ADC Verification Using Quantization Noise for Fault Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing analog-to-digital converters (ADCs) in power delivery systems often fail to accurately detect faults, leading to unintended operations due to false or inaccurate data, which can result in system malfunctions and safety risks, as current error detection methods are either costly, inaccurate, or take too long to detect failures.
Innovation Solution
The proposed solution involves verifying the quantization noise properties of ADCs using noise-shaping techniques, such as delta-sigma modulators, by monitoring deviations from expected noise characteristics, employing bandpass filters to compare RMS values across frequency bands, and using threshold comparisons to determine ADC health and detect errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional error detection methods are used in ADCs, then system cost increases, but detection accuracy does not improve significantly
Solution Approach 1:
The ADC system performs self-verification by monitoring its own quantization noise characteristics. The verification module analyzes the noise properties of the ADC output without requiring external test equipment or additional sensors, enabling the system to detect its own failures autonomously
Solution Approach 2:
The invention converts the harmful quantization noise, which is normally considered a source of error and distortion, into a useful verification signal. By analyzing the statistical properties and spectral characteristics of this noise, the system can detect ADC failures without requiring separate test signals or additional hardware
2Reliability
If traditional error detection methods are used in ADCs, then system reliability decreases, but detection time increases
Solution Approach 1:
The verification module continuously monitors the quantization noise characteristics of the ADC in real-time during normal operation. This continuous verification ensures that failures are detected immediately when they occur, maintaining high system reliability without introducing detection delays
Solution Approach 2:
The system performs preliminary verification of ADC health by continuously analyzing quantization noise properties before critical failures occur. The verification process checks statistical parameters and spectral characteristics proactively, enabling early detection of degradation trends before they lead to complete system failure
3Measurement precision
If quantization noise verification is implemented, then ADC failure detection accuracy improves, but computational complexity increases
Solution Approach 1:
The verification process is segmented into distinct analytical stages: extracting quantization noise from the ADC output, computing statistical parameters (mean, variance, skewness), performing spectral analysis, and comparing results against reference thresholds. This segmentation allows each computational task to be optimized independently and implemented efficiently in hardware or firmware
Solution Approach 2:
The system implements verification at appropriate levels of detail without unnecessary complexity. Rather than performing exhaustive analysis of all signal characteristics, the verification module focuses on specific quantization noise properties (statistical moments and spectral density) that are most indicative of ADC health, achieving high detection accuracy with computationally efficient algorithms
Data Source
AI summary
Disclosed are systems and methods for identifying and reporting failures of an analog-to-digital converter (ADC). Specifically, the systems and methods described herein evaluate quantization noise properties of ADCs, including delta-sigma ADCs and successive approximation register (SAR) ADCs, to verify functionality and/or identify failures. Quantization noise properties can be evaluated in the frequency domain by, for example, comparing RMS values, magnitudes, frequency spectrums, and the like, in various frequency bands to threshold values and/or to verify an expected noise shape. Quantization noise properties can additionally or alternatively be evaluated in the time domain by, for example, comparing counts of pulse widths, average pulse widths, and/or number of transitions within a sequence of pulses to threshold values and/or to similar identifiable characteristics in other pulse width bands.