UC Riverside

Demand for high-performance analog-to-digital converter (ADC) integrated circuits (ICs) with optimal combined specifications of resolution, sampling rate and power consumption becomes dominant due to emerging applications in wireless communications, broad band transceivers, digital-intermediate frequency (IF) receivers and countless of digital devices. This research is dedicated to develop a pipeline ADC design methodology with minimum power dissipation, while keeping relatively high speed and high resolution.

Pipeline ADC is a mixed-signal system, which consists of sample and hold amplifier (SHA), sub-ADC, multiplying digital-to-analog Converter (MDAC) and bandgap voltage reference, comparator, switch-capacitor circuits and biasing circuits. This project set up a pipeline ADC design flow. It links all the specifications between the system levels and circuit levels together. With this design flow, if the overall ADC specifications are given, such as resolution, sampling rate, voltage supply and input signal range, all the sub-block circuitry specifications are achieved.

This paper studies all the sub-block circuits of pipeline ADC first, and then come up with all the constraints and limitations for all the circuitry in term of speed and noises. Then a system level speed and power trade off consideration is explored in order to optimize the overall performance.

As verification of the proposed design methodology, a 10-bit 40MHz pipeline analog-to-digital converter prototype is developed in commercial TSMC 90nm CMOS technology: using op-amp sharing, dynamic biasing methods, it works in two modes: pipelined ADCs for high speed, cyclic ADC for low speed (only last stage runs, other stages are power off to save power). For pipeline mode, the total power consumption decrease as the sampling frequency drops.