Taiwan Turns a Page--How Two Companies Produced Award-Winning Products
Elaine Chen / photos Hsueh Chi-kuang / tr. by Phil Newell
June 1994
Personnel are the essential condition for R&D. But for successful products it is necessary to stir in many other factors. The companies which this year won the national product image Gold National Awards of Excellence all have very strong R&D structures. But behind the perfect appearance of the products which won awards lay many disappointments and failures. Take for example the fingerprint recognition system developed by Startek Engineering--they had a setback when they misjudged the market. And the development team for Brilliance computer monitors at Philips Taiwan were for a time "pariahs" of the company because they fell so far behind schedule.
How did they transcend their difficulties? What did they learn from their experiences?
At the Celebration of Excellence Awards Ceremony this year, the two most watched companies were Philips Taiwan and Startek Engineering.
Startek originally was just one of many small, unknown companies in the Hsinchu Science-Based Industrial Park, and its only product is a fingerprint recognition system, so how can it carry the heavy burden of raising the image of Taiwan products? Philips, on the other hand, is a famous international corporation, so why would it work so hard to be recognized within Taiwan as making superior products? Can Philips represent "Made in Taiwan"?
"Philips invested in Taiwan 27 years ago, and certainly at first the Taiwan branch was just an offshore production unit with cheap labor," says Cheng Cheng-wen, executive vice president of Philips Taiwan. But because Philips' chairman served as the manager of the Kaohsiung factory, he understood that Taiwan has many outstanding engineers, and could do much more. Therefore they established a design department beginning in the 1980s, and now, in the 90s, Philips Taiwan has been set up as an independent operating entity. It has fully independent units for everything from market research to sales to R&D to manufacturing to postsales service. "It is indubitably and completely' Made in Taiwan.'"
Cheng further argues that the world already recognizes the quality of Taiwan computer monitors, and putting the "National Award of Excellence" label on the product actually has a positive impact in the market.
Last year, Philips applied to participate in the awards, but was excluded for being a "foreign enterprise." They didn't despair, and came back again this year with all the data to prove that they deserve the "Made in Taiwan" status, and in the end won the Award of Excellence with their new 17-inch high-resolution, low-radiation color monitor christened the "Brilliance."
Startek, on the other hand, is a classic case of an enterprise started on the side by high-technology academics.
"Our company only has 30 people. The successful development of the fingerprint recognition system was a result of the collective efforts of the production, management, and academic people. Startek wants the whole world to know that we are not only 'made in Taiwan,' we are created in Taiwan!" These are the effusive remarks made by Startek founder and Tsing Hua University professor Hsu Wen-hsing at the Celebration of Excellence Awards Ceremony.
"In fact the market has been waiting for this type of biotechnology information product for a long time," says Michael Lee, Sales and Marketing Department Director at Startek. People have always hoped to be able to distinguish identities based on unique biological features, so that people won't have to carry so many keys and credit cards and ID cards, which can be lost or stolen or falsely used at any time. Of the unique biological characteristics people have--fingerprints, the retina, and DNA--fingerprints are the most convenient. Moreover, there is already a theoretical foundation, and it is internationally standardized and interchangeable information. Thus, ten years ago, when he was still studying in the Graduate Institute of Electronics at Tsing Hua, he got interested in computerization of fingerprints. Later, when he went to study at Keio University in Japan, Hsu did specialized research in image processing automation.
How can a computer differentiate fingerprints? Simply put, the principles of optics are applied to turn the fingerprints into images, which are recorded in a fingerprint memory. When new fingerprints are entered in, image processing techniques are used to find the special cross-hatching, swirls, and end points of each print. These special features are then compared to match up fingerprints.
Later the Image Processing Laboratory at Tsing Hua, under Hsu's guidance, did basic research and feasibility studies of fingerprint differentiation. With the research in hand, Hsu felt he could open a company and use a corporate structure to proceed with development and production. Therefore he brought together a tightknit group of friends from college days to form Startek.
The development of any product is rooted in human needs. It's just that some products also carry heavy responsibilities for companies while the products are being developed. "As of 1993, and still today, Philips has been second in production volume of computer monitors only to Samsung of Korea. In terms of value-added we are also second in the world, behind only NEC," states Cheng Cheng-wen. Philips Taiwan, aiming to make itself the worldwide audiovisual center for Philips, decided that it would aim to be number one in the world in both volume and value before 1995.
Therefore Philips Taiwan began market research and compared world-class products, and then the marketing and R&D departments jointly determined the product specifications for the future.
"We wanted to have all the functions that others lack, and we wanted to insure that the things we produce have zero defects," proclaims S.T. Huang, manager of Electrical Engineering at Philips Electronics Industries (Taiwan). They aimed to sell worldwide, and wanted to sell 200,000 units in the first three months on the market. If there were problems with the products, they would have no time to go back and slowly fix them, and they would end up a dismal failure. And because electronics products have a very short product cycle, they felt this product would have only about a two-year lifespan, so they decided they would have to develop it in a year.
Speed and product quality are often at odds with each other, and it is hard to do both well at the same time. Philips' method was to use a comprehensive system.
After many years of accumulated experience, Philips has developed a patterned way to compete. For example, a special team is set up for each new product; the team includes colleagues from twelve different departments, such as electronics, machinery, quality control, purchasing, trial production, testing, automation, product strategy, and post-sales service....
Of these, product strategy must figure out what types of products the market is looking for, while the after-sales service people have to look at product while it is still in development and determine whether, from the consumer's point of view, it will be sufficiently user-friendly or easy to maintain. Trial production people are at the juncture of development and manufacturing, and judge whether the products will be easy to mass-produce, whether there are any problems with the production process, and how efficient production will be.
Startek, on the other hand, has none of this.
"The company has been formally established, but in terms of R&D we're still more like a laboratory," says Gau Ya-wei, manager of the R&D department. For example, they are all former schoolmates, and they live at the company. Often everyone works together into the wee hours of the morning, then comes in a little late the next day. "We're not like your typical company, where we have a strict plan and schedule for developing a product. You could say that we are exploring new territory with the fingerprint recognition technology. First we have to get the technology completed, then we'll talk about other things," concludes Gau.
Nevertheless, it has been a hard road for Startek in commercializing its product. "Academia is too far away from commercialization," explains Hsu Wen-hsing. For example, academic research doesn't have to take into account outside factors, like the temperature, humidity, or what will happen if something is knocked around. They just make a sample in the lab, which is an "incidental success." But, he continues, "if you want to turn it into a product you have to be able to guarantee the stability of the product under a wide variety of circumstances."
For example, products age, so there must be lifespan tests in order to get a handle on what the product will be like after two or three years of use. It may be necessary to change some parts whose functions are less than ideal or to throw out part of the design.
Optics handling is one of the weakest links not only for Startek but for Taiwan as a whole. Fortunately, just at the right moment the Industrial Technology Research Institute founded its Optics Laboratory. Startek cooperated with ITRI to develop a very high grade optical camera, marking a major breakthrough in terms of fingerprint hardware. Skillful use of external resources has been a key factor which has allowed Startek to remain small while thinking big.
Startek was founded four years ago, and has been in the Hsinchu Science-Based Industrial Park for two years. "In these two years, half the companies founded at the same time folded," says Julie Shih, general manager of Startek and also Hsu's wife. Pointing out the window, she sighs that most investors only provide for a couple of years, and after that time their money has run out. If there has not been any result by then, few investors will put in more money, and the company will probably fold.
Therefore, Startek is very frugal with its daily expenditures. They only rent an office-cum-factory of 210 ping, and there are no more than 30 people in the whole company. Of these, almost all the R&D people are fellow alumni from Hsu's days in graduate school at Tsing Hua, and they have long been accustomed to working together, thus saving the time that might be spent in training and learning. As for the costs of product development, because it is considered priority technology by the government, one-half of costs are subsidized.
"Our company doesn't have anyone here for the salary or benefits. If you want responsibility, and are willing to say what you think, this is for you," says Kang Chii-yuan, senior engineer in the R&D Department. They are all still young, and don't really care how much money they make; what's important is to get plenty of hands-on experience. "Working at Startek presents a very rare opportunity. This is the first time that we in Taiwan have the chance to lead the world in technology."
The problem for Philips Taiwan isn't money, but the time pressures of commercial competition.
"Basically this type of product has a high degree of difficulty. Moreover, for all our tests we take the highest standards set in the US and Europe and then add 10%. Thinking back, for 49 weeks straight--a whole year--there has been no Saturday or Sunday. We were still telling purchasers to bring in materials even on Chinese New Year's Eve, and we went right back to overtime even before the holiday was over," recalls Charles Hsieh, project manager for Brilliance monitors.
During the development process, the Brilliance was put through more than 200 tests including those for magnetic interference, shipping, heat and humidity. Under the "money is no object" principle of testing, they went through 350 prototypes.
Naturally there were a lot of glitches in the process. For example, some parts they first wanted to use, made by a certain Japanese firm, didn't meet their needs, so they had to change suppliers. But this meant they were compelled to adjust many specifications, which cost them a lot of time.
These weren't the only things troubling the R&D people. The different viewpoints of the various departments, and their unwillingness to compromise, also created headaches.
Charles Hsieh notes that in many local enterprises the R&D department is the largest, and other departments have to accommodate it as much as possible, so it has a lot of flexibility.
"But in Philips Taiwan, every department presents a challenge to the other." He takes a case in point: When the production department felt that the design was too expensive to produce, then there was no way to get around them. Yet sometimes there were simply problems getting the design just the way they wanted it. Often departments had fierce arguments.
In order to get the various units to cooperate, meetings became essential. Eventually these went from one a week to once a day at 3:30.
This situation created a lot of pressure on the R&D people. Electrical engineer Liu Feng-yi often felt that, "as soon as I had settled down enough to get back to work, somebody would be right back arguing with me." But there is undoubtedly some value in such a system--design problems can be uncovered through strict supervision and balancing effects.
Because the product was a difficult one from the start, and demands were made so high, the project began to fall behind schedule. Later, after the prototype had been sent to the laboratory, many problems cropped up after 2000 hours of use. The case became extremely hot to handle, and one of the two project managers tendered his resignation and left Philips.
As a top level manager, S.T. Huang, who has had over 20 years of R&D experience with the company, saw that things were not going well, and decided to take the helm himself. He went right to the front line to directly work with everyone to solve problems. Besides regular hours and overtime, he often brought problems home with him, figured out ways to resolve them, and then on the following day invited the relevant people to get together to try out his ideas. To overcome initial design problems, he spent a couple of months trying out new materials and designs and putting in a lot more capital. But they had already lost several months, affecting the time when they could go on the market in Europe.
"The R&D Department always had the custom of throwing a party whenever a new product came out to thank everyone for their efforts. But no one suggested that after Brilliance monitors were done," says Charles Hsieh. From the company point of view, the delay in R&D cost commercial opportunities, and that is a very serious matter.
Later, when analyzing the reasons why progress on the monitors slipped out of control, the main problem was in the initial evaluation--the technological level demanded on the product was too far beyond the original foundation. There were too many variables that could not be controlled, yet no additional time allowance was made. Also, before the process began, there was no proper risk analysis, and alternative plans should have been thought up at the outset to get around sudden problems.
Philips divides its research plans into A, B, C, and D grades depending on the level of new technology needed. After this experience, it was decided that top-level managers--who originally were to do only administrative work--should personally participate in the technical side for Class A projects. "This makes the R&D process much more stable," says Charles Hsieh.
For Startek, there were technical breakthroughs on the fingerprint recognition technology after seven or eight years. They won a number of domestic and international prizes or their patents. Nevertheless, between research and production there were many things that were difficult to control. And in the later stages of commercialization, Startek met a serious setback.
Just as the technology was becoming mature, investors urged Startek to commercialize it as soon as possible and to get it on the market. At that time many people advised Hsu Wen-hsing to produce fingerprint time-clocks for employees and door lock systems. But the market response was poor.
"I thought for half a year, and in the end I discovered that the problem was with the definition of the market," says Hsu, who drew his own conclusions from this: For every product there is a period of concentration, of dispersion, and of personal use. From the example of computers, in the 1960s and 1970s only specialists used computers. In the 1980s came personal computers, and now there are portable ones.
"Things have to proceed a step at a time. The technology must be mature, and only when the price is reasonable can it become a commercial product for personal use," argues Hsu.
Applying this lesson to the fingerprint differentiation system, this is still the era of specialists, when only police and intelligence experts are using them. Further, there are still problems with the technology, such as children not knowing where to put their fingers, or the optical lens not being able to get a clear picture for the computer if a person's hands are too dry. And then there's the price--NT$300,000 (more than US$10,000) is rather expensive for an employee time-clock.
Hsu Wen-hsing feels there were reasons for this failure. He has decided to allow his fingerprint reading technology to be used as a part for others to develop other products. "We can't go and decide what, say, Japanese might need in their daily lives. A small company like ours doesn't have the ability to advance into the global market."
After this redefinition of goals, plus the appearance of the 486 personal computer on the market, so that small computers now have the capability to rapidly handle the Startek fingerprint identification, prices have dropped to NT$100,000 per unit, and orders are beginning to come in.
For example, the largest security company in South Africa installed the Startek fingerprint system in their mobile cash trucks to prevent deception in collection of social welfare payments. The trucks can now go anywhere in the country, even to the smallest village, and all people have to do is stick their fingers down, have their identity confirmed, and collect their money. Indeed, fingerprint technology could be applied to anything that requires personal identification, such as safe deposit boxes, car locks, or identification cards.
Hsu believes that if Startek simply continues to make progress on fingerprint recognition technology, they can play a role similar to that of Intel, which has produced the 286, 386, and 486 processors one after the other while allowing computer manufacturers around the world to develop products. "In the past it was always the advanced nations that produced the core technology and the developing nations that did the assembly, but today we're turning that around!" says Hsu, unable to conceal his pride.
It is particularly timely for both Startek and Philips that they received the National Award of Excellence this year.
Initially the Brilliance monitor was not so-well received in the company because of the problems with the schedule. It was not expected that the product would win a series of awards after finally coming out. It not only won the design award at the Hanover Computer Fair, it finished first in many critical functions in comparisons of more than ten of the world's leading monitors done by PC magazine. It was even more shocking when it won the Award for Excellence in Taiwan. The boss happily treated everyone to dinner, so that the project team finally got its celebratory party.
"I think they really deserved it," says project manager Charles Hsieh, who was known as the "set father" (father of the product) inside the company.
As far as R&D people are concerned, there is no perfect product, and there's always the possibility of improvement. At the same time as they move forward, there are others yapping at their heels. The greatest challenge now is whether or not they can maintain their leadership. This is like when people asked the Philips Taiwan R&D people: Was winning the prize and getting recognition from the president like eating sugar cane-sweeter as you go along?
"Not at all. New challenges are coming." Charles Hsieh says this in a quiet way, and the determination of R&D people to take on all new challenges rings clear in that sentence.
[Picture Caption]
p.16
The Giant Corporation raises their Award for Excellence flag every morning to remind its staff to continue to pursue excellence.
p.18
After more than 200 tests, the "Brilliance" monitor produced by Philips has really been through the wringer.
p.18
Strengthening R&D is the road Taiwan business must take. (photo by Pu Hua-chih)
p.20
The Startek team that developed the fingerprint recognition system (including Hsu Wen-hsing in the dark suit and his wife Julie Shih at right) pose with their many prizes.
p.21
The Peruvian authorities are already using Startek's "excellence Made in Taiwan." (photo courtesy of Startek)
p.22
R&D is an endless road, and it is always necessary to be prepared to meet the next challenge. (photo by Cheng Yuan-ching)

After more than 200 tests, the "Brilliance" monitor produced by Philips has really been through the wringer.

Strengthening R&D is the road Taiwan business must take. (photo by Pu Hua-chih)

The Startek team that developed the fingerprint recognition system (including Hsu Wen-hsing in the dark suit and his wife Julie Shih at right) pose with their many prizes.

The Peruvian authorities are already using Startek's "excellence Made in Taiwan." (photo courtesy ofStartek)

R&D is an endless road, and it is always necessary to be prepared to meet the next challenge. (photo by Cheng Yuan-ching)