Implementation of an OCT sensor for remote laser welding

Share this on social media:


Following his presentations at AKL and LASYS, Thibault Bautze, sales manager at Blackbird Robotersysteme, describes the benefits of using optical coherence tomography to assist remote welding applications

Remote laser welding gained popularity in the automotive industry as a versatile tool for joining applications. It allowed innovative part designs and improved the duty cycle, especially in combination with on-the-fly welding. Still, this technology requires improvement in terms of usability, profitability and functionality. The use of sensors can improve the overall welding process, such as offering coaxial seam tracking and inspection. On the application side, welding fillet seams can help reduce the weight of structures and also avoids a zinc degassing gap. But because of small deviations between parts and the need to guide the laser beam precisely onto the seam edge, pre-process edge tracking needs to be implemented.

The state of the art for quality assurance tasks is mostly based on the interpretation of in-process camera pictures, optical process emissions and post-process laser triangulation. Every method has its benefits and drawbacks, and all come with severe restrictions – laser triangulation, for instance, results in a forced welding direction and small working area, negating the large field size of a remote welding scanner and the ability to weld in any direction. Camera and photodiode-based systems may need recalibrating after small changes in the part’s quality or the process.

It has been shown that a remote laser welding scanner with an OCT-based sensor can perform quality assurance and edge tracking tasks, while avoiding some of the restrictions mentioned above. Additionally, the OCT sensor adds new capabilities to the remote welding system, such as measuring the gap between metal sheets in fillet weld configurations and adjusting welding parameters in real time.

For omni-directional use of OCT within the full working envelope of the 3D scanner, the sensor is adapted coaxially to the welding head. By using a beam splitter in the collimated beam, the laser beam and the OCT measuring light are combined. Together, they are focused by optics mounted between the beam splitter and the mirror galvanometers. This optical setup, sometimes referred to as pre-focus scanner, prevents chromatic aberrations that can occur when using f-theta optics, normally mounted after the mirror galvanometers.

The OCT sensor has its own four degrees of freedom. Two mirror galvanometers and motorised collimation are required to place the OCT focal spot anywhere around the laser focus. A further adjustable, optical reference path is required to displace the measuring field of the OCT. To enable tracking and quality assurance with full on-the-fly capability, especially for curved seams and in the z-direction, exact knowledge of the position and the movement of all axes is essential. The three axes of the welding scanner, the four axes of the OCT scanner and the six axes of the robot have to be perfectly synchronised. Also, the tracking error, acceleration and deceleration times of the galvanometers need to be considered.

The seam tracking functionality is activated within the user interface. As a first test run, the welds can be recorded without activating the edge tracking algorithm. All records are saved into a database with a timestamp, serial number (if provided) and welding parameters, and can be accessed via the software, or be exported. Edge tracking is activated if OCT height profiles can be acquired. Similarly, OCT seam inspection can be activated for every welded seam. Immediately after the weld has been made, the height profile is available through the system’s own database.

The proposed solution – a 3D scanner equipped with a coaxial OCT sensor able to acquire height profiles in the pre- and post-process areas – is a versatile tool. It combines the efficiency of a conventional remote laser welding system with the flexibility of fixed optics, being equipped with one or more optical sensors. Fillet joints can be welded in any direction and virtually within the complete working envelope of the scanner. Seam tracking is performed online without any limitation on the joint geometry.

Future challenges include robust measurement of the keyhole depth. In contrast to fixed optics, the high processing speed of scanners and quick changes of direction make it very difficult to get a stable signal from the bottom of the keyhole. For sequential pre-, in- and post-OCT scans, the keyhole depth measurement would be interrupted, complicating the interpretation of the keyhole depth signal. The trend of a spatial laser beam oscillation contradicts the use of OCT as it’s done today, as the OCT beam succumbs to the same oscillation pattern, meaning precise acquisition of height profiles cannot be achieved. New scanner and software concepts will remedy this limitation in the mid-term.

Since its first public appearance in 2013 as a tool to improve the laser welding process, the OCT technology has been tested and implemented in numerous configurations. The remote laser welding process asks for further developments until it can fully benefit from the advantages of this technology, but current limitations are known and will be addressed. As a result, the OCT-assisted remote laser welding technology integrated within one single tool may address a majority of today’s laser welding and process monitoring tasks.

Related articles 

OCT inspection to enhance automotive welding