Over the past 20 years, remote laser welding has grown to be a widely used process in day-to-day manufacturing across many different markets. The versatility and speed it provides compared to other joining processes warrants its use in numerous applications, most of which are high volume in nature.
As products and applications evolve, the materials and joint requirements for laser welding have gone from standard overlap style configurations in steel to those using more exotic materials with discrete weld placement requirements. Additionally, the parts vary more in shape and size, greatly increasing the complexity of the process.
A major growth point for the application of laser welding falls in line with the evolution of electric vehicles and their rise in popularity. The challenge in the production of these vehicles, namely the battery packs and motors, is that they use materials and joint configurations often not readily weldable with standard technologies.
The evolution of weld depth control, laser sources and integrated process monitoring has greatly helped. However, there are still gaps in today’s technology to address these roadblocks. Implementation of features such as seam tracking integrated to the optic have helped, but it very likely will take more than just seam tracking to solve most of these issues.
While the standard laser triangulation-based seam tracking provides a great deal of benefit, certain parts exceed the capability of this technology due to their shape, stack-up and orientation. In these instances, shape recognition serves as the next level of technology that can overcome previous technology shortcomings.
Shape recognition is a technology that uses a camera-based approach where pixilation, the discernment of light and dark portions of the image, and the use of advanced machine algorithms are used for dictating not only part shape, but location of the joint in multiple axes. The technology was developed to help address the complexities of welding the copper hairpins used in electric motors.
A growing number of manufacturers are using copper hairpins instead of copper winding in the production of stators, shown in Figure 1. The heads of these hairpin-shaped parts are inserted into the stator and welded together by a laser beam. Each stator has 160 to 220 hairpins that are processed in a time window ranging from 60 sec. to 120 sec. This technology reduces the overall size of electric motors and improves performance.
The biggest challenge thus far has been to reliably detect the position of the hairpins. This is because mechanical preprocessing of the hairpin surface results in different levels of reflectivity, which makes image processing more complicated. The laser welding process is extremely demanding because copper has very low absorption at room temperature. During the welding process, absorption and temperature increase sharply along with the high thermal conductivity and low molten bath viscosity of the material.
In Figure 2, a depiction of the process screen is shown wherein a stator with hairpins is processed. The blue outline of the pin is the user-defined contour to search and find, where the red lines depict the user-defined welding contour.
Scansonic is a global market leader in laser-based joining systems for body construction. The company’s optics, sensors and process monitoring systems are primarily used in the automotive industry, rail vehicle construction and energy technologies. Since 2008, Abicor Binzel has provided global sales, service and support for the Scansonic product line around the world. New Scansonic developments include the RLW-S laser optic that features the company’s Integrated Shape Recognition, which is already being used successfully in production by automotive suppliers.
Integrated Shape Recognition offers a number of key advantages. The system uses laser scanners to precisely guide the laser beam to within 0.1 mm. This is made possible by an integrated camera that accurately detects the processing point and enables the welding to be completed according to a predetermined pattern. The camera also ensures continuous process monitoring and high process reliability. This gives customers maximum dependability with strong, pore-free seams at high production rates.
The main capabilities of this optic are derived from the two main components of the system, namely the scanners and the camera system utilized in the optics package, shown in Figure 3. The system utilizes a high-end CMOS digital camera with up to 100 frames per second processing in a field of view of 40 mm (X) by 33 mm (Y). The RLW-S with Integrated Shape Recognition system uses standard 2-in. optics to keep system operation costs down as it utilizes standard cover glasses and requires a fraction of the air to support cross jet functions that is typically seen with F-Theta lenses used in other remote offerings.
One of the key benefits of the system is tied to its ability to use the camera system for automatic finding of shapes and related control of scanner positioning. Being that the camera operates coaxially with the laser beam, the beam will follow the same path of the camera. When a position is detected, the scanner moves to accurately detect the position. Additionally, the system adjusts automatically to height variations through an integrated autofocus approach where adjustments in the collimation lens position are made based on data received from the process.
This is all done in conjunction with an integrated illumination package, which provides a homogenous bright illumination from four bench lights (64 W per bench, 256 W total) located at 90-degree intervals. This illumination requires no safety measures. The system then interprets the bright field and dark field portions of the part to discern the part shape. This lends itself to do parts other than hairpins on electric motors, such as heat exchangers (tube in plate) or plug welds.
In the case of heat exchangers, the system’s multi-shape recognition capabilities ensure that every tube in the camera’s field of view is recognized and welded individually. Extrapolation of position is not required, and, therefore, numerical position errors do not occur. As seen in Figure 4, the system processes all parts noted in the field of view, taking additional pictures after each individual weld to accommodate possible process variations. From there, it moves on to process the next group of welds in the adjoining field of view.
The optical ratio was optimized at 1:2.7, providing a process freedom that would offer an in-focus spot size of 270 microns with 1,000-micron fiber and 455 microns with 200-micron fiber with an extremely low focal shift between 500 W and 2,000 W.
Integrated Shape Recognition is suitable to use with disc laser (1,030 nm), rod laser (1,064 nm) and fiber laser (1,070 nm) generators with a maximum power of 8 kW. It has a 400-mm focal length and utilizes various fieldbus types for process communication between it and the robot/controls system. To achieve maximum accuracy, the system utilizes scan frequencies up to 1,000 Hz, which provide highly dynamic reorientation capabilities at speeds up to 2 m per sec.
The high-speed, high-accuracy scanners open up new process possibilities such as optimized heat input and increased productivity. Figure 5 shows some of the dynamic capabilities of producing spiral and square shaped welds. The hardware in combination with internal controls allows the scanners to precisely guide the laser beam to within 0.1 mm. This is especially evident in the ability to change direction on the edges of the 4-mm square along with the tight concentricity of the spiral shape.
Based on the ever-changing needs of the market, Integrated Shape Recognition is poised to address some of the more demanding processes where high-precision, high-quality laser welding is needed. With the integrated process controls for adaptive welding, best-in-class scanner speeds and camera-based shape recognition capabilities, this technology is ready to emerge for the next generation of automotive technology.