In my engineering experience, the best efficiency comes from fewer, larger blades rotating at slower speeds (all of which reduce turbulent flow). The actual driving force of this device comes from a very small, very fast, multibladed propeller.

I soon thought out a lot of frills, like applications and its attributes in combat,but unlike the picture of the model, I was thinking in terms of a ramjet at one end. I also considered having a jet at each end, but rejected it in favour of a jet at one end and controls, camera, weaponry etc at the other. I was uncertain about the practicality of controlling such a device, and suspected that the only practical means would be by warping the surface and adjusting the angle of the engine. piezoelectric control sounded marvellous, but I did not think that it would be practical. Then I saw on television that a professor of aircraft construction in some southern state (Arizona, Texas, New Mexico,??? Can't remember) was in fact applying such controls experimentally. I e-mailed him, and he was quite encouraging about my idea but whether he took it any further I cannot say. Certainly I could do nothing of the kind, not having any skills that way. Anyone interested in discussing some of my initial ideas about the attributes and possible applications can e-mail me.
Jon Richfield

There are a few drawbacks to this, of course - the camera (and any RF video link) will have to be mostly self-contained and run off their own power source, since there's a pivot between them and the flight batteries, and the mechanical complexity (and therefore durability) is somewhat higher than with the computer-witchcraft approaches. Still, in at least some applications it seems like a reasonable price to pay for the simpler implementation.

Hm, on second viewing of the big model I guess I overestimated the RPM. I was thinking a servo wouldn't be fast enough to twist the blade up and down once every turn, but at about 4 rotations per second it should be doable.

If I were to make a small one, I'd probably try an electromagnetic "servo", as seen on some cheap foam parkflyers. You could adjust the effective control throw by varying the duty cycle of the magnet. An IR sensor could detect a stationary IR source (maybe on the transmitter) to sense rotation!

Now I want one!