You want 200 microns collimated beam ? Well .. my optics knowledge is pretty basic, but you need convex lens to focus it into point .. and then concave lens to expand it again into parallel beam. Focus points of both lens should be in the same point (which will be on the opposite side of the lens then the light source). For low spherical aberration, the lens should have one planar side, which will be inside the couple, with curved side toward collimated beam on both sides.
Now that all is just theory, I have no idea if 0.2mm is indeed achievable with this. Also such beam will have large divergence due diffraction.
Edit: now I realized the lens strength will need to be in the same ratio as the beam size. So if you want to go from 4mm to 0.2 mm, to concave lens will be 20 times stronger (shorter focal distance) then the convex lens. Then you place them so the focal point is in the same place, and that's it.
Btw. What do you use for the simulation ?