Differences and uses for XFDTD,HFSS, and CST

[Rancid]

Hello,

Could someone give me a quick overview of what the advantages of each of these programs are, and why you would pick one over the other?

Thanks

 

[Mustafa_Engineer]

As I know, the main difference between the commercial softwares is the computation method used , for instance:

XFDTD: FDTD based software
HFSS: FEM based software
Ansoft Designer: 2D planar MOM based software
.
.
.
.
etc

 

[iyami]

You might want to tell us what applications you have in mind.
Depending on
- size of the model
- materials used (types of materials)
- accuracy/speed requirements
- frequency range wideband/narrowband/single frequency
- ratio between wavelength / geometric details
the answer may change.

 

[Rancid]

I don't have a particular project (yet). I was just curious if there were particular rules of thumbs you could follow to let you know which programs might be more sufficient. I've used XFDTD in the past to do some antenna simulations, that's about it.

 

[jpsmith123]

One "problem" with CST is that it's so user friendly it has me spoiled.

With that said, here are some of my complaints:

(1) The TD and FD solvers in MWS don't seem to agree with each other very well.

(2) I've never been able to simulate anything involving a helical coil using the TD or Eigenmode solver in MWS, e.g., a helical resonator. (Is there a good TD solver anywhere that could use a tetrahedral mesh?)

(3) Some CST products don't seem to be very well documented.

(4) Unlike XFDTD for example, CST MWS lacks an ability to directly simulate the switching of charged conductors, e.g., Blumleins.

(5) When I simulate various configurations of coils and cores using the EMStudio magnetostatic solver, I get significantly different results depending on whether I'm using a hexahedral or tetrahedral mesh.

(6) When I've simulated certain relatively simple coil configurations, e.g., Brooks coils, for which there's an analytical solution, I got results that differed by 5 to 10% or so, IIRC.

(7) EMStudio field plots are generally lousy, especially with default settings, IMO. Many times, I've seen simple E and M problems with cylindrical symmetry for example, result in jagged, asymmetrical, artifactual, completely unphysical looking field plots. In order to get realistic looking, useful results, I've often had to go to a ridiculously dense mesh and set the "plot quality" to maximum. This is in contrast to Ansoft Maxwell, which seems to give much better looking (although not perfect) results with default settings.

 

[KP18]

I am wondering the same question. I am trying to model a very electrically small antenna in CST for the 400MHz band which all my geometry options for my application are going to give me very small efficiencies. I like in CST how I am able to easily co-simulate my matching network and recalculate my farfield. to see the improvement. However, CST is giving me an error saying the total efficiency is less than 1e-4 and farfield is innaccurate. Is this a limitation of the transient solver? Would HFSS do a more accurate job for this type of antenna?

 

[Tiberius8]

HFSS might help if you are looking at:
1) Single frequency
2) No lossy dielectrics

If not, you might want to look at another time domain code such as xfdtd from Remcom.

 

[tallface65]

HFSS might help if you are looking at:
1) Single frequency
2) No lossy dielectrics

If not, you might want to look at another time domain code such as xfdtd from Remcom.

Neither 1) or 2) are true. Please be wary of the difference in opinions and facts. HFSS more than sufficiently handles both lossy dielectrics and frequency sweeps with accurate results. Make your own choice, jsut be careful of those who religiously defend one code or another...including me

 

[volker_muehlhaus]

1) Single frequency

Indeed, there is a difference between the time domain and the frequency domain solvers.

The time domain solver will give a wide band frequency response from a single analysis (single excitation), but for n ports, it has to run n analysis to get the full S matrix. This is no big deal for a simple 2-port, but for complex model with many external + internal + tuning ports, the total time to get the full S-parameters will increase with the number of ports.

For frequency domain solvers, you will get the full S-parameters from a single analysis, but only for one frequency point. To get the wide band frequency response, the solver has to run multiple analysis, one for frequency point.

 

[jason_mri]

I don't have a particular project (yet). I was just curious if there were particular rules of thumbs you could follow to let you know which programs might be more sufficient. I've used XFDTD in the past to do some antenna simulations, that's about it.

hi Rancid, I am trying matching a microstrip coil for MRI in xFdtd, the coil resonates at 128MHz. Do you know how to match the coil to 50Ω coaxial cable?
Thanks in advance.