Circuits and systems

A Time-Domain CNN Engine With Adaptive-Precision Computing and Threshold-Controllable Prediction for Edge Computing

A Time-Domain CNN Engine With Adaptive-Precision Computing and Threshold-Controllable Prediction for Edge Computing 150 150

Abstract:

With the growing demand for energy-efficient convolutional neural network (CNN) accelerators in edge intelligence, conventional digital CNN processors with fixed precision incur excessive switching energy and limited scalability. This work presents a time-domain CNN (TD-CNN) engine that achieves adaptive precision and computation reduction for ultralow-power operation. The main features include: 1) …

View on IEEE Xplore

A 6.8–14-GHz Ring-Based Sampling-PLL Achieving 69.3-fs Jitter Under 50-mV Supply Noise

A 6.8–14-GHz Ring-Based Sampling-PLL Achieving 69.3-fs Jitter Under 50-mV Supply Noise 150 150

Abstract:

This article presents a type-III wide-bandwidth ring-oscillator-based analog phase-locked loop (PLL) optimized for low-jitter performance in noisy supply environments. The design uses an 812.5-MHz reference frequency and a high-gain sampling phase detector to achieve a closed-loop bandwidth over 100 MHz, effectively reducing the intrinsic phase noise of the ring oscillator. To …

View on IEEE Xplore

A 10.1-ENOB 8kHz Bandwidth 95–250nW PVT-Robust DT Level-Crossing ADC for Sparse and Generic Signals

A 10.1-ENOB 8kHz Bandwidth 95–250nW PVT-Robust DT Level-Crossing ADC for Sparse and Generic Signals 150 150

Abstract:

This article presents an event-driven discrete-time level crossing analog-to-digital converter (DT-LCADC) that is energy-efficient in converting both sparse and generic signals and is robust against process voltage and temperature (PVT) variations. The proposed DT-LCADC uses the comparator delay to classify each level-crossing event as slow (produced by a small input …

View on IEEE Xplore

A 28-nm System-in-One-Macro Computing-in-Memory Chip Utilizing Leakage-Eliminated 2T1C and Capacitor-Over-Logic 1T1C eDRAM

A 28-nm System-in-One-Macro Computing-in-Memory Chip Utilizing Leakage-Eliminated 2T1C and Capacitor-Over-Logic 1T1C eDRAM 150 150

Abstract:

Computing-in-memory (CIM) is a promising paradigm for energy- and area-efficient implementation of the heavy general matrix multiplication (GEMM) operations, especially in the evolving deep learning algorithms. Though existing CIM macros have demonstrated remarkable energy/area efficiency, the corresponding metrics of the system-level CIM chips degrade due to the peripheral components, …

View on IEEE Xplore

A MEMS-Free 4096-Pixel CMOS E-Nose Array With MOF-Based Molecular Selectivity, In-Pixel Thermal Regeneration, and a Compact Single-Coefficient Bandpass Sigma–Delta ADC

A MEMS-Free 4096-Pixel CMOS E-Nose Array With MOF-Based Molecular Selectivity, In-Pixel Thermal Regeneration, and a Compact Single-Coefficient Bandpass Sigma–Delta ADC 150 150

Abstract:

This work presents a CMOS-only [micro-electromechanical systems (MEMS)-free] electronic nose (e-nose) for concurrent multi-gas-sensing applications. The proposed system integrates 4096 capacitance-to-digital converter (CDC) pixels, each implementing a compact bandpass sigma–delta ( $\Sigma \Delta $ ) ADC with a single feedback coefficient and no additional feedforward or feedback paths, achieving each pixel footprint …

View on IEEE Xplore

A 16-MHz CMOS RC Frequency Reference Achieving Accuracies of ±125 ppm From −40 °C to 85 °C and −560/+580 ppm After Accelerated Aging

A 16-MHz CMOS RC Frequency Reference Achieving Accuracies of ±125 ppm From −40 °C to 85 °C and −560/+580 ppm After Accelerated Aging 150 150

Abstract:

This article presents a low-power, high-accuracy CMOS RC frequency reference featuring a capacitively modulated RC time constant (CMT) generation and a die-to-die error removal (DDER) technique for precise frequency generation with a low-calibration cost. Unlike resistive trimming, the temperature dependence of the on-chip resistor is compensated by a $\Delta \Sigma $ …

View on IEEE Xplore

A 40k-Pixel Multimodal Biophysiology Monitoring Platform With 10k Concurrent Electrophysiology Channels and a Mixer-Embedded ΣΔ Impedance Sensor

A 40k-Pixel Multimodal Biophysiology Monitoring Platform With 10k Concurrent Electrophysiology Channels and a Mixer-Embedded ΣΔ Impedance Sensor 150 150

Abstract:

Understanding complex physiological processes requires the ability to monitor multiple biological modalities concurrently, as electrical, ionic, and biochemical processes are tightly coupled and cannot be holistically described by single-mode sensors. This article presents a 40 nm CMOS multimodal cellular physiology monitoring platform that integrates 40 960 reconfigurable pixels at $10.8~\mu $ m pixel pitch. …

View on IEEE Xplore

A 16× Interleaved 32-GS/s 8b Hybrid ADC With Self-Tracking Inter-Stage Gain Achieving 44.3-dB SFDR at 20.9-GHz Input

A 16× Interleaved 32-GS/s 8b Hybrid ADC With Self-Tracking Inter-Stage Gain Achieving 44.3-dB SFDR at 20.9-GHz Input 150 150

Abstract:

This article presents a 32-GS/s 16-channel hierarchical time-interleaved (TI) hybrid analog-to-digital converter (ADC). The prototype utilizes the intrinsic high-speed quantization of time-domain (TD) ADC to reduce the interleaving factor. By incorporating hierarchical sampling and a cascode sampler, the compact TI-ADC achieves 44.3-dB spurious-free dynamic range (SFDR) at 20.9-GHz input. …

View on IEEE Xplore

An 800-MS/s 13-b 2× TI Pipelined-SAR ADC With Rapid Digital Amplification

An 800-MS/s 13-b 2× TI Pipelined-SAR ADC With Rapid Digital Amplification 150 150

Abstract:

This work proposes a rapid digital amplification (RDA) with residue-aware reference, offering an equivalent open-loop (OL) gain enhancement of 25 dB and reducing the interstage gain error (ISGE)-induced SNR degradation by 20 dB, with an extra amplification latency of only 200 ps. It is implemented in an 800-MS/s 13-b two-way time-interleaved (…

View on IEEE Xplore