Focus on Aluminum Fabrication & Solution Providing
single unit orders profitable with self-taught robot: robotic mig welding of a wide variety of irregularly shaped, thin aluminum components might not seem the best situation for robotic welding. however, a 50-employee manufacturer of fans finds it an idea
As many manufacturers are familiar with today, a rapid transition has become possible, \"explains John Moore, president of Moore fan Limited (Marceline, MO).
Fans used in these processing industries are the basic units of production technology and are much more complex in design and manufacturing than fans with stamped steel or aluminum blades.
These fan blades are wings and are similar in shape to the aircraft wings. [
\"Usually, we have very few orders, usually at 4--
Six weeks, \"added Moore.
\"But all we need is 1--
We have to meet our fast order guarantee and it\'s even harder these two weeks.
Therefore, we are constantly seeking to improve the cycle time.
\"Fast order guarantee to customers, the company will be 24-
Hours after receiving the order
Moore explained that the company\'s advantages include making unique shapes from 0.
063 \", 5052 Aluminum plate, shaped into a double layersided airfoil.
The company may be unique because it doesn\'t like outsourcing.
Its philosophy generally rejects the option of buying from outside and makes all of its own parts, such as the use of in-
Home lathe, bar feed, extrusion and CNC center.
\"At the tip of the blade, we have attached a bracket that is inside the blade.
John\'s brother, Robert D, said: \"This stand has a tip that extends about 1/2 outside the blade to reduce noise and improve operational efficiency . \"Moore,Jr. (David)
Management member of the company.
\"However, in some cases of winter operation, ice accumulates at the tip of the fan front, which can both reduce efficiency and damage the tip of the end of the blade.
If the rivets are carried out only in one or two places, the frozen ice will bend the plate from the blade.
David Moore said: \"While we have already made rivets on blades in the last 10 years, we think welding will produce better products with better strength, enhanced aerodynamics and better
According to Lambert, the gas was used in the first welding attempt after 90 s. tungsten-arc (GTA)
It is processed on several fans, but the typical slow welding speed is not satisfactory.
\"Nexteffort uses this gas. metal-arc (GMA)
But this traditional technology can\'t meet our needs . \" Tool and mold/welding supervisor for gas and welding product supplier understand the target of atMoore fan and think robot arcwelding is an unexplored option
2000 recommend John and David to Panasonic Factory Automation (Elgin, IL).
Panasonic\'s Brian Doyle reviewed the application and recognized several complex issues.
\"Given the large amount of air shape and length, we have to assume a large amount of arrangement,\" Doyle recalls . \".
This makes programming tasks daunting.
Then the variability of the part. to-part.
\"It\'s basically impossible,\" said John Moore, \"because thin materials are welded, it\'s basically impossible to describe the exact part position in the space, which is a key task before welding . \".
During 2001, the first assessment was for optical sensing using a laser range machine, and the attempt was not entirely successful.
Although the performance on the convex surface of the blade is acceptable, according to Doyle, the performance on the concave surface is not necessarily due to the reflection of aluminum.
\"We just don\'t feel like we have a strong enough solution to produce, so we put it on hold for the time being,\" Doyle said.
John Moore added, \"I will send an email on a regular basis to see if they have something new and make sure they don\'t forget us. \" [
Panasonic proposed four in 2003.
Way solution with new technology: High-
Voltage touch sensing; Windows[R]CE-
Robot control platform based on G2AC-
200, this is an artificial intelligence variable polarization dual inverter AC power source and meters Gree servo wire feeding system specially designed for welding aluminum.
\"The voltage sensor is in contact with the welding wiring and works well with cold rolled steel,\" Doyle observed . \".
\"But for larger plates or materials coated with lead primer, low
Voltage touch sensing is not reliable.
We want to know high.
Voltage touch sensors previously developed for these applications sense the refractory oxide layer on aluminum and do so without bending the wires.
Doyle explained the method, noting that a set of contactor isolated the welding wiring from the welding power supply while the transformer applied highvoltage, low-
The current potential of the welding cable with the workpiece.
This happens when the robot goes into the part in search mode.
When the wire touches the workpiece, the potential disappears and the robot remembers where it finds it.
Combination of 400-
Volt AC signals and sharp points on the electrical line create small sparks, which are often seen before robots get there.
\"What makes this work super in practice
The rapid response of the G2 controller and its standard 64-bit processor allows the robot to be parked at a dime before the wire is bent, \"Doyle explains.
The last factor is the migree servo pull gun, which, thanks to its unique orbital planetary drive system, actually pulls the wire straight when it sends the wire.
The answer is to contact with a reliable straight wire.
\"We can now handle the changes in the parts better,\" Doyle said . \".
The next important task is to model the geometry of the part to be welded.
\"Once we prove that we can find the part by touch sensing, we are confident that we can weld it. Welding thin-
Gagealuminal benefits from the ARC Properties of the variable polarity AC power supply. Burn-
Due to the combination of low current, the penetration is lower by being eliminated, and in the direct polar part of the arc cycle.
Due to the almost complete control of the wire feeding speed of the migree torch, our arc stability is very good, so the weld seam is smooth, small and reliable.
But how we produce all the programs is the main problem, \"said Panasonic Welding Engineer.
Engineers realize that it is correct to describe the geometry, even for offline programming systems, too cumbersome. Self-
The teaching blades are tapered and can be accurately described as aircraft wings of different twists and turns and lengths.
After an early attempt to program the configuration with a teach pendant, the engineer worked with Bill Carter, an information technology expert for Moore\'s fans.
He believes that the availability of a working touch sensing system will be a new way.
\"I believe we can develop a program that allows robots to teach themselves instead of trying to program for every arrangement of fans.
\"If the touch sensing can be used to find the part that is placed or deviated, maybe it can be used to find the point of the weld path first,\" Carter said . \". [
Using position variables and math functions in the Panasoniccontroller, Carter is able to develop routines that address setting time challenges.
\"With the maturity of the G2 operating system and information from the top
Voltage touch sensing system.
I am sure that when a new part geometry appears, we can create a routine that enables the robot to learn the welding path on its own.
But, obviously, we still need to modify the program for each part in the run, because the change ratio of the part /-0.
\"The tolerance requires the welding process,\" Carter explained . \".
Carterdid completed all the heavy work at this stage and eventually became the key to resolving the installation time issue and making the project successful.
Doyle explained that Carter had evolved three.
Step process to achieve the goal.
The first is what Carter said about the setup program in which the robot displays the length of the fan convex surface and the part on the concave surface in teaching mode.
The robot moves manually along one axis, storing a point on each side.
Doing so will tell the length of the robot part, so it knows the starting point to start the next step, which Carter calls the discovery plan.
The Discovery program runs in automatic mode, during which therobot initially moves to the cut line station and then to the point at the end of the blade, learned in the previous installer.
Then, it adopted the high
Voltage touch sensing, alternately touch the vertical side of the blade, then move down to touch the horizontal end, then index forward and repeat.
In this way, it teaches itself a highly complex geometry.
In this discovery program, a large search window is used and the robot searches at a nominal speed of 1 m/min.
Unlike traditional touch perception, the robot does not find an offset in the Discovery program.
It is creating a new location point, actually creating the path to use in the production program it writes.
Carter calls it the last step, the production program, in which the robot still uses touch sensing, not to create the path, but to correct the fragmentation --to-
Part changes in fan blades being welded.
In a production program, the search window is much smaller, and perception takes only a small amount of time.
Once the induction is completed, the welding begins.
It is worth noting that the controller does not need to store all program variants generated.
\"There is only one production program in memory, and every time we change the part geometry, it will be rewritten automatically,\" Carter explains . \".
\"This is the beauty of it.
When we go back to the part we have done before, creating new programs is faster and more accurate than finding and reusing old versions.
We can achieve cost-effectiveness on a lot of dimensions of a part, and I don\'t have any headaches with the archiving program or the backup system.
JohnMoore concluded: \"Despite the complexity of this approach, the bottom line is that the team\'s efforts have made great achievements for us.
We need a quick process.
Considering our regular small orders
This new program runs well and achieves our main goal of having an automated teaching program and being able to process a small order without a lot of teaching time.
As far as welding itself is concerned, this is also beyond our expectation.
Wenow production month.
060 \"angle weld about 60-in. /min. travel speed.
This is the smallest we have ever produced. They are uniform and attractive.
\"Our next step will be to increase the output of the robot by adding a station and parts loading fixtures so that the robot does not stop welding,\" concluded John Moore . \".
Panasonic Factory Automationrsleads. com/410mn-