Make sure the prism is used in its wavelength region. Find a polarized hene with an internal Brewster window to check the polarizer against (If Visible or near IR)
There is the ANCIENT stack of transparent slides at Brewster's angle method to ensure you have a highly polarized beam, essentially a stack of Brewster Windows in series. Then use that known polarized beam coming through the stack to check your beam. Keeping track of all the Fresnel reflections requires a really linear optical sensor.
Speaking of optical sensors, what are you using for your measurement? I'd really look at the linearity and bias of a photodiode before assuming I have a problem.
I'd also look at the quality and alignment of my rotator. 90' incidence with no tilt while rotating is important.
Or call up some one like Medowlark Optics or and have a chat about what to do.
The Poincare sphere is an interesting thing. Really incredible polarization is hard to achieve because the last optic always has some minor thickness difference or phase shift or internal inhomogeneity across the beam.
There are other techniques like "Law of Malus" experiments or methods involving reflected beams and waveplates etc...
Steve
Qui habet Christos, habet Vitam!
I should have rented the space under my name for advertising.
When I still could have...