-
随着集成电路进入纳米时代,电路的可靠性问题越来越严重。目前,软错误已经成为影响集成电路可靠性的主要问题,其中由高能粒子诱发的单粒子翻转(single event upset,SEU)是存储元件中软错误的主要来源[1-2]。因此在纳米工艺下,SEU的加固设计对于提高电路的可靠性具有重要的意义。
由于集成电路特征尺寸的不断缩减,电源电压的不断下降,电路节点的关键电荷不断减少。相关研究表明,随着集成电路的特征尺寸进入90 nm后,电荷共享导致的多节点翻转已经成为一个问题[3]。电荷共享是单个高能粒子轰击硅材料,产生的电荷被多个敏感节点收集的一种辐射效应。国内外学者对电荷共享和多节点翻转进行了大量的研究,文献[4]对SEU加固单元的多节点翻转进行了3D器件模拟研究。文献[5]研究了电荷共享对于软错误率的影响。文献[6]指出单粒子多节点翻转所诱发的软错误愈发严重。单粒子多节点翻转给加固锁存器设计提出了更高的要求。目前大多数SEU加固锁存器设计[7-10]都是针对单粒子单节点翻转的防护,没有防护单粒子多节点翻转的能力。单粒子多节点翻转已经成为纳米工艺下抗辐照芯片的主要挑战。
针对单粒子多节点翻转问题,文献[11]提出了基于版图的加固技术。该技术通过在版图上采取分离敏感节点等方法,以降低电荷共享导致的多节点翻转的概率。但是该技术的加固性能有限,同时也很难有效地应用于大规模集成电路的自动化设计。区别于版图加固技术,文献[12]提出了一种基于DICE[7]单元的SRAM结构。文献[13]提出了一种将DICE单元与级连电压开关逻辑 (cascode voltage switch logic,CVSL)逻辑门相结合的锁存器设计,但是该锁存器的面积和功耗开销非常大,在实际应用中价值不大。文献[14]提出一种基于DICE单元和隔离思想的锁存器设计,但是该锁存器只能对部分单粒子多节点翻转实现容错。
本文针对单粒子多节点翻转问题,以及现有方案的一些不足,提出了一种新颖的加固锁存器设计。该锁存器使用代码字状态保存单元(code word state preserving,CWSP) [15]构成三模互锁结构,并在锁存器末端使用一个CWSP单元实现对单粒子多节点翻转的容错。该锁存器不仅具有单粒子单节点翻转的容错能力,而且具有单粒子多节点翻转的容错能力。使用HSPICE工具对该锁存器进行了广泛的SEU故障注入实验,实验结果表明,该锁存器能够很好地容忍单粒子多节点翻转。
-
为了评估研制的锁存器的性能,本文比较了TMR锁存器、FERST锁存器、HRPU[18]锁存器、DNCS-SEU锁存器以及该锁存器的SEU容错能力;比较了上述各锁存器的面积、延迟、功耗开销。由于集成电路特征尺寸的缩减,工艺偏差对于电路的影响也越来越严重[19]。为了评估锁存器对工艺偏差的敏感性,本文对上述各锁存器进行了工艺偏差分析。
-
根据前文分析,各锁存器的SEU容错能力如表 1所示。由表 1可得,本文的加固锁存器具有单粒子单节点翻转和单粒子多节点翻转的容错能力。
-
使用HSPICE仿真工具比较了各锁存器的延迟和功耗开销,并计算了功耗延迟积(power delay product,PDP)。仿真中使用PTM 45 nm模型,电源电压为1 V,系统时钟为500 MHz,温度为30℃。各锁存器的开销如表 2所示。计算了本文的锁存器相比于其他锁存器的开销变化,则有:
$$\Delta \text{=}\left( 本文的锁存器-其他锁存器 \right)/其他锁存器$$ (1) 表 2 锁存器的开销
结果如表 3所示。为了更好地比较两种容忍单粒子多节点翻转的锁存器的开销,依次在PTM 45、32、22 nm工艺下比较了两种锁存器的PDP,结果如图 7所示。由图 7可知,相比于DNCS-SEU锁存器,本文的锁存器PDP有大幅的下降。
表 3 本文的锁存器相比于其他锁存器的开销变化
-
对于器件,最主要的扰动是沟道长度和阈值电压。在HSPICE工具中运用蒙特卡罗分析来评估沟道长度和阈值电压变化对于电路性能的影响。设定沟道长度、阈值电压的变化都服从高斯分布,并设定最大的偏差为10%。仿真中使用PTM 45 nm模型,电源电压为1 V,系统时钟为500 MHz,温度为30℃。图 8所示为10 000次蒙特卡罗分析的实验结果。
由上述实验结果可得,相比于TMR锁存器和DNCS-SEU锁存器,本文的锁存器受工艺偏差的影响更小;与FERST锁存器相比,两者受工艺偏差的影响相当。所以在上述比较的各锁存器中,本文的锁存器对工艺偏差具有较低的敏感性。
A Single Event Multiple Upset Tolerant Hardening Latch with Triple Interlock
-
摘要: 为了能够容忍单粒子多节点翻转,提出了一种新颖的三模互锁加固锁存器。该锁存器使用具有过滤功能的代码字状态保存单元(CWSP)构成三模互锁结构,并在锁存器末端使用CWSP单元实现对单粒子多节点翻转的容错。HSPICE仿真结果表明,相比于三模冗余(TMR)锁存器,该锁存器功耗延迟积(PDP)下降了58.93%;相比于容忍多节点翻转的DNCS-SEU锁存器,该锁存器的功耗延迟积下降了41.56%。同时该锁存器具有较低的工艺偏差敏感性。Abstract: In nanometer process, a single event induced multiple upset cannot be ignored. A novel triple interlock hardening latch is proposed for tolerating single event multiple upset. The proposed latch employs code word state preserving (CWSP) cell which has the filtering function to compose triple interlock. At the end of latch, the CWSP cell is also exploited to tolerate single event multiple upset. The simulation results of HSPICE suggest that compared to triple modular redundancy (TMR) latch and DNCS-SEU latch, the power delay product of the proposed latch is reduced by 58.93% and 41.56% respectively. Meanwhile, the proposed latch has less sensitiveness to process variations.
-
Key words:
- hardening latch /
- multiple node upset /
- soft error /
- triple interlock
-
表 1 锁存器的SEU容错能力比较
表 2 锁存器的开销
-
[1] BAUMANN R C. Radiation-induced soft errors in advanced semiconductor technologies[J]. IEEE Transactions on Device and Materials Reliability, 2005, 5(3):305-316. doi: 10.1109/TDMR.2005.853449 [2] PETERSEN E, KOGA R, SHOGA M A, et al. The single event revolution[J]. IEEE Transactions on Nuclear Science, 2013, 60(3):1824-1835. doi: 10.1109/TNS.2013.2248065 [3] ZHU X W, DENG X W, BAUMANN R, et al. A quantitative assessment of charge collection efficiency of N+ and P+ diffusion areas in terrestrial neutron environment[J]. IEEE Transactions on Nuclear Science, 2007, 54(6):2156-2161. doi: 10.1109/TNS.2007.908758 [4] 刘必慰,郝跃,陈书明. SEU加固存储单元中的多节点翻转[J]. 半导体学报,2008, 29(2):244-250. http://www.cnki.com.cn/Article/CJFDTOTAL-BDTX200802010.htm LIU Bi-Wei, HAO Yao, CHEN Shu-min. Multiple node upset in SEU hardened storage cells[J]. Journal of Semiconductors, 2008, 29(2):244-250. http://www.cnki.com.cn/Article/CJFDTOTAL-BDTX200802010.htm [5] BLACK J D, DODD P E, WARREN K M. Physics of multiple-node charge collection and impacts on single-event characterization and soft error rate prediction[J]. IEEE Transactions on Nuclear Science, 2013, 60(3):1836-1851. doi: 10.1109/TNS.2013.2260357 [6] BLACK J D, BALL D R, ROBINSON W H, et al. Characterizing SRAM single event upset in terms of single and double node charge collection[J]. IEEE Transactions on Nuclear Science, 2008, 55(6):2943-2947. doi: 10.1109/TNS.2008.2007231 [7] CALIN T, NICOLAIDIS M, VELAZCO R. Upset hardened memory design for submicron CMOS technology[J]. IEEE Transactions on Nuclear Science, 1996, 43(6):2874-2878. doi: 10.1109/23.556880 [8] FAZELI M, PATOOGHY A, MIREMADI S G, et al. Feedback redundancy:a power efficient SEU-tolerant latch design for deep sub-micron technologies[C]//37th Annual IEEE/IFIP International Conference on Dependable Systems and Networks. Edinburgh, UK:IEEE, 2007:276-285. [9] NAN H Q, CHOI K. High performance, low cost, and robust soft error tolerant latch designs for nanoscale CMOS technology[J]. IEEE Transactions on Circuits and Systems, 2012, 59(7):1445-1457. doi: 10.1109/TCSI.2011.2177135 [10] RAJAEI R, TABANDEH M, RASHIDIAN B. Single event upset immune latch circuit design using C-element[C]//2011 IEEE 9th International Conference on ASIC.[S.l]:IEEE, 2011:252-255. [11] ZHANG C Y, WANG Z S. A novel reliable SEU hardened latch to mitigate multi-node charge collecrion[C]//IET Internation Conference on Information Science and Control Engineering. Shenzhen, China:IET, 2012:1-4. [12] ALESSIO M D, OTTAVI M, LOMBARDI F. Design of a nanometric CMOS memory cell for hardening to a single event with a multiple-node upset[J]. IEEE Transactions on Device and Materials Reliability, 2014, 14(1):127-132. doi: 10.1109/TDMR.2012.2206814 [13] CASEY M C, BHUVA B L, BLACK J D, et al. Single-event tolerant latch using cascode-voltage switch gates[J]. IEEE Transactions on Nuclear Science, 2006, 53(6):3386-3391. doi: 10.1109/TNS.2006.884970 [14] WEY I C, YANG Y S, WU B C, et al. A low power-delay-product and robust Isolated-DICE based SEU-tolerant latch circuit design[J]. Microelectronics Journal, 2014(45):1-13. http://cn.bing.com/academic/profile?id=1982484741&encoded=0&v=paper_preview&mkt=zh-cn [15] NICOLAIDIS M. Time redundancy based soft-error tolerance to rescue nanometer technologies[C]//17th IEEE VLSI Test Symposium. Dana Point, CA, USA:IEEE, 1999:86-94. [16] KATSAROU K, TSIATOUHAS Y. Double node charge sharing SEU tolerant latch design[C]//IEEE 20th Internation On-Line Testing Symposium (IOLTS). Platja d'Aro, Girona:IEEE, 2014:122-127. [17] LIU B W, CHEN S M, LIANG B, et al. Temperature dependncy of charge sharing and MBU sensitivity in 130 nm CMOS technology[J]. IEEE Transactions on Nuclear Science, 2009, 56(4):2473-2479. doi: 10.1109/TNS.2009.2022267 [18] RAJAEI R, TABANDEH M, FAZELI M. Single event multiple upset (SEMU) tolerant latch designs in presence of process and temperature variations[J]. Journal of Circuits, Systems and Computers, 2015, 24(01):1550007. doi: 10.1142/S0218126615500073 [19] NAN H Q, CHOI K. Novel radiation hardened latch design considering process, voltage and temperature variations for nanoscale CMOS technology[J]. Microelectronics Reliability, 2011(51):2086-2092. http://cn.bing.com/academic/profile?id=1971440286&encoded=0&v=paper_preview&mkt=zh-cn