Walterk, you're right, ... but I can't resist (Ill keep it short):
Saabluster, I'm not a professor (no one said so (except you)), but I have a doctor's degree in experimental physics (related to optics). I bring up arguments from Ra, because they are correct, also arguments I learned in university, for the same reasons. All that stuff isn't actually new. Furthermore, gcbrian and I know each other already from another forum.
Back on topic, beam expanders:
For beam-expanders / telescope / eyepieces (working all the same ) there are several principles based on plain convex lenses, all effective.
Yes, eye pieces can be very complex, mostly for two reasons: avoiding chromatic aberration and getting a good field of view with low image distortion. For expanding a collimated beam, the latter is often unimportant, because the beams have rather small angles to the optical axis already, so a simple setup often suffices, although the use of achromatic lenses (duplets) is advisable for white light.
Back to pre-collimating:
I present my newest super-thrower, the
Sloppy270:
I guess you can see why it got that name.... and I didn't even apply a battery yet (and probably never will; on the other hand, I might take it out for some field test...)
There are three versions:
ver1: no pre-collimator,
ver2: low-NA (high-f#) pre-collimator, f=150mm
ver3: higher-NA (lower-f#) pre-collimator, f=150mm
The spot brightness, measured at 18.3m, is nearly the same for all three, it's a bit lower for ver3 because the pre-collimator is only a spheric lens with quite some aberrations at that NA (or f#).
While the spot brightness is roughly the same, with the pre-collimator the spot size increases (5cm/10cm/15cm).
And of course a bigger spot with the same illuminance (lux) means more flux (lumen).