基于SET传播特性的软错误率研究

发布时间：2018-04-26 21:34

本文选题：单粒子效应 + 单粒子瞬态　；参考：《电子科技大学》2015年硕士论文

【摘要】：随着电子技术的发展,大规模集成电路被广泛应用于航空、航天领域。空间环境中存在大量辐射粒子,由这些粒子引发的软错误干扰电子设备正常工作,甚至造成设备故障。软错误主要来源于时序电路的单粒子翻转效应和组合逻辑的单粒子瞬态效应。为了评估软错误效应,一方面要考虑寄存器中出现的单粒子翻转效应,另一方面要考虑组合逻辑中产生的单粒子瞬态效应。有数据表明,当集成电路工艺技术进入纳米级以后,组合逻辑中软错误发生概率有超越存储电路的趋势。本文研究内容分为三个部分:一是研究了单粒子瞬态脉冲的展宽、缩减和滤除特性,并对瞬态脉冲传播模型进行分析验证;二是研究了组合电路的软错误分析方法并进行建模;三是选择ISCAS’85作为测试电路,对本文的模型进行分析验证,并与基于路径模型的估算结果进行对比。具体内容如下:本文第三章基于130 nm工艺条件,分析了瞬态脉冲在传播过程中的脉冲效应:展宽效应、缩减效应、滤除效应。分析了基于脉冲效应的瞬态脉冲传播模型,得到该模型估算值和Hspice仿真值平均误差为1.302%。分析了电路节点扇出、晶体管大小以及传播延时对脉冲传播特性的影响。在反相器链仿真分析中,通过Hspice仿真,验证了瞬态脉冲在传播过程中的脉冲效应临界点(Wp/Wn)在2~3之间。本文第四章对瞬态脉冲的产生概率、传播概率、捕获概率评估模型进行建模,提出了基于概率模型的软错误评估方法。在脉冲产生概率评估模型建模中,提出了基于线性模型的临界电荷估算模型,该模型是被轰击节点处驱动和扇出晶体管大小的函数。在脉冲传播概率评估模型建模中,采用查找表法来估算单个逻辑门传播概率l,l有助于评估不同类型逻辑门的软错误率,验证了随着逻辑门输入端数目的增加使脉冲传播概率降低。在脉冲捕获概率评估模型建模中,依据注入脉冲宽度和传播延时关系,对捕获到存储单元的脉冲采用脉冲传播模型来估算,将脉冲效应和锁存效应相结合。本文第五章以ISCAS’85作为测试电路,实现对本文所采用软错误评估模型的分析验证。验证了瞬态脉冲产生概率随着临界电荷的增大而减小,呈指数关系。验证了瞬态脉冲捕获概率随着寄存器捕获到的脉冲宽度的增加而增加。最终得到了所采用电路的基本信息和电路的软错误率,软错误率估算值近似在1E-05数量级,和已有经验模型估算值近似在同一数量级,验证了本文所采用模型的可行性。
[Abstract]:With the development of electronic technology, large-scale integrated circuits are widely used in aviation and aerospace fields. There are a large number of radiation particles in space environment. The soft errors caused by these particles interfere with the normal operation of electronic equipment and even cause equipment failure. The soft error mainly comes from the single particle flip effect of sequential circuit and the single particle transient effect of combinational logic. In order to evaluate the soft error effect, on the one hand, the single-particle flip effect in the register must be considered, on the other hand, the single-particle transient effect in the combinational logic should be considered. Some data show that the probability of soft error in combinational logic is higher than that of memory circuit when integrated circuit technology enters nanometer level. This paper is divided into three parts: first, the characteristics of single particle transient pulse broadening, reduction and filtering are studied, and the transient pulse propagation model is analyzed and verified; second, the soft error analysis method of combinational circuit is studied and modeled. The third is to select ISCAS'85 as the test circuit to analyze and verify the model and compare it with the estimated results based on the path model. The main contents are as follows: in the third chapter, based on the 130 nm process conditions, the pulse effects of transient pulse propagation are analyzed: broadening effect, reducing effect and filtering effect. The transient pulse propagation model based on pulse effect is analyzed, and the average error between the estimated value of the model and the simulation value of Hspice is 1.302. The effects of circuit node fan-out, transistor size and propagation delay on the pulse propagation characteristics are analyzed. In the simulation analysis of inverter chain, the critical point of pulse effect during transient pulse propagation is verified by Hspice simulation, and the critical point Wp / Wn is between 2 and 3. In the fourth chapter, the generation probability, propagation probability and capture probability evaluation model of transient pulse are modeled, and a soft error evaluation method based on probability model is proposed. The critical charge estimation model based on linear model is proposed in the model of pulse generation probability evaluation. The model is a function of the size of the driven and fan-out transistors at the bombarded nodes. In the modeling of pulse propagation probability evaluation model, the method of lookup table is used to estimate the propagation probability of a single logic gate, which is helpful to evaluate the soft error rate of different logic gates. It is verified that the pulse propagation probability decreases with the increase of the number of logic gate inputs. In the modeling of pulse acquisition probability evaluation model, according to the relation between injection pulse width and propagation delay, pulse propagation model is used to estimate the captured pulse to the memory cell, and the pulse effect and latch effect are combined. In the fifth chapter, ISCAS'85 is used as the test circuit to realize the analysis and verification of the soft error evaluation model used in this paper. It is verified that the probability of transient pulse generation decreases with the increase of critical charge and is exponentially related. It is verified that the transient pulse acquisition probability increases with the increase of the pulse width captured by the register. Finally, the basic information of the circuit and the soft error rate of the circuit are obtained. The estimated value of the soft error rate is approximately in the order of 1E-05 and approximately the same order of magnitude as the estimated value of the existing empirical model, which verifies the feasibility of the model adopted in this paper.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN47

【相似文献】