Impossible microwave interference? I was doing a microwave experiment with the following set up: there is a Gunn diode which emmits microwave radiation and a receiver (both work with polarised light). 

The strange thing is that when I  put a metal plate along the line connecting the emmiter and the receiver, if the plate is perpendicular to the polarisation plane, no radiation is detected. Instead, it's deflected about 20° from both sides of the plate. However, if the plate is parallel to the polarisation plane, radiation is detected as expected.
One explanation might be that polarised light will "collide" with the laterals of the plate (and that could produce interference). But when the polarisation plane is parallel to the plate, light will past without noticing. Still, my professor had no idea what was happening.
Addendum 2013/10/26
It's important to note that I doesn't matter if it a metalic or wooden plate, so it's nothing related the Fresnel equations. 
The effect is this:

The image is not precise at all.
The black zone is where no microwaves are detected. The red line is where the maximum is detected. The incident wave is blue. 
Another possibility is that there is no diffraction at all, because the plate blocks the incident waves (like if you shelter on a side of a wall when someone's shooting at you).
 A: Generally speaking, you might use the exact solution of the problem of diffraction on a half-plane (see, e.g., http://www.physics.princeton.edu/~mcdonald/examples/sommerfeld.pdf ). I don't have time to get the results, so I looked at the results for diffraction on a perfectly conducting cylinder. The idea was that the leading edge of the plate is, on the one hand, similar to the thin cylinder, on the other hand, it contributes most to scattering. The (linear) scattering cross-section is big (of the order of the wavelength) when the electric field in the incident wave is parallel to the axis of the thin cylinder, and very small when the magnetic field in the incident wave is parallel to the axis of the cylinder (see, e.g., the problems in the book by Batygin/Toptygin http://www.amazon.com/Problems-Electrodynamics-V-V-Batygin/dp/0120821605 ). Thus, I would suspect that scattering is significant when the electric field in the incident wave is parallel to the edge of your plate and negligible when it is orthogonal to the edge. But that seems to contradict your results, unless your definition of polarization plane is different from mine.
EDIT (10/26/2013):Looks like your problem is rather hard, but it was considered earlier. For example, I looked at the book by P. Ufimtsev (who is, by the way, is "considered the seminal force behind modern stealth aircraft technology" - http://en.wikipedia.org/wiki/Petr_Ufimtsev ) "Theory of edge diffraction in electromagnetics", Tech Science Press, Encino, California, 2003 (I used a Russian translation http://www.vixri.ru/d/Ufimcev%20P.Ja.%20%20_Teorija%20difrakcionnyx%20kraevyx%20voln,%202012,%20375s.pdf ). In Section 8.2, he starts considering a plane wave diffracted on a perfectly conducting ribbon with electric field in the incident wave parallel to the ribbon, and at the end of Section 8.5 he says that scattered field does not arise for grazing incidence. This seems consistent with your experimental results, so my initial response does not seem correct. However, Ufimtsev's derivation is rather difficult to follow, so, again, looks like this is a hard problem, but you can look at the book and references there. Looks like Fialkovskiy's article on diffraction on a ribbon can be relevant.
A: Your explanation seems to be right if the polarized photons are in linear polarization, as it can only be explicitly detected when the receiving plate is in a similar orientation to the emitting plate, because linear polarized photons can only travel in a straight line, not in waves. An example of this would be how polarized sunglasses block out light that is coming from a vertical angle.
That's what I think is happening
PQx
