RogerBW's Blog

I'm trying to learn more about practical multirotor design with a view to designing and building my own. A multirotor flies essentially by replacing mechanical complexity with software complexity, which is lighter. Here's a bit more detail.

A conventional helicopter is mechanically complex: the angle at which each blade cuts through the air must be varied both during each rotor revolution (cyclic pitch, to control the aircraft's pitch and roll) and over the longer term (collective pitch which varies thrust, feeding into yaw with the tail rotor), and this is achieved by a mechanical linkage, the swashplate. Many small parts have to move perfectly and repeatedly.

A multirotor, by contrast, can use single-piece propellers: a quadrotor with direct drive has just four moving parts, the motor-propeller assembly for each rotor. The important factor is that half the rotors rotate in each direction. So a pitch forward is achieved by increasing power to the rear rotors; a roll left increases power to the right rotors; and a yaw right increases power to the anticlockwise-spinning rotors, which to avoid control linkage are a diagonally opposed pair.

The minimal design uses three rotors, but there are obvious asymmetries. Four works pretty well; six and eight cut down the required power per rotor for heavier aircraft. (Some of the larger ones use three or four arms and mount a pair of rotors on each one in a push-pull configuration, but generally it's one rotor per arm.)

This was actually tried quite early in the development of helicopters. The Breguet-Richet Gyroplane of 1907 was the first example, indeed the first time a rotary-wing aircraft had lifted both itself and its pilot off the ground, though records are unclear as to whether it ever achieved free flight; in its early tests it needed several men on the ground to hold it steady.

Etienne Oehmichen's Helicopter No. 2 was more successful though a less "pure" design: it had four rotors on the ends of an X-shaped frame, like the Gyroplane, but also five propellers for horizontal stabilisation, one in the nose for steering, and two more for propulsion. In spite of this, it was moderately successful, establishing the first FAI distance record for helicopters (360m in 1924). George de Bothezat also developed a helicopter in the 1920s, also using separate propellers for control; it never exceeded 5m above ground level.

Going back to pure quadrotors, the 1950s saw the Convertawings Model A and the Curtiss-Wright VZ-7; both of them flew reasonably well, though not sufficiently better than conventional helicopters for anyone to try switching.

What really makes the multirotor concept viable now is electronics, specifically cheap and light gyroscopes and accelerometers. Every little variation in thrust or air density leads to an instability and potential toppling, so the on-board electronics need to be able to tweak the power to each motor very quickly in order to keep the platform stable. External commands (whether from a radio control unit or an onboard navigation processor) take the form of "rotate right" or "tilt forward", and these are combined with position and orientation feedback to generate control outputs to the individual motors.

The multirotor concept can certainly be scaled up; Bell-Boeing is working on the Quad TiltRotor, a derivative of the Osprey that would have roughly the speed and lift capacity of a C-130, but be able to land in its own length. It's not clear to what extent this will need control surfaces, though in aircraft-mode flight they seem likely to be important; I suspect it would also need variable-pitch propellers for efficiency, which starts to bring mechanical complexity back in.

I could easily see a future generation of attack helicopter using this technology: the simplicity (suggesting increased ease of maintenance and even resistance to damage) might well be worth any slight drop in performance. Of course, with current trends, it probably won't need room for a pilot.

One could do the same thing with jet engines, but there are two problems: jets aren't as immediately responsive to small changes in power, so stabilisation would be more work, and they don't produce useful amounts of torque, so you'd still need some other sort of control system for yaw.

Apart from simplicity, the multirotor can use smaller propellers: a quadrotor needs only about half of the blade diameter of a single-rotor aircraft in order to have the same total rotor disc area, though they generally have to spin faster. This still lets individual rotors be lighter and have less kinetic energy, helpful in case of a collision.

What I'm trying to learn at this point is a way of calculating how much thrust I can get out of a given combination of motor and rotor, and therefore how much power I need to feed in…