Possible to define goals at multiple frequencies in OPTFEKO?

bouvy

Hi, I am running FEKO 2017.2. For a two-port antenna that I am designing, I would like to define two optimization goals at separate frequencies: minimize the S11 at f1, and minimize the S21 at f2. Is it possible to define two optimization goals at separate frequencies? 

 

Thank you,

Alex

Mel

Use two separate SParameterConfigurations and two goals. One configuration per frequency, with the frequency defined per configuration.

Then define two goals. One goal operates on the 1st S-parameter request and the other goal on the 2nd request.

Alternately, consider using a mask for S11 or for S21. An example in the Example Guide example "Pattern Optimisation of a Yagi-Uda Antenna"

Multiple configurations, where you can group together different sources and frequencies and output requests, are explained below. 

bouvy

Perfect. I've only used used the global frequency before, didn't realize you could set it by configuration. Thank you!

bouvy

Hi Mel, I'm getting some results that I don't fully understand that I'm hoping that you might be able to clarify.

Below are my two configurations and optimization goals:

Note that one port is a discrete port while the other port is a waveguide port.

 

In my optimization parameters, I have three parameters that govern the S21 of the antenna, and two parameters that govern the S22 of the antenna. The goal I am trying to achieve is to minimize the S21 of the antenna via the distributed feedback mechanism I have in my antenna design. However, when I run the optimizer, it instead is simply making the S22 very high, which makes the S21 very small (minimized). I was hoping that adding the second optimization goal to minimize the S22 would fix this issue. When I run the optimizer now, it runs for ~5 iterations. My S21 is very low as desired, but my "minimized" S22 is very close to 1 (nearly the maximum possible value). It seems to me that the two optimization goals are not being weighted equally, as I know that the S22 can be reasonably made much lower by modifying the parameters I've provided to the optimizer. 

 

Do you have any insight into this issue? I wasn't expecting perfect results from OptFEKO based on the way I have set up the problem, but based on the results it seems like something is going wrong with the way I have set things up. As a note, I am leaving the optimizer set to "automatic" but I notice it selects the Simplex Nelder-Mead method. 

 

Any help is much appreciated.

 

Thanks,

Alex

 

Mel

A few ideas:

 

From your screenshots it shows you are using a value of 1 for the weigthing of both goals. If you have not tried yet, give the S21 goal more weight.

 

You could also maybe define the goals to be more specific - say S21 in dB should be smaller than -10 and something similar for S22.

 

Try the GRSM method instead of Auto (which usually selects Simplex)

 

Try a optimisation with regards to a mask: in the Example Guide there is an optimisation example where a mask is used to define the desired output response. "A.6 Pattern Optimisation of a Yagi-Uda Antenna"

bouvy

Hi Mel, thanks a lot for the feedback, I'll take a look into these suggestions.