Q&A: Weld Sensing
August 17, 2011
The Role of Weld Sensing
Canadian Industrial Machinery asked Motoman Robotics Technology Leader for Welding Chris Anderson to describe the role weld sensing plays in fabrication shops.
CIM: What are the main types of sensing devices?
Anderson: We distinguish between two primary types of weld sensing: seam finding and seam tracking.
Seam finding, done before welding, will find an offset at one particular point. It may be done by touch sensing with the weld wire, which can take a few seconds, or with a laser or camera and be done in a second. Multiple features or points can be detected at the start and end or along a seam and recalled during welding to track a weld, based on previewed data.
Seam tracking follows a seam in real time while the robot is welding. Through-arc seam tracking is the most popular and least expensive. It requires the robot to weave in the joint and then compares the current values at each leg of the weave to keep the stickout the same.
Laser cameras can also be used to measure the width and depth of the joint. They are mounted just ahead of the welding torch to guide the robot.
CIM: How should a manufacturer choose a seam-tracking system?
Anderson: The robot can be equipped with a low-cost seam-finding sensor to find a start point or series of points prior to welding. Real-time seam-tracking systems are more expensive, but required for contoured parts or irregular shapes. Seam-tracking systems are higher technology and require more training than seam-finding sensors. Robot manufacturers or integrators can help a manufacturer select a sensor by reviewing its application.
CIM: Describe the relationships among the scanning technology, software, and robot.
Anderson: The robot will repeat a programmed motion very precisely. The sensor is used to detect the joint location and needs to be matched to the joint type, material condition, and part type. The software should be geared toward making the sensor easy to use, while providing application flexibility. The software determines the shift or offset amount from the programmed path to the actual part. Laser cameras are sophisticated enough to require their own software to set up the camera to identify different joint types and a profile to track.
CIM: Is it important to be able to track weld quality while the part is still in production?
Anderson: Yes, and this technology is still evolving. Some manufacturers add arc data monitors to the robot to verify that weld parameters do not vary from established procedures, but this does not guarantee detection of a bad part.
Laser cameras have been equipped with inspection software to analyze the profile of a completed weld for surface conditions such as weld size, convexity, even larger size porosity, but it does not analyze depth of penetration.
Robot software can limit the window of seam-tracking variation to prevent gross tracking errors. However, most users rely on the robot operator to inspect the welds after being produced on a robot with weld sensing.
CIM: Is real-time feedback necessary?
Anderson: A majority of applications can be performed just with preview sensing of joints or part location. Real-time feedback is necessary for complicated geometries or parts that will vary from distortion while being welded.
CIM: What type of information is collected, and how is it used?
Anderson: For touch sensing, a voltage is applied to the weld wire and the robot moves toward the joint until the circuit is closed when the wire touches. A laser beam can be used in a similar fashion, moving toward a workpiece until it reaches a focal point.
Through-arc seam tracking measures current while welding occurs with a weaving motion in the joint. The current is proportional to the wire stickout, and the robot "steers" by keeping the stickout equal on each leg of the weave.
A laser camera generates a line of laser light and a camera detects the distortion of this line by the joint geometry. The line bends at geometry features, and the camera can calculate depth information from the line's position in the field of view. The laser camera is suited to welding because it is less sensitive to surface conditions and ambient lighting in a factory. A laser camera can provide additional joint information, such as gap or area, beyond just the position of the seam.
Gray-scale cameras can be applied to seam finding, but they detect variation in an X-Y plane and do not provide depth information. They are good at finding patterns such as the diamonds in a sheet of expanded metal.
CIM: What happens if a gap or imperfection is detected?
Anderson: The reaction of the robot to a seam imperfection needs to be programmed. It can either vary parameters or skip the weld if the gap is too large.
CIM: How do these systems eliminate rework and scrap parts?
Anderson: Adding sensors to robots allows them to accurately place welds on varying parts. However, manufacturers are also using seam-finding or gray-scale sensors to errorproof welding operations.
[This must be done] by using the sensor to call up the proper weld program based on parts loaded into the fixture or verifying that manual clamps are closed prior to welding. Part sensing can add cycle time and cost, but it can be less expensive than adding sensors to the tooling or dealing with rework and stoppages caused by improper fixture loading.
CIM: Is this more commonly seen in thin or thick material processing?
Anderson: Sensors have been widely used for fabricating agriculture and construction machinery because the material is thick and structures are large, making it more difficult to control tolerances. Sensors are less prevalent in automotive applications where material is thinner, volume drives tighter tolerances, and joint sensing adds a significant percentage of cycle time.
CIM: How can a company determine the ROI for one of these systems?
Anderson: Some companies must use sensors because of the size of the parts. The application of robotic welding typically forces a company to implement manufacturing controls to keep joint tolerances.
One manufacturer that has embraced vision sensors reversed this approach. It specifies steel that has the desired forming characteristics. However, its purchasing department often can save money on steel that is outside of specifications and rejected by other manufacturers. Robots equipped with sensors give this manufacturer the ability to weld parts that have bigger tolerances because of the less expensive steel.
Manufacturers that produce part families may want to incorporate sensors on the robot to detect the part size or type and use this feedback to control the weld location. This may reduce the cost of tooling to accommodate different part sizes. Cameras can be used to verify that various brackets are in place and in the proper position.
blog comments powered by Disqus
