10.7 A Single-Channel 70Db-Sndr 100Mhz-Bw 4Th-Order Noise-Shaping Pipeline SAR ADC with Residue Amplifier Error Shaping

Yanbo Zhang,Junyan Hao,Shubin Liu,Zhangming Zhu,Yan Zhu,Chi-hang Chan,R. P. Martins
DOI: https://doi.org/10.1109/isscc42615.2023.10067689
2023-01-01
Abstract:Emerging wireless standards demand ADCs with hundred megahertz (MHz) bandwidth (BW) with an inband dynamic range (DR) better than $60\text{dB}$. Continuous-time sigma-delta modulators can accommodate a high DR under low OSR for a wideband target [1]. Nevertheless, power-hungry integrators limit the energy efficiency of CTSDMs with a best reported $\sim 166\text{dB}$ Schreier $\text{FoM}(\text{FoM}_{\mathrm{s}})$, while suffering from the STF peaking issue. Noise-shaping SAR (NS-SAR) ADCs are free from the aforementioned constraints, and well-suited to multi-standard wireless communication systems with an energy-efficient, fully passive integrator and an easily configurable architecture. Furthermore by timeInterleaving the NS-SAR (TINS-SAR), the BW boosts to $80\text{MHz}$ with $\sim 66\text{dB}$ SNDR, maintaining an outstanding efficiency of 171.2 $\text{FoM}_{\mathrm{s}}$ [2]. However, the NS order and efficiency are difficult to enhance due to the lack of active summation nodes, limiting the achievable SNR and DR. Moreover, the necessitated coarse-fine quantization and midway feedback scheme $[3] \text{prevent a high interleaving factor in the TINS}$ -SAR ADC, thus holding back the BW from a further push. Rather than TI, hybridizing the pipeline with NS-SAR (NS-PiSAR) also can extend the BW while simultaneously enabling a high NSefficiency [4]. Previous arts demonstrate a $1^{\text{st}}$ -order NS-PiSAR ADC with $\mathrm{a}40\text{MHz}$ BW and $\sim 75\text{dB}$ SNDR, but requiring a background-calibrated residue amplifier (RA). When pursuing hundred MHz BW, the impairments from the RA are a bottleneck. This work presents a single-channel NS-PiSAR ADC with effective $4^{\text{th}}$ -order NS, achieving a peak SNDR of 70. $15\text{dB}$ over $100\text{MHz}$ BW. With a one-time-only trimming, the ADC is robust under VT variations.
What problem does this paper attempt to address?