A Highly Stable 8T SRAM Cell: P.Rama Koteswara Rao and D.Vijaya Kumar
A Highly Stable 8T SRAM Cell: P.Rama Koteswara Rao and D.Vijaya Kumar
A Highly Stable 8T SRAM Cell: P.Rama Koteswara Rao and D.Vijaya Kumar
2, 2012
Abstract
This paper describes stability problems in 6T SRAM cell and their solutions with 8T SRAM cell. 8T SRAM with the virtual ground concept has been proposed which increases the cell stability, increases read SNM and reduces leakage power. Comparison of 8T SRAM cell proposed with conventional 6T SRAM cell with respect to leakage power, read SNM are proved by simulation and experimentally using the tool cadence (180nm technology).
Keywords: Cell stability, Leakage Power, Read Static Noise Margin, Virtual Ground 1. Introduction
The embedded memories account to 65 percent of the area in modern SoC. The memory area increases with the complexity of SoC and by the year 2018, it is expected to go to 90 percent. Most of the memory is composed of the SRAM cells. With increasing variability in the future CMOS manufacturing processes, the SRAM cell stability, which depends on the balance of transistors, becomes a major concern. As the device dimensions decrease, the SRAM cell becomes more and more susceptible to process variations. The effect of random variations in logic paths can be overcome to a certain extent by placing multiple stages, but in the case of memory, each SRAM cell must function according to the specifications [1]. In addition to the data stability issues, the increasing leakage energy consumption of the embedded memory circuits is also a growing concern. In modern high performance microprocessors, more than 40 percent of the total active mode energy is consumed due to leakage currents [2]. In the conventional six transistor SRAM cell, the read and write operations impose conflicting constraints on the transistors. Due to this any improvement done to enhance the stability for one operation will result in performance degradation in the other operation. The problem of improving the SRAM cell stability is further aggravated by lower supply voltages. As the supply voltage decreases, the threshold voltage variation account for a large fraction of the supply voltage. Many design techniques have been proposed to overcome the variability problems. A higher supply voltage exclusively for the SRAM array apart from the normal supply voltage is one way to enhance the noise margins of the SRAM cell [3]. In this case, the supply voltage of the SRAM cell will not scale with the technology. Further, it may increase if the variability problems become prominent. The supply voltage can be dynamically modulated for read, write and standby operations to achieve the required noise margin [4].These separates the read and write operation from the standby operation. While these techniques provide better noise margins, they add to the complexity of the circuit. Also these techniques may result in increase of dynamic power consumption during the read and write operations. An 8T SRAM cell with two extra transistors and virtual ground has been proposed which is used to separate the read and write current path and avoid accidental cell flipping during read operation. This cell provides significant large SNM during Read operation improves cell stability, reduces leakage power and read power consumption.
February Issue
Page 31 of 84
Figure1: Schematic diagram of the conventional 6T SRAM cell showing the current flow during read operation
February Issue
Page 32 of 84
Figure 4: Proposed 8T SRAM cell during read operation 3.1.3 Role of virtual ground Whenever Q is 0 and Read_Word_Line is not enabled M7 causes unnecessary leakage. This can be avoided by connecting M7 to the virtual ground instead of ground. Virtual ground acts as ground when Read_Word_Line is enabled and in other case it provides logical 1. On overall this reduces a considerable amount of leakage power.
February Issue
Page 33 of 84
consists of 128 rows and 8 columns array of proposed 8T SRAM is designed. Another 128*8 conventional 6T SRAM is also designed for comparison.
Figure5: Comparison of the static noise margins of various SRAM cells SRAM _6 _2B, SRAM _6_ 3B and SRAM _6_4B represent the conventional 6T SRAM cell with of 2, 3 and 4 respectively. SRAM_8T represents the eight transistor SRAM cell proposed. The 8T SRAM have a separate read mechanism, and hence = 1.
February Issue
Page 34 of 84
6. CONCLUSION
In this paper the stability and leakage power reduction in 8T SRAM cell is shown with the use of virtual ground concept when compared to conventional 6T SRAM cell signal. By designing 8T SRAM cell with the help of virtual ground concept, the read SNM is increased (almost doubled).The cell also supports low power operation. This new asymmetric cell structure is capable of using differential sense technique for high speed read operation. REFERENCES
[1] L. Chang, R. K. Montoye, Y. Nakamura, K. A. Batson, R. J. Eickemeyer, R.H.Dennard, W. Haensch, and D.Jamsek, An 8t-sram for variability tolerance and low-voltage operation in highperformance caches, IEEE Journal of Solid State Circuits, vol. 43, no. 4, pp. 956963, 2006.
February Issue
Page 35 of 84
[2] G. Sery, S. Borkar, and V. De, Life is cmos: why chase the life after?in Proceedings of the 39th conference on Design automation, 2002, pp. 7883. [3] J. Davis, D. Plass, P. Bunce, Y. Chan, A. Pelella, R. Joshi, A. Chen, W. Huott, T. Knips, P. Patel, K. Lo, and E. Fluhr., A 5.6GHz 64kb dual-read data cache for the power6tm processor, in Digest of Technical Papers. IEEE International Solid-State Circuits Conference, 2006, pp. 25642571. [4] K. Zhang, U. Bhattacharya, Z. Chen, F. Hamzaoglu, D. Murray, N. Vallepalli, Y.Wang, B. Zheng, and M. Bohr, A 3-ghz 70-mb sram in 65-nm cmos technology with integrated column-based dynamic power supply, IEEE Journal of Solid-State Circuits, vol. 41, no. 1, pp. 146151, 2006.
7. Authors Profile
Mr. P Rama Koteswara Rao is currently working as Associate Professor in ECE Department, Usha Rama College of Engineering & Technology, Telaprolu, AP, India. He is working towards his PhD. at Andhra University College of Engineering, Visakhapatnam, AP, India. He received his M.Tech from JNTU College of Engg, Anantapur. He has thirteen years of experience in teaching undergraduate students and three years industrial experience. His research interests are in the areas of speech signal processing, embedded systems and Memory Design.
Mr D. Vijaya Kumar received his M.Tech degree from Institute of Technology, Banaras Hindu University, Varanasi India in 2010 and B.Tech from Malaviya National Institute of Technology, Jaipur, India. He is a member of IEEE. He is currently working as an Assistant Professor in ECE Department, Usha Rama College of Engineering & Technology, Telaprolu, AP, India. His areas of interests include Solid State Physics, VLSI and Memory Design.
February Issue
Page 36 of 84