Low energy digital circuit design using sub-threshold operation

Abstract

Scaling of process technologies to deep sub-micron dimensions has made power management a significant concern for circuit designers. For emerging low power applications such as distributed micro-sensor networks or medical applications, low energy operation is the primary concern instead of speed, with the eventual goal of harvesting energy from the environment. Sub-threshold operation offers a promising solution for ultra-low-energy applications because it often achieves the minimum energy per operation. While initial explorations into sub-threshold circuits demonstrate its promise, sub-threshold circuit design remains in its infancy. This thesis makes several contributions that make sub-threshold design more accessible to circuit designers. First, a model for energy consumption in sub-threshold provides an analytical solution for the optimum VDD to minimize energy. Fitting this model to a generic circuit allows easy estimation of the impact of processing and environmental parameters on the minimum energy point. Second, analysis of device sizing for sub-threshold circuits shows the trade-offs between sizing for minimum energy and for minimum voltage operation.

(cont.) A programmable FIR filter test chip fabricated in 0.18pum bulk CMOS provides measurements to confirm the model and the sizing analysis. Third, a low-overhead method for integrating sub-threshold operation with high performance applications extends dynamic voltage scaling across orders of magnitude of frequency and provides energy scalability down to the minimum energy point. A 90nm bulk CMOS test chip confirms the range of operation for ultra-dynamic voltage scaling. Finally, sub-threshold operation is extended to memories. Analysis of traditional SRAM bitcells and architectures leads to development of a new bitcell for robust sub-threshold SRAM operation. The sub-threshold SRAM is analyzed experimentally in a 65nm bulk CMOS test chip.