| Author |
 Topic  |
|
|
Sven
Germany
10 Posts |
 Posted - Mar 19 2020 : 12:19:42
|
Hello together,
I am right now working on parasitic extraction of RLC of PCBs.
But for the modeling of ground planes I have some problems.
On my groundplanes I have some conducters. In FastHenry I modelled a ground plane and used the holes directive in order to cut holes for the conducters.
Is there any possibility to do the same in FasterCap as well?
So far I haven't found any clues neither in the documentation or in the forum.
It would be very nice if someone could help me.
PS: I am using this FasterCap file format (not this special file format for FastCap)
|
|
|
Enrico
530 Posts |
Posted - Mar 20 2020 : 12:05:06
|
Hi Sven,
your question in the FastHenry forum topic was clear, while I'm not 100% sure about your doubts about FastCap / FasterCap.
FastCap / FasterCap uses planar triangles or quadrilaterals as basic modelling element. You need to mesh your object conforming to the shape you have if you want to model complex geometries. Therefore if you have a ground plane with holes, you need to approximate its surface using triangles.
Could you also elaborate a bit more about what you mean by "this special file format for FastCap"? What special file format do you refer to? Patran maybe?
Best,
Enrico
|
 |
|
|
Sven
Germany
10 Posts |
Posted - Mar 20 2020 : 18:33:12
|
Hello Enrico,
thank you very much for your reply.
Maybe an example may clear things up:
I have a ground layer (conductor) and on this ground layer I have a wire (another conductor).
In FastHenry I create a ground plate, create some holes in it, and create some nodes and elements in order to create a wire inside that hole.
So is there an option to do this in FasterCap as well?
Or do I need to "dismember" that holey plane and design it with several rectangular and triangular shapes?
The second option sounds very tedious.
I am using those .lst and .txt files (not .qui files from FastCap2). |
|
|
|
Enrico
530 Posts |
Posted - Mar 23 2020 : 15:56:29
|
Hello Sven,
the file format between FastCap/FasterCap and FastHenry is completely different. The former uses surface patches as primitives, while the latter uses segments (i.e. parallelepipeds).
Manual generation is of course difficult for complex structures. The options normally used are:
1. scripting the generation of the geometry
2. start from a 3D model, mesh it and generate the FastCap/FasterCap input file
Regarding the latter, we have created a set of scripts you can use in FreeCAD (w*w.freecadweb.org) to ease the generation of the FasterCap input file. The end goal is to have them in a workbench just like implemented today for FastHenry in the E.M. workbench for FreeCAD, however this activity is second priority unless we find a sponsor willing to fund us for this improvement, so we can kick-up the priority (the software is all open source and free and we work on consultancy / services).
Kindly let me know if you are interested and if this option would be valid for you.
Best Regards,
Enrico
|
|
|
|
Sven
Germany
10 Posts |
Posted - Mar 27 2020 : 13:14:34
|
Hello Enrico,
thank you for clearing this up.
It looks like I am going with the meshing mehtod for my application.
Do you have an idea if the mesh size affects the simulation time?
For example: Is there a signifiant difference if I use a mesh of 100x100 with 1mm˛-squares or if I use a 1.000x1.000 mesh with 0.1mm˛-squares.
Best
Sven |
|
|
|
Enrico
530 Posts |
Posted - Mar 28 2020 : 18:14:52
|
Hi Sven,
if you are using FasterCap, it handles the discretization refinement for you, so as detailed in the embedded online help, you should use the most coarse discretization possible that still represent your geometry.
FastCap is different as it is not able to refine the discretization and will use the input one. So you need to 'manually' progressively refine the input file until you reach convergence (i.e. the result of two consecutive simulations with a significative finer discretization yields the same result within 1%).
Using from time zero a hugely discretized geometry will only needlessly increase the simulation time - even if usually the bottleneck is not the speed, but the memory (that in the end affects speed as the OS will start using swap memory from disk and everything will slow down to a crawl; unfortunately under Windows there is not an official method to avoid that)
Best Regards,
Enrico
|
|
|
| |
Topic |
|
|
|
Holes in Plane
