• JSTS:Journal of Semiconductor Technology and Science
• /
• v.9 no.1
• /
• pp.37-50
• /
• 2009

Evaluation results about area scaling capabilities of various SRAM margin-assist techniques for random $V_T$ variability issues are described. Various efforts to address these issues by not only the cell topology changes from 6T to 8T and 10T but also incorporating multiple voltage-supply for the cell terminal biasing and timing sequence controls of read and write are comprehensively compared in light of an impact on the required area overhead for each design solution given by ever increasing $V_T$ variation (${\sigma}_{VT}$). Two different scenarios which hinge upon the EOT (Effective Oxide Thickness) scaling trend of being pessimistic and optimistic, are assumed to compare the area scaling trends among various SRAM solutions for 32 nm process node and beyond. As a result, it has been shown that 6T SRAM will be allowed long reign even in 15 nm node if ${\sigma}_{VT}$ can be suppressed to < 70 mV thanks to EOT scaling for LSTP (Low Standby Power) process.

• JSTS:Journal of Semiconductor Technology and Science
• /
• v.13 no.2
• /
• pp.87-97
• /
• 2013

Negative bias temperature instability (NBTI) and positive bias temperature instability (PBTI) are critical circuit reliability issues in highly scaled CMOS technologies. In this paper, we analyze the impacts of NBTI and PBTI on SRAM $V_{MIN}$, and present a design solution for mitigating the impact of NBTI and PBTI on SRAM $V_{MIN}$. Two different types of SRAM $V_{MIN}$ (SNM-limited $V_{MIN}$ and time-limited $V_{MIN}$) are explained. Simulation results show that SNM-limited $V_{MIN}$ is more sensitive to NBTI while time-limited $V_{MIN}$ is more prone to suffer from PBTI effect. The proposed NBTI/PBTI-aware control of wordline pulse width and woldline voltage improves cell stability, and mitigates the $V_{MIN}$ degradation induced by NBTI/PBTI.

• Journal of the Korea Computer Industry Society
• /
• v.6 no.5
• /
• pp.757-766
• /
• 2005

There have been great demands for higher density SRAM in all area of SRAM applications, such as mobile, network, cache, and embedded applications. Therefore, aggressive shrinkage of 6T Full CMOS SRAM had been continued as the technology advances, However, conventional 6T Full CMOS SRAM has a basic limitation in the cell size because it needs 6 transistors on a silicon substrate compared to 1 transistor in a DRAM cell. The typical cell area of 6T Full CMOS SRAM is $70{\sim}90F^{2}$, which is too large compared to $8{\sim}9F^{2}$ of DRAM cell. With 80nm design rule using 193nm ArF lithography, the maximum density is 72M bits at the most. Therefore, pseudo SRAM or 1T SRAM, whose memory cell is the same as DRAM cell, is being adopted for the solution of the high density SRAM applications more than 64M bits. However, the refresh time limits not only the maximum operation temperature but also nearly all critical electrical characteristics of the products such as stand_by current and random access time. In order to overcome both the size penalty of the conventional 6T Full CMOS SRAM cell and the poor characteristics of the TFT load cell, we have developed $S^{3}$ cell. The Load pMOS and the Pass nMOS on ILD have nearly single crystal silicon channel according to the TEM and electron diffraction pattern analysis. In this study, we present $S^{3}$ SRAM cell technology with 100nm design rule in further detail, including the process integration and the basic characteristics of stacked single crystal silicon TFT.

• Journal of the Korean Institute of Electrical and Electronic Material Engineers
• /
• v.19 no.4
• /
• pp.314-321
• /
• 2006

There have been great demands for higher density SRAM in all area of SRAM applications, such as mobile, network, cache, and embedded applications. Therefore, aggressive shrinkage of 6 T Full CMOS SRAM had been continued as the technology advances. However, conventional 6 T Full CMOS SRAM has a basic limitation in the cell size because it needs 6 transistors on a silicon substrate compared to 1 transistor in a DRAM cell. The typical cell area of 6 T Full CMOS SRAM is $70{\sim}90\;F^2$, which is too large compared to $8{\sim}9\;F^2$ of DRAM cell. With 80 nm design rule using 193 nm ArF lithography, the maximum density is 72 Mbits at the most. Therefore, pseudo SRAM or 1 T SRAM, whose memory cell is the same as DRAM cell, is being adopted for the solution of the high density SRAM applications more than 64 M bits. However, the refresh time limits not only the maximum operation temperature but also nearly all critical electrical characteristics of the products such as stand_by current and random access time. In order to overcome both the size penalty of the conventional 6 T Full CMOS SRAM cell and the poor characteristics of the TFT load cell, we have developed S3 cell. The Load pMOS and the Pass nMOS on ILD have nearly single crystal silicon channel according to the TEM and electron diffraction pattern analysis. In this study, we present $S^3$ SRAM cell technology with 100 nm design rule in further detail, including the process integration and the basic characteristics of stacked single crystal silicon TFT.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.46 no.9
• /
• pp.14-24
• /
• 2009

Memory cell array and peripheral circuits are designed for system on panel style frame buffer. Moreover, a parallel test methodology to test multiple blocks of memory cells is proposed to overcome low yield of system on panel processing technologies. It is capable of faster fault detection compared to conventional memory tests and also applicable to the tests of various embedded memories and conventional SRAMs. The various patterns of conventional test vectors can be used to enhance fault coverage. The proposed testing method is also applicable to hierarchical bit line and divided word line, one of design trends of recent memory architectures.

• ETRI Journal
• /
• v.22 no.1
• /
• pp.20-29
• /
• 2000

In this paper, we present a design of video and audio single chip encoder/decoder for portable multimedia application. The single-chip called as video audio signal processor (VASP) consists of a video signal processing block and an audio single processing block. This chip has mixed hardware/software architecture to combine performance and flexibility. We designed the chip by partitioning between video and audio block. The video signal processing block was designed to implement hardware solution of pixel input/output, full pixel motion estimation, half pixel motion estimation, discrete cosine transform, quantization, run length coding, host interface, and 16 bits RISC type internal controller. The audio signal processing block is implemented with software solution using a 16 bits fixed point DSP. This chip contains 142,300 gates, 22 Kbits FIFO, 107 kbits SRAM, and 556 kbits ROM, and the chip size is $9.02mm{\times}9.06mm$ which is fabricated using 0.5 micron 3-layer metal CMOS technology.

• ETRI Journal
• /
• v.40 no.6
• /
• pp.759-773
• /
• 2018

Uncore components such as on-chip memory systems and on-chip interconnects consume a large amount of energy in emerging embedded applications. Few studies have focused on next-generation analytical models for future chip-multiprocessors (CMPs) that simultaneously consider the impacts of the power consumption of core and uncore components. In this paper, we propose a convex-optimization approach to design heterogeneous uncore architectures for embedded CMPs. Our convex approach optimizes the number and placement of memory banks with different technologies on the memory layer. In parallel with hybrid memory architecting, optimizing the number and placement of through silicon vias as a viable solution in building three-dimensional (3D) CMPs is another important target of the proposed approach. Experimental results show that the proposed method outperforms 3D CMP designs with hybrid and traditional memory architectures in terms of both energy delay products (EDPs) and performance parameters. The proposed method improves the EDPs by an average of about 43% compared with SRAM design. In addition, it improves the throughput by about 7% compared with dynamic RAM (DRAM) design.