隨著半導體製程技術的進步,電腦中央處理器CPU所能提供的性能也同步上升,此意味著單位體積內的熱通量也不斷上升,傳統的氣冷式散熱器已不足以滿足此需求,依照摩爾定律,在不久的將來CPU的性能將面臨散熱問題而無法提昇。 熱電致冷技術是應用Peltier效應,透過主動控制輸入電流,將熱量由TEC冷端傳遞到熱端去,有效降低發熱源的溫度,在搭配智慧型類神經網路控制理論及適應性控制理論為基礎,分別設計兩種不同的控制器,實際進行熱電致冷模組於CPU之溫控研究,並提供未來研究人員進行相關實驗之參考。 本研究將熱電致冷模組與CPU結合在一起進行溫控實驗,分別在CPU啟動時及CPU未啟動時進行實驗,實驗結果顯示在未開啟CPU時之溫控效果,不論在暫態或穩態,兩種控制器皆能有效的控制溫度。而在啟動CPU後,同樣能達到暫態及穩態之溫控表現。
As semiconductor manufacturing technology progress rapidly, the performance of CPU(central processing unit) has risen accordingly ,which means that the heat flux per unit volume has been increasing. The traditional air-cooled radiators can no longer satisfied the requirement. In accordance with Moore’s Law, the performance of CPU will soon be restricted by the challenge of heat dissipation in the near future. Thermoelectric cooling technology is the application of Peltier effect. By controlling the input current of thermoelectric cooler, the heat would transmit from the relative colder side to the hotter side. Older technologies showed that Intelligent Neural Network Control and Adaptive Control theories fit well at non-linear or time-varying systems. This study used these two theories to design CPU thermal controller modules. Experimental results showed that both of these two controllers’ settling time and steady state error are outstanding. This indicated the performance of CPU could be enhanced and reliability by this cooling module.