Characterize and Quantify the Production Inspection Capability of the AXI of HiP (Head in Pillow) Defects


Project Leaders:


Robin Hou, IBM




Herb Holmes, Intel

 

Recent Presentations 

End-of-Project Report: Characterize and Quantify the Production Inspection Capability of the AXI of HiP (Head in Pillow) Defects Project (June 9 & 10, 2019)

Statement of Work and Project Statement

Background

HiP (Head in Pillow) solder joint defects have a greater risk of occurrence with increasing BGA package size (modules/CPU sockets), thicker boards (increased layers) and non-complimentary warpage characteristics.

In addition, HiP defect Detection with in-line AXI (Automatic X-ray Inspection) equipment is a high risk FMEA (Failure Modes and Effects Analysis) element where algorithms used are not proven to be robust enough leading to:

  • High probability of false passes (algorithm based)
  • High probability of false passes (operator over rides) – Training/Shade of Gray scale resolution and defect determination.

Current algorithms are based on simplistic models (supplied by X-ray equipment manufacturers) and are composed on differently generated X-ray images such as Transmissive, Laminography, and Tomosysnthesis techniques. This leads to correlation inconsistencies and subsequent low Agreement Analysis (kappa value less than 0.7).

Limited electrical coverage test (ICT/FCT) post AXI in process becomes a reliability concern when released to the field for (e.g., enterprise server systems, other high-reliability applications). In addition, there is a lack of “golden” sample with various HiP defects for calibration and correlation.  As a result, a need exists to quantify and correlate HiP defects across all AXI equipments’ algorithms and levels of HiP conditions.

Purpose of Project

A pre-competitive investigation to characterize and quantify production inspection capability of the AXI of HiP (Head in Pillow) defects.

Past Presentations


For Additional Information

Mark Schaffer