| 摘要: | 隨著科技發展，光纖收發器朝向小體積高功率的方向發展，對於系統設備商而言熱
管理的壓力不斷提高，因此提供簡化的數學模型將能有效的減少其開發成本。
本研究針對 QSFP(Quad Small form-Factor Pluggable, QSFP)的熱測試載具(Thermal
Test Vehicle, TTV)進行風洞實驗，再透過風洞實驗驗證運用 DELPHI(DEvelopment of
Libraries of PHysical models for an Integrated design environment)方法學概念生成的
CTM(Compact Thermal Model, CTM)，熱測試載具共 7 個熱源，總功率19.729W。研究
內容包括針對熱測試載具進行風洞實驗以及透過商業模擬軟體Simcenter FloTHERM建
立計算流體力學(Computational Fluid Dynamics, CFD)模型，再透過與實驗完成驗證的
CFD模型生成熱測試載具的CTM。CTM的生成包括設計熱阻網絡、使用田口法建立邊
界條件集以及運用MATLAB來進行CTM熱阻值配置的最佳化。
生成 CTM 後將進行誤差評估，首先探討 CTM 的邊界條件獨立性(Boundary
Condition Independence, BCI)，在所有代表不同環境的邊界條件集中CTM與DTM(Detail
Thermal Model)的誤差均小於 10%，說明 CTM 具有良好的適用性。接著將 CTM 輸入
FloTHERM，模擬風洞實驗的條件來和CFD模型進行誤差評估，結果發現熱源溫度最大
誤差來到13.55%，CTM與風洞實驗的熱源溫度最大誤差則來到10.52%，可見本研究生
成的CTM雖然有一定的準確性，但仍然有改進空間。未來可使用本研究提供的方法針
對實際的光纖收發器開發CTM，但若要應用於系統級的開發則還要進一步改善誤差。;With the advancement of technology, fiber optical transceivers are moving towards smaller
sizes and higher power, placing increasing thermal management pressure on system equipment
vendors. Therefore, providing simplified mathematical models can effectively reduce their
development costs.
This study focuses on conducting wind tunnel experiments on the Thermal Test Vehicle
(TTV) of the Quad Small form-Factor Pluggable (QSFP), and verifies the Compact Thermal
Model (CTM) generated using the concept of DELPHI(DEvelopment of Libraries of PHysical
models for an Integrated design environment) methodology. The TTV includes 7 heat sources
with total power of 19.729W. The research involves wind tunnel experiments on the TTV and
uses the commercial simulation software Simcenter FloTHERM to creat TTV’s Computational
Fluid Dynamics (CFD) models. With the CFD model validated by the experiment, we generate
a CTM for the TTV. The CTM generation process includes designing thermal resistance
networks, using the Taguchi method to generate boundary condition sets, and optimizing CTM
thermal resistance configurations using MATLAB.
After generating the CTM, error assessment is conducted. Boundary Condition
Independence (BCI) of the CTM is first discussed, errors between CTM and DTM(Detail
Thermal Model) are all less than 10% across all boundary condition sets representing different
environments, demonstrating the good applicability of the CTM. Subsequently, the CTM is
input into FloTHERM to simulate wind tunnel experiment conditions and evaluate errors
against CFD models. Results show maximum errors in heat sources temperatures are less than
13.55%, while the maximum error between CTM and wind tunnel experiment heat sources
temperatures is 10.52%. This indicates that while the CTM generated in this study has a certain
level of accuracy, there is still room for improvement.
The methods provided in this study can be applied to develop CTMs for actual optical
ii
transceivers in the future. For system-level development applications, the accuracy of CTM
waits further improvements. |
