A 440-μW, 109.8-dB DR, 106.5-dB SNDR Discrete-Time Zoom ADC With a 20-kHz BW

Abstract

This article describes a discrete-time zoom analogto-digital converter (ADC) intended for audio applications. It uses a coarse 5-bit SAR ADC in tandem with a fine third-order delta-sigma modulator (M) to efficiently obtain a high dynamic range. To minimize its over-sampling ratio (OSR) and, thus, its digital power consumption, the modulator employs a 2-bit quantizer and a loop filter notch. In addition, an extra feed-forward path minimizes the leakage of the SAR ADC's quantization noise into the audio band. The prototype ADC occupies 0.27 mm 2 in a 0.16-μm technology. It achieves 109.8-dB DR, 106.5-dB SNDR, and 107.5-dB SNR in a 20-kHz bandwidth while dissipating 440 μW. It also achieves state-of-the-art energy efficiency, as demonstrated by a Schreier FoM of 186.4 dB and an SNDR FoM of 183.6 dB. Index Terms-A/D conversion, asynchronous SAR analogto-digital converter (ADC), audio ADC, delta-sigma ADC, discrete-time (DT) delta-sigma, dynamic zoom ADC, inverter-based operational transconductance amplifier (OTA), low-power circuits, multi-bit quantizer. I. INTRODUCTIONA UDIO applications often require analog-to-digital converters (ADCs) with high dynamic range (DR), high energy efficiency, and low area [1]-[3]. By combining a lowpower successive-approximation register (SAR) ADC with a high-resolution delta-sigma modulator (M), zoom ADCs can meet all these requirements [4], [5]. The SAR ADC determines the coarse references of the fine M, drastically reducing loop filter swing and enabling energy-efficient design. The overall digital output is then obtained by simply summing the outputs of both converters.Recently proposed Ms with finite impulse response (FIR) DACs and negative-R-assisted integrators are also capable of satisfying the requirements of audio applications [2], [3], [6], [7]. An FIR DAC essentially filters out the fed back quantization noise and, thereby, also relaxes

A 516 μW, 121.2 dB-SNDR, and 125.1 dB-DR Discrete-Time Sigma-Delta Modulator with a 20 kHz BW

“…One architecture adopts a multi-bit quantizer and input feedforward path [7]. Another is the 0-L multistage noise shaping structure, for which the first stage uses a multi-bit Nyquist ADC and the second stage uses L-order DSM [8]. On the circuit level, using switched capacitors or dynamic amplifier to implement integrators can effectively reduce power consumption [9,10].…”

A 590 µW, 106.6 dB SNDR, 24 kHz BW Continuous-Time Zoom ADC with a Noise-Shaping 4-bit SAR ADC

“…The SAR output (k) updates the references of ∆Σ DAC with 1 LSB of over-ranging such that VREF-= (k-1).VLSB < VIN < (k+2).VLSB = VREF+, where VLSB is the quantization step-size of the fine ∆Σ DAC. As in [1], the resulting 3 LSB range can then be well matched to the steps of the 2-bit ∆ΣM.…”

“…Although these interferer signals can be easily removed by using a low pass filter at the input, a sharp cut-off filter is undesirable in a CT ∆ΣM that should benefit from relaxed anti-aliasing filter requirements. For improved robustness to OOB interferers, this work uses a 4-bit SAR ADC as a coarse quantizer rather than the 5-bit SAR of previous designs [1,2,3], thereby relaxing the antioverloading filter requirements. However, this introduces more quantization noise and fuzz in the zoom ADC's output spectrum.…”

“…Zoom ADCs meet these requirements by using a low-power coarse SAR ADC to update the references of a fine delta-sigma modulator (∆ΣM). However, previous discrete-time (DT) zoom ADCs [1,2] place stringent requirements on the linearity of the input and reference drivers needed to charge their sampling capacitors with large signaldependent peak currents. The resistive input of a continuoustime (CT) zoom ADC [3] is easier to drive but a power-hungry OTA is needed to ensure that its first integrator achieves sufficiently high linearity.…”

“…Another limit on the linearity of zoom ADCs is the non-unity STF of their fine ∆ΣMs which causes in-band leakage of the SAR ADC's quantization noise (also referred to as "fuzz"). This can be mitigated, at the expense of complexity, by adding a digital filter after the SAR ADC [2], or by using a feedforward path in the fine ∆ΣM to enforce a unity STF [1]. However, the effectiveness of both techniques is limited by analog spread.…”