plane wave propagation problem in HFSS

Hello,
I have a problem with a very basic simulation. I am trying to simulate plane wave propagation from one medium to another. The problem is not the setup, but the phase of the wave.

Setup: All lengths in the setup are parameterized. The basic length unit is the free space wave length lambda (@14Ghz). Die setup is 4lambda long, with 2lambda for the first medium and 2 for the second.
An incident wave source is placed at (0, 0, -1lambda) in the medium1. The k-vector looks in the direction (0, 0, 1). The setup is surrounded by two Master/Slave (M/S) boundaries and at both ends of the QUADER I used perfect matched layer (PML). The M/S boundaries are also used over the whole setup, i.e. the boxes of the PML are also covered by the M/S.

Now, the problem is that there is a big difference between the calculated phase and the measured one. Also, the difference is not constant over the way.
For example, I measured the E vector at the point (0, 0, lambda/2). So the wave moved one lambda in the medium1 with epsilon_r = 1.1 and then half lambda in the medium2 with epsilon_r = 2.1. The measured phase is -23.7414 and the theoretical phase is -81.5840. This difference is too large to be just a numerical problem. Does anybody know what to do or did anyone experience this problem before?
Wave Propagation.zip
I also tried the setup with just one medium. And when you do the same setup with just one medium, you can see waves behind the wave incident wave source. Is it possible that there are reflections on the perfect matched layer?

Thank you very much in advance
Trifu

Problem solved by using a Floqet Port on one side of the setup and PML on the otherside.