Академический Документы
Профессиональный Документы
Культура Документы
Abstract - In this paper we propose a novel design of a low power method, by using a charge sharing technique the bit-lines
static random access memory (SRAM) cell for high speed voltage swing has been reduced [4]. Another technique known
operations. The model adopts the voltage mode method for as half swing pulse mode technique is also proposed to reduce
reducing the voltage swing during the write operation switching the dynamic power [5]. Hierarchically divided bit-line
activity. Dynamic power dissipation increases when the operating approach is shown to effective for reducing the active power in
frequency of the SRAM cell increases. In the proposed design we SRAMs by reducing bit-line capacitance [6]. Multiple valued
use two voltage sources connected with the Bit line and Bit bar technique (MLV) is also a very useful method for reducing the
line for reducing the voltage swing during the write “0” or write voltage swing of the output. MLV circuits are designed with
“1” operation. We use 90 nm CMOS technology with 1 volt of
both voltage mode [7-8] and current mode [9-10].
power supply. Simulation is done in Microwind 3.1 by using
BSim4 model. Dynamic power for different frequencies is This paper presents a novel design technique for low
calculated. We compare it with conventional 6-T SRAM cell. The power high speed SRAM circuits. In this paper the proposed
simulation results show that the power dissipation is almost SRAM cell, unlike the conventional 6T SRAM, contains two
constant even the frequency of the proposed SRAM model extra transistors connected to two voltage sources for reducing
increases. This justifies the reduction of the dynamic power the voltage swing during write operation. It is found that for
dissipation for high frequency CMOS VLSI design. high frequency of operation, the dynamic power dissipation is
almost constant for the proposed SRAM cell.
Keywords— CMOS; Dynamic power; SRAM; Voltage Mode;
Voltage Swing The paper is organized as follows: in section II, the design
of the conventional 6T SRAM cell is discussed. Section III
I. INTRODUCTION describes circuit design and its working principle of the
Reduction of power consumption makes a device more proposed novel SRAM cell. The simulations results and
reliable. The need for devices that consume a minimum amount discussions are presented in section IV and finally section V
of power was a major driving force behind the development of concludes the paper.
CMOS technologies. As a result, CMOS devices are best
known for low power consumption. However, for minimizing II. CONVENTIONAL 6T SRAM CELL
the power requirements of a board or a system, simply Figure 1 shows the circuit diagram of a conventional
knowing that CMOS devices may use less power than SRAM cell. Word line is used for enabling the access
equivalent devices from other technologies does not help much transistors M1 and M2 for write operation [9]. BL and BL
[1]. lines are used to store the data and its compliment. For write
Low power design has become a critical issue in VLSI operation one BL is high and the other bit line is on low
design, especially for high speed systems [2]. Usually, dynamic condition. For writing “0”, BL is Low and BL is high. When
power dominates the power dissipation in most digital systems. we assert the word line high transistor M1 and M4 is on and
Dynamic power dissipation mainly depends on the switching
frequency, supply voltage, and the output voltage swing. any charge stored in the BL goes through M1- M4 path to
Reducing the supply voltage is the most effective approach to ground. Due to zero value at Q, the M5 transistor is ON and
decrease dynamic power dissipation. Unfortunately, lower M6 is OFF. So the charge is stored at Q bar line. Similarly in
supply voltage degrades performance dramatically [3]. Lower the write “1” operation,BL is high due to this M6 is ON and
supply voltage decreases the threshold voltage which will
increase the sub-threshold current or leakage current so the the charge is stored on the Q is discharged through the M2-
static power dissipation increases. Limiting the output voltage M6 path and due to this low value on the Q , M3 is ON and
swing is another method to reduce dynamic power and delay.
M4 is OFF so the charge is stored on the Q.
Different approaches are used for reducing the voltage
swing during the switching activities in SRAM design. One of
the well known method is Charge Sharing Technique. In this
The comparison of the power dissipation between area are increased in comparison to conventional SRAM cell
conventional SRAM and the proposed model has been shown but low power dissipation even at very high frequency can
in Table II. easily overcome this drawback. This proposed SRAM cell can
be used to provide low power solution in high speed devices
TABLE II. COMPARISON OF POWER DISSIPATION BETWEEN THE like laptops, mobile phones, programmable logic devices etc.
PROPOSED VS CONVENTIONAL SRAM CELL
REFERENCES
Operating Power Dissipation in 6T Power Dissipation in [1] A. P. Chandrakasan and R. W. Broderson, "Minimizing power
Frequency SRAM cell (μW) Proposed SRAM cell consumption in digital CMOS circuits," Proc. of the IEEE, vol. 83,no.4,
(μW) pp. 498-523, April 1995.
500 MHz 4.890 4.977 [2] P. K. S. Vasundara, and K. S. Gurumurthy, "Quaternary CMOS
1 GHz 8.002 5.250 Combiniational Logic Circuits," IEEE Internatinal Conferece on
2 GHz 11.979 6.281 Information and Multimedia Technology, pp. 538-542,Dec. 2009.
[3] B. Ebrahimi, S. Zeinolabedinzadeh, and A. A. Kusha,"Low Standby
Power and Robust FinFET Based SRAM Design", IEEE Computer
Table II shows that if we double the operating frequency of Society Annual Symposium on VLSI, pp. 185-190, Jan 2008.
the conventional SRAM cell the power dissipation is almost [4] Gu Ming, Yang Jun, Xue Jun, "Low power SRAM design using charge
double, but in case of proposed SRAM cell power dissipation sharing technique," 6th International Conference On ASIC, ASICON,
at 1GHz and 2 GHz is almost same as that of 500 MHz. pp.19-23, Oct.2005.
[5] K.W. Mai, T. Mori, B. S. Amrutur, R. Ho, B. Wilburn, M. A. Horowitz,
So from these results it is evident that the proposed I. Fukushi, T. Izawa, and S. Mitarai, "Low power SRAM design using
approach for CMOS based SRAM design is suitable for high half-swing pulsemode techniques", IEEE J. Solid-State Circuits, Vol.
speed VLSI design. 33,no.1, pp. 1659-71, Jan. 1998.
[6] A. Karandikar,; K. K. Parhi, "Low power SRAM design using
In the proposed SRAM cell the crosstalk voltage values are hierarchical divided bit-line approach," International Conference on
increased for bit lines, word line (WL) and for outputs in Computer Design: VLSI in Computers and Processors, ICCD '98, pp.82-
comparison to conventional SRAM cell but these values can 88, 5-7 Oct 1998.
be controlled with the help of proper sizing of width (W) and [7] I. Thoidis, D. Sourdris, I. Karafyllidis, A. Thanailakis, and T. Stoursitis,
"Design Methodology of Multiple-Valued Logic Voltage-Mode
length (L) of the transistor. StorageCircuits", IEEE Internatinal Symposium on Circuit and System,
pp. 125-128, March 1998.
V. CONCLUSIONS [8] R. C. G. da Silva, H. Boudinov, L. Carro, "A Novel Voltage-Mode
Dynamic power dissipation has become a major issue in high CMOS Quaternary Logic Design", IEEE Transaction on Electron
Devices, vol. 53, no. 6, pp. 1480-1483, June 2006.
speed VLSI designs. In this paper we propose a novel design
[9] Keivan Navi, Roshanak Zabihi, Majid Haghparast, Touraj Nikobin, "A
for low dynamic power SRAM cell by using voltage mode Novel Mixed Mode Current and Dynamic Voltage Full adder", World
approach. This proposed SRAM cell has two voltage sources Applied Sciences Journal, Vol.4,no.2, pp. 289 -294, 2008.
which are used for reducing the voltage swing during [10] A. Ghorbannia Delavar, K. Navi and O. Hashemipour, "Very Fast
switching activity. The reduction in voltage swing results in Current Mode Logic Gates," CSIT Conference, Yerevan, Armenia, pp.
reduction of dynamic power dissipation. Dynamic power 19-23, June 2005.
dissipation for proposed SRAM cell is almost constant for [11] Neil H. E. Weste, David Harris, Ayan Banerjee, "CMOS VLSI Design",
Pearson Education, 3rd Edition, pp. 55-57.
high speed operations. Although numbers of transistors and