I did a new simulation extending the optical path (of 45 cm), using 4 additional mirrors (fig 1).

The best simulation is shown in fig 2 where the first lens is 47.5 cm far from PBS and the second one is 86.6 cm far from PBS, both with the same focal of 1020 mm.

This solution is more robust than the previous one (entry #1283)  as shown in fig 3 and fig 4. Moving the lenses of +/- 1 cm we have a mismatch < 10%.

Next step:

- Simulation for the FC reflected beam (first idea in fig 5).