ADC Background Calibration Using Dither for Nonlinearity Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing calibration techniques for analog-to-digital converters (ADCs) face challenges in efficiently detecting and correcting non-linearities and non-idealities in analog circuits, leading to high power consumption and limited performance, particularly in pipelined ADCs where MDAC amplifiers dominate performance and power consumption.
Innovation Solution
An improved calibration technique using a dither signal injected into the analog circuit to expose non-linear or non-ideal behavior in the digital domain, allowing for the estimation and correction of second and third-order non-linearities through a counting-based approach independent of input distribution, without requiring analog changes to the circuit.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If existing calibration techniques are used for ADCs, then calibration can be performed, but power consumption is high and performance is limited
Solution Approach 1:
The patent replaces traditional analog calibration circuits with a digital calibration approach. A dither signal is injected into the ADC, and the output is processed digitally to extract non-linearity information. The calibration data is stored in lookup tables and applied through digital correction, eliminating the need for complex analog calibration circuitry and reducing power consumption while maintaining calibration accuracy.
Solution Approach 2:
The patent introduces a dither signal as an intermediary to expose non-linearities in the ADC. By injecting a known dither signal and analyzing the correlated components in the output, the system can accurately measure non-linearity without requiring complex calibration hardware. This intermediary signal enables precise measurement and correction while keeping the calibration system simple and low-power.
2Measurement precision
If traditional calibration methods are applied, then non-linearities can be detected, but convergence is slow and accuracy is limited
Solution Approach 1:
The patent performs preliminary action by injecting a dither signal during normal operation and continuously accumulating correlation data in registers. The calibration process begins with pre-computed lookup tables that provide initial correction values. This preliminary setup enables fast convergence because the system starts with accurate reference data and continuously refines it without requiring lengthy calibration sequences.
Solution Approach 2:
The calibration process operates continuously during normal ADC operation rather than requiring separate calibration phases. The dither signal is injected throughout operation, and correlation accumulation occurs continuously, allowing the system to constantly refine calibration data without interrupting useful work. This continuous operation dramatically reduces effective calibration time while maintaining high accuracy.
3Reliability
If analog calibration circuits are used, then calibration can be performed, but device complexity increases
Solution Approach 1:
The patent replaces complex analog calibration circuits with digital processing. Instead of using analog switches, variable gain amplifiers, and complex feedback networks, the system injects a dither signal, digitally correlates the output, and applies corrections through lookup tables. This substitution dramatically reduces circuit complexity while maintaining or improving calibration functionality.
Solution Approach 2:
The patent extracts the calibration functionality from the main signal path by using a separate dither injection path and digital correlation process. The calibration measurements are taken out from the normal signal flow and processed independently, allowing the main ADC to operate unchanged while calibration occurs in parallel. This extraction simplifies the overall device architecture by separating calibration functions from the primary conversion path.
Data Source
AI summary
Analog circuits are often non-linear, and the non-linearities can hurt performance. Designers would trade off power consumption to achieve better linearity. An efficient and effective calibration technique can address the non-linearities and reduce the overall power consumption. A dither signal injected to the analog circuit can be used to expose the non-linear behavior in the digital domain. To detect the non-linearities, a counting approach is applied to isolate non-linearities independent of the input distribution. The approach is superior to and different from other approaches in many ways.