Voids in solder on printed circuit boards are a problem. Where there is an air gap, there is no conduction locally, which leads to hotspots on the PCB, reducing performance or even causing failure.
The human eye is good at picking out voids in grey-scale absorption images, but this is hard to automate, especially where the background intensity is non-uniform. Edge-detection algorithms don’t work well on intensity gradients.
IBEX has succeeded in automatically detecting voids of various sizes and shapes on PCBs. In the example shown, the metal pad is about 700 µm wide. The smallest void detected is about 10 µm across. A void on the join between the component and the pad and voids on the intensity gradient have been clearly identified.
Voids may then be classified according to various criteria, such as total void area, largest single void or number/area of voids under the solder pad. This technique offers the possibility of automating the void detection process in a wide variety of applications - electronics inspection and beyond.
The data were collected at 120 kV, 3.9 W, with a total image acquisition time of 320 ms. The copper tracks are 35 µm thick (1 oz).