By RICHARD MANDEL, senior editor

By retrofitting this Fanuc robot with Auto-Axcess, DTR increased productivity 25 percent. Part of the increase occurred because the Auto-Axcess' Accu-Pulse process cut arc-on time by 50 percent.

The DTR anti-vibration engine mount. The thumb rests in the main ring, the battery bracket is above the fingers, and the difficult-to-weld wiring harness mount is at the upper right of the part.

In this weld bead cross-section from DTR's quality department, the weld bead on the left shows questionable penetration, while the bead on the right shows good penetration at both the root and toes of the weld.

DTR production engineer Eric Walton uses a pendant to adjust torch angle on one of the Fanuc welding robots.

Christmas miracles do come true, or at least they did for the associates and engineers managing the robotic welding stations on DTR Industries, Inc.'s anti-vibration engine mount line. The miracle? In 2005, for the first time since this tier-1 automotive supplier began production at its Midway, Tenn., facility, everyone on the engine mount line had weekends off between Christmas and the New Year.

"We normally run the engine mount line six days per week, as well as some Sundays, to build up inventory," says Eric Walton, production engineer, welding. "The way we were MIG welding was very slow and difficult. The robotic welding cell had to run perfectly all day to meet our production target, and we would rarely get it."

New welding systems retrofitted on five Fanuc robots, a task that began in October '05, changed throughput. However, the decision and execution was the result of a great deal of advance planning.

The cost of quality
Since DTR's parent company, Tokai Rubber Industries of Japan, established its first North American facility in January 1988, the company has been working to globalize its corporate activities to respond to the local procurement needs of its customers, primarily automobile manufacturers, as well as adapt to the intensifying international competition and changing business environment. The Midway DTR operation, which has a total of 34 robotic welding stations, promises its key customers (which include Toyota, Honda, Nissan), that products will be delivered on time, per customer requirement, with no delivery errors and no late shipments.

Allowing even a single non-conforming part to reach the customer is not an option. However, achieving a perfect output on the engine mount line often required going over the number of man-hours allotted in the production plan, which in turn drove up piece-part costs. DTR incurred these additional hours largely due to some of the industry's most common welding problems: excess spatter, poor arc starts and insufficient weld penetration.

Weld spatter especially causes problems in two areas of the engine mount. If spatter enters the main ring, a bushing pushed into the hole may not go in correctly. Any spatter on the bottom of the battery bracket, where it mates to the frame of the vehicle, may cause bolt hole misalignment. In general, spatter can also cause clamping problems with welding fixtures, as well as create a rough appearance after the part is e-coated.

"We capture spatter problems at the welding cell so parts do not go out with spatter, but we had to work a lot of overtime to make our production goals," says Walton.

The configuration of the wiring harness mount and associated fix-turing also caused problems. Because of limited access, the welding wire (an .045-in. diameter ER70S-6) needed to stick-out approximately 1 in. beyond the contact tip. Long stick-outs typically cause arc starting problems, as most systems cannot compensate for the voltage drop created by the long stick-out.

"Having a long wire stick-out causes a lot of spatter," says Walton.

Tim Taylor, production engineering assistant manager with DTR, notes that long stick-outs also cause poor arc starts, as does having a large "ball" on the end of the welding wire. Both conditions lead to other problems.

"If the welding power supply doesn't start and stabilize the arc within a certain amount of time, the Fanuc system will detect this and set off a machine alarm," Taylor says. "The poor arc start might jeopardize weld penetration, so our policy dictates that machine alarm parts automatically go into our 'red reject bin' and we start all over."

An arc start failure may also create a "collision sensor clutch break over," where the cold wire being fed will push the spring loaded arm holding the torch past its break-over point. Essentially, the robot's collision sensing mechanism gets fooled into thinking that the torch collided with an object, so it stops the system.

The wiring harness bracket is difficult to weld, as well, without burn-through, because the 0.5 mm piece is much thinner than the other components of the engine mount, which are up to 2.4 mm thick.

In fact, the engine mount, which is made of mild steel and has 15 total welds made at six different stations, is by far DTR's most complex and challenging component to weld. To ensure quality, in addition to the visual inspection and quality policies in place at each welding station, DTR randomly pulls one part per welding cell per shift for destructive testing.

"After cutting a part and etching it with acid, we use a 10X magnifying lens to measure weld penetration," says Bonnie Collins, supervisor of quality at DTR. "If there is any question, we then take the part to our DVC comparator, which can measure at the microscopic level."

If any weld penetration issues occur, DTR stops production, corrects the weld, conducts a test on a second part, and quarantines all parts back to the last known acceptable sample. Ensuring quality (and subsequently losing too many pieces due to weld penetration issues) was one of the reasons the engine mount associates typically had to work three out of four Saturdays.

The limits of Kaizen
Like nearly all Japanese and many American firms, DTR first sought to achieve its production, scrap rate, man-hour and piece-price goals through Kaizen events and working based on data-driven decisions. Kaizen events, or "continuous improvement" events, are generally not major changes, but are based on making little changes on a regular basis, always improving productivity, effectiveness and reducing waste.

Small efforts, however, can reach a point of diminishing returns. "When you have exhausted all avenues of kaizen with your current system," says Taylor, "you have to branch out into new technology or different ways to manufacture the part."

Making major changes is generally not part of the Japanese automotive manufacturing philosophy, and switching brands of welding equipment and using a new MIG welding process is perceived as a significant change with a large risk. In a Kaizen-based shop, a significant change is undertaken only after exhausting all efforts to produce desirable results through Kaizen. Further, making a change may require informing customers, which in turn can take an enormous amount of work and data to prove the viability of the change. Lastly, the new system could fail to produce the desired results. If that happened at DTR, the company would lose the time it took to install the new equipment, remove it and then reinstall the old equipment and resume production all of which would jeopardize delivery.

"We are very concerned about losing production anytime we implement a change, as we don't usually have the luxury of a large inventory to fall back on," says Walton.

Solving weld problems
DTR allowed teams from Miller Electric and Holston Gases, DTR's gas distributor, to retrofit the DTR robot with Auto-Axcess systems from Miller to handle the most difficult challenge: the long stick-out on the engine mount's wiring harness bracket. DTR believed that the Auto-Axcess could weld this part because, through weld demonstrations, Taylor and Walton had seen that the system's Accu-Pulse process could address their three primary welding problems: poor arc starts, spatter and weld penetration.

Accu-Pulse is a software-driven CC/CV (constant current/constant voltage) process that combines with the power supply's faster response rate to sense poor arc conditions and correct fluctuations in contact tip-to-work distance. The process adjusts the weld current to a predetermined level at the beginning of each peak and background phase. Once the target current is reached at the beginning of each phase, CC control of the current switches off and the CV control loop turns on. The current floats within a limited range based on the actual arc condition present, and the voltage is kept constant, allowing Accu-Pulse to maintain a much shorter actual arc length. The shorter arc length provides control characteristics and allow for optimized arc performance. Without large current overshoots, the very short, stable, and controllable arc limits explosive starts and spatter.

Another feature of Auto-Axcess is SureStart technology, which electronically ensures that a large ball is not left on the end of the wire when welding stops. This provides a predictable condition for the next arc start.

Retrofit battle plan
Taylor notes that, "Eric [Walton] and I have a very good relationship with Mr. Satoru Ueno, our Japanese coordinator, and our production management team. We have built trust and confidence in each other, so Mr. Ueno supported us very well. He went to production management and said, 'We really need to conduct a trial of the Auto-Axcess instead of producing parts this Saturday.' That opened up robot time on a weekend for our team and the teams from Miller and Holston."

Jason Blair, an account representative with Holston, explains the need for a retrofit battle plan.

"In the past, we had some installation situations with some welders that were just a debacle because we had a hard time getting them to communicate with different robots. Because we knew we would only have one chance to get it right with DTR's production schedule, we started preparing for the Auto-Axcess installation two months ahead of the actual installation date. We used our store as a staging base and had all the materials and personnel assembled on Friday night so we would be ready to roll at 6 a.m. Saturday."

Miller personnel, including regional account manager Curt Beck, application manager Kevin Summers, and industrial application specialist Stan Henn, prepared by taking the serial numbers from DTR's Fanuc robots to make sure the Axcess "plug and play" adaptor kits matched the interface on the Japanese robots. They pre-measured the mounting hole pattern for the Axcess wire feed system so it would mount without difficulty, pre-measured cable lengths and considered grounding, tooling and torch angle issues.

Key to a successful installation, Beck believed, would be Auto-Axcess' AutoCAL (automatic calibration) function, and the fact that Axcess requires just one connector and one set of secondary cables.

"Many robotic welding customers literally fear swapping out a welding power source because connecting and calibrating a power source to a robotic controller and robot I/O board has historically been painful," says Beck, who has more than 22 years of experience and has installed hundreds of automated welding systems. "Cables and interfaces frequently aren't compatible, and there are a lot of them. Further, plotting data points (voltage and wire feed speed reference values) often takes hours because it requires making a weld, plotting a point and repeating this enough times to achieve linear control."

Most robotic systems communicate using analog signals. The AutoCAL feature uses the Auto-Axcess' internal Device Net digital technology, communicating through an internal digital-to-analog converter to allow simple, automatic scaling and synchronization to an analog robot controller. This assures consistent and repeatable results with minimal downtime when installing Auto-Axcess power sources into existing robot cells.

Mission accomplished
The AutoCAL function, prior preparation and having the DTR, Miller and Holston teams on site enabled retrofitting the first robot and striking an arc by 9 a.m. The team spent another hour moving some "teach" points and adjusting torch angles with a pendant, and they were making parts by 10 a.m. — just four hours after starting. "I have a great deal to do with installations, and I didn't think we could do it that quickly," says Walton. "We actually gained production the first night after installing the Auto-Axcess system. In fact, our first part had zero weld failures." With competitive welding systems, installing a new system and reaching production typically takes 8 to 10 hours.

The team retrofitted three robots within the next two months. The first installation, as noted, was trouble-free, as was the third. The second robot cell, however, had grounding and arc wander issues. Spatter and anti-spatter compound associated with the previous welding process had built up in the cell and prevented making good contact.

"To solve the problem, we cleaned the tooling and fix-turing to get a better ground for the weld," says Miller's Henn.

Taylor says that, "The net effect over a six-month period has been a productivity increase of 25 percent. We also cut 19 seconds from the welding cycle and cut arc-on time by 50 percent, which also reduced shielding gas use by 50 percent. I calculated a return on investment of 143 percent and a payback time of .43 years. This easily justified buying the Auto-Axcess welding systems."

Additionally, the Auto-Axcess system reduced weld penetration failures to just two failures in the fourth quarter of 2005, a 12-fold improvement compared to previous quarters.

"Our previous system cost us a lot of money in scrap and rework on the five engine mount welding cells because of fluctuations in weld penetration. That's why we decided to try the Miller welders in those cells," says Collins. "We had zero failures in October, two failures in November and zero failures in December, once we changed to the Miller welders."