Editor's Note: True or false?
Steve Scheiber, Technical Editor -- Test & Measurement World, 5/1/2004
Perhaps the biggest complaint from board-makers about AOI is the inordinate number of false calls. Increasing a system's tolerances to avoid failing good boards would allow bad boards to escape.
Many AOI systems make pass/fail decisions by comparing the current board with "known-good" examples. Such a system can easily detect whether a component is missing but not whether it is the right component or correctly positioned.
Today's crowded boards and tiny components require an AOI system with higher resolution for accurate results, so manufacturers must endure correspondingly longer test times.
Higher resolution can reduce false calls but it generally means a smaller field of view and therefore requires the system to acquire and process more images. The most time-critical step remains mechanically (and precisely) repositioning the camera. Board manufacturers need smaller, lighter cameras that can be positioned more quickly and that also have a large-image capability so they need to be repositioned less often. Fortunately, camera makers are progressing in both of those areas.
New approaches are also emerging to analyze the images. Joe Vilella, president and CEO of Vectron (San Diego, CA) heralds his parametric technique as one effective answer. The human-like recognition technique introduced by Teradyne might provide another; Landrex now owns the patents on the technology (see p.17 of the main issue).
We need AOI in the inspection/test arsenal. We also need to be able to trust its results.
Contact Steve Scheiber at sscheiber@aol.com.
