Design of continuous-time delta sigma converters for high fidelity audio applications
Leow, Yoon Hwee
Date of Issue2016-10-27
School of Electrical and Electronic Engineering
Audio signal processing is one system that demands high resolution A/D converters. The healthy human ear can decipher sound with intensities varying from 0~3dB to more than 160dB (Sound Pressure Level) SPL with a theoretical frequency bandwidth of up to 20~25kHz. Traditional audio signal processing channels rely on standalone modules where each component is optimized in their own chosen technology. Consequentially, high fidelity but huge products result. The luxury of having a higher supply voltage for the noise sensitive analog circuitries whereas a lower supply voltage in speed demanding digital processing circuits addressed the ADC resolution. Recent audio codec systems present a set of different requirements. Heavy Digital Signal Processing (DSP) fueled with cost and aesthetics of the final product demand a consolidation of components along the audio signal processing channel. The Continuous-Time (CT) Δ∑ modulator offers a number of advantages over its discrete counterpart in some fields of application. Traditionally, the DT Δ∑ modulator has dominated this market. The moderately low bandwidth requirement (i.e. 20 kHz–25 kHz) means the design of OTAs with a reasonable speed consuming an acceptable power can still be achieved. In order to maximize the suppression of quantization noise, ADC designers resort usually to either high order (4-6) or multibit feedback DAC structures; not to mention the usage of high OSR of above 64. However in the advent of nanoscale technology, the former apparent choice of DT architectures becomes less justifiable. Feature size down scaling with technology advancements gives rise to weaker devices that can only tolerate low supply voltages. The input dynamic range of the frontend Sample and Hold (S/H) ADC buffer has to be maximized under the constrained VDD supply while maintaining linearity. This warrants the use of more complicated switching control techniques such as boot strapping. On the other hand, the inputs to a CT Δ∑ modulator can be built without a S/H buffer with the possibility of reaching high input dynamic ranges via linear resistors of a opamp RC integrator. This generates interest in the research for high resolution ADCs for high fidelity audio applications in advanced nanoscale technologies. This work researches into the design of CT Δ∑ modulators targeted for high fidelity audio applications. The literature review chapter presents an evolution of oversampling CT Δ∑ modulators in this field of area. Subsequently, high level detailed modelling will be discussed leading to the development of a circuit-based model for rapid verification that shorten optimization turn-around time. Two CT Δ∑ modulators with different architectural approaches are presented. Chapter 5 described the realization of a 4th order CT Δ∑ modulator. System level non-idealities such as clock jitter uncertainties and process variation tolerance were discussed and relaxed with proposed circuitries. This modulator, implemented in 180nm technology with a supply of 1.5V was fully verified in simulation and achieved an SNR/SNDR of 90.6dB/88.7dB with a reasonable input dynamic range of differential peak-to-peak of 2V while consuming a current consumption of only 153µA. Chapter 6 proposed a third order audio CT Δ∑ modulator developed in advance 65nm standard CMOS technology. To enhance the NTF, Noise Coupling (NC) was applied in a second order CIFF architecture. The input dynamic range achieved a differential peak-to-peak voltage of 2V even at the ultra-low supply of 1V considering such a high resolution design target. The modulator achieved a measured DR/SNR/SNDR of 103.1dB/100.1dB/95.2dB while dissipating 0.8mW.