Residue Amplifier Nonlinearity Correction in Pipelined ADC Stages
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Pipelined analog-to-digital converters (ADCs) face challenges in compensating for non-linear gain errors, which lead to non-linearity in transfer characteristics, requiring complex algorithms and large silicon areas and power consumption, making them impractical for high-speed communication applications.
Innovation Solution
A piecewise linear representation is used to model the non-linearity of the gain in residual amplifiers, segmenting the amplifier into linear operating regions and adjusting digital codes with calibrated gain factors to correct for non-linearities, reducing gain errors efficiently.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If complex algorithms are used to correct non-linearity in residual amplifiers, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent divides the non-linear gain correction into multiple linear segments using piecewise linear representation. Instead of using a single complex non-linear correction algorithm, the transfer function is segmented into multiple linear regions, each with its own correction parameters. This segmentation approach maintains correction precision while significantly reducing computational complexity.
Solution Approach 2:
The patent transforms the non-linear gain correction problem into a parameter adjustment problem by calibrating digital codes with gain factors. The non-linearity is corrected by changing the parameters (gain factors) associated with different input ranges rather than applying complex real-time non-linear calculations, thereby simplifying the device complexity.
2Measurement precision
If complex correction algorithms are implemented, then measurement precision is improved, but power consumption increases
Solution Approach 1:
By segmenting the non-linear correction into piecewise linear segments, the computational burden is reduced. Each segment requires simpler linear correction calculations rather than complex non-linear algorithms, leading to lower power consumption while maintaining acceptable precision across the full input range.
Solution Approach 2:
The patent uses pre-calculated lookup tables and fixed gain factors stored in memory rather than performing complex real-time calculations. This approach trades minimal storage space for significant reductions in computational power requirements, effectively using 'cheap' stored data instead of 'expensive' computational resources.
3Measurement precision
If complex correction algorithms are used, then measurement precision is improved, but silicon area increases
Solution Approach 1:
The piecewise linear segmentation approach requires storing correction parameters for multiple segments rather than implementing complex correction logic. This trades minimal additional memory space for significant reductions in logic circuit area, as the correction is achieved through simple parameter lookup and application rather than complex computational blocks.
Solution Approach 2:
The patent replaces complex computational mechanics with simpler data lookup and parameter application mechanics. Instead of using large logic circuits to perform real-time non-linear calculations, the system uses pre-computed correction data stored in memory, substituting computational hardware with storage hardware which is more area-efficient.
Data Source
AI summary
In a method and apparatus for compensating non-linearity of a gain of a residual amplifier (RA), a pipelined analog-to-digital converter (ADC) converts an analog input to a digital output (DO). The ADC includes a plurality of pipelined stages (PPS). Each stage, which includes an instance of the RA, provides a digital code corresponding to an output of the RA included in a preceding stage. A memory stores a piecewise linear representation for modeling the non-linearity of the gain. A calibrated gain of the RA corresponding to each region of a plurality of linear operating regions of the RA is stored in the memory. A gain adjuster adjusts the digital code for each one of the PPS in accordance with a gain factor derived from the calibrated gain for each one of the PPS. A constructor constructs the DO from the adjusted digital code received from each one of the PPS.