Today merely a few monitoring systems for in-process detection of laser welding defects are commercially available. Despite a trend towards cameras, industrially more robust is still a photodiode, measuring in an optically filtered spectral window either the thermal emissions from the melt and vapour or the laser beam reflections. The monitoring rule for each application is identified empirically through correlations between the signal dynamics and welding defects, as the linking mechanism is non-trivial and therefore hardly understood. Thus the method does not provide a systematic guideline to detect a certain welding defect.
The here presented approach studies the context between the photodiode signal, the welding defects and the vapour, melt pool, keyhole and temperature dynamics. Simultaneous laser-illuminated high speed imaging is compared to photodiode monitoring at three spectral windows in order to identify through simultaneous timing any linking dynamics in a qualitative manner. Supportive methods are emission modelling and thermal imaging. Several cases of joints, materials and defects were studied to develop an illustrated theoretical description of the defect-signal correlations.