FDTD Simulation stops on its own,

fdtd

#1

HexSiRodsOnAirSlabFluxField.fsp (537.3 KB)

Hi ,

I have started to play with the FDTD to simulate the power spectrum above a H1 cavity of a photonic crystal, FDTD predicts required time of about 15 hours, yet it stops only a few minutes in. The memory requirements are within bounds of the local resources, so I am hoping someone more experienced will be able to aid me.

Thanks

Yameng


#2

Hi Yameng,

There are a few points:

  • Symmetry BC: The way it used is likely wrong. I think you would like the sym/anti-sym BCs on X or(and) Y instead. Please refer to this page for proper application of the BC.
  • Q analysis group: There are low and high groups at different location. I think you can choose one of them and stick with it. You could run a simulation to location the highest field location for the analysis group to find the highest Q.
  • Simulation bandwidth: I am not sure why this is checked in the FDTD region. If this is a linear simulation, I think you should stick with the source bandwidth.
  • There are some application examples that I think is worth looking at, particularly the getting started example. And an example for the Tang’s cavity.
  • I disabled the 3D monitor because it takes a lot of the memory. Instead, I would suggest adding some 2D monitors to show the profile you are interested in. Once you are confident with some initial simulation results, you can turn the 3D monitor back on with downsampling options.

HexSiRodsOnAirSlabFluxField (1).fsp (507.6 KB)


#3

Hi Kchow,

I followed your advice, the result I got did not make any sense, because the power spectrum was just a flat line, constant in wavelength, whereas the source is a dipole with Gaussian pulse.

I simplified the construction, with no Photonic Crystal, just a dipole source, emitting in air towards a 2Dmonitor that records the integrated power. The result still shows a line, instead of a Gaussian like the source.

What is going on?

Thanks
Yameng


#4

FDTD Solutions returns the normalized power spectrum when CWnorm is on. This is to return the CW response while it is true that a pulse is injected by the source. That said, the power injected by the dipole should not be a flat line but it should follow these analytical calculation for homogeneous environment.