Single event transient sensitivity analysis of different 32 nm CMOS majority voters designs

Impact of BTI and HCI on the reliability of a Majority Voter

“…Since TMR systems are used for applications in harsh environments where circuits are more exposed to radiation, the sensitivity of majority voters to SETs has been a topic of interest in the literature [3]- [5]. The impact of an energetic particle on a circuit node can be modeled with a double-exponential current pulse [10]: Fig.…”

“…It is important to consider which nodes in the circuit are relevant depending on whether there is a logical sensitized path between the node and the output [5]. Considering the nodes in the schematic of Fig.…”

“…Types of failures, defined as deviations of the expected service of a component, can be generally divided into four groups [7]: [5], which are computational, soft errors that do not cause permanent degradation of parameters. A clear example is a single-event upset (SEU): when an energized particle hits a sensitive area of a circuit, it may produce a voltage glitch, a socalled single-event transient (SET).…”

Design exploration of majority voter architectures based on the signal probability for TMR strategy optimization in space applications

“…However, the majority voter robustness against radiation effects is crucial for the fault detection and correction efficiency of TMR strategies. Few works in the literature have provided design studies on the radiation robustness of majority voters in terms of power, delay, area and SET cross-section [5][6][7][8]. The work developed in [5] provides heavy-ion experimental data and proposes a relative efficiency criterion for the majority voter selection according to the TMR strategy.…”

“…In this case, the NAND-and NOR-based voter circuits have shown a similar soft-error rate (SER) due to the symmetric sizing of the PFET and NFET devices provided by the strain engineering and width quantization at FinFET technology [7]. The work in [8] adopts a layout-level analysis through stick diagram to evaluate the diffusion area of majority voters and to calculate a fault masking ratio. Despite the layoutbased approach, charge sharing between internal nodes and very importantly the intra-cell charge sharing effect in the multi-level circuit implementations are not considered, which can considerably increase the voter sensitivity or reduce it through pulse quenching effect [9].…”