radiation efficiency hfss

时间：03-22 整理：3721RD 点击：

Hi all,
I am simulating a p@tch antenna. The S11 plot looks okay and radiation patterns too are okay but when I use the "Compute Antenna parameters" feature ( by right clicking the infinite sphere used to calculate the far fields ) , it gives a radiation efficiency of 1.013 . How can this happen ? I am using sufficiently big airbox ( more than Lambda/4 from each edge of the model ). Even when I do a tutorial from ansoft, it is giving 1.0043 as the radiation efficiency. I am highly puzzled. Can you point out what the reasons could be ? All comments are welcome. Thanks for your time.

-svarun

I've had this happen too. It's pretty easy to fix once you understand why it's occuring. The adaptive meshing is trying to converge the S parameters which is local to the p@tch, e.g., most of the tets will be near the feed of the p@tch. However the radiation efficiency is calculated from the power absorbed by the boundary, and this usually has very few tets.

In other words, since the field at the boundary has little effect on the S parameters, it is not calculated accurately enough to insure that the radiation efficiency is less than one. For your case there is at least a 1.3% error in the field at the boundary. To fix increase the mesh density at the boundary by seeding the mesh there.

can you explain this a bit further? how can I seed the mesh places I want.

svarun

This happens to other full-wave simulators as well. Your first thought on the cause of this matter was the airbox size. Well, this is not always the major cause of this. Several things need also to consider.

1. Your structure is assummed loss-less, therefore the structure has dielectric loss tangent set to 0, no conductivity value for conductive material, and background material is set as vacuum.

The above setting is enough to obtain 100% radiation efficiency. But do you realise computation does round-off decimal numbers, with literally millions of calculations of long decimal numbers take place during one simulation, soon the many round-offs could reach a final value a little over the exact value. As you have seen in your simulation, your exact value is 1, but the round-offs have made it a little over 1. That is something I would expect to see happening for a loss-less structure.

At this point, I hope you won't be telling me that you are using air as your background material and both the substrate and metal have lossy parameters. In which case, I personally don't think I ever got a radiation efficiency of greater than 98% for a p@tch antenna. If you do you should consider submitting an "PATENT" application. :D I only just woke up from a good night sleep, finding a 100% efficiency antenna was enough to wake me up to the full. An over 100% efficiency has wondered how long I have slept.

If assume you took loss into account and still got over 100% efficiency. Then you might want to go back to your radiation box problem and consider the following

2. Generally setting airbox boundary quarter-wavelength distance away from the nearest radiating edge (i.e. edge of the substrate) is sufficient. If you have a high performance PC with decent amount of memory >1GB, then have the radiation box larger say half-wavelength.

3. But best thing to do isn't make larger of the radiation box. Instead keep it at a quarter-wavelength distance. Then build PML layers on the faces of the radiation box which the strong radiative wave is most likely to hit. I would suggest start with just a PML layer on the face of the radiation box perpendicular to the boresight of the antenna or the face facing the p@tch element.

What you might find is this might not help, as the result of this is that we have produced a very good abosrbing boundary would usually improve leads to slight improvement in the radiation efficiency. Another way to look at this is less reflection is caused by the radiation boundary and minimise distortion with the on-coming radiation.

I hope this help. I am a very busy this few days. I hope someone else could give you advice during this time. All the best!

Sassyboy

can you explain this a bit further? how can I seed the mesh places I want.

Have you tried to use Gain / Directivity to calculate the radiation efficience? How is the result, still great than 1?

hi .....
I face the same problem now. Have anybody solve the problem? Can you tell me how to do???

Thanks...

Hi,
Gain/Directivity ratio was the first thing I looked at and as expected in this scenario, it is greater than one. By the way , seeding the mesh in the airbox to λ0/6 also did not help. Increasing the airbox size to λ0/2 in both -z and +z direction also did not help ( It gives 1.0045 as radiation efficiency )

Admittedly my structure is lossless ( Lossless dielectrics and PEC ) but I guess not all fields are radiated . Some part of it is still in the substrate and actually some surface modes might be excited ( The TM0 mode will always be excited ). The same structure simulated in IE3D or MWS gives something like 78% radiation efficiency and in one case at a very high frequency IE3D gave only 45% efficiency which is perfectly understandable. I am a bit confused about the cause of this. Is it a bug or am I making some silly error somewhere and getting into these problems ? I am using HFSS 9.2.1 on a P4 machine I would be grateful to anyone who can throw more light on this.

-svarun As my expectaion, 1.0045 is accurate enough, the error is only 0.45%, I can accept it. You model is lossless, all the energy should radiate out. If you really need to increase the accuracy further, I suggest you use PML to cover your whole model which is λ0/2 away your model. And the convegency criteria of the port S-parameter, try to set it as small as you can(very time consuming).
Once you introduce the metal loss and dielectric loss, you will find the calculated radiation efficiency is quite good enough.

Best Regards,

hi...
I try a case for monopole. I don't use wave port and coaxial line and I use one iron club and iron plane by lumped port to simulate monopole.
I also get the efficiency is greater than 100% and even change the iron to lossy material.
I try other antenna case. When the reflection coefficient =-1, the efficiency=56%!!!!
I can't believe it!!!!!

Thank......

Hi all,
I simulated another p@tch antenna with copper ground and p@tch and air as the air box material and a lossy Gallium Arsenide substrate ( Loss tangent = 0.006 ). I have big air boxes and fine mesh and the far field sphere computed efficiency is still 93 %. The same structure in CST-MWS gives 83 % . Thanks a lot for all your replies.

-svarun

Hi Haydar,
This is what is highly puzzling to me. Almost all other parameters are in excellent agreement except for the radiation efficiency part ( which also means that I have to doubt either directivity or gain values ) . I cannot use any finer mesh (lambda0/6) nor can I increase the airbox size as I am already runnung out of disk space and time taken is approaching days instead of hours. So I guess I have to take MWs results as accurate as most of the MWS results seem quite reasonable. It is a pity that I cannot do any fabrication and measurements in our lab. The facility is still coming up. Please let me know about any measurements that any one has done even for simple p@tch antennas.

-svarun

hi svarun
In our antenna lab, there is a tool designed for paatch antenna far field pattern and other applications. I have not used it, because I study on near field. however, in these days, some of the instructors are planning to use that tool.
If I will be able to help you about measurements, I will let you know. before that, I should learn some details of the tool.
regards

I made one PIFA in GSM band. The measured efficiency is . The simulation result is 91-93% by HFSS and 84-96% by CST MWS. The material are loss.(Copper,conductivity 5.7e7 S/m, Polycarbonate:2.7;0.02). Hope to get some ideas from profesionals.

Added after 6 minutes:

The measured efficiency is 31%-45%

Hi, Rainli,

What kind method you used to measure the radiation efficiency?

Regards,

I used 2D chamber and network analyzer to test gain in AZ, EL0 and EL90 plane. Then make a approxiate caculation of efficiency.