CST Discrete port problem

[steven werbrouck]

cst discrete port

Hey,

I've got a problem when using a discrete port in CST Microwave Studio 5.1.2. It's a pretty basic thing, I think.

I want to do a transient analyses for a simple monopole structure. So I choose a discrete port between the monopole and the PEC surface. In the manual they state that such a discrete S-port has to be small enough, at least 1/10 smaller than the wavelength. So I choose it pretty small.

But when I sweep the length of that port, I get results that are very different: the resonance frequency stays more or less the same (cause the total length of the monopole remains constant), but the S11 at resonance frequency varies a lot. It seems that the length of that port plays an important role! Obviously this shouldn't be.

Has anyone an answer on this?

thx,
Steven.

 

[loucy]

discrete port

To the contrary, the length of that port should "Obviously" play an important role. In fact that is the main reason a "discrete port should be small". I would say for your wire antenna, lamba/10 is not small enough. You should try to sweep the length toward smaller value and see if there is some "trend of convergence", and make comparison with the result from a MoM code. Both the "thin wire" and the "small gap" (of the port) are tough to handle with FDTD (or FIT). But these two problems are well-studied in MoM.

 

[steven werbrouck]

cst discrete port same mesh locatio

Thank you for your reply.

" It seems that the length of that port plays an important role! Obviously this shouldn't be. "

I ment that the real result can't be dependent on the length of some discrete port (that doesn't exist in reality). So the simulation result can only be a good estimation of the reality if that length does not effect the simulation too much.

I did a lot of sweeps (from the smallest length possible to length = lambda/10), but no convergence is met. I thought of comparing with MoM-method too, and indeed, there is some match for a certain length of discrete port. The problem is that I don't see any clear connection with e.g. the wavelength, the size of the monopole... So this means that I have to do a MoM-simulation everytime I use a discrete port, just to determine a good length for it.

That isn't a good way of working, I think. I use a simple and free program for MoM. It cannot construct complex structures. So there's the limit of using a discrete port in CST for me...

 

[snkhan]

Hi Steve:
I think in CST(as in manual), dicrete port is a good way to get approximate S11 rather than exact value, although it gives exact radiation pattern.

 

[Pushhead]

Hi steven,

How about trying the same model with a "real" port. I'd suggest you build a coaxial waveguide to feed your monopole. It shouldn't be so time consuming in computation (Don't forget to mesh it right in order to get the right impedance)

In that way you will be able to compare discrete-port results with "reality".

Good luck,
P.