Issue with consistency in ADS-EMDS co-simulation results

Hello Everyone,

I am facing consistency issues while simulating a simple structure with ADS EMDS (FEM), using different mesh parameters' values. Thus, I am not able to decide on proper dimensions, to go ahead for fabrication.
I created a quarter-wave impedance transformer of 57.002-Ohms characteristic impedance. It is tested by using a load resistance of 64.9845600hms, as this impedance is converted by the impedance transformer into 50-Ohms. Simulation set-up and results are given in the file attached.quarter_wav_Z_xfrmr.pdf
In sum, I found two main issues with the EM-circuit co-simulation, done using EMDS (FEM) and ADS-2008:
1. For same dimensions of the microstrip line, used to implement the transformer, matching frequency varied with change in FEM settings
(no. of 'Consecutive Iterations of Delta Error Required' and 'Mesh Seeding').
2. As length of the microstrip line is increased, change in the matching frequency shows illogical behavior. I mean logically the matching frequency should decrease, but it does not.

May anyone please help me out in this regard OR direct me to the appropriate expert. This has severely halted my work.

By the way, I may also wish to know if any modifications are also done in EMDS, as ADS version changed from 2008 to 2009 and then to current 2011?

Thanks and Regards,
Vivs

I am thankful unto Anurag Bhargava, Applications Consultant at Agilent Technologies (http://abhargava.wordpress.com/).

He pin-pointed the issue with use of 'internal' port in the circuit. This port being uncalibrated, contains inductive parasitics. This led to inconsistent results. Ensuring port 2 as 'Single Mode' has mitigated the problem and is leading to consistent results.

That's actually true for most uncalibrated lumped ports in 3D EM: they introduce a thin current path that depends on meshing. Refining the mesh will change that path, and might actually introduce even more parasitic series inductance as the mesh is refined.