Static Random Access Memory (SRAM) is an indispensable part of most modern VLSI designs and dominates silicon area in many applications. In scaled technologies, maintaining high SRAM yield becomes more challenging since they are particularly vulnerable to process variations due to 1) the minimum sized devices used in SRAM bitcells and 2) the large array sizes. At the same time, low power design is a key focus throughout the semiconductor industry. Since low voltage operation is one of the most effective ways to reduce power consumption due to its quadratic relationship to energy savings, lowering the minimum operating voltage (Vmin) of SRAM has gained significant interest.This thesis presents four different approaches to design and analyze robust low voltage SRAM: SRAM analysis method improvement, SRAM bitcell development, SRAM peripheral optimization, and advance device selection.We first describe a novel yield estimation method for bit-interleaved voltage-scaled 8-T SRAMs. Instead of the traditional trade-off between write and read, the trade-off between write and half select disturb is analyzed. In addition, this analysis proposes a method to find an appropriate Write Word-Line (WWL) pulse width to maximize yield.Second, low leakage 10-T SRAM with speed compensation scheme is proposed. During sleep mode of a sensor application, SRAM retaining data cannot be shut down so it is important to minimize leakage in SRAM. This work adopts several leakage reduction techniques while compensating performance.Third, adaptive write architecture for low voltage 8-T SRAMs is proposed. By adaptively modulating WWL width and voltage level, it is possible to achieve low power consumption while maintaining high yield without excessive performance degradation.Finally, low power circuit design based on heterojunction tunneling transistors (HETTs) is discussed. HETTs have a steep subthreshold swing beneficial for low voltage operation. Device modeling and design of logic and SRAM are proposed.
【 预 览 】
附件列表
Files
Size
Format
View
Design and Analysis of Robust Low Voltage Static Random Access Memories.