I'm working with a high Er (Er=36), 10 mil substrate which I have defined dielectric vias for it and so I must use FEM. When I use the linecalc to calculate for example the width of the microstrip line for Z0=15 ohm, the answer is 26.3 mils. I draw a 100 mils length and 26.3 mils width line in layout and simulate it using FEM and Momentum simulators from 1 GHz to 20 GHz. The return loss of the FEM results is not better than -25 dB for Freq>10 GHz but the results of Momentum analysis seems correct. I have checked this simulation using CST frequency solver and I saw that results are similar to ADS Momentum but differs from ADS FEM.
I tried to change FEM EM setup to get better results which are similar to ADS Momentum and CST but nothing happens.
I should say ADS FEM results for high Er material in high frequency seems to be incorrect. What can I do with this problem?