Robot Welding Aluminum Car Seats.

During 2000 I was requested by an engineer at a tier one supplier to analyze the welding performance of their ABB Arcitec robot/welding equipment. This plant produces extruded aluminum parts, the welded parts in question are used in (SUV's) car seats. The car seat parts require small welds which are made on thin gage 6061 aluminum.

ABB Flexible Automation (the robot welding division), of Fort Collins, Colorado was awarded the contract by Criterion Machinery. Criterion is the system integrator for the seat frame parts. Two robot cells were ordered, ABB advised Criterion that their "Swedish Arcitec" welding power source could handle the thin gage aluminum welding requirements. The Arcitec power source is unique in that it's integrated into the ABB robot controls

Two ABB robot cells containing the Arcitec power sources were shipped to Criterion for initial fixture and weld tests, the weld tests carried out at this time were minimal. The two units were then shipped to tier one supplier.

Since the installation of the robot cells continuous production of optimum weld quality parts has been impossible due to the issues documented in this report. Weld reject rates have averaged sixty percent and the robot down time per hour has averaged 20 to 30 minutes. Many issues are documented in this report, however the primary cause of the extensive weld quality and productivity issues was the Arcitec welding power sources provided by ABB. This weld equipment had extensive electronic problems, the equipment welding results were inconsistent and unsuited to the needs for the high volume, small aluminum welds.

The Arcitec Power source History.
The Arcitec power source is manufactured by ESAB in Sweden. The Arcitec is an inverter pulsed GMAW power source. This power source was integrated into the ABB S4/S4C robot controls in an attempt to improve the communication "response time" between the robot controller and welding power source. I have extensive knowledge of Arcitec power source issues. Four years ago, I was employed by ABB, a Swedish manufacturer of robots. My position with ABB was senior weld process engineer. I was employed at the robotic welding division, Fort Collins, CO. Part of my welding responsibility was to evaluate the application performance of the new Arcitec invertor power source.

I have evaluated many power sources during the last three decades and cannot recall any welding power source that even came close to the amount of welding Issues generated by the Arcitec. The bottom line, this equipment was introduced to the market without sufficient testing by either ESAB or ABB. Also in contrast to more electronically sophisticated power sources this equipment was and still is limited in its "pulsed" application potential.

 

As a result of the never ending software issues and erratic performance, and in conjunction with the Miller/ABB agreement, ABB North America reduced its focus and promotion of the European ESAB package and instead focused on promoting either the Miller Invision, the Lincoln Power Wave or the OCT pulsed equipment for use with its robots, specifically when aluminum welds were to be made.

When requested by VAW to evaluate this installation I was under the impression I would be examining a new ESAB welding power source that had evolved electronically from the Arcitec equipment I had evaluated four years previously. Four years is a lifetime in the evolution of electronic power source technology. What I saw at VAW was I believe two Arcitec power sources manufactured four years earlier, they certainly had the same unique soft ware glitches which I was familiar with, and the power sources still produced an inconsistent output weld voltage and weld performance which was also found in this equipment at the time I was with ABB.

Shortly after the seat manufacturer accepted the robot cells, "three Arcitec power sources were replaced" by ABB. The typical life of a traditional MIG power source is 10 years without requiring repairs. With the more sophisticated, inverter pulsed equipment one would anticipate that this equipment should at least last till the warranty runs out, which is typically two to three years.

For five months after the robots were installed the two robots were unable to produce consistent, acceptable weld production or quality. Reject rates were in the 60-70%% range and the production cycle efficiency rarely attained 40%.

The Primary Weld Issue:
From a weld perspective the most serious issue was the numerous welding wire burn backs which were generated during the arc starts. The wire burn-backs have also been a dominant cause in the abnormal weld reject rate .

A MIG wire burn-back when aluminum welding can affect the weld in the following manner. During a burn-back the welding wire disintegrates in the arc and either cleanly breaks off or gets trapped (welds itself to the contact tip). With aluminum the arc generated in the arc ignition burn back can damage the protective aluminum skin, this can result in the arc start location with a highly oxidized aluminum surface, (dirty gray appearance = poor electrical conductor). When the robot tries to re-strike an arc on this contaminated surface either the arc will not re-ignite or the weld will be damaged with the contamination.

A burn back as mentioned can also result in the welding wire trapped in the contact tip. The tension that results can cause the wire to spiral around the drive rolls. In a one hour production run on the 19 0ct, 2000, supervised by ABB personnel after they completed their fine tuning of the robot weld data, we witnessed on one robot the burn-backs occurring four times in one hour causing a 22 minute or 32% per hour production loss.

 

 

It should be noted that a primary cause of wire burn-backs with aluminum is poor wire feed tension. However on these push pull wire feed systems in the final weld test we had excellent wire tension as tested when the wire exited the gun tip. Other causes of wire burn-backs are;

[1] The power source electronics and dynamics designed to address the aluminum needs. Without question the Arcitec power source did not have the electronics necessary.
[2] If the ARC ignition wire feed rate is set too low. On this application I ensured that this was set correctly
[3] If the arc ignition weld voltage is set too high, I ensured it was set correctly.
[4] If the arc ignition delay time is not sufficient for the arc starts, it was set at optimum.
With respect to the Arcitec welds, the final weld data applied at the ABB runoff was optimum and yet the burn-backs were too common.

 

 

 

 

 

 

Weld issue, power source inconsistencies.
As was evident to anyone standing close to the ABB cells when welding, the resulting Arcitec arc sounds were extremely erratic and inconsistent, yet only two weld schedules were used. The abnormal weld parameter changes that occurred during the weld were noted on voltmeters.. With robot "welds" and fixed nozzle to work distances one should expect minimal arc changes (consistent arc sounds) with voltage fluctuations of typically no more than one volt. The weld voltage variations noted with this weld equipment were extensive.

During the run off both cells were set with the same data yet there was no similarity with the welds and arc sounds attained. At the arc starts the ignition voltage would frequently flare up to excessive levels damaging the parts and causing wire burn-backs

When welding thin gage aluminum the degree of difficulty is increased when the welds are small in length. With a small weld the power source must have the capability of "responding inside the weld cycle times". In this case the weld cycle times average two to four seconds. With the Arcitec the welds required specific
[a] weld start data,
[b] weld data, and
[c] weld end data.
These units did not have electronic sophistication necessary to handle this short cycle application. I noted the exact same power source symptoms on ABB/Arcitec units installed at a Mexican car seat plant approximately three years previous. After two days of evaluation of the welding equipment, I contacted ABB about the Arcitec power sources, the engineers response to the problems was that the people at the auto plant were incompetent, and not capable of handling the ABB technology. Two days later ABB sent one of their most experienced reps. After many hours he was unable to run either unit in an acceptable weld production mode.

 

 

 

I advised we pull out the Arcitec equipment and replace with welding equipment I knew could handle the application. We brought in a very competant intergrator and two days later we were able to produce consistent quality welds. Issues outstanding, the need to improve wire feed consistency, With the OTC equipment we did not use a push pull system, with the 4043 0.046 wire this is required. Also the parts should clean, free of lubricant.

Weld Process Controls,
Implement the following for weld process controls. This data should be posted where the robot operator can view it

Before Production Commences.
[1] Check fixture, ensure weld spatter does not cause issues and fixture clamps are fully functional.
[2] Check amount of wire available, have a spare Alco Tec De-reeler with quick connections.
[3] Check wire tension at shift start.
[4] Check the correct part program is in use and positioner is in correct position.
[5] Run automatic TCP control at the shift start., and ensure this function is frequent during production
[6]Start shift with new tip, repeat after lunch.
[7] Ensure nozzle is clean, and nozzle reamer and anti-spatter is used frequently.
[8] Purge gas, check gas flow.
[10] Check wire helix at shift start, adjust wire straighten if necessary.
[11] The first two production parts should be approved by the programmer or supervisor before production commences.
[12] Check part tube bore for excess weld penetration.
[13] Operators must watch for repeat weld issues and report them to programmer,
[14] Operators must check that all the welds are in place and the welds are acceptable.
[15] Ensure all parameter settings for weld schedules, ensure wire feed, voltage and amps are adhered to.
[16] Ensure operators are aware of what they can, and cannot adjust, operators cannot change the pre-qualified weld data.
[17] Ensure operators use templates for positioning gun height or angles.

 

The robot operators must be trained to recognize welding issues
Each shift will require an individual capable of addressing programming issues.
Provide operators and programmers with well defined job description.
Recognize the importance of the programmer position, provide bonus incentive for quality and productivity goals. Consider in future the use of PLC controls for the robot cells to help simplify the operation of the robot. Consider in future the use of positioner's that hold more than two fixtures.Tag equipment with all settings; provide lock for OTC control panel

 

Wire Feed Issues.
[1]With drive rolls disconnected, check wire tension as it exits wire feed inlet guides, this determines feed-ability from Alco Tec Dee-reeler.
[2] With OTC drive rolls released, pull wire from contact tip to check tension through gun and liners. Try 062 "plastic" liner if tension problems exist
[3] With OTC drive rolls in tension, check wire tension as it exits gun, must exceed force of two fingers holding wire.
[4] Keep eye on wire helix issue
[5] Check that the contact tips slide easily over wire, oversize wires and undersize tip bores are not unusual
[6] Use slight tension on wire feed rolls, avoid marking or crimping wire.
[7] Determine PM program necessary for guns and robot cell.
[8] If wire feed issues still occur, evaluate best pull gun available.

Spares
One Power source
One Wire feeder
Two Weld guns
Robot spares???
For each two robots use a spare AlcoTec De-reeler, it reduces wire feed out time and is a spare if a problem occurs.

Weld Issues
Crater fill data at present not functional with the OTC power source, check with OTC for method to control craters. Ensure anyone who programs is aware of the importance of wire stick out at the weld starts, ideal is 7/16 wire from contact tip to work.
As mentioned small welds are more inconsistent than longer welds a 25 mm weld length is preferred over a 12 mm length. If a 15 mm weld length is required produce an 18 mm long weld, therefore craters without cracks should not be an issue.
The wire feed welding range is 170 to 230 ipm.
Only two weld schedules are required.
Voltage sets the arc length. If the arc length is too long it will produce a quite spray sound, set arc length for intermittent crackle sound. The weld voltage in the ABB pendant is not calibrated to the actual voltage. If voltage changes necessary first check wire stick-out.

OTC Settings
Pulse Arc On
WP/WO/OFF to off
Fuzzy alum on
Constant Pen On
Gas check on weld
Air Water on Air
Fuzzy mode superior to regular pulsed especially in arc consistency and cleaning action of alum oxides. However it's also the mode that at present is responsible for the craters. OTC for assistance

Establish
(1) Weld Qualification Procedure,
(2) Weld Procedure,
(3) Weld Process Control Program.
Establish type of daily quality check, macro and destructive,
Determine frequency of quality check.

 

Conclusion.
The problems were extensive and had gone on for more than a year. This was not a robot issue it was simply a weld process issue. It took the intergartor team and myself five days to get this unit running at over 85% efficiency with minimum weld rework. There are many lessons that can be learnt from this installation. The prime lesson when it comes to welding, talk is cheap. Before purchasing a robot welding cell have the robot supplier prove the welding process. Only deal with robotic companies or integrators that recognize and show some concern for the consequences of their recommendations.

 

This page is currently under development.