SGH model in openEMS shows similar results to CST in the E-plane but different in the H-plane.

[dirixmjm]

For a couple of days I have been trying to generate a simple NRL-based SGH model in OpenEMS and have been comparing results to a similar model created in CST Studio.
While working on the model I have reviewed and tried to incorporate some of the responses to #220 but to no avail yet. Probably I have missed some key advise.

It seems that in the E-plane direction the OpenEMS and CST models are quite comparable, however in the H-plane cut I see clear differences. For as far as I can tell I have created the models equal in terms of mechanical sizes, having the following requirements:
Frequency: 8.2 GHz
Waveguide internal size: 22.86x10.16 mm
Horn antenna length: 289.8 mm
Horn antenna aperture: 194x144 mm
Wall thickness 2 mm.

I have attached the following:
(1) Horn_Antenn3.m file to simulate with OpenEMS
(2) Comparison between CST and OpenEMS results E- and H-Plane plots.
(3) CST exported STEP file, should it be helpful to check the model used in CST.

Any advise at this point would be welcome.

SGH1_t_2.0.zip
Horn_Antenna3.zip