car antenna hfss
Another question is about the limits of HFSS, someone knows about the limit size mesh that HFSS can simulte?
cristiano shalom,
Theoretically, HFSS can solve anything :)
Practically it is limitted by the hardware resources (memory ussually).
The first thing I would do is to post here your simulation profile of a model that convereged and gives you satisfying results (Results->Solution Data->Profile. Click the export button). This will allow to evaluate the memory needed by the solver.
HFSS users measure the size of the model in wavelenghts. By increasing the mesh adaptation frequency from 500 MHz to 1GHz you have decreased the wavelength by 2, but increased the size of the model by (at most) 8 (2X*2Y*2Z=8*X*Y*Z).
You should contact an Ansoft AE and ask about the following:
1. Use the new HFSS v10 model healing /defeaturing that allows you to automatically convert the mechanical model into a simplified electromagnetic model.
2. Use the new HFSS v10 mesh model resolution feature to reduce the mesh before the solver.
3. Use mesh surface approximation in order to trace the curved contours with a rougher mesh (the default is intended for antennas, but who cares if the metal around the wheels will not be exactly round).
4. If nothing helps you would need a EM64 bit computer with more RAM. I do not know how much more. You should ask the AE to approximate based on your lower frequency solutions how much more RAM will be needed in the higher frequencies. FYI, 32 bit Windows XP with 4GB RAM will allow the HFSS solver to use 3GB RAM. Above that 64bit is needed.
Regards,
Itai
Did you really design the Car fullsize?
I think it is sufficient to simulate a smaller car. I'm sure it makes no difference in the results, whether the car has a length of 10*lambda or 2*lambda.
Depending on the excitation you may be able to split the car with symmetry planes Perfect_E or Perfect_H.
I can confirm what itaifrenkel said about the RAM. A 32bit Prozessor cannot adress more than 4GB of memory. Your RAM minus Grafics-RAM and some other memory leaves not much more than 3GB for Windows, my XP shows 3.25GB available. HFSS v10 has a 64bit Solver but there is no ready version of the new Windows out yet.
thanks a lot. I′m trying to use the symmetric plane but the results isn't so good yet. maybe the 64bit computer will cover my spectactive but I cant how to test one. I hope to find same tools that optmize the solution of HFSS. Sorry for my english, because i'm not so good.
I think such a structure (a full sized car) is very large (you are simulating a structure that is around 12 lambda * 5 lambda * 7 lambda). For such a large structure matrix inversion used in FEM used in HFSS would be extremely lengthy. Have you tried time domain solvers, such as CST Microwave Studio or XFDTD.
Best Regards,
Adel
I think you may be try CST software to model it.
I agree with the other posts, this is not the correct problem for HFSS. Try a higher-frequency tool or hybrid approach. You can remove the tires and whatever's in the engine compartment if you are looking for ways to cut down the model size. The shell of the car should be sufficient.
Again, use another tool if possible. Something that is much better suited to this type of thing such as FEKO or a Shooting and Bouncing Ray (SBR) code. You can try MicroStripes, which is a TLM code (also time-domain). CST is also time-domain which may be somewhat better.
I agree with antennas321 that it is not a good idea to use FEM-based code such as HFSS for full-size car simulation, especially at high frequency. You may try some MoM based code, such as FEKO, which also has Multilevel fast multiple method (MLFMM) built-in and good for such applications. You may find attached a paper doing similar simulations with a full-size car using FEKO.
I think that CST does not work with this car. You should use Feko or similar. If you want to try with CST and HFSS, you scould not model the wheels and the engine. More the strusture is simple more the result is good. Reducing the size of the car is not a good physical approximation. Are you working on EMC problem?
Hi,
that is a pretty challenge doing this in HFSS.
Actually, if you work is EMC related, the pioneer and leader is Flomerics, with its 3D EM solution MICROSTRIPES.
one of paper I remember was "Comparison of measured and computed electric field distributions due to vehicle-mounted antennas" done by Mira (automotive)
you may want to check MIcroStripes, where their Transmission Line Matrix (TLM) method to solve Maxwell?s equations in the time domain would better suit your application....
good luck,
AW
Hi,
I just checked the CST application web site. They have a short article on CAR simulations (Cable Harness and antenna placement) on their web site.
http://web.cst.com/Content/Articles/article264.aspx
Just have look. For electrical very large structures FEM codes are not particulare well suited. Time Domain or MoM (MLFMM) codes are better suited for this kind of problems.
F.
hi,
antennas231, can you suggest/upload a free SBR code?