I have simulated a rectangular fractal dielectric antenna, resonating at 9.5 GHz,
and all is nice and fun, until i check the directivity. I have too many main lobes (hope that this is the right translation), and there should be only one.
i tried playing with height and size, but i cannot seem to get one main lobe. i keep getting tons of them (at least 8 ).
in one simulation, i changed εr to 36, and toyed with size again, and i managed to get one ugly, 30 degrees leaned single lobe, but not close to any readalbe frequency.
where is your excitation port?
it's not necesary to change εr if you need to get rid lobes. Good directivity is based on optimal coupling between the dimensions of the 5 composants of your antenna. I proposed you to optimize dimensions paramaters using "transient solver-par.sweep"
if you look at the design its like 5 patches making the antenna. Each patch will have its some what main lobe. In patch array feeding network is designed such that the main lobe of all antennas meet and you get single lobe (combined).
Since you do not have feeding network , try making the dimensions of the patches different to one another and see what happens.