Memory Interface Receiver Offset Calibration via Data-Eye Convergence
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional memory interface circuits require additional switches to adjust receiver offset, leading to complex circuit architecture and potential side effects such as narrowed signal bandwidth and inaccurate adjustments.
Innovation Solution
A method and apparatus for self-calibration of receiver offset without shorting differential input terminals, utilizing a training control circuit to adjust offset control codes and reference voltage to converge data eye widths across multiple receivers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If additional switches are implemented to adjust receiver offset, then receiver offset calibration can be performed, but circuit architecture becomes more complex
Solution Approach 1:
The receiver performs self-calibration by using its own internal resources (DLL circuit, existing switches) to automatically adjust its offset without requiring external calibration equipment or additional dedicated calibration switches. The system serves itself by utilizing the data strobe signal and existing circuit components to achieve offset calibration.
Solution Approach 2:
The existing switches in the circuit are made multi-functional by using them for both normal signal routing operations and offset calibration operations. The DLL circuit is also used for dual purposes: maintaining clock synchronization and providing delay control during calibration. This eliminates the need for dedicated calibration switches.
2Measurement precision
If additional switches are implemented for offset calibration, then calibration can be performed, but signal bandwidth is narrowed
Solution Approach 1:
The calibration process uses the existing DLL circuit and data strobe signal already present in the system, avoiding the need for additional calibration switches that would introduce resistance and narrow bandwidth. The system leverages its own internal signals and circuits for calibration purposes.
Solution Approach 2:
The invention changes the calibration approach from using physical switches (which have resistance characteristics limiting bandwidth) to using controllable delay parameters through the DLL circuit. By adjusting delay values digitally rather than physically switching, the full signal bandwidth is preserved while still achieving offset calibration.
3Measurement precision
If additional switches are implemented for offset calibration, then calibration can be performed, but leakage paths are introduced making calibration less accurate
Solution Approach 1:
The receiver uses its own internal DLL circuit and data strobe signal for calibration, avoiding external or additional switches that would introduce leakage paths. The calibration is performed using signals and circuits already present in the receiver architecture.
Solution Approach 2:
The invention extracts the calibration function from the normal signal path by using the data strobe signal specifically for calibration measurements without affecting the main data signal path. This separation prevents leakage paths from interfering with either the data signal or the calibration accuracy.
Data Source
AI summary
A method and apparatus for performing self-calibration of receiver offset without shorting differential input terminals of a receiver are provided. The self-calibration includes: inputting input signals carrying predetermined data patterns into a plurality of receivers; performing data eye width measurement on the input signals received by the plurality of receivers to obtain multiple first data eye widths and multiple second data eye widths respectively corresponding to first and second data bytes; performing first offset calibration to make the multiple first data eye widths converge to a first common data eye width; performing second offset calibration to make the multiple second data eye widths be equal to the multiple first data eye widths, respectively, and converge to the first common data eye width; and performing reference voltage calibration on a reference voltage to optimize the multiple first data eye widths and the multiple second data eye widths.