LLC諧振轉換器由於具有初級側零電壓切換(zero-voltage-switching, ZVS)、次級側零電流切換(zero-current-switching, ZCS)、以及寬廣輸入範圍操作等操作特性,因此廣泛被用於資料中心、運算伺服器、儲存設備等應用,以滿足高效率與高功率密度之需求。同時,具有變壓器中心抽頭整流架構之LLC諧振轉換器,由於具有最少的整流元件數,在低電壓高電流的應用中,能夠減低整流元件產生的功率損失。然而,在硬體電路實現中,由於次級側整流迴路的電路雜散參數不相等,可能導致變壓器產生磁通偏移(flux walking)現象,使得變壓器可能進入磁通飽和狀態,同時也會造成次級側整流元件的功率損失不平衡,進而導致功率損失較大的整流元件容易損壞。針對此問題,本論文提出了一種磁通平衡控制策略,以改善磁通偏移問題。 本論文提出的控制架構是基於電壓模式控制架構,其中電壓控制迴路通過控制開關頻率以實現輸出電壓調整。在這個架構中,引入了磁通平衡控制迴路,通過調整初級側開關元件的責任週期,作為激磁電流直流成分的控制變數。由於LLC諧振轉換器的激磁電流無法直接感測,因此本論文提出了一種利用次級側整流二極體電流的感測方法,並通過數位化的取樣和估測機制間接獲得激磁電流的直流成分。此外,本論文亦針對次級側電路雜散參數的不對稱性對激磁電流的影響進行了數學分析,並推導出描述激磁電流直流成分與次級側電路雜散參數不對稱性之間關聯性的方程式。同時,根據這些結果,加入初級側責任週期控制變數,以調整激磁電流直流成分為零的目標,並推導出在考慮次級側電路雜散參數不對稱性的條件下所需的責任週期修正量。 此外,本論文也針對所提出的控制架構進行了控制迴路分析,定義相關的開迴路增益,以便於分析和控制器設計。本論文所提出之控制架構硬體實現是基於低成本之定點運算數位信號處理器(digital-signal-processor, DSP),因此,針對數位實現之相關議題,也一併在本論文中探討。最後,本論文分別使用電路模擬軟體PSIM建立整體數位控制模擬平台,同時建構硬體實驗環境,以輸入電壓380 V、輸出電壓20 V、輸出額定功率200 W為電路規格,藉由模擬與實驗相互驗證所提出控制方法有效性與可行性。
LLC resonant converters are widely used in applications such as data centers, cloud computing servers, and storage equipment due to their characteristics, including zero-voltage-switching (ZVS) on primary-side, zero-current-switching (ZCS) on secondary-side, and a wide input voltage range. These features make them suitable for meeting the demands of high efficiency and high-power density. Additionally, LLC resonant converters with a transformer center-tapped rectifier architecture have the advantage of requiring fewer rectification components, reducing power losses in applications with low voltage and high current requirements. However, in practical hardware implementations, the asymmetry in the parasitic parameters of the two secondary-side rectification loops can lead to magnetic flux imbalance (flux walking) in the transformer. This phenomenon may cause the transformer to enter a state of magnetic saturation and result in uneven power losses among the secondary-side rectification components. Consequently, rectification components with higher power losses are more susceptible to damage. To address this issue, this dissertation proposes a flux balance control strategy to mitigate the effects of flux imbalance. The control architecture proposed in this dissertation is based on a voltage-mode control framework, where the voltage control loop adjusts the output voltage by controlling the switching frequency. In this framework, a flux balance control loop is introduced, regulating the duty ratio of the primary-side switching elements as a control variable for the DC magnetizing current. Since the magnetizing current in LLC resonant converters cannot be directly sensed, this dissertation also presents a method of indirectly obtaining the DC magnetizing current using secondary-side rectification diode currents and digital sampling and estimation mechanisms. Furthermore, this dissertation conducts mathematical analysis of the impact of asymmetrical parasitic parameters in the secondary-side circuit on the magnetizing current. It derives equations that describe the relationship between the DC magnetizing current and the asymmetry in the secondary-side parasitic parameters. Based on these findings, a modification to the duty ratio is proposed to achieve the goal of the zero DC magnetizing current while considering the asymmetry in the secondary-side parasitic parameters. In addition, this dissertation also conducted control loop analysis for the proposed control architecture, defining relevant open-loop gains for analysis and controller design. The hardware implementation of the proposed control architecture is based on a low-cost fixed-point digital signal processor (DSP). Therefore, related issues regarding digital implementation are also discussed in this thesis. Finally, this dissertation establishes an integrated digital control simulation platform using circuit simulation software PSIM and constructs a hardware experimental environment. The circuit specifications are set at an input voltage of 380 V, an output voltage of 20 V, and a rated output power of 200 W. The effectiveness and feasibility of the proposed control methods are verified through mutual validation between simulation and experimentation.