# Different transmission results for broadband vs single frequency source

dear all
recently I have simulated a 2D FDTD grating coupler .In the simulation file, I use the gauss source instead of the mode source in the fiber.But when I find a question about the transmission of the monitor(“T”).When I
set the center wavelength to 1.55um and set wavelength span to 0,the transmission of the T at 1.55um wavelength is 0.509 ,as shown below.

But when I set the center wavelength to 1.55um and set wavelength span to 0.1um,the transmission of the T at 1.55um wavelength is 0.47 ,as shown below.

why the transmission of monitor(“T”) at 1.55um is difference in the two conditions and how to make the transmission of monitor(“T”) at 1.55um to be the same in the two wavelength?
finally,which transmission is really true ,just coindicent with the measured value in fact in the two condition setting. 1.fsp (478.6 KB)

Dear @jbwei

I think the discrepancy between the two results is due to meshing.

Auto non-uniform meshing algorithm uses a few factors including source wavelength, material properties, and the structure geometry to mesh the structure (see Mesh setting in this link). For a mesh accuracy of two, for example, software uses 10 points per wavelength to mesh the structure. For the broadband pulse, meshing algorithm uses the smallest wavelength to mesh the structure.

Below I showed how meshing varies (for a 1*1 um^2 FDTD region and mesh accuracy of 2) based on the source wavelength in free space:

When results changes with mesh size (in your case due to change in source frequency), this simply means that your results are not reliable and require convergence testing. Thus, non of the results are precise unless you get to a point where results do not vary when you change the mesh by making them smaller.

I used a mesh accuracy of 8 and results of two cases are in good agreement:

You could also add mesh override region to precisely mesh the grating and structure to check if my results make sense.