How to use openEMS. Discussion on examples, tutorials etc

Moderator: thorsten

Hale_812
Posts: 162
Joined: Fri 13 May 2016, 02:54

After being absent from the RF topic for almost a year, I am back, and restarting OpenEMS again.

While checking for weird results from the scripts, which composition I already forgot, I found interesting things.

In some tutorials Thorsten uses antenna efficiency definition as e_a=nf2ff.Prad' ./ interp1(freq, port.P_acc, f0) ; sometimes called nu_rad.
(here transposing is mine, intended for multi-frequency analysis)
So it looks like in HFSS. Supposing nf2ff.Dmax is calculated from both E and H*, the antenna gain is simply G=e_a*nf2ff.Dmax

So, here are first unclear things:
- nf2ff.Prad' ./ interp1(freq, port.P_acc, f0) often gives larger than 100% efficiency (104%-105%, easily)
- is nf2ff.Dmax really calculated from ExH* ?
- What is nf2ff.P_rad ? How is it related to nf2ff.Prad? Maybe the calculation is not good?

Next, I see more weird thing in the Horn Antenna example:
G_a = 4 .*pi.*A./(c0./f0).^2;
e_a = nf2ff.Dmax./G_a;
wawqawawawhat is that!? Isn't gain proportional to efficiency? G_a looks kind like ideal gain (with 100% efficiency). And e_a looks like INVERSE from the efficiency.
ea={G(c/f)^2}/{4pi(A_width*A_height)}

But well, ideal gain is NOT(!) physical Gain. So that's why G_a/Dmax, where Dmax is an "physical" value gives numbers close to 200%.
I think, getting value below 100% is not very good justification for inverting e_a

Here I mean "effective" is the electrical effect of accepting power with the feeder, "physical" is related to antenna geometry, how it emits power, and "ideal" is for an abstract section of the wave phase surface.

AFAIK, technically, e_a is(should be) just a ratio of radiated power to accepted power: e_a=Prad/Pacc
Where radiated is related to final gain, like accepted is related to accepting geometry of the antenna. (see above)

Now again, it is a question, why nf2ff.Prad gets so often larger than port.P_acc. It can be related to interpolation interp1(freq, port.P_acc, f0). But my observations on dense arrays tell me it is not the only possible reason.

I guess, something does not converge well, while it should give unconditional convergence. Something in numerical errors maybe? I know the mesh can be too coarse, particularly for horn antenna with oblique surfaces. But the power conservation should be satisfied.