

May 1, 2024 at 6:32 pmeugenio_digaetanoSubscriber
Hello,
I am using MODE with FDE to calculate the mode at device facet and generate how the mode would be into freespace propagation after a certain distance using the "Farfield Setting" option with "Planar" as projection method.
However, the results as function of projection distance are quite weird as the mode appears expander for short propagation distance (even for 0 um), it would have a size narrowing up to a distance (similar to a waist of a Gaussian beam), and then expand again. This doesn't match the propagation done using the farfield with angular projection method.
Is this expected for planar method? Is this method giving accurate results only above a certain propagation distance?

May 1, 2024 at 8:58 pmGuilin SunAnsys Employee
I would not use FDE to simulate the facet as it has less control. It is better to use FDTD. Please refer to this example: https://optics.ansys.com/hc/enus/articles/360042304894Facetreflectionandtransmissioncalculations
It is physical that as long as the mode leaves the waveguide, it spreads since there is no confinement anymore. I do not know if you have any theory to support
" it would have a size narrowing up to a distance (similar to a waist of a Gaussian beam), and then expand again."Â I did a test using FDTD and the fields spreads after leavning the waveguide. I am not sure if there is any mechanism for the narrowing. Here is my result for a ridge waveguide:
Fig. 1: farfield from the monitor immediate the facet
Fig.2: far field from the monitor 3 um away from the facet.
FIg.3 The waveguide mode profile
Fig.4 The field in the plane, where mode source is located at z=00.3um, and the facet it at 0um.
I did not use farfieldexact to project fields at specified distance. Please use Â farfieldexact3d  Script command for your application

May 2, 2024 at 12:32 pmeugenio_digaetanoSubscriber
Hello Guilin,
Thanks for your answer but I probably was not really clear in my explanation. I completely understand that FDTD is better for facet simulation but quite resourceconsuming hence I used FDE mode overlap for a quick verification and the planar projection to simulation a propagation in free space in a certain distance.
Your comment "It is physical that as long as the mode leaves the waveguide, it spreads since there is no confinement anymore." is exactly what I expected but Farfield projection in Planar method gives those weird results, see the mode diameter for different projection distances.
It seems to be a waist around 20 um distance before the expansion of the more that I would expect.
My question is: Is this behaviour normal for the Planar Farfield projection function in MODE? Maybe this method is unreliable for short projection distances, I cannot find a guide on this in Lumerical website.

May 2, 2024 at 3:22 pmGuilin SunAnsys Employee
Â
In the settings https://optics.ansys.com/hc/enus/articles/360034917353FDEsolveranalysisModalAnalysisTabÂ it did not mention what method it is used for plannar projection. It might be the regular farfield projection with scaling distance, which is not accurate for small distance. Consideting its fast calculation, it might not use farfieldexact, which requires raw data in the uniform material: https://optics.ansys.com/hc/enus/articles/360034914713FarfieldprojectionsinFDTDoverview
So there is a price to pay when you try to simplify the calculation.Â The correct method is to use FDTD:
https://optics.ansys.com/hc/enus/articles/360034914833Adjustingtheprojectiondistanceinfarfieldprojections
Â

May 2, 2024 at 6:26 pmeugenio_digaetanoSubscriber
Great, thanks for the suggestion. If so, I would try to avoid the planar projection if so as I need to simulate the mode expansion between 1 and 10 um from the facet.
I tried your suggested example "Facetreflectionandtransmissioncalculations" and works well for what I need. My aim is to calculate the power coupling between two waveguides at a specific distance from each other. The example is really useful to calculate the mode distribution of one of the waveguides after a distance but how can I check its power coupling with the other one? I could use a full 3DFDTD simulation with mode expansion monitors but I would prefer to use an overlap analysis as quick test before a long simulation as final one. I noted that FDTD has the overlap function between two monitor modes, is there a function to find the best position for coupling such as "optimizeposition"?
Alternatively, I was wondering whether I can export the mode from a monitor in zbf format and load it in the dcard of the FDE overlap analysis tab. It seems possible but does not recognise the overlap.

May 2, 2024 at 6:54 pmGuilin SunAnsys Employee
If you want to use optimization, you will need to get the overlap calculated in an analysis group, andÂ output the overlap. In optimization you can choose the output as FOM.
Set up the monitors in one analysis group. However if it is from two different simulation files, you can save the mode data into ldf file or matlab file, and load it into the analysis group, then calculate the overlap with scripting. This is more flexible.
Please refer to the overlap calculation: https://optics.ansys.com/hc/enus/articles/360034405254overlapScriptcommand to get the formula. In such case the output is only the ovderlap, not the power coupling.
Ideal method is to do two simulations if the gap is large with the farfield projection as the import source. If the gap is small you can place the two waveguides into one simulation file.
Â

May 2, 2024 at 7:13 pmeugenio_digaetanoSubscriber
Thanks Guilin, this is exactly what I meant to do, I was wondering how to optimise the two modes position in FDTD but I found an easy solution. I just simulate the mode profile at different distances from the facet in FDTD using several monitors and export the results with savedcard function. I then import it in MODE and use the standard Overlap Analysis either manually and by script with the mode at facet from the second waveguide.Â

May 2, 2024 at 9:31 pmGuilin SunAnsys Employee
Â yes, then you will need two products.

May 2, 2024 at 10:20 pmeugenio_digaetanoSubscriber
This method is working fine but I have a couple of doubts: 1) is the overlap function always using normalsed version of the mode? In case of two mode overlap from MODE the modes both are normalised with field maximum being 1 whereas, if I import the second mode as dcard from the FDTD monitor, this second mode is not normalised to have filed maximum being 1 but I seems function is still giving reasonable results. 2) I tried to free space waveguide1 mode from facet in FDTD and overlap with waveguide2 facet in FDE (as I stated in the previous post), I replicated the same swapping the two waveguides and the results are slightly different, though I would expect the process to be symmetrical. Is this expected for some other reason?
Â

May 3, 2024 at 1:07 amGuilin SunAnsys Employee
A1: no requirement for normalization;
A2: there are interpolation errors and overlap is approximation for coupling! the best way to do coupling is using FDTD and check the S parameter result.Â

May 3, 2024 at 10:27 ameugenio_digaetanoSubscriber
Thanks again for the suggestions. My first question was because the overlap was giving weird (negative!) coupling if I use a monitor where mode is propagating backward for example. For the second question, I understand that the FDTD of the full system with S parameter is the best possible simulation but I need to simulate several configurations (probably hundreds) and each simulation is 23 hours. Therefore, I prefer to use the full FDTD only to confirm the results, overlap analysis would be a good compromise to guide me towards the best configuration.

May 3, 2024 at 2:03 pmGuilin SunAnsys Employee
Negative sign is ok, as you can see from the definition https://optics.ansys.com/hc/enus/articles/360034405254overlapScriptcommand it uses absolte value.

 You must be logged in to reply to this topic.

461

225

200

177

162
Â© 2024 Copyright ANSYS, Inc. All rights reserved.