[SOLVED] trouble designing a dipol antenna in cst (port problem?)

hi guys,
i'm new to CST. i did a tutorial (koax T), which was no problem.

Then i startet with a simple lambda/2 dipol. Two tubes (l=325mm, fo=400Mhz, discance 5mm) and a discrete port connected to both. E-field and S11 parameter looked pretty good.

Then all i did was adding a dielectricum (450mm*150mm*70mm with muscle) around the tubes, and a vessel with blood. the antenna lies inside the vessel.
with the new epsilon of muscle (exactly i used 2/3 epsilon of muscle), i calculatet the new wavelenth/2 to about 60mm.
Here starts the trouble: when i uses the dipol, s-parameter made absolutly no sense. a wild curve, no real resonance. i thought perhaps the mesh is not fine enough (i'm limited to 100 000 cells) and used a groundplane+lambda/4 antenna. the result looked good, as long as my discrete port goes through the ground plane. when i connect it to the plane itself, the s11 makes no sense.
used open (add space) boundaries, an a E=0 boundary with the lambda/4 ground.

thanks for your help

is there nobody who can help me?=)

perhaps i should outline my question a little bit more?

I'm not sure how i should connect the port when i use a lambda/4 antenna. is a discrete port right at all?
Could the ratio wavelength /distance be too big, and if yes, how can i simulate the dipol otherwise?
Would a frequency domain solver be better as the time domain solver?
Did I make a mistake using the mesh grid?

hope anyone can help me...

solved it by myself, was a mesh grid problem:

-surrounding material ->background material
-symmetrie: simulating just a quarter of space
-submeshing
-2 cylinders with background material properties to cause a finer meshing at the port

--> nice curve