compare results hfss momentum sonnet
I designed a set of microstrip hairpin filters at different frequencies. I simulated them using Momentum and HFSS and I had a chance to compare it with the measured data.
I was amazed to see that the Momentum Simulation matched exactly with the measured S parameters while the HFSS Simulations predicted a higher resonance frequency and a higher Bandwidth..ie.. a wider skirt for the band pass filter.
For the entire set of measured Vs Simulated data, HFSS predicts a higher resonance frequency... say by 0.4 GHz at 20 GHz. Momentum and HFSS simulations were done exactly on the same design under identical conditions. So far, from my experience, HFSS seems to predict the magnitude pretty accurately but is shifted high in frequency. This is the case with matching networks too. I need to design 0.5 GHz higher in frequency with HFSS so as to match at the center frequency.
I thought HFSS being a full 3D simulator was the most accurate among the lot, but based on the measured results, I feel Momentum is much better .. at least for planar circuits.
I would like to know if other EM software users have similar issues with microstrip filter simulations or matching networks. I would appreciate your responses.
Thanks,
Xyman
Hi,
Just a few words about my experiences and opinions. There is an old story that a person with one watch in his house allways knows exactly what is the time whille one with two watches can never be too sure.
The same seems to be with EM simulators. The more you compare them the less you can be sure which one is the best. As was writen in some books, no one is perfect for every job, so you can solve the problem in two ways - 1. find the best for each and every particular type of problem or - 2. find the one that will not be perfect and most accurate for any particular job but will have acceptable error for all the problems.
Both approaches can be justified and one should chose the one that suites him most. Personaly I think on each simulator as a solution 2 - after all, EM simulators are just tools, not the absolute truth.
To come back to your comparison of Momentum and HFSS. I also have some very good experience of Momentum accuracy in analysing planar structures, but have to say the solutions from IE3D and Sonnet were insignificantly different. HFSS can do planar structures, but it is not the field where it can give most - it is ideal for waveguides, antenna and other similar, real 3D structures.
Finally, filters and resonant structures are very tricky to analize precisely, if you make a very small error in reactive part you will get considerable shift in resonance.
Regards
flyhigh
If you observed systematic difference between HFSS results and the measurement (always 0.5G lower as you said), then you should check whether your model has been set correctly to correspond with your measurement. The measurement, especially the calibration, is also tricky.
If the HFSS results are offset randomly (sometimes higher, sometimes lower), then you have more of a case against it.
I think both HFSS and Momentum are quite adequate for simulating microstrip filters. (They might not be fast enough to be used for optimization/synthesis). It is very likely that the model has not been set correctly.