Optics: polarisation alignment

[Quasi Particle]

With a combination of wave plates in the order \lambda/2 - \lambda/4 - \lambda/2 it should be possible to achieve any polarisation. But I don't understand why: the first wave plate switches the polarisation direction, the second changes linear to circular polarisation and vice versa, and the last one switches it back again. How does this align the polarisation?
Thank you for any answers, or hints, or links.

 

[Claude Bile]

http://en.wikipedia.org/wiki/Jones_calculus

You need to take into account the possibility that the wave plates may not have their fast axis' aligned.

Claude.

 

[Quasi Particle]

Thank you. I calculated the problem now, but the solution is lengthy and I guess it'll be easier to just try it out with the accordant waveplates

I figured that I only need two waveplates to produce any desired linear polarisation. A quarter-wave plate will turn elliptical into linear polarisation, which can be aligned into the desired direction by the half-wave plate.