can i used imepdance boundary for antenna

[aniakhan]

hi
i am simulating antenna but when i try to smiluate it with radiation boundary my simulation dose not complete but when i give impednace boundary to airbox it works and i got good results,can anybody help me to understand how does impedance boundary condition work in air box?does it acts like a resistor?ur ans will be great favour

 

[rodgerwxh]

At the beginning, you can understand impedance boundary as a equivalent resistor. It matches the impedance at the airbox boundary so that the wave sees nothing. However, this is ok in a way, but not rigorous.

The trouble is waves propagate along all directions, impedance boundary (IB) only perfectly absorb waves coming vertically. However, in most cases, this is not guaranteed at all, especially at gaze incident angle, impedance boundary almost absorb nothing. Either high-order ABC or PML or Boundary integral have to be applied in such a case.

So if you can predict that your field inside your problem domain is mainly TEM waves hitting the IB, then your result should not be bothered much. However, if your problem has all kinds of waves going out. Then the reflection might destroy your result.

 

[aniakhan]

hi.
rodger,
thanks for guiding me your ans really helped me,i have one more question by using impedance boundary i cant see radiation patterns and only see results of s parameter and resistance etc so how can i find radiation pattern?secondly i have tried to simulate same design with radiation boundary at air box and it took 15 hours for single antenna to be simulated and at end my results were very bad s11 was found to be 5 and impedance 0-40 ohm etc so is there any problem with my model or i should consider results of impedance boundary defined air box's result ok?hope you will help me again thanks for your answer

At the beginning, you can understand impedance boundary as a equivalent resistor. It matches the impedance at the airbox boundary so that the wave sees nothing. However, this is ok in a way, but not rigorous.

The trouble is waves propagate along all directions, impedance boundary (IB) only perfectly absorb waves coming vertically. However, in most cases, this is not guaranteed at all, especially at gaze incident angle, impedance boundary almost absorb nothing. Either high-order ABC or PML or Boundary integral have to be applied in such a case.

So if you can predict that your field inside your problem domain is mainly TEM waves hitting the IB, then your result should not be bothered much. However, if your problem has all kinds of waves going out. Then the reflection might destroy your result.

 

[rodgerwxh]

Generally, in EM simulations, many factors can lead to bad results, you need to do it step by step and accumulate your clues of what is exactly going on. For your case, if you said it takes 15 hours, then I am wondering what kind of solver your are using, FDTD, FEM? Frequency-domain solver or time domain solver? 15 hours sounds like a very complicated geometry or some of your criteria is too high?

Something I found very useful is that before you do your real model, start with something you know and simple, for example, a dipole antenna, or a simple ideal dipole source if your solver support it. For dipole antenna, you can calculate your S11 even by hand and you know the how the near field looks like and of course you know the radiation pattern is a donut. If your solver can predict all these without any problem, then at least you know your solver is good for simple cases and your port, field solver and absorbing boundary condition all works well without any critical error. Then investigate your model with careful steps.

If your real model is still problematic, try to reduce its complexity gradually.
For the question of calculating far-field pattern, for differential based solver, virtual box is used. And for time domain solver, you have to FFT your result in order to see pattern at a single frequency. Many text books (e.g. Allen's FDTD book) talked about this. Not a simple sentence to describe. Share more if you need further assist.

 

[aniakhan]

hi,
i am using hfss for simulation of cross bow tie antenna for working at 100 ghz,my problem is that when i take dimensions of antenna few mm(.4*.2mm) then i normally got results after 15 minutes,but if i consider antenna dimension such as 5*5 mm then with radiation boundary it took long time for single simulation.also can you help me in understanding one more thing?normally when we give impedance to port we want to match it with antenna's and if it happens we got good s11.but my problem is that my s11 is -10 db for very small range of frequency but when i check impedance of antenna it is 25 ohm only can you please help me to understand it?thanks for ur answers it is very informative for me

 

[rodgerwxh]

First of all, radiation boundary should not take that long to run. Even for high-order ABC it should be very fast, I implemented high-order ABC before, it should be very fast. PML and Boundary Integral method is relatively slow. I am not familiar with HFSS after version 9 on how they handle the radiation problem. (I used 7 years ago.)

Yes, when port impedance is equal to (match) the load impedance, you get maximum power output. However, antenna input impedance has resistance and reactance which forms the real and imaginary part of your impedance in frequency domain representation. And such a impedance is frequency dependent, this is how you get your S11 right? by calculating the reflection coefficient.

I do not quite understand your second question, S11 is the reflection, it is actually (ZL-ZS) / (ZL+ZS), where ZL is load impedance, ZS is source impedance. Make sure you are compare apple to apple and be clear of what you want to look at.

 

[aniakhan]

hi,
rodger,
in my case i have given lumped port resistance of 100 ohm and reactance of -100 ohm but in full port impedance i write 50 ohm as with 100 ohm i was getting s11 above -10 db.my question is that now i got s11 below -10 db in required frequency band so it means my antenna has impedance near about 50 ohm?also when i draw graph of z parameter then my result shows real part of impednace is almost below 60 ohm,but i have given 100 ohm at star to antenna is it ok to conclude that my antenna has resistance of 60 ohm or not?i am attaching my results of impedance(may be u dont have hfss 12)so its in ppt format,
secondly can you help me to understand what factors caused radiation patterns different from normal shape,bkz as i have modeled crossed bow tie so expecting to get radiation pattern of shaped 8 but it is quite different plz if possible guide me
thanks for answering
Yes, when port impedance is equal to (match) the load impedance, you get maximum power output. However, antenna input impedance has resistance and reactance which forms the real and imaginary part of your impedance in frequency domain representation. And such a impedance is frequency dependent, this is how you get your S11 right? by calculating the reflection coefficient.

I do not quite understand your second question, S11 is the reflection, it is actually (ZL-ZS) / (ZL+ZS), where ZL is load impedance, ZS is source impedance. Make sure you are compare apple to apple and be clear of what you want to look at.[/QUOTE]

 

[rodgerwxh]

Aniakhan, I was on travel these days and just got back. I say your example and want to give it a run on my own software and let compare it. Can you give me the detailed dimension of your design? Also where did you put your ground? From the file you attached, I did not see the ground defined, did you used one of the exterior boundary as ground? Also is your lumped source put right in the middle of the object? What kind of substrate you are using? You might want to check the mesh of your structure since I am doubting the sharp corners of your object might cause very tiny and fine mesh. That might cause your run to take 15 hours.

- - - Updated - - -

Aniakhan, I was on travel these days and just got back. I say your example and want to give it a run on my own software and let compare it. Can you give me the detailed dimension of your design? Also where did you put your ground? From the file you attached, I did not see the ground defined, did you used one of the exterior boundary as ground? Also is your lumped source put right in the middle of the object? What kind of substrate you are using? You might want to check the mesh of your structure since I am doubting the sharp corners of your object might cause very tiny and fine mesh. That might cause your run to take 15 hours.

 

[aniakhan]

hi,
roger,
sorry for late reply.i have not used ground,my antenna is on quartz substrate having thickness of .1 mm.my dimensions are l=.5mm and width =.42mm,alpha for bow tie is 40.now my problem is that on simulating it i am getting negative gain which is not usual behaviour of crossed dipole and whenever i changed width of lumped port its values varies from -13 to -5 db does it due to uncorrectly defined lumped port???but whenever i check validation i got no error.
secondly can you help me to under stand impedance concept? i have given lumped port resistance of 100 ohm and reactance of -100 ohm but in full port impedance i write 50 ohm as with 100 ohm i was getting s11 above -10 db.my question is that now i got s11 below -10 db in required frequency band so it means my antenna has impedance near about 50 ohm?also when i draw graph of z parameter then my result shows real part of impednace is almost below 60 ohm,but i have given 100 ohm at star to antenna is it ok to conclude that my antenna has resistance of 60 ohm or not?