Big discrepancy in CAD tools

These days we are trying to design a band pass filter at 77GHz based on coplanar waveguide technology. Although filters are known by their systematic design approaches, we faced surprises as usual. The Schematic simulations for ADS are far different than the momentum one. Moreover, the full field solver HFSS give us completely different results!
Any clue?

HFSS should be the best.

I don't agree. More degress of freedom to model things, but also more degress of freedom to make mistakes. From my experience in EM solver support, the number one reason for bad results in EM simulations are mistakes/misconceptions by the user.

The difference between ADS schematic and EM might be (missing) airbridges, where the circuit models don't predict all spurious modes.

This is going to go against the grain of what younger engineers have been taught, but you should NOT BELIEVE what ANY CAD program tells you! You can take the CAD results as an "advisement", to be treated with skepticism until lab results confirm what the program is predicting.

77 GHz, of course, is high enough so that closed-form models of structures are probably not fully developed yet. You should develop your own models in that case. Decisions running Emag simulators will get more difficult to clearly make.

That's especially true when you start using a new tool or technology. To get started, it's a good idea to simulate something where results are already known. Push things to the limits, see where your tool (or models) break. Then, after you have established trust in your tools and methodology, you can starting analyzing more complex, unknown devices.

You are right I have compared the three simulators at 2 GHz and the discrepancy is much less. But now the difference is huge, although I am using the finest meshing and most accurate convergence test!

We are now trying Sonnet solver to judge the results. I wish it does not give another different result :(

Good luck

For Sonnet, here is a description how to choose the correct port configuration for CPW.
http://www.sonnetsoftware.com/suppor...r_waveguid.htm