A 103.6Db-Sndr 760Mvpp-Input-range 7.8Gω-Input-impedance Direct-Digitization Sensor Readout with Pseudo-Differential Transconductors and Dummy DAC
Jianhong Zhou,Yijie Li,Kaiwen Zhou,Yuying Li,Tian Dong,Zhiliang Hong,Jiawei Xu
DOI: https://doi.org/10.1109/cicc60959.2024.10529076
2024-01-01
Abstract:High-precision and power-efficient sensor interfaces have become increasingly ubiquitous as they form an integral part of multiple fast-growing sectors such as instrumentation control, medical devices, and agricultural automation. Much progress has been recently made in direct-digitization sensor readout circuits (DD-RO). By skipping the instrumentation amplifier (IA) and connecting the ADC directly to the sensor, this architecture reduces system-level complexity compared to conventional IA plus ADC solutions. However, several challenges must be considered. First, in the absence of an IA, it requires a high dynamic range (>90dB) ADC to process the cross-scale input signals and disturbances spanning from several μV to hundreds of mV while retaining excellent sensitivity and linearity, which would usually entail excessive power dissipation. Such demanding DR can be relaxed by a spectrum-shaping $\Delta-\Delta\Sigma$ ADC [1] (Fig. 1). However, $\Delta-\Delta\Sigma$ ADCs are better suited to the spectral characteristic of bioelectrical signals and disturbances (electrode offset, power-line interference, and motion artifact), whose amplitudes decrease approximately with frequency. Second, conventional ADCs typically exhibit finite input impedance that hampers their direct interaction to high-impedance sensors due to the deterioration of total common-mode rejection ratio (T-CMRR) and gain accuracy. Prior art DD-ROs rely on a positive feedback loop (PFL) [2] or pre-charging buffers [3] for impedance boosting, but the former requires dedicated manual calibration while the latter suffers from additional noise, power, and area overhead. Alternatively, a $\Delta\Sigma$ ADC employing the $\mathrm{G}_{\mathrm{m}}-\mathrm{C}$ integrator provides high input impedance (Fig. 1 bottom left), and $\mathrm{G}_{\mathrm{m}}$ can be linearized by degeneration [2], [4]–[6] or fully differential transconductors embedded in current feedback [7] [8] (Fig. 1 bottom right). However, $\mathrm{G}_{\mathrm{m}}$ often dissipates more power to accommodate out-of-range input signals while retaining linearity.