Narrow Results

Data stability, write ability and leakage power are major concerns in submicron static random access memory (SRAM) cell design. This paper presents an 11T SRAM cell with differential write and single-ended read. Proposed cell offers improved write ability by interrupting its ground connection during write operation. Separate read buffer provides disturb-free read operation. Characteristics are obtained from HSPICE simulation using 32nm high-performance predictive technology model. Simulation results show that the proposed cell achieves 4.5$\times$ and 1.06$\times$ higher read static noise margin (RSNM) as compared to conventional 6T (C6T) and PNN-based 10T cells, respectively, at 0.4V. Write static noise margin (WSNM) of the proposed design is 1.65$\times$, 1.71$\times$ and 1.77$\times$ larger as compared to those of C6T, PPN-based 10T and PNN-based 10T cells, respectively, at 0.4V. Write “1” delay of the proposed cell is 0.108$\times$ and 0.81$\times$ as those of PPN10T and PNN10T cells, respectively. Proposed circuit consumes 1.40$\times$ lesser read power as compared to PPN10T cell at 0.4V. Leakage power of the proposed cell is 0.35$\times$ of C6T cell at 0.4V. Proposed 11T cell occupies 1.65$\times$ larger area as compared to that of conventional 6T.

In this paper, a 11-T static random-access memory (SRAM) cell has been examined that shows a fair reduction in read power dissipation while upholding the stability and moderate performance. In the presented work, parametric variability analysis of various design metrices such as signal to noise margin, read current and read power of the Proposed 11T cell are presented and compared with few considered topologies. The Proposed cell offers single ended write operation and differential read operation. The improvement in read signal to noise margin and write signal to noise margin with respect to conventional 6T SRAM is 10.63% and 33.09%, respectively even when the write operation is single ended. Mean hold static noise margin of the cell for 3000 samples is $1.75\times$ times higher than considered D2p11T cell. Sensitivity analysis of data retention voltage (DRV) with respect to temperature variations is also investigated and compared with considered topologies. DRV variation with temperature is least in FF process corner. In comparison to conventional 6T SRAM cell, the write and read delay of Proposed 11T cell gets improved by $2.55\times$ and 1.64%, respectively. Proposed 11T topology consumes least read energy in comparison with considered topologies. In comparison with another considered 11T topology, i.e., D2p11T cell, Proposed cell consumes 13.11% lesser area. Process variation tolerance with Monte Carlo simulation for read current and read power has been investigated using Cadence virtuoso tool with GPDK 45-nm technology.

This paper presents a circuit-level technique of designing a low power and half select free 10T Static Random-Access Memory Cell (SRAM). The proposed cell works with single end read operation and differential write operation. The proposed bit-cell is free from half select issue and supports bit interleaving format. The presented 10T cell exhibits 40.75% lower read power consumption in comparison to conventional 6T SRAM cell, attributed to reduction of activity factor during read operation. The loop cutting transistors used in core latch improve write signal-to-noise margin (WSNM) by 14.94% and read decoupled structure improve read signal-to-noise margin (RSNM) by $2.02\times$ as compared to conventional 6T SRAM. In the proposed work, variability analysis of significant design parameters such as read current, stand-by SNM, and read power of the projected 10T SRAM cell is presented and compared with considered topologies. Mean value of hold static noise margin of the cell for 3000 samples is $1.75\times$ times higher than the considered D2p11T cell. The proposed 10T cell shows $1.83\times$ and $1.22\times$ narrower read access time and write access time, respectively, as compared to conventional 6T SRAM cell. Read current to bit-line leakage current ratio of the proposed 10T cell has been investigated and is improved by $2.75\times$ as compared to conventional 6T SRAM cell. The write power delay product and read power delay product of the proposed 10T cell are $4.21\times$ and $2.79\times$ lower than conventional 6T SRAM cell. In this work, cadence virtuoso tool with Generic Process Design Kit (GPDK) 45nm technology file has been utilized to carry out simulations.

With technology scaling, stability, power dissipation, and device variability, the impact of process, voltage and temperature (PVT) variations has become dominant for static random access memory (SRAM) analysis for productivity and failure. In this paper, ten-transistors (10T) and low power eight-transistors SRAM cells are redesigned using floating-gate MOS transistors (FGMOS). Power centric parameters viz. read power, write power, hold power and delay are the performance analysis metrics. Further, the stochastic parameter variation to study the variability tolerance of the redesigned cell, PVT variations and Monte Carlo simulations have been carried out for 10T FGMOS SRAM cell. Stability has been illustrated with the conventional butterfly method giving read static noise margin (RSNM) and write static noise margin (WSNM) metrics for read stability and write ability, respectively. A comparative analysis with standard six-transistor SRAM cell is carried out. HSPICE simulative analysis has been carried out for 32nm technology node. The redesigned FGMOS SRAM cells provide improved performance. Also, these are robust and reliability efficient with comparable stability.