Fault Detection For ASIC Design Reliability On Resistive Delay Faults And Strength-Based Soft-Errors

Soft-errors (SEs) and delay faults (DFs) frequently occur in modern high-density, high-speed, low-power VLSI circuits. Therefore, SE hardened design and DF testing are essential. This thesis introduces two novel methods for soft-error detection and delay fault propagation in nanometre technology. A new idea is proposed to propagate those delay faults that are not causing logic failure at the site of the defect, but the delay makes the circuit more prone to soft-errors that manifest the effect of delay faults. This approach propagates the fault from the fault location by mapping a nine-valued voltage model on top of a five-valued voltage model to convert delay faults to static faults. This original idea reduces the complexity of delay fault propagation. This thesis introduces an original approach toward soft-error detection based on the strength violation in the circuit. This research shows that transient pulses of less than threshold voltage will cause soft-errors without altering the logic value at the strike location. This method will increase the Soft-Error Rates (SER) for all existing methods if strength-based Soft-Error detection will be considered. The offered approach uses a novel coding system that carries both logic and strength which applies to certain logic functions that are sensitive to strength variations. A wide range of soft-errors are the result of strength violation in switch-level that have never been investigated before.

Fault Detection For ASIC Design Reliability On Resistive Delay Faults And Strength-Based Soft-Errors

Soft-errors (SEs) and delay faults (DFs) frequently occur in modern high-density, high-speed, low-power VLSI circuits. Therefore, SE hardened design and DF testing are essential. This thesis introduces two novel methods for soft-error detection and delay fault propagation in nanometre technology. A new idea is proposed to propagate those delay faults that are not causing logic failure at the site of the defect, but the delay makes the circuit more prone to soft-errors that manifest the effect of delay faults. This approach propagates the fault from the fault location by mapping a nine-valued voltage model on top of a five-valued voltage model to convert delay faults to static faults. This original idea reduces the complexity of delay fault propagation. This thesis introduces an original approach toward soft-error detection based on the strength violation in the circuit. This research shows that transient pulses of less than threshold voltage will cause soft-errors without altering the logic value at the strike location. This method will increase the Soft-Error Rates (SER) for all existing methods if strength-based Soft-Error detection will be considered. The offered approach uses a novel coding system that carries both logic and strength which applies to certain logic functions that are sensitive to strength variations. A wide range of soft-errors are the result of strength violation in switch-level that have never been investigated before.

Testing and Diagnosis of Delay Faults in Finfet VLSI Circuits using Non-Incremental Genetic Algorithm

FinFet transistors are used in major semiconductor organizations which play a significant role in the development of the silicon industries. Due to few embedded memories and other circuit issues the transistors have specific faults in manufacturing, designing of the circuit etc. This paper presents an advanced test algorithm to diagnose those faults. The circuit with different gates is designed to identify the places having faults. In addition, algorithms such as non-incremental algorithms is used to find critical path, path delay and PDF of Critical path delay and Genetic Algorithm for optimisation of Critical path delay for sensitive test vector and no of iterations. The transfer characteristics curve is plotted along with the delay curve which helps in finding out the simulation parameters such as noise margin, propagation delay. The results in the methodology calculate the probability density function of the critical path by estimating mean, standard deviation and variance. The advantages of the integration of the two algorithms in this paper help in analyzing the specific faults in the circuits and the error correction of the broken link in the path analysis and has enhanced performance. Furthermore, more complicated circuits are analyzed for fault detection with different approach. In this paper the research work on testing, diagnosis, estimation of Critical path and PDF of Critical path delay faults for FinFET based Combinational Circuits for 20nm and 32 nm Technologies are presented for the first time using latest Non Incremental Genetic algorithm.