the SWGC design as the paper


dear all
recently I have learnt the simulation of the uniform SWGC just as the picture show below

I have built the same 2D FDTD model, setting the relative parameter as same as the author of the paper optimized (just as show below) and wanted to reproduce the same results as the paper’s.

however the results of my simulation is much different from the results of the paper. I dont know what cause so big difference.Can you help me check my file? Thanks a lot.

2015 Apodized Focusing Fully Etched Subwavelength Grating Couplers.pdf (1.9 MB)
SWGC_TE.fsp (563.8 KB)


Hi, I have downloaded the file and took a quick look at it. Once obvious thing is that you had the offset the other way around. You should offset the source (and the monitor) such that it has a cross section with the fiber. In the current setting, you are injecting an unphysical mode to the device. You can take a look at this example to see how the offset should be properly done. Please give it a try and let us know how it goes.


dear kchow
Thanks for your reply.But in fact from the example you mentioned,I can’t get any information about the offset modifying.I am sorry that I don’t know how to do the offset.Can you give me some more guidance about it.furthermore I want to ask another question abou the GC example.When the GC structure is used as an input grating,what kind of source should be used? Is the mode source in the fiber or the Gauss source more accurate which we should use as the input source when we do the GC simulation as input grating?. Hope to get your feedback. Thanks a lot in advance.


dear all
Can anybody help me solve the question above? Thanks in advance.


Dear all
Now I have modifed the 2D FDTD file and the source (and the monitor) has a cross section with the fiber.However the transmission I get from monitor(“T”) is still very low,as shown below.

Can anyone help me solve the question? Thanks a lot in advance.SWGC_TE.fsp (566.1 KB)


It will depend on your actual setup. If you are using the fiber for input, then you should use the Mode Source since you will need the Mode Source to calculate (and inject) the fiber eigenmode. On the other hand, if you have it free space coupling, then you can consider using the gaussian source.

It looks like you have already figured out the offset. I will take another look at your file very soon. Do you expect the coupled waveguide mode to mainly propagate to the -x or +x direction?


dear kchow
Thanks for your reply.As the FDTD file shows,I hope the coupled waveguide mode to mainly propagate to the -x direction and I can record the transmission of monitor(“T”). Hope to get your feedback.Thanks in advance.


dear kchow
Do you have any idea about the question? When I directly use the gauss source instead of the mode source of the fiber,it work well and I can get the right transmission of monitor(“T”). But if I use the mode source of the fiber,the simulation results is still wrong,just as I show in previous post.Until now I can’t still find out what is wrong with the mode source.Hope to get your feedback.Thanks a lot


I took at quick look at the file again. I think it might make sense to step back a little bit to see if the grating coupler can emit light like shown in fig 2b), at -31 degree (TE injection). I did a quick test on it and it looks like it out-couples light at -21 degree. You may want to start from there and figure out what exactly has caused this, I think it could be the structure parameters, mesh, or other reasons that can cause this. SWGC_TE (2).fsp (574.6 KB)

Once you have the angle correct, then you can try to see what the (roughly) the best position of the fiber should sit in. The near field data from the “above” monitor should be able to tell you a rough estimate. Then you can follow what is discussed in this KB example to sweep the fiber position and optimize the simulation.


dear kchow
Sorry for replying late.In fact I can get the right result by just using gauss source and I also set the tilt angel at -31 degree, just as the same of the mode source in the fiber. So I think it is not the tilt angel setting that cause this problem.The result is shown below. What confused me is that why I can’t get the same result by using mode source in the fiber.

Do you have any idea and advice about it? Hope to get your feedback.Thanks a lot.


Gaussian beam can be diverging, so it may have a range of angles for the fields when it hits the grating. Although the mode exiting the fiber may have similar behavior but it is hard to justify that they have exact same effect when light lands on the grating.

In my previous reply, it is clear that light is emitting at about -21 degree. Due to reversibility of light, it means that the optimum injection angle of the fiber is also -21. Obviously, it did not match what is reported.

Before you use a fiber to couple light at -31 degree (in-coupling), you should make sure that light can be injected at -31 degree (out-coupled).