雜訊相依模型與模組設計最佳化運用在離散時間積分三角類比數位轉換器

Abstract

傳統高階積分三角調變器的設計上，主要是依賴行為模擬的方法。然而此方法相當耗時。本論文是第一個提出使用模組化設計的方法去設計積分三角調變器。在速度上，使用模組化設計的方法將會比使用行為模擬的方法快上萬倍。由於先前非理想雜訊以及失真模型的不完備，模組化設計的方法一直無法真正的去實現。如今，由於積分器充放電雜訊模型以及放大器非線性直流增益諧波失真模型的推出，使得模組化設計方法得以實現。然而，使用模組化設計方法將會遭遇到雜訊相依的問題；本論文將會提出雜訊相依模型以便解決這個問題。除此之外，本論文也同時提出了積分三角調變器最佳化設計流程。對照於一篇積分三角調變器設計實例，模組化積分三角調變器最佳化設計將可以使用更短的時間去達到更高的訊號對雜訊以及失真比，同時降低積分三角調變器的功率消耗。

The conventional high-level SDM synthesis is mainly based on behavior simulation which is very time-consuming. This thesis is the first one in the literature to propose model-based SDM synthesis. Model-based method can be at the order of ten thousand times faster than simulation-based method, but it is never realized before due to the incompleteness of non-ideality models. The recent establishment of settling noise model [18] and OTA distortion model [17] facilitates model-based SDM designs. Nonetheless, new problem associated with model-based method arises, notably correlation between noises. Noise correlation models here are derived. In addition, a SDM design optimization scheme is proposed, which incorporates a comprehensive power consumption model. This model-based optimization is tested against a published SDM design, achieving higher SNDR and lower power results in a much shorter design time.