Deterministic Vector Clustering for IC Test Data Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing test compression schemes for Integrated Circuits face inefficiencies due to limited encoding capability, high test data volumes, and long test application times, particularly when dealing with complex and diverse designs, as they struggle to effectively handle conflicting test cubes and require additional memory and resources.
Innovation Solution
The approach involves deterministic vector clustering of incompatible test cubes, forming a parent pattern with incremental and control patterns to encode test data efficiently, allowing for a tri-modal decompressor to load test data into scan chains, reducing the number of specified bits to be encoded and optimizing the structure for compression.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If conventional test compression schemes are used, then test data volume is reduced, but encoding capability is limited and additional memory resources are required
Solution Approach 1:
The patent merges incompatible test cubes by forming parent patterns that can accommodate multiple conflicting test cubes simultaneously. This merging approach allows the system to handle test data with conflicts without requiring additional memory resources, as the parent pattern structure inherently manages the conflicts through its design with incremental and control patterns.
Solution Approach 2:
The parent pattern structure serves multiple functions: it stores the common portion of test cubes, manages incremental patterns for conflicting bits, and controls the decomposition process. This multi-functional design eliminates the need for separate memory structures that would otherwise be required to handle different aspects of test data compression independently.
2Productivity
If more test cubes are merged, then compression ratio improves, but conflicts between test cubes increase
Solution Approach 1:
The patent segments test cubes into parent patterns and incremental patterns. The parent pattern contains the common portion of merged test cubes, while incremental patterns handle the conflicting bits. This segmentation allows multiple test cubes to be merged (improving compression ratio) while conflicts are systematically managed through the incremental pattern mechanism rather than accumulating as harmful interference.
Solution Approach 2:
The control pattern acts as an intermediary that manages the relationship between parent patterns and incremental patterns. It stores information about which bits are incremental and which are specified, thereby mediating the conflicts that arise when merging multiple test cubes. This intermediary structure enables high compression ratios while systematically resolving conflicts without requiring additional memory resources.
3Quantity of substance
If LFSR coding is used, then test patterns are compressed, but encoding capability is limited by LFSR size
Solution Approach 1:
The patent transitions from the traditional LFSR sequential generation approach to a parent-pattern-based structure that organizes test data in a different dimensional framework. Instead of relying on LFSR size to determine encoding capability, the system uses parent patterns that can accommodate multiple test cubes simultaneously, effectively adding a new dimension to the compression approach that is not constrained by LFSR register size.
4Ease of manufacture
If static LFSR reseeding is used, then encoding is performed, but loading and unloading scan chains requires two non-overlapping phases
Solution Approach 1:
The patent enables continuous operation by allowing the loading of parent patterns and incremental patterns to overlap with the scanning process. Unlike static LFSR reseeding that requires sequential phases, the parent pattern structure allows test data to be loaded and scanned simultaneously through the decompression mechanism, eliminating idle time and maintaining continuous useful action throughout the test application process.
Data Source
AI summary
The test data compression scheme is based on deterministic vector clustering. Test cubes that feature many similar specified bits are merged into a parent pattern in the presence of conflicts. The parent pattern along with a control pattern and incremental patterns representing conflicting bits are encoded efficiently. A tri-modal decompressor may be used to decompress the test data.