openEMS.de

The changes made back then are still there and not much was changed in recent years on this front. I do not see an immediate reason or problem.
What nf2ff files are being generated by the MPI openEMS version? Do they not exist or are they invalid?

Well in ports.py is a class calles "UI_data" that handels with U/I (voltages, currents) data and the DFT. As you said usually the ports handle all this calculations (and even probe creations). But back to your question. What sense does it make to calculate U/I in time domain? I currently do not see ...

Yeah, qt4 and vtk5 are out of date. Just use the latest qt and vtk. Should just run fine...
I really need to update this infos. Trouble is I do not use Ubuntu anymore...

regards
Thorsten

What you try to do looks complicated and I see a lot of issues. First I don't think you should use cylindrical coordinates. If you want to have a perpendicular waveguide you loose any gain but keep all the disadvantages of this mesh type. Secondly your excitation is inside/behind the PML? regards Th...

Maybe use 800MHz (800e6) as fc, that should give a nice short pulse. A feed must connect two metal signal layers. If it is on top of one it may be just short-circuited or at least unable to create a connection between ground and your antenna. As far as I see it, currently is your antenna not connect...

openEMS::SetupFDTD: Warning, the timestep seems to be very small --> long simulation. Check your mesh!? Excitation signal length is: 100000 timesteps (2.43824e-08s) Max. number of timesteps: 100000 ( --> 1 * Excitation signal length) Well openEMS agrees: your excitation is even longer than your req...

Well first, it looks like your feed is on top of the feed line instead of between the feed line and the ground?
Secondly using a pulse with a width of only 1 MHz is very narrow and thus takes a very long time to simulate.
I recommend using a large pulse bandwidth (short pulse)

Well I do not see an obvious problem. (except the missing mirror for the 3D far fields, but that does not change the cut planes) I'm running it now, but it takes a while. What you could try is to run the non-mirrored simulation, but with the cut nf2ff recording and apply the mirror then. This way sh...

Finding problems like this may be hard to find indeed. Try the "--debug-PEC" option to see what the parts look like after meshing...

Looks like static charges/currents that cannot go anywhere. Solutions/workarounds: 1) Make the model more realistic by adding losses (which should affect the DC parts too) 2) Try to use a narrower excitation pulse and exclude the DC (low frequency) range, downside: longer excitation and possibly lon...