- 學位論文
基於輪廓圖方法之微機電慣性感測器最佳化設計
Contour-Plot Methodology for MEMS Inertial Sensor Design Optimization
摘要
隨著微機電慣性感測器已深入各領域中，其商用之產品，如：陀螺儀(Gyroscope)、加速度計(Accelerometer)等等商品充斥於整體市場，但深入其設計方法不難發現，由於此類裝置之幾何複雜性，多數這類慣性感測器，在設計階段皆仰賴於有限元素法；其根本原因即為因為其繁雜之取捨關係(Trade-off)，導致任何一點的幾何參數調整，都可能起到牽一髮動全身之效果，往往在無提前設想的情形下，調整設計只造成更大的混沌；因此多透過有限元素法來反覆驗證與試驗結果，便是設計方法之大宗；但其耗時耗費不茲，且當有新的設計需要調整時，往往就需要更多嘗試，對於設計者來說，也不夠直觀使用，彷彿只是將參數丟入黑盒子內便能得到答案，難針對特定目標優化。故本研究希望透過輪廓圖(Contour plots)方法來改善此一問題，其架構為首先了解該裝置之問題定義 (幾何參數、製程限制、規格等等)，再透過其相關之物理方程式，從物理的角度上，細分拆解這些參數，並同時利用數值計算，將大量的設計點位透過數學軟體計算，得出每種組合之解，再透過物理之特性，將其結果依據變數劃分。便能得到特定物理量對幾何參數之輪廓圖，利用輪廓圖清晰，趨勢明顯，可多重疊加之特性，協助設計者找出符合目標之設計。
並列摘要
As the MEMS inertial sensor has penetrated into various fields, its commercial products, such as gyroscope, accelerometer, etc. are command devices with numerous amounts in our life. Due to the geometric complexity of such devices, most of these inertial sensors rely on the finite element method in the design stage, the fundamental reason is that the complicated trade-off relationship leads to any point of geometry parameter adjustment that may have a huge effect on its behavior. Therefore, repeated verification through the finite element method is the most command design method. Obviously, it is time-consuming and costly, and when there is a new design that needs to be adjusted, it often requires more attempts. For the designer, it is not intuitive to use, as if it can be obtained by just throwing the parameters into a black box. The de-sign is difficult to optimize for specific goals.Therefore, this research hopes to improve this problem through the Contour plots design methodology. The framework is to first understand the problem definition of the device, such as geometric parameters, process constraints, specifications, etc. And then use the related physical equations to obtain the relationships between geometry parameters and physics behavior. At the same time using numerical calculations, a large number of design points are calculated through mathematical software to obtain the solution of each combination, and then through the physical characteristics, the results are divided according to variables. The contour map of the specific physical quantity versus the geometric parameter can be obtained. With the advantage of con-tour plots, such as data trend is obvious, and be able to observe multiple overlays, can be used to assist the designer in finding a design that meets the goal.