Multi-Mode ADC Architecture Using Bandpass Subband Oversampling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional analog-to-digital converters (ADCs) face limitations in multi-mode applications, particularly in achieving high-resolution conversions for both narrowband and wideband signals, due to limitations in input bandwidth and effective conversion resolution, which are constrained by practical implementation impairments such as sampling jitter and thermal noise.
Innovation Solution
The implementation of a Multi-Channel Bandpass Oversampling (MBO) technique that decomposes continuous-time signals into distinct frequency subbands, independently processes each subband, and combines them to preserve bandwidth, using continuous-time quantization-noise-shaping circuits, sampling/quantization circuits, and digital bandpass filters to enhance resolution and reduce noise sensitivity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional ADC approaches are used, then narrowband signals can be converted with high precision, but wideband signals are limited to moderate precision due to sampling jitter and thermal noise
Solution Approach 1:
The patent divides the wideband input signal into multiple narrowband frequency subbands using bandpass filters. Each subband is then processed independently by a dedicated noise-shaping ADC, allowing high-resolution conversion for each narrowband segment while collectively covering the wide input bandwidth. This segmentation resolves the contradiction by enabling both high precision (through narrowband processing) and wide bandwidth (through parallel subband handling).
Solution Approach 2:
The patent transitions from single-channel time-domain processing to multi-channel frequency-domain processing. By decomposing the signal into parallel frequency subbands and processing them simultaneously, the system achieves high resolution for each subband while maintaining wide overall bandwidth, effectively adding a frequency dimension to the conversion process.
2Measurement precision
If high-resolution conversion is implemented, then quantization noise is reduced, but input bandwidth is limited to a few GHz due to sampling jitter
Solution Approach 1:
The patent segments the wideband signal into multiple narrowband subbands, each processed by an independent noise-shaping converter. Since each converter handles a narrow bandwidth, sampling jitter has minimal impact on resolution, enabling high effective bit conversion for each subband while the aggregate system maintains wide input bandwidth.
Solution Approach 2:
The patent converts the harmful effect of sampling jitter into a benefit by using noise-shaping techniques that push quantization noise and jitter-related errors out of the signal band and into higher frequencies, where they can be filtered out. This allows high-resolution conversion despite the presence of sampling jitter.
3Measurement precision
If oversampling is used to improve resolution, then quantization noise is reduced, but the converter complexity increases
Solution Approach 1:
The patent uses parallel oversampling in multiple independent noise-shaping converters, each handling a narrow subband. This distributed oversampling approach achieves high resolution through the combined output of multiple converters, balancing the complexity increase with improved performance and noise rejection.
Solution Approach 2:
The patent employs feedback-based noise-shaping in each parallel converter, where the quantization error is fed back through a noise-shaping filter and subtracted from the input. This feedback mechanism shapes the quantization noise spectrum, pushing noise out of the signal band and improving effective resolution without requiring extreme oversampling ratios.
Data Source
Figure 1A~1B
Figure 2A~2B
Figure 3
AI summary
Provided are, among other things, systems, methods and techniques for converting a continuous-time, continuously variable signal into a sampled and quantized signal. According to one implementation, an apparatus includes multiple processing branches, each including: a continuous-time quantization-noise-shaping circuit, a sampling/quantization circuit, and a digital bandpass filter. A combining circuit then combines signals at the processing branch outputs into a final output signal. The continuous-time quantization-noise-shaping circuits include adjustable circuit components for changing their quantization-noise frequency-response minimum, and the digital bandpass filters include adjustable parameters for changing their frequency passbands.