應用雜訊轉換矩陣設計一雜訊最佳化之混和架構寬頻低雜訊放大器

Abstract

現今之科技逐漸朝著互聯網(IoT)發展，將生活中各種用品智慧化儼然成為一種趨勢，而各種智慧產品則需要靠無線通訊技術將之連結在一起。例如發展了約十年的智慧型手機，以及智慧型手錶、智慧型眼鏡等等，甚至有許多家電產品也已經智慧化，並有智慧型管家來語音控管所有產品。在這波互聯網熱潮中，無線通訊技術的發展顯然越來越重要。 在通訊技術中，所有工程人員無不在解決改善的就是雜訊問題。而影響雜訊品質好壞最大的部分，就是射頻電路(radio frequency)最前端之低雜訊放大器(Low-noise Amplifier)，本篇論文主要在研究 LNA 之雜訊，除了一般的研究方法之外，我們使用了一個新的雜訊計算方法，雜訊轉換矩陣(Noise Transformation Matrix)，來研究在如何讓 LNA 之雜訊指數(Noise figure)最佳化。並且將此理論套用在 Shunt-Shunt feedback 架構之 LNA 設計上，此種架構的特點在於能達成阻抗匹配的頻率範圍較寬，能應用於寬頻應用的設計上。我們將設計一簡單的電路，並以晶片量測結果來驗證我們推導出來的設計理論。電路使用 pHEMT 0.15um 的製程技術來實現。最後根據此理論，我們進階設計了結合 Shunt-Shunt feedback 架構以及 Source-degenerated inductor 架構之LNA，大幅增加操作頻寬。電路使用 UMC 0.18um 的製程技術來實現。

Nowadays, the trend of technology has gradually been leaning toward the Internet of things (IoT). Each portable thing will be inevitably connected to internet via wireless communication. In addition to the best-known smart phones, smart watches and smart glasses are also well-developed, even are many appliances, have been intelligent. All smart appliances are now controlled by smart housekeepers, like Google home and Amazon Echo. In this intellectual trend, the development of wireless communication technology is obviously more and more important. In communications technology, engineering encounters noise problems. The noise is mainly affected by Low Noise Amplifier (LNA), which is the front-end of radio frequency (RF) circuits, thereby this paper will focus on the noise of LNA design. In addition to the general method, this paper uses a new noise calculation Method-the Noise Transformation Matrix-to study how to optimize the noise figure (NF) of the LNA. Then we apply this method to the Shunt-Shunt feedback architecture LNA. The characteristic of this architecture is wide range of frequency for impedance matching, and it can be applied to broadband LNA design. We will design a simple circuit to verify the result which is calculated by Noise Transformation Matrix. The circuit was implemented by using pHEMT 0.15um (P15) process technology. Finally, based on this theory, we will move one more step further as enhancing the design of an LNA which combines the Shunt-Shunt feedback architecture and the Source-degenerated inductor architecture to increase operating bandwidth. The circuit was implemented by using UMC 0.18um (U18) process technology.