Moved on from #meep to #openEMS. I think I can probably do the same types of things in meep, but it was just taking too much time. One of those 'right tool for the right job' type scenarios.

Here are some pretty plots of reflections and impedances from a simple dipole simulation in openEMS.

Not quite as pretty as the EM wave pictures from the other day in MEEP, but its more useful info for antenna design.

#electronics #RF

Great, I think I'll stick with these parameters for antenna B (in #openEMS I simulate each antenna separately).

The example in the help has little to do with AddWire, I should remember to tell the author.
#openEMS

I will probably repeat some of the described experiments (simulations), especially related to reflectors, but in #openEMS. It will take some part of the night:)

Another peculiarity when modeling complex antennas in #openEMS: when using 'Translate' to create primitives, the argument is a string. When converting a number into a string there is an understandable loss of accuracy and you can then look for a long time for an error in your grid and other things if you don't remember about it:)

Note to #openEMS users: the difference between PLA_8 and MUR boundary conditions is the difference between 39 million cells per second and 200 million cells per second 😉
Very useful for rough experiments.

It's better, and so much better that I can't believe it:)
Energy radiated/energy received (I mean through the port minus reflected back). There is nowhere to go wrong, well almost: I need to plot the reflection graphs and compare.
#openEMS

Nothing to do with reality, but beautiful.
I messed up with the anonymous functions and closures somewhere🤔
#openEMS #octave

@Alexis_Krier@mastodon.social
#openEMS and #paraview?
yes, although the first must be built from source, and the second has disappeared from #DragonFlyBSD ports for some reason, maybe there are still in Linux packages:)
https://openems.de/start/

In principle, I am happy with the picture that is emerging. I knew that I had to compromise on the parameters for appearance reasons: I cannot spread the antennas too far apart, I cannot make a reflector of the recommended width because I will make a half-meter monster, which does not look like a room antenna at all, and the modem will not withstand the conditions outside.
#radio #openems

This time I'm purposely making the shape of the antenna ``broken'' in order to get several S11 peaks and thus expand the effective bandwidth.

#radio #diy #openEMS

This is what you can get after a few hours of simulation if you did not write the command quite correctly🤪
Well, I can still say that the peak efficiency is probably at the frequency I need, 1800MHz😉
The slowdown happens when the swap is engaged, it's raining outside, but maybe I'll go for some DIMMs after all.
#radio #diy #openems

Well, all the geometry is parameterized, I can turn off the visualization and start experimenting:)
Although the insulation of the antennas is too good, I probably made a mistake somewhere.
#radio #diy #openems

``The transformations are applied in the order as given inside the {} brackets.''
It didn't immediately dawn on me that this isn't a list at all, but rather the order in which the unique slots of operations are applied: I can't do { 'Rotate_X', pi/3, 'Transform', [2, 3, 0], 'Rotate_X', pi/10 } --- only the last Rotate_X will work.
Thank goodness there is another rotation slot 'Rotate_Origin', and 'Matrix' as the last resort.

https://openems.de/index.php/Primitives.html#Transformation

#openEMS

The idea of a different element arrangement for the new MIMO antenna. I need to install #OpenEMS and test this design.

No doubt I will have to edit a bunch of files again to make this thing work with Qt5. :(

I really respect people who make things like #OpenEMS, #KiCad or #FreeCAD, in fact I consider them wizards, but why in God's name would they choose such shitty graphical toolkits?
I know it's multiplatform and I know it's going to be a pain in the ass:)

By manipulating the distance to the reflector and the rounding radius, I achieved a cool s11 graph!😜
It was only when I launched the #FreeCAD to design the antenna support elements that I discovered that in one place I was mistaken with the dimensions by about half☹️
Have to repeat the simulation from the very beginning.

#openems #radio

Now that I understand what's what and how to interpret the simulation results, I don't have serious troubles with the mesh and I don't have to wait 3 hours because of poorly selected regions.
Now I can add copper parameters to the model, add a reflector and try different edge lengths for good reception at 2.6GHz.😉
#radio #noob #openems

Not bad! The difference between the theoretical and simulated perimeter of the antenna is: | 114.06-113.89 | = 0.16717mm
I had to rotate the antenna along the grid, and create the grid itself very carefully.
#openems #radio

If I tilt the dipole at 45 degrees, then the antenna parameters are floating. And this is understandable: #openEMS is bad for non-grid oriented solid conductors🙁
nec-2 proved to be better, but it has dragonian restrictions: the ratio of segment length to radius cannot be less than 6. I have a wire with a radius of 1 mm, so the power gap will be> 6 mm, but 1/4 of the wavelength of 2.6 GHz = 28.8mm! That is, I will not get a square, but something indecent.
#radio

The first result is with my geometry. It’s not very clear what the black lines are in the center. I also need to add a reflective screen and see how this simulation system behaves, whether the wave will be reflected😀

If #openEMS breaks on this, then it doesn’t matter --- may be I can replace the reflection with pseudo sources.