| Summary: | 碩士 === 國立臺灣科技大學 === 電子工程系 === 105 === DWDM (Dense Wavelength Division Multiplexing) owns a promising performance in coupling many numbers of input signals with the certain designated wavelength (multiplexing) and splitting them again into several outputs (de-multiplexing). The dense channel spacing of DWDM is capable of overcoming the massive demand of data capacity needs nowadays. However, this easily causes high adjacent channel crosstalk. It is believed that the flat transmission passband of each channels in DWDM device would relax the need of precise equipment and polarization performance. Thus, the flat passband would effectively reduce the operational cost. In consideration of its critical functions, this thesis will focus on how to develop 1x5 DWDM with low insertion loss, low crosstalk, and flat passband transmission. All simulations were run through Photon Design commercial software, EPIPPROP collaborated with FIMMPROP.
At beginning, the study was conducted to optimize the parameters that have significant impact in reducing the adjacent channel crosstalk. Various output waveguide position along the Rowland circle was tuned to observe its effect in crosstalk reduction. The results showed that putting the waveguides at the higher angle position of Rowland circle would effectively reduce the channel linewidth and the adjacent channel crosstalk as well. Then further simulation showed that using the narrow waveguide width and chirped grating also offer a significant noise floor reduction.
Several techniques have been proposed to obtain the flat passband transmission with low insertion loss and low adjacent channel crosstalk. In general, one can modify the waveguide geometry or tune the grating focal point to flatten the passband transmission. In this thesis, a new approach of passband flattening have been simulated by intentionally tilting the grating facet more to the certain degree. By such, the focal point of diffracted light would be shifted and result in a flat passband once it is convoluted by the output waveguide mode. However, this technique still suffer from high insertion loss. This proposed approach performance was also compared to the MMI coupler method that has been widely utilized in the passband flattening and demonstrated a better channel isolation (± 39 dB) by using narrow width on the input waveguide, which cannot be implemented in the MMI coupler method.
|
|---|
