ISBN 9783843901024

978-3-8439-0102-4, Reihe Informatik

Daniel Zaum
System Level Analysis of Mixed-Signal Systems using State Space Models

151 Seiten, Dissertation Universität Hannover (2011), Hardcover, A5

Zusammenfassung / Abstract

Past years have shown that a further downsizing of microelectronic circuits alone is no longer sufficient to enable the design of innovative future devices. Today, integrating versatile functionality on a single chip has become one of the most defining trends in microelectronics. Hence, the manufacturing of digital and analog components on a single die grows more and more important. As today’s efforts to further automate the art of analog design do not yield satisfactory results, the analog parts of mixed-signal systems consume a disproportionately high amount of design and verification effort. In practice, despite the implementation of new design methodologies, their design and simulation is stuck at transistor level. The resulting skyrocketing simulation runtimes for mixed-signal designs incorporating such analog blocks pose a threat to the verifiability of future designs.

In order to achieve satisfactory performance, mixed-signal simulation has to be transfused to higher levels of abstraction. This thesis describes a simulation tool for mixed-signal systems with the capability to accelerate transient simulation compared to evaluating the analog blocks using a traditional analog simulator. The speedup is achieved by automatically precomputing input-output behavior of the analog blocks. A simulation kernel for these models is then interfaced via SystemC to digital discrete event models. The result of the precomputation are multiple numerical models completely describing the transient output behavior of the analog parts. Evaluation of these models driven by an interfaced digital simulator is a magnitude faster than a traditional analog simulator and results in an even greater speedup of the overall mixed-signal simulation.

Several practice-oriented mixed-signal example circuits are used to examine performance and accuracy of the simulation methodology. Thereby, a comparison with commercially available simulators is performed.