Inference Engine Detects Stuck Components in Automated Test Sockets
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Automated test systems lack the ability to detect stuck components in test sockets, leading to repeated testing of stuck components and incorrect binning of untested components, which is exacerbated by the increasing popularity of thin device packaging formats like CSPs and InFO packages.
Innovation Solution
A method and system that utilize a handler system to place a device under test in a test socket, apply stimulus signals, and record parameters, with an inference engine classifying these parameters to determine if the test socket is stuck, generating an alert signal if necessary, using decision trees or neural networks to assess the socket status.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If automated test systems test components at high volume without detection capability, then productivity increases, but reliability deteriorates due to undetected stuck components causing repeated testing and incorrect binning
Solution Approach 1:
The system continuously monitors DUT parameters during testing and feeds this information back to the inference engine, which classifies whether the DUT is stuck. This closed-loop feedback mechanism enables real-time detection of stuck components without reducing testing throughput, resolving the contradiction between high-volume productivity and testing reliability.
Solution Approach 2:
An inference engine acts as an intermediary between the test execution subsystem and the handler system. It receives DUT parameters, classifies socket status, and generates alerts when stuck components are detected, enabling reliable detection without disrupting the high-speed automated testing flow.
2Reliability
If the system implements stuck component detection through parameter classification, then reliability improves by detecting stuck components, but device complexity increases due to the inference engine and additional processing
Solution Approach 1:
The inference engine serves multiple functions: it classifies DUT parameters to detect stuck components, determines socket status, and triggers appropriate alerts. This multi-functionality consolidates detection logic into a single component, improving reliability without proportionally increasing overall system complexity.
Solution Approach 2:
The system uses the existing DUT parameters already being collected during normal testing operations. The inference engine processes these existing parameters without requiring additional sensors or external inputs, enabling stuck component detection while minimizing added complexity.
3Productivity
If multiple components are tested simultaneously in parallel sockets, then productivity increases, but the risk of stuck components increases, worsening reliability
Solution Approach 1:
Each test socket has its DUT parameters monitored independently, with the inference engine receiving and classifying parameters from each socket separately. This per-socket feedback mechanism enables simultaneous testing of multiple components while maintaining the ability to detect stuck components in any individual socket, resolving the contradiction between parallel productivity and reliability.
Data Source
AI summary
An inference engine determines whether a device under test has become stuck in a test socket of an automated test system by analyzing measurement data sampled during normal execution of a production test program for the device. The inference engine is trained to identify stuck patterns in the measurement data for a particular device design. In the event a stuck condition is detected by the inference engine, testing by the automated test system is halted and an alert is transmitted to an operator.