This is what I have found:
[ame="http://en.wikipedia.org/wiki/Control_moment_gyroscope"]Control moment gyroscope - Wikipedia, the free encyclopedia[/ame]
The links to an MIT paper and the 6154691 us patent look promising. On the todo list - power requirements (I never trust wiki, it promises "a few hundred watts").
EDIT: L3 DGCMG (of the ISS vintage) specs (assume they are state-of-the-art, or 5-10% behind)
design life: 10 yrs, angular momentum 4760 Nms, nominal speed 6600 rpm,
MAX output torque: 258 Nm
MAX gimbal rate: +-3.1 deg/sec
Size: 1.3x1.37x1.25, mass: 272 kg (funny, but the spec sheet says weight)
Alas, no electricity consumption numbers in the datasheet, will attempt scouring through the ISS manuals etc.
---------- Post added at 03:09 AM ---------- Previous post was at 02:26 AM ----------
Here goes Honeywell M50 (the data are of course sanitised so higher perf is clearly possible, plus it's quite light):
http://www51.honeywell.com/aero/com...og-documents/M50_Control_Moment_Gyroscope.pdf
Ang.momentum 25-75 Nms
Speed: 4500,5000,5500,6000,6500,XXXX rpm
Gimbal rot range: 360 deg
bandwidth: 10 Hz
gimbal acc: 3 rad/s^2
Output torque: <0.075 to 75 Nm
POWER CONSUMPTION:
Standby 11 W
Run-up 113 W
Peak torquing 95 W
Quiescent 35 W
Mass: 28 kg