Discussion Forum

If, by "project a 3D radiation hemisphere pattern to a uv 2D plane," you mean that you want to create traditional 2D polar radiation pattern plots, you can certainly do that with the built-in post-processing plotting features! Here's how. In the Model Builder, right-click Results. Choose Polar Plot Group. Right-click Polar Plot Group and choose More Plots --> Radiation Pattern. Now, you'll have to play with the various settings to get the plot you want. Be sure to put in the right expression to be plotted (same as you would use for a 3D pattern plot). Under Evaluation, be sure to set the Normal Vector and Reference direction suitably. For the Normal vector, define a direction that is normal (i.e., perpendicular) to the plane of the polar-cut plot that you want! (You can also flip the plot around, by changing the sign of that unit vector.) For the Reference direction (the "zero" direction of the polar plot), experiment with various unit vector choices until you understand what is happening. You'll learn fastest by doing.

Hi Robert, thanks for the post. No, what I mean is projection of the hemisphere of the unit sphere onto a circle, so that we can easily see the beam pattern for the whole hemisphere without having to limit ourselves to a slice. The disadvantage is that the plot suffers from distortion on the edge of course. Please check this paper Section 4 gives some simple expressions to do this transformation.

I've found that the 'regular grid' in COMSOL takes the grid from the model space itself when exporting normdBEfar which is a bit confusing as the far field shouldn't depend on the model size or distance from the origin. I guess this must be an internal thing in how COMSOL works. I've extracted some data based on 'grid' with x,y,z swept from -1 to 1 and get the vector directions to the unit sphere and it seems OK, but still struggling with the coordinate transforms a bit.

I guess one way would be to create a sphere in the model then plot normdBEfar on the surface of that sphere and position the camera to make the projection but its a bit hacky.

Ah. In the years prior to Comsol providing far-field pattern plots in post-processing (and before that, computing any far-field quantities at all), if you wanted to compute and plot something like that, you could place a sphere around the origin and create/write a set of your own equations, to compute and plot the quanties of interest on the surface of that sphere. I used to do that, long ago. I also remember looking at spheres along the axis, and making plots (on the screen) just like the projection you mentioned. I presume that same technique still works. I've attached an example image. Alternatively, if you want to actually make such plots on a disk (instead of looking at a sphere from a particular perspective) you ought to be able to do that via judicous use of coupling variables. I suggest you investigate along that path.