光纖收發器為通訊傳輸系統中重要的電子設備，當前朝向高功率和小型化封裝體積的方向發展，對於大型系統的開發來說，使用簡化的數學模型能夠有效減少計算資源和時間。 本研究針對具有五個發熱源且總消耗功率為25 Watts之光纖收發器QSFP (quad small form-factor pluggable)，運用DELPHI (DEvelopment of Libraries of PHysical models for an Integrated design environment)方法學的概念，開發出具有邊界條件獨立性之緊湊熱模型，為光纖收發器建構不同的熱阻網絡拓樸，運用田口法設計出適當的邊界條件集合，再使用商業模擬軟體Simcenter FloTHERM為光纖收發器建立計算流體力學模型，輸出所需的熱數據做為優化緊湊熱模型之依據，最後結合MATLAB編寫緊湊熱模型之優化過程，找出緊湊熱模型中最佳的熱阻值配置。 在本文使用的不同網絡型式之緊湊熱模型中，每個緊湊熱模型皆擁有良好的預測性與邊界條件獨立性，與詳細熱模型比較，發熱源溫度預測之平均相對誤差在2%以內，壁面熱傳量預測之平均相對誤差在8%以內，其中又以先進雙分流型網絡緊湊熱模型之預測能力為最佳，由此可知網絡拓樸的改進能有效改善緊湊熱模型之預測性。其後將緊湊熱模型應用於FloTHERM軟體中，模擬實際應用的環境，在發熱源溫度預測的平均相對誤差為6%左右，壁面熱傳量預測的平均相對誤差為10%左右，但最大相對誤差達到了20%以上，代表本文所開發出的緊湊熱模型仍有進步空間。未來若要將光纖收發器之緊湊熱模型實際應用在系統熱分析上，還須要儘可能地縮小緊湊熱模型的預測誤差，才能更準確地模擬光纖收發器之熱行為。 ;The fiber optical transceiver is an important electronic device in the communication system. Currently, it is evolving towards high power and miniaturized packaging. For the development of large-scale systems, the simplified model can effectively reduce computational resources and time. In this study, we focused on the quad small form-factor pluggable (QSFP) optical transceiver, which has five heat sources and a total power dissipation of 25 Watts. We applied the concept of DEvelopment of Libraries of PHysical models for an Integrated design environment (DELPHI) methodology to develop a compact thermal model with boundary condition independence. Different thermal resistance network topologies were constructed for the optical transceiver. Using the Taguchi method, appropriate sets of boundary conditions were designed. The commercial software Simcenter FloTHERM was then used to create a computational fluid dynamics model for the optical transceiver, which provided the thermal data for optimizing the compact thermal model. Finally, the optimization process of the compact thermal model was implemented using MATLAB to find the optimal thermal resistances. Among the different network topologies of the compact thermal models used in this study, each model exhibits good predictive accuracy and boundary condition independence. The average relative error in predicting heat source temperatures is within 2%, and the average relative error in predicting wall heat flux is within 8%. The advanced double shunted network compact thermal model has the best predictive capability, which indicates that improving the network topology can effectively enhance the predictive accuracy of the compact thermal model. When applying the compact thermal model to the computational fluid dynamics software FloTHERM for simulating realistic environments, the average relative error in predicting heat source temperatures is around 6%, and the average relative error in predicting wall heat flux is around 10%. The maximum relative error exceeds 20%. This indicates that there is still room for improvement in the developed compact thermal model in this study. In the future, if the compact thermal model of the optical transceiver is to be applied in system-level thermal analysis, it is necessary to minimize the predictive errors of the compact thermal model. This will enable more accurate simulation of the thermal behavior of the optical transceiver.
