cst patch antenna tutorial

Hi,
I am currently working on a project which I need to design a probe-fed type circular patch antenna using CST MWS (please see the attachment). I am trying to simulate the S11 parameter of the patch antenna but I do not know how to model the coaxial feed which is connected underneath the patch (since I need to define a waveguide port for it).
Could anyone please help? Many thanks.
Sim

That's not difficult. u may simply take a look at a example in CST, a four circular patch antenna array which are all fed by coaxial line from the bottom of the Ground Plane. The Tutorial.pdf gives every detail and the corresponding model could be found in "Examples".
good luck.

hi drunkbear,
Many thanks for your help. I have a question related to the outer and the inner radius parameter of the probe feed. From the tutorial, I notice that the inner radius = 1.12mm, outer radius = 4mm with dielectric constant of 2.33 (same as the substrate). Does it try to simulate an actual coaxial cable properties (which is then connected to the patch via a coaxial connector) in this case?
May I know is there any guideline in choosing the values for the inner radius, outer radius as well as the dielectric constant for the feeding structure? I am trying to simulate a 50-ohm SMA connector actually. Is there any links that I can refer regarding that?
Thanks.

Try in CST: Macros - Calculate - Calculate analytical Line Impedance, there you can calculate coax. Is the same as SMA or another connector, you must only calculate impedance (50 or 75 Ohm).

hi mauloftin,
Thanks for your reply. I have a question related to the dielectric material used for the feed part in the CST circular patch antenna tutorial. If I am not mistaken, the dielectric used for the coax is Solid PE which has a dielectric constant of 2.33. In that example, the substrate of the patch is also having a dielectric constant of 2.33.
(please see the tutorial attached)
What if the patch substrate is a FR4 material (for example) which has a dielectric constant of 4.7. In that case, do I need to model the dielectric of the coax to 4.7 as well? Or does it depend on the type of coaxial cable used?
Please help. Thanks.

hi zhiweisim,
when it comes to the physical size of the coax line, basically, it largely depends on the size of your antenna. It's not appropriate to use a very thick coax line to feed an antenna with compact size. In addition, I suggest you first contact the company where you're going to send to fabricate you model, and make it clear what size of the SMA connector they usually apply. of course, it is the basic policy to make sure the characteristic impedance of the connetor is 50 ohm.
that example in the tutorial is an exception. in most cases, the dielectric constant of the substrate and that of the filling material in the coax line is different. however, the outer part of the coax line (except the inner pin) only penetrates through the ground(should choose "insert" in boolean operation in CST), but not through in the upper substrate. so the two dielectrics just contact, not intersect with each other. don't worry about that.
u won't forget to tick "helped me" if i really helped u, will u? hoho~

hi drunkbear,
I followed your instructions and have performed a simple CST simulation based on a probe fed circular patch (please unzip and see the attachment). For your information, my objective is to design a circular patch antenna which is fed to a 50 ohms coaxial connector (+ cable) and the antenna will have a resonant frequency of 866MHz.
Since the dielectric constant of the coax is solid PE = 2.33, I have calculated the needed outer and inner radius of the coax in order to obtain line impedance of 50 ohms.
outer radius = 4mm
inner radius = 1.12m
Also, the patch radius size is 6.55cm with a RT5880 substrate having dielectric constant of 2.2. After the simulation, I found that at 866MHz, the S11 is -20.72 dB. However, when I checked the line impedance of the waveguide port (2D result from Ports), it is 42.80 ohms and not 50 ohms.
Therefore, may I know whether the probe feed structure in my simulation is correct? What I want to simulate is a patch antenna connected to a 50 ohm probe fed.
Many thanks.

hi
well i simulated your structure and i got the line impedance of port equal to 49.7ohm
so i dont know why your reults are different from mine
but one important point , u will never get exactly equal to theorictal value of 50 ohm
as it depends on meshing density in the port region, the more dense mesh u have the more close you will get to therotical value
i hope this helps
regards

hi shahid78,
Thanks for your help. So, may I confirm whether my simulation approach is correct for the above scenario?
In the above example, the inner and outer value was chosen based on the CST tutorial: inner radius = 1.12mm, outer radius =4mm, dielectric constant = 2.33.
which gives a line impedance of 50 ohms. But what if my patch antenna is connected to a commercially available 50 ohms coax connector (when it is fabricated)? In that case, do I need to create an extra connector drawing in the simulation? Also, if the coax used for future antenna testing is a RG58 cable, should I change the inner, outer radius as well as the dielectric constant property of simulated structure based on the manufacturer's specs?
As you can see from the attached pic (antenna_fullground), the dielectric of the coax penetrates a thick layer of ground. In reality, the ground plane of the substrate will be fairly thin. It is made thick in the simulation due to ' port region must be homogeneous for at least 3 mesh lines in longitudinal direction' requirement. Can I trim off the ground plane so that it will look like this (antenna_trimgnd)? May I know which is the correct simulation method? In that case, can I still set the boundary coundition of Zmin as Et=0 or do I need to change it to open (add space)? Zmin was set as Et=0 previously because it is a thick ground plane.
Please help. Thanks.

hi zhiweisim
well both models are correct. however when you use trim ground plane model u cant use E=0 for Zmin
it should be open or open add space
u dont need to change the inner and outer radius of coax in the model because it doesnt matter which coax cable you are using practically. its characteristic impedance will be 50ohm
so as you have already modeled 50 ohm coax in simulation so no need to change its dimension
i hope this answers ur question
regards

hi zhiweisim,
i almost agree with shahid78.
at the initial step of our simulation, due to the limited discretization of the model, the results of the port (e.g Zc) may not be that accurate. personally, Zc between 45ohm and 55ohm(even 43 or 42ohm) is acceptable. it gives a more accurate result after the mesh refinement is conducted.
to ensure accuracy, i always construct the coax model with the exact same parameter as that of the manufacturer provides.
you don't have to make the GND that thick, i think the "3 mesh line" demand only requires that the length of the coax projecting backward from the GND should be more than 3 mesh line.

Hi shahid78/drunkbear,
Thanks you guys for the advice.
Do you mean that I need to add the structure of the coax connector for the simulation? or will a 50 ohms coax cable suffice enough to get the correct results?
I tried to simulate the structure based on 3 conditions:
1) thick ground plane , zmin set as et =0
2)trimmed ground plane, zmin set as et=0
3) trimmed ground plane, zmin set as open added sapce
I get three different S11 results (please see the attached pic). The difference is so large for the thick ground plane (Zmin set as Et=0) and trimmed ground plane (Zmin set as open added space)
May I know which one is the correct approach in simulating a patch connected to a 50 ohms probe feed coax? Please help.
Thanks.

hi
well in view there is not much difference in results if we are talking about resonant frequency.
so all of them are reasonably corrrect
as u will be having different mesh settings in each case so there must be a small difference in charactristic impedance leading to different dB values
can you tell me what is the difference in the 2 trimmed ground models?
if you post their models or picture that would help to se e the difference
but i hope this answers ur question

Hi shahid78,
The difference between the two trimmed ground model is that for the boudary condition, one of the Zmin was set as Et=0 and another was set as open added space.
From the initial_result pic (the attached JPG), you can see that at 866MHz, each has different S11 reading.
Also, I have extracted the minimum S11 from each cases:
1) thick ground plane (Zmin set as Et=0) = -33.24dB at 865.8 MHz
2) trimmed ground plane (Zmin set as Et=0) = -39.42dB at 867MHz
3) trimmed ground plane (Zmin set as open added space) = -31.72dB at 866.4MHz
I assume (2) is not the correct approach since the ground plane has been trimmed to thinner layer andf Zmin cannot be set as Et=0. Therefore, may I know (1) or (3) is a better approach?
Thanks.

hi shahid78,
Sorry my mistake. Correction to my previous post.
1) thick ground plane (Zmin set as Et=0) = -33.24dB at 865.8 MHz
2) trimmed ground plane (Zmin set as open added space) = -39.42dB at 867MHz
3) trimmed ground plane (Zmin set as Et=0) = -31.72dB at 866.4MHz
Assuming (3) is not the correct setting for boundary conditions, may I know (1) or (2) is the better approach? Resonant frequency for (1) and (2) are quite close, but why their S11 value are so different? I 've checked the line impedance of both ports and they are the same at 49.73 ohms. Therefore, may I know what is actually affecting the S11 parameter in this case?
Thanks.

hi
well as i said before its not possible to get 100% same results for different structure.
so in my view all your results are correct
however i think its better not to use Zmin boundary conditon as Et=0 for the trimmed ground plane case. as there will be small air gap between the actual ground plane and the boundary conditon Et=o at Zmin
i hope u understand what i want to say
regards