ADC Background Calibration Without Substitute Comparators
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Analog-to-digital converters (ADCs) in small technologies face mismatches due to size and power constraints, leading to inefficiencies in correcting for aging, voltage, and temperature variations, which traditional calibration methods exacerbate with large and power-intensive switching networks.
Innovation Solution
Implement an ADC with background calibration that takes one comparator off-line for recalibration without a substitute, using neighboring comparator values or random inference to maintain accuracy, reducing the need for complex switching networks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional calibration methods are used with substitute comparators, then calibration reliability is improved, but device complexity and power consumption increase due to large switching networks
Solution Approach 1:
The patent extracts the calibration function from the normal operational path by taking one comparator off-line for calibration while the rest continue operating. This separates the calibration process from the signal conversion path, eliminating the need for complex switching networks to route signals through substitute comparators.
Solution Approach 2:
The patent creates a simplified model of comparator behavior by inferring the off-line comparator's output from neighboring comparators or by random inference. This copying approach avoids the need for physical substitute comparators and their associated switching infrastructure.
2Measurement precision
If traditional calibration methods are used with substitute comparators, then calibration accuracy is improved, but power consumption increases
Solution Approach 1:
The calibration process is extracted to occur off-line in parallel with normal operation, allowing calibration activities to consume power independently from the signal conversion path. This eliminates the need for power-intensive switching networks that would otherwise be required to route signals through substitute comparators.
Solution Approach 2:
The patent enables continuous calibration by taking comparators off-line sequentially rather than stopping the entire ADC. Multiple comparators remain operational during calibration of one comparator, ensuring continuous signal conversion while maintaining calibration accuracy through iterative refinement.
3Measurement precision
If comparators are taken off-line for calibration, then calibration precision is improved, but productivity decreases due to operational interruption
Solution Approach 1:
The ADC is segmented into multiple independent comparator units that can be calibrated individually. By dividing the calibration process into discrete segments (one comparator at a time), the system can maintain operation with the remaining comparators while calibrating each unit separately, thus preserving overall productivity.
Solution Approach 2:
The calibration process is designed to continue the useful action of signal conversion by having multiple comparators operate simultaneously during the calibration of one comparator. This parallel operation ensures that the ADC maintains its productivity while achieving high calibration precision through iterative off-line calibration cycles.
Data Source
AI summary
Analog-to-digital converters (ADCs) with background calibration processes are disclosed. In one aspect, an ADC with a plurality of comparators that each compare an input voltage to voltages that are generated at taps across a plurality of references (e.g., a reference resistor ladder). The comparators are initially calibrated with foreground calibration routines and continuously recalibrated to compensate for aging, voltage, and temperature variations without interrupting operation of the ADC by randomly taking one comparator of the plurality of comparators off-line to run calibration processes without replacing that comparator. The value for the off-line comparator may be reliably inferred from values from neighboring comparators or, in some cases, guessed randomly. While possible errors may be introduced, such errors may be driven to a mean square quantization noise level through exemplary aspects of the present disclosure.