Learning Forum

[sme8]

Hey,

I use an embedded high-Q analyzer in the FDTD module. Essentially, I use ring-down to estimate the quality factor of ultra-high Q factors. However, I realized that the Q factor it calculates is highly dependent on the source settings. For example, I set all the parameters the same for two simulations and only modified the wavelength of the source. I found that if I set my source wavelength to 700-900 nm, the quality factor is different from another simulation with a wavelength of 740-760 nm. Assume my desired resonance occurs at 750 nm. The resonant wavelength does not change with different source wavelengths. I use an auto non-uniform mesh for my simulations.

I also modified the time domain properties of my source to have a very narrow band signal in the frequency domain (consequently wide in the time domain) near my resonant wavelength, and again, the Q factor changed dramatically.

What do you think? Which one is more accurate?

Thank you for your support and help

[Guilin Sun]

Hi Seyed, It is common that when the source spectral range changes the result will be different, mainly because the mesh and the material fitting is not the same. It converges if much higher mesh accuracy is used. However if the material fitting is different the resonance will still shift. To avoid this, please fix the wavelength range in material fitting.

In addition, you may also check the Q analysis group and result and see if it is reasonable, in particular when FFT method is used. If "findresonance" findresonances - Script command  is used, it might be better since it uses analytical method to get the resonant frequency and Q from a few cycles of signals containing the resonant information.