Why the injecteed power from an input port of a two port network is not "1"?

arefin21arefin21 Member Posts: 2

I am trying to simulate a two-port network across a waveguide. The input port is injecting power. In principle the injected power should be equal to 1 and will decay across its way.

However, if I plot the power profile of the input port it shows only 0.3 and after a certain distance of around 50 µm of travel it becomes 0.1. Why the power "P" profile does not show me 1?


  • gsungsun Posts: 954Ansys Employee
    edited September 7

    I am not sure what is the meaning " power profile" here. Do you mean you plot the source power as a function of frequency? if so, what is the peak value at the center frequency? Please post the screenshot here so I can have a better understanding in order to help resolving this.

  • arefin21arefin21 Posts: 7Member

    Hi Gsun, yes exactly. Thanks for your reply. Thats what I meant. The power out from the input port should be 1. That is my understanding is. I am attaching the screenshot of the power profile of input port.

  • gsungsun Posts: 954Ansys Employee

    I guess you mean the Poyinting vector:

    It is not power, but power density. Its integration in space is the absolute power of the source. In FDTD, the source power is very low, as by default, we set the E field max as 1V/m. The source power is in the order of -16 (10^(-16) Watts.

    However, the transmission and reflection are normalized to source power: transmission - Script command

  • arefin21arefin21 Posts: 7Member

    Hi gsun,

    Again thanks for the reply. My query is that how do I confirm the normalized power injected by the source is equal to 1. I plotted the S and T value which does not add up to 1.



  • gsungsun Posts: 954Ansys Employee

    Please try a short straight waveguide and see the transmission.

  • arefin21arefin21 Posts: 7Member

    Hi Gsun,

    Agian thanks for your reply. I simulated with a 10 µm waveguide. With and without grating. Still my transmission value is not 1. My question is from the input source the amount of normalized power should be 1. But why does it have modulation and not sums to 1. I am attaching two photos with and wiothout grtatings. I am okay to share my file. But cannot upload here. Can you have a solution?


  • gsungsun Posts: 954Ansys Employee

    Are you simulating a small cavity and the ports are positioned on the cavity end faces?

    It seems not correct device configuration. Usually the ports are positioned on the input and output waveguides, with gratings in between. However I did not find them.

    From your results, the green line has positive and negative values. This indicates either you are not simulate long enough (probably due to artificial cavity) or you are plotting real(s), not abs(s)^2. Please check your device and modify the settings accordingly.

    On the other hand, the normalization does not guarantee R+T==1, due to 1) numerical error 2) fields not confined well in waveguide and other losses.

  • arefin21arefin21 Posts: 7Member


    As you told to simulate a small waveguide and see, thats what I did. I have simualated keeping the input and output port exactly at the input and output of the waveguide. Right now, I have run 2 different simulations. One with 10 µm waveguide length and another with 50 µm waveguide length. I am attaching the photo of both. However, I am wondering in the input port the transmission value should be 1. It makes sense that in the output port the value will drop due to absorption and propagation loss. But why is it the case for input port. Also, I am not sure what is the source of the oscillation in the transmission curve of both spectrum?

  • gsungsun Posts: 954Ansys Employee

    Sorry, There is mis-communication. When I say "short waveguide", I mean the waveguide is enclosed and truncated by PML. Please significantly increase the waveguide length to extend it outside of PML. Otherwise you are simulating a waveguide cavity with air interfaces. From this

    you can clearly see the cavity effect!

    Please use the 50um long waveguide, but set the FDTD span smaller in the propagation direction to be 10um or 20um. then do the test. Make sure the waveguide ends outside of PML.

  • arefin21arefin21 Posts: 7Member

    Now it makes sense!

    Thank you so much. Only question remains is why there is an oscillation in the spectrum and its not flat.

  • gsungsun Posts: 954Ansys Employee

    Since it is grating, it ay need longer simulation time, and smaller autoshutoff min. To get better result, you will need to use override mesh to well resolve the grating teeth, and make sure each teeth and gap to have integer number of cells.

    In addition, it is better to have no grating after the ports. Ports should be in I/O waveguides.

Sign In or Register to comment.