2.5d simulator accuracy - some info

[toonafishy]

vivaldi .hfss

I just finished a microstrip balun design and simulated it in M/W/O/(2.5D) and H/F/S/S (3D). The results were similar but not exactly the same. which got me thinking. In you experience, which 2.5D simulator results most closely matched your lab measurements of the final circuit. I'm thinking of M/W/O, S/o/n/n/e/t, Mo/m/e/n/t/u/m, E/n/s/e/m/b/l/e, etc. When I build it in the lab, I'll let you know how closely the results matched the sims.

 

[loucy]

I think the measurement is also very tricky. The error there should be taken into acount.

 

[shou]

I also interesting in the accuracy of em tools such as H@f55 ,Mom, MW0 cT5 and 5snnet.
for antenna design, i think the @gi1ent Hf55 is accurate.

 

[sugm]

for 2.5D EM tool
accuracy
planar structure modeling
Ensemble, IE3D, Sonnet
planar antenna
IE3D

 

[loucy]

I think Momentum performs best in terms of accuracy.

 

[shou]

what are you design using MOM,?

 

[harkonnen]

The accuracy all depends on many factors.
For example, is it a MMIC balun or a PCB microstrip balun?
Is the gap between the lines in the same magnitude as the thickness of the metal lines? It also comes down to fact if the skin depth is smaller than the metal thickness.

And most importantly: do you really know the exact geometry and dielectrics of your structure?

You have to take much care in evaluating your simulations and measurements.
The "best" simulator always is only "best" for a specific structure, application and goal (design time?).

 

[filterman]

1. Prof. David Pozar published a paper some years comparing those various tools:IE3D(Zeland),HFSS,Momentum, ensemble,etc.

He wrote that results were quite close, Momentum and IE3D were very accurate (but IE3D was much faster).
I designed a filter not long ago , and there were differences between Momentum, MWO and HFSS. we decided to follow the HFSS results for the production prototype.
Now the biggest difference was in the Return Loss parameter, which for filters is the most sensitive parameter in pass-band. There were differences in the rejection also , so we decided to implement HFSS version. But we still don't have results.
An collegue of mine compared MMIC passive simple components(caps,inductors,etc) and he claimed that Sonnet, MWO,Momentum, HFSS gave for simple cases very similar results compared to mesurements. so he is convinced that they all are the same.
Well for precise work you have to use the correct tool, and as we all know for certain work they should perform the same, but for other work each method has its advantages and cons.

 

[eirp]

1. Prof. David Pozar published a paper some years comparing those various tools:IE3D(Zeland),HFSS,Momentum, ensemble,etc.

Please, do you have these papers? Can you upload it??
Thanks in advance!

Eirp

 

[loucy]

Someone posted a conference paper written by the group of Pozar. I saw it in the disscussion on benchmarking, but I couldn't find it now. IE3D 4.? was used, I think. There is not much useful information on improving the accuracy of simulation.

 

[sviodo]

about 2.5D simulators:

In my experience properly used sonnet gives most accurate results just before momentum,
ie3d is good for "optimalization" because of very fast analysis but it's wise to finish project in one of above simulators.
Never used ensemble, no comment about this proggy then.

about 3D simulators:
I've just checked cst mw studio and i have to say that it's very usefull software for education purposes because of very friendly interface and easy postprocessing however hfss is much more PRO. If you use macros and you are able to bear strong cpu/ram requirements it's still the best choice. Moreover I don't agree with people who claim that this proggy lacks on accuracy.

With Regards,

 

[filterman]

D. M. Pozar, S. M. Duffy, S. D. Targonski, N. Herscovici, “A Comparison of Commercial Software Packages for Microstrip Antenna Analysis”, IEEE International Symposium on Antennas and Propagation, Salt Lake City, Utah, July 2000.
As you may observe it is not so old, they did not try to find the best, but to compare various criteria by tests run be experienced user and not by the companies themselves as is done regularly in comercial papers.
One interesting aspect was that software that might seem fast to learn for beginners might not be the fastest when used by experienced users of that software.

 

[loucy]

It seems to me that Sonnet and MWOffice give the same results when the circuit fits into a uniform rectangular grid. The main difference between these two is Sonnet provides a "diagonal fill" for irregular geometry, but this doesn't always improve the accuracy. Sonnet also offers some optimization features--sweeping some geometry parameters automatically.

When comparing Sonnet with IE3D or Momentum, there is always some shift in frequency, no matter how big the enclosure is in Sonnet, or whether the enclosure is simulated in IE3D.

 

[harkonnen]

Here is the paper

 

[Lupin]

I've realized microstrip patch antennas (probe-fed, single and multilayer structures), microsrtip and stripline components in UHF,L,S,X Bands.
The e.m. adopted code has been Ensemble 5,6,7,8 and CST Mws.
I can say that the measured data matches the predicted data in the most of the case.

Hope it can be useful.
Regards

Lupin

 

[Lupin]

I've no idea bout HFSS.. can you say something about it and its accuray?

Thanks in advance
Lupin

 

[toonafishy]

For HF$$ vs M*W*O* I find HF$$ underestimates the loss but M*W*O* underestimated the phase balance. This is for PCB microstrip stuff I'm working on.

 

[eirp]

BTW, IE3D 9.38 is out, fixing some minor bugs.
Z/e/l/a/n/d guys are very active with work on IE3D :lol:
Regards,
Eirp

 

[filterman]

EM software in general cannot calculate the influence of roughness on the loss, leading usually to optimistic loss.
To calculate loss they need to apply correction formulas based on experience or measurment. tHAT CAN CHANGE RESULTS BY 0-100% OR MORE.

 

[shou]

All the eda tools are just a tool, they can make our design easily, but they can not instead of experiment.