Which one is better, HFSS or CST ?

[loucy]

How well they are? 0.5dB, 1% shift of resonance? Numbers are more useful.

 

[shou]

An 16 elements plane antenna, with 3.5% shift from the design resonance freq. the gain is about 1.7dB lower than design.7.5GHz

 

[yuyu]

Anyway, HFSS is very inefficient about memory resource since is a FEM based and this numerical method is not practical for radiation problems.

 

[boy]

I think speed is importance, coz when the structure is build, it still needs some fine tune. The simulator just give a guide line. That why i like CST MWS.

 

[yuyu]

Hi,

MWStudio is based on FIT (Finite Integral Technique) that is applied to Cartesian grids and combined in time domain with the leapfrog scheme; FIT is applied to the integral form of the Maxwell equations and is computationally equivalent to FDTD (FDTD is applied to the differential form). Thus, FIT shares all of FDTD advantages regarding memory needs.
Maxwell equations are solved in integral form instead of differential form and that is more suitable to develop locally conformal schemes.

Regards

 

[edgerate2222]

i've experience many memory resource problems with hfss myself.

 

[wolf69]

seem great softwares.. (HFSS and CST)
but how much do they cost?
Anyone is able to do a comparsion?

Bye All!

 

[eirp]

Hi, friends!

Just for fun... - my friend has built corner-reflector antenna designed by CST MWS. Input parameters has been also measured using Sitemaster (Anritsu). In attachment you can find comparison - I think it fits well.. Due to non-smooth measured curve I suppose that Sitemaster had some problem with calibration...
I can send more data and mws file if you're interested, someone can simulate in HFSS too ...
Best regards,
Eirp

 

[goxy]

Eirp, it's very interesting, so please send and other data.

 

[eirp]

Here goes the structure in MWS. It isn't my project, I only know that it's designed for highest gain, not best matching.
From the file I see it's also a bit "overmeshed" - 30 cells/lambda...
Somebody can simulate it in HFSS for comparison??
If you don't have MWS, I'll post picture of this simple antenna.
Best regards,
Eirp

 

[plasma]

hfss vers. cst Horn

:? :? :?

hello

attached here an exemple off high frequecy 18-40GHz Horn
not metched
you can see that s11 same but in higher freq. hfss
get s11 more then zero.

RGZ

 

[eirp]

Re: hfss vers. cst Horn

:? :? :?

hello

attached here an exemple off high frequecy 18-40GHz Horn
not metched
you can see that s11 same but in higher freq. hfss
get s11 more then zero.

RGZ

Hi, plasma!

Thank you for your nice example. Please can you give us also times needed for solving by both programs?

Cheers
Eirp

 

[plasma]

Simulation time

eirp

its a small problem so it almost the same time
12min 41sec CST and 13min HFSS

plasma

 

[pewang]

Both HFSS and CST MWS are power tool.
But HFSS is more professional than CST MWS.Althoug this will lead itself more complex and hard to control.But it is true that HFSS is more helpful than MWS for solving complicate problem.
CST MWS is easy to use,good GUI.But it leave small space for user to setup more detail.For some problem,this is annoying.
also MWS has shortage when solve resonate problem.
For Antenna problem ,MoM will be the most suitable method because no absorb bondary problem,I think,although I have never simulate one antenna.

 

[yuyu]

Both HFSS and CST MWS are power tool.
But HFSS is more professional than CST MWS.Althoug this will lead itself more complex and hard to control.But it is true that HFSS is more helpful than MWS for solving complicate problem.
CST MWS is easy to use,good GUI.But it leave small space for user to setup more detail.For some problem,this is annoying.
also MWS has shortage when solve resonate problem.
For Antenna problem ,MoM will be the most suitable method because no absorb bondary problem,I think,although I have never simulate one antenna.

Dear Pewang,
I don't agree about your considerations.

1) What do you mean with "HFSS is more professional than CST"? It is a not clear definition.
2) It is not true that HFSS is more helpful than MWS for solving complicate problem; true is on the contrary.
3) It is not true that MWS leave small space for user to setup more detail; probably you don't know well MWS.
4) Resonant problems can solved by autoregressive filters or by frequency domain feature.
5) MWS is much more suitable to solve antenna problems than HFSS; more efficient (speed, memory resources) and more accurate. MoM is not so general purpose method as FDTD method.

Regards.

 

[cheng]

HFSS vs. CTS

Yuyu,

When you claim the frequency domain solver in MWS is any good please be honest and tell me if you have ever used it. I doubt you did.
This solver is so fu..in slow and inaccurate that it renders useless your resort to it.

When you say MoM is not so general method - who cares how general a method is as long as you can use it to solve your problem. And MoM is an awesome technique as well as FEM and FDTD.

Yours truly,

cheng

 

[yuyu]

Cheng,

I used MWS for 4 years from version 2. The problem is that; many people
talk about MWS but they didn't use it. It is difficult to speak about a thing not known. Frequency domain is an efficient feature and further I want to remember the other technique based on autoregressive filters.
MoM method is not a general method as FDTD or FEM; that is obvious.

Regards.

 

[yuyu]

Hi,

I want to add another technique to overcome resonant problems with MWS.
Since simulation time is increased by strong resonances, very often it is possible slice the structure such that each part has no resonances. Then, it is possible to calculate separately S parameters for each part using MWS without waiting for a long time. Finally it is possible combine the S parameters by using a circuital simulator (ADS, Microwave Office, Aplac, Serenade, Design Studio ...).

Regards.

 

[cheng]

HFSS vs CST

well, the freq. domain solver in MWS is lagging far behind the solver in HFSS and that is quite understndable.
I can understand the folks in MWS to be "all inclusive" but sorry to tell'em there is no such thing. And will never be. And that's the whole beauty - else we will be hostages in their hands. The problem with the separation of the problem (or fraction it down) is quite popular but is not a magic. Especially if your structure needs be simulated with a number of modes (for filters most of them are below cut-off) where accuracy deteriorates rapidly.

When you take a mode below cut-off in a WG it does have some 'initial', (at the port) level of S11. It should be 0 theoretically but numerically is not. If now I take a lenght of the WG that is a few lambda (and if my mode is the 3-rd below cut-off with a huge Gamma) I may end up with S21 that is +50dB (no mistake). And when you combine data this will lead to nill. Therefore it is very important the wideband, frequency domain calculations on a port level (and close to the discont.) and HFSS does an excellent job there. CST is trying hard to match there but is quite far yet. For ppl that have designed narrowband (or even wideband) filters, it is not easy to utilize CST (here you should trust me ).

Again, that is not to say that CST is a piece of chunk - definitely it is a very valuable code. My only point is that such "comparison" is strongly irrelevant. One should be able to use every tool possible that leads to a good design. The rest is just a speculation.

cheng

And one other thing I hate about CST - they continously put down others' tools and get even irritated if one dares even mention the existence of other solvers than their miraculous tool. They get fu..in offensive. I hate that bud

 

[yuyu]

Hi Cheng,

1) I never said "separation of the resonant problem is magic"; that is a chance to overcome long time simulation. Sometimes is possible to slice the model with dominant mode at interfaces, sometimes not; it is obvious. Anyway there are two other possibilities; frequency domain or autoregressive filters.

2) I never said "MWS is a miraculous tool". I replied to Pewang's considerations that, in my opinion, are wrong.

Regards.