ADS EM simulation error

Disable Krylov Subspace Solver and compare results.

Change Compression Level from "Normal" to "Reduced" in Options-->Solver tab in emSetup Controller.
Or
Change the Solver from "Automatic" to "Direct Dense"

simulation finished with error.



im trying other suggestions now



again simulation terminated with error!

You have too little memory (RAM) for the number of mesh cells. Maybe you can simplify the mesh.

the ram is 128GB. is it too little?

Of course it's not little but if your structure is too big and mesh density is to dense, the memory problem may occur.
Have you ever tried it with FEM ? It sometimes gives more information about the structure.

It depends on your model. The disadvantage of MoM solvers: the maxtrix size grows with N^2 where N is the number of mesh cells.

Recently I did some inductor + dummy metal fill models, with 130k metal polygons. For that I used FDTD method because FDTD memory requirement grows with N^1 only. This is an advantage for models with large polygon counts. I don't think my model with 130k polygons would ever be possible with MoM because you "hit the wall" with matrix size.

http://muehlhaus.com/support/rfic-em...-filler-effect

ADS have 3 Em simulators. Momentum RF, Momentum microwave and FEM. I'm using momentum microwave because momentum RF is giving an error for my frequency range. how FEM better than momentum microwave in solving the accuracy warning issue i have reported above?



is it means that i have to switch to FEM EM simulator? i cannot find FDTD solver under momentum microwave EM simulator. Matrix solve method available for Momentum microwave simulator are auto-select, direct dense, iterative dense and direct compressed.



I'm trying to simulate a layout for an RF design. this layout will be made into a model and to be simulated with the lumped elements in the schematic later. its not a standalone EM structure like a balun transformer you shared. so is FDTD solver is suitable for just interconnect layout like mine?

You can try FEM. FEM memory dows not grow as fast as MoM, it might work. If you have never used the FEM engine, do some reading before you start, and try a few small testcases to learn using FEM for RFIC properly.

FDTD would be best for this. ADS does not have an FDTD simulation engine, that's why I switched to a different tool (superfast FDTD implementation). Keysight EMPro has FDTD, but that is much slower than the tool that I used (Empire XPU).


How complex is your model then? You might need to divide your layout into multiple pieces that can be simulated independently.

Is FDTD suitable? I can't answer without seeing the model.

Dear Volker,
i have added you as a friend in edaboard so i can share the design. i read your articles about Empire XPU. its sounds very interesting. for my current project, i almost done so i shall try Empire XPU for next projects. i shall reccommend Empire XPU to the university since its sounds interesting.



my design is not RF IC design but just discrete design. discrete design is PCB design.

Hi pragash, you can send me your model by email, so that I can have a look why it takes so much memory.
https://muehlhaus.com/contact

(deleted, mistake)