以統計方法產生的觸控板感測器模型與最佳化

Abstract

為提供消費者更簡單、直覺使用介面的電子產品，感測器的使用越來越普遍， 觸控板就是最常見的一種。依據感測模式，我們可以把觸控板分成三類：電容式、 電阻式，以及即時影像捕捉式。本論文針對電容式觸控板分析，試圖提出一加速 設計流程的方法。在電容式觸控中，x 方向、y 方向各有數條感測器，在手指沒 有靠近的時候，感測器即有一基本電容值；一旦手指靠近感測器，電場的改變， 會使感測器的電容值變化，稱為感應電容值。處理器在收集不同位置感測器的感 應電容值後，可估計出手指所在的座標。 如何衡量觸控板的品質，有兩主要項目。第一，觸控板對手指有無的感應靈 敏度；第二，觸控板對手指位置的解析度。針對上述項目，我們能藉由感測器的 佈局調整做某種程度的改善；然而在感測器佈局最佳化的過程中，每次改變佈局 後都需重新抽取電容以驗證偵測效果。目前抽取電容有兩種主要方式：在三維空 間中解波松方程、隨機遊走法(蒙地卡羅)，皆相當耗時。為減少抽取電容的數目， 本篇論文提出一統計方法，選擇特定情形做抽取電容，並以其結果建立電容模組， 再利用此電容模組做最佳化，找出佈局設計。

To provide users in consumer electronic products more convenient interface, there are kinds of sensors applied in modern electronic products. Touch pad is the most popular one, and it can be classified into three kinds: capacitance-type, resistance-type, and real time camera. We will focus on the capacitance-type touch pad and propose a methodology to accelerate the design flow in this thesis. In a capacitance-type touch pad, we design sensors in trace shape and place them crossed in horizontal and vertical directions. Once a finger is closing to a sensor, the electrical field changing will affect the capacitance value of the sensor. The processor will collect the capacitance value of all traces and calculate the coordinate where the finger locates. The quality of a touch pad is decided by two measures. One is the sensitivity of detecting whether a finger exists or not. The other is the resolution of finger coordinates. We can improve touch pad quality by modifying sensor layout. In the flow of sensor layout optimization, we must take time for capacitance extraction once the layout is modified. The state-of-art capacitance extraction is to solve Poisson equation in 3D or random walk-based method (Monte Carlo). Both of them are time-consuming. To reduce the number of capacitance extraction, we propose a statistical methodology that constructs a capacitance model relying on some results of capacitance extraction. And by applying this model, we can generate the optimal layout design of sensor.