CHAPTER 3 PERFORMANCE OF A TWO INPUT NAND GATE USING SUBTHRESHOLD LEAKAGE CONTROL TECHNIQUES

Transcription

1 CHAPTER 3 PERFORMANCE OF A TWO INPUT NAND GATE USING SUBTHRESHOLD LEAKAGE CONTROL TECHNIQUES

2 41 In this chapter, performance characteristics of a two input NAND gate using existing subthreshold leakage control techniques such as stack technique, MTCMOS technique, sleepy keeper technique, and SCCMOS technique are compared with the conventional CMOS logic technique in 45nm CMOS technology. Performance characteristics such as static power dissipation, dynamic power dissipation, and propagation delay of a two input NAND gate are analyzed. Reduction in static power dissipation by reduction factors of 1.13x, 1.27x, 1.40x, and 1.44x are observed by using stack technique, MTCMOS technique, sleepy keeper technique, and SCCMOS technique respectively in comparison with conventional CMOS logic technique for a two input NAND gate. SCCMOS technique lowers the subthreshold leakage power dissipation most effectively. Hence, this technique effectively reduces the overall static power dissipation in digital logic circuits in deep submicron CMOS technologies. All simulations are performed at a temperature of 27 0 C, and a supply voltage of 0.40V in 45nm CMOS technology. Microwind ver. 3.1 EDA tool is used in designing and simulating the layout of a two input NAND gate in various technologies using BSIM4 MOS parameter model. Finally conclusion is provided in the end of this chapter. 3.1 INTRODUCTION Power dissipation in digital logic circuits can be broadly divided into two categories: dynamic power dissipation and static or leakage power dissipation. Dynamic power dissipation is mainly caused by the current flow due to charging and discharging of parasitic capacitances in the logic circuit. Static power dissipation occurs during the static input states of the device. With the down scaling in technology, contribution by static power dissipation increases in the overall power dissipation (Elgharbawy et al. 2005). In deep submicron CMOS technologies, the role of subthreshold leakage power dissipation becomes dominant among other leakage power components (Deepaksubramanyan et al. 2007) because of down scaling in technology. Under such condition, the static power dissipation is approximately equal to the subthreshold leakage power dissipation and is expressed as (3.1) A general formula for the total power dissipation in a digital logic circuit in deep submicron CMOS technologies can be expressed as

3 42 (3.2) (3.3) 3.2 CIRCUIT DESIGN USING VARIOUS TECHNIQUES In this section, circuit diagrams of a two input NAND gate are designed using conventional CMOS technique and existing subthreshold leakage control techniques such as stack technique, MTCMOS technique, and SCCMOS technique CONVENTIONAL CMOS TECHNIQUE A conventional CMOS circuit (Kang et al. 2003) consists of a pull up network and a pull down network. A pull up network consists of pmos transistors while a pull down network consists of nmos transistors. In this technique, all pmos transistors must have either an input from the power supply voltage (V DD ) or from another pmos transistor. Similarly, all nmos transistors must have either an input from ground or from another nmos transistor. Fig. 3.1 shows the circuit diagram of a two input NAND gate using this technique. Fig. 3.1 Circuit diagram of a two input NAND gate using conventional CMOS technique

4 STACK TECHNIQUE In stack technique (Johnson et al. 2002), a MOS transistor of width W is replaced with two series connected MOS transistors, each of width W/2. Fig. 3.2 shows the circuit diagram of a two input NAND gate using this technique. In this figure, each MOS transistor (width W) of a CMOS two input NAND gate is replaced with two series connected MOS transistors (each of width W/2). When two half size stacked MOS transistors are turned off together, induce reverse bias between them results in the reduction of the subthreshold leakage power (Chen et al. 1988). However, increase in the number of transistors increases the overall propagation delay of the circuit. Fig. 3.2 Circuit diagram of a two input NAND gate using stack technique MTCMOS TECHNIQUE In MTCMOS technique (Mutoh et al. 1995), a high-threshold voltage pmos transistor (sleep pmos transistor) is inserted between the power supply voltage (V DD ) and the logic circuit; and a high-threshold voltage nmos transistor (sleep nmos transistor) is placed between the logic circuit and ground. Fig. 3.3 shows the circuit diagram of a two input NAND gate using this technique. During active mode of

5 44 operation, the high threshold voltage MOS transistors (sleep transistors) are turned on; while during sleep mode, these high V TH transistors are turned off. Fig. 3.3 Circuit diagram of a two input NAND gate using MTCMOS technique SLEEPY KEEPER TECHNIQUE In sleepy keeper technique (Kim et al. 2006), an additional high threshold voltage nmos transistor is connected in parallel with the sleep pmos transistor (high V TH sleep pmos transistor) and an additional high threshold voltage pmos transistor is connected in parallel with the sleep nmos transistor (high V TH sleep nmos transistor). Fig. 3.4 shows the circuit diagram of a two input NAND gate using this technique. In sleep mode, the sleep transistors are in cutoff state. So, when sleep signal is activated, then the high threshold voltage nmos transistor connected in parallel with the sleep pmos transistor is the only source of power supply to the pullup network and the high threshold voltage pmos transistor connected in parallel with the sleep nmos transistor provides the path to connect the pulldown network with ground.

6 45 Fig. 3.4 Circuit diagram of a two input NAND gate using sleepy keeper technique SCCMOS TECHNIQUE In SCCMOS technique (Kawaguchi et al. 2000), a pmos transistor (sleep pmos transistor) having the same threshold voltage as that of pmos transistors of the logic circuit is inserted between the logic circuit and the power supply voltage (V DD ); and an nmos transistor (sleep nmos transistor) having the same threshold voltage as that of nmos transistors of the logic circuit is inserted between the logic circuit and ground. Fig. 3.5 shows the circuit diagram of a two input NAND gate using this technique. During active mode of operation, sleep transistors are turned on by connecting GND at V GS1 and V DD at V GS2 ; while during sleep mode, these sleep transistors are turned off by providing positive and negative gate voltages at V GS1 and V GS2 respectively.

7 46 Fig. 3.5 Circuit diagram of a two input NAND gate using SCCMOS technique 3.3 SIMULATION RESULTS AND OBSERVATIONS Performance characteristics such as static power dissipation, dynamic power dissipation, and propagation delay for a two input NAND gate using subthreshold leakage control techniques such as stack technique, MTCMOS technique, sleepy keeper technique, and SCCMOS technique are compared with the conventional CMOS technique. All simulations are performed at a temperature of 27 0 C and a supply voltage of 0.40V in 45nm CMOS technology. Microwind ver 3.1 EDA tool is used for the layout design and simulation of a two input NAND gate using advanced BSIM4 MOS parameter model. Figs show the layout of a two input NAND gate using conventional CMOS technique, stack technique, MTCMOS technique, sleepy keeper technique, and SCCMOS technique respectively. MOS transistor parameters such as channel length, channel width, and aspect ratio used to design a two input NAND gate using various techniques are mentioned in Tables Aspect ratio of a MOS transistor is the ratio of its channel width and its channel length. The threshold voltage of a MOS transistor increases with the increase in its channel length. So, the channel lengths of high threshold voltage transistors and sleep transistors are chosen greater than the

8 47 normal threshold voltage transistors (low V TH transistors) in MTCMOS and sleepy keeper techniques. Table 3.6 shows all possible static input combinations for measuring static power dissipation in a two input NAND gate. Fig shows the waveform to measure the dynamic power dissipation of the logic gate using conventional CMOS and stack techniques. Similarly, Fig shows the waveform to measure the dynamic power dissipation of a two input NAND gate using MTCMOS and sleepy keeper techniques and Fig.3.13 shows the waveform to measure the dynamic power dissipation using SCCMOS technique. Fig. 3.6 Layout of a two input NAND gate using conventional CMOS technique

9 48 Fig. 3.7 Layout of a two input NAND gate using stack technique Fig. 3.8 Layout of a two input NAND gate using MTCMOS technique

10 49 Fig. 3.9 Layout of a two input NAND gate using sleepy keeper technique Fig Layout of a two input NAND gate using SCCMOS technique

11 50 Table 3.1: Physical aspects of MOS transistors using conventional CMOS technique Parameters nmos transistor pmos transistor Channel width (μm) Channel length (μm) Aspect ratio Table 3.2: Physical aspects of MOS transistors using stack technique Parameters nmos transistor pmos transistor Channel width (μm) Channel length (μm) Aspect ratio Table 3.3: Physical aspects of MOS transistors using MTCMOS technique Parameters nmos pmos Sleep nmos Sleep pmos transistor transistor transistor transistor (low V TH ) (low V TH ) (high V TH ) (high V TH ) Channel width (μm) Channel length (μm) Aspect ratio Table 3.4: Physical aspects of MOS transistors using sleepy keeper technique Parameters nmos pmos Sleep nmos Sleep pmos transistor transistor transistor transistor (low V TH ) (low V TH ) (high V TH ) (high V TH ) Channel width (μm) Channel length (μm) Aspect ratio

12 51 Table 3.5: Physical aspects of MOS transistors using SCCMOS technique Parameters nmos transistor (low V TH ) pmos transistor (low V TH ) Sleep nmos transistor (low V TH ) Sleep pmos transistor (low V TH ) Channel width (μm) Channel length (μm) Aspect ratio Table 3.6: Static input combinations for measuring static power dissipation Static input A Static input B Output Y Fig Waveform to measure the dynamic power dissipation using conventional CMOS and stack techniques Fig Waveform to measure the dynamic power dissipation using MTCMOS and sleepy keeper techniques

13 52 Fig.3.13 Waveform to measure the dynamic power dissipation using SCCMOS technique Static power dissipation is obtained by combining all possible static input combinations. The overall static power dissipation is calculated as the average of power dissipation in all possible static input combinations. This power dissipation for a two input NAND gate using various techniques is measured for 50ns time interval. Dynamic power dissipation in a logic circuit is mainly due to the switching power dissipation. From Figs , it is observed that the input signal frequency (A and B) is equal to the switching frequency of the output waveform. So, the output switching frequency is equal to the input signal frequency. The supply voltage, V DD is equal to 0.40V. Dynamic power dissipation is obtained by applying two clock signals, A and B, of same frequency of 200MHz. The amplitude of each clock signal is fixed at 0.40V. This power dissipation for a two input NAND gate using various techniques is obtained for 50 ns time interval. In SCCMOS technique, dynamic power is obtained by connecting V GS1 to GND and V GS2 to V DD. Similarly, in MTCMOS and sleepy keeper techniques, sleep signal and sleep signal are connected to GND and V DD, for obtaining the dynamic power dissipation. In this way, dynamic power dissipation can be obtained with the simulated waveforms for a logic circuit by using various conventional and reported techniques. Propagation delay of the logic gate by using various techniques is measured from the trigger input edge reaching 50 % of V DD to the circuit output edge reaching 50 % of V DD. Table 3.7 shows the performance characteristics of a two input NAND gate using conventional CMOS, stack, MTCMOS, sleepy keeper, and SCCMOS techniques in 45nm CMOS technology. Table 3.8 shows the static power dissipation reduction factors for a two input NAND gate using stack, MTCMOS, sleepy keeper, and

14 53 SCCMOS techniques in comparison with the conventional CMOS technique. Fig shows the static power dissipation comparison of a two input NAND gate by using various techniques. Fig shows the static power dissipation reduction factors for a two input NAND gate using existing subthreshold leakage reduction techniques, such as stack, MTCMOS, sleepy keeper, and SCCMOS techniques in comparison with the conventional CMOS logic technique. Reduction in static power dissipation by reduction factors of 1.13x, 1.27x, 1.40x, and 1.44x are observed by using stack technique, MTCMOS technique, sleepy keeper technique, and SCCMOS technique respectively in comparison with conventional CMOS logic technique for a two input NAND gate. It is found that SCCMOS is the most effective technique to reduce the static power dissipation in comparison with other existing subthreshold leakage control techniques. It is also important to note that in this technique the dynamic power and delay remains almost same as that of conventional CMOS logic technique. Table 3.7: Performance characteristics of a two input NAND gate using various circuit techniques Techniques Static power Dynamic power Delay (in μw) (in μw) (in sec.) CMOS x Stack x MTCMOS x Sleepy keeper x SCCMOS x 10-12

15 54 Table 3.8: Static power dissipation reduction factors for a two input NAND gate using stack, MTCMOS, sleepy keeper, and SCCMOS techniques in comparison with the conventional CMOS logic technique Techniques Static power dissipation reduction factors in comparison with conventional CMOS technique Stack 1.13x MTCMOS 1.27x Sleepy keeper 1.40x SCCMOS 1.44x Fig Static power dissipation comparison of a two input NAND gate using various techniques

16 55 Fig Static power dissipation reduction factors for a two input NAND gate using subthreshold leakage reduction techniques in comparison with CMOS technique 3.4 CONCLUSION In this chapter, performance characteristics such as static power dissipation, dynamic power dissipation, and propagation delay of a two input NAND gate are analyzed using various techniques. Performance characteristics of a two input NAND gate using subthreshold leakage control techniques such as stack technique, MTCMOS technique, sleepy keeper technique, and SCCMOS technique are compared with the conventional CMOS logic technique in 45nm CMOS technology. Among these existing techniques, SCCMOS technique effectively reduces the subthreshold leakage power dissipation while the dynamic power and delay remain low. So, in deep submicron and nanoscale technologies, this technique can be utilized for designing digital circuits with effective reduction in the static power dissipation, which is mainly due to the subthreshold leakage power dissipation.