SRAM Cell Stability: A Dynamic Perspective

Abstract

SRAM cell stability assessment is traditionally based on <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">static</i> <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">criteria</i> of data stability requiring three coincident points in DC butterfly curves. This definition is based on static (DC) characteristics of the cell transistors. We introduce the <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">dynamic</i> <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">criteria</i> of cell data stability knowing that the cell operates in a dynamic environment alternating between access and non-access conditions. The proposed definition of the dynamic data stability criteria introduces a new bound for the cell static noise margin (SNM). It reveals that the true noise margin of the cell can be made considerably higher than the conventional SNM once the cell access time is sufficiently shorter than the cell time-constant. This phenomena can be used to extend the noise margin in (partial) subthreshold SRAMs. Moreover, a simulation method for verification of the dynamic data stability criteria is presented. Silicon measurement results in 130 nm CMOS technology confirms the concept of dynamic data stability and designer's ability to trade timing and static parameters . Finally, it is shown that the long time constant due to the subthreshold operation of the cell can be exploited to maintain data stability with proper choice of access and recovery time.