CST MWS - PCB Structures - Transient Solver Settings

[Leegit]

Hello all,

I have been working with CST MWS for about a week now and I am simulating PCB structures, specifically microstrips, both single-ended and differential. I have a lot of experience with 3D modeling and EM solvers. Basically, I was hoping we could share our experiences on what are the ideal settings for the transient solver for ONLY these cases... here is what I have uncovered...

I start by loading up the EDA template so that my units are in [mm/GHz/ns] and my boundary conditions are set to normal with magnetic = 0 for xyz directions...


Transient Solver Parameters
Accuracy = -30 dB or -40 dB [not entirely sure on how this plays a role in the simulation?]
Normalized to fixed impedance -> ON with 50 ohms [electrical world is always compared to ideal 50 ohm ports]

Under Specials -> Line Impedance Adaptation -> ON with 0.1% accuracy and maximum passes = 8. [For my work, I need very high accuracy on Line Impedance, could be higher]
Under Specials -> Electric Shielding -> OFF
Under Specials -> Inhomogenous Port Accuracy Enhancement -> ON


Port Dimensions
Also, a very important part of the simulation is the port setup. I have my port dimensions set to the following:

Port Height = 5*trace_width & Port Width = 10*trace_width [These are the suggested values for port dimensions in IEEE Microwave Magazine Dec. 2008]


Question?
Has anyone come across alternative settings or setups which produce high accuracy while keeping the simulation time under a few minutes?

 

[hal_owaimer]

I'm working on UWB antenna design using CST Microwave Studio
I tried to simulate (without normalizing the impedance), then I normalized the impedance to 50 ohm. if the software is normalizing to 50 ohm by default, the results must be identical but the results appeared with me were really different.
so far I couldn't explain it.

also, I tried before to simulate with -80 dB accuracy, it sometime requires increasing the (maximum number of pulses) under (special>steady state). this is because the solver will continue simulating until the reflected signal at the port becomes less than -80 dB. for complicated shapes, I get different S11 when setting the accuracy to -80 dB
but for simple shapes like rectangular patch antenna I got almost identical results.