Welding tasks are a frequent application and can be easily automated if the workpieces are precisely fixed. For the follow-up acquisition of the seam for inspection, profile sensors that output a height profile by means of laser triangulation are best suited. To scan the weld seam completely, the sensor or the test part must be moved. The main advantages over conventional 2D cameras are the additional height information and stable imaging despite the varying colour and surface structure of the weld seam.
With our standard software VisionTools V60, defect features such as pores, holes, binding defects and weld spatter are reliably detected from a defect size of 0.3 mm. Sensor or part can be moved at up to 100 mm/s to capture the height profiles. Speed and path must always be the same.
The seam geometry is trained on a sample part. For the evaluation, only the tolerance in the Z-direction and the minimum area of the defects as a threshold have to be entered in a user dialogue. A 3D pattern comparison is used to evaluate the data. All pixels outside the Z-tolerance are marked, then the area of the pixels is determined. A classification based on the number and size of the defect areas can also be used as a criterion for defect evaluation. Many small defects can thereby be defined as a defect criterion, as well as one large defective area.
The project software is able to manage any number of variants with different seam geometries, selectable by default via the interface.
The evaluation computer and profile sensor form a lean hardware concept. Besides a 24VDC power supply, no additional hardware like controllers are required. An additional visualisation at a further location or in the rework area is optionally possible.