Single Frequency Injection Instability Problem



Hello Everyone,

I’m trying to work on an application where the source needs to be a continuous wave mode source (i.e. single valued frequency mode source), however, the transmission looks very weird and a lot of instability is observed.

Is there something with these conditions?

N.B. When I make the source a pulse in the time domain (broadband source), the transmission looks fine. That’s why I think the problem is from the source and not from the structure given that the structure is a straight silicon waveguide.

short_optical_isolator 3.fsp (313.3 KB)


Hi @ahmad.ayoub
You want to consider a 600x220[nm] silicon waveguide in a single valued frequency. In the most cases, the mode source is set in frequency/wavelength. Then, in wavelength start and stop you can write the single frequency. Also, i would like to show you something else. Select FDTD and click write, then “edit object”. Go to the tab “advanced options” and check the button “set simulation bandwidth”. Then, you can type the frequency or the wavelength that you want.

In “min” and “max” type the same frequency(not 0.4 and 0.7).
Also, i would recommend you to use the 3D-FDTD and not the 2D.
But , i have a question. Why does exist the monitor_1 ? I think that “monitor_1” is not useful. Additionally, the “monitor” should be two dimensional , i mean that the waveguide should be enclosed by the “monitor”.

I hope that it could help.


I’m doing 2D simulations, that’s why the Monitor is a linear Y type. I’m asking about the transmission. It diverges with me. Why the transmission reaches this large values?


Hi @ahmad.ayoub
I would like to ask you something , to make it clear. Do you want to extract only the transmission or the S parameters of the waveguide ? For example , if you want to find the S parameters you have to add more objects.


I want to calculate the transmission only.


Hi @ahmad.ayoub,
Sorry for your overlooked question since it is marked as solved. It is actually never accurate to check the transmission for CW signal due to numerical issues. This is why Sources in FDTD Solutions always inject a pulse (i.e. broadband), even when the user is only interested in collecting data at a single frequency.

But if you are interested in a real CW excitation, perhaps for non-linear applications. I suggest starting with this example:

Hope this is helpful. Any further discussion on this issue is most welcome. :slight_smile:


Thanks it’s almost fixed.