15-01-2011, 12:33 AM
SURVEY OF SEAM TRACKING SYSTEMS FOR ROBOTIC ARC WELDING
JUSTIN JOSE
S7 MECHANICAL
JUSTIN JOSE
S7 MECHANICAL
[attachment=8243]
CONTENTS
1 INTRODUCTION
2 THRU ARC SEAM TRACKING
3 ARC VOLTAGE CONTROL
4 LASER VISION SYSTEM
5 TOUCH SENSING
6 OPTIONAL FEATURES
7 SENSOR COMPARISION CHART
8 CONCLUSION
INTRODUCTION
Several types of continuous arc welding process can be accomplished by industrial robots capable of continuous path motion.These processes include gas tungustion arc welding(GTAW),gas metal arc welding(GMAW) e.t.c. These kind of operation are traditionly performed by welders who must often work under conditions which are hot uncomfortable and some times dangerous. Such conditions make this a logical candidate for application of industrial robots.however there several problems associated with arc welding which hindered the widespread use of robots in this process.dimensional variations occour in components to be welded robots cannot compensate for this as a human worker .one of the main process problems of robotic arc welding is the consistency which needs to be maintained while making part after part. A quality weld can only be achieved if theweld seam of each part moves less than ½ the diameter of the weld wire from the programmedweld path. Movement of a weld seam can be the result of poor fixturing or variations in themetal forming process. A standard robotic arc welding system does not have the ability to seethe changes in the joint location. If unacceptable weld joint variation occurs, an independentmethod of finding the location of the weld seam must be used to provide the robot controllerwith information to adjust the position of the weld path. In order to select the correct method, amanufacturer must be familiar with the strengths and limitations of each of the different types ofsensor systems available. A sensor can only track the position of the joint if it can recognize and use the features of the joint. This paper will give an overview of four of the common robotic arc welding sensor processes used today. These processes include:
1. Thru arc seam tracking
2. Arc- voltage control
3. Laser based vision system
4. Touch sensing
TAST
For Gas Metal Arc Welding (GMAW), a low cost method of tracking a weld seam is
Through Arc Seam Tracking (TAST). This method uses the welding arc as a sensor to measurevariations in the welding current which are caused by changes in arc length. For example, achange in stick-out is determined by using the inversely proportional relationship between arclength and arc current. By monitoring the arc current feedback, the robot can adjust the torch’svertical position to maintain a constant stick-out. The lateral location of the seam is determinedby using the weave function of the robot. As the torch weaves over the seam, the weld currentfeedback will oscillate. A valley in the current feedback signal indicates the torch is passingover the seam, while a peak in the signal represents the torch is at either edge of the weavecycle. Changes in the value of the peak current signal indicates to the robot that the torch ismoving away from the joint and its position should be corrected.Due to the violent nature of the welding arc, the weld current feedback signal contains alarge amount of noise. This noise can cause trouble in how the robot interprets the feedback signal and therefore, must be filtered out. By performing this filtering and creating a specific algorithm to work with only the necessary feedback information, the TAST process has been able to track lap joints with a top plate as thin as 2mm with 55ipm travel speed. TAST control methods typically have a large number of variables which are used to optimize tracking performance for a variety of applications. This makes TAST relatively difficult for operators to program during the integration period. A good understanding of how these variables effect the TAST process will allow the user to optimize the setting for each application. A utility which helps the user determine the settings of the TAST parameters can create a much more user friendly package. Once these variables are optimized, it is critical that a stable welding process is maintained. Since the weld current is dependent on the weld process, any changes in this process will effect the feedback going to the robot. The cause of these changescan be as minor as a worn contact tip or as significant as a change in the weld schedule. The robot will only know that the weld current feedback has changed and try to compensate by
adjusting the torch position.TAST is an inexpensive method of tracking weld joints. The only hardware requirementis a weld current sensor. The robot must also have software which can interpret the current feedback to modify torch position.
In many gas metal arc welding (MIG) applications, the weld joints are not repeatable. Typically, these applications cannot be satisfactorily welded by a robot without some means of adaptive control.Inconsistent forgings, stampings and castings, tolerance stack-up, distortion, and fixturing are some of the common causes of repeatability problems. Sensors adapt the path of the robot to the weld seam to ensure consistent weld quality. Through-Arc Seam Tracking (TAST) is used in constant voltage gas metal arc welding (GMAW), also known as MIG, processes. In these processes, the current varies as a function of the distance between the contact tip and the weld puddle. TAST can be used with SINE type weaving that includes vertical and lateral tracking or without weaving that includes vertical tracking only. Also TAST can be used with linear or circular motion. TAST supports any ferrous metal welding where the feedback current signal is in a steady state and stable condition. TAST can be used with these kinds of processes:
Gas metal arc welding
• Short circuit
• Globular
• Spray
• Pulse (50 to 150 Hz)
Shielding gases
• Ar and Ar--C02
• C02
• Ar and O2
TAST allows the robot to track a weld seam both vertically, in the distance between the torch and work piece, and laterally, across the seam by monitoring changes in the weld current. The information provided by TAST enables the system to adjust the robot path to keep the weld centered in the joint.
The robot path can be adjusted for the weave plane and the vertical plane (z-direction of the tool). You can use vertical tracking with or without lateral tracking, and with or without weaving.
ADVANTAGES OF TAST
1. Low cost method.
2. External sensor is not required.
3. No additional cyclic time.
4. Minimum lap thickness(2mm).
5. low maintanance requirment.
DISADVANTAGES
1. No adaptive capability.
2. Not a seam finder(an arc need to be establishedto start with).
3. Difficult to programme.
4. Noise interferance.
5. Change in welding process affect considerably.
AVC
When using Gas Tungsten Arc Welding (GTAW), the same sensing principle used in
TAST can be used to track the weld joint. Instead of using weld current, Arc Voltage Control (AVC) is used to monitor changes in the welding voltage. The weld voltage has a directly proportional relationship with arc length. (Fig. 1) The robot uses this relationship in the same way as TAST to maintain the torch’s relationship to the weld seam. Due to the low sensitivity of the feedback signal, AVC is mainly used for vertical tracking, but theoretically can also be used for lateral tracking.
Figure 1
Like TAST, AVC can be relatively complicated to set-up and maintain. If the GTAW
process is only used for vertical tracking and a wire feed system is not used, the amount of variables effecting the AVC settings is significantly reduced. Since the AVC settings depend on the individual weld parameters, a good weld must be made before developing the AVC settings. AVC is also an inexpensive option that only requires some filtering hardware and a robotic software interface. The voltage sensor hardware can be made to work with pulsed or constant weld current.
In many gas tungsten arc welding (TIG) applications, the weld joint location varies to a degree that weld quality is not acceptable. Typically, these applications cannot be welded satisfactorily by a robot without some means of adaptive control.
Inconsistent forming, castings, tolerance stack-up, distortion, and fixturing are just some of the common causes of repeatability problems. Sensors adapt the path of the robot to the weld seam to ensure consistent weld quality.
Automatic Voltage Control (AVC) is used in constant current welding processes. In these processes, the voltage varies as a function of the distance between the electrode and the weld puddle. AVC can be used on linear or circular paths. AVC can also be used with or without weaving.
AVC can be used with these kinds of processes:
Gas Tungsten arc welding
• DC electrode negative (straight) or electrode positive (reverse).
• AC
• Pulsed .1 to 10 Hz
Shielding gasses
• Ar
• He
• Ar/He
AVC allows the robot to track a weld seam by monitoring changes in the weld voltage both vertically and across the seam. The information provided by AVC enables the system to adjust the robot path to keep the weld aligned with the joint.
ADVANAGES OF AVC
1. Inexpensive.
2. No additional cyclic time.
3. Maintanance requirment is low compared to TAST.
4. No external sensor requirment.
5. Less programming difficulty.
Disadvantages
1. No adaptive capability.
2. Complicated to setup and maintain.
3. Not a seam finder.
4. Increased lap thickness.
Laser Based Systems
When material or process conditions create an application that is not feasible for the arc
sensors to track a joint, an external sensor can be added to the robot. Of the external sensors available, a laser based sensor provides the most flexibility in the size and type of joint that can be tracked. For sheet metal applications, a laser sensor can track lap joints with material thickness less than 1mm, or butt joints with less than a 1mm gap. In order to track these joints, the laser sensor must be placed in front of the welding torch in such a manner that the laser can scan across the weld joint. A camera inside of the sensor monitors the laser light to determine the location of the weld joint. The joint information is then passed on to the robot which makes any adjustments to the torch location.
Unlike TAST or AVC, the laser sensor does not need to have an arc established in
order to get joint information. The sensor can be used to search for the joint location before starting the weld, allowing the robot to place the wire directly on the joint before the arc start. Once the weld is started, the laser can be used to track the joint. The laser sensor can also be used to determine information about the weld joint, such as gap or mismatch. As this joint information changes, the robot can adaptively modify the weld parameters to match the optimal settings for the changing conditions. A laser sensor creates a relatively complex system which can be affected by the everyday rigors of a harsh production environment. Since the sensor package is attached to the weld torch, it can become an obstruction which limits the torch access to some joints. The sensor package is relatively fragile and the programmer must be careful to create programs while avoiding collisions with the parts and fixtures. Some laser packages also come with a mechanical rotator which turns the sensor around the torch to help position the laser on the joint. While a rotator can reduce the complexity of positioning the laser, it will add to the bulkiness and cost of an already expensive system. In order to justify the cost of a laser tracking system, a study should be performed which indicates the laser sensor would have a significant reduction in weld repair costs.
Principle of operation
The sensor head contains a CCD camera and either one or two laser diodes. The lasers act as a structured light source and project a stripe of laser light at a pre-set angle onto the workpiece surface under the sensor. Sensors are available with varying field of view to accommodate a range of working distances, seam types and sizes. This allows the camera to view the stripe directly under the sensor in any situation. In front of the camera is an optical filter that allows the light from the lasers to pass through, but filters out nearly all other light, such as the welding arc. The sensor is therefore able to operate very close to the welding source.
When the system is tracking, the welding speed and look ahead distance are used to calculate the vertical and horizontal corrections to keep the welding torch positioned over the seam. Tracking systems are interfaced to either a robot or to a welding machine PLC via analogue, digital or serial interfaces. In the case of welding machines the corrections are made via motorised slides.
Robust design for welding environments
Sensor head optics are protected from the smoke and spatter of the welding process by a copper spatter shield. This spatter shield holds a clear plastic disposable window, which is changed periodically when a build up occurs on its surface.
The sensor head can be cooled by air or welding gas (clean, dry and oil free) in order to maintain the temperature of its electronics below 50ºC and protect the optics against fumes. If necessary a water cooled mounting plate can be used to provide additional cooling. Conversely if the laser diode temperature is likely to fall below +5ºC, an optional heater can be added to the sensor head.
Safety
Lasers used are Class 3B devices, according to EN60825:1992.
Laser Wavelength (nm) Max output power per laser diode (mW)
Visible 650 - 699 30 (laser output is not pulsed)
Unique 3D Machine Vision for Robot Welding
In robotic welding, the robot is directed by a program which states the overall dimensions and general trajectory of a job.
However, the robot will not know exactly where the start of a new seam is. Even if located in a jig or fixture, the work may be slightly away from its intended position, introducing errors.
These errors will, obviously, have a detrimental effect on the quality of the weld, perhaps resulting in scrap or the shipping of bad parts.
The CSS Weld Sensor can remove this uncertainty by exactly locating (down to 25µm) the seam start, end or multiple points in between.
Operation
Within 150ms, the CSS will determine a seam’s:
Position, Width, Depth & Trajectory
This information is used to automatically correct the robot path in 6 degrees of freedom, whilst the gap and depth information can also be used to control welding parameters.
Box Corners & Fillets
The CSS Weld Sensor’s circular trace allows the position of all faces to be defined in one measurement, enabling fast, accurate location of internal or external corners.
Circular Scan Advantages
Using a circular scan, rather than a line, the CSS is capable of obtaining extensive three dimensional data from a single measurement. It will also cross a seam twice, allowing the trajectory to be calculated.
The CSS Weld Sensor is vital when welding thin materials, where torch position errors can result in the metal being cut rather than welded, and essential for short tacks (or safety welds) when all the necessary seam information is obtained in the minimum of time.
In the vast majority of cases, seam location is a quick and easy alternative to full seam tracking and, because the sensor is not used whilst welding, the front aperture can be covered by a shutter to prevent ingress of dirt.
Seam location for safety welds Corner Location Start of seam detection
CSS Weld Sensor - Seam Tracking
ADVANTAGES
1. Flexibility.
2. minimum lapthikness and gap distance.
3. Seam finder.
4. Adaptive capabilities
5. low repair cost.
6. Can be used wirh all types of welding processes.
7. Can be used to welde all types of joints.
8. No arc need to be established .
Touch Sensing
Another method of adjusting a robot path for movement in a weld seam is called Touch
Sensing. Touch Sensing does not function as a seam tracker, but instead finds the weld seam
and adjusts the entire weld path before starting the arc. The robot finds a seam by using the
welding electrode or a separate pointer to make electrical contact with the part. A search
pattern is performed by the robot so it can touch the part until it finds out how far the seam has
shifted and rotated in up to three dimensions. An offset can then be applied to every weld
which is on this seam. This method can also be applied to determine if there is a gap in the weld joint which requires a change in the weld schedule. The requirements which must be met in order to use touch sensing are very straight forward. Most robotic welding power supplies contain a circuit which can be used for the touch sensing technique, and keep this a very low cost method. The biggest disadvantage comes from the cycle time increase that is added by the robot to perform the search routines. The weld joint must also have an edge which can be found by the sensor. For lap joints, this requires a top plate thickness of 2mm.
Touch sensing allows the robot to change a path automatically to compensate for object displacement.
Touch sensing consists of:
• Moving the robot tool center point (TCP) toward the object using pre-defined robot motion, speed, and direction.
• Using an input signal to indicate that the robot has come into contact with the object.
• Storing the found location of the object, or position offset information, in position registers.
• Using the stored position to move the robot to the stored position, or using the stored position offset information to shift one or more positions in your welding program.
• Support for coordinated motion
ADVANTAGES OF TOUCH SENSING
1. Less programming complexity.
2. Low maintanance requirment.
3. Suitable with most welding processes
4. Lap thickness of 2mm.
5. Can be used to weld all materials.
DISADVANTAGES
1. Additional cyclic time is more(1.5sec).
2. Limited adaptive capability.
3. cannot be used for all type of joints.
4. lap thickness of 2mm.
Optional Features
Each of these sensing methods can be enhanced with additional options on the robot controller. Coordinated motion with an auxiliary servo table provides additional flexibility in part programming and design. For large weld joints requiring a number of passes, root pass memorization (RPM) can be used to record any joint offset information from the sensor. For each additional pass, the sensor can be turned off and the RPM offset can be played back to maintain the torch placement.
Dynamic Seam Tracking
This photo shows a Reis V15 6-axis robot welding a saddle joint on a pipe section placed on a 2-axis workpiece positioner. The robot and positioner are fully synchronized, the positioner automatically rotating the workpiece so that the robot can access the entire seam. A unique feature of this system is that only a single starting point needs to be taught to the robot and positioner. Once this starting position has been taught, the robot system is simply instructed to weld for a specified distance. This system is based on a form of end-point control that we refer to as "Dynamic Seam Tracking" or DST. An MVS laser profiling sensor located ahead of the welding torch is used to detect the seam position, orientation, and geometry. As long as the seam is detectable from the starting position, the system will automatically track and weld the seam for the specified distance, without any a priori knowledge of the seam trajectory. Another unique feature is that the torch travel speed along the seam, as well as the torch standoff and orientation with respect to the seam, can all be controlled in real time. This enables highly sophisticated weld process control techniques to be implemented, including feed-forward (adaptive) fill, and real-time closed-loop feedback control of the welding parameters. A further unique feature is that workpiece setup on the positioner is much less critical than would normally be required. This is due to the technique used in which the sensor drives the entire system, and completely integrates the workpiece with the positioner and the robot. Fixturing and setup are thus greatly simplified, which facilitates low volume industrial applications. Indeed, since path programming is also eliminated, this system can effectively be used for one-off production. This unique (patented) Dynamic Seam Tracking system was developed under the Robotic Welding project funded by MRCO, and forms the basis of our research in the next generation of intelligent robotic welding systems.
ROOT PASS MEMORIZATION
Root pass memorization (RPM) records position offset information provided by a tracking sensor. The recorded information provides accurate weld seam information during welding. RPM is used with Multipass (MP) welding. RPM/MP is an option that is included with TAST or AVC Tracking.
Multipass welding is repeatedly welding the same seam to increase the weld size. The multipass instruction offers ways to offset the different weld passes. Offsetting the weld passes allows proper fill placement for quality bead profile and weld appearance. Multipass welding can be used with or without root pass memorization.Root pass memorization (RPM) is the process of recording position offset information at specified intervals during the root, or first, welding pass. Position offset information is the difference between the robot positions you recorded during programming of the weld, and the robot positions that a tracking sensor indicated were best to weld the seam. Tracking sensors include Thru-Arc Seam Tracking (TAST), Automatic Voltage Control (AVC), and others. These offsets occur because of variations in welding conditions, such as part fixturing, and welding materials that can have an effect on part fit-up. The recorded positions plus the position offsets provide the true route the robot should take when welding the seam.
Conclusion
Each type of sensing method presented in this paper has its own advantages and
limitations. By fully understanding what these aspects are, a user can best determine what is needed for their robotic application to allow for the maximum productivity. Table 1 gives a comparison of how each of the different sensors can be applied. The user must remember that a welding sensor will not improve the welding process, it will only maintain the correct alignment of the torch to the part. By adding one of these sensor methods to the manufacturing process, additional complexity is introduced to a robotic system. The best way for a robotic arc welding user to consistently make acceptable welds is to develop a manufacturing process which gives consistent weld joint placement, so these sensors are not needed.
Table 1
Sensor comparison chart
References
1) Jackson, C.E., “The Science of Arc Welding”, The Welding Journal, April 1960,
129-s to 140-s
WWW.CSS Weld Sensor - Seam Location.htm
WWW.Meta Vision Systems - sensors, laser vision, laser tracking.htm
WWW.Seam-Tracking and Control of Arc & Laser Welding Robots.htm
WWW.FANCROBOTICS\TOUCH SENSING_J536 - Product detail.htm
WWW.FANUCROBOTICS\ WEAVING_J504 - Product detail.htm
WWW.FANUCROBOTICS\AVC_J526 - Product detail.htm
WW.FANUCROBOTICS\ARC SENSOR - TAST_J511 - Product detail.htm
