Two-Stage ADC Reference Selection for Low-Noise Temperature Sensing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing analog-to-digital converters (ADCs) for temperature sensors, particularly those based on electrothermal filters (ETFs), face challenges with nonlinearity and quantization noise, which affect the accuracy and precision of temperature measurements in advanced deep-submicron CMOS technologies.
Innovation Solution
A two-stage digitization process using cascaded averaging converters, where a first stage performs coarse digitization with a set of references and a second stage refines the conversion using a subset of references dynamically selected based on the digital value, reducing nonlinearity and quantization noise.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single-stage ADC is used for temperature measurement, then the device complexity is low, but the measurement precision is limited due to nonlinearity and quantization noise
Solution Approach 1:
The ADC is divided into two stages: a first averaging converter that performs coarse conversion with a first set of references, and a second averaging converter that performs fine conversion with a second set of references. This segmentation allows each stage to operate with optimized reference ranges, reducing overall nonlinearity and quantization noise while maintaining manageable device complexity through modular architecture.
2Measurement precision
If a fixed set of references is used in the ADC, then the device complexity is low, but the measurement precision deteriorates due to nonlinearity across the full reference range
Solution Approach 1:
The reference set is made dynamic through a selector that chooses between a first set of references for coarse conversion and a second set of references for fine conversion. This dynamic reference selection allows the ADC to optimize linearity for different conversion stages, with the second set of references being a subset that provides finer granularity around the instantaneous analog signal value, thereby reducing nonlinearity errors.
3Measurement precision
If coarse quantization is used to reduce device complexity, then the quantization noise increases, but if fine quantization is used, the device complexity increases
Solution Approach 1:
The quantization process is segmented into two passes: coarse quantization by the first averaging converter using a broader reference set, followed by fine quantization by the second averaging converter using a narrower, optimized reference set. This two-pass segmentation reduces quantization noise effectively because the second stage operates with reduced input range requiring fewer reference levels, while the overall structure remains modular and manageable.
Data Source
AI summary
An analog-to-digital converter (ADC) is described. This ADC converts an analog signal into a digital value using a two-pass digitization process. In a first operation, coarse digitization is performed by an averaging converter based on a set of references. Then, in a second operation, fine digitization is performed by either another averaging converter or the same averaging converter based on a subset of the set of references that is progressively closer to an instantaneous value of the analog signal. For example, the coarse digitization may be performed by a low-resolution ADC stage and the fine digitization may be performed by a sigma-delta ADC, such as a single-bit sigma-delta ADC. Moreover, the other averaging converter may use dynamic element matching to shuffle reference elements used to generate the subset. In this way, the ADC may provide high resolution with reduced nonlinearity and quantization noise.