# DGTD sources

Member Posts: 1

I am simulating the interaction of a particle with light in DGTD to see the farfield light distribution. I notice that I can define surfaces with different shapes as a reference for the incident light, say rectangle, sphere or plane as shown below. But with different geometry for the source I got different results, even the original location of the plane wave and other parameters are the same. Can someone explain how the geometry is used for the incident light, and why I got different farfield projection?

• Posts: 2Member

I enclosed the DGTD file and the figures obtained by using a rectangular source (field, ID and index). And another question I would like to ask is:

In the field figure, there is a jump of intensity/phase on the interface between the source rectangle and the open space, especially in the upper part, which is not correct. How can I fix this?

• Posts: 583Ansys Employee

Sorry to reply you late as our team is busy in other projects. As you may know, Ansys staff are not allowed to download and open client's files. So I can only reply according to your description and the images you posted.

Regarding to "jump of intensity/phase on the interface between the source rectangle and the open space, especially in the upper part", It is correct, since inside TFSF it is the total field, and outside the TFSF, it is the scattering field. they are not continuous. Please refer this post: Ansys Insight: FDTD 初学者： 平面波和全场散射场tfsf光源有什么区别

As for the different result using different source surfaces, what result you are referring? how large the difference it is? due to discretization, small change of the result is due to the calculation error, which is normal. In theory, plane wave has plane wavefront. We can simplify the plane wave to be injected from a sphere surface, as we know analytically the phase difference at each point on the sphere surface, without simulation. By this way the simulation volume can be smaller than a rectangle. I would suggest that you check the mesh at different excitation method, and see if there is any difference.