商用高密度直流/直流電源模塊之傳導性電磁干擾模擬與分析研究

Abstract

電磁雜訊一直是電源工程師的設計瓶頸。即使近年來在設計經驗上已有所進展，工業界在電磁干擾的問題上仍在摸索階段。雖然過去幾年在使用模擬軟體來預測電力電子電路之傳導性電磁干擾特性上做了許多的努力，然而其研究成果仍停留在學術研究領域，是否能應用於商用產品的開發仍令人質疑。針對此議題，本論文以最大輸出功率40W、切換頻率320kHz、且輸出電壓15V之商用高密度直流/直流電源模塊為研究對象，藉由電腦模擬軟體來探討其傳導性電磁干擾的特性。 本研究論文中，闡述電腦輔助模擬傳導性電磁干擾的方法。將電力元件、電路板與器件包裝中的重要寄生成分納入模擬電路後，使用IsSPICE模擬電路的時域特性，並將模擬結果作快速傅利葉轉換，進而得到電路的頻域電磁干擾資訊。 提出能減化模擬的複雜度卻不失去在10MHz以下頻率範圍之模擬精確度的方法。即使精確度受限制，分析結果足以提供產品之電磁干擾控制上許多有價值的資訊與實務的建議。對於超過10MHz之頻域模擬結果與實驗量測結果差異較大的情況並不意外，因為器件包裝間的近接效應並未全盤考量，因此未來需要作更深入的探究。

Electromagnetic Interference (EMI) has always been a major concern for power electronic circuit design engineers. Despite the progress made in recent years, there is still “black magic” associate with fixing the EMI problems in the industry. Attempts have been made in the past few years to use simulation tools to predict conducted EMI behavior in power electronic circuits and results were reported in research articles. However, the results reported in these articles were obtained in research environment. Doubts were expressed if such results can still be meaningful in a commercial environment. To address such an issue, efforts were spent to investigate the conducted EMI of a commercial 40-Watt 320-KHz 15-volt output high-density DC/DC power module using computer simulation tools. In the thesis, methodology of computer simulation of EMI behavior is outlined. Parasitic elements of importance, including those of the power components, the printed circuit board, and the assembly package, were included in the simulation. IsSPICE was used for time-domain simulation and Fast-Fourier Transform function was used to converter the simulation results into frequency-domain EMI data. It is concluded from the investigation that simulation complexity can be reduced to a practical level without losing the accuracy for frequency range below 10MHz. Although the accuracy is limited, analysis of the results already provides useful information and practical guidance for EMI reduction for the product. For frequency beyond 10MHz, the simulation results deviate significantly from the experimental results. This is not unexpected. Package proximity effects, which are still not fully understood, were not incorporated in the simulation. Further works along this line is needed in the future.