Measurement, Identification and Modeling of Acoustical, Electrical and Mechanical Systems

• Main Authors: HERNANDEZ, DIEGO, 王小迪
• Other Authors: HUANG, JIN-HUANG
• Format: Others
• Language: en_US
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/7gpwzq

博士 === 逢甲大學 === 機械與航空工程博士學位學程 === 106 === This thesis focuses on the measurement, the identification and the modeling of acoustical, electrical and mechanical systems. An audio emulation platform implemented on Field Programmable Gate Array (FPGA) is presented. This platform allowed real-time implementations of Wave Digital Filter (WDF) models controlled via graphical user controls. Linear electronic circuits such as, an audio tone control and crossover filters for loudspeakers were emulated. A nonlinear Junction Field-Effect Transistors (JFET) model was presented and used to emulate audio distortion circuits. Furthermore, WDF linear and nonlinear models for moving coil loudspeakers were presented, comprising electrical, mechanical, thermal and acoustical parts. The WDF models presented in this thesis were found to be computationally faster than SPICE and Runge-Kutta methods, obtaining identical output results. Besides, emulations proved the high accuracy and real-time capability of the WDF models by measuring the signals from the FPGA I/O modules. Sequentially, this thesis addresses to the measurement, characterization and parameters identification of moving coil loudspeakers. Traditional measurement techniques analyze the impedance function to obtain the electrical parameters of the loudspeaker. However, in order to obtain the mechanical parameters, a second measurement has to be performed by adding a mass or a volume. The loudspeaker measurement system presented in this work uses a laser sensor to measure the diaphragm displacement, allowing the identification of electrical and mechanical parameters in a single measurement. The excitation of the loudspeaker under test was carried out by a DC-coupled amplifier. In order to identify the loudspeaker parameters, the measured signals were fitted in time and frequency domain by using nonlinear least-squares to the ones obtained by loudspeaker models. When the driver under test was excited by a large amplitude signal, the measurement of the diaphragm displacement provided information about the nonlinear symptoms of the driver. Then, the identification of the nonlinear parameters provided the physical cause of the distortion. The nonlinear parameters identification was carried out by full-dynamic and point-by-point methods. By using the WDF loudspeaker models, the parameters identification was performed 32 times faster than traditional Runge Kutta 4th order solver. For verification purposes, the parameters of six loudspeakers and four microspeakers were identified with very low fitting error. After the parameters identification, simulations of displacement and distortion were compared with measured data, matching the results. In addition, further verifications were performed by comparing with Klippel measurement system. Moreover, reliability and repeatability tests demonstrated the measurement system is stable under different working conditions.