Transient simulation is a main challenge to achieve system level ESD failure prediction. During the turn-on of the protections, complex phenomena introduce complex transient behaviors. In this paper we investigate the parameters that have to be added to perform accurate transient simulations and we propose a methodology to extract them by measurements.

ElectroStatic discharge (ESD) protection devices have non-linear and complex behavior that makes system level design robustness predictions complex. To obtain a precise simulation, a model reproducing the turn-on behavior is needed.In this paper we propose a complete measurement and computation setup to get access to an equivalent frequency model of devices under strong pulse injection. To validate our proposed frequency model a comparison of measurements and simulations is performed on passive and linear components first, then on a protection device (TVS).

In order to ensure reliability of systems early in the design phase, it is becoming crucial to have models able to predict the behavior of systems exposed to ElectroStatic Discharge (ESD). This is an increasing necessity since the number of embedded electronic products is growing and since they are employed in applications where people's safety is a requirement, such as automotive and aeronautic applications. Until now, quasi-static models of protection devices have succeeded in providing fairly good results in failure predictions (mainly hard failures). Today, the increased frequency range of such devices requires dynamic models able to reproduce their transient behavior. In this paper, we investigate if conventional methods for modeling linear devices, generally used in the frequency domain, could be used to obtain an equivalent frequency model for ESD protection devices, which exhibit non-linear behavior. A methodology to extract an ESD protection SPICE model from Transmission Line Pulse (TLP) measurements to address transient and frequency simulation is proposed and detailed. We demonstrate that, in well-defined conditions, such frequency models can give accurate results to predict overshoots related to protection device triggering delays. Validation of the models is performed under TLP and Human Metal Model (HMM) conditions on three off-the-shelf devices.

No abstract