To the outer limits of design

Afew years ago, part and mold designers would have walked awayfrom the challenge presented by Pitney Bowes' industrial designteam-a concept part so complex that even today, many experts woulddeem it impossible to mold. Instead, a talented team made themiracle happen, staying true to the original concept and, in theprocess, expanding the window of possibility for injection moldedpart design.

This dream team consisted of four main players: Tony Sgroi,product design engineer, and Nizarali Virani, senior engineerand project coordinator, both of Pitney Bowes (Danbury, CT); DavidAkin, an independent mold designer from Akadco (West Dennis, MA);and Roger Poirier, vp of toolmaker Osley &Whitney (Westfield,MA).

Dubbed "the collar," the amazing part is one of 11molded components that make up the housing for the Spark mailingmachine, Pitney Bowes' newest product line addition. Accordingto industrial designer David Beckstrom, the company wanted theSpark's design to be just as groundbreaking as the technologythat it will bring to the mailing market.

"With Spark, we are strivingto create a forward-looking image that connotes the same messageabout the unseen technologies inside," he says. "Wealso wanted to keep the footprint small, yet not allow the productto become visually bulky.

Our in-house industrial designers-JosephSugrue and Patrick Thrailkill-decided to break the form into twoparts that are functionally distinct. As a result, the user interface(or meter) stands upright and at an angle to make its presentationmore user-friendly."

Modeling the Future
Designing the Spark in this way meant that the meter would beinserted into the machine at an angle close to 90°. The collarwas developed as the main nest for the meter, so that it too wouldbisect the horizontal machine. As a result, all of the other housingparts mate with the collar.

After industrial designers created the foam model and mastercurve file for the entire housing, they initiated a 3-D mastermodel file in Pro/E that would be used throughout the project.Using an SLA 500 machine at Pitney's Shelton, CT facility, theycreated a one-piece SLA prototype of the entire housing from thePro/E master file. Also, the decision to use plastic injectionmolding was made.

From this point, the project was handed to the Product PackagingDepartment, also in Shelton. After design engineer Tony Sgroireceived the SLA part and the master model file, he began breakingdown portions of the housing into separate components. "Itwas necessary to divide the housing into 11 parts, and the collaris a direct result of those part breaks, which are based on thesurfaces that industrial design needed,"Sgroi explains.

Approaching Reality
When the main surfaces for all the parts were defined, it becameclear to Sgroi that the collar would be the heart of the housing."It provides all of the structural integrity because theother main housing parts bolt into it, which stiffens the entireassembly structure," he says. "It acts both as the mainnest for the meter and as the mating surface for all other parts.Most of its surfaces are contoured, with a few flat areas appearingas ribs or internal features. Roughly, the dimensions are 9 by7.5 inches, with a 4.75-inch depth."

Luckily, Sgroi has a mold design background, and could approachthe part design with a knowledge of tool construction. His firstimpulse was to simplify the tool design. However, that would havemeant changing geometry from the original concept design, whichwas forbidden.

Sgroi realized that in its original state, the part appearedimpossible to tool. Rising to the challenge, however, he begana preliminary mold design (Figure 1), generating surfaces forthe core, cavity, a cavity slide designed for wire EDM, and threeother slides. In operation, the cavity would pull up at a 90°angle, and the cavity slide would retract at an 8° angle relativeto that normal direction. There was also a slide pulling throughthe core. Matters were quickly becoming complicated.

When toolmaker Roger Poiriersaw the part and mold designs, he also had ideas aimed at simplification.First, he suggested molding the collar in two parts to reducetooling cost and make the component easier to mold. But becauseof structural requirements, it couldn't be done. Secondly, hewanted to build the tool with the part laying on its side, ratherthan standing straight up in the mold. With this scheme, though,witness lines would have been at a spot where the customer wouldoften be touching the housing. Aesthetic concerns nixed this idea.

Adding Steel
Poirier contacted mold designer Dave Akin, and they began workingon the concept mold design sent by Sgroi. Akin imported Pro/Efiles of the collar into his system (Cadkey with FastSurf), andcreated all of the molding steel as 3-D wireframes, along witha surface model for parting lines and mold surfaces.

"The major actions and parting surfaces were already defined,"recalls Akin, "and my task was to design the steel aroundthese components." But as the tool design progressed, morefunctionality and features were being added from Pitney Bowes'end. An added circuit board mounting, for example, required anadditional slide.

Akin decided to use the existing slide that pulled throughthe core to activate this additional slide. "As the coreslide pulls, the other one cantilevers out of the part,"he explains. All of the slides are also interlocked in the B sideof the mold to make them stronger. (For slide positioning, seeFigure 2.)

To give an idea of the tool design complexity, Akin pointsto 28 E-size sheets required for detailed drawings of the moldlayout. These show tooling details that include a floating plateon top of the mold to which the core is mounted with sliding gibs.All of the sliding actions are driven by in-mold mechanics, nothydraulics. "Without CAD/CAM, we could not have producedthis part," he says.

Tooling Up
Poirier agrees that advances in technology play a big part inthis molding miracle. "Building the tool brought us to anew level of trusting our CNC equipment," he says. "There

wasn't any one feature that was more difficult than the others-we've produced unusual slides before-but we had never put these in combination with a geometry this complex. We had to machine directly from CAD files, and there was no way to know if the pieces would fit when we put them all together. Yet they did."

After getting the wireframe and surface files from Akin, andthe Pro/E master file from Sgroi, Osley &Whitney toolmakersand programmers went to work. "We couldn't always get filesinto our Mastercam system, so we bought a seat of Pro/E for thisproject," says Poirier. "When files come in, we needto define the surfaces for EDM electrodes or CNC machining better,and with this project, we spent an equal amount of time programmingand machining."

Programmers interrogated the Pro/E file for dimensions, bringingevery surface on the molded part and every piece of steel intothe Mastercam file. To verify that electrode dimensions were correct,toolmakers used CMM system outputs and compared them against theCAD file. About 75 percent of all surfaces were EDM machined becauseof the deep and unusual shapes, and slides were machined fromblocks of steel. All of the other surfaces were CNC machined.Finally, heavy venting was added.

While the molded part itself weighs only 150g, the resultingtool tips the scales at a whopping 10,000 lb. "We had tosample the tool on a 500-ton press at Bayer's facility in Springfield,MA because it was too heavy for the cranes at Pitney Bowes' plant,"he recalls. Bayer also supplied the PC/ABS material (BayblendFR110) used to mold the collar. Injection pressures ranged from10,000 to 12,000 psi.

When the dust settled, Poirier had a chance to appreciate howtalented his programmers and toolmakers proved to be. The firstshots off the tool were good, with only normal debugging requiredfor a small amount of flash. Close-up views of the mold showingthe core, cavity, and several slides can be seen in Figure 3.

Learning Experience
Other than the radical design, a factor that made this projectso surprising is that the design geometry did not change one micronfrom the concept to the production stage. Both Sgroi and Viraniagree that Pitney Bowes learned the trade-offs required for strictadherence to a high-end aesthetic design. At the Danbury facility,a captive and custom molding operation, Pitney Bowes is currentlymolding all of the housing parts except the collar (see sidebar,below).

 

Moldingthe impossible part

PitneyBowes does much of its own injection molding, and might have moldedthe collar itself, if it weren't for the size of the 10,000-lbmold. The molding facilities at Pitney Bowes don't have a cranethat can handle the load. The job went instead to Alliance PrecisionPlastics in Rochester, NY, a regular molder for Pitney Bowes.

Mart Raidmae is the director of engineering and tooling atAlliance and was tasked with running the mold. But running themold is not what makes this part so challenging. Raidmae saysAlliance was invited to the project late in the schedule and didnot get mold dimensions until it arrived at the plant.

"When the tool first got here, lo and behold, we couldn'thang it in our press. It was too big," says Raidmae. "Itwas slated to run in a 500-ton press, and with the mold put togetherthere wasn't enough tiebar clearance to get it in there."

Alliance was forced to improvise by putting the tool into the500-ton Toshiba one half at a time. "We put the ejector sidein first," Raidmae reports. "We turn it sideways, dropit between the tiebars, and then swing it into position."The other half of the mold is not so easy. "We load the coverside of the tool horizontally between the tiebars," he explains.Using a forklift and crane, the mold slides in sideways, withless than 2 inches total clearance between the tiebars (Figure1). "It becomes a really complicated setup," says Raidmae.

Once it's in, however, Raidmae reports that the mold runs cleanlyand well, despite the complicated set of slides and lifters thatfills the tool. The part fills easily and, aside from a few last-minuteengineering changes performed by Alliance, the mold performs aswell as the molder could hope.

Raidmae says his primary concern right now is residence time.The 62-oz barrel on the 500-ton machine is too large for the part,which weighs only 150g (5.3 oz), though the mold does have a 7-inchcold sprue. So a smaller barrel may be in store for the future.Still, the machine runs the PC/ABS part in 1-minute cycles withoutincident. Raidmae says he's only sorry Alliance, also a moldbuilder,wasn't given a chance to build the tool. "I would have lovedthe opportunity to give it a shot. It's really amazing,"he says.-- By Jeff Sloan