LED PCB Defect Detection
Detect manufacturing errors in complex electronic components
Graphical programming environment for deep learning-based industrial image analysis
The energy efficiency and versatility of light-emitting diode (LED) lighting have led to its increasing use in automotive, medical, and consumer products. LEDs are controlled by printed circuit boards (PCBs). These PCBs are complex, with a large number of densely packed components in a very tight geometry.
The large number of components and connections means they can suffer from a wide range of possible defects. If defects are missed, defective parts could be installed in final assemblies, resulting in poor performance or premature product failure.
The complex background presented by the board, the large number of smaller components, and the wide range of potential defects make automated visual PCB inspection with conventional machine vision difficult.
Cognex Deep Learning is designed to reliably performs multiple complex board solder inspections simultaneously. It trains on an image set for a range of PCBs that have been thoroughly tested by other means and are confirmed to be defect-free. Deep Learning can then distinguish good from bad boards on a large number of dimensions. The defect detection tool will detect open solder, bridged solder, missing components, misaligned components, and other subtle errors, many invisible even to manual inspection, and will highlight them on the image for further processing.
Whenever there is a change in board design, or if the criteria for acceptability change for any reason, the defect detection tool can quickly be retrained on a new set of confirmed good boards, without requiring any programming.