Figure 2: At left, principle of recombining optics for N collimated coudé beams. In order to avoid the differential field rotation otherwise caused by the coudé trains, the complete system, except the flat mirrors FM, rotates rigidly around point C with its axis remaining pointed towards the observed star. A simple mechanical linkage ML rotates the flat mirrors FM at half the pointing rate. They must also be translated to match the ellipticity of the telescope ring. Concave mirrors CM bring all beams to a common focus F, relayed by mirror RM, which is located close to the pupil DP. Unlike the elliptical ring locus of the entrance telescopes and mirrors FM, both the entrance pupil and the densified exit pupil DP remain circular. RM can be segmented and equipped with piston actuators for adaptive phasing. Each segment can also be translated axially together with the corresponding mirror CM, in order to stabilize the sub-pupil's axial position while the primary collecting telescopes (not shown) are being translated for optical path correction. The recombined image enters the wave analyzer WA and the spectro-imager SI. If needed for convenience, a small rotatable flat mirror can be installed in C to provide a fixed coudé output. Risley prisms RP (front view PF) with adjustable equivalent angle and thickness correct the atmospheric dispersion for respectively the diffraction and the interference functions.

At right, a more compact recombiner uses, to correct field rotation, a small convex paraboloidal recombination mirror (inset) which is rotated around its focus to keep its axis aimed in the star direction.