Transmission Higher than One




After I finished the simulation for a ring of a quarter circle, I found that the transmission power is larger than one. This cannot be correct due to the losses that are caused because of the bending of the waveguide. Why does this occur and how can I correct this result?


Hi @g15081147

I did a couple of modifications into your simulation file:

  1. Used a 32 layer of standard profile for PML rather than custom PML. This seemed to be the most important factor.

  2. Increased the FDTD, source, and monitor span specially in the z-direction. This is to make sure that field is decaying around 5 orders of magnitude or more before reaching PML.

  3. Used multifrequency mode calculation for the mode source.

Below is the results with the modified simulation file:

testing_ring.fsp (333.2 KB)

Hope this addresses your inquiry.


The simulation works correctly now, but I did not understand either the reasons for increasing the number of layers nor using multifrequency mode source or when we choose to change these parameters?


Hi @g15081147

When you expect a reflection from PML, you need to change the PML settings or number of PML layers as part of convergence testing.

Once a light reaches into PML layer, it has to be completely absorbed:

And here is a good link that explains why you need to use multifrequency beam calculations for broadband simulations:


Thanks for sharing the links. They are helpful to understand some of the physics of the simulator. Nevertheless, I still have some problems when simulating simple structure like a straight waveguide. I tried to change the parameters as you did in with the ring, but still I get transmission power higher than one.

I have other questions as well. Is it advisable to use the multifrequency option with any structure? For example, when I tried to simulate a directional coupler without using the multifrequency option, the results were more reasonable.
Here you can see the transmission power from FDTD simulation without using the multifrequency option

While the transmission power from FDTD simulation using the multifrequency=1

As you can see from the second image, the value of transmission power decreases at higher wavelengths which is not reasonable. Why does this happen?


Hi @g15081147

Can you please share the public link of your files with us? Currently it is not public and it requires permission.

As is explained in our documentation, multifrequency option has to be enabled when the field profile changes significantly with wavelength. If the profile does not change, both simulations should return identical results. Simulations with multifrequency beam enabled are more trustworthy if the simulation is setup properly.


Hi @bkhanaliloo

I edited the sharing option and you can download and view the file now.

Can you please help me to understand the behavior of the directional coupler (DC) after enabling the multifrequency option? Why the transmission from the coupled arm of the DC decreases in the range of 1.56um to 1.59um then it increases again at 1.6um?


Hi @g15081147

Thank you for sharing the simulation file.

Regarding the results for the straight waveguide case: The transmission may exceed 1 at some weavelength but this is generally less than 1% error. I think this numerical error can be ignored. However, if you want to improve the results even further you can:

  1. Try convergence testing specially test different PML profile for a proper absorption:
  2. Use finer mesh over the waveguide region.
  3. Increasing the simulation time and decrease the auto shutoff level and make sure that simulation triggers the auto shutoff.
  4. Use multifrequency beam injection.

Regarding the directional coupler case:

It would be quite helpful if you could share the simulation files with me. Regardless, the transmission in directional coupler is a function of coupling coefficients and thus wavelength:

If the simulation is set properly, and you are seeing wavelength dependent results after you performed convergence testing, the explanation above may describe it.



Hi @bkhanaliloo,

Thank you for your detailed and helpful reply. I tried all the suggestions in your reply, but I still get transmission higher than one.

In the attached file, I am running s-parameters sweep to find the s-matrix, but I still get transmission higher than one
Strip_WG_Ports.fsp (1.1 MB)

Can you help me please to get a more accurate result of the transmission?


Hi @mohamedmansour

I reduced the FDTD x-span to lower the simulation time. I also increased the FDTD y- and z-span (and port objects) to make sure that waveguide mode is not truncated by the PML. I added a mesh override over the waveguide region to make sure that mesh cells match the edge of the objects. Below I provided the results as well as the simulation file for your review:


Strip_WG_Ports.fsp (1.6 MB)

Transmission goes above 1 at some wavelengths that is due to some numerical error that I expect to be resolved by using a finer mesh or increasing the number of frequency points in the Modal properties of port object.