| Author |
 Topic  |
|
|
rick
2 Posts |
 Posted - Dec 28 2023 : 05:28:49
|
Dear Enrico,
1.When using FasterCap, while dealing with multiple dielectrics, especially those with a high permittivity ratio, the results are not very satisfactory. Issues such as asymmetry in the diagonal of the capacitance matrix and non-negative elements occur. Does FasterCap have the capability to handle high permittivity ratio problems?
2.For FasterCap, what kind of mesh is considered the optimal mesh? I have tried inputting the original CAD model, or meshes processed using some mesh tools, but FastCap still requires refining the mesh to achieve good results.
Best Regards,
Rick |
|
|
Enrico
530 Posts |
Posted - Jan 03 2024 : 00:14:41
|
Hi Rick,
quote:
1.When using FasterCap, while dealing with multiple dielectrics, especially those with a high permittivity ratio, the results are not very satisfactory. Issues such as asymmetry in the diagonal of the capacitance matrix and non-negative elements occur. Does FasterCap have the capability to handle high permittivity ratio problems?
Yes, you are right here; for the details, you can reference J. Tausch, J. White, "Multipole Accelerated Capacitance Calculation for Structures with Multiple Dielectrics with High Permittivity Ratios", ACM Proceedings of the 33rd annual Design Automation Conference, 1996 (it is cited in the Literature->Publications section of our web site under FastCap2).
In a previous paper by Keith Nabors (that anyway fails to pinpoint the exact source of the issue, as done later by Tausch and White) it is mentioned that using a Galerkin method helps reducing the impact of the issue. You can try that in FasterCap - but it is only a mitigation.
quote:
2.For FasterCap, what kind of mesh is considered the optimal mesh? I have tried inputting the original CAD model, or meshes processed using some mesh tools, but FastCap still requires refining the mesh to achieve good results.
Best Regards,
FasterCap will run at least two simulations in automatic mode, as it needs to compare the results of a second simulation with the first, to understand if it reached convergence or not. You can manually force FasterCap not to refine the input geometry, if you believe that is already sufficient for your goal. In this case you must play with the -m parameter and set the mode to manual (un-select automatic mode).
Consider however that the best results from FasterCap are obtained when FasterCap can refine the mesh on its own; you should always start with a mesh as coarse as possible, provided that it can represent correctly your geometry (e.g. if you have a square plane, a single big square works best; if you have a sphere, you'll need instead enough triangles to mimic the curvature to the level of your satisfaction).
Best Regards,
Enrico
|
 |
|
| |
Topic |
|
|
|
In FasterCap, the high permittivity ratio
