Not sure if this is what you are looking for but give it a try
Seeing as you are adding the vectors to get the rotations I doubt using the same absolute position would be ok as one has already taking in transformed data. Using the already transformed vector position by the earlier rotation seems to give more stable results.
Not a 100% sure the transform is correct as it causes some slight nausea so some parameter might still be off.
Time based
Alternatively you could use a mesh and a material like this:
Then internally rotate the mesh component
Another option is to input the vector parameter directly into two axis
This does not give you exact speed controls of the rotations but one axis takes on the for of a fraction of the rotation speed. (seems one of the more sane options)