| T O P I C R E V I E W |
| Michael |
Posted - Oct 28 2011 : 10:06:49
Hi,
i'm new to this forum. I have some questions regarding workflow.
I'm on Windows XP. Not the best if you are forced to use open source sientific software, I know.
I need to model following:
1. pcb 50mmx50mmx1.6mm FR4
2. some conductive shapes Cu 35um thick
3. a floating conductive shape 0.5mm thick flying 1mm above the pcb (steel)
This will make the heart of a rotational sensor.
FastCap2 will not do the discretisation itself, right?
So i need a meshing tool.
I came across Gmesh and Netgen/Ngsolve. The windows port of Salome seems also promising.
Are there others?
What is the best tool for this?
How do i tell the tool that i need finer meshing at the edges?
What is the best mesh type (triangular or rectangular) for my entities and how do i tell this to the tool (pcb ist rectangular)?
I use Allibre to do 3D construction.
I was able to export iges and step and can read both into Gmesh/Netgen/Salome.
The pcb is a dielectric, do i need a mesh of the volume?
I think yes, because the field enters it.
Is a surface mesh of the conductors sufficient, as a static field can not enter?
Should i do the workflow separate for each element (pcb, each conductor) and combine it by a list file. Or can i export a whole 3D model from Alibre (as a part or as an assembly) and tell the meshing tool what is what and what is conductor or dielectric and do the appropriate meshing? Can FastCap then know that these are separate entities?
Should i model the conductors as surfaces or as volumes, as they are thin?
How do i get the mesh files from Gmesh/Netgen/Salome into FastCap2?
Hope that somebody can help, because i'm lost.
Best Regards
Michael |
| 1 L A T E S T R E P L I E S (Newest First) |
| aj21 |
Posted - Nov 09 2011 : 20:45:11
Hi Michael,
I hope you've had some success with your simulation.
I too am not quite clear on the limitations of the tool.
I am trying to model a 3 x 3 bus except unlike the busgen, the two buses are really thin (in um range), so is the spacing between them, so I created a custom code to for generating a qui file for this. But, I do not get good data with this setup.
With no descritization I get warnings regarding bad diagonal matrix and bad data, and with descritization = 3 with 0.1 Inner to Edge Ratio the setup gives more warnings: "capacitance matrix has non-negative off-diagonals", " Warning - capacitance matrix is not strictly diagonally dominant". Then it freezes.
I am guessing that FastCap will have problems simulating a thin 3D conductive parallel plates as well.
I am trying to find a way through to simulate this.
To answer your questions:
- Using Gmsh with FastCap
I believe there is an answer in the forum regarding this.
- Regarding your questing about using a comprehensive geometry file or a modular list file with component geometry files:
I believe if you are modeling two parallel plates for example with air between them, you would get the same result if you created a comprehensive file or if you created a list file with 2 single plate geometry file. In other words, unless you are planning on changing the dielectric between your conductive surfaces to something other than 1, then comprehensive file would be sufficient.
- Modeling Thin Conductors
Since I have modeled mine as 3D surfaces and it does not seem to work , perhaps we have to model them in 2D
- Meshing Files
I am not sure about this since, after I meshed mine the simulation started to freeze.
In order to mesh, yes you can use GMSH, but you probably have to create a program to parse the various vertices of the quadrilaterals and/or triangular mesh.
Or an option would be if your surface is not complicated you could write a script yourself.
I hope that helps.
Perhaps you or some other forum member could help me out with my issue :)
- AJ |
|
|
