| T O P I C R E V I E W |
| Lavdim |
Posted - May 05 2012 : 01:51:38
Hi!
Is it possible to scip inversion and get directly the Potential Matrix, instead of Capacitance Matrix, from FastCap2?
If its possible, can it be used with addition of delay term to simulate time retardation problems?
Best Regards
Lavdim |
| 3 L A T E S T R E P L I E S (Newest First) |
| Enrico |
Posted - May 08 2012 : 20:36:53
The potential matrix in FastCap2 is calculated and stored in compressed format, using the multipole approximation. This is the 'trick' used by FastCap to solve the system in O(n) operations (for each conductor), where n is the number of panels.
So if you need the whole potential matrix, you loose the benefit of the fast solve. There are debug functions in FastCap2 that could dump this matrix, but the matrix is huge, and so useless for medium to large scale simulations. The same goes for FasterCap.
Anyway, to use retarded potentials, you should also know the relative positions in space of the panels; and you should use a circuit solver tool able to handle retardation (I know some papers suggesting using Spice TX line models; but this is feasible only for small models). The 'real' solution would be to use some technique of model order reduction (MOR), but here we are outside the scope of FastCap2 / FasterCap.
Best Regards,
Enrico
|
| Lavdim |
Posted - May 07 2012 : 22:12:11
Hi!
As I read, last phase of FastCap2 algorithm is inversion of "cofficients of potential matrix", and mentioned as time-consuming. That matrix is used in PEEC modeling. In quasi-static model Capacitance Matrix can be used, but for retarded problems, retarded potetials are used insted. So, I was wondering if its possible to get that matrix before doing last inversion.
Best Regards
Lavdim |
| chromatik |
Posted - May 07 2012 : 17:17:08
Hi Lavdim,
By potential matrix, you mean the differents voltages applied to each conductor? The method is basically applying 1V to a conductor, and grounding the rest, then switching the conductors and so on...
As FastCap is based on quasi-static approximations, I am quite sure it wouldn't deal with time retardation problems.
Regards |
|
|
