Bandpass Oversampling DAC Architecture for Wideband High Resolution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional digital-to-analog converters (DACs) face limitations in high sample rates and instantaneous bandwidth, resulting in reduced resolution and accuracy due to quantization noise, sampling jitter, and thermal noise, which restrict their ability to handle very high-frequency signals effectively.
Innovation Solution
The proposed solution involves a Multi-Channel Bandpass Oversampling (MBO) converter that uses multiple parallel processing branches with discrete-time noise-shaping/quantization circuits, multi-bit-to-variable-level signal converters, and analog bandpass filters to achieve higher resolution and bandwidth. This includes the use of multirate delta-sigma modulators, adaptive non-linear bit-mapping, and digital pre-distortion linearizers to mitigate noise and distortion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If conventional DAC architectures (resistor ladder networks or switched current sources) are used to achieve high instantaneous bandwidth, then bandwidth is improved, but resolution is degraded due to component mismatches, rounding errors, sampling jitter, and thermal noise
Solution Approach 1:
The input digital signal is divided into multiple parallel channels, each processing a subset of the signal. Each channel includes its own DAC and noise-shaping circuitry, allowing independent optimization of bandwidth and resolution for each channel while collectively achieving high overall performance
Solution Approach 2:
Digital noise-shaping circuits are introduced as intermediary components between the digital input and analog output stages. These circuits pre-process the digital signal to shape quantization noise away from the signal band, thereby improving effective resolution without sacrificing bandwidth
2Measurement precision
If Nyquist-rate conversion is used to maintain signal fidelity, then resolution can be achieved, but instantaneous bandwidth is limited to one-half the converter sample rate
Solution Approach 1:
The system dynamically allocates processing resources across multiple parallel channels operating at different rates. By distributing the conversion task across channels with different sampling rates and using noise shaping to manage quantization effects, the system achieves effective high-resolution conversion beyond the Nyquist limit of individual channels
3Measurement precision
If oversampling techniques are used to improve resolution by averaging, then conversion resolution is improved, but the bandwidth is reduced due to the low-pass filtering required
Solution Approach 1:
The signal spectrum is segmented into multiple frequency bands, with each parallel channel responsible for a specific band. Each channel performs noise-shaping and conversion optimized for its frequency range, allowing the system to achieve high resolution across the entire bandwidth without requiring aggressive low-pass filtering that would limit overall bandwidth
4Measurement precision
If high-order noise-shaping is used to achieve high resolution, then conversion resolution is improved, but the complexity of the converter architecture increases
Solution Approach 1:
The high-order noise-shaping function is segmented across multiple parallel first-order or low-order noise-shaping circuits. Each channel implements simpler noise-shaping logic, but the collective effect achieves high-order noise-shaping performance, reducing the complexity burden on any single circuit while maintaining high resolution
Solution Approach 2:
Multiple parallel conversion channels are merged at the output stage, with their individual contributions combined to produce the final high-resolution analog signal. This merging approach allows the system to achieve high resolution through parallel processing rather than requiring complex high-order noise-shaping in a single channel
Data Source
Figure 1A~1B
Figure 2A
Figure 2B
AI summary
Provided are, among other things, systems, apparatuses, methods and techniques for converting a discrete-time quantized signal into a continuous-time, continuously variable signal. An exemplary converter preferably includes: (1) multiple oversampling converters, each processing a different frequency band, operated in parallel; (2) multirate (i.e., polyphase) delta-sigma modulators (preferably second-order or higher); (3) multi-bit quantizers; (4) multi-bit-to-variable-level signal converters, such as resistor ladder networks or current source networks; (5) adaptive non-linear, bit-mapping to compensate for mismatches in the multi-bit-to-variable-level signal converters (e.g., by mimicking such mismatches and then shifting the resulting noise to a frequently range where it will be filtered out by a corresponding bandpass (reconstruction) filter); (6) multi-band (e.g., programmable noise-transfer-function response) bandpass delta-sigma modulators; and/or (7) a digital pre-distortion linearizer (DPL) for canceling noise and distortion introduced by an analog signal bandpass (reconstruction) filter bank.