Voltage and impedance postprocessing for wave ports

awslabs / palace

3D finite element solver for computational electromagnetics

https://awslabs.github.io/palace/dev

Apache License 2.0

237 stars 49 forks source link

Voltage and impedance postprocessing for wave ports #172

Open sebastiangrimberg opened 8 months ago

sebastiangrimberg commented 8 months ago

One way to extract the impedance for a wave port mode is:

Compute the mode voltage using two user specified points (p1, p2) which lie on the ground and conductor. Then, the voltage can be computed by a line integral from p1 to p2 of the electric field mode shape (this can be evaluated using numerical quadrature across elements, using GSLIB). This works as long as the two points can be connected by a straight line (a generalization allowing intermediate waypoints for the line integral should be straightforward).
The port modes are normalized to have unit power. Thus, Z = V^2 / P to estimate the mode impedance.

nico-arnold commented 2 months ago

Hey Sebastian,

any news or examples on this one? That would be a really great feature for the simulation of EMC problems in circuits 👍

best wishes, Nico

sebastiangrimberg commented 2 months ago

Hi Nico, unfortunately no progress here but we would very much welcome a PR working towards this and would be happy to help with it. The idea is not so complicated, and I can try to outline the steps here for an interested developer:

User interface: Provide two coordinates (p1, p2) between which the line integral E \cdot dl will give the port voltage
Sample the computed electric field mode on the port at a number of points between p1, p2 as given by the quadrature rule of a desired order, using GSLIB (see palace/fem/interpolator.hpp, in particular this new function).
Compute the line integral to get V, and impedance as Z = V^2 / P (where P is the port power, already computed in waveportoperator.cpp).

Sorry for the lack of progress but I hope that someone will be able to take up this feature soon!

nico-arnold commented 2 months ago

Hi Sebastian,

nice, thanks for the update + explanation! I will take a look into that. Still, I'm out of "real" C++ for a while (only did smaller embedded stuff recently) and didn't even manage to compile the package without using Spack, so it will definitely take a while for me to get there. Let's see (:

best wishes and have a great week, Nico