Energy-efficient data converter design in scaled CMOS technology

Data converters bridge the physical and digital worlds. They have been the crucial building blocks in modern electronic systems, and are expected to have a growing significance in the booming era of Internet-of-Things (IoT) and 5G communications. The applications raise energy-efficiency requirements for both low-speed and high-speed converters since they are widely deployed in wireless sensor nodes and portable devices. To explore the solutions, the author worked on three directions: 1) techniques to improve the efficiency of the low-speed converters including the comparator; 2) techniques to develop high-speed data converters including the reference stabilization; 3) new architecture to improve the efficiency of the capacitance-to-digital converter (CDC). In the first part, a power-efficient 10-bit SAR ADC featured with a gain-boosted dynamic comparator is presented. In energy-constrained applications, the converter is usually supplied with low supply voltage (e.g., 0.3 V-0.5 V), which reduces the comparator pre-amplifier (pre-amp) gain and results in higher noise. A novel comparator topology with a dynamic common-gate stage is proposed to increase the pre-amplification gain, thereby reducing noise and offset. Besides, statistical estimation and loading switching techniques are combined to further improve energy efficiency. A 40-nm CMOS prototype achieves a Walden FoM of 1.5 fJ/conversion-step while operating at 100-kS/s from a 0.5-V supply. To further improve the energy-efficiency of the comparator, a novel dynamic pre-amp is proposed. By using an inverter-based input pair powered by a floating reservoir capacitor, the pre-amp realizes both current reuse and dynamic bias, thereby significantly boosting g [subscript m] /I [subscript D] and reducing noise. Moreover, it greatly reduces the influence of the input common-mode (CM) voltage on the comparator performance, including noise, offset, and delay. A prototype comparator in 180-nm achieves 46-μV input-referred noise while consuming only 1 pJ per comparison under 1.2-V supply, which represents greater than 7 times energy efficiency boost compared to that of a Strong-Arm (SA) latch. The second part of this dissertation focuses on high-speed data converter techniques. A 10-bit high-speed two-stage loop-unrolled SAR ADC is presented. To reduce the SAR logic delay and power, each bit uses a dedicated comparator to store its output and generate an asynchronous clock for the next comparison. To suppress the comparator offset mismatch induced non-linearity, a shared pre-amp are employed in the second fine stage, which is implemented by a dynamic latch to avoid static power consumption. The prototype ADC in 40-nm CMOS achieves 55-dB peak SNDR at 200-MS/s sampling rate without any calibration. A key limiting factor for the SAR ADC to simultaneously achieve high speed and high resolution is the reference ripple settling problem caused by DAC switching. Unlike prior techniques that aim to minimize the reference ripple which requires large reference buffer power or on-chip decoupling capacitance area, this work proposes a new perspective: it provides an extra path for the full-sized reference ripple to couple to the comparator but with an opposite polarity, so that the effect of the reference ripple is canceled out, thus ensuring an accurate conversion result. The prototype 10-bit 120-MS/s SAR ADC is fabricated in 40-nm CMOS process and achieves an SNDR of 55 dB with only 3 pF reference decoupling capacitor. Finally, this dissertation also presents the design of an incremental time-domain two-step CDC. Unlike the classic two-step CDC, this work replaces the OTA-based active-RC integrator with a VCO-based integrator and performs time domain (TD) ΔΣ modulation. The VCO is mostly digital and consumes low power. Featuring the infinite DC gain in phase domain and intrinsic spatial phase quantization, this TDΔΣ enables a CDC design, achieving 85-dB SQNR by having only a 4-bit quantizer, a 1st-order loop and a low OSR of 15. The prototype fabricated in 40-nm CMOS achieves a resolution of 0.29 fF while dissipating only 0.083 nJ per conversion, which improves the energy efficiency by greater than 2 times comparing to that of state-of-the-art CDCs