HFSS solution frequency

[elektr0]

Hello,

the solution frequency of HFSS is the frequency, at which the mesh
adjustment is done.
For larger frequency ranges 1..10 or 1...30 GHz the results for lower frequencies get erroneous, if I use the highest frequency as the solution frequency.
How can I fix the problem. If I use different solution setups, with different sol. freq. the result plots are not continuous on the transitions.


Another question, is it possible to force HFSS to calculate a minimum frequency points, when doing the interpolation sweep ?


Thanks for any reply.

 

[cd79]

The use of different setups is a good way. If the solution plots are not continuous, this means, that your tolerated error is too large. In this case you must increase the number of converged passes or minimum passes. Increasing this step by step you will see how the plots become nicer.
Of course the simulation time rises with every pass, but you also prove the correctness of your work when combining different setups. Normally, when I have a solution I change the setup frequency slightly, e.g. 600GHz -> 650GHz. If the results here look similar, I know that there is no calculation error in the initial solution. This can appear, if the setup frequency is located close to a zero of any physical value.

 

[elektr0]

@cd79.
Thanks for reply.

There is no possibility to force the interpolating sweep algorithm to do a minimum number of frequency points. RIGHT ?

The result data from the interpolater is ok, but at low frequencies it shows strange behaviour, which can be fixed by placing a frequency point.
If i perform a second sweep the curves wont have a continuous transition, due to differences of 0.001db insertion loss. It's ok from a physical point of view, but if i show the results. This will provoke questions. *g*

I tried the discrete sweep. It is not possible to integrate further measuring frequencies later on. It is always like a second result.

What can I do ?

 

[cd79]

It's ok from a physical point of view, but if i show the results. This will provoke questions.

Thats right and its good! You need to mention the numerical solution and the FEM presenting these results. And the small jumps in the plot are a good possibility to explain it. It will help people understand and estimate the acceptable error.