Second-Order DEM Segmentation for Three-Level DAC Mismatch Noise
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing multi-bit sigma-delta digital-to-analog converters (DACs) face challenges with harmonic distortions due to element mismatch, requiring higher over-sampling ratios and being inefficient with three-level unit elements, especially at low signal levels and in terms of area usage.
Innovation Solution
A 2nd-order dynamic element matching (DEM) logic that achieves highpass noise shaping with three-level unit elements, using multiple layers to distribute and match signal bits, and a dynamics enhancer to increase data activity, allowing effective noise shaping at lower over-sampling ratios and low signal levels.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If 2nd-order DEM logic is used to shape higher order noise, then noise shaping capability is improved, but circuit area increases
Solution Approach 1:
The input signal is segmented into multiple parallel processing paths (e.g., 4 paths for a 2^2 segmentation). Each path processes a portion of the signal through simplified DEM logic, achieving overall 2nd-order noise shaping without requiring a single complex circuit. This segmentation reduces the area of individual DEM units while maintaining the higher-order noise shaping capability across the entire system.
Solution Approach 2:
The patent transitions from a single-path sequential processing approach to a multi-path parallel processing architecture. By adding the dimension of parallelism with multiple processing paths, the system achieves 2nd-order noise shaping functionality distributed across several simpler units, thereby reducing the area footprint compared to a single complex 2nd-order DEM circuit.
2Measurement precision
If higher over-sampling ratios are used to improve noise shaping, then signal-to-noise ratio is improved, but processing complexity and area increase
Solution Approach 1:
The signal processing is divided into multiple parallel paths that can operate with lower individual over-sampling requirements. By segmenting the processing workload, each path achieves adequate noise shaping with reduced over-sampling ratios compared to a single high-order path, thereby reducing overall processing complexity while maintaining high signal-to-noise ratio.
Solution Approach 2:
Instead of applying excessive over-sampling to a single processing path, the patent distributes the processing across multiple paths with partial over-sampling. The combined effect of multiple paths with moderate over-sampling ratios achieves the same noise shaping performance as a single path with excessive over-sampling, thereby reducing processing complexity.
3Measurement precision
If existing 2nd-order DEM logic is used, then harmonic distortions are reduced, but it does not work with three-level unit elements
Solution Approach 1:
The patent modifies the DEM logic in each processing path to handle three-level unit elements specifically. By adapting the local processing units to accommodate three-level elements, the system achieves both harmonic distortion reduction through 2nd-order noise shaping and compatibility with three-level unit elements, resolving the limitation of existing DEM logic.
Solution Approach 2:
The DEM logic is designed to be universal and adaptable to different unit element types, particularly three-level elements. The multi-path architecture with configurable processing units can handle various element configurations, making the system versatile while maintaining the ability to reduce harmonic distortions through effective 2nd-order noise shaping.
4Measurement precision
If existing DEM techniques are used at low signal levels, then conversion is performed, but noise shaping effectiveness deteriorates
Solution Approach 1:
By segmenting the input signal into multiple parallel paths, each path receives a reduced signal level that is better suited for effective noise shaping. This segmentation prevents any single path from being overwhelmed by the full signal amplitude, thereby maintaining noise shaping effectiveness even when processing low-level signals across the entire system.
Solution Approach 2:
The patent adds the dimension of parallel processing paths to handle low signal levels effectively. Instead of relying on a single path that may struggle with low signal-to-noise ratio, multiple paths process the signal in parallel, each benefiting from improved noise shaping at their respective signal levels, thereby maintaining overall effectiveness at low input levels.
Data Source
AI summary
A method and circuit to perform noise shaped splitting of a digital input signal may include using multiple layers to process the input signal. In the first layer, the most significant bits of the input signal may be distributed to a plurality of branches. Dynamic element matching may be performed using the least significant bits of the input signal. Based on the results of the dynamic element matching, values may be added to the plurality of branches. If there is insufficient data activity, dynamics enhancement may be performed to increase the data activity. The output signals of each of the plurality of branches in the first layer may be provided to a second layer, in which these steps can be repeated on each of the output signals. The outputs of the second layer may be provided to a plurality of three level unit elements.