隨者半導體製程技術尺寸的縮小,使得近幾十年超大型積體電路(VLSI)的設計在電路的效能及層級上有了很大的改進。然而元件尺寸的縮小,除了會有操作電壓降低和操作頻率增加這些隨著製程演進而產生的好處之外,同時也伴隨了設計可靠度降低的問題。而在近代的數位設計系統中,適應性電壓調變被視為是最有效能夠達到功率節省的設計方法,而且透過這個技術讓系統能動態偵測系統的操作,並給予即時的反應,因此能夠讓設計更加的健全。然而,這種低功耗技術在數位電路的設計階段時比較難以付諸實行。 標準元件流程設計在數位設計上最常被運用在客製化的晶片設計上。而標準元件庫正是這個流程不可或缺的元素之一。因此,在這個前提的假設下,對於標準元件庫的設計,我們提供了一套最佳化元件的流程挑選出具有較低消耗功率的元件尺寸,此元件庫支援操作在標準值以下電壓,另外由於動態電壓調整的考量,我們另外提出了一套後段的實現流程。最後在整個系統中,由於加入了適應性電路的調整,系統便可以順利找出系統電壓的最佳點。另外利用電路中的錯誤偵測電路,我們就能偵測系統在運作時的功能正確性,讓系統更加的完備。 最後,測試電路使用台積電的0.18m的製程進行晶片下線驗證,根據電路之模擬結果,動態電壓調整操作在64-QAM的調變方式下跟原來的保守估計的電壓相比,將會有約50%的功率改善。
With the continuous down scaling of the technology, the performance and levels of CMOS VLSI circuit has been on a tremendous boost over the past decades. However, device scaling, power supply lowering, and frequency increasing of operation that accompany with the technology progress will induce the reliability problem. In modern digital integrated circuit, adaptive voltage scaling is regarded to be the most efficient way to achieve low power design and which can also make a design more robust. However adopting this low power technique to the digital circuit is difficult in engineering practice at design stage. In digital system design, cell-based methodology is regarded to be the most popular design flow in the various customized IC design market. Based on this fact, we proposed a simple cell optimization flow to develop a cell library which enables a low voltage operation under nominal value. Meanwhile, voltage scaling technique can be adopted to the modified cell-based design flow. By adding the adaptive detect circuitry, the voltage value of the system will be adjusted to its optimal point. By the error detection signal report at runtime, we can observe the system correctness and make a more robust design. Finally, a test circuit is implemented under TSMC 0.18um 1P6M CMOS process. And it shows a dynamic voltage system will have almost 50% power saving comparing to the full margin design.