In practice, the current is measured using a Hall Effect sensor or a current shunt。 A low-pass filter is used to depress noise from the signal。 From a control point of view, sensor data can be analyzed continuously using template matching or only at the turning points using differential control。 Due to the relatively low accuracy of the sensor data regarding precision, etc。, a differential control is in most cases sufficient。 This should be seen in the context that the sensing principle needs a few mm weaving amplitude to be able to measure a reliable change and difference in arc current。 It also indicates that achievable tolerance are restricted for those cases which can accept a weaving motion, which produce a wider weld rather than a
straight motion。 However, in medium to thick plates this is usually not a problem (thicknesses above about 3 mm)。
Figure 3。6。 Left: definition of Tool Center Point (TCP) and weaving directions during through-arc sensing。 Right: the optimal position for seam tracking in arc sensing [5]
Experiments based on the through-arc sensing principle indicate that it is possible to detect a variation in contact tube to work-piece distance of about ±0。25 mm。 This shows that the method can be used for tracking with quite small amplitudes, in the order of 1 mm。 However, it is important that the control of the motion is robust so that the robot does not lose the weld joint during tracking since there is no obvious way to get back on track。 This is due to the fact that the weld joint can only be detected during weaving and welding and that the field of view is equal to the weaving motion of the welding gun。 If no weld joint is detected, there is no information available on how to find the joint。
In practical implementations the tracking functionality is usually combined with a search function, specifically for the start-up phase, so that the robot starts welding with a weaving motion and, if no joint is found, gradually moves in a predefined direction perpendicular to the main nominal weld path。 A search function can be implemented in basically two different way, (i) move the weld torch towards the respective plates one at a time until contact occurs, usually by measuring an electrical contact, and from that information calculate a starting position of the weld joint, or (ii) during welding define a start position for the weld and also define a direction of gradually moving the weld torch during weaving until it detects the weld and tracks the seam in a normal way。
The information acquired from through-arc sensing can be retrieved and used in basically two different ways, either continuous measurement of the current or measurements at the turning points of the weaving motion。 If a height control is included, a measurement should take place at the center of the weld joint as well。 Based on the measurement principles above, different control principles can be
applied which are usually based on differential control and/or template matching of the signal。 If template matching is used together with continuous measurement of the current, a more precise control can be made that can also take into account for non-symmetric shapes of the joint profile。
Figure 3。7。 Example of the functionality of the through-arc seam tracking over segmented plates that deviate both sideways and in height [5]
Figure 3。8。 A T-pipe representing a type of work-piece that should benefit from a seam tracker which can compensate for both position and orientation changes [5]
As for seam tracking using laser scanners, a nominal path is used。 In multi-pass welding the robot controller has a function to memorize the tracked path and use that as a template to make subsequent welds。 Similar to tracking using lasers, tracking is performed to change the position of the weld torch so that it will be aligned with the weld joint。 In general, no compensation will be made for orientation of the weld joint。 However, a control scheme can be made that generates a polynomial and vector description of the weld path as described above and the sensor principle can be used to drive the robot with subsequent target positions during welding。 A sample simulation of through-arc seam tracking is shown in Figure 3。7 which displays the principle of tracking over segmented plates