Crosspoint Switch Redundancy Using Row-Column Remapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cross point switching systems face challenges in achieving redundancy due to position sensitivity, leading to increased power dissipation, path-dependent skew, and silicon wastage, making it difficult to efficiently replicate defective cells and maintain yield.
Innovation Solution
A redundant cross point switching system that maps a redundant column/row of point cells to enable an alternate path with similar electrical characteristics, allowing for the replacement of defective cells without additional power or area, using a control circuit to remap switching devices and maintain functionality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multiple cross point switch die are packaged together to increase yield, then the yield is improved, but power dissipation increases due to intra-die I/O cells
Solution Approach 1:
The system segments the cross point switching function into multiple interchangeable die, where each die can independently implement the switching logic. This allows selective activation of die based on functional requirements, enabling power-efficient operation by activating only necessary segments rather than requiring all packaged die to operate simultaneously.
Solution Approach 2:
The redundant die are designed with universal functionality to perform the same cross point switching operations. Any die can replace another in the package, allowing the system to utilize only the functional die for active operations while keeping defective or excess die in standby or deactivated states, thereby reducing overall power consumption.
2Reliability
If multiple cross point switch die are packaged together to increase yield, then the yield is improved, but path-dependent skew increases due to connections spanning intra-die space
Solution Approach 1:
The switching system is divided into modular die segments, each with localized I/O connections. By confining signal paths within individual die or using standardized inter-die interfaces, the system reduces the variability in signal propagation delays that would otherwise occur across large intra-die spaces in monolithic designs.
Solution Approach 2:
The system changes the physical architecture parameter from monolithic large-die to multi-die packaging with controlled interconnects. This parameter change allows better control over signal path lengths and characteristics, reducing path-dependent skew while maintaining the yield benefits of multiple die.
3Reliability
If redundant cross point cells are added to replace defective cells, then yield is improved, but silicon area increases
Solution Approach 1:
The system merges the functionality of primary and redundant cross point cells into shared physical resources. The redundant die are packaged in the same package as functional die, and the control logic dynamically selects between them, effectively combining their functions without requiring separate dedicated areas for redundancy within a single die.
Solution Approach 2:
Instead of duplicating redundant cells within the same die, the system creates copies on separate die that are then packaged together. This allows the redundant functionality to be implemented using standard manufacturing processes without requiring additional area overhead on the primary die, as the redundancy is achieved through package-level integration rather than on-die duplication.
4Ease of operation
If cross point switching systems use position-sensitive cell locations, then switching functionality is achieved, but redundancy design becomes difficult
Solution Approach 1:
The system implements universal die where each cross point switch die can perform the same switching function regardless of its physical position in the package. The control circuitry dynamically assigns functional roles to different die based on their operational status, allowing any functional die to replace a defective one without being constrained by fixed position-sensitive connections.
Solution Approach 2:
The system transitions from static, position-fixed cell assignments to dynamic, reconfigurable assignments. The control logic can dynamically remap connections to use redundant die in place of defective ones, and can adapt the switching fabric configuration in real-time based on which die are functional, thereby simplifying redundancy design while maintaining position-sensitive switching functionality.
Data Source
AI summary
A redundant cross point switching is achieved by mapping a redundant column/row of point cells and enabling at least one of the switching devices which is associated with each column/row to define an alternate path around the defective point cell which replicates the function of the switching location of the defective point cell.