Learning Forum

[sean]

Simulating the following silicon photonic circuit in Interconnect did not give me the results I expected and I don't know why. Any help would be appreciated:

Am new to silicon photonics so have probable missed something simple.  I was expecting the above to behave as a normal Mach Zehnder interferometer, as I varied the length of the fibre optic. 

I was expecting to see the transmission dip for each 180  degress of phase difference, as the  ΔL between the two paths changed.

However the transmission output remained very similar pretty much overlapping the original graph regardless of the length difference, see below

         

My aim here is to make an interferometer that can measure displacements of about a 1 foot i.e. ~30 cm.

Any insights would be appreciated

[Greg Baethge]

Hi Sean,

Thanks for your post. I suspect this is caused by the optical fiber model. I ran some quick test, using waveguides instead of fibers, the results were more what you would expect when changing the length difference. I have to check with my colleagues, it's not clear to me if the group index is defined. Since the FSR of the Mach-Zehnder interferometer directly depends on the group index, it could explain this result.

On a side node, the transmission given by the ONA is the complex transmission. To get the fraction of power, use abs(T)^2.

[sean]

Thanks Greg. 

Much appreciate you takling the time to help out.

You are right, with waveguides this works fine and. interestingly. if you put another MZI down stream of this that also works fine? So it does seem to be related to the built in fibre optics elements in Interconnect and like you say "the lack of change is FSR would be consistent with a missing group index in fibres"? 

I see no way to set the Group indexattribute on the built in fibres. Note I get the same result on all three built in fibre types.

So maybe this is just something that Lumnerical needs to update?

[Greg Baethge]

Hi Sean,

I had the chance to check this with my colleagues, this behaviour is expected: the fiber model was designed for long distance propagation (typically, on the order of km) where users don't necessarily care about the delay, but more about the dispersion, for example.

With that in mind, you can use a waveguide element instead, that will include this delay. You can either use a MODE waveguide element, or a Straight waveguide element. The tricky thing is to find the right parameters as they're usually not provided by the manufacturers!