Auto-BET-AMS: An Automated Device and Circuit Optimization Platform to Benchmark Emerging Technologies for Performance and Variability using an Analog and Mixed-Signal Design Framework

Angada B. Sachid1,  Rajesh A. Thakker1,  Chaitanya Sathe2,  Maryam Shojaei Baghini1,  Dinesh K. Sharma1,  V. Ramgopal Rao1,  Mahesh B. Patil1
1Department of Electrical Engineering, Indian Institute of Technology Bombay, 2University of Illinois, Urbana-Champaign


Abstract

In this paper, we present Auto-BET-AMS, an automated device, circuit and system-level simulation platform suitable for benchmarking emerging technologies at the end of the CMOS roadmap. This platform is suitable for technologists and circuit designers alike. One of the features of Auto-BET-AMS is that no advanced knowledge of device and circuit design is needed to perform a fair evaluation of emerging technologies. To enable this, the platform comes with a versatile multi-variable optimizer that can be used to quickly optimize devices and circuits for a set of specifications. Using Auto-BET-AMS it is possible to accurately design digital and analog circuits and assess them in conventional and emerging technologies. The platform can handle definitions of charge-based devices in either the compact model form or a look-up table form. The latter is needed for devices which do not have mature compact models developed for them. Several representative digital and analog circuits like buffer chain, SRAM cell, two-stage Miller Op-Amp, three-stage low-voltage Op-Amp, temperature compensated current reference and Miller OTA are optimized by considering PVT variations using Auto-BET-AMS. Additionally, the impact of parametric variations on these circuits is studied.