Star Couplers for AWGs - Why an odd number of outputs?


#1

Usually AWGs are for 2,4,8,32,64…etc. channels (sometimes 40 or 80 as well), so why are there an odd number of outputs for the “AWGs and Star Couplers” example in Lumerical?

For the options in the output_star.lms geometry, it says there are 8 outputs, but I count 17 output waveguides after the second free propagation region. What is going on here? I would imagine the number of channels supported by the AWG would equal the number of output waveguides.

Thanks,
Christian


#2

Hi @cgbottenfield
Generally AWG devices serve as multiplexers, demultiplexers, filters, and add-drop
devices in optical WDM and DWDM applications. Figure 1 shows a schematic
representation of the MxN AWG. The device consists of two concave slab waveguide
star couplers (or free propagation zones/ranges, FSZ), connected by a dispersive
waveguide array with the equal length difference between adjacent array waveguides.

Generally, there are two kinds of AWG: 1xN (M=1) and NxN (M=N). The number of the wavelength channels N is selected with the exponent of 2 such as 16, 32, 64, and 128.

The number of the array waveguides P is not a dominant parameter in the AWG design
because the Δλ and N do not depend on it. Generally, P is selected so that the number of
array waveguides is sufficient to make the numerical aperture (NA), in which they form a
greater number than the input/output waveguides, such that almost all the light diffracted
into the free space region is collected by the array aperture. As a general rule, this number
should be bigger than four times the number of wavelength channels.
For example, you can have an AWG with M=1 (number of input ports), N=16 (number of output ports), P = 60 (number of array waveguides).

Best Regards
Konstantinos


#3

Hello every one

Please put some more light on that,
For the options in the output_star.lms geometry, it says there are 8 outputs, but I count 17 output waveguides after the second free propagation region. What is going on here? I would imagine the number of channels supported by the AWG would equal the number of output waveguides.thanks.


#4

Hi @ashar
Could you please provide me with the link which refers that there are 8 outputs ?
Nevertheless, i will try to explain it. I found a link in Lumerical’s knownledge which demonstrates the modeling procedures of the optical AWG based on various components https://kb.lumerical.com/en/index.html?pic_circuits_awg-circuit-design.html . In this example there is a design of a AWG circuit design. As you can see there is an input star coupler with 41 inputs but only 21 of them have input signals. And there are 41 outputs. The output star coupler has 41 input signals but only 8 outputs are measured from the network analyser. From a number of 41 outputs, only 8 are measured. Please check this example.

Best regards
Konstantinos


#5

Hello @konslekk
Thank you so much for your reply.Please check “input_star.lms” or “output_star.lms” files provided in this link https://kb.lumerical.com/en/pic_passive_waveguide_couplers_awgs_and_star.html. you can see in the edit section of “input_star.lms” that the number of input/output ports are defined as 5 and 8 respectively but in practical there are 11 input ports and 17 output ports.Please explain.


#6

Can anyone Please check “input_star.lms” or “output_star.lms” files provided in this link https://kb.lumerical.com/en/pic_passive_waveguide_couplers_awgs_and_star.html. you can see in the edit section of “input_star.lms” that the number of input/output ports are defined as 5 and 8 respectively but in practical there are 11 input ports and 17 output ports.Please explain.


#7

Yes, this is the confusion I still have regarding this model…still haven’t found an answer to this…


#8

Dear @cgbottenfield and @ashar

Your comments are valid and I will take a better look at this in the coming days to see why we set the script in a way that it always creates an odd number of channels. But here is the reason why this happens:

If you go to the script, you will see that software takes 2 * num_inputs + 1 to create the channels:

You can edit the script to be:

theta_input_array = linspace(-theta_input*num_input/2,theta_input*num_input/2,num_input);

This should solve the current problem (and will work if only num_input is an even number), but again I will need to check with the developer to make sure that I am not missing any other considerations.

Thanks


#9

Thank you so much.Please help me to understand one more thing. what is 300 e-15 in the script file “set_source_star.ls” in this link fhttps://kb.lumerical.com/en/pic_passive_waveguide_couplers_awgs_and_star.html#introduction

As we know then what is 300e-15.?


#10

Dear @ashar

I think, as you mentioned, what is important is the relative phase between each arm and the results should not change by changing the value of 300e-15.

I also checked with the application developer about setting star coupler. It looks like that example was for demonstration purposes and what we did before (changing the script) should not cause any additional problem.

Thanks