2D-FDTD-Gaussian Pulse Propagation
Please, I need some help about setting a Gaussian pulse as source of a 2D FDTD grid (At the middle of the grid). I′m using the Yee model to define the discretization and the Taflove's equations for analysing of wave propagation - TMz mode . My problem is: Even considering a lossless medium, the wave starts with attenuation/distortion of its shape, since the begining of propagation.
Reading some comments on the web, I suppose that this problems surges because of the source definition, but this issue is still a little confuse to me. I believe that some adjusts of the electric or magnetic field values near of the source point are needed... but I don′t know how. I appreciate if someone can give any help about that. Thanks in advance for your attention.
What a coincidence, I'm currently doing the exact same thing.
If the wave isn't propagating properly, its most likely a cause of your discretization. Are you meeting the CFL condition?
Hi,
Thanks for the reply,
Trying to ensure stability, I am using the rule stated in Taflove, ie, how the grid has two dimensions, S = 1/(square(2)); "S" = Courant number.
If I understand your hint, I can keep the same "S", but should improve the refinement of the grid?
Have you managed to solve it then? We could change the codes in matlab to see the effects?
Thanks again for the information.
PS.: By the way, what you called the CFL is the Courant Factor?
Honestly I'm not familiar with the Courant number, but the C in CFL does stand for courant, so I'm guessing we're referring to the same thing, which is that in a 2d grid,
vΔt ≦ (((1/Δx)^2) + ((1/Δy)^2))^(-1/2)
I've attached what I just coded - there's a bit of reflection off of the 2nd order Mur ABC boundaries, but I'm sure propagation is correct.
FTDT_GaussPulse_EDA.pdf
if it is a 2D problem with open boundaries and a point in the middle, you would expect the shape of the signal change as propagates for sure. Since the energy of the wave is spreading out to every direction, thus the magnitude is changing for sure. Test your code with a parallel plate waveguide case with top and bottom set as PEC and side walls set as PMC, then give a surface excitation, the shape will not decay anymore in this case since its TEM wave now, and it is just like a 1D problem. You need to understand the physics of wave propagation better when judging your simulation results. If you still think the result is wrong, post a figure and lets take a look.
Enjoy coding.