Regression Model-Based AMS Circuit Optimization Technique Utilizing Parameterized Operating Condition

An analog and mixed-signal (AMS) circuit that draws on machine learning while using a regression model differs in terms of the design compared to more sophisticated designs. Technology structures are advanced than conventional CMOS processes, specifically fin field-effect transistor (FinFET) silicon-on-insulator (SOI), have been proposed provide higher computation performance required meet various specifications. As result, latest research AMS optimization has enabled enormous resource savings procedures but remains limited with regard reflecting intended operating conditions parameters. Hereby, we propose what is termed supervised artificial neural network (ANN) as means by which define an model. This approach allows for rapid searches complex dimensions, including variations metrics caused process–voltage–temperature (PVT) changes. The method also reduces considerable expense simulation-program-with-integrated-circuit-emphasis (SPICE) simulations. Hence, flow generates highly promising output target requirements showing excellent ability match performance. To verify potential promise our flow, successive approximation register analog-to-digital converter (SAR ADC) designed 14 nm process kit. In order show maximum single ADC (6-bit∼8-bit resolution few GS/s conversion speed), set three different targets. Under all SS/TT/FF corners, ±6.25% supply voltage variation, temperature variation from −40 ∘C 80 ∘C, SAR yields remarkable figure-of-merit energy efficiency (approximately 15 fJ/conversion step).