QMC5883P compass) | 1 | $229.99 | $229.99 | Decided 2026-09-28 (Ryan, th-telemetry): replaces the H-RTK F9P Ultralight ($279) and the LC29H rover plan. UBX to ArduPilot’s u-blox driver, GPS L1/L5 and Galileo E1/E5a; RTK from the hangar’s LC29H base (bench-test the RTCM before first flight). Ø47.04 x 37.00 mm dome, 24 g, sunk under the radome on Fuselage Mid B at Y 404 (on the Q43 Seat). UART and I2C on a GH1.25 8-pin; to the LUX F765’s GPS port (6-pin SH 1.0) through the Y lead, row 9f (th-avionics) | Waveshare |
| 10a | Q43 RF shield floor | MG Chemicals 503, single-sided copper-clad FR4, 1 oz copper, 1.57 mm, 76.2 x 127 mm (DigiKey 473-1000-ND) | 1 | $11.09 | $11.09 | Required on Fuselage C (th-telemetry b9bafe6, 2026-10-10): the Lucid 4-in-1 sits 39 mm under the Q43, so the Q43 seat gets a copper floor at least 67 mm across, copper face toward the Q43, with a short ground lead to the flight controller ground. Plain or tin-plated copper only: never nickel EMI tape or steel, which are magnetic beside the compass. Cut it to th-blender’s seat outline; about 14 g at 70 x 70 mm (ESTIMATE). Price from DigiKey’s related-item list, not read on the product page | DigiKey |
| 10b | External compass (SUPERSEDED: do not order) | DFRobot Fermion SEN0697: BMI323 IMU, BMM350 3-axis magnetometer and BMP581 barometer on one breakout, I2C (DigiKey 1738-SEN0697-ND) | 0 | $26.81 | $0.00 | SUPERSEDED (Ryan, 2026-10-15): it cannot be bought, so the SEQURE M10-25Q (row 10d) is the compass. Was: Ryan confirmed it on 2026-10-10 (via th-blender): on Fuselage C the Q43’s own compass is off and this board’s BMM350 is the compass, in the Y 404 keep-out, on th-blender’s seat. Ryan’s SEN0696 stays out: ArduPilot cannot read it (th-avionics). Its 3.3 V supply is th-eda’s to specify. DigiKey price from a related-item list, not read on the product page; DFRobot sells it direct for $17.90 | DFRobot wiki, DFRobot store |
| 10c | Compass supply parts (SUPERSEDED: do not order) | Diodes AP2112K-3.3TRG1 3.3 V 600 mA LDO, SOT-23-5 (DigiKey AP2112K-3.3TRG1DICT-ND), and KEMET C0805C105K3RACTU 1 µF 25 V X7R 0805 (DigiKey 399-C0805C105K3RACTUCT-ND) | 0 | $0.28, $0.19 | $0.00 | SUPERSEDED with row 10b (2026-10-15): the M10-25Q runs on 5 V. Was: the SEN0697 is 3.3 V only and the LUX’s GPS port gives 5 V, so its lead carries one regulator with a 1 µF capacitor on each side (th-eda’s choice, 2026-10-10; wiring in docs/electronics-notes.md, th-avionics bbb10e6); one regulator and two capacitors spare. The same DigiKey numbers as the lights node’s U5 and C33, so they can go on the row 13d order. Prices from the lights BOM’s Part-DB run |
hw/lights-node/digikey/lights-bom.md (DigiKey prices via Part-DB) |
| 10d | GPS2 and flight compass | SEQURE M10-25Q GPS module: u-blox M10 class, L1 (GPS, GLONASS, BeiDou, Galileo), QMC5883L compass, 5 V, UART (115200 from the factory, UBX; ArduPilot sets it to 230400 at detection, so SERIAL4_BAUD stays 230) and I2C on an SH1.0 6-pin connector, 25 x 25 x 8 mm, 12.2 g | 1 | about $28.00 | $28.00 | ESTIMATE price (retailers list about US$27 to $29; Amazon’s price not read). Ryan, 2026-10-15, in place of the SEN0697 (row 10b), which cannot be bought: GPS2 on the LUX F765’s SERIAL4 (UART8, underside pads), GPS_AUTO_SWITCH 4 with the Q43 RTK as primary while it has 3D, and its QMC5883L the only flight compass; the Q43’s QMC5883P stays off (th-avionics f7b6ae3f, e05774a2). On its PETG hanger under the Bay roof on the SEN0697’s old dowel holes (th-blender W52.1). Check its pinout against the LUX’s SERIAL4 pads and I2C before you make the lead (th-eda), and re-check the LUX’s free memory with it wired (th-avionics) | Amazon (Ryan’s), Rotorama specs |
| 11c | Airspeed, first flights | MS4525DO kit (4525DO-DS5AI001DP: ±1 psi differential, I2C address 0x28, 5 V) with its own straight pitot tube, 99.5 mm x 4 mm (not used since W104: our own brass total tube, row 11g, takes its place) | 1 | on hand | $0.00 | Ryan has it (2026-10-04). On the LUX F765’s GPS port I2C, through the branch of the Y lead (row 9f) (ARSPD_TYPE 1, ARSPD_BUS 0). Measure 5 V on the GPS port’s pin 1 before you connect it. The total tube mounts in the nose cone unit (th-blender W104). The DroneCAN DLVR-L10D waits until CAN is on (Follow-on below) |
|
| 11d | Static port tubes | K&S Precision Metals 8125 round brass tube, 1/16 in OD (1.59 mm) x 0.014 in wall, 0.035 in ID (0.89 mm), 12 in long, 3 pieces, alloy 260/272 | 1 pack | $5.99 | $5.99 | The flush static ports in the Fuselage Front (nose ring, pack pocket), one each side at Y 40.0, Z -10 on Fuselage C v2.4B (th-blender W77, de8f23b8; Y 46 on the v2.3 nose ring) (th-blender e1f6bdc, Ryan 2026-10-06). Cut two 13 mm lengths, bond them flush in the 1.7 mm holes (0.11 mm diametral clearance) and sand to within 0.05 mm with no burr. The ID is 0.89 mm, not the 1.0 mm drawn: no effect on a static port, which has no flow. About 6 mm of each tube stays inside the wall for the line. The pack gives spares for a second nose | K&S |
| 11e | Static line tee | Component Supply FBNT-062-10, nylon barbed tee, 1/16 in barbs, pack of 10 | 1 pack | $9.50 | $9.50 | Joins the two port branches into the trunk to the Pitot Coupler’s static spigot (th-blender W111 c4d7ee16): the two branches are equal within 0.5 mm, so the tee cancels sideslip | Component Supply |
| 11f | Static port line | Component Supply SCT-063A silicone tubing, 1/16 in ID x 1/8 in OD x 1/32 in wall, peroxide cured, per foot | 2 ft | $0.75 | $1.50 | ESTIMATE price: the pages show $0.46, $0.72 and $0.75 a foot; check at checkout. Three cuts, routed inside the nose (th-blender W111 c4d7ee16, W111.2 d0d22013: the port branch rerouted clear of the radio cradle lid, the starboard one lengthened to match): port branch 106.7 mm and starboard branch 106.7 mm from the ports to the tee, and the trunk 71.1 mm from the tee to the coupler on route B (Ryan, 2026-10-20; th-blender W117.1 88ac2c95: a loop down the starboard wall at Y 80, held by the Static Line Clip), 284.5 mm in all. The coupler’s static spigot is 3.4 mm OD, too large for this line, so the trunk’s 1/8 in (3.18 mm) OD end goes into a short cut of the 3 x 5 line (row 11h), which stretches over the spigot: an 8 mm cut, 5 mm on the spigot and 3 mm over the trunk’s end (th-blender’s fit: 2.1 mm to the MS4525 board, 6.1 mm to the Front). The kit’s line is not used | Component Supply |
| 11g | Pitot total tube stock | K&S Precision Metals 9822 round brass tube, 4mm OD x 0.45mm wall (ID 0.122 in, 3.1 mm), 300mm long, 3 pieces, alloy 260/272 | 1 pack | $9.99 | $9.99 | Our own total-pressure tube in place of the MS4525 kit’s tube (Ryan, 2026-10-19: “own tube with 15 camera”; th-blender W104 888383ba; the cut list is pitot-tube’s HANDOFF 3.6): cut 83 mm, trim to 80.90 mm, open bevelled mouth, with a soldered collar from rows 11l and 11m. About 83 mm of the 900 mm is used | K&S |
| 11h | Pitot total-pressure sleeve | Adafruit 4661 silicone tubing, 3mm ID x 5mm OD, 1 m | 1 | $2.50 | $2.50 | An 11 mm cut is the total line’s joint in the W104 nose (th-blender 888383ba): 5.16 mm over the own tube’s plain aft end, a 0.84 mm butt gap, then 5.0 mm over the Pitot Coupler’s 4.0 mm barbed spigot, dry, with no grease or tie (th-avionics). It is the joint that lets go when the cone unit slides off, so do the blow-and-hold check after every cone fit; a second cut, 8 mm (5 on the spigot, 3 over the trunk; th-blender’s fit), joins the static trunk (row 11f) to the coupler’s 3.4 mm static spigot; the rest is spare | Adafruit |
| 11i | Tube bender (SUPERSEDED: do not order) | K&S spring tube bender set, 5 pieces: 1/16, 3/32, 1/8, 5/32 and 3/16 in OD (Micro-Mark 60360) | 0 | $9.95 | $0.00 | SUPERSEDED (th-blender W75.1, 2026-10-17): the brass return is three straight soldered pieces, so nothing is bent. Was: the 5/32 in spring for a 90 deg bend on a 4 mm radius, which would kink | Micro-Mark |
| 11j | Brass soldering flux | Harris Stay-Clean liquid soldering flux (zinc chloride), 30 ml bottle | 1 | $5.75 | $5.75 | For the own tube’s collar joints (rows 11g, 11l, 11m; W104), with ordinary soft solder. Its residue is corrosive: wash each joint in warm water, blow the bore clear and dry it before the tube goes near the electronics | Western Metal Art Supply |
| 11k | Airspeed, starboard boom (SUPERSEDED: do not order) | Matek ASPD-AUAV digital airspeed sensor: All Sensors AUAV-L30D, ±30 inH2O (7500 Pa), used in its I2C mode (JST-GH 4-pin: 5V, SCL, SDA, GND; 20 cm GH-4 cable included), 4.5 to 5.5 V, 36 x 32 x 7.3 mm, 5 g | 0 | $97.00 | $0.00 | SUPERSEDED (Ryan, 2026-10-18: no wing booms; W93.2 60f05e70 flies one MS4525, row 11c, beside the camera). Was: The second sensor for the wing booms on both sides (Ryan, 2026-10-18; th-blender W86): the port boom keeps the MS4525DO (row 11c, I2C 0x28), and this one is read by ArduPilot’s AUAV driver at I2C 0x26 on the LUX’s one bus (ARSPD2_TYPE 19, AUAV-30in; th-avionics, AP_Airspeed_AUAV.cpp). Matek’s spec sheet pairs the I2C port with ARSPD_TYPE 19 (‘AUAV-L30D I2C mode’), but it does not say how I2C mode is selected: confirm on the bench that the sensor answers at 0x26 before it is fitted. Mount it on the fuselage side of the wing joint, so only the pneumatic lines cross the joint. Price at fpvfaster, stock not shown (RMRC: $113.35, back-ordered, 2026-10-18). Fallback: a QioTek ASP5033 module ($32), whose about ±40 Pa error is too coarse near the stall | fpvfaster |
| 11l | Pitot tube collar, inner | K&S Precision Metals 9823 round brass tube, 5mm OD x 0.45mm wall (ID 0.162 in, 4.1 mm), 300mm long, 3 pieces | 1 pack | $10.99 | $10.99 | A 1.5 mm slice over the 4 mm tube (row 11g), the collar’s inner ring (th-blender W104; pitot-tube’s HANDOFF 3.6); the rest is spare | K&S |
| 11m | Pitot tube collar, outer | K&S Precision Metals 9824 round brass tube, 6mm OD x 0.45mm wall (ID 0.202 in, 5.1 mm), 300mm long, 2 pieces | 1 pack | $10.99 | $10.99 | A 1.5 mm slice over row 11l’s ring, so the collar is 6 x 1.5 mm and seats in the Nose Cone Plate’s pocket (W104); the rest is spare | K&S |
| 12 | RC receiver | RadioMaster RP4TD ELRS 2.4GHz diversity | 1 | $27.92 | $27.92 | Named; update to ELRS 3.5+ for MAVLink | RMRC |
| 13 | Battery | Nexus Battery Systems 6S2P Molicel P50B 21700, 10 Ah, 21.6 V nominal (25.2 V full), 216 Wh, 139 x 48 x 72 mm, 900 g, 120 A continuous, 10 A charge, 16.2 V recommended cut-off, 15 V minimum; configured with the XT90-S female discharge connector on 10 AWG, balance leads only (no BMS) and the 7-pin JST-XH 2.54 balance lead, not the JST-HY 2.0 of Nexus’s Daly BMS option, which the charge node cannot mate (the charge node reads the cells through a 6S JST-XH, th-eda) | 1 | $252.00 | $252.00 | The female mates with the XT90-S male on the aircraft lead (row 15a). Decided 2026-10-01 (Ryan), in place of the Tattu 6S 4500 LiPo: in SITL with the battery model draining correctly (th-simulator 1daccb3), a 100 ha survey in an 8 m/s wind uses about 74 Wh, and the 100 Wh LiPo gives the low-battery warning during the landing (th-avionics). This pack is the one battery.md measured into the bay (148 mm the limit; the 156 mm RED V3 and SkyVolt packs do not fit). About 93 min modelled. Li-ion needs its own voltage parameters (battery.md), and the 230 g extra moves the CG about 9 mm forward. The pack has no thermistor, so the charge node uses the H7 NTC probe taped to the pack, as before. The charge node needs no change for Li-ion (th-eda, Molicel datasheet 80122). 216 Wh is over the 160 Wh airline limit, so it ships as hazmat and cannot fly as luggage. WREKD sells the same pack at the same price. The $165 Snack Packz pack on eBay gives no dimensions, so it may not fit |
Nexus |
| 13a | Charger supply | Mean Well LRS-600-12: 12 V, 50 A, 600 W enclosed switching supply, 90 to 132 or 180 to 264 VAC by switch, fan cooled | 1 | $59.90 | $59.90 | For the ISDT P20 on hand (Ryan, 2026-10-06). The P20 draws 35 A at most from its input, so at 12 V it charges at about 420 W in all: one P50B at 1C (about 280 W with losses), or two at less. A 24 V supply would let it reach its full power. Set the AC input selector before the first power-up, earth the FG terminal, and keep the mains terminals covered. Needs a grounded 3-wire mains cord and a 10 or 12 AWG lead to the P20’s XT60 input | Mouser |
| 13b | Lighting battery | Roaring Top 1300mAh 2S 25C 7.4V, standard LiPo (4.2 V a cell), 9.6 Wh: 69 x 33 x 12 mm, 67 g with wire, plug and case, Deans main lead, JST-XH balance lead | 1 | $10.42 | $10.42 | Chosen (Ryan, 2026-10-06; via th-avionics): the night lights’ own battery, charged from the flight pack only when disarmed by th-eda’s lights node, which also protects and balances it (no BMS). It carries the lights and the 1.25 W always-on rail with about 50 % in hand on a 55 min night survey (docs/lights.md). The only candidate that fits: flat on top of the flight pack under the Hatch Front, riding the slider (th-blender). Change its Deans plug to an XT30 male (row 13c) |
Altitude Hobbies |
| 13c | XT30 connectors | Amass XT30U male and female, 10 pairs, bare | 1 | $11.99 | $11.99 | The male in place of the lighting battery’s Deans plug (row 13b), the female on the lights node’s battery lead; the rest are spares. The same pack as hangar row H2 | Amazon |
| 13d | Night lights, DigiKey parts | The lights boards’ parts (lights option 3, Ryan 2026-10-11: the node and the position-light driver board, th-eda d889b02; the charger board is withdrawn, the lighting battery charging at the dock, th-eda 200b4dc) and the strobe and position light LEDs and housings, for one node board, on th-eda’s dome strobe list (Ryan, 2026-10-10: printed domes over a ring of tilted LEDs in place of the hexagonal pods; hw/lights-node/digikey/, th-avionics 5cd2d35; the belly strobe moved into the keels, six LEDs at 2.2 A, Ryan 2026-10-11): 95 lines, the off-board quantities from th-eda’s off-board list, digikey-upload.csv loads into DigiKey’s BOM manager |
1 | $216.20 | $216.20 | From Part-DB’s DigiKey prices at the time of the run; 0402 and 0603 passives carry two spares each. The 14 Carclo 10415 strobe lenses left with the domes; TDK ACM2520 chokes on the position outputs and Fair-Rite ferrite sleeves on the strobe and landing leads (th-eda, th-telemetry). R23 is the Stackpole RMCF0402FT2M49, and C33, C36 and C39 are the KEMET C0805C105K3RACTU (th-eda dd75a55); R20 and R38 are the Stackpole RMCF0402FT4K70 and RMCF0402FT76K8 in the design, the Yageo parts being out of stock (th-eda b7849c0). Not from DigiKey and not in this price: the Carclo 10508 landing optic (Mouser; the 10509 is out of stock everywhere), the node board itself (th-eda, JLC), the LED stars (Cutter), three SM04B-GHS-TB headers and two XT30PW-M (LCSC or Mouser), GHR-02V-S housings, crimps and wire, the battery (row 13b), and the strobe domes’ clear domes, base discs, foil tape and strobe tab boards, which have no supplier until th-blender settles the dome construction (the “Not from DigiKey” table in lights-bom.md) |
hw/lights-node/digikey/lights-bom.md (DigiKey prices via Part-DB) |
| 13e | Thermal tape, lights | 3M 8810 thermally conductive adhesive transfer tape, 0.25 mm (10 mil), 0.6 W/m·K, 2 in x 3 yd sample roll (3M 7010379057; DigiKey 19-88102”X3YD-ND, not stocked there, about 4 weeks) | 1 | $18.96 | $18.96 | th-eda’s pick (2026-10-10): bonds the twelve strobe facet boards and the three position-light stars to their PETG mounts, and can be reworked. Shelf life 24 months from manufacture | Masterworks |
| 13f | Thermal adhesive, landing light | Arctic Silver ASTA-7G two-part thermal adhesive, 7 g (two 3.5 g syringes), over 7.5 W/m·K, permanent | 1 | $15.95 | $15.95 | th-eda’s pick (2026-10-10): the landing light’s 20 mm three-up star onto its aluminium plate, about 6.5 W of heat. Arctic Silver is phasing it out, so stock is thin: any two-part thermal epoxy at 1 W/m·K or better substitutes, for example MG Chemicals 8329TFF-25ML (DigiKey 473-1367-ND, about $39 to $51, sets in 15 minutes, UL 94 V-0) | AeroCooler |
| 13g | Landing light plate stock | 6061-T6511 aluminium flat bar, 1/8 in (3.175 mm) x 2 in, mill finish, the 4 in length (Remington Industries) | 1 | about $10.00 | $10.00 | ESTIMATE: the listing starts at $8.75 and the 4 in price was not read. For the landing light’s 40 x 40 x 3.175 mm heat plate (th-blender sizes the Landing Light Frame’s pocket for 3.2 mm, 2026-10-10): one cross cut and one rip from the 2 in width give one plate, the 4 in length two (one spare). The plate does not fit at the full 50.8 mm width, because the belly pocket is 40 mm wide | Remington Industries |
| 13h | Keel strobe ferrites | Fair-Rite 2643250302, 43 material shield bead: 6.35 mm OD, 3.18 mm ID (2.95 mm per RS Components), 15.88 mm long, 122 Ω at 100 MHz (DigiKey 1934-1337-ND) | 3 | $0.46 | $1.38 | th-telemetry (6be39f5, 2026-10-11): one on each keel’s strobe pair where it enters the keel, the strobes having moved into the keels (Ryan, 2026-10-11); one spare. The same bead as the charge node’s shoe lead (row H12). Slide it on before the pair is terminated, and check that the twisted pair passes the bore. The common-mode choke at the lights node end is not here: th-eda decides on-board (preferred) or a snap-on core | DigiKey |
| 13l | Keel lighting charge lead ferrites | Fair-Rite 2643001301, 43 material bead: 3.55 mm OD, 1.65 mm ID (minimum), 5.95 mm long, 48 Ω at 100 MHz; RS sells it in 25s | 1 | about $2.00 | $2.00 | th-telemetry (2026-10-11): one on each keel’s 22 AWG lighting charge lead at its sole end (th-blender 11ca1d8: the lead cannot share the main charge lead’s bore), one spare from the 25; 22 AWG ETFE (1.3 mm OD) passes its bore. ESTIMATE: about $0.08 each in 25s, backordered at RS US on 2026-10-11 | RS |
| 13i | Keel strobe heat strips | uxcell 1060 aluminium sheet, 100 x 100 x 1.0 mm, 2 pieces | 1 | $9.99 | $9.99 | th-eda’s spec (EDA cd1c6c7, 2026-10-11): six strips (three a keel since the keel strobes went to six LEDs, th-blender c125cf8), 1.0 mm thick and 3.2 mm wide (th-blender 2c88445, narrowed from 3.8 for the strobe pair’s channel), about 16 mm long flat, bent twice at 90°: one leg under each keel strobe tab’s foot, one up through the motor pad’s slot, one flat under the lift motor’s base. Cut them from th-blender’s flat pattern (2c88445), not from these numbers. th-eda names 1100-H14 or 3003-H14 (5052-H32 acceptable); 1060 is purer than 1100, so it conducts heat better (about 230 W/m·K) and bends at 1 mm without cracking (th-eda confirmed). One sheet makes all six. 1060 is soft: shear it or score and snap it, do not use a cutting disc, and deburr the slot leg so it does not catch in the pad’s 1.3 mm slot | Lowe’s |
| 13j | Thermal paste, keel strobes | ARCTIC MX-4, 4 g syringe (ACTCP00031B), 8.5 W/m·K, silicone-free, not electrically conductive | 1 | $7.99 | $7.99 | th-eda’s pick (2026-10-11): under each heat strip’s flat leg where the lift motor’s base clamps it. The gap is about 0.15 mm, thinner than any gap pad, which would tilt the motor; the motor screws clamp the paste. One tube does all six joints. The strip’s other leg is bonded under the tab’s foot with the thermal adhesive (row 13f) | Walmart |
| 13k | Keel strobe window sheet | Clear polycarbonate sheet, 0.040 in (1.0 mm), cut to about 4 x 4 in (TAP Plastics cut-to-size) | 1 | about $15.00 | $15.00 | ESTIMATE: TAP’s cut-to-size polycarbonate starts at $15. The keel strobe windows are flat pieces cut from this sheet, not printed (a printed window is translucent, and the strobe intensity figures assume clear): four side windows, each a 12 x 21.5 mm slot with 2.5 mm corner radii, and four tail windows on the boss’s aft closure, bevelled where each pair meets on the centre plane (th-blender c125cf8; the outlines are Keel B Strobe Window (clear PETG).stl and Keel B Strobe Tail Windows (clear, placed).stl). Bond each pane flush: at its forward end the skin may stand up to 0.1 mm proud, but at its aft end the pane must be flush or the skin lower, never a forward-facing step; if the bond cannot promise that, chamfer the skin’s edge there lightly after bonding, and the same at the tail windows’ aft edges (th-aeronautics). 1 mm clear PETG sheet also works. The top strobe dome is 0.5 mm sheet, so it is a separate item |
TAP Plastics |
| 14 | Smoke stopper | iFlight XT90 Smart Smoke Stopper, XT90 in and out, 2S to 8S (7 to 29.6 V), trips at 1.2 A, buzzer | 1 | $10.99 | $10.99 | The list recommends one for first power-up. XT90 and 29.6 V to suit the P50B (XT90, 25.2 V full); it replaces an XT60 unit rated 7 to 25 V, which needed an adapter and was under the full-charge voltage (th-aeronautics, 2026-10-04). Bench only: it trips long before the motors spin | iFlight |
| 15 | XT60 pigtails | XT60H male + female, 12AWG | 1 | $11.99 | $11.99 | Pusher ESC tap and a spare | Amazon |
| 15a | XT90-S connectors | Amass XT90-S anti-spark, 5 male (anti-spark) and 5 female plugs with cases | 1 | $14.25 | $14.25 | Ryan approved the anti-spark lead on 2026-10-04 (th-aeronautics, wiring by th-avionics 03f3faf): the XT90-S male goes on the aircraft lead at the re-leaded PM02 V3, and the pack keeps its standard XT90 female (row 13), which mates with it. Its built-in resistor charges the bus capacitance (the 4-in-1’s 470 µF, the pusher ESC, the PM) before the main contacts close, so the plug-in spark does not pit them; pitted contacts heat at the 50 to 75 A transition current. Genuine Amass marks the anti-spark male green. th-avionics, 2026-09-28: re-lead the PM02 V3 with 10 AWG and XT90, because its XT60 and 12 AWG are rated 30 A continuous and 60 A burst, against about 60 to 75 A in transition (estimate). The spare females re-lead the Tattu tuning pack. The XT90 smoke stopper (row 14) plugs in with no adapter. Stock not shown on the page; MPI Hobby sells single genuine pairs at $5.99, in stock on 2026-10-04 | Arris Hobby |
| 15b | 10 AWG wire | Silicone wire, 10 AWG, 3 ft red and 3 ft black | 1 | about $12.00 | $12.00 | ESTIMATE, price not shown. The PM02 V3’s new leads and the pack-side main feed to the junction where the pod and pusher feeds split. Log battery current on every flight: events above 80 A move us to th-eda’s 200 A sense module, which replaces the PM02 with no harness change | Amazon |
| 16 | Telemetry radio | Seeed Wio-SX1262 with XIAO ESP32S3 kit (SKU 102010611): ESP32-S3 board and SX1262 LoRa module, 915 MHz, IPEX antenna connector (very likely MHF1: check on receipt) | 5 | $10.99 | $54.95 | Order this kit SKU only. Its radio board joins the XIAO ESP32-S3 by a 30-pin board-to-board connector and is sold only in the kit. The pin-header “Wio-SX1262 for XIAO” (SKU 113010003, also in the XIAO nRF52840 kit 102010710) has the same radio on different pins: it does not fit our cradle, firmware or the first-flight card (Seeed wiki). Ryan, 2026-10-04: in place of the Holybro SiK V3 pair, with our own mLRS firmware target (MAVLink, th-telemetry). Aircraft 1 (the LUX F765’s TELEM2, SERIAL2; RTK corrections and SITE_CLR), dock 1, a bench pair for firmware work while the flight pair is fitted, 1 spare. The keel coil and its RG316 run stay: the aircraft module plugs in through a U.FL-to-SMA bulkhead pigtail (row 16e), and the run’s SMA plug (row 16k) screws onto it (joint J2 is SMA, Ryan 2026-10-06). Fit the 100 uF capacitor (row 16i) at its 5 V input. Keep each kit’s WiFi antenna in the bag: on the flight and dock units the ESP32’s 2.4 GHz radio stays off and its U.FL stays empty, because ELRS shares the band (th-telemetry); the kit’s FPC LoRa antenna is for the bench pair only | Seeed |
| 16a | Keel coax | RG316, 50 Ω, 2.5 mm, per foot | 15 ft | $0.90 | $13.50 | With the two-radio LoRa diversity (Ryan, 2026-10-06) four runs, about 2.0 m fitted: 15 ft leaves room for one mis-cut run (th-telemetry). The keel antenna runs enter the kit’s 25 x 10 mm motor-wire channel at its mouth in the Fuselage Mid stub (Y 283 to 308), not a drilled exit (the drill is cancelled: its bore is the wing spar’s, th-blender 2026-10-10). Cut lengths, from the radio end to the MMCX plug just under each keel’s root exit (bay / channel / spar / keel, tools/wire_routes.py --keel-coax, ESTIMATE): the port LoRa run 602 mm (220 / 130 / 190 / 62) from the single J2 on the port bay wall; the starboard LoRa run 435 mm (53 / 130 / 190 / 62) from its SMA plug and 50 Ω load on the Fuselage Front’s starboard wall over the front spar (the J2 split, Ryan 2026-10-09; th-blender 63fa640; no added length: the beacon board’s J2 goes at the parked plug, th-telemetry); ELRS A 494.5 mm (starboard) and ELRS B 498.8 mm (port), to the RP4TD’s SMA brackets on the Spar Box walls (th-blender W121 08799381, ESTIMATE; was 458 and 550 mm to a bracket beside the receiver). A pair balances on loss, not length: the LoRa pair stays within about 350 mm and the ELRS pair within about 210 mm, and no run gets extra coax to match (th-telemetry ace46c7). th-blender’s drawn routes govern where they differ. Plus spare (cut lengths of record: docs/agent-threads/harness/). Threading order: fit the radio-end connector first, push the bare keel end from the bay through the channel mouth, the channel, the spar and the 3.2 x 6 mm root slot, and crimp the keel-end MMCX plug last, at the cassette. No connector passes the spar slot or the root slot |
JEFA Tech |
| 16c | Coax connectors, MMCX plugs | POBADY MMCX straight crimp plug (male) for RG174 and RG316, 10 per pack: gold-plated brass, DC to 6 GHz, VSWR 1.15 or less; knurled body 4.0 mm across and 18 mm long, crimp ferrule 2.9 mm OD with a 1.5 mm bore (measured by Ryan; th-telemetry) | 6 | on hand | $0.00 | Ryan’s own, in place of the Amphenol 262101 (th-telemetry, 2026-10-22): 4 fitted, at the keel end of each of the four runs, where the plug snaps onto the cassette jack (16d, joint J1); 2 spare, because crimps fail. Since J2 became SMA (Ryan, 2026-10-06) the radio ends take 16k instead. With the 18 mm body, check the mated pair’s length against the cassette bay before keel B prints | Amazon B0D9BBDSYX |
| 16d | Cassette jacks | Fairview Microwave FMCN5182, MMCX jack (female), end launch, solder, 50 Ω, DC to 6 GHz, brass body and phosphor-bronze contact, gold over nickel over copper, 7.62 mm long (Fairview datasheet; files in vendor/fairview-fmcn5182/) | 6 | on hand | $0.00 | Ryan bought these for the cassette’s antenna side in place of the Amphenol 262141 (th-telemetry, 2026-10-22). Two per keel, both keels (ELRS and LoRa in each cassette): 4 fitted and 2 spares. The RG316 run’s MMCX crimp plug (row 16c) snaps onto it. The jack and a mated plug set the 5 mm cassette bay, so measure both with calipers before keel B (g17, g18) prints. The 262141 (Mouser 523-262141, $28.63 on 2026-10-20) is no longer needed | Fairview |
| 16g | Cassette pigtail coax | 1.13 mm micro-coax, 50 Ω | 0.5 m | about $5.00 | $5.00 | ESTIMATE. From each cassette jack to its element’s feed: ELRS 72 mm, SiK 30 mm (th-blender’s cassette draft, c82a475); about 0.25 dB at 2.4 GHz over 72 mm (estimate) | part to choose |
| 16h | Pigtail ferrite beads | Fair-Rite 2661000101, 61 material shield bead: 3.51 mm OD, 1.30 mm ID, 3.25 mm long, 30 Ω at 100 MHz (DigiKey 1934-1467-ND) | 18 | $0.086 | $1.55 | 12 are fitted, so 18 leaves 6 spare: five at the feed of each ELRS pigtail (16.3 mm, in the 20 mm pocket) and one on each LoRa pigtail, in place of two, which makes room for the feed joint in th-blender’s cassette with no loss in NEC2 (th-telemetry 36af7f0, 2026-10-11; was two each, 14 fitted, from 2026-09-28 and the two-radio LoRa diversity, Ryan 2026-10-06). Each bead is 3.25 mm long; a bead’s impedance on a wire scales with length × ln(OD/ID), so one real bead is about 70 % of the modelled choke, and five give about 117 %. No NEC2 re-run | DigiKey |
| 16e | U.FL to SMA pigtails | Taoglas CAB.011: right-angle U.FL (I-PEX MHF1) plug to SMA female (jack) bulkhead, REAR mount (its 8 mm hex body at the cable end), standard SMA, 95 mm of 1.13 mm coax, DC to 6 GHz; 1/4-36 UNS-2A thread 11.4 mm long with a D-flat 5.8 mm across (+0.1/-0.2); one hex 8 nut 1.8 mm thick and a star lock washer in the bag; DigiKey 931-1098-ND | 4 | $4.44 | $17.76 | 2 fitted (the two RP4TD ports) and 2 spare: U.FL pigtails are fragile (th-telemetry, 2026-10-06). The nose LoRa radio’s pigtail at J2 is row 16p since 2026-10-20. The beacon carrier’s jack A on Fuselage C is not fitted until the beacon board’s radio A flies (th-telemetry, 2026-10-11); its pigtail is row 16o. RG316 is too thick for a U.FL, so each RP4TD port and each LoRa module gets this pigtail, and the run’s SMA plug (16k) screws onto its bulkhead jack in a printed bracket: the single J2 on the port bay wall for LoRa, beside the receiver for ELRS. J2 is rear mount (Ryan, 2026-10-07, in place of the front-mount TE Linx CSI-SGFE-100-UFFR at $9.12): the thread goes through the bracket from the radio side, keyed by the D-flat in the bracket’s D-hole, and the washer and the one nut go on the plug side. One length, 95 mm, serves both (LoRa wants 80 to 100 mm, ELRS 50 to 100 mm). Order it with follow-on row 16o (the beacon board’s four), 10 in all, at $3.77 each. $4.44 single, 43,680 in stock at DigiKey (th-eda’s Part-DB lookup, 2026-10-07) | DigiKey, datasheet |
| 16p | U.FL to SMA pigtail, nose radio at J2 | Taoglas CAB.697: right-angle U.FL (I-PEX MHF1) plug to SMA female (jack) bulkhead, rear mount, standard SMA, 100 ±3 mm of 1.13 mm coax; 1/4-36 UNS-2A thread 11.4 mm long with a D-flat 5.8 ±0.08 mm (Taoglas drawing); Mouser 960-CAB.697 | 2 | $5.14 | $10.28 | 1 fitted and 1 spare. In place of CAB.011 (row 16e) for the nose radio’s lead to J2 (th-telemetry 58575840): CAB.011’s 95 mm is too short for the W113 route, and the vendors’ minimum bend for 1.13 mm coax (6.8 mm on this sheet) rules out a tighter route. The radio cradle lid’s channel holds it at the 6.8 mm minimum (th-blender W119 9baa3f8b). Measure two things on arrival: the sleeve behind the hex (“17 REF” on the sheet, no tolerance; 17.7 mm or less keeps the J2 arc at 6.8 mm or more) and the hex across flats (drawn but not sized; the J2 bracket’s pocket is 8.3 mm across flats, 2.0 mm deep). Mouser had 451 on 2026-10-19 (Part-DB’s Mouser provider) | Mouser |
| 16f | VNA adapter | MMCX to SMA adapter or pigtail | 1 | about $6.00 | $6.00 | ESTIMATE. So the NanoVNA-F V3 can tune each cassette on the bench | part to choose |
| 16i | LoRa supply capacitor | 100 uF low-ESR electrolytic, 10 V or more | 1 | about $0.50 | $0.50 | ESTIMATE, part to choose. At the aircraft LoRa board’s 5 V input, against its transmit bursts of about 125 mA (th-telemetry) | part to choose |
| 16j | Starboard LoRa run termination | Amphenol RF 132361, SMA female (jack) 50 Ω terminator, standard SMA, brass body, gold-plated contact, DC to 18 GHz | 1 | $6.48 | $6.48 | Until th-eda’s two-radio beacon board flies, the starboard LoRa coil and its RG316 run are fitted with no radio. Since J2 became SMA (Ryan, 2026-10-06) the run’s radio end is an SMA plug (16k), so this female load screws straight onto it and the coil is not an unterminated parasitic beside ELRS A (th-telemetry, 2026-10-06). The run goes in now, while the spar is open: the bore is too full to thread it later. Nothing transmits into the load, so its power rating does not matter. Replaces the MMCX male load (Field Components FC37M-TERM1-1) and the MMCX female-female adapter (FC37F-FC37F-1), which are dropped from the cart. Price is DigiKey’s quantity-1 listing (ACX1252-ND), read through search; confirm at checkout | DigiKey |
| 16k | Coax connectors, SMA plugs | Amphenol RF 132114, SMA straight crimp plug (male), standard SMA (not RP-SMA), for RG316, RG174 and LMR-100A (cable group B), gold-plated brass, solder centre pin, crimp ferrule, DC to 18 GHz | 6 | $5.90 | $35.40 | Joint J2 (Ryan, 2026-10-06): 4 fitted, at the radio end of each of the four runs, screwed onto the female bulkhead jack of the radio’s pigtail (16e); 2 spare. Fit this end first, before the run is threaded (row 16a). Tighten by hand, then to about 0.6 N·m with the SMA torque wrench | RF Parts |
| 16l | SMA torque wrench | KCR Products KCR-3125S, 5/16 in, preset 5 in-lb (0.56 N·m), standard length | 1 | $27.50 | $27.50 | Joint J2: hand-tight, then about 0.6 N·m (Ryan, 2026-10-06; brass SMA wants 4 to 5 in-lb, 8 in-lb is for stainless). Order the 5 in-lb option, not the 8 in-lb default | Kendon Park |
| 16m | ELRS dipole wire | Remington Industries bare copper buss wire, 18 AWG (1.02 mm), solid, natural, the smallest spool | 1 | $15.41 | $15.41 | ESTIMATE price from Remington’s listing (spool sizes from $15.41), confirm at checkout. The ELRS dipoles in the keel cassettes: two arms of 1.0 mm wire, 27.4 mm each with 2 mm trim allowance (docs/keel.md); four dipoles with spares take under 0.5 m (th-telemetry, 2026-10-07). Copper, not brass: brass conducts about a quarter as well |
Remington |
| 16n | LoRa coil wire | Remington Industries magnet wire, 20 AWG enamelled copper (0.81 mm bare, 0.85 mm over the enamel), 155 °C solderable polyurethane with polyamide overcoat (NEMA MW 80-C), 2 oz, 40 ft (20SNSP.125) | 1 | $13.19 | $13.19 | ESTIMATE price from Remington’s listing, confirm at checkout. The LoRa coils on the SiK Coil Formers: 0.8 mm wire, about 4.8 turns per arm on the 4.4 mm former, about 0.2 m per coil with its leads; two coils and spares take about 1 m (th-telemetry, docs/keel.md). The enamel solders through, so the ends need no scraping |
Remington |
| 17 | CG trim slider: servo | FeeTech FS90R continuous-rotation micro servo (Adafruit 2442): 4.8 to 6 V, 1.3 to 1.5 kg·cm, 10 g | 2 | $7.50 | $15.00 | Moves the pack through the rack. One spare, since it is cheap and a stripped gear grounds the slider. The generator’s envelope is FS90R-size (ESTIMATE): measure the real one before th-blender’s tower prints | Adafruit |
| 17a | Slider sensor MCU | Seeed Studio XIAO RP2040 (102010428) | 1 | $4.90 | $4.90 | Reads the AS5600 and the VTX NTC, and sends $CGP and $CGT to the LUX on USART6 | DigiKey |
| 17b | Angle sensor | Adafruit AS5600 Magnetic Angle Sensor, STEMMA QT (6357): I2C, no magnet | 1 | $5.95 | $5.95 | Reads the pinion’s angle over I2C, as the firmware does. th-eda’s choice (2026-10-06): 25.4 x 17.78 mm, the chip at the board centre, four 2.5 mm holes on 20.32 x 12.70 mm; th-blender draws the slider’s sensor shelf to it | Adafruit |
| 17c | Sensor magnet | 6 x 2.5 mm diametric NdFeB disc, N35SH, 140 °C (mjbots) | 2 | $2.00 | $4.00 | In the pinion’s top pocket (6.2 x 2.6). It must be DIAMETRIC: an axial magnet gives nonsense angles. One spare. ams’s own part is AS5000-MD6H-2 (same size; no US price found). mjbots showed a “sold out” variant note: check stock | mjbots |
| 17d | Pinion and sensor screws | Essentra 50M020040P006: M2 x 6 nylon pan head, Phillips, 4.0 mm head | 5 | $0.116 | $0.58 | One holds the pinion on the servo spline, and two hold the AS5600 board chip-down under its shelf (th-blender b86402b); two spare. Nylon, because it sits about 1 mm under the AS5600 magnet; a steel screw distorts the field. Its 4.0 mm head is the generator’s HUB screw_head. It is a machine screw into the servo’s plastic output shaft (made for a self-tapping horn screw): check the bite on the bench | DigiKey |
| 17e | VTX temperature sensor | TDK B57861S0103F040 NTC, 10 k, B25/100 3988 K, glass bead, 40 mm leads (DigiKey 495-2142-ND; LCSC C209896) | 1 | $1.97 | $1.97 | On the Walksnail VTX heatsink, to the slider XIAO’s A0 (th-eda) | DigiKey |
| 17f | NTC divider resistor | Yageo RC0603FR-0710KL, 10 k 1 % 0603 (DigiKey 311-10.0KHRCT-ND) | 10 | $0.025 | $0.25 | 3.3 V to A0. Ten, because an 0603 is easy to lose | DigiKey |
| 17g | NTC filter capacitor | Samsung CL10B104KB8NNNC, 100 nF 50 V X7R 0603 (DigiKey 1276-1000-1-ND) | 10 | $0.027 | $0.27 | A0 to GND, and one on each servo-rail divider into A1 and A2 (row 17k) | DigiKey |
| 17h | Sensor wire | Adafruit silicone stranded-core wire, 30 AWG, 2 m, 0.8 mm OD (2051), in red, black, yellow, white | 4 | $0.75 | $3.00 | The NTC twisted pair (about 250 mm, VTX on the camera arm to the servo tower, th-blender’s route; wire_routes to confirm), the XIAO to the LUX’s USART6 pads and 5 V (TX, RX, GND, 5 V, about 150 mm), and the AS5600 to the XIAO (four short leads). Nothing in the cart covers it: H15 (28 AWG) is in the follow-on harness table and not chosen | Adafruit |
| 17i | NTC bond | High-temperature polyimide (Kapton) tape, 1 cm x 33 m (Adafruit 3057) | 1 | $4.95 | $4.95 | Holds the NTC bead flat on the VTX heatsink, no glue. Adafruit’s category page showed it out of stock: any 1 cm polyimide tape serves | Adafruit |
| 17j | Slider joint screws | Essentra 50M020040H004B: M2 x 4 nylon 6/6 flat countersunk head, 90°, slotted drive, 4.0 mm head, 1.2 mm head height, bag | 10 | $0.12 | $1.20 | ESTIMATE price: Mouser’s listing read through an aggregator (datasheetarchive.com), not on Mouser’s own page; confirm at checkout. Three join the CG Slider (port) and (starboard) halves at the lap joints at Y 95, 153 and 215, one each, from the top, flush in 4.2 mm countersinks, with 0.5 mm under the tips in the tray’s channels (th-blender a4aca09; Ryan, 2026-10-06: the slider must come out with the tray bonded in). Seven spare, because a nylon slot strips easily and the slider comes out often. Nylon, as row 17d, to keep steel away from the AS5600. Drive with a 2 mm flat blade, by hand | Mouser, Essentra |
| 17k | Servo-rail divider resistors | Yageo RC0603FR-0722KL, 22 k 1 % 0603 | 10 | about $0.03 | $0.30 | ESTIMATE price. Two dividers, one per Micro BEC (row 9j), into the slider XIAO’s A1 and A2: each is two of these with one 100 nF from row 17g, from that BEC’s 5 V output, so the flight log shows each servo’s supply (th-eda, EDA 5b6cb9d) (th-avionics, firmware f16e0a2). Ten, because an 0603 is easy to lose | part to confirm at DigiKey |
| 18 | LUX USB-C pigtail | USB-C 2.0 flat ribbon (FPC) extension, USB-C male plug with a side-angle (left or right) 90 degree bend to a straight USB-C female socket, about 10 cm: part to choose. Most listings are up/down-angle, which does not fit: the LY tech B07PMSYWLC first listed here is up-angle. Look in ADT-Link’s T1C-series heads or a modular FPC kit with a side-angle head | 1 | about $8.00 | $8.00 | ESTIMATE price and part to choose. On Fuselage C v2.4B it stays plugged into the LUX F765’s USB-C port, tied at the Lucid Edge Bracket’s bridge, with its socket under the Front’s aft roof, so the LUX is reached without lifting the Lucid: its plug goes in only with the Lucid off its bosses (th-blender W36, W43). The LUX’s USB-C opens forward on its forward edge, about 8 mm behind the Lucid bracket, and the route turns to port at Z 19 in the board’s plane: a straight plug does not fit, an up bend meets the Q43 copper floor’s no-lead zone above Z 26.7, and a down bend meets the plate and the spar (th-avionics, 2026-10-15). USB-C is reversible, so left and right are the same part. th-blender’s route is about 58 mm (ESTIMATE); 5 cm is too short, so 10 cm, and the ribbon folds the extra. Dry-fit it on the real LUX in the printed plate before you tie it: the board positions come from retail photos (ESTIMATE). USB 2.0 is all ArduPilot needs | ADT-Link (example of the family) |
| Total | $1,952.52 | The lift ESC is the 4-in-1, rows 2b to 2e, and row 2g is the video unit’s capacitor; the S50 pod ESCs are the follow-on below. Above the list’s $450 to $980 band. RTK adds $286: the GPS is $227 more and the radio pair is new. The DroneCAN airspeed sensor adds $51 |
Follow-on: DroneCAN
Section titled “Follow-on: DroneCAN”The DroneCAN parts wait until CAN is on: the LUX F765 that flies first has one CAN port, off for the first flights (Ryan, 2026-10-04), and our own flight controller has two buses. The pod path also stays for after the first flights (Ryan, 2026-09-29): one AM32 ESC at each lift motor pod, first the bought S50s, then th-eda’s pod board one pod at a time. The airframe takes either layout. Not in the cart total; order when our flight controller is ready.
| # | Item | Product | Qty | Unit | Line | Why | Link |
|---|---|---|---|---|---|---|---|
| 2 | Lift ESCs, DroneCAN pods | Vimdrones AM32 DroneCAN ESC S50, CAN version, one at each lift motor pod, plus one bench spare | 5 (a 4-set and a single) | $20.58 | $102.90 | Decided 2026-09-27 (Ryan): fly first on bought S50s, then fit our own pod board (th-eda, a 34 mm disc over the pod pad, 2026-10-05, with the same STM32L431 and AM32 DroneCAN firmware) one pod at a time, so a design fault shows on one motor, not four. The shop’s data shows every option available, but stock versus backorder shows only at checkout. Check the cradle’s hold-down holes, the CAN connector and the capacitor space against the S50 (th-blender). Each needs about 2 x 220 uF low-ESR at the board for its 0.4 m DC leads; the shop’s add-on pack is row 2a. The 4-in-1, row 2b, flies first (Ryan, 2026-09-29) | Vimdrones |
| 2a | ESC bulk capacitors | Vimdrones add-on: 6 x 470 uF 35 V Panasonic electrolytic | 1 pack | $6.50 | $6.50 | One at each of the 4 pods (the S50 cradle’s bay takes a 10 x 20 mm can), one at the video unit’s 12 V input, one spare. Order with row 2 | Vimdrones |
| 11 | Airspeed, DroneCAN | Holybro High Precision DroneCAN Airspeed Sensor, DLVR-L10D board plus PT40 pitot tube | 1 | $110.99 | $110.99 | For when CAN is on: the LUX F765’s one CAN port stays off for the first flights, and our own flight controller follows (2026-10-04). Named DLVR class, on CAN (decided 2026-09-26). L10D is the 2500 Pa range for a 16 to 30 m/s aircraft. Board $103.00 and pitot $7.99 bought separately because the bundle is sold out | Holybro |
| 16o | Beacon board LoRa pigtails | Taoglas CAB.011: SMA female (jack) bulkhead, rear mount (front nut), standard SMA, to a right-angle U.FL (I-PEX MHF1) plug, 95 mm of 1.13 mm coax; DigiKey 931-1098-ND | 4 | $4.44 | $17.76 | For th-eda’s two-radio beacon board (boards/beacon, two Seeed Wio-SX1262 LoRa modules since EDA 99ed160, in place of the HT-RA62; both have a first-generation IPEX, so the pigtail is unchanged), from each radio to its single SMA bulkhead on the spar-box walls (th-eda with th-blender, 2026-10-07): two per airframe and two spares, because a U.FL takes about 30 matings. $3.77 each at 10. Active, 43,680 in stock (th-eda, Part-DB DigiKey lookup, 2026-10-07). Order with the beacon board, not for the first flights On Fuselage C, when radio A flies and jack A is fitted, sleeve its RG316 where it crosses the port lift phase bundle at X 107.8 with about 15 mm of PTFE tube or two layers of heat shrink: the two stack 9.5 mm in the 10 mm channel and would rub. Keep every coax clipped in the channels so it cannot saw along a phase bundle (th-telemetry, th-blender 64f720e). | DigiKey |
One airframe
Section titled “One airframe”Ryan builds one airframe, on Fuselage B with the spine (2026-10-04); the earlier plan for two airframes sharing one
set of electronics is withdrawn. Its parameters are in params/airframe.parm (th-avionics, 8d40d05).
RTK notes
Section titled “RTK notes”- The LC29H base in the hangar sends RTCM 3 MSM corrections for GPS L1/L5, Galileo E1/E5a, BDS
B1I/B2a and QZSS L1/L5. The Q43 NEO-F9P tracks all of these signals. Test the pair on the bench
before the first flight: corrections between two receiver makes usually work, but we have not
tested this pair. Before mapping, give the base one fixed coordinate (see
site-mapping.md). - The corrections reach the aircraft over the LoRa telemetry link (row 16: Seeed XIAO ESP32-S3 +
Wio-SX1262 at both ends, our own mLRS target, Ryan 2026-10-04) as MAVLink
GPS_RTCM_DATA. Use mLRS at 31 Hz, 2,000 B/s up and 2,562 B/s down, on all of 902 to 928 MHz (25 hops; a 915 to 928 MHz slice has too few channels for FCC 15.247). Keep MAVLink 2 on every hop. - Send MSM4, not MSM7. MSM4 for 20 to 28 satellites is about 400 to 520 B/s (about 580 B/s in
GPS_RTCM_DATAfragments). WithSITE_CLR, the wind values and heartbeats, the uplink carries about 790 B/s, 39 % of the 31 Hz mode. MSM7 does not fit at 19 Hz. The research and arithmetic are indocs/agent-threads/gps/20261004T130536Z-th-telemetry-lora-link-research.md. - Mission uploads and terrain data share the uplink with RTCM, with no priority. Upload missions on the ground and preload terrain on the FC’s microSD card.
- The ELRS link stays the RC link only. Its MAVLink uplink is too small for RTCM.
- The ESP32-S3 has a 2.4 GHz WiFi and Bluetooth radio. Our firmware keeps it off at both ends, because ELRS uses the same band. Do not fit an antenna on its U.FL connector.
- ArduPilot does not configure the constellations of the F9P. Set them one time with u-center.
GPS L5 is not yet operational, so the satellites mark it unhealthy. Set bit 5 of
GPS_DRV_OPTIONS(value 32), or the F9P ignores L5. - UARTs: see
params/board-luxf765.parmfor the LUX F765 that flies first; on the NxtPX4v2 alternative (params/board-nxtpx4v2.parm, th-avionics):SERIAL1the telemetry radio (the SiK until 2026-10-04, the LoRa link since),SERIAL2the ELRS receiver (CRSF) on J5 (UART4: 5 V, TX, RX), not the RC connector J7,SERIAL3the Waveshare Q43 NEO-F9P on J3 (UBX; its QMC5883P compass on J3’s I2C, the board’s only external I2C, because the board has no compass),SERIAL4the HD VTX connector (DisplayPort OSD),SERIAL5the 4-in-1’s telemetry on the ESC lead (J6, UART7 RX).SERIAL6andSERIAL7are not used. The MS4525DO airspeed sensor shares J3’s I2C through the Y lead (row 9f). The 6C Mini’s port map is inelectronics-notes.md. - The GPS sits under the radome of Fuselage Mid B, on a keyed seat that comes out through the
payload bay (
antennas.md). Keep it away from the carbon spars, the ESCs and the power wiring. The printed airframe is radio-transparent. Carbon is not. When the GPS and the camera are both fitted, measure the antenna-to-lens offset for the mapping software.
Sources, 2026-09-28: Waveshare Q43 NEO-F9P documentation, Waveshare Q43 NEO-F9P product page. Sources, 2026-09-26: Holybro SiK Telemetry Radio V3, PX4 RTK bandwidth figures, Emlid ArduPilot integration, ExpressLRS MAVLink.
4S alternate (swap these; ESCs already cover 4S)
Section titled “4S alternate (swap these; ESCs already cover 4S)”Keep the 10x7 pusher prop. With the 900KV on 4S a 10-inch prop draws about 40 A at full throttle on a fresh pack, where the 12x8 would draw 64 A (Propulsion).
| Item | Product | Qty | Line | Link |
|---|---|---|---|---|
| Lift motors | EMAX ECO II 2807 1700KV, “19x19mm mounting” | 4 | $77.96 | EMAX USA |
| Pusher motor | T-Motor VELOX V3115 900KV | 1 | $62.99 | RDQ |
| Battery | CNHL G+Plus 6000mAh 4S 70C (ships XT90, re-terminate) | 1 | $60.99 | CNHL |
| Saving against the 6S cart | $143.20 less | The 6S cart with its lift motors, pusher motor and pack replaced by these three |
Optional
Section titled “Optional”| Item | Product | Price | Link |
|---|---|---|---|
| FPV, earlier choice | RunCam WiFiLink 2 (OpenIPC, IMX415, M12 lens, 5.8 GHz), primary from 2026-09-28 until the Walksnail (2026-10-06) | $129.99 | RunCam |
| FPV lens | Commonlands CIL292-F2.1-M12A650: 2.3 mm f-theta M12, F/2.1, 7.1 mm image circle, 650 nm IR-cut, 3 g; replaces the WiFiLink 2’s stock lens (Ryan, 2026-09-28) | $59.00 | Commonlands |
| FPV, alternative | DJI O4 Air Unit (non-Pro) | $139.99 | RDQ |
| Radio | RadioMaster Boxer, fan-cooled internal ELRS module, FCC version (ships at 100 mW; the LBT version is limited to 100 mW) | $153.99 | RadioMaster |
| Tuning LiPo | Tattu 4500mAh 6S 25C XT60 (the cart’s pack until 2026-10-01): 100 Wh, 0.67 kg. Fits by design. Re-lead to XT90 (row 15a). battery.md suggests one LiPo for tuning flights that want more hover power |
$108.19 | Tattu |
| FPV goggles | Walksnail Avatar HD Goggles L, chosen 2026-10-06 (Ryan): one 4.5 inch 1080p 60 Hz screen, built-in LHCP patch antennas (4.9 dBi, which match the kit’s LHCP), head tracking, 350 g. They take an external battery, 6 to 25.2 V on a 5.5 x 2.1 mm DC plug | $199.99 | Caddx |
| Debug adapter, for our FC | Holybro Pixhawk Debug Adapter (#18073): adapter board with an FTDI serial port, JST SH 6-pin and 10-pin cables, USB-C cable. Off the first build (th-avionics, 2026-10-04): it carried the 4-in-1’s telemetry into the 6C Mini’s FMU Debug port; on the LUX F765 that telemetry goes to pin 4 of the ESC #1 port. Useful later on our own FC’s SWD and debug header | $17.99 | Holybro |
| Folding pusher blades, later | Aeronaut CAM Carbon Z 10x7 folding blades, pair (8 mm root, rated on a 42 mm yoke). The folding prop dropped for now (Ryan, 2026-10-04); for a later build, with the yoke and bushing below and printed stop plates. Right-hand blades (Aeronaut 723931) with the pusher reversed in the ESC; left-hand 722430 for a motor in its normal direction | $12.20 a blade, 2 pairs | Aloft Hobbies |
| Folding yoke, later | Aeronaut 7242/22 aluminium centre piece, 42 mm, 0°, 8 mm bore. A bare yoke, not a spinner; needs printed front and rear stop plates | $10.00 | Graves RC |
| Folding prop bushing, later | Aluminium bushing, 5 mm bore, 8 mm outside, over the motor’s M5 thread; the yoke and both stop plates sit on it and the motor’s flange nut clamps the stack against the bell. ESTIMATE, part not chosen; check that the stack fits the motor’s shaft | about $2.50 | ItsQv guide |
| Flight controller, alternative | MicoAir HKUST NxtPX4v2: STM32H743, two BMI088 IMUs, no CAN, 20 x 20 mm; the checked alternative to the LUX F765 (params/board-nxtpx4v2.parm, its plate Avionics Plate (NxtPX4v2) in parts/) |
$71.99 | Pyrodrone |
| Rangefinder, optional payload | Benewake TFmini-S micro lidar: 0.1 to 12 m, UART or I2C, 4.5 to 6 V. Off the first build (Ryan, 2026-10-05: RTK and the barometer carry the height); it rides lens-down on the Payload Cover (TFmini-S) in a seat pad under a clip-on PETG lid, with a TPU lens bezel (th-blender, 0900ef8 and e446ad8), in place of the Survey3W cover. A 4-wire lead with a plug at the cover (so the cover comes off) goes to the LUX F765’s GPS2 pads on the underside (UART8: TX8, RX8, 5 V, GND); cross TX and RX, the sensor’s TX to RX8 (th-avionics, params/payload-tfmini.parm, 635b2a3). The lead leaves the board flat to port, with Kapton over the outer pad row it crosses, and cuts to about 120 mm plus enough slack to lift the cover with the plug still in (wire_routes.py --tfmini-cover) |
$43.90, in stock 2026-10-05 | SensorLidar |
| FPV, first build | Caddx Walksnail Avatar HD Moonlight kit (WN11-4K14B): Micro camera and 4K VTX, two LHCP antennas, 38.5 g without the antennas, 7.4 to 25.2 V, chosen 2026-10-06 (Ryan; see FPV above) | $226.99 | RDQ |
| FPV spectator receiver | Walksnail Avatar VRX (WN02-FP001): HDMI out, LHCP antennas, 7 to 25.2 V, microSD recording. A second viewer: turn Broadcast on in the Goggles L, and the VRX takes the aircraft’s VTX ID from its Share list. Keep the goggles, VRX and VTX on current firmware (sharing came in 33.39.10). Not the Ascent VRX (WN02-FP018): Ascent does not work with Avatar HD cameras such as the Moonlight | $219.00 | Caddx |
| BEC, earlier choice | Matek BEC12S-PRO, 9 to 55 V in, 5 A (9 A peak), jumper-set 5.2, 8 or 12 V, 5 g: planned at 12 V for the video and at 5.2 V for the elevon servos (rows 9h and 9b until 2026-10-08), in place of which the two Matek Power Hubs fly. Rated to 55 V, so it is the part to come back to if a hub suffers from the brake spikes | $18.99 each | ProgressiveRC |
Lighting battery candidates not chosen. Ryan chose the Roaring Top 1300 mAh 2S (row 13b, 2026-10-06), because it is the only candidate that fits (th-blender: flat on top of the flight pack under the Hatch Front, on the slider). Kept for reference: the Tattu 800mAh 2S 45C (TAA8002S45X3, 52 x 30.5 x 15 mm, 43 to 50 g ESTIMATE, $10.29, GensTattu) and the GNB 850mAh 2S LiHV long pack (GNB8502S60AHV, 18 x 18 x 70 mm, 38 g, $12.99, Pyrodrone). Ryan’s Zeee 2S 5000 (210 g) gives less coverage than no lighting battery, and the Tattu 2S 300 HV (2.3 Wh) is too small for a night survey.
Bench test kit, candidates
Section titled “Bench test kit, candidates”For radio bench test 7 (receiver desense, docs/antennas.md), so the fix is on the bench the day the
test runs (th-telemetry, th-avionics, 2026-09-28). Bench diagnostics in steel SMA bodies, not flight
parts. Not in any total, and not ordered until Ryan confirms. If test 7 shows desense, th-telemetry then
chooses small filters for flight.
| Item | Product | Qty | Price | Why | Link |
|---|---|---|---|---|---|
| ELRS band-pass filter | Mini-Circuits VBFZ-2340-S+, SMA inline, 2020 to 2660 MHz; about 1.5 dB loss in band, about 55 dB rejection near 1 GHz (typical) | 2 | to confirm at DigiKey | One in each ELRS antenna lead, to keep the SiK’s 915 MHz out of the RP4TD | datasheet |
| SiK low-pass filter | Mini-Circuits VLF-1000+, SMA inline, DC to 1000 MHz, under 1 dB loss, stopband 1900 to 5000 MHz | 1 | to confirm; Mouser showed 2 in stock and a 22-week lead | In the SiK antenna lead; also cuts the SiK’s 1830 MHz second harmonic. Fallback: VLF-1200+, $40.57 at DigiKey, once its rejection at 2.4 GHz is checked | datasheet |
| U.FL to SMA pigtails | U.FL (IPEX1) to SMA, about 10 cm | 4 | to confirm | The RP4TD’s antenna ports are U.FL: one pigtail each side of each ELRS filter | |
| RP-SMA to SMA adapters | RP-SMA male to SMA female, and SMA male to RP-SMA female | 1 each | to confirm | The SiK V3 has an RP-SMA female antenna port (Holybro): one adapter each side of the SiK filter | |
| mLRS reference receiver | MATEKSYS MR900-30: mLRS MAVLink 900 MHz receiver, SX1262 with SE2435L PA, 29 x 21 mm, 5.3 g; IPEX MHF1 dipole and heatsink in the box | 1 | $27.99, in stock 2026-10-04 | Approved by Ryan (2026-10-04, th-telemetry): the Rx end of a known-good mLRS link, to test each XIAO firmware build against proven hardware, and a fallback link. Set it to 20 dBm (the mLRS default; the US cap is 24 dBm with 25 hops); its levels are 17, 20, 24, 27 and 30 dBm, whatever the “30” in the name (th-telemetry, 2a315e6). Antenna connector IPEX MHF1 | RaceDayQuads |
| mLRS reference transmitter | MATEKSYS mR900-30-TX module kit: the mR900-30-TX module, IPEX MHF1 to SMA and RP-SMA cables and a 5 V fan; no SMA antenna or case | 1 | $41.99, in stock 2026-10-04 | Approved by Ryan (2026-10-04, th-telemetry): the Tx end of the reference pair, at the site node on the bench. Its MAVLink input is LT1/LR1 or USB-C, not the Tx1/Rx1 the silkscreen suggests. Set it to 20 dBm (the mLRS default; the US cap is 24 dBm with 25 hops). Not compatible with ELRS 900 MHz or FrSky R9 | RaceDayQuads |
| mLRS reference antenna | ALFA Network ARS-915P: 915 MHz half-wave dipole, 902 to 928 MHz, 2 dBi, VSWR 2.0 or less, SMA male, 195 mm, 90 degree elbow | 1 | $11.97, in stock 2026-10-04 | For the mR900-30-TX kit, which comes with no antenna: the same spec as the dock’s post dipole, so the bench matches the dock (th-telemetry). The bench XIAO pair uses the LoRa FPC antennas in their kits (Seeed: for testing). At close range, run every radio at 17 dBm, the lowest level, so the receivers do not overload; our XIAOs fly at 20 dBm, as the reference pair | Rokland |
| RF dummy loads | Inmet (Spectrum Control) 3004M SMA male 50 Ω termination: DC to 18 GHz, 2 W average (1 W at +125 °C), VSWR 1.3:1 or better; 3 pieces, pre-owned | 1 lot | $14.00, free US shipping (2 to 4 days), eBay seller 73volts; new, $22.91 each at RF Parts | An mLRS transmitter transmits from power-up, and an open connector can damage its power amplifier (th-telemetry, 98c931d): one load on the mR900-30-TX’s SMA pigtail whenever its factory firmware could run, one on a lone XIAO bench board through a U.FL-to-SMA pigtail (bench test kit). Ships from the US, so it arrives before the Matek units; a known-good 50 Ω load also checks the NanoVNA-F V3 and Ryan’s VNA at 915 MHz (th-telemetry) | eBay, RF Parts |
R&D wishlist
Section titled “R&D wishlist”Ryan’s wishlist for research, not for the build (2026-09-28). These are not orders and are not in any build total. Do not order without Ryan.
| Item | Product | Price | For | Link |
|---|---|---|---|---|
| High-permittivity filament | Nanoe Zetamix Epsilon 10: εr 10 ± 0.5, tanδ 0.001 (both at 9.4 GHz, not at GNSS frequencies), TiO2 in polyolefin, 1.75 mm, 500 g | €137.50 per spool | Formers for the custom L1/L5 helix (R&D, benchmarked against the Q43 NEO-F9P that flies), test coupons measured on the NanoVNA-F V3, dielectric resonators and 5.8 GHz lenses | Filament2Print |
| Hardened nozzle | Prusa ObXidian HF 0.4 | price not checked | Printing the filament above: it is 80 % TiO2 by mass and would wear the brass HF0.4 | Prusa |
| Alternative filament | Avient PREPERM ABS, grades εr 3 to 12 (ABS1000: εr 10, tanδ 0.003), 1.75 mm, 750 g | quote only | The same work, if Zetamix is not available. ABS base, so easier to print; abrasiveness not confirmed | Avient |
Future: a printed TPU hinge wiper (Ryan, 2026-09-28). The first build seals each elevon’s hinge line with hinge-seal tape (print plan, “Seal the wing seams”). A tape-free option to try later: a thin flexible TPU strip, printed and fixed to the wing along the hinge line, that rides on the elevon and closes the gap. It needs a design, adds a part and some friction on the elevon, and needs a check that the servo keeps its full travel. Nothing is designed yet.
The helix study (Loaded Helix) sets expectations: a 2 mm tube former of εr 10 brings the L5 helix from 74 mm to about 45 mm with about 74 % efficiency, but its bandwidth is short of the full L5 band. A solid former is shorter still and loses more than half of that efficiency.
Zetamix asks for 270 to 300 °C, a 110 °C bed, a 120 °C chamber and 2.5 mm/s. The Core One+ reaches about 290 °C and has no 120 °C chamber, so small formers and coupons are possible but slow. Measure printed parts: εr falls when a part is not fully dense.
Mapping payload
Section titled “Mapping payload”Terrain mapping (photogrammetry) with a nadir camera in the aft bay of the Fuselage Mid. The camera looks down through a floor opening. Checked 2026-09-25; prices from mapir.camera. The camera writes one JPG about every 1.5 s, an average that MAPIR says depends on the card, so the card must meet MAPIR’s U3 / V30 minimum. Any U3 / V30 card of a known brand also works.
| Item | Product | Qty | Price | Link |
|---|---|---|---|---|
| Mapping camera | MAPIR Survey3W, visible light RGB, 87° HFOV, 12MP, 59 x 41.5 x 36mm, 76g with battery | 1 | $400.00 | MAPIR |
| Trigger cable | Survey3 HDMI PWM trigger + exposure strobe cable, 20cm, Dupont ends | 1 | $12.00 | MAPIR |
| Strap | 20mm hook-and-loop strap through the sleeve’s slots | 1 | on hand | |
| Memory card | MAPIR Camera Memory Card, 128 GB microSD, Class 10, U3, V30, A1 (MAPIR’s minimum) | 1 | $50.00 | MAPIR |
| Total | $462.00 |
The first build carries the camera in the belly payload bay of Fuselage B (Ryan, 2026-10-03; th-blender’s
tools/make_payload_bay.py, files in parts/):
- Payload Cover (Survey3W), PETG, 85g sliced. It closes the bay and holds the camera in its sleeve. The Survey3W Lens Seal, TPU, seals the lens against the cover. Five M3 x 8 countersunk screws hold the covers in heat-set inserts in the bay.
The kit fuselage variant, without the bay, prints these instead, from tools/make_survey3_mount.py:
- Survey3 Belly Mount, PETG, about 12g. A sleeve holds the lower 16mm of the camera, lens down, on a bezel plug that fills the floor opening. The face of the plug follows the measured belly skin, keel strip included, so it sits flush. Its window curves up to the camera face on a 3mm radius. Glue the edges of the plug into the opening. The strap holds the camera in.
- Fuselage Mid (Survey3 cutout), LW-PLA, replaces the Fuselage Mid from the kit. It is the same part with a 64.4 x 46.9mm floor opening, centred 373mm aft of the nose datum.
Why this camera: it is built for an FC to trigger it. ArduPilot fires it with a 2000us PWM pulse. The cable returns a 3.3V exposure pulse, so each photo is tagged with the exact position of the FC. Its fastest shutter is 1/2000s. That is about 8mm of blur at 16 m/s cruise, well under its 5.5cm/px ground resolution at 120m. At 1.5s per JPG and 16 m/s, that is a frame every 24m, about 85% forward overlap at 120m with the wide lens. The narrow Survey3N gives 2.3cm/px but only about 65% overlap at that rate. Thus it needs slower or higher flight.
Fit and balance, checked against the STLs:
-
On the kit fuselage variant, the camera clears the rear hatch by 6.6mm and the FC mount by 9mm. The sleeve clears the FC mount by 6.8mm. The lens face sits 2.6mm above the belly skin. The belly skin is 30mm off the ground here.
-
MAPIR does not publish the position of the lens on the camera face. The bezel window stays out of the 87 x 70 degree view if the lens is up to 19mm off-centre along the span and 11mm fore-aft. Measure the real camera. If the lens is outside those limits, set
LENS_OFFSETin the generator. -
Balance with the P50B pack and the nose package (Ryan, 2026-10-05; th-blender 1dbefba, 7bcbbd8), on Fuselage B with the payload bay (th-aeronautics,
cg.md3fef9c2, with each part sliced alone; ESTIMATES from the build-up, about ±20 mm until the aircraft is on the balance, and the measured CG decides). The pack lies on the CG Slider over Battery Tray B (slide): its front moves from Y 80 to Y 96 (86 as drawn), and the slider’s aft fence holds it (no battery stop). The lighting battery (row 13b) lies flat on top of the pack in its TPU saddle, and the aft strap goes round both (th-blender a8bf553). The pack’s balance lead and 7-pin JST-XH plug cannot lie on top any more: stand the plug upright behind the pack on the starboard side, clear of the XT90 on the port side, and hold it to the aft strap with a short tie so it moves with the pack. At the aft stop it is 19.5mm clear of the front spar and 28mm above the 4-in-1’s feed (th-blender). The Nose Ballast Cups are gone.The beds print keel A until keel B’s root coupon passes. With nose rev C, the CG trim slider, the all-in gear masses and the Walksnail Moonlight (th-aeronautics,
cg.md5856776), the pack moves on the slider from a front of Y 80 (the forward stop) to Y 96:Build Gear CG, pack at the forward stop Pack front for 268mm For 265mm (stall test) All-up mass Survey3W fitted keel A (prints now) 265.6mm Y 87.3 pack at Y 80 and about 9g about 2.87 kg Camera off, cover on keel A (prints now) 262.7mm Y 96, the aft stop pack at Y 86.9 about 2.79 kg Survey3W fitted keel B (decided) 267.4mm Y 81.9 pack at Y 80 and about 32g about 2.84 kg Camera off, cover on keel B (decided) 264.4mm Y 90.6 pack at Y 81.6 about 2.76 kg No trim weight is necessary for 268mm on either gear: the balance sets the slider’s stop pin. Keel B with the camera fitted must not fly with the pack front aft of Y 88 (aft of the 270mm hard limit). Only the stall test needs trim weight, attached inside the nose to the Camera Arm at about Y 55 (ESTIMATE until the balance).
-
The avionics layout in the manual has the pusher ESC and the GPS in this area. Route them forward or to the sides. Keep the GPS away from the ESCs and the power wiring.
-
Power the camera from a 5V pad over USB, so it switches on with the aircraft. Its own battery can stay out, to save 26g.
Wiring (the LUX F765 and the NxtPX4v2 alike): connect the PWM lead to output 8, in timer group 7 to 8 with the pusher,
which is also plain PWM. Connect the exposure lead to the AUX2 pad (GPIO 82) for camera feedback
(params/board-nxtpx4v2.parm, th-avionics). The 6C Mini’s pads are in survey3.md and electronics-notes.md.
In ArduPilot, set the function of that output to camera trigger, and set the trigger PWM to 2000us with 1000us idle. Point the camera feedback pin at the exposure input. Check the exact parameter names against the ArduPilot camera docs for the firmware you flash.
Charge node harness and mounting
Section titled “Charge node harness and mounting”The off-board parts for the charge node: connectors, wire, the thermistor, inserts and screws for the brackets, and the long drill for the motor-wire exit. The boards themselves are a separate order from the fab (see charge-node.md); the harness table there says which connector goes on which end. Checked 2026-09-27 on Amazon in a browser session. Not in the flight cart total.
| # | Item | Product | Qty | Unit | Line | Why | Link |
|---|---|---|---|---|---|---|---|
| H1 | XT30 pigtails | Amass XT30U male and female pigtails, 16 AWG, 10 cm, 5 pairs | 1 | $8.99 | $8.99 | The three male pigtails on the power stage (two shoes, pack); spares | Amazon |
| H2 | XT30 connectors | Amass XT30U male and female, 10 pairs, bare | 1 | $11.99 | $11.99 | Female ends on the two shoe leads and the pack tap’s XT30 branch | Amazon |
| H3 | 16 AWG wire | Silicone wire, 16 AWG, 3 m red and 3 m black | 1 | $7.99 | $7.99 | Shoe leads through the spars and the bay; the pack tap branch | Amazon |
| H4 | JST-GH kit | GH 1.25 mm housings 2 to 10 pin with pre-crimped 28 AWG leads, no tool needed | 1 | $18.99 | $18.99 | The 8-pin balance-plus-thermistor plug, the 5-pin link at the sense board, the 60 mm CAN stubs and the terminator plugs. Its leads are 150 mm, double-ended, so the long CAN runs use H13 to H15 | Amazon |
| H5 | JST-PH kit | PH 2.0 mm housings 5, 6 and 7 pin with crimp terminals | 1 | $9.99 | $9.99 | The 5-pin link plug at the power stage. Needs a JST crimper | Amazon |
| H6 | Balance extension | JST-XH 6S balance extension leads, 22 AWG, 200 mm, 5 pieces | 1 | $8.99 | $8.99 | One is cut and re-terminated GH-8 as the balance-plus-thermistor adapter; the rest are spares | Amazon |
| H7 | Thermistor | NTC 10 kΩ, B 3435, probe on 1.5 m lead, 5 pieces | 1 | $9.39 | $9.39 | Pack temperature for the interlock; the design page’s polynomial is fitted to this part | Amazon |
| H8 | Heat-set inserts | ruthex RX-M2x4, M2 × 4 mm brass inserts, 70 pieces | 1 | $9.99 | $9.99 | One in each of the power bracket’s four bosses; the bosses have the 3.2 mm hole that ruthex specifies. H1 in the board is an M2.5 hole, and an M2 screw goes through it | Amazon |
| H9 | M2 screws | M2 socket head screws, 4 to 16 mm, 660 pieces | 1 | $7.99 | $7.99 | M2 × 5 for the power stage into the inserts, M2 × 6 for the sense board into the floor tray (th-blender, 6053c96) | Amazon |
| H10 | Long drill | HSS twist drill, 10 mm, 200 mm long, 155 mm flute (uxcell, 2 pieces) | 1 | about $10.00 | $10.00 | WITHDRAWN (th-blender, 2026-10-10): do not drill. The bore at X ±101 is the 8 mm wing spar’s seat, and the motor wires use the kit’s 25 x 10 mm channel through the Fuselage Mid stub. ESTIMATE, price not read. The motor-wire exit is 10 mm since 2026-09-29, for the 4-in-1’s phase leads (was 8 mm); it needs 65 mm of reach past the drill guide, and a jobber bit is too short | Amazon |
| H11 | CAN terminator resistor | 120 Ω, 1/4 W, 1 % metal film, 5 pieces | 1 | about $6.00 | $6.00 | ESTIMATE, price not shown. Both CAN buses end in terminator plugs, not on nodes (th-avionics, 2026-09-28): a GH-4 housing from H4 with this resistor across CAN high and low. Make four: two for bus 1, one in the free header at each front pod (the joiner’s in phase one, our ESC’s in phase two), and two for bus 2, one in the joiner beside the airspeed module and one in the CAN2 joiner at the FC. The airspeed module is a 60 mm stub off its joiner with its software terminator off, so a module swapped or a plug worked loose loses only the airspeed, never the bus termination (th-avionics, 2026-09-28). The module sits in the Fuselage Front’s starboard lobe; it does not fit in the nose cone. Every ESC and every node keeps its terminator jumper open, so a spare or swapped ESC, or an added or removed node, never changes termination. One resistor is spare | Amazon |
| H12 | Shoe-lead ferrite | Fair-Rite 2643250302, 43 material shield bead: 6.35 mm OD, 3.18 mm ID, 15.88 mm long, 122 Ω at 100 MHz, about 1.8 g (ESTIMATE from its size) (DigiKey 1934-1337-ND) | 3 | $0.46 | $1.38 | One per keel, in th-blender’s sole boss, and one spare: about 3.6 g in all at the keels (ESTIMATE). The top bead inside the spar end is dropped: the bore is full of phase leads and RG316 (th-blender), and in th-telemetry’s model no top bead costs 0.5 dB at the worst azimuth and gains 0.5 dB in mean gain (antennas.md ed35fde, 2026-10-06). 43 material is past its peak at 915 MHz but still lossy there. It replaces the 0461164281 snap-on, which is 26 g each. Fair-Rite lists no 61-material bead in this size, and 43 is acceptable at 915 MHz (th-telemetry). Slide it on before the lead is terminated. The bore is 3.18 mm per DigiKey but 2.95 mm per RS Components, so the keel stretch is thin-wall fluoropolymer 16 AWG, M22759/16-16 at 2.0 mm (see the charge lead in the keel B table) | DigiKey |
| H13 | GH crimp contacts | JST SSHL-002T-P0.2, the GH socket contact for 26 to 30 AWG, loose | 100 | $0.02 | $2.00 | The CAN runs longer than H4’s 150 mm leads: bus 1 at 330 to 575 mm and bus 2’s nose run at about 400 mm (harness thread 20260928T051244Z). Not MINI-SSHL-002T-P0.2, which is the same contact at ten times the price | TME |
| H14 | GH crimp tool | Engineer PA-09 micro crimping pliers, 20 to 32 AWG | 1 | $55.84 | $55.84 | Engineer lists SSHL-002T-P0.2 as a compatible contact; JST’s own tool (WC-GH2630) costs far more. Not the IWISS SN-025, whose dies are too large for GH. Users report it also crimps PH, which H5 needs. The Engineer PAD-11 (smaller dies, 0.7 to 2.2 mm) is the better-regarded tool for GH-size contacts, at a higher price (TME); crimp and pull-test a few contacts with either before the harness | TME |
| H15 | CAN wire | 28 AWG stranded hook-up wire, thin wall, four colours (red, yellow, green, black), about 3 m each | 1 | about $10.00 | $10.00 | ESTIMATE, product not chosen. The insulation must fit the GH housing’s cavity: compare it with an H4 lead before buying. Twist CAN high and low by hand, about one turn per 25 mm | |
| H16 | CAN joiner board | Our CAN joiner rev A.1 (th-eda, terrahawk-eda/boards/can-joiner/, order files frozen at fab/releases/can-joiner-revA1-2026-09-30/, 8a79efb): 29 x 8 mm, 0.8 mm FR-4 (the seat was reopened for it, th-blender 8e474d0), three JST SM04B-GHS-TB side-entry headers on one edge, all four pins bussed, no terminator footprint, assembled. It ships as a 6-up panel, 74.1 x 72.1 mm, JLCPCB Economic PCBA, one side, 18 header placements a panel. Rev A.1 only moved the buses so the top edge can carry the panel tabs; outline, headers and pinout are unchanged |
30 | about $1.17 | about $35.00 | ESTIMATE from JLCPCB’s published fees (th-eda, 2026-09-30): 5 panels of 2-layer 0.8 mm board about $5 to $10, Economic PCBA $8 setup, stencil about $1.50, one extended-part feeder fee about $3, 90 connectors about $9 to $14, placement a few dollars; about $30 to $40 before shipping. The quote when Ryan uploads the zip sets the price. 30 because JLCPCB’s minimum is 5 panels. The first build flies with CAN off and the I2C MS4525DO, so it needs none for flight; the night lights node takes one of its own when the lights are fitted (on the LUX’s one CAN port at first, on bus 2 with our own flight controller, about 0.2 m from the node: trunk in, trunk out and the node’s stub, or the LUX lead, the stub and a terminator plug; th-eda, 2026-10-11). The pod path needs seven: four pods, CAN1 and CAN2 at the FC, and the airspeed module. 21 in all; the 30 cover it. At order time, check the connector rotation in JLCPCB’s placement preview (th-eda’s README). No bought splitter fits the pod seat (the smallest, Vimdrones’ 5-port board, is 35 x 10 mm and has a terminator jumper) | LCSC header |
| Total | $214.53 |
Airspeed lines. On Fuselage C v2.4B the total pressure comes from our own 4 mm brass tube beside the camera, starboard at X -18, 7 deg tip down, its open bevelled mouth at (-18, -25.3, 12.22) (Ryan, 2026-10-19; th-blender W104 888383ba; pitot-tube’s HANDOFF 3.6). It is total pressure only: it has no static outlet. Its soldered collar (rows 11l, 11m) seats in the Nose Cone Plate’s aft pocket, and the M2 x 6 clamp screw at nose Y 21.5 holds it, so it comes off with the cone unit. The static pressure comes from two flush ports, one on each side at Y 40.0, Z -10 on Fuselage C v2.4B (th-blender W77, de8f23b8; since 2026-10-06, Ryan, th-blender e1f6bdc). Join the tube’s aft end to the Pitot Coupler’s spigot on the MS4525 with an 11 mm sleeve of 3 x 5 mm line (row 11h): 5.16 mm on the tube, a 0.84 mm gap, 5.0 mm on the spigot, dry. Connect each port tube (row 11d) with a short leg of 1/16 in line (row 11f) to the tee (row 11e), and the tee’s third leg, the 85 mm trunk, through a short cut of the 3 x 5 line onto the Pitot Coupler’s static spigot (lengths in row 11f, th-blender W111 c4d7ee16). Cut the static lines to th-blender’s W111 lengths (row 11f): the two branches must stay equal within 0.5 mm. Mark the total-pressure line and its nipple with the same colour, a band of coloured heat shrink (th-avionics). ARSPD_TUBE_ORDR 2 also accepts either order, but a line that is not fully on reads low, so do the pitot check after you disconnect a line.
The charge node needs CAN, so this harness waits until CAN is on: the LUX F765 that flies first keeps its one CAN port off (2026-10-04). Already in the cart: the XT60 pigtails (row 15) make the pack tap’s Y-lead. The CAN cables come with the DroneCAN airspeed sensor (Follow-on). The H4 GH kit also makes the bus 2 terminator plug (H11) and the GH1.25 8-pin end of the Q43 lead: for the first flights the Q43 goes to the LUX F765’s GPS port through the Y lead (row 9f), SH 1.0 6-pin at the board end. Zip ties for the power bracket’s tunnel are on hand.
Hangar landing gear (keel B)
Section titled “Hangar landing gear (keel B)”Keel B, chosen 2026-09-29 in place of keel A: an airfoil-section rear keel blended into the rear mount and a
blended front stilt per side, each an LW-PLA blade on a PETG mount at a keyed seam, on PETG canoe soles; see
docs/keel.md. The printed parts wait for th-aeronautics’ strength gate. Listed here, not in the cart total. The FR4 ENIG contact plates and the pad pins are th-hangar’s and th-eda’s parts.
| Item | Product | Qty | Price | Link |
|---|---|---|---|---|
| Keel cores, E-glass | 6 x 4mm pultruded glass-fibre tube, 1m: four 125mm (front at Y 433, rear at Y 462), 500mm (th-blender, keel B; keel A’s front cores were 137mm). Glass, not carbon, because the ELRS and SiK antennas stand in the keels beside the cores (th-blender, d0f1489) | 1 | ESTIMATE, from about €4 (price not read; the range starts at €3.78), plus shipping from Germany | Carbon Composite PGT06040-1 |
| Front stilt cores, carbon | 6 x 4mm pultruded carbon tube: two 125mm | from stock | $0.00 | the 300mm left of the elevon-spar tube |
| Zip ties | 2.5mm nylon, about 100mm | 4 | on hand | |
| Charge lead, keel stretch | 16 AWG thin-wall fluoropolymer wire, M22759/16-16 (ETFE, 600 V, 150 °C, 19 strands of tin-plated copper, 2.0 mm OD nominal, 2.06 mm at most), or the thinner M22759/32-16; one colour per side, since both shoes on a side share a conductor (th-telemetry). From the sole bead through its own 3.0 mm channel at Y 490 and up into the spar end. Join it to the 16 AWG silicone shoe lead inside the bay or at the XT30. 16 AWG because the pack charges at up to 10 A (docs/charge-node.md, th-telemetry 2026-10-06; was listed as 22 AWG in error). Length to be re-measured on the keel B files |
about 0.4m per foot, 2m | to price | ProWire, spec |
| CAN pair | 28 AWG twisted pair, starboard-front foot only | 0.5m | on hand | |
| Sole contact strips, brass | Cambrian Model Rail brass strip, CZ121, 8.0 x 0.5 x 250 mm. th-eda (2026-10-11), keel B W10 (th-blender 11ca1d8): no ready-plated 8 x 0.5 mm nickel-brass stock at hobby scale, so bare brass and plating. Each keel takes a flight strip (84 mm) and a lights strip (28 mm), cut from one 116 mm length; one 250 mm strip makes both keels’ four with the kerf, the second is a spare. Plate after cutting, then solder the leads to the back with an active flux and keep solder off the contact face. The dock pins press at 0.4 N | 2 | £2.25 each | Cambrian Model Rail |
| Sole contact strips, plating | Electroless nickel, about 5 to 10 µm, after cutting and deburring: Caswell One-Plate 1 kit (Mini) or a local plating shop for the four short strips (Ryan’s call: the kit is $148.35, mostly left over) | 1 | $148.35 (kit) or a shop quote | Caswell |
| Lighting charge leads, keel | 22 AWG ETFE wire, M22759/16-22 (about 1.3 mm OD): L+ red (M22759/16-22-2) in the port keel, L- black (M22759/16-22-0) in the starboard keel, each about 0.6 m (168 mm in the keel plus about 400 mm to the lights node, ESTIMATE until routed), in the keel’s 1.6 mm bore (th-blender W10.1) with a 2643001301 bead at the sole (cart 13l) | 1 m of each colour | to price | ProWire |
The carbon 6 x 4 tube is the elevon-spar stock. The 1m already ordered leaves 300mm after the two elevon spars, enough for the two 125mm front stilt cores. The keel-as-foot mounts (2026-09-28) grip the cores in their root sockets, so the collars and their cross pins are gone.
Dock weather station
Section titled “Dock weather station”The hardware for the dock’s weather station (hangar.md, “Weather station”): Harry’s Weather Station One, parts 3,
4 and 5 (Printables,
CC BY 4.0), with th-hangar’s changes. Both wind shafts read an AS5600 angle sensor, so the author’s reed contacts,
multiplexer and 4 x 1 mm magnets are not needed, and there is no 433 MHz radio. The station has its own controller,
a weather node in an IP65 box on the post at 1.35 m, on the dock’s CAN bus (th-hangar, hangar.md a62aaaa). It can
run before the dock exists, on a plug-in 12 V supply, and log the site’s wind. Prices read on 2026-10-01 for a US
buyer; Qty is the number of units or packs to buy. Not in the flight cart total. All metal parts stainless: the gauge
and the wind unit get wet inside.
| Item | Product | Qty | Price | Link |
|---|---|---|---|---|
| Angle sensor | AS5600 magnetic angle sensor board, the round-PCB version with GND, 5V and PWM pads, 2-pack | 1 | $8.49 | Amazon |
| Diametric magnet | Neodymium N38, diametrically magnetized, 8 x 1 mm, 50-pack (the only 8 x 1 mm diametric disc found from a US store) | 1 | $10.95 | BuyNeoMagnets |
| Weather sensor | BME280 board, the 4-pin one-hole-mount version, 3-pack. Its silkscreen says “BME/BMP280”: on arrival, check that it reads humidity, which only the BME280 does | 1 | $8.49 | Amazon |
| Rain gauge reed contact | Reed contact, plastic body, 2 x 14 mm, normally open, 5-pack; ships from China, 6 to 16 days | 1 | $13.54 | Amazon |
| Rain gauge magnet | Neodymium N48, 5 x 1 mm, 50-pack | 1 | $6.99 | Apex Magnets |
| Miniature bearings | S693ZZ, 3 x 8 x 4 mm, stainless 440C, shielded: 4 in the wind unit, 2 in the rain gauge, 2 spares. Not the plain 693, which is chrome steel and rusts | 8 | $1.99 | RCBearings |
| Axle screws | M3 x 30 socket cap, stainless A2, DIN 912; the cup axle’s screw goes head down | 4 | $0.10 | Bolt Depot |
| M3 x 10 screws | Socket cap, stainless A2: 14 wind unit, 2 BME280 housing, 5 rain gauge, 4 spares | 25 | $0.06 | Bolt Depot |
| M3 x 20 screws | Socket cap, stainless A2 (M3 x 16 or 18 also fits): rain gauge body to base | 6 | $0.07 | Bolt Depot |
| M6 rod and nuts | Stainless A2 M6 threaded rod, 1 m, cut into the wind unit’s and the rain gauge’s base mounts, with 6 M6 nuts ($0.66) | 1 | $8.69 | Bolt Depot |
| M4 rod and nuts | Stainless A2 M4 threaded rod, 1 m, cut into 3 for the radiation shield’s plates (tools/make_hangar_parts.py), with 8 M4 nuts ($0.56), which the list had left out |
1 | $7.45 | Bolt Depot |
| Sensor cable | Tensility 30-01138, 6-core 26 AWG, shielded, 4.5 mm outer diameter: about 4 m of short runs to the weather node on the post and one run of about 10 m to the bay (12 V, ground, CAN H, CAN L) | 15 m | $5.93 | DigiKey |
| Arms | 2020 slotted aluminium extrusion, 610 mm: cut 400 mm from one and 2 x 250 mm from the other | 2 | $7.50 | Adafruit |
| Filament | White PETG, and TPU 95A for the seals | about 300 g, ESTIMATE | on hand | print the shield in white |
| Controller | Espressif ESP32-S3-DevKitC-1-N8R8, the PCB-antenna version; WiFi is only for the stage before the dock (Mouser 356-EP32S3DVKTC1N8R8; DigiKey was out of stock) | 1 | $15.00 | Mouser |
| CAN transceiver | Waveshare SN65HVD230 CAN Board, 3.3 V, SKU 3945; its 120 Ω termination stays fitted, the node is a bus end. Ships from China | 1 | $4.99 | Waveshare |
| 12 V to 5 V | Pololu D24V10F5, #2831. It has no reverse-polarity protection, so the diode below goes in series. The page allows backorders: check stock | 1 | $12.95 | Pololu |
| Reverse-polarity diode | 1N5819 Schottky, 1 A 40 V, DO-41, in series with the 12 V input; 9 spares | 10 | $0.14 | Mouser |
| Logging | Adafruit MicroSD card breakout board+ #254 ($7.50) and a SanDisk 32 GB Ultra microSD card ($25.21, Amazon) | 1 | $32.71 | Adafruit |
| Enclosure | LeMotech IP65 ABS junction box, inner 108 x 83 x 48 mm, 2-pack. Its lid screws are not rustproof: fit stainless ones | 1 | $14.59 | Amazon |
| Cable glands | Altech 5306107, PG7, clamps 2.5 to 6.5 mm: three sensor runs, the soil probe and the bay cable. Check that a locknut comes with it | 5 | $2.78 | Mouser |
| Breather vent | Amphenol LTW VENT-PS1YBK-O8002, M12 pressure-equalising vent with nut; stops condensation in the box | 1 | $3.30 | Mouser |
| Passives | 1/4 W resistors, Yageo MFR-25, 10 of each: 100 Ω (reed input), 10 kΩ (reed pull-up), 2.2 kΩ and 3.9 kΩ (the AS5600 PWM dividers), 4.7 kΩ (the soil probe’s 1-Wire pull-up to 3.3 V), about $0.42 each value; 100 nF 50 V capacitors (10, reed debounce); proto-board 3-pack | 1 set | $5.82 | Mouser |
| Supply, first stage | CUI SWI12-12-N-P5, 12 V 1 A wall adapter, 2.1 x 5.5 mm barrel, centre positive, until the dock’s 12 V reaches the post | 1 | $17.59 | Mouser |
| Soil probe | DS18B20 waterproof probe, stainless sheath 6 x 50 mm, 5 m lead, 5-pack (4 spares): buried beside the RTK mast footing at the footing’s depth, so the weather node logs frost under the base (th-hangar, hangar.md 1deec64). The lead jacket is rated to about 85 °C, which is enough here |
1 | $23.99 | Treedix |
| Probe conduit | JM Eagle 67447, 1/2 in Schedule 40 PVC conduit, 10 ft, rated for direct burial, 2 lengths joined at the bell ($11.36), for the buried run of about 4 m from the mast footing to the beacon post; 2 x 1/2 in Schedule 40 90° conduit elbows, the wide-radius electrical kind, not plumbing elbows, one turning down at the footing and one up at the post (about $1.18 each, Menards BPF49DB001); Oatey 31012 PVC cement, 4 oz (about $6.99). Prices from cached pages, ESTIMATE: buy at a local store | 1 set | $20.71 | Home Depot |
| Total | $353.74 | before shipping and tax; the unit prices above are rounded to the cent |
- AS5600 supply. Run the AS5600 boards on 5 V and divide each PWM line, 2.2 kΩ over 3.9 kΩ (5 V to 3.2 V), because
the ESP32-S3’s inputs are not 5 V tolerant (th-hangar,
hangar.md96b2084). These boards feed the chip’s VDD5V pin, which needs 4.5 to 5.5 V, so 3.3 V is not an option. - Soil probe. Wire the DS18B20 in powered mode (three wires: 3.3 V, data, ground), not parasite power, with the 4.7 kΩ pull-up from data to 3.3 V at the weather node. The 5 m lead goes through the buried conduit and into the box through the fifth gland. Seal the conduit’s top end around the lead so it does not fill with water. The probe’s sheath sits in the soil just past the conduit’s lower end, at footing depth, not inside the conduit: it must read the ground, not the air in the pipe.
- Stores. Bolt Depot takes all the stainless fasteners in one order. Mouser takes the controller, diode, glands, vent, passives and adapter, about $56, which is near its free-shipping limit. Treedix takes the soil probes; buy the conduit locally.
Dock controller
Section titled “Dock controller”The dock’s Rover board (th-hangar, hangar.md a9e054ba, with the reasoning and the port map). th-avionics builds its
firmware: stock ArduPilot Rover with one storage patch, so the signing key is not kept on the SD card. It takes its
power from th-hangar’s 12 V dock rail through the PM02 V3 that comes with it. The lid drive and the
bay’s supplies follow th-hangar’s design. The first flights do not use the dock, and the time to buy is Ryan’s. Prices read on 2026-10-19. Not in the flight cart total.
| Item | Product | Qty | Price | Why | Link |
|---|---|---|---|---|---|
| Dock flight controller | Holybro Pixhawk 6C Mini, Model A (revision), with the PM02 V3 (12S) power module | 1 | $197.99 | The PM02 V3 powers the board from the bay’s 12 V rail and reports that rail’s voltage and current to it, so the dock watches its own supply (th-hangar, hangar.md 6c6fd9a4). Holybro gives its input as 2S to 12S, so 12 V is inside it, and its output as 5.2 V, 3 A maximum, with 120 A current sensing (PM02 V3, 2026-10-19). Board only is $179.00 | Holybro |
| Dock flight controller microSD card | microSD card, 4 to 32 GB, FAT32, the same choice as row 9g | 1 | about $8.00 | ArduPilot needs it for logs; Holybro’s page does not say that one is included. ESTIMATE, part to choose: any name-brand card | part to choose |
| Lid gearmotor | goBILDA 5203 Series Yellow Jacket planetary gear motor, 13.7:1, 435 rpm, 8 mm REX shaft, 3.3 to 5 V encoder (5203-2402-0014): 12 V, 0.25 A no load, 1.83 N·m (18.7 kg·cm) and 9.2 A at stall | 1 | $54.99 | Drives the roll-off lid (th-hangar, the gentler lid in hangar.md after 6534957b). Against th-hangar’s running load of about 0.45 N·m it turns about 330 rpm on its published line (ESTIMATE), so the roof opens in about 70 s; stall torque is about four times the load. Its 9.2 A stall is the dock’s stall figure, in place of 20 A. Not sealed: it sits inside the bay | goBILDA |
| Lid motor driver | Pololu G2 High-Power Motor Driver 18v17 (#2991): 6.5 to 30 V, 17 A continuous, PWM, DIR and SLP inputs for 1.8 to 5 V logic, current-sense output about 20 mV/A, adjustable current limit, no over-temperature shutoff | 1 | $44.95 | Its current-sense output gives the lid’s stall detection; its current limit is set to 8 A, under the motor’s 9.2 A stall, which still gives about 1.6 N·m, 3.5 times the running load (th-hangar, hangar.md a11fda93; ESTIMATE). Pololu calls the limit imprecise when set this low, so measure the trip on the bench. In place of the Cytron MD13S, which has no current sense | Pololu |
| Bay 12 V supply | Mean Well HLG-240H-12A: 12 V, 16 A, 192 W, 90 to 305 VAC input, free-air convection (no fan), IP65, Vo and Io set on built-in potentiometers, 244.2 x 68 x 38.8 mm | 1 | $78.80 | The dock’s 12 V rail; its peak is about 14 A (th-hangar). Potted and fanless, so it takes the bay’s condensation, and it has no 115/230 V switch to set. Specs from Mean Well’s HLG-240H datasheet; price and 998 in stock at TRC. Fallback: LRS-200-12 (17 A, fanless, not potted) | TRC |
| Servo rail supply | Pololu 6V, 11A Step-Down Voltage Regulator D42V110F6 (#5673): 6 V ±3 %, 6 to 60 V input, 11 A typical at 42 V input, ENA and PG pins, reverse-voltage, over-current and over-temperature protection | 1 | $69.95 | 12 V to the dock’s servos, at least 6 A (th-hangar). The output current rises as the input voltage falls, so more than 11 A at 12 V (ESTIMATE). Stock is rationed at Pololu | Pololu |
Dock charger board, early buys
Section titled “Dock charger board, early buys”Parts for th-eda’s dock charger board (hangar-io) that are bought before the board is ordered, because stock is short. Ryan approved the buy on 2026-10-14. Not in the flight cart total. When hangar-io joins Part-DB, its BOM line uses this stock.
| Item | Product | Qty | Price | Why | Link |
|---|---|---|---|---|---|
| Charger float divider, top | Yageo RT0603BRD07634KL, 634 k 0.1 % 0603 thin film, 25 ppm/°C (DigiKey 13-RT0603BRD07634KLCT-ND) | 10 | $0.065 | R59 of the LT3652 float divider (EDA d585f17): 0.1 % holds the 2S lights float inside th-avionics’ 8.30 to 8.48 V window (hangar.md c1d141f5). DigiKey had 74 in stock on 2026-10-14, so ten now; one is fitted, nine are spares. The 412 k partner (TE CPF0402B412KE1, 3,054 in stock) waits for the board order |
DigiKey |
Bench tools
Section titled “Bench tools”Bought parts for the bench tools that build the aircraft’s antennas and its pitot tube. The first is th-blender’s coil winder Mk2 (1ec3daab; Ryan approved it on 2026-10-12, and its beam frame on 2026-10-14, 83b37cec): two PLA cheeks on a printed Base Beam and 4 mm steel rods, a quarter-turn detent and a turns scale, for winding the LoRa coils on their formers. It replaces the Mk1 cradle, thumbwheels and wire guide (printed as job 22 on the Gen 2); the ELRS dipole gauge from that job stays. The main PLA plate is about 93 g as sliced (3 h 22 min, 2026-10-14), and the parts are bench tools, so the print hold does not apply. Not in the flight cart total. Prices to confirm: these are hardware-store and hobby-shop items.
| Item | Product | Qty | Price | Why | Link |
|---|---|---|---|---|---|
| Winder rods | 4 mm silver steel or 4 mm ground shaft, cut from 330 mm lengths to 2 x 128 mm and 1 x 142 mm | 3 pieces | to confirm | The two frame rods are 128 mm for the beam frame (2026-10-14): if they are already cut to 135 mm, cut 7 mm off each. They join the cheeks low at the back and carry the carriage, pressed into the cheeks’ 4.0 mm holes; the third carries the wire spool between the spool cones (th-blender’s generator) | |
| Wire eyelet | Brass tube, 1.5 mm OD x 1.0 mm ID, cut to 6.5 mm | 1 | to confirm | The carriage’s eyelet that lays the 20 AWG wire into the groove; countersink both ends 45 degrees (th-blender) | |
| Hub screws | M3 x 12 socket head | 2 | local | Hold the Crank Wheel and the Tail Knob to their hubs; each cuts its own thread in the hub’s 2.8 mm hole (th-blender’s generator) | |
| Crank handle screw | M3 x 20 with an M3 nut | 1 | local | The crank wheel’s spinner handle turns on it | |
| Feet (SUPERSEDED: do not order) | Self-adhesive rubber feet, 12 mm x 3.5 mm | 0 | to confirm | Replaced by printed TPU feet that snap into through holes in the cheeks (Ryan, via th-telemetry; th-blender W28) | |
| Spool brake pad (optional fallback) | Self-adhesive felt, 6 x 8 x 1.6 mm | 1 | to confirm | Optional: a printed TPU snap-on brake pad (two thicknesses, standard and +0.4 mm) is the first choice (Ryan, via th-telemetry; th-blender W28). Fit the felt only if the TPU pad does not give an even drag: about 0.3 to 0.6 N of wire tension on the spring beam in the tail cheek (th-blender, ESTIMATE) | |
| Beam screws | M3 x 16 socket head | 4 | local | Hold the Base Beam’s dovetail ends up in the cheeks’ sockets from below; each cuts its own thread in the cheek (th-blender W31, 83b37cec) | |
| Pitot tube bevel bit | CIC Workshop VSC30: 30° V-groove bit, 3 flutes, solid carbide, 1/4 in (6.35 mm) shank and cutting diameter, 7/16 in cutting height | 1 | $34.99 | Cuts the bevelled mouth of the own total tube (row 11g), turned by hand in a T-handle tap wrench, not in a spindle (how-to: Build the airspeed tube, 5050da76). The bevel reaches 3.8 mm across, so the cone must be wider than that: a 1/8 in shank bit’s cone stops at 3.175 mm (th-docs-site, 2026-10-19). Any 30° bit with a 1/4 in shank does; SpeTool’s W06010 ($19.99) and Inventables’ 3-flute bit were sold out on 2026-10-19 | CIC Workshop |
Thrust stand
Section titled “Thrust stand”The electronics for the thrust stand’s measurement box (thrust-stand-ae, spec in ../thrust-stand/docs/esp32-box.md;
part choices th-eda, stock and prices from Part-DB’s DigiKey and Mouser lookups on 2026-10-12). The stand tests the lift
and pusher motors against the pass marks in docs/propulsion.md, so it measures current itself: the 6S feed runs through
the box, and an INA228 reads a Kelvin shunt in it. Not in the flight cart total: $311.05 with the recommended HX711, the corded EPO, the rotary encoder, the temperature and air sensors and their two GX12 connectors, and the optional resistors (2026-10-18: the 2.0 in TFT is Ryan’s own), plus the M3 and M4 steel from a local A2 stainless assortment (th-eda’s advice). The XT90 plugs come from row 15a’s spares. Ryan already has the 10 kg load cell and an HX711 (2026-10-12, via thrust-stand-ae). The stand borrows the
flight ESCs, the Lucid 4-in-1 (row 2b) and the Aria (row 5): record each one’s AM32 settings before any bench change,
and restore them before flight.
| Item | Product | Qty | Price | Why | Link |
|---|---|---|---|---|---|
| Current monitor | Adafruit 5832, INA228 breakout; remove its onboard 15 mΩ shunt and wire the external shunt to its sense pads | 1 | $14.95 | 20-bit current and voltage on I2C; with the 1 mΩ shunt it resolves 0.31 mA (th-eda) | DigiKey 1528-5832-ND |
| Current shunt | Bourns RSA-100-100: 100 A, 100 mV (1.000 mΩ), ±0.25 %, Kelvin terminals | 1 | $38.71 | 100 mV in place of the spec’s 75 mV: the exact 75 mV part (Ohmite SHD1-100C075DE) had 3 in stock and is ±0.5 %. The 7.3 A hover reading is limited by the shunt’s 0.25 %, not the ADC (th-eda) | Mouser 652-RSA-100-100 |
| 5 V supply (SUPERSEDED: do not order) | Recom R-78K5.0-1.0, 6S to 5 V 1 A switching regulator | 0 | $3.98 | Replaced by th-eda: its 36 V input limit is inside the range the Lucid’s brake-on-stop regen spikes can reach on the bus (low to mid 30s V, th-eda’s ESTIMATE), and the stand runs braked tests. A 9 to 72 V input regulator and a P6KE39A TVS take its place; their rows are added at the end of this table when th-eda confirms the parts | DigiKey 945-R-78K5.0-1.0-ND |
| 5 V isolation diode | MCC 1N5819-TP Schottky | 2 | $0.31 | One keeps the regulator and the XIAO’s USB 5 V apart; the second is the buzzer’s flyback diode (th-eda) | DigiKey 1N5819-TPCT-ND |
| Load cell amplifier | Adafruit 5974, HX711 breakout with a 10/80 SPS slide switch | 1 | $9.95 | Recommended: thrust-stand-ae designs the box around its outline and holes. Ryan’s own HX711 can replace it only if it runs at 80 SPS and fits the same mount. Set to 80 SPS for the thrust log | DigiKey 1528-5974-ND |
| 5 V supply | Recom R-78HB5.0-0.5L, SIP-3 switching regulator with bent pins (lies flat), 9 to 72 V in, 5 V 0.5 A | 1 | $18.32 | Replaces TH-T-3: its 72 V input clears the Lucid’s braked regen spikes. Fed from a fused tap off the 10 AWG line (th-eda). The bent-pin L version, not the upright R-78HB5.0-0.5: upright it stands 17.5 mm off the perfboard, under the arm switch and the button (thrust-stand-ae’s CAD); same part and pinout | DigiKey 945-1614-5-ND |
| Regulator input capacitor | TDK FG23X7R2A335KNT06, 3.3 µF 100 V X7R, through-hole | 1 | $3.08 | At the regulator’s input (th-eda) | DigiKey 445-181578-1-ND |
| Regulator output capacitor | KEMET ESH226M025AC3AA, 22 µF 25 V electrolytic | 1 | $0.18 | At the regulator’s output (th-eda) | DigiKey 399-6562-ND |
| Tap surge clamp | MCC P6KE39A-TP TVS diode, about 33 V standoff, 54 V clamp | 3 | $0.43 | Across the regulator’s tap, after the fuse; two spares (th-eda) | DigiKey P6KE39A-TPMSCT-ND |
| Display (SUPERSEDED: do not order) | Adafruit 326, 0.96 in 128 x 64 SSD1306 OLED, I2C | 0 | $17.50 | Replaced by the 2.42 in Adafruit 2719 at the end of this table (th-eda, thrust-stand-64) | DigiKey 1528-1220-ND |
| Arm switch (SUPERSEDED: do not order) | CW Industries GTS447A101HR toggle switch, SPST | 0 | $6.87 | Replaced by the DPDT GTS448B101A1HR at the end of this table, the hardware interlock that cuts the ESC signal (thrust-stand-ae, th-eda) | DigiKey CWI285-ND |
| Arm switch cover (SUPERSEDED: do not order) | CW Industries GT-4R safety cover for the toggle | 0 | $5.72 | Not needed with Ryan’s keyswitch (th-eda with thrust-stand-21, 2026-10-18, esp32-box.md Safety; Mouser via Part-DB). Was: Keeps the arm switch from being flipped by accident | DigiKey CWI312-ND |
| Start and abort button (SUPERSEDED: do not order) | Same Sky PBM2-M-R2GR3-R-BA-67 panel pushbutton | 0 | $9.58 | Ryan uses his own 18.5 mm button, on hand, row at the end of this table (th-eda with thrust-stand-21, 2026-10-18, esp32-box.md Safety; Mouser via Part-DB). Was: Start a sweep, or abort to zero throttle. DigiKey had only 98 in stock on 2026-10-12 | DigiKey 2223-PBM2-M-R2GR3-R-BA-67-ND |
| Buzzer | Same Sky CMI-1295-0585T magnetic buzzer | 1 | $2.08 | Warns before the motor spins | DigiKey 102-CMI-1295-0585T-ND |
| Buzzer driver | Microchip TN0702N3-G N-channel MOSFET, TO-92 | 1 | $1.52 | Switches the buzzer from a XIAO pin | DigiKey TN0702N3-G-ND |
| Load cell jack | Same Sky SJ1-43502PM, 3.5 mm 4-pole panel jack | 1 | $2.46 | The load cell’s four wires to the box | DigiKey CP-SJ1-43502PM-ND |
| Fuse holder | Littelfuse 0FHM0003ZXJS inline holder for ATM mini blade fuses | 1 | $6.39 | On the regulator’s tap off the 10 AWG line | DigiKey 5346-0FHM0003ZXJS-ND |
| Tap fuse (SUPERSEDED: do not order) | Bussmann ATM-1, 1 A mini blade fuse | 0 | $5.31 | Replaced by the Littelfuse 0297002.WXNV 2 A MINI fuse at the end of this table (thrust-stand-ae’s reviewed list, with th-eda) | Mouser 504-ATM-1 |
| Power wire | DFRobot FIT0581, 10 AWG silicone wire, 1 m red and 1 m black | 1 | $7.32 | The 6S path through the box: XT90-S in, the shunt, XT90 out | DigiKey 1738-1431-ND |
| Strain relief grommets (SUPERSEDED: do not order) | LAPP S2507 M12 cable gland | 0 | $2.44 | The XT90s now panel-mount on the box in printed mounts (Ryan, 2026-10-12, via thrust-stand-64), so the glands and pigtails go | DigiKey 2181-S2507-ND |
| XT90 plugs (not ordered) | XT90-S male in (anti-spark, battery side, back wall) and XT90 female out (to the ESC, right wall), panel-mounted in two printed mounts (L340’s XT90 Panel Mount Unisex, Printables 1530294, CC BY 4.0) | 0 | From the spares in row 15a: one male and one female go to the stand. th-eda checked the genders against row 15a and the pack lead. The 10 AWG now runs only inside the box, from the XT90 cups to the shunt lugs (thrust-stand-64) | ||
| Box controller | Seeed XIAO ESP32-S3, the bare board (SKU 113991114, no Wio kit) | 1 | $7.49 | The box’s own: the five in row 16 belong to the aircraft (thrust-stand-ae). $7.49 at DigiKey and at Mouser (713-113991114), status active, from Part-DB’s provider lookup on 2026-10-12; stock not reported | DigiKey 1597-113991114-ND |
| Arm switch (SUPERSEDED: do not order) | CW Industries GTS448B101A1HR toggle switch, DPDT ON-ON | 0 | $8.40 | Ryan uses his own 2-terminal keyswitch, on hand, row at the end of this table (th-eda with thrust-stand-21, 2026-10-18, esp32-box.md Safety; Mouser via Part-DB). Was: The hardware interlock: off, it cuts the ESC signal, so nothing spins (thrust-stand-ae, th-eda) | DigiKey CW163-ND |
| Tap fuse | Littelfuse 0297002.WXNV, 2 A MINI blade fuse | 5 | $0.39 | In the tap’s holder, before the regulator and the TVS; four spares. 2 A on purpose: the tap draws about 0.15 A, 2 A still protects the holder’s 18 AWG leads, and the TVS clears the fuse or fails short into it (th-eda) | Mouser 576-0297002.WXNV |
| Shunt current lugs | Panduit PV10-14R-D, 1/4 in stud ring lug, 10 to 12 AWG | 2 | $0.51 | On the shunt’s two current studs (th-eda) | DigiKey 298-10070-ND |
| Shunt sense terminals | Panduit PV18-8R-MY, #8 ring terminal, 18 to 22 AWG | 2 | $0.45 | On the shunt’s sense screws, for the 22 AWG Kelvin pair to the INA228 (th-eda) | DigiKey 298-10437-ND |
| Board inserts | PEM SI IUTB-M2.5, brass heat-set insert | 14 | $0.39 | For the Adafruit boards in the box, and four for the lid’s corners (th-eda, thrust-stand-ae’s CAD) | Mouser 153-IUTB-M2.5 |
| Board screws | Essentra 50M025045P006, M2.5 x 6 pan Phillips | 10 | $0.12 | The listing gives no material: check it on Mouser before ordering; nylon is preferred against the board pads (th-eda) | Mouser 144-50M025045P006 |
| DShot series resistor | Yageo MFR-25FBF52-47R, 47 Ω 1 % 1/4 W | 2 | $0.10 | In the ESC signal line; one spare (th-eda) | DigiKey MFR-25FBF52-47R-ND |
| Wetting resistor | Yageo MFR-25FBF52-2K2, 2.2 kΩ 1/4 W | 2 | $0.10 | ESC signal wetting pull-up, switch common to 3V3: bidirectional DShot idles high, and the ESC’s eRPM reply needs the released line held high (th-eda with thrust-stand-21, 2026-10-18; was a pull-down to GND); one spare | DigiKey MFR-25FBF52-2K2-ND |
| Arm-sense pull-up | Yageo MFR-25FBF52-1K, 1 kΩ 1/4 W | 2 | $0.10 | One spare (th-eda) | DigiKey 1.00KXBK-ND |
| Button LED resistor (SUPERSEDED: do not order) | Yageo MFR-25FBF52-150R, 150 Ω 1/4 W | 0 | $0.10 | The 270 Ω resistor at the end of this table feeds the ring LED from the SW8B’s WP3 in its place (th-eda with thrust-stand-21, 2026-10-18: the Serial Wombat SW8B now drives the buzzer FET, the ring LED and the TFT reset). Was: For the button’s red ring LED; one spare (th-eda) | DigiKey 150XBK-ND |
| Load cell screws, bracket | M4 x 12 socket head cap screw, A2 stainless, any maker (a local metric assortment) | 2 | local | Bracket into Ryan’s 10 kg cell: through the bracket’s 4 mm socket wall and 8 mm into the cell (thrust-stand-64’s thick-socket bracket). The cell is M4, 10 mm apart: Ryan test-fitted the printed bracket (2026-10-12). It was first read as M5 from a caliper across the threads, which cannot tell M4 from M5; check the length against the cell’s thread depth | |
| Load cell screws, pillar | M4 x 16 socket head cap screw, A2 stainless, any maker (a local metric assortment) | 2 | local | Pillar into the cell; M4 as above. Check the length against the cell’s thread depth (thrust-stand-ae) | |
| Pillar mounting screws | M4 x 30 socket head cap screw, A2 stainless, any maker (a local metric assortment) | 4 | local | Pillar to the 18 mm plywood board: heads counterbored from below, into the square nuts in the pillar’s four pockets (Ryan, via thrust-stand-64; was two screws into hex nuts). Fallback screw: Keystone 25045 titanium, $4.26 (th-eda) | DigiKey 36-25045-ND |
| OLED bezel screws (SUPERSEDED: do not order) | Essentra 50M025045I010, M2.5 x 10 countersunk, nylon | 0 | $0.14 | The 0.96 in display’s bezel is gone with it (thrust-stand-64) | DigiKey 145-50M025045I010-ND |
| Lid screws | Essentra 50M025045P008, M2.5 x 8 pan head, nylon | 4 | $0.14 | Into the lid’s corner inserts (th-eda) | DigiKey 145-50M025045P008-ND |
| Shunt screws | M4 x 16 countersunk, A2 stainless, with M4 nylon-insert lock nuts, any maker (the local metric assortment) | 2 | local | Clamp the 0.1 kg shunt to the box floor. STEEL, not nylon: the shunt heats to about 2.4 W at 49 A, and nylon creeps when warm (thrust-stand-ae). Nylon is only for the electronics side: the OLED bezel and the lid. Steel does not change a resistive shunt’s reading; keeping the Kelvin pair twisted is what matters (th-eda) | |
| Perfboard | BusBoard PAD1, 80 x 50 mm (not 70 x 50), FR-4 1.6 mm, double-sided plated, 2.54 mm pitch, 31 x 19 holes, four 3.18 mm holes on 71.12 x 40.64 mm centres | 1 | $4.15 | Carries the regulator, the capacitors, the TVS and the buzzer driver. The box CAD was drawn for 70 x 50: thrust-stand-ae checks the fit (th-eda, datasheet) | Mouser 854-PAD1 |
| ESC signal lead | DFRobot FIT0033, servo extension 300 mm, male to female (JR/Futaba) | 1 | $0.70 | From the box to the ESC under test (th-eda) | DigiKey 1738-1331-ND |
| Kelvin pair wire | SparkFun 08865 (red) and 08867 (black), 22 AWG stranded, 25 ft each; insulation not stated | 1 pair | $10.85 | The twisted Kelvin pair from the shunt’s sense screws to the INA228: $5.26 red plus $5.59 black (th-eda) | DigiKey 1568-08865-ND |
| OLED bezel nuts (SUPERSEDED: do not order) | Keystone 4687, M2.5 hex nut, nylon, 5.0 mm across flats x 1.8 mm | 0 | $0.27 | The 0.96 in display’s bezel is gone with it (thrust-stand-64) | DigiKey 36-4687-ND |
| XT90 mount screws | M3 x 10 with M3 nuts, A2 stainless, any maker (the local metric assortment) | 4 | local | Two per XT90 panel mount, nuts inside the box (thrust-stand-64) | |
| Display, 2.42 in (SUPERSEDED: do not order) | Adafruit 2719, 2.42 in 128 x 64 OLED, SSD1309, I2C at 3.3 V | 0 | $39.55 | Ryan changed the box display to his own 2.0 in TFT modules, row below (2026-10-18, via thrust-stand-21); not ordered. Was: The only 2.42 in SSD1309 module either distributor stocks (Mouser 212 in stock; DigiKey 0, $43.07). Not the Waveshare 2.42 in module: its TXB0108 level translator is not for I2C, on the bus the INA228 shares (th-eda). Caliper the received board before the lid and clamp bars print: Adafruit gives no board drawing (thrust-stand-64) | Mouser 485-2719 |
| Display clamp screws (SUPERSEDED: do not order) | M2.5 x 8 pan head, steel, self-tapping into the lid’s bosses | 0 | local | They held the 2.42 in OLED’s clamp bars, which go with it (thrust-stand-21, 2026-10-18). Was: Two per printed clamp bar, which presses the display board on its outer pads clear of the panel’s flex cable (thrust-stand-64) | |
| TFT reset pull-up | Yageo MFR-25FBF52-10K, 10 kΩ 1 % 1/4 W | 2 | $0.10 | Its role changes: the TFT’s RES pull-up, WP4 to 3V3 (th-eda with thrust-stand-21, 2026-10-18: the Serial Wombat SW8B now drives the buzzer FET, the ring LED and the TFT reset); one spare. Was the display’s power-on RC reset resistor | DigiKey 10.0KXBK-ND |
| Display reset capacitor (SUPERSEDED: do not order) | KYOCERA AVX SR215C105KAR, 1 µF 50 V X7R, radial, 5.08 mm leads | 0 | $2.46 | The RC reset is no longer needed (th-eda with thrust-stand-21, 2026-10-18: the Serial Wombat SW8B now drives the buzzer FET, the ring LED and the TFT reset). Was: The RC reset; one spare. The usual KEMET C320C105K5R5TA is out of stock at both distributors (th-eda) | DigiKey 478-SR215C105KAR-ND |
| Display reset diode (SUPERSEDED: do not order) | Vishay 1N4148-TAP small-signal diode | 0 | $0.26 | The RC reset is no longer needed (th-eda with thrust-stand-21, 2026-10-18: the Serial Wombat SW8B now drives the buzzer FET, the ring LED and the TFT reset). Was: Discharges the reset capacitor at power-off; one spare (th-eda) | DigiKey 1N4148-TAPCT-ND |
| Motor screws | M3 x 16 socket head cap screw, A2 stainless, any maker (a local metric assortment) | 4 | local | The motor under test to the bracket’s 12.95 mm plate, about 3 mm of thread; check against each motor’s thread depth before tightening, so no screw reaches the windings (thrust-stand-64) | |
| Pillar square nuts | M4 square nut, DIN 557 (7 x 7 x 3.2 mm), A2 stainless; a thin DIN 562 also fits | 6 | local | Four in the pillar’s base pockets (7.25 x 3.6 mm slots, open to the top face), two spares (thrust-stand-64) | |
| Thread locker | Medium-strength, removable (blue), for example Loctite 243 | 1 | local | On the M4 screws into the load cell, the motor’s M3 screws and the base screws at the square nuts (Ryan). Anaerobic thread lockers can stress-crack some plastics: put a small amount on the screw threads only, and keep it off the PETG. The aircraft’s blue thread-locker (hardware list) serves if it is the same type | |
| Display, 2.0 in TFT | AITRIP 2.0 in IPS TFT, ST7789V, 240 x 320, 4-wire SPI, VCC 3 to 5 V, pins GND, VCC, SCL, SDA, RST, DC, CS, BL; marked 2.0TFTSPI GMT030-03-8P ver1.21; 37 x 62 mm as listed (caliper one before the lid is drawn) | 1 | on hand | The box display in place of the Adafruit 2719 (Ryan, 2026-10-18, via thrust-stand-21). Ryan has a 5-pack; driver and pins from his listing. BL ties high: the module drives its own backlight through an onboard transistor (Q2, R6, R7) | Amazon B0H8N7MNRW |
| I2C coprocessor | Serial Wombat SW8B (PCB0029), I2C pin expander and coprocessor, 5-pack | 1 pack | $9.99 | Frees XIAO pins for the TFT’s SPI (Ryan, 2026-10-18, via thrust-stand-21): one goes in the box, four are spares. Price as Ryan’s listing shows it | Amazon B0FNQ8CD4Q |
| TFT backlight resistor (SUPERSEDED: do not order) | Through-hole resistor, value to set | 0 | to choose | SUPERSEDED (thrust-stand-21, 2026-10-18): the module has its own backlight transistor and resistor (Q2, R6, R7), so BL ties high and needs no resistor here. Was: in series with the TFT’s backlight LED, for th-eda to size | |
| Ring LED resistor | Yageo MFR-25FBF52-270R, 270 Ω 1 % 1/4 W | 2 | $0.10 | Feeds the button’s red ring LED from the SW8B’s WP3 (th-eda with thrust-stand-21, 2026-10-18: the Serial Wombat SW8B now drives the buzzer FET, the ring LED and the TFT reset); one spare. Mouser had 19,203 on 2026-10-18 (Part-DB’s Mouser lookup) | Mouser 603-MFR-25FBF52-270R |
| Buzzer FET gate pull-down | Yageo MFR-25FBF52-100K, 100 kΩ 1 % 1/4 W | 2 | $0.10 | Gate to source on the TN0702, so the buzzer stays off while the SW8B boots (th-eda); one spare. Mouser had 17,470 | Mouser 603-MFR-25FBF52-100K |
| TFT clock series resistor | Yageo MFR-25FBF52-33R, 33 Ω 1 % 1/4 W | 2 | $0.10 | In series in the TFT’s SCK line at the XIAO, for ringing (th-eda: 22 to 33 Ω, 33 the pick); one spare. Mouser had 4,066 | Mouser 603-MFR-25FBF52-33R |
| Buzzer FET gate resistor (optional) and I2C 3V3 series resistor | Yageo MFR-25FBF52-100R, 100 Ω 1 % 1/4 W | 3 | $0.10 | Optional series gate resistor for the TN0702 (th-eda); one spare. The third is in series in the SENSORS connector’s 3V3 pin, to limit a short on the exposed male pin (th-eda, 2026-10-19). Mouser had 52,054 | Mouser 603-MFR-25FBF52-100R |
| TFT screws | M2 x 5 self-tapping pan head, steel (M2 x 6 also fits), any maker (a local assortment) | 4 | local | The 2.0 in TFT into the lid’s four bosses at its M2 holes (58 x 30 mm): the 1.6 mm board plus up to 3.9 mm of thread in the 1.6 mm pilots (thrust-stand-21, 2026-10-18) | |
| ESC signal header | 3-pin straight male header, 2.54 mm, cut from any strip (a local assortment) | 1 | local | Held in the tub’s slot behind the right-wall ESC signal port (S + -), so the ESC’s servo extension plugs in from outside (thrust-stand-21, 2026-10-18) | |
| ESC signal jumper | 3-pin Dupont female-to-female jumper, about 10 cm (a local assortment) | 1 | local | From the header’s tails to the perfboard (thrust-stand-21) | |
| ESC servo extension | Standard 45 cm servo extension | 1 | on hand | Ryan’s own, from the box’s ESC signal port to the ESC under test; not ordered (thrust-stand-21) | |
| HX711 SCK pull-down | Yageo MFR-25FBF52-10K, 10 kΩ 1 % 1/4 W | 2 | $0.10 | Holds the HX711’s SCK low while the XIAO ESP32-S3 boots: the pin floats during boot, and the HX711 powers down if SCK stays high for more than 60 µs (th-eda’s pin-map review, via thrust-stand-21, 2026-10-18); one spare | DigiKey 10.0KXBK-ND |
| EPO comparator | Microchip MCP6547-I/P dual open-drain comparator, DIP-8 | 2 | $0.69 | EPO window circuit; one spare. ONLY 3 IN STOCK at Mouser on 2026-10-18: the backup with the same pinout is TI TLV3402IP (595-TLV3402IP, $4.92) (th-eda with thrust-stand-21, 2026-10-18, design in ../thrust-stand/docs/esp32-box.md; Mouser via Part-DB) | Mouser 579-MCP6547-I/P |
| EPO analog switch | TI SN74HC4066N quad analog switch, DIP-14 (one section used) | 2 | $0.89 | EPO window circuit; one spare (th-eda with thrust-stand-21, 2026-10-18, design in ../thrust-stand/docs/esp32-box.md; Mouser via Part-DB) | Mouser 595-SN74HC4066N |
| EPO delay capacitor | TDK FG14X7R1H225KRT06, 2.2 µF 50 V X7R, leaded, 2.5 mm | 1 | $0.61 | RC delay on the switch control; Mouser had 2,098 (th-eda with thrust-stand-21, 2026-10-18, design in ../thrust-stand/docs/esp32-box.md; Mouser via Part-DB) | Mouser 810-FG14X7R1H225KRT6 |
| EPO, ADC and IR 100 nF capacitors | Vishay K104K15X7RF5TL2, 100 nF 50 V X7R, leaded, 2.5 mm | 9 | $0.29 | SENSE filter, two supply bypasses, WP0 debounce, and the KEY RC debounce on GPIO44; one at each ADS1115 input, one across the MLX90614’s VDD and VSS pins, and a spare (th-eda, 2026-10-19; Mouser via Part-DB; th-eda with thrust-stand-21, 2026-10-18, design in ../thrust-stand/docs/esp32-box.md; Mouser via Part-DB) | Mouser 594-K104K15X7RF5TL2 |
| Encoder filter capacitors | Vishay K103K15X7RF5TL2, 10 nF 50 V X7R, leaded, 2.5 mm | 2 | $0.28 | Encoder A and B filters (th-eda with thrust-stand-21, 2026-10-18, design in ../thrust-stand/docs/esp32-box.md; Mouser via Part-DB) | Mouser 594-K103K15X7RF5TL2 |
| EPO, encoder and NTC 10k | Yageo MFR-25FBF52-10K, 10 kΩ 1 % 1/4 W | 13 | $0.10 | R_top, the cord’s R_id, the bypass plug’s R_id, the SENSE series to WP5, the EN read-back series to WP6, the RC delay, and the encoder’s 2 pull-ups and 2 series; the two NTC dividers’ top resistors (3V3 to the node) and a spare (th-eda, 2026-10-19; Mouser via Part-DB; th-eda with thrust-stand-21, 2026-10-18, design in ../thrust-stand/docs/esp32-box.md; Mouser via Part-DB) | Mouser 603-MFR-25FBF52-10K |
| EPO 40.2k | Yageo MFR-25FBF52-40K2, 40.2 kΩ 1 % 1/4 W | 2 | $0.10 | EPO window divider (th-eda with thrust-stand-21, 2026-10-18, design in ../thrust-stand/docs/esp32-box.md; Mouser via Part-DB) | Mouser 603-MFR-25FBF52-40K2 |
| EPO 19.6k | Yageo MFR-25FBF52-19K6, 19.6 kΩ 1 % 1/4 W | 1 | $0.10 | EPO window divider (th-eda with thrust-stand-21, 2026-10-18, design in ../thrust-stand/docs/esp32-box.md; Mouser via Part-DB) | Mouser 603-MFR-25FBF52-19K6 |
| EPO EN pull-up | Yageo MFR-25FBF52-100K, 100 kΩ 1 % 1/4 W | 1 | $0.10 | EN pull-up (th-eda with thrust-stand-21, 2026-10-18, design in ../thrust-stand/docs/esp32-box.md; Mouser via Part-DB) | Mouser 603-MFR-25FBF52-100K |
| EPO hysteresis (optional) | Yageo MFR-25FBF52-1M, 1 MΩ 1 % 1/4 W | 2 | $0.10 | Optional comparator hysteresis (th-eda with thrust-stand-21, 2026-10-18, design in ../thrust-stand/docs/esp32-box.md; Mouser via Part-DB) | Mouser 603-MFR-25FBF52-1M |
| EPO button | IDEC XA1E-BV302-R, 16 mm e-stop, 29 mm red mushroom, push-lock with pull/turn reset, 2 NC (both wired in series) | 1 | $49.89 | The hand pendant’s stop. Open: the contacts’ minimum switching load at the loop’s 0.17 mA is not confirmed (on thrust-stand-21’s check list; IDEC’s datasheet did not download) (th-eda with thrust-stand-21, 2026-10-18, design in ../thrust-stand/docs/esp32-box.md; Mouser via Part-DB) | Mouser 199-XA1EBV302R |
| EPO pendant box | Bud HP-3651-B hand-held box, 3.6 x 2.6 x 1.1 in | 1 | $3.20 | The pendant body, or a printed one (thrust-stand-21’s call) (th-eda with thrust-stand-21, 2026-10-18, design in ../thrust-stand/docs/esp32-box.md; Mouser via Part-DB) | Mouser 563-HP-3651-B |
| EPO cord gland | Amphenol AIO-BMCS-EN-2SL cord gland, M12 x 1.5, for 3 to 6.5 mm cable | 1 | $0.44 | At the pendant. If the cord measures under 3 mm, use AIO-BMCSBE-0SR (654-AIO-BMCSBE-0SR, $2.28, 2 to 5 mm) (th-eda with thrust-stand-21, 2026-10-18, design in ../thrust-stand/docs/esp32-box.md; Mouser via Part-DB) | Mouser 654-AIO-BMCS-EN-2SL |
| EPO gland nut | Amphenol AIO-BMCSBL-01 lock nut, M12 x 1.5 | 1 | $0.39 | For the cord gland (th-eda with thrust-stand-21, 2026-10-18, design in ../thrust-stand/docs/esp32-box.md; Mouser via Part-DB) | Mouser 654-AIO-BMCSBL-01 |
| EPO cord | TE / Raychem 44A0121-24-0/9-MX, 24 AWG 2-core, per foot | 7 ft | $1.55 | About 2.1 m from the box to the pendant. Ryan’s PJ31640 plugs (bought) go on it and on the labelled NO EPO bypass plug (th-eda with thrust-stand-21, 2026-10-18, design in ../thrust-stand/docs/esp32-box.md; Mouser via Part-DB) | Mouser 650-44A0121-24-09MX |
| Rotary encoder | Bourns PEC11R-4220F-S0024, 24 detents / 24 pulses, push switch, 20 mm flatted shaft, M7 x 0.75 bushing | 1 | $2.29 | In the lid (th-eda with thrust-stand-21, 2026-10-18, design in ../thrust-stand/docs/esp32-box.md; Mouser via Part-DB) | Mouser 652-PEC11R-4220F-S24 |
| Encoder knob | Davies Molding 1230-J, push-on, for a 6 mm shaft with a 4.5 mm flat | 1 | $1.81 | (th-eda with thrust-stand-21, 2026-10-18, design in ../thrust-stand/docs/esp32-box.md; Mouser via Part-DB) | Mouser 5164-1230-J |
| Arm keyswitch | Ryan’s own on/off keyswitch, 2 terminals | 1 | on hand | The arm interlock: the key powers the reed relay below, a second, independent cut in the ESC line, and its state is read on GPIO44 (th-eda with thrust-stand-21, 2026-10-18, esp32-box.md Safety; Mouser via Part-DB) | |
| Start and abort button, Ryan’s | Ryan’s own 18.5 mm panel pushbutton | 1 | on hand | In place of the Same Sky PBM2. Whether it has an LED is not known yet, so the 270 Ω ring LED row stays (th-eda with thrust-stand-21, 2026-10-18, esp32-box.md Safety; Mouser via Part-DB) | |
| Arm reed relay | Standex MEDER SIL05-1A72-71D reed relay, 1 Form A, 5 V coil, built-in diode, SIL | 2 | $3.52 | Its contact is the second cut in the ESC signal line, powered through the keyswitch; one spare. Mouser had 1,214. Coil resistance and pinout are ESTIMATES until checked against Standex’s SIL datasheet (about 10 mA at 5 V) (th-eda with thrust-stand-21, 2026-10-18, esp32-box.md Safety; Mouser via Part-DB) | Mouser 876-SIL05-1A72-71D |
| KEY divider, top | Yageo MFR-25FBF52-18K, 18 kΩ 1 % 1/4 W | 1 | $0.12 | The keyswitch sense divider into GPIO44 (th-eda with thrust-stand-21, 2026-10-18, esp32-box.md Safety; Mouser via Part-DB) | Mouser 603-MFR-25FBF52-18K |
| KEY divider, bottom | Yageo MFR-25FBF52-33K, 33 kΩ 1 % 1/4 W | 1 | $0.10 | (th-eda with thrust-stand-21, 2026-10-18, esp32-box.md Safety; Mouser via Part-DB) | Mouser 603-MFR-25FBF52-33K |
| I2C panel connector | Kinghelm KH-GX12-4P, GX12 4-pin set: panel socket with nut and cable plug | 2 | $1.72 | Labelled SENSORS: the I2C lead to the BME280 and the MLX90614; one spare. Its pin count differs from the TEMP connector’s, so the two leads cannot be swapped (th-eda with thrust-stand, 2026-10-19). Mouser has no GX12; LCSC had 94 on 2026-10-19 (Part-DB’s LCSC provider) | LCSC C18203262 |
| NTC panel connector | Kinghelm KH-GX12-3P, GX12 3-pin set: panel socket with nut and cable plug | 2 | $1.66 | Labelled TEMP: the two NTCs, at the ESC and the motor mount; one spare (th-eda with thrust-stand, 2026-10-19). LCSC had 83 on 2026-10-19 (Part-DB’s LCSC provider) | LCSC C18203261 |
| Temperature ADC | Adafruit 1085, ADS1115 16-bit 4-channel ADC breakout (STEMMA QT) | 1 | $14.95 | I2C 0x48 (ADDR to GND, the default): AIN0 the ESC NTC, AIN1 the mount NTC, AIN2 the 3V3 rail for a ratiometric read, AIN3 spare. Its 10 kΩ I2C pull-ups, with the BME280’s and the box’s, come to about 3.3 kΩ, which is correct at the bus’s 100 kHz (th-eda, 2026-10-19; Mouser via Part-DB; 550 in stock) | Mouser 485-1085 |
| Bell IR thermometer | Melexis MLX90614ESF-BCF-000-TU, IR sensor, 3 V, 10° field of view, gradient compensated, TO-39, no breakout | 1 | $31.13 | Reads the motor bell: at 20 mm the 10° BCF sees a spot about 3.5 mm across, where the usual 90° BAA ($10.15) sees about 40 mm, which takes in the prop hub and the bench. Gradient compensation, because the motor heats the sensor’s can unevenly. Wired direct to the SENSORS lead, with a 100 nF at its pins (row above) (th-eda, 2026-10-19; Mouser via Part-DB; 442 in stock) | Mouser 482-90614ESFBCF000TU |
| Air pressure, temperature and humidity | Adafruit 2652, BME280 breakout (I2C/SPI) | 1 | $14.95 | I2C 0x77 (the default), outside the box in still, shaded air, for air density (th-eda, 2026-10-19; Mouser via Part-DB; 782 in stock) | Mouser 485-2652 |
| NTC thermistors | TDK B57861S0103F040, 10 kΩ 1 % NTC thermistor, B 3988 | 3 | $1.75 | One at the ESC, one at the motor mount, and a spare; on the TEMP lead to the ADS1115 (th-eda, 2026-10-19; Mouser via Part-DB; 41,821 in stock) | Mouser 871-B57861S103F40 |
thrust-stand-ae’s full sourced list, with dimensions, is ../thrust-stand/docs/box-bom/thrust-box-bom.md.
Corvon / Aero Selfie H743 instead of a wing board
Section titled “Corvon / Aero Selfie H743 instead of a wing board”This comparison was written against the Kakute H743-Wing when it was the board in the cart. The Lucid H7 Wing is at least as good on every row.
| CORVON743V1 ($49.99 FC only) | Kakute H743-Wing ($82.99) | |
|---|---|---|
| Servo power | No servo rail; 5V 3A shared with FC, RX, GPS | Dedicated 6V/8V rail, 6A |
| Current sensor | None; reads the 4-in-1 ESC pin only, pusher draw invisible | Onboard 90A cont / 220A peak |
| UARTs | 7 | 7 |
| I2C for airspeed | Yes (SDA/SCL pads) | Yes |
| Outputs | 10 in groups 1-4 / 5-6 / 7-10 | 14 |
| Voltage sense | To 6S | 3-8S |
The Corvon is acceptable for a basic 6S build under three conditions. The servos go on outputs 7 and 8. The 5V budget is respected. Current logging and current-based failsafes see only the lift ESC. The $33 saving buys a board without a servo rail or a real current sensor.
Motor bolt pattern, resolved
Section titled “Motor bolt pattern, resolved”Both printed mounts have four M3 holes on a 13.44mm square, 19.00mm corner to corner. The “19x19” or “19mm” in motor vendors’ data is the bolt-circle diameter. Thus every recommended motor fits.
Three sources confirm this. GetFPV’s motor guide says “the diameter of a circle with its centre placed in the middle of the motor shaft”. OpenELAB’s V3115 sheet says “19 mm mounting circle”. SpeedyFPV’s ECO II 3115 sheet says “19mm bolt circle”.
The EMAX ECO II 3115 has conflicting reseller data (one says 16x16). EMAX’s own dimension drawing settles it: 4 x M3 on a 19 mm circle. It is the pusher in the cart (2026-09-27).
Caveats
Section titled “Caveats”- The V3115 640KV is sold out at every US shop checked (2026-09-27), so the cart has the ECO II 3115 640KV, in stock at RMRC, EMAX USA, RaceDayQuads and GetFPV.
- Every DroneCAN airspeed sensor was short on 2026-09-26. Holybro direct had the L10D board and the PT40 pitot in stock (Hong Kong, US tariff applies) and the L20D bundles, which are the wrong range for this aircraft. Matek’s ASPD-DLVR ($110.32) and ASPD-AUAV ($113.35) were on backorder at RMRC and sold out at WREKD. The US-made mRo unit was $136.90 with a 3 to 4 week lead.
- The TBS Lucid H7 Wing is on backorder at RMRC (October 2026) and out of stock at TBS, and it has one CAN bus. The cart had the Pixhawk 6C Mini instead (2026-09-27), had the NxtPX4v2 on 2026-10-04, and has the LUX F765 for the first flights since that day. Earlier fallbacks, the Kakute H743-Wing and the BLITZ Wing H743, are single-IMU boards.
- The lift ESC is the list’s TBS Lucid 4-in-1, 6S version (Ryan, 2026-09-29). The single DroneCAN ESCs at the pods (decided 2026-09-27) are the follow-on, and the airframe takes either.
- The elevon servos are Hitec MD65MG, not the list’s ES08MD II (decided 2026-09-27). Hitec CS sells at most 20 of one part number online.
Source: docs/parts-list.md in the repo. Edit it there; this page is generated.
Comments
Comment on the whole page here, or on one section or step with the button beside it.