PLASMADRONE

Prototype delivery tool · PD-1 · Rev A · 2026-10-18

PD-1 Saucer

A 3–5 ft hobby flying saucer built from parts you can buy today. One electric coaxial rotor does the lifting. A cold-gas ring at the rim does the fast corrections. The high-pressure tank is a certified paintball COPV, filled on the ground. The intelligence is trained on a ground station and flashed into an embedded chip.

Phase-1 goal: stable hover, vertical climb, side-to-side translation — on a tether, then free.

The mission: deliver what matters, and land it alive. When the cargo is plasma or rescue gear, payload survivability outranks delivery proximity — if power fails, PD-1 autorotates to the safest reachable spot and spends every stored joule on one soft touchdown, even when that spot is farther from the receiver.

PD-1 in hover. Rotor lifts, rim ring corrects, body stays level.

48 in

Baseline Ø (36–60 scalable)

7.4 kg

All-up weight target

~71%

Hover throttle

Recessed gas stations (8–16 by size)

~8,000

Pulses per tank fill

01 · Concept

Three jobs, three systems

Everything in PD-1 follows one rule from the research: decouple lift, attitude, and recharge. Each job goes to the physics that is actually good at it.

LIFT — ELECTRIC

Coaxial rotor core

Two contra-rotating 18 in propellers in a central 19 in duct carry 95%+ of the weight. Contra-rotation cancels the spin torque, so no tail rotor and almost no yaw fight. The spinning mass also adds gyroscopic stiffness that resists tip-over.

ATTITUDE — PNEUMATIC

Cold-gas rim ring

A 120 psi ring manifold at the outer rim feeds recessed solenoid nozzles — 12 on the 48-in baseline, 8 on the 36, 16 on the 60 — every port flush with the skin so the hull stays smooth. Torque is force × radius, and the rim is the longest lever arm on the craft: millisecond pulses snap the saucer level using almost no air.

RECHARGE — ON THE GROUND

Certified COPV, filled between flights

A 48 ci / 4,500 psi carbon-wrapped paintball tank is the air battery. It refills in minutes from a hand pump or SCBA fill station. No onboard compressor in Phase 1 — see the physics note below.

Physics check — why no turbine top-off

A hobby turbine or EDF reaches a pressure ratio of about 1.2:1 (a few psi). Topping off a 4,500 psi tank needs over 300:1 — the tank's check valve simply wins and the flow reverses. A motor-driven multi-stage piston pump can do it, but the turbine→generator→motor→pump chain multiplies out to roughly 3.7% efficiency and more dead weight than the air is worth. So PD-1 treats the tank as a battery you swap-charge on the ground, and keeps a piezo micro-compressor trickling a low-pressure accumulator on the Phase-3 research list — low pressure only, never the main tank.

02 · Preliminary drawings

Orthographic views

Baseline geometry at Ø48 in (1,220 mm), 13 in tall at the hub. The same layout scales from 36 to 60 in.

PD1-100 · Top view. Eight rib bays; tank and battery pods oppose each other across the duct so the center of gravity stays on the rotor axis. Eight down-firing nozzles (squares) handle pitch/roll and four tangential (arrows) handle yaw — 12 stations on this 48-in baseline, 8 on the 36, 16 on the 60, every port recessed flush with the skin.

PD1-110 · Side section. The whole duct tilts ±12° on a two-servo gimbal to vector thrust for forward / back / sideways travel while the body stays level — the helium annulus and sensors never leave the horizontal plane.

PD1-120 · Bottom view. Downwash exits the central annulus; landing pads sit outside it so ground effect doesn't tip the craft. The fill port and electronics hatch are reachable without opening the hull. Every nozzle port is recessed — nothing protrudes below the skin.

PD1-130 · Exploded view. Six assemblies, all serviceable: the shells lift off, the He annulus is a drop-in ring, the core chassis carries every heavy component on the rotor axis, and the valve ring unbolts as one unit with its manifold.

Detail drawings

The three features the prototype lives or dies on, drawn to be built.

PD1-210 · Gimbal detail. Two nested axes, ±12° each, one 60 kg·cm HV servo per axis. Hall sensors on the pivots close the loop so the policy knows the true duct angle, not the commanded one.

PD1-220 · Valve station. One per station — 12 around the 48-in baseline rim (8 on the 36, 16 on the 60). Each nozzle sits in a recessed well with its exit flush to the skin: nothing protrudes and the hull stays smooth. Normally-closed spring return: power loss means the air stays in the tank.

PD1-230 · Duct section. The two numbers that set hover efficiency: a rounded inlet lip (a ducted rotor gains static thrust from it) and a tight, constant tip gap — which is why the duct is rigid and the rotors are balanced before the shell ever goes on.

03 · Variants

Permutations

One architecture, explored across three sizes and four configurations. The tank, ring, valves and avionics carry over unchanged between all of them — only the airframe and lift stack rescale.

PD1-400 · Size permutations, drawn to a common scale. Helium only starts paying rent at 60 in — exactly what the lift-per-liter math predicts. The recessed rim ring (diamonds) is standard on every size and scales with it: 8 nozzles at 36 in, 12 at 48, 16 at 60. Every size has a job: the Scout's small payload is still enough to drop a transceiver or locator beacon so the delivery party can establish comms before the heavy lift arrives; the Baseline flies camera and training sorties; the Cruiser hauls the cargo that must land alive.

PD1-410 · Configuration permutations. The highlighted element is what changes; everything drawn in neutral — including the recessed rim control ring on every model (8 nozzles at 36 in, 12 at 48, 16 at 60) — carries over between variants.

PD1-420 · Variable-pitch hub & capacitor-discharge landing — Cruiser-60 only. A Sikorsky-style collective tilts the 24-in blades to chop more or less air at constant RPM. The failure case is why it earns its mass: the instant the flight computer sees pack voltage collapsing it flattens the blades into autorotation — the windmilling rotor back-drives the motors and every remaining watt charges a supercapacitor bank. Baro, GPS and IMU telemetry time the one discharge that matters: at flare height the cap dumps into the motors as the collective bites, trading rotor inertia and stored charge for a touchdown at minimum impact acceleration. Landing policy is payload-first: carrying plasma or rescue gear, the computer steers the glide to the safest reachable spot even when that is farther from the receiver — payload survivability outranks delivery proximity. The 36 and 48 stay fixed-pitch — their rotors don't store enough energy to buy the flare.

Variant	Vectoring	Attitude	Helium	Status
V-A Ringflat	none — fixed duct	recessed nozzles do everything	no	first tether article
V-B Vector	servo gimbal ±12°	ring assists	optional	baseline · Phases 1–2
V-C Magring	magnetic rim bearing	ring assists	yes	Phase-3 R&D
V-D Cruiser-60	EM rim gimbal (servo fallback)	ring assists	full 400 L annulus	Phase-3 concept

Variant details — flight-sequence animations, EM rim gimbal concept drawings, mission profiles →

04 · Systems

Pneumatics

Store high, regulate low, spend in milliseconds. Two regulator stages take the certified tank's 4,500 psi down to a 120 psi working ring, which the flight computer spends in 5–20 ms solenoid pulses.

PD1-310 · Air path. Working numbers at 120 psi through a 1.5 mm nozzle: ≈1.8 N thrust per station, ≈1.0 N·m of leveling torque at the 24 in lever arm, ≈34 mg of air per 10 ms pulse — roughly 8,000 correction pulses per fill from the 290 g of air the tank holds.

Electrical

One 6S pack feeds everything through a power distribution board; three BECs make the 8.4 V servo rail, the 12 V valve rail and the 5 V avionics rail. Signals are the dashed lines — every protocol is stock drone wiring.

PD1-320 · Electrical schematic. Solid lines are power, dashed are signal. The valve rail is fused separately from the lift rail, so a stuck valve can never brown-out the motors.

Concept — perovskite solar skin, the recharge that comes with sunrise

The mylar hull is also the biggest flat surface on the craft, and flexible perovskite photovoltaic film laminates onto exactly this kind of taut substrate: a perovskite-impregnated skin whose faint cell grid sits under the aluminum-doped finish without adding wrinkles or protrusions. The payoff is survivability, not cruise power. If conditions force a variant to spend every stored joule on a safe landing, the skin trickle-charges the pack where it sits — hours, not minutes — until the craft can resume its mission or return to base on its own. A downed PD-1 with sun on its back is not a recovery job; it is a delayed sortie. Phase-3 research item: cell-grade perovskite film is only now reaching hobby hands, so the laminate flies after the plain skin does.

Brains: train on the ground, fly on a chip

The fast loop never waits on a big model. An adversarial reinforcement-learning setup on the ground station (a simulated "disturber" throws gusts and dropouts at the flight policy until it stops crashing) trains a small network; its weights compile to plain C and flash onto the flight controller, which runs the 1 kHz attitude loop. The LLM supervises from above — mission goals, tuning, telemetry review — never inside the control loop.

PD1-330 · Control stack. One-way trust: the trained policy flies; the Pi suggests at 50 Hz; the ground LLM only ever tunes and reviews. If every link drops, the flight controller alone can land it.

Buoyancy assist — the honest math

Helium lifts about 1 g per liter. The sealed annulus in a 48 in PD-1 holds ≈170 L → ≈170 g of lift, about 2.3% of all-up weight — it offsets the avionics, not the airframe. At 60 in it grows to ≈400 L / 400 g and starts earning its keep. So the bladder is a Phase-2 option: the bays are reserved in the structure, nothing in Phase 1 depends on it.

Design rule carried through every phase

The saucer body never tilts to travel. The gimballed duct vectors thrust while the hull stays level — so the helium volume stays balanced, the disc never catches wind like a sail, and the state estimator works on a flat plane. Phase-3 R&D may replace the servo gimbal with the magnetic rim bearing from the research notes (electromagnet quadrants on the duct wall, filament ring on the fan shroud); the servo version flies first because every part of it is a catalog item.

05 · Mass budget

7,410 g, accounted for

Assembly	Contents	Mass
Airframe	EPP ribs, carbon rods, film skin, duct, core chassis	2,170 g
Lift stack	2× U8-class motors, 2× 18 in CF props, 2× 80 A ESC	820 g
Vectoring	Gimbal frame, 2× HV servos (60 kg·cm)	360 g
Energy	6S 10 Ah Li-ion pack	1,250 g
Pneumatics	COPV 48 ci + stage-1 reg, stage-2 reg, ring + tubing, 12 valves, recessed nozzle wells	1,940 g
Avionics	FC, GPS, flow/ToF, ELRS RX, companion Pi + cam	220 g
Power & wiring	PDB, BEC, harness	330 g
Ground interface	Landing gear ×4, hatch hardware, misc + margin	320 g
All-up weight	coax static thrust ≈10.5 kgf → hover ≈71% throttle	7,410 g

Scaling: a 36 in build drops to ≈5.5 kg on 15 in props; a 60 in build runs ≈9.5 kg on 24 in props and gains the helium margin above.

06 · Bill of materials

Everything is a catalog part

No custom pressure vessels, no machined one-offs. Printed PETG brackets and foam/carbon airframe are the only fabricated pieces.

Part	Spec	Where it comes from	Est.
Carbon HPA tank + reg	48 ci / 4,500 psi, DOT/TC certified, 850 psi output	Paintball (Ninja, First Strike)	$180
Stage-2 regulator	Adjustable 60–150 psi (SLP airsoft/PCP type)	Airsoft HPA / PCP airgun	$70
Solenoid valves ×12	2/2 NC, 12 V, fast-acting, 1/8 in ports (8 on the 36, 16 on the 60)	Pneumatics suppliers	$195
Ring manifold	6 mm nylon tube loop + push-fit tees, printed recessed nozzle wells	Pneumatics suppliers	$50
Lift motors ×2	U8-class ≈170 KV, 6S	Multirotor hobby market	$340
Props 2×18 in + ESC 2×80 A	CF folding or rigid, one CW + one CCW	Multirotor hobby market	$170
Gimbal servos ×2	HV digital, ≥60 kg·cm	RC hobby market	$85
Flight controller	STM32H7 (Matek H743 / Pixhawk 6C mini)	Drone electronics	$90
GPS + flow + ToF	M10 GPS, optical-flow + lidar module	Drone electronics	$75
RC link + companion	ELRS RX, Raspberry Pi Zero 2 W + cam	Drone electronics	$55
Battery	6S 10 Ah Li-ion (21700 pack)	Battery vendors	$190
Airframe stock	EPP sheet, 3 mm CF rods, covering film, PETG spools	Hobby / maker suppliers	$160
Fill gear	4,500 psi hand pump (or dive-shop SCBA fills)	PCP airgun market	$130
Prototype total	plus wiring, hardware, spares		≈ $1,790

07 · Build roadmap

Four phases, each one flyable proof

Phase 0 — Bench. Core chassis on a thrust stand: map coax thrust vs throttle, characterize valve response (open/close latency, thrust per pulse), leak-test the 120 psi ring for 24 h. Nothing flies until the numbers match this page.
Phase 1 — Tether. Full airframe on a 3-point tether rig. Goals, in order: stable hover → vertical climb/descent → side-to-side translation. PID first, then the ground-trained RL policy. This phase is the deliverable.
Phase 2 — Free flight. GPS position hold outdoors in calm air, helium annulus installed and trimmed, RL policy flying full-time with PID fallback.
Phase 3 — Research. Magnetic rim bearing to replace the servo gimbal; piezo micro-compressor trickling a low-pressure accumulator; LLM supervisor moved onboard (Jetson-class). Each item lands only if it beats the part it replaces on weight and reliability.

08 · Safety rules

Non-negotiables

Never fabricate or modify a pressure vessel. Certified tank only, in-date hydro stamp, factory burst disc in place. A home-wrapped COPV fails explosively.
120 psi is the ceiling downstream of stage 2. Every ring component is rated 150 psi+; the stage-2 relief vents before the tubing does.
Mesh over both duct openings. 18 in blades at hover power take fingers. The intake and exhaust get guards before the first spin-up.
Tether until Phase 2. Three-point tether, hard throttle ceiling, kill switch on the transmitter and on the ground station.
Failsafe is passive-safe. Link loss: valves spring closed, gimbal centers, motors ramp to controlled descent. LiPo charging and storage per standard RC practice. No flight over people, ever.