In the new design, I removed the mirror and track the position of the photo diode via a reflection in the prism.
This technique has a weakness.

The weakness is as follows, if the center of the rotor is not stable but moves microns during  the rotation.
This will propagate to the photo-diode.

To simplify if the rotor translates in the old design nothing happens;

In the figure below the prism is moved upward by 10 mm..  Still the photodiode is hit at the same position.

In the figure below the prism is moved 2 mm sidewards
This is very different than the original
This behavior already becomes a problem at say 20 microns. The error is propagated back twofold. The relation seems non linear and become more complex at larger displacement.

The angular position is assumed to be stable due to angular momentum build up.