ANNOUNCEMENT - The community will be down for maintenace this Thursday August 13 from 12:00 AM PT to 02:00 AM PT. As a precaution save your work.
cancel
Showing results for 
Search instead for 
Did you mean: 
cancel

WDM Devices — AWG with Flat Response

467
Views
0
Helpful
0
Comments
Beginner

Why is Flat Response required? How to obtain a flat response for AWGs? The principles of different proposals and the loss penalties of each proposal will be discussed in this article. 

 

Why Is Flat Response Required?
In the all optical network (AON), the optical signals passed tens of nodes before reaching the destination node, as shown in Fig.1. The ROADM nodes are usually composed of wavelength selective switches (WSS), multiplexers/demultiplexers and optical switches. The wavelength multiplexers/demultiplexers are optical filters, including TFF-based WDM devices, arrayed waveguide gratings (AWG) and optical interleavers.

图片1.jpg

The spectral response of an optical filter without optimization is usually Gaussian-type, as shown in the left of Fig.2. The response of cascaded filters becomes narrower when the optical signal passes multiple nodes. As we know, wavelength jittering is inevitable for laser diode (LD). Loss increases rapidly when the wavelength of LD deviates from the ITU-grid, which is intolerable for the optical communication system. The spectral response of an optical filter after optimization is shown in the right of Fig.2. The response of cascaded filters is still flat. The wavelength jittering of LD doesn’t introduce much loss.

图片2.jpg

The spectral response of the TFF-based WDM devices is usually block-like due to the characteristics of TFF filters. However, the response of an AWG before optimization is usually Gaussian-type, which doesn’t meet the requirement by optical communication system.

 

Cause for Gaussian Passband
As we know, a standard AWG includes an input waveguide, an input star coupler, hundreds of arrayed waveguides, an output star coupler and tens of output waveguides. The wavelengths are dispersed in the output star coupler as shown in Fig.3. Different channels (λ1, λ2, λ3, …) are focused at the endface of different output waveguides, as shown in Fig.3(a). Then we enlarge one of the channels (λ3) at the corresponding output, as shown in Fig.3(b). We can see that different wavelengths (λ31, λ32, λ33) of the channel are focused at different positions. λ32 is focused at the center of the output and the best coupling ratio is obtained. λ31 and λ33 are focused at the edge of the output and experience the highest power loss. That’s to say, when we enlarge one of the DWDM channels, the edge wavelengths experience more loss than the central wavelengths. Thus the spectral response of an AWG before optimization is usually Gaussian-type.

 

....

 

For more please refer to HYC blog.

 

 

This widget could not be displayed.