ive got a couple ideas floating around upstairs. the cheaper of the two is a 3 axis sim with 360 degree yaw radias and id say 90 degrees for pitch and roll. the yaw axis frame rotates on rails which are elevated above the floor and is driven by wheels by means of a gear motor, a simple digitaal compass provides the sensor here. the frame is a hexagonal ring about 6 feet in diameter and made from steel tubing. single wheel assembalys mounded on all 8 sides, 2 of which are driven by motors. the inside is left free for the pitch frame to rotate
your pitch axis frame which fits inside the yaw frame. it is mounted on bearings. the bearing rods are fixed to the yaw frame while the bearings are in the pitch frame. a gearwheel is fixed to the inside of the braring rod, gear motors with worm drive is mounded in such a way as to rotate the frame. the sensor emplyed here is a simple ball baring in a wire ring sensor (which can be made fairly easiy). this frame would also house the bearings motors and worm gears that articulate the roll frame.
the roll frame is a rectangular frame about 5 feet in length. the plot seat is positioned slightly behind the center of gravity. the computer, panel, controls, and screen(s) will be mounted in front to balence it out. fixed bearing rods and gears mount in the pitch frame for articulation. a rotation sensor like the one used in the pitch frame would also be used.
the computer would use a solid state ram based hard drive, which should survice being tossed around. a semi-decent video and sound card. id make it powerfull enough to run the game im building the simulator for. in the case of freespace 2 that means a gig of ram and a good 2ghz processor. it would need to have enough io ports to control the 3 sets of motors and read the 3 sensors. for screens a fairly large flat panel would do. however if the game supported multi monitors id use a pair of sli cards to drive 4 displays, which would be aranged in a hood with screens positioned to simulate front side and up windows for a virtual canopee. id use ch stick, rudder, and throttle. or sence saitek is finally coming out with rudder pedals that match my x-52 id use those.
depending on the torque of the motors being used, it might be nessicary to make the machine ultra balenced, meaning an adjustable pilot seat which could slide forward or back. this simulator could potentially hafe a full 360 degrees of rotation on all axes if you could solve the problems of running power/signal data through rotational couplings. you could just couple the power and piggyback the signal data onto the current (the same technology is used to set the clocks in many an office building).
--or--
this is a more exotic idea i just came up with. get a big sphere and hover it using either an air cushion or magnets. and at the 6 corners of the sphere have a "thruster quad" essentially 4 nozzels pointing on each direction along the plane prependicular to each axis. this gives you essentially 6 sets of 4 thrusters to controll all 3 axes of rotation. you drive theese using tanks of compressed air inside the sphere. this simulator is self contained, it requires several tanks of compressed air (which are "charged" by an external air compressor pump). your computer would also need to be on batteries (an alienware laptop
) as theres no way to connect power and keep the sphere freely rotatable. also balence is critical, the pilot seat would need to be fore/back and up/down adjustable so that balence may be tweaked to perfection. to hover the thing magnets would be the cooler solution, as they dont make any noise and would interfere less with the rcs thrusters. an ir cushion would be the easy way, you just put your sphere in a ring with a seald base, and use about 6 carpet blowers to provide the thrust.
