AutoML Tool for Automatic PSP Model Card Extractions

This paper presents an AutoML tool that automates the extraction of 59 parameters from the PSP Bulk MOSFET model. The user introduces measurements in the tool and a predicted model card that describes the capacitance, linear and saturation behavior is returned. The tool generates a simulated database, trains a deep neural network on it and then uses it in inference, streamlining the traditionally complex and iterative process of model card extraction. The core contribution of this paper, distinguishing it from previous works, lies in the validation of this approach across three distinct technology nodes, making it a versatile tool that can operate on any given technology.