The face of a new technology: The inside story of Philips's decision to launch the Digital Compact Cassette and its engineers' struggle to build it.

'When you hit a wall with a sledgehammer, you do tend to think more clearly,' says Gerry Wirtz, recalling the day in 1988 when he was knocking down a wall in his house. Wirtz is an electronics expert who had been pulled out of his work on security devices at Philips, the Dutch electronics group based in Eindhoven, to head a group to predict the future of digital recording and plan the new products that Philips should be making. As he worked on his wall, 'I asked myself what my neighbours would buy,' he recalls. 'They might have a CD player at home, and a cassette player in the car. They would not be interested in a completely new recorder just because it was digital. But if I came to them with a digital machine that would play a new type of cassette and it also played their old tape - now that they might like.'

Out of this moment of inspiration was to come the Digital Compact Cassette recorder. Launched late last year, DCC offers consumers the opportunity to make digital recordings on a new kind of tape with the quality of a compact disc. And just as Wirtz dreamt, it has 'backwards-compatibility', so that it can play conventional analogue tapes. All eyes are now on the battle for market share between DCC and Sony's new Minidisc system (see 'Head to head in the recording wars,' New Scientist, 17 October 1992). But behind DCC is an even more fascinating series of struggles as Wirtz and his supporters fought to win Philips's support for the machine, research engineers battled to find new ways to compress data onto tape without losing sound quality, and Philips executives tried to win support in the music industry.

The story started in 1987, when Cor van der Klugt, Philips's president, brought in Jan Timmer to take charge of consumer electronics. As head of Polygram, the record company of which Philips owns 80 per cent, Timmer had an excellent track record. At Philips, his brief was to make the division more profitable - a great challenge because all consumer electronics companies were facing hard times, even before the recession. Most people who wanted and could afford a colour TV, video recorder and hi-fi, already had several. Equipment was now built to last, so manufacturers could only make money by continually persuading people to buy new products.

At the time of Timmer's appointment, Philips was still pinning its hopes for future profits on DAT, the digital audio tape system intended to replace the analogue compact cassette by providing two hours of CD quality stereo from a cassette the size of a thick credit card. It had DAT recorders ready to sell, but the record companies were angry about the new technology. They said it made home copying of CDs too easy and would let people 'clone' music, with each digital copy as good as the original. The leading record companies refused to release prerecorded music on DAT until the copyright issue was resolved, forcing Philips to delay the launch. All the Japanese electronics companies were holding back, too, for fear of triggering a wider trade war with the West.

Concerned that DAT would never take off, Timmer started looking for other new products. The Japanese tend to throw as many new ideas at the home entertainment market as their research laboratories dream up, in the hope that some stick. But Trimmer wanted a more focused approach. Market research told him that consumers wanted digital recording machines, so at the beginning of 1988 he set up a working party to predict the likely future of digital recording. He took Wirtz away from the DAT project, and made him coordinator of this Future Digital Audio Recording Group.

Inside Philips there was a strong feeling that the future home recorder would use blank CDs. Everyone expected the working party to confirm this within six weeks. Instead the party took four months over its deliberations, and then said that Philips should reject the CD option altogether. This decision did not go down well with the CD divisions within Philips, which had invented the technology. But the facts spoke for themselves.

A recorder that uses 12-centimetre discs would not be portable. The high-power lasers needed to record onto a disc are expensive, and so the technology would attract only small numbers of well-heeled enthusiasts. What Philips wanted was a mass-market product. To Wirtz, once the sledgehammer had concentrated his mind, the answer was a digital cassette recorder that was 'backwards-compatible', so that people could play their old analogue tapes on the new digital machines. The working party quickly saw the logic. More than 200 million analogue cassette mechanisms are sold every year, most for use in personal stereos and cars. Backwards-compatibility would allow a new machine to fill the replacement slot in the market. If the new digital players were not too expensive, the public - and car manufacturers - would pay the extra for a player that plays both types of tape.

Radical thinking
In 1988 this was a radical idea. Both DAT and CD were incompatible with previous technologies. Research engineers tend not to worry about catering for the past. They also have scant regard for copyright problems, or the need to design a product with such a clear commercial target. Wirtz anticipated these developments in 1987, shortly after he started working on DAT, when he remarked: 'Digital technology is so powerful that you cannot just do as we have done before and throw it over the wall of the laboratory and into the marketplace, to see what happens. Decisions have to be taken before the technology goes on sale. Once it escapes into the marketplace it is too late.'

Philips's working party came to three conclusions. First, the price of a digital recorder that was backwards-compatible had to be as low as possible though not less than that of existing cassette recorders. This criterion ruled out DAT because of its complexity. Secondly, a new recorder could not succeed without a deal on copyright - the public would not buy new equipment without prerecorded music. Thirdly, the decision to build the new system on tape, rather than disc, meant the project might become bogged down in company politics.

Timmer quickly gave the working party the go-ahead to see if it was feasible to produce a backwards-compatible digital recorder. He also helped to solve problems. His crossover from a record company to an equipment manufacturer was the start of a bridge between the two copyright factions. Working through the International Federation of the Phonographic Industry, he persuaded the two sides to discuss digital copying. Timmer also recruited Angelique Hoogakker, a young lawyer working for the IFPI, as manager of public affairs. She was responsible for liaising with Timmer's office, the music industry and the new digital cassette working party. Hoogakker understood copyright law and knew the record companies. The working party understood the technology and talked to Hoogakker who talked to Timmer who talked to the group. So the working party obtained the resources it needed, without time-consuming formal meetings and without breaching the Philips tradition of talking only to the higher management through middle managers, often with vested interests to protect.

Wirtz, meanwhile, became the crusader for DCC within Philips. He also made it his business to find out what expertise was available within the company, making sure that the right people came together at the right time. Membership of the working party was deliberately kept fluid, with Wirtz pulling in specialists on demand. He had learnt the benefits of such a strategy the hard way, during his time on the DAT project.

With the mass market in mind, the working party decided that its new digital cassette machine should use recording heads that remain stationary while the tape moves over them. Stationary heads are potentially much cheaper to make and more reliable than the rotary heads used in the DAT recorder. This machine worked like a miniature video recorder, with its heads mounted on a drum that spins rapidly to scan the slow-moving tape, recording and reading tracks narrower than a human hair. Another advantage of stationary heads is that tapes can be copied at more than 64 times faster than normal speed. DAT music tapes cannot be duplicated by running the master and copy tapes at high speed and copying from one to the other, because the drum speed is already so high.

In late 1988, Wirtz asked Philips's research laboratory - the Naturkundig Laboratory, or Nat Lab as it is known - to make a prototype stationary head. The laboratory was able to do this in two months because, as Wirtz was aware, it had already designed stationary heads for multi-track sound recorders used to log police calls.

The working party wanted the new digital tape to run at 4.76 centimetres a second - the same speed as analogue cassette tape. This would mean that factories would not have to redesign the tape mechanisms already being built into analogue recorders. But CD-quality sound requires 1.4 million bits of data a second, which cannot be crammed onto such a slow-moving tape. The data had to be compressed to a quarter of the amount. The compression system the working party devised for DCC was born of a mix of natural curiosity, good fortune and hard work.

In 1985, Raymond Veldhuis was working at the Philips Nat Lab looking at ways of compressing digitally encoded speech. He would break down the 20-kilohertz spectrum of the audible sound signal into narrower sub-bands, and encode each one separately. (Veldhuis now works for the Institute for Perception Research in Eindhoven, funded partly by Philips and partly by the Eindhoven University of Technology.) He recalls how 'I was just interested in what I heard, and saw no commercial application'. When he heard that researchers at the University of Duisburg in Germany had been trying to compress music in a similar way, he decided to try his speech encoder on music. The results were encouraging; instead of using the CD standard of 16 bits to encode the music, his encoder needed only 3 bits.

At the end of 1989, the Nat Lab hired Robbert van der Waal, a university student from Duisburg, to do further research on compression coding, but still with no commercial application in view. He looked at sub-band masking - splitting sound into frequency bands and analysing which sounds will mask others to the ear. Van der Waal's research laid the groundwork for DCC's coding system, Precision Adaptive Sub-band Coding (PASC). It works on the same principle as the human ear, which ignores a note swamped by a louder note of the same or similar frequency.

Wirtz knew of the work by Veldhuis and van der Waal, and wondered if it might be suitable for the digital cassette. Each year, the Nat Lab holds an open day, to which all Philips employees are invited. There they see demonstrations of work in progress put together by the researchers. Wirtz made sure that the members of the working group were at the 1989 open day to see a demonstration of sub-band coding. They thought it might work, and mentioned it to Timmer. His immediate response was: 'Make me a working model in six months.'

The group used the hand-built stationary head that Nat Lab had made the previous autumn, and a large rack of prototype digital compression circuitry. To prove the feasibility of mass production they chose a cheap and cheerful plastic cassette player, put in the head and connected it to the external electronics. The make-shift recorder used ordinary tape in a conventional cassette - and worked astonishly well. The working party made a point of demonstrating it to Timmer in JJuly 1989, six months to the day after receiving the instruction to make the prototype.

No wow or flutter
Abraham Hoogendoorn and Paul van der Plas, who had both moved to the digital working party in 1988 as Philips pulled out of DAT, recall how the meeting with Timmer went: 'It was a marvellous demonstration. We used a very cheap cassette deck, which gave terrible wow and flutter when playing analogue tapes. Then we put in a digital cassette and shook the cassette deck violently. There was no wow or flutter. The timing circuits of the digital decoder just compensated for the errors.'

Timmer was thrilled with the DCC prototype. And he had some good news for the working party about copyright. At an IFPI meeting in Athens during June, the delegates hammered out a deal on copyright. Under the terms of the Athens agreement, the music industry agreed to withdraw its objections to the sale of DAT recorders provided that the electronics industry agreed to incorporate technology that restricted copying. The technology, called Serial Copy Management System or SCMS, lets the owner of a digital recorder make one digital copy, but then prevents anyone making a digital copy of that copy. Sony is now the chief supplier of DAT machines, mainly to professional users who can afford them.

Although the Athens agreement covered mainly DAT, it was not long before it was extended to embrace every conceivable method of making a digital recording. Timmer, however, wanted to be absolutely certain that the music industry would back the new recording format before giving the go-ahead for the mass production of DCC. In late 1989, he arranged a demonstration for Bhaskar Menon, then head of IFPI and EMI Music. At the IFPI's next board meeting, Menon advised the chief executives of all the world's record companies to go to Eindhoven and see for themselves what was on offer. One by one they went and Timmer told them: 'If you are interested, then help us specify the system. Without your support we will not go ahead and make it.' All were enthusiastic, except CBS.

Through his contacts, Timmer also persuaded a hardware manufacturer, Matsushita of Japan, to come to Eindhoven to see the demonstration. Matsu-shita was impressed and quickly gave its support to the DCC project - support crucial to Philips if it wanted to sell the system in the all-important Japanese market. Soon after, Timmer arranged for Sony engineers to see the same demonstration. The Philips engineers recall how the Sony engineers were 'visibly shocked' at the sound quality available. But by then, Sony was secretly developing Mini Disc - a miniature, recordable version of the compact disc - that was to be a competitor for DCC. And Sony owns CBS.

In Japan, Hoogendoorn and van der Plas found Matsushita's factory engineers still busy with DAT and all the manufacturing problems that went with its complex design. The Dutch engineers showed them their DCC prototype. 'Very simple, very simple, very simple,' said the Japanese. 'It was nice of them to say that, but they were not so right,' says Hoogendoorn.

DCC seemed very simple, and its backwards-compatibility reinforced this impression. So the working party got a rude awakening when they tried mounting their stationary head and compression circuitry on other cassette machines - and discovered that the sound would mute unexpectedly. It took a long time to find out why: in the demonstration model, the tape, which is always slightly abrasive, had cut a channel into the plastic. The machine worked only because the channel held the tape at exactly the right angle to the tiny heads. Once the problem was identified, engineers arranged for appropriate guides to be designed for mounting on the head.

Ready for production
Meanwhile, the electronics engineers thought that when they did get sound from the tape, it was good. But when recording engineers from Polygram's Dutch recording studio in Baarn and from Decca in London heard the system in action they said it sounded 'awful'. Veldhuis and van der Waal continually changed the compression coding algorithms, which Philips then checked with the Polygram music engineers. But still the music engineers weren't satisfied. Then Wirtz enlisted the help of Adrianus Houtsma, professor of psychoacoustics at the Institute for Perception Research, who identified the problem in three days. The laboratory researchers had been testing each of the stereo channels separately, while the music engineers were listening in stereo. While the music signal that spreads between the two stereo channels is correlated, any background noise is random and different in each channel, standing out when both channels are heard together. The solution was to use less compression in the low-frequency range, which produced a signal that masked the noise.

Despite the setbacks, Hoogendoorn looks back with obvious pleasure. 'It was a marvellous time. We were just sitting together as if we had nothing to do with Philips. We made all our own decisions. The important thing was that Timmer believed in the project. If you do not have that kind of confidence from the top then you get small groups opposing each other.'

At the end of 1990, Timmer, by then president of Philips, felt he had enough support from the music industry to give the order to 'build' DCC. Now the working party had to concentrate on the problems of mass production.

For encoding and compressing the audio signal, Wirtz took the large prototype circuits designed by the Nat Lab to the Philips semiconductor centre at Southampton. Here the circuits were converted into the photolithic designs needed to make microchips at the Philips fabrication plant in Nijmegen.

Philips could not afford to build its own factory to make the stationary heads, but if the market for DCC took off it would need millions of heads a year to satisfy demand from other firms wanting to make DCC recorders under licence. The heads rely on thin-film technology, and are also made like microchips: a thin layer of semiconductor material is deposited on a slice of silicon, and etched in the shape of the heads. One of the members of the working group saw that the DCC heads have a lot in common with the heads produced by the computer industry for use in hard disc drives. Philips struck a deal with Seagate, an American manufacturer of hard disc drives for computers. Seagate now makes the thin-film microchips and Philips handles the tricky task of assembling the head, which involves mounting the chips on a block and connecting the etched parts to tiny wires with microscopic accuracy. Everything was now in place to satisfy whatever demand exists, with a head that was easy to insert into a mass-produced recorder.

The original plan was to launch DCC in spring 1992, and then September. But as summer approached, it looked as if there would not be enough of the vital thin-film heads. The manager of the Eindhoven factory that assembles the head went to his workforce and asked them to cancel their summer holidays and make heads instead. Eighty per cent agreed.

Then, with only weeks to go, Japanese engineers found that some of the prerecorded digital music cassettes that Philips had made to support the launch seemed faulty. In hot and humid conditions they sometimes shed residue onto the heads, causing digital errors and muting the sound. So, again, Philips delayed the launch, recalled all the cassettes and started duplication all over again, with a slight modification to the inside of the cassette.

No one yet knows whether DCC will be a success, or whether Sony will capture the public imagination with its Mini Disc. But the outstanding feature of DCC that could make the format a long-term winner is its backwards-compatibility. Wirtz explains the sea change in policy: 'Society is now overloaded with innovation,' he says. 'Producing a nice little new toy is not enough. Society is more competitive, so the window of opportunity for getting a new product accepted is smaller. It took seven years for the world to find out that the cassette recorder was not just a recorder, it was also the ideal mobile source of prerecorded music. The early video recorders did not even have a tuner to tape programmes off-air. In those days the industry had some time to get it right. Now you have to shoot at exactly the right time or you are out.'

From issue 1882 of New Scientist magazine, 17 July 1993, page 30