Newbie struggling with mesh setup

[piranha32]

I'm completely new to EM modelling, and since I need to design and build a discone antenna, I used it as an opportunity to learn openEMS.
Based on the examples and tutorials, I created the antenna model, set up simulation, and I can generate simple graphs for S11 and antenna impedance. The excitation port is connected between the center of the top metal disk, and the center of a metal disk capping the top end of the cone. One thing that I can't get over is the setup of the simulation mesh.

As a starting point, I use antenna dimensions generated by an online calculator, and I'd expect to see from the model results roughly matching the requirements. Unfortunately, what I get is nowhere close to my expectations (not very surprising, the results from the calculator may be off), and drastically changes with even small changes to the mesh structure.

I got to using mesh with resolution of lambda_min/30 for the volume enclosing the antenna (with margins), and lambda_min/20 for the rest of the simulation volume (with 3/4 of max dimension of margin along each of the axes), and I still can't obtain trustworthy-looking results.

I would appreciate if you could help me improve the experiment setup.
Attached is a python program with full simulation code.

disconelab.txt

[Sarajevo67]

Unfortunately, openEMS isn't best suited for simulation of curved 3D shapes. I tried to simulate thick wire helical antenna, but only for the purpose of visualisation of currents and fields. No trustworthy impedance simulation could be expected (as stated by @thliebig ).
My solution was to use NEC MoM simulation for helix impedance and openEMS for 2D shapes, like MSL impedance matching transformer.

[Sarajevo67]

You took too wide frequency range for port results. Reduce it to 0.7-3.4GHz to remove that deep reactance drop.
Use discone antenna calculator from this site There you will find ANSYS simulation results for this discone with Fmin=800MHz. Compare them with openEMS.

Best Regards,

DD

[piranha32]

I used calculator from https://www.changpuak.ch/electronics/DisconeAntennaCalculator.php, and the dimensions came out almost identical. Both calculators look so similar, that I think they may be based on the same code.
Thanks for the link to the ANSYS analysis. I'll compare the results with my simulation. I got to the point where the simulation results don't change when I refine the mesh even further, so hopefully I've got it right.

As for the too wide frequency range, is this too wide for the simulation or for the antenna?

[Sarajevo67]

Calculator references Fmin=800 Fmax=3400 MHz for antenna. For simulation excitation I would use Fc 2.5GHz with 2 GHz from each side (0.5-4.5 GHz)

[glompos21]

After some testing i found the discussion. As of now (25/12/2024) does the opemEMS capable of simulating a 3d shape?
I try to simulate an asymmetrical bionical antenna (for start will be a symmetrical one).
I used Freecad for the design, the macro to generate the python code. I have put it on the following GitHub repo.
Am at the right road to get some results or i have to find another software?

[thliebig]

openEMS should be able to simulate this. The only issue is to create a mesh that does recognize the curved shapes correctly.
Ideally the wall thickness is larger than the mesh size. Otherwise the PEC (metal) surface is not closed.

At some point in the future openEMS needs a feature to map a metallic surface safely to the FDTD grid.
I recommend to use the --debug-PEC option...