實作與驗證單事件效應容忍壓降轉換器

Abstract

隨著火箭發射成本降低，衛星通訊成為未來通訊發展的關鍵。衛星系統中，壓降轉換器可能受輻射引起的單事件效應(SEE)影響，導致輸出電壓擾動甚至造成系統損壞。 本研究以雙指數電流模型模擬SEE，並且以LET=15 (MeV*cm^2)/mg作為標準測試傳統電壓控制壓降轉換器。模擬顯示，運算放大器為傳統電壓控制壓降轉換器中最敏感之電路，將運算放大器組成緩衝器模擬SEE發生時，緩衝器在輸出電壓0.9V下輸出電壓擾動最大達到85mV。當運算放大器發生SEE時，傳統電壓控制壓降轉換器最大輸出擾動達到10mV。根據模擬使用類比冗餘、CVSL、延遲冗餘濾波器及DICE栓鎖器改善設計。模擬結果顯示單事件容忍緩衝器擾動自85mV降至26mV，單事件容忍電壓控制轉換器輸出電壓擾動自10mV降至3.3mV，有效減緩單事件對電壓控制壓降轉換器造成之影響。 本論文設計之傳統與單事件容忍電壓控制壓降轉換器皆使用台積電180奈米 CMOS製程製作，在質子測試中，使用動能230MeV、通量〖10〗^8 #/(s*cm^2 ) 姪子束照射三分鐘，傳統與單事件容忍緩衝器皆未觀察到SEE；雷射測試中，我們使用800nm波長雷射，其重複率為75MHz並且最大功率為30mW(單發雷射400pJ)。傳統緩衝器輸出恆定誤差隨雷射功率增加，單事件容忍緩衝器則無恆定誤差，抗輻射設計有效抵抗單事件效應干擾。與緩衝器相似，在模擬中傳統電壓控制壓降轉換器隨雷射能量增加輸出恆定誤差增加，並且在雷射注入電流超過5μA後無法穩定輸出電壓，而單事件容忍電壓控制壓降轉換器在任何注入電流下均保持輸出電壓穩定且無恆定誤差，證明設計有效抵抗單事件效應。

As the cost of rocket launches decreases, satellite communications have become the key to future communications. In satellite systems, buck converters may be affected by single-event effects (SEE) caused by radiation. SEE can cause output voltage disturbances or even system damage. This thesis uses a double-exponential current model to simulate SEEs, using LET=15 (MeV*cm^2)/mg as the standard for testing conventional voltage control buck converter. Simulations show that OP-AMP is the most sensitive circuit in voltage control buck converter. When OP-AMP is formed into a unit gain buffer(UGB) and tested under SEE, the UGB output voltage disturbance reaches 85 mV with 0.9V output voltage. For conventional voltage control buck converters, radiation hits OP-AMP, which can lead to 10mV voltage disturbance at the buck converter output. To mitigate these effects, analogue average, cascode voltage switch logic(CVSL), redundant delay filter, and DICE latch are used and tested. Simulation results show that the SE-tolerant UGB reduces disturbances from 85 mV to 26 mV, and the SE-tolerant voltage control buck converter reduces output disturbances from 10 mV to just 3.3 mV. Simulation shows SE-tolerant designs effectively minimize SEE-induced impacts. Conventional and SE-tolerant voltage control buck converters were fabricated using 180 nm CMOS process. In proton testing, we use proton beam with 230MeV kinetic energy and flux is 〖10〗^8 #/(s*cm^2 ).Neither the conventional nor SEE-tolerant buffers observed any SEE. In laser testing, we use an 800 nm laser with a 75 MHz repetition rate and max power of 30 mW (400pJ per shot). The conventional UGB showed a steady-state error that increased with laser power. In contrast, the SE-tolerant buffer showed no error, confirming the effectiveness of the SE-tolerant design. Similarly, simulations showed that the conventional voltage control buck converter's steady-state error increased with higher laser energy and became unstable beyond 5 μA of injected current. In contrast, the SE-tolerant voltage control buck converter’s output maintained stable and no error under all tested current injection conditions. Test and simulation results validate that SE-tolerant designs can against SEEs.