Probably, you already tried to implement the Mach-Zehnder interferometry scheme, but in case you didn't, you might think in this direction. In this scheme only one beam-splitter can be used, and interference is observed by precise positioning two mirrors. For example, you can look for a setup in this article - DOI https://doi.org/10.1039/C7TC00061H.

In one hand using polarizing beam splitter is the best way recombine beams with low losses. On the other hand beams with orthogonal polarisations will NOT interfere. So if you want to do collinear interferometry, you must use non-polarizing beam splitters. Both Michelson and Mach-Zehnder interferometers have 2 outputs. If you have constructive interference in one output, you will have destructive one in the other. Therefore the power split ratio between the outputs depends on the phase difference between your two beams.

Hi,

The lossing power you mentioned may be caused by the misalignment of polarization between polarized beams when utilize a polarized beamsplitter. You can try using a polarization insensitive beamsplitter and a half-wave plate to rotate their polarization and recombe them.