Turnstile Antenna in CST MWS: Quadrature Feed and Weird Results

Hello everybody,

I'm new to CST and I've run into a few problems. I'm going to simulate a turnstile antenna for 437 MHz consisting of two crossed half-wavelength dipoles with 2 cm feed gap. One along the x-axis and the other along the y-axis, so the maxima is supposed to be in the z-direction. I've made one (discrete) port for each dipole and performed the transient analysis to obtain the far-field simulations.

Q1: My first problem is that I don't know the port impedance. I saw another guy just put it to 73 ohms while making a simple dipole but suppose you do not know the impedance. What do you do then? I tried to put a current source of 1A at both ports instead. I got the warning "Balance is not available for all excitations." What does this mean?

I read somewhere that in order to simulate the antenna pattern when the two dipoles are fed in quadrature you simply execute the transient solver, which will then perform transient analysis on each dipole subsequently, and then you combine the results afterwards.

Q2: When I run the transient solver I got two far-field patterns. One for each dipole. I assumed that the pattern from for instance the dipole along the x-axis would be that of toroid with nulls at the x-axis and maxima in the yz-plane. Instead I get the maxima in another plane which is rotated around the z-axis (approx. phi=45 degrees) to be between the yz-plane and the xz-plane. It seems a bit like the two elements are excited in-phase in each of the two far-field plots eventhough the manual says that they should be excited subsequently. I can not understand that it could be any kind of coupling either. If that was the case; the configuration has a certain symmetry so there would be no reason for the plane of maximas to be at phi=45 degrees rather than phi=-45 degrees (one dipole is at phi=0 degrees, the x-axis, the other at phi=90 degrees, the y-axis).

To combine the results I read that i should choose "Results->Combine Results" and then choose the two far-field solutions and add them with equal amplitude and a phase-difference of 90 degrees. So I tried to choose amplitude 1 or 0.707 and phases 0 and 90, or 0 and -90 (I haven't taken care of whether I'm looking at RHCP or LHCP so I tried both).

Q3: Is this the correct way of simulating the quadrature feed?

Q4: Again, I get a weird pattern, probably because my single port patterns are weird in the first place?

Q5: Is the phase in the Combine Results dialog box in degrees? (that's what I assumed but I tried degrees also)

It's a lot of questions but I will be very grateful for any answers. I've been struggling with some of these for many hours now.

Kind regards,
marholm

:-D Tips looks great

thanks mate

Hi again,

I have partly solved the problem myself, but I need some furhter help.

The gap of the two crossing dipoles is in the same plane (xy-plane) and port 1 (discrete) goes between the edges of one dipole while port 2 (also discrete) goes between the edges of the other dipole. This means that the ports cross each other. I didn't think this mattered when I choose to use discrete ports as I thought it was not a physical thing. It turns out, however, that when I make a tiny extension (in the z-direction) to one of the dipoles and define its port on the extension, the pattern, and the result combination works just fine. Of course the extension introduce a non-ideal phase-shift, but as the extension is small, so is the phase-shift.

So here's my (new) questions:

Q6: Why can discrete ports not cross each other?

Q7: I still like to have the antenna structure like i first made it (without the extension). But that means that the gaps will be in the same plane again, so how can i set up ports which doesn't cross each other then?

Q6. The excitation is placed to along the edge of the mesh, so when two ports cross each other, they have electrical connection, which is not the kinda feed we are interested.
Q7. You may try something like the given in the attached simple dipole model. You will still keep the original phase difference. To be even safer, make sure that those feed lines are at least two mesh cell apart (so that they don't belong to same cube (i.e. hexahedral)) You can see your mesh from "mesh view". Select "z normal" and by hitting up and down arrows find the plane that your ports are at.

chameleon said:

Q6. The excitation is placed to along the edge of the mesh, so when two ports cross each other, they have electrical connection, which is not the kinda feed we are interested.
Q7. You may try something like the given in the attached simple dipole model. You will still keep the original phase difference. To be even safer, make sure that those feed lines are at least two mesh cell apart (so that they don't belong to same cube (i.e. hexahedral)) You can see your mesh from "mesh view". Select "z normal" and by hitting up and down arrows find the plane that your ports are at.
View attachment 63633

Click to expand...

Thanks you very much for your answer chameleon.

This is actually similar to what I finally did. I extended the edges of one dipole with PEC material in the +z direction whereas I extended the edges of the other dipole in the -z direction. That way there's an equal phase shift from the excitation ports to the dipoles and I get the desired pattern.

Q8: Is this what you meant too? The reason I'm asking is that I saw in your figure that the "lines along the z-axis" is colored red and blue so I start thinking that maybe it is possible to kind of "bend" the discrete ports instead of extending the dipoles' edges with PEC material.

Q9:In my case, I also think the extra PEC material added changes the input impedance of the dipole. I would like to be able to measure Zin as well. Do you have any good ideas here?

The reason I want to measure Zin is that I'm placing a box of PEC material close to the antenna and expect it to influence my input impedance. When I sometime want to connect the antenna to a tranceiver it must be matched to it's characteristic empedance.

Again; thank you very much.

Q8: I meant just what you did. You can use infinite thin very narrow PEC plates to do this extension, to have the least effect.
Q9: As I said infinite thin PEC plates with very narrow widths may help improving the effect. How do you plan to feed the fabricated model?
Even though the numbers of real part of impedance and imaginary part of the impedance may look changing a lot, this may be due to small movements on the Smith Chart, especially if your antenna is not matched that well. Try visualizing your antenna impedance with Smith Chart.
I'm not sure if I understood your use correctly. So, are you trying to measure the effect of the PEC box, like an RCS measurement? If that's the case, you may try using "target"-"no target" (target - background) subtraction. To do that you need to simulate the two seperately and do the subtraction. Making sure that your mesh does not change between the two measurements will improve your accuracy. To do that, model everything in your setup.
Not to change geometry you can try to do following:

For each model part, click on "Local mesh properties" to turn automatic refinement off. Then in "no target" case, change the material assignment of PEC box to Vacuum. Check the number of unknowns to see that you are actually not changing the mesh.

If you don't want to decrease your refinement for PEC surfaces (because automatically refines it by 2, if not otherwise chosen), just put your own number. The key point is to not let CST do automatic refinement.

Let me know if this makes sense for your application.