Perform a tidy3d grating coupler simulation from a GDFactory component

[joamatab]

It would be great to be able to perform a tidy3d grating coupler simulation from a GDFactory component

gdsfactory/gplugins#539

[momchil-flex]

Could you provide the code to test with?

[joamatab]

Hi Momchil,

yes, you can test it with with this notebook for +5 deg positive angle
https://gdsfactory.github.io/gplugins/notebooks/tidy3d_00_tidy3d.html#get-simulation-grating-coupler

or with this UBC notebook for the negative angle -31 degrees
https://gdsfactory.github.io/ubc/notebooks/11_sparameters_gratings.html

in both cases the center wavelength should be 1.55um

[joamatab]

any updates on this?

[momchil-flex]

Sorry - fell through the cracks as no one was assigned! We'll take a look.

[joamatab]

Any updates on this?

[tomflexcompute]

@bzhangflex could give it a try?

[bzhangflex]

Will have a look today, sorry for the long wait

[joamatab]

Any updates on this ?

[bzhangflex]

Hi Joaquin, I tested the function you provided and found some issues. I left a couple of comments in your PR gdsfactory/gplugins#539 that will hopefully solve some of the issues.

Some additional comments

• The default grid resolution (10 steps per wl) is a bit low, I would recommend using at least 20 steps per wl, ideally 30, although I'd warn that it will greatly increase the cost of 3d sims
• The default source bandwidth is very large, and the simulation outputs warnings that some of the material models are not defined across the entire source bandwidth. I would recommend reducing it to the center wavelength +/- 50 nm
• The default waveguide monitor position is too close to the left PML, I'd recommend moving it so it's at least half a wavelength away
• Coupling loss will be highly dependent on the gaussian source position (see figure below), just something to note if you see lower than expected loss