Discussion Forum

Wait a minute... do you possibly just have the wrong sign on the voltage or current that you used when constructing V/I? How did you define the impedance?

Hi Robert,

I used the global result emw.Zport_1. And things get even stranger when I look at the complex components. Here is what I sent to COMSOL support as well:

It looks like emw.Zport_1 actually returns a (wrong) real part only. One example at a specific frequency:
emw.Zport_1 = -10 Ohm
real(emw.Zport_1) = -10 Ohm
imag(emw.Zport_1) = +15 Ohm

emw.Zport_1 should be sqrt(real^2 + imag^2) = sqrt((-10)^2 + 15^2) = 18.02

Surprisingly the results are the same when I calculate the impedance and its complex components form emw.Vport_1 / emw.Iport_1.

According to the definition the absolute value of the imdedance cannot be negative and I always presumed that emw.Zport is defined as the absolute value of the impedance. I assume this is a bug in COMSOL.

I attach the impedance plot. Please note that the blue and the green line are on top of each other.

I appreciate your input.
Best regards
Edgar

Interesting. (I didn't even know that Comsol had introduced a global impedance variable; these are the sorts of things we used to have to define for ourselves...) Anyway, in regard to your assumption that the expression emw.Zport_1 should default to showing its magnitude, I should point out that for most complex quantities in Comsol Multiphysics, if you specify the complex variable without using real() or imag(), then it typically defaults to the real value. So at least that is consistent with the way Comsol treats other complex variables.

Now, that doesn't address your question about the negative value of the real part of the impedance. Perhaps someone else here on this forum can speak to that?

Hi Robert,

yes I also learned that impedance returns the real part only, and you are right, that this is indeed how complex values are treated in the post processing. I was proabably looking at it from a too practical point of view.
So this is not really a problem since we can use the abs(), real() and imag() operators.

What still drives me nuts is the negative real impedance. I also addressed this to COMSOL support. And I hope I can find some time to cross-check that in a simple model like a dipole.

Cheers
Edgar

Just a little update: The problem could be resolved with the help of COMSOL support. The reason was an accidential wrong phase setting in the solutions.
Now the model and measurements are in good agreement.

Cheers
Edgar

Hi Edgar,

I'm experiencing the same problem simulating an Alderman-Grant-resonator at ~2.3GHz.

The imaginary part of the frequency-dependent impedance behaves as expected. The real part gives a symmetric peak, as expected, but with negative sign. Correspondingly, the S11 parameter shows a peak with positive values.

Could you tell me where to find that phase setting that solved the problem in your case?

Best regards,

Oliver

Hi Oliver,

In my case the problem was that the 'solution at phase angle' was set to 90° in some of the solutions. Setting it to the default value 0 resolved the problem.

Cheers and HNY
Edgar

I have a similar problem, could you be more specific as to how one should set the phase value to 0?

Regards,
Mantas

No need for explanation I found it.