Site: CMET, Inc. and Asahi Denka
7F Kamata Tsukimura Bldg.
15-8 5-chome, Kamata, Oota-ku
Tokyo 144, Japan
Tel: (03) 37396611; Fax: (03) 37396680
Asahi Denka Kogyo K.K.
4-1, Higashi-ogu 8-chome
Tokyo 116, Japan
Tel: (03) 38922111; Fax: (03) 38193243
Date Visited: 11 December 1995
JTEC/WTEC Attendees: Paul Fussell (report author), K. Narayanan
Dr. Satoyuki Chikaoka
Dr. Kazuo Ohkawa
CMET was founded in 1988, principally by Mitsubishi, building upon the developments of Dr. Yoji Marutani (of Osaka Sangyo University) in liquid photopolymer rapid prototyping. Hidetaka Narukawa joined CMET in 1990, and CMET was incorporated on 2 November 1990.
Dr. Marutani commented to the JTEC/WTEC team that he still provides some limited consulting to CMET, but primarily on philosophical rather than specific machine design issues. This is consistent with the team's observation of Dr. Marutani's lab and research interests. CMET is currently owned by several companies:
Mitsubishi 54% Asahi Denka 20% NTT/Data Communications 20% Various machinery manufacturers 6% (including Toyo Denko Saso, and Y.A.C.)
The owners each contribute to CMET. For example, Asahi Denka provides all of the resin, and NTT provides software expertise.
CMET representatives provided the team with a copy of the May 1995 Rank Prize (London) for optoelectronics, which was jointly given to Hideo Kodama and to Charles Hull (Kodama received £15,000, and Hull received £10,000).
As of September 1995, CMET had sold 70 machines in Asia. This compares to about 30 sold by Sony/D-MEC, 20-25 by 3D Systems, 10-15 by Teijin Seiki, and 24 by Denken. CMET's sales rate is shown below:
1988 89 90 91 92 93 94 95 (June) 2 8 4 14 7 10 9 13
The Japanese market for rapid prototyping devices has apparently been as follows:
1992 93 94 95 18 22 54 29
CMET uses technology that is similar to other liquid photopolymer systems. The company's engineering appeared to be at Japan's typically high level of excellence.
CMET's original work was an XY plotter mechanism with the laser light being delivered to the working head by fiberoptic cable. Professor Nakagawa still has and operates one of these devices in his lab at Tokyo University. In some versions of this device, CMET engineers were able to vary the spot size of the beam from 0.1 mm to 2 mm. The SOUP 850P (P for plotter) has a working volume of 850 mm x 600 mm x 500 mm and is still used by the sand casting industry to make master models for green sand pattern boxes. There are both HeCd and Ar-ion (ArI) versions of this device.
The galvanometer-based devices (the SOUP 400G and SOUP 600G) are newer designs, again using either HeCd or ArI lasers.
One of the newest machines is the SOUP 250GH, which has the ability to vary the spot size (perhaps through a zoom lens arrangement). CMET researchers have also developed the SOUP 1000GA, currently only for the Japanese market (one of the dimensions in this device is 1,000 mm). The team's hosts indicated they were working on a SOUP 6000GA device. They also indicated they were able to control the shape of the spot at the liquid surface as they traversed the area of the vat. (This was perhaps accomplished by the use of a flat field lens in the optic path.)
CMET researchers indicated they had a recoating mechanism that was more sophisticated than just a doctor blade; however, they had only partially solved the trapped volume problem. They understood that trapped volumes are difficult because of resin flow under the recoater as it traverses the trapped area. The recoater did not appear to have any auxiliary hoses, tubes, or other active means of providing resin to the device.
The liquid level of the vat is maintained by placing a machined weir at the front of the vat; liquid simply flows over the weir into a channel surrounding the vat.
The CMET representatives felt their scale factors for building parts were very small, but the JTEC/WTEC team was unable to understand how they applied either scale factors or beam (line) compensation factors.
CMET was unwilling to provide dimensional information taken from a North American 3D Users Group test part (the part originally designed by Ed Gargiulo of DuPont). CMET managers felt the Japanese customer had little interest in this part, and they preferred to compete with other equipment manufacturers by building geometries provided by the customer. (This was consistent with views expressed by other vendors as well.)
CMET management understands that software is critical to the company's success. It currently has an internal version of the software used to prepare and control builds that will automatically add support structures, but customers seem to be required to manually design the supports. The team's hosts appeared to be aware of, but uninterested in, automatic support generators such as Magics or Bridgeworks. In part, this might be due to the fact that Bridgeworks is distributed in Japan by INCS, Inc., a competitor. The STL file repair approach is topology-based, but the team was unable to understand what problems it is able to repair.
CMET's principal competitor is the small Japanese prototype shop. There are approximately 3,000 pattern makers able to convert a relatively complex part drawing into an accurate part in a three-day turnaround; CMET could only be competitive with these shops if the original design had been done in a solid modeler (a seat of Pro/E costs about $40,000) and had relatively complex surfaces.
CMET had little interest in other technologies for making plastic parts: liquid photopolymer-based systems were, in the opinions of the CMET representatives, as poor in terms of accuracy as the Japanese market would tolerate, and all other methods were poorer. The improved material properties of, say, a nylon-like material held no appeal: they firmly believed the Japanese user of these technologies demanded very high accuracies and would permit nothing less than this high accuracy in their shops.
CMET representatives also felt that the Japanese market requires mass production orientation, and that this implies that methods used to fabricate metal must mold quickly and accurately. Thus, CNC milling is the other principal competitor of their business. They also observed that the Japanese engineer is generally able to mentally visualize a part, and therefore does not need as many visualization or review parts to be made in rapid prototyping as does the North American engineer.
The Japanese medical market has found rapid prototyping to be too expensive. (One skull model was priced at about $2,000.) Doctors have little motivation to explore the technology, either for evaluation of a patient or as a means toward a cure.
The facility that the JTEC/WTEC team visited had three machines in place: one was an older demonstration machine, one was used to develop customer benchmarks with customer data, and the third was used to test new hardware and software.
Asahi Denka has provided epoxy-based resin systems from about 1990. The team was not able to discover from Asahi Denka how these resins differ from the current systems available in Europe and North America.
There are three primary and one experimental argon-ion-based resins: HS671, HS672, and HS673 are the primary resins, and HSX-A-4 is the experimental resin. All of these are low-viscosity materials, relatively strong, with a basically brittle nature (1-4% elongation at failure). The Asahi resins use all three argon-ion wavelengths of 333, 351, and 364 nm.
The HeCd resins include HS660, HS661, HS662, HS663, and HS666. HS662 is castable in a thin-shell method. It has an Ecrit of 3 and a Dp of 100-120 µm.
CMET was unwilling to provide either liquid or cured samples of these materials.
CMET recommends running the resin at 40°C during build (the build chamber is heated to this extent). CMET representatives expressed little interest in glass-filled resins: their objection was that cycle time to build a part is much longer with the glass-filled material.
The team saw a casting part, a model of a shoe sole, at the CMET facility. CMET is able to build shell parts (perhaps similar to 3D Systems' QuickCast parts) with an internal T-shaped support. The support is regularly placed:
The team explored the CMET representatives' opinions of the 3D Systems patent application (JP 3-9660 [B2] opposition document) that is about to be issued as a Japanese patent. They observed that there were four "rounds" to a patent application process and that the 3D patent was just exiting round two (successfully coming out of opposition). They professed to be aware of the patent situation, but did not feel the current situation would affect their business. They said that if the 3D patent passed round four, they would have to cross-license with 3D Systems; and they felt this would be a "hard negotiation."
Team members asked why, in the opinion of our hosts, other large Japanese companies had rejected the panel's request to visit and discuss rapid prototyping. They believe that, fundamentally, Japanese companies are concerned about giving information and receiving little in return. They mentioned that Toshiba has an internal show, and that it would not be useful for the team to see Matsushita. CMET representatives, in particular, expressed little interest in exporting to the U.S. market, for patent reasons (although they indicated that they have an exclusivity agreement for distributing the Asahi Denka resin and their policy is to not sell the resin in markets other than those they are selling in), and therefore they were not disposed to provide too much information about their rapid prototyping technology.
Mr. Narukawa, Mr. Hayano, Dr. Chikaoka, and Dr. Ohkawa were extremely hospitable hosts and gave us considerable insight into elements of Japanese culture and the Japanese business environment.