Date Visited: 12 December 1995
JTEC/WTEC Attendees: E. Sachs (report author), C. Atwood, M. Wozny
Teijin Seiki Company, Ltd., (TS) was founded in 1944. It has 1,745 employees and sales of ¥62 billion. Its four product lines are (1) textile machinery, (2) aircraft equipment, (3) hydraulic machines, and (4) new technologies. Its RP effort is under the last category. The company's RP work has had the following milestones:
|Oct. 1991||licensed Soliform Solid Forming System from DuPont|
|May 1993||started R&D for applications|
|June 1993||developed new resin, TSR-730 (filled resin for making tooling - see below)|
|Nov. 1993||developed a "short run" injection die system that uses SOMOSTM 2100 to create a pattern, then aluminum-filled epoxy to create the cavity from the SOMOSTM master|
|April 1994||developed vacuum casting and injection molding die by soliform directly using TSR-730 resin for up to 20 pcs of ABS|
|June 1994||exhibited in "Design Engineering '94" and started business|
|Aug. 1994||developed CATIA interface|
|Sept. 1994||developed new resin, TSR-750, for injection molding die (inj 50 pcs of ABS)|
|Oct. 1994||exhibited in "Concurrent Engineering Fair 94" with Shonan Design (a service bureau that uses 3 TS machines) and Nissei Plastic Industrial (a manufacturer of injection molding machines)|
|March 1995||developed new resin, TSR-752, for injection molding die (inj. 200 pcs of ABS)|
|April 1995||exhibited injection molding machines in "Inter Mold 95" with Fujitsu, Shonan, and Fanuc|
|Sept 1995||exhibited in "Concurrent Engineering Fair 95"|
Sales Performance 1992 1 machine 1993 1 machine 1994 7 machines 1995 20 machines (estimate) Total 29 machines Employees Sales and service 8 Planning 3 Engineering/software 4 Engineering/hardware 3 Manuf./hardware 8 Manuf./scanner 7 Resin R&D 3 Resin Manuf. 2 Total 44 (including some subcontractors)
There are four market segments of relevance to Teijin Seiki, as shown in Table TS.1.
Of the "big 3" in Japan (CMET, D-MEC and TS) Teijin Seiki representatives believe that TS has the best resins; that CMET is #1 in hardware; and that D-MEC is #1 in software.
The team asked the Teijin Seiki representatives why there is no Japanese development of powder and other alternative processes. They replied that companies are interested in high accuracy, which means liquid systems.
Resins are perceived by TS to be a competitive advantage (see Competitive Strategy section). Table TS.2 shows the various TS resins with physical properties, together with a column showing comparative properties for ABS.
On the other hand, Teijin Seiki uses layers as thin as 50 microns and so must guarantee that there are no particles larger than 40 microns. The team saw this resin running in one of Teijin Seiki's machines. It appeared possible to spread a new layer in approximately 15 sec (see Machine section). There was some bubble formation by the spreader mechanism; however, these bubbles formed far from the part and perhaps are not a big issue. TSR-752 is UV- and then thermally post-cured (the UV cannot penetrate well due to the filler). The resin must be stirred periodically (although not during a build) to keep particles suspended.
Teijin Seiki's TSR-800 resin is intended to provide properties as close as possible to those of ABS. The goal is to provide resins that achieve properties similar to those of ever more demanding materials. Managers recognize that ABS is a commodity plastic and that they will have to graduate to engineering plastics such as Nylon-4,6 and then to super engineering plastics such as PEEK.
They view the requirements for resins as follows:
TS does not have an epoxy resin at the present time. DuPont has an epoxy resin (6100) that TS managers hope to use. However, they recognize that it does not have sufficient performance, being inferior to the CMET epoxy resin. As noted above, 3 employees are in resin R&D and 2 in resin manufacture. Resin efforts are also subcontracted to 2 companies. Resin R&D is also supported by DuPont (which is Teijin Seiki's resin supplier) and Tokyo Institute of Technology (TIT). TIT is advisor to TS and in the future they will codevelop a new resin with Professor Endo.
As noted above, Teijin Seiki's claim to fame is a filled resin intended to allow the creation of tooling for injection molding directly by Soliform. The tooling is intended for prototype applications only. The JTEC/WTEC team was shown several examples of tools created with TSR-752. One of the tools was said to be unfinished. This tool exhibited a bump in the top surface of the part which reappeared consistently on all top surfaces of this part. The team was also shown several examples of finished tools. In one case, the tool had been coated with a fluoropolymer (Shonan design developed the coating), which was said to increase the usable life of the tool.
The team also saw tools that had been finished but not coated. In use, the tools are clamped to metal backing plates into which cooling lines are built. Ejector pins are machined into the tooling. The tools showed evidence of degradation, especially in the form of chipped corners. Most if not all of the molded ABS parts the team saw showed a great deal of flashing in shut-off areas and at parting planes. There was evidence of short-shotting (or perhaps trapped air) in some of the molded parts. The molded parts' surfaces also showed evidence of roughness that looked as if it were due to tool wear. Parts were injected at a temperature of 170-180°C, and the tool was maintained at a temperature of 70-80°C.
TS also does vacuum casting using SOMOS( 2100 as the tool material. The casting is performed at an elevated temperature, at which temperature the 2100 resin softens a bit, aiding in the demolding process (like silicone rubber, but less extreme). The TS representatives cited an example product made by vacuum casting (a portable telephone case) where 40 hours is spent on CAD, 14 on Soliform, and 1 hour on cleaning and post-curing, leading to urethane castings in a lot of more than 20 pieces.
The INCS site report (pp. 69-72) contains additional comments on TS tooling efforts.
TS uses 4100 resin to create patterns for investment casting. The company obtains funding from the Organization of Small and Medium Enterprises (SME) for its work on investment casting.
TS has made medical models and had some to show the team. This does not represent a major source of income for TS at the present time.
The team's hosts felt that at the present time, resin accounts for more than 50% of part variation from intended dimension, software accounts for 30%, and the rest can be attributed to the machine. They considered machine improvements to be the least important factor governing accuracy at the present time. See Machine section for more on accuracy.
The basic architecture of the TS machine stems from DuPont technology. This includes the following basic features:
TS uses an argon laser with 1 W output from Coherent (45 amps). It uses the 365 nm line.
The scanners are of TS design and are manufactured at a TS plant in Gifu, which is 400 miles from the site we visited. The scanners are digital. TS uses purchased servo motors (motor/encoder package) and mirrors. They are said to be able to scan at speeds of up to 24 m/sec while maintaining maximum errors of 40 microns (simultaneous specs). Note that their scanning is raster, and that this relieves some aspects of the positioning challenge. More typical speeds are 15 m/sec. TS makes its own controller. The company does not have flat field optics. It does distortion correction by calibration and with software.
The JTEC/WTEC team spent considerable time discussing the issue of accuracy of beam location. What is clear is that the digital resolution (corresponding to one pulse) is 6.25 microns. Further, TS engineers "control to within 12.5 microns." When asked how accurately a 10 mm dimension would be scanned, the team's hosts answered that it would be ±12.5 microns. In the TS catalog maximum error is specified as ±40 microns, but in correspondence, Takakuni Ueno specified that this means only digital motor error: "We use AOM shatter also for positioning assist, and our real positioning error is according to digital motor positioning and AOM positioning. When the digital motor is overrun, AOM shuts the laser beam at a more accurate position, so we can control [to] ±12.5 microns [when] positioning [the] laser beam."
Interestingly, when the tooling resin is run, a wave of material is visible ahead of the first knife. After the recoat is completed in about 15 sec, the mechanism immerses into the resin again. When lower-viscosity resins are used, a brush replaces the wire mesh basket. As the team watched the tooling resin run, the build area was maintained visibly above the free surface of the resin (perhaps 100-200 microns). This is deliberate. TS researchers have experimented with above, at, and below the free surface, and the best results depend on the resin. In any event, this may represent a control issue for TS.
TS managers feel that they must reduce the cost of their machine to 1/3 of current level. Currently a model 500 costs $500,000. They are interested in other lasers, including HeCd, CO2, excimer, and others. The laser running cost must be reduced.