Analog-to-Digital Converters for Radio SoCs (LMSADC)
Cheongyuen (Bill) Tsang, Yun Chiu, Johan Vanderhaegen, Sebastian Hoyos, Robert W. Brodersen, Paul R. Gray and Borivoje Nikolic
In modern digital communication systems, the analog-to-digital converter is one of the key functional blocks that bridges the signaling between analog and digital worlds. High-performance radios usually dictate high dynamic range, high sampling rate, or both from these converters. With the advance of the digital IC technology, the new generation of radio system-on-a-chip exhibits less and less chip area (and power dissipation) consumed by DSP circuits while analog circuitry stays largely unaffected by scaling. The motivation is strong to reconcile approaches of implementing conventional analog functionalities in deep-submicron CMOS technologies.
This project will investigate the possibilities of enhancing ADC performance using digital signal processing techniques. Specifically, the Wiener filtering concept is applied to correct analog impairments. By means of an LMS learning algorithm, the performance of correction can be adapted to track environmental changes. With this approach, analog circuits can be made simple to leverage on high switching speed of short-channel CMOS devices, while accuracy of conversion solely relies on the back-end digital processing.
Figure 1: Chip micrograph of the ADC prototype