An Analog Model Template Library: Simplifying Chip-Level, Mixed-Signal Design Verification

Abstract

Real number models, which are computationally efficient analog functional models, are now indispensable in verifying complex mixed-signal systems on chip (SoCs); yet, creating and validating these models remain difficult. To remove this problem, we created a framework for building analog functional model templates. Each template covers a class of circuits (e.g., oscillators or amplifiers) and can generate functional models for any implementation of this circuit class, regardless of pin configurations, circuit topologies, and process technologies/corners. This ability to cover many implementations with one template allows one to create a “standard model template” library that covers all the basic analog building blocks used in mixed-signal SoCs. Like a digital standard cell library, this template library only contains the low-level building blocks. Models for more complex systems, such as a phase-locked loop, are constructed using a number of these models. Our template library also contains validation routines for each of the models, measuring the key performance parameters of the modeled function. These routines can also be used to drive the circuit-level implementation of this function, making it easy to update the parameters of the functional models to match their circuit implementations. We demonstrate the utility of our template library on several mixed-signal IPs integrated into an SoC. Using the proposed framework, designers can easily generate a working model of an analog functional block in a second and thus construct a complete mixed-signal IP model for the design exploration in less than a couple of hours, while it typically needs a few days when the models are manually written. The regeneration of the models to reflect the behaviors of real implementations can be seamlessly done in less half a day from the same template library. In addition to the vast improvement in model creation speed, the generated model from the template library is less error-prone by archiving designer’s experience from previous mistakes to the template library.