The year 1979 is one of the most important years in CD development history. World renowned Dutch electronics manufacturer Philips brought a prototype CD player to Japan to demonstrate it. The team from the Netherlands included Lou Ottens, technical director of the Philips Audio Industry Group, Joop Sinjou, also from the Audio Industry Group, and chief engineer Jacques Heemskerk. They visited each of the major manufacturers in turn. The intention was clear: Philips was thinking about proposing a new global standard and Japan, brimming with prominent electronic manufacturers, was the first target.
The prototype had a huge impact on Sony. Nakajima recalls, “We hadn’t even thought as far as a commercial product, but our eyes were opened and we were very impressed by Philips’ approach”. As mentioned previously, the reaction from most manufacturers other than Sony was to take a “wait and see” stance. This was because most Japanese manufacturers at the time were caught up in the appeal of the video disc and had disregarded the idea of commercializing an audio-only disc. Another reason for this reaction is that most manufacturers were developing contacttype signal reading technology and had little confidence in a contactless optical system.
Meanwhile, Sony had improved digital recording and error correction technology to achieve higher density recording and longer playing time and was now at the stage of having to define a specific product image. The question was how to market several hours of music software. At that point, the surprisingly good proposal came from Philips, and Sony leapt on board to resolve its own issues. With its “eyes opened” to Philips’ idea of a small scale, dedicated audio disc, Sony worked with Philips on standardization based on the Philips prototype. The company was quick to reach a consensus, and the Philips/Sony joint task force was formed in August 1979.
At the same time, Sony’s senior management was moving quickly. In April 1979, a month after Philips’ demonstration in Japan, Sony Vice President Ōga Norio met with W. Zaiss, Philips VLP division manager. Sony chairman Morita Akio met with Philips vice president Cor van der Klugt in May, while Sony president Iwama Kazuo visited PolyGram in Eindhoven and Hanover. This rapid succession of moves by senior executives was keen and agile, and deepening these bonds between business is thought to have been the greatest factor in the success of the CD.
Philips prototype specifications
The Philips prototype had the following specifications.
- Disc diameter: 115mm
- Disc thickness: 1.5mm
- Sampling frequency: 44.3kHz
- Quantization: 14 bit
- Rotation: 500-215 rpm, starting from center
- Replacement for LP records
- Dedicated audio
- Wider scope of application by downsizing
- Uses a semiconductor laser
Other details, such as the recording density and error correction, are still under investigation and have not yet been published. Fig. 2-3-3 shows the Philips demonstration prototype. There was a considerable amount of digital circuitry under the table, but as advances were made in LSI, this all was made to fit inside the small sized player. Philips and Sony had been unable to negotiate to an agreement over household video cassette recorders and both had experience bitter failure. Now faced with shrinking music production, this made the bond between the two companies stronger and probably became the driving force behind the new-concept product, the CD-DA
Philips/Sony Discussion Memorandum - The six meetings
With “eyes opened” by the compact disc product proposal from Philips, Sony had a rapid succession of questions and suggestions for Philips, confident in its decade of experience in digital audio technology. Despite the fact that it had its own music company, CBS Sony (with the power to resist a new format), it was less than a month after the demonstration that the company reached a consensus to base its technology on the optical format compact disc. This was the first time in the company’s history that a basic agreement on how to proceed had resulted from urgent meetings between senior managers. Based on that agreement, relevant staff from the two companies held a series of six meetings in Eindhoven (Netherlands) and Tokyo between August 1979 and June 1980 to establish specifications. Discussions between the companies began in Eindhoven and resulted in a basic agreement within six months. Within a year, a provision standards book, the so-called “Red Book”, had been put together. The negotiated outcomes between Sony and Philips were added into the detailed report (memorandum) by Mizushima Masahiro (Sony R & D manager). Mizushima was the only Sony representative present at all of the meetings. Since the senior managers of the companies had already agreed on the product concept, the necessary items to be discussed regarding digital audio disc specifications included: (1) the outer diameter, inner diameter, thickness and other physical characteristics of the disc; (2) playback length; (3) sampling frequency and quantization bit number; (4) modulation method; (5) error correction method; (6) standard laser frequency (780nm semiconductor laser).
First meeting: August 27-28, 1979, Eindhoven
The attendees were Nakajima Heitarō, Miyaoka Senri (manager of the Disc Development Division), Doi Toshitada (deputy director of the technical laboratory) and Mizushima Masahiro from Sony, and Pieter Bögels, Marino Carraso (manager of the Optical Disc Research Division, Research Center), Joop Sinjou and Jacques Heemskerk from Philips. It was agreed to aim to release new products in 1982-1983 and to finalize a summary of specifications by the end of the year. The details of the discussion are given below.
Disc Outer Diameter
The demonstration disc Philips brought to Japan had a diameter of 115mm. This diameter matched the diagonal length of the compact cassettes that were the main audio system at the time. Philips was conscious of the car audio industry and wanted to maintain company consistency with a size that matched that of the compact cassette. The shared development task of the developers was to go “from compact cassette to compact disc”. However, this did not mean Philips was fixated on 115mm. The requirement was to have the player fit within the DIN standards, which meant any disc diameter between 105mm and 125mm was possible.
Both companies were of the same opinion regarding playback duration. Philips proposed a playback duration of 45-60 minutes, to which Sony agreed. The echoes of four-channel audio were still present throughout the audio industry at the time. Sony insisted on the importance of a four-channel system that was bi-directionally compatible with two-channel systems. Frequency division and time division were also discussed, but there were questions about both. There were also doubts about the marketability of a four-channel system, so the discussion did not go into depth. However, it was standardized in the Red Book that one bit would be allocated to four-channel and if there was market demand, this could be achieved by halving the playback time. Unfortunately, no four-channel audio has emerged to date.
Sampling Frequency and Quantization Bit Number
Sony believed that the strongest candidate was 50.4kHz, as discussed by the Digital Audio Standards Committee of the Audio Engineering Society, and advocated for a sampling frequency of 50.4kHz and 16-bit quantization. However, Philips argued that this was too much, since 20kHz was adequate for audio bandwidth. Regarding the quantization bit number, Philips spoke on behalf of its subsidiary music company, PolyGram, with the view that “since 16-bit is used for professional mastering, 14-bit is enough for consumer use”. Either way, both companies agreed that “one level down from professional use is adequate for consumer use”.
At this meeting, Philips explained that the demonstration model had used modified Miller modulation, the so-called M2 method, for its modulation. Philips also discussed the effect of the molding precision of the pit (the shape the signal forms on the disc) on jitter and highlighted the importance of controlling the low frequency component of the modulation signal. Sony agreed that jitter depends on molding precision. Regarding the effect of the low frequency component, Sony asserted that the emphasis should be on having the minimum inversion intervals during high density recording, and promised to present data on the verified four to twelve modulation (FTM) at the next meeting.
Error Correction Method
Philips showed the huge amount of data it had compiled on disc errors and claimed that most of it was random error using 3 bits or less. Sony insisted on the importance of countermeasures for burst error as well as random error and the importance of error signal correction with long interleave length. The two companies had very different opinions. As the debate went on, Philips admitted the importance of burst error countermeasures, but insisted that burst error is caused by hardware mistracking or defocusing and that a method to completely prevent the cause from occurring would be sufficient. Sony repeatedly advocated the idea that if long burst error is presumed to occur, then a suitably capable error correction method should be built into the system from the outset.
The companies also discussed the concept of control and display, the forerunner to subcode, and how to implement it.
Second meeting: October 3-5, 1979, Tokyo
The two companies first presented and explained their modulation and error correction methods. These were at odds, with Sony favoring (1) FT modulation and (2) cross-interleaved error correction and Philips favoring (1) M2 modulation and (2) convolution error correction (2,3). They agreed to technically compare, measure and verify both claims and to reciprocally send engineers to work on trial disc manufacturing. Laying the groundwork for proceeding with discussions based on data made for significant progress in that it was the first step towards joint development.
It has become something of a legend that the team sent to Philips were true engineers with very little English. Dr. Heemskerk, chief engineer at Philips, shared the following recollection. “Those engineers were very brave. Fluent English is not always necessary to have a discussion about technology, and they were able to fully enter the discussion using the universal language of technology. In the end, we became a single team, regardless of nationality or management.” As a success story for future generations, this proves that no matter what the situation, the determining factor is “having the courage to take the plunge”. It has long been said of many events that “it was easier than we thought”.
At the meeting, Sony insisted that if it was possible to have a disc diameter of less than 125mm while still ensuring more than 45 minutes of playback time, then the standard sampling frequency should be 50.4kHz with 16-bit quantization. Philips commented that if the disc was less than 115mm, it had no issue with Sony’s proposal.
However, despite understanding that the basic concept for the CD was a dedicated audio system, Sony unusually proposed that it should have some components in common with video disc players. The situation at the time was probably affected by the introduction of video disc systems on the market, and so this came up for discussion. Specifically, this included making factors such as disc thickness, center hole diameter and signal area start diameter the same as video discs. The conclusion was that it would not be possible to implement, given the significant difference between the concepts, and Sony agreed. This resolved the first question of compatibility with video discs and the matter was settled that the CD would be a dedicated digital audio disc. The CD continued to be developed as a dedicated digital audio system for another decade until around 1990, with the emergence of the video CD, a digital video disc using MPEG1 video signal processing technology.
Third meeting: December 17-19, 1979, Eindhoven
The test results of the jointly developed experimental discs agreed on at the previous meeting were reported.
In terms of modulation method, the FT method proposed by Sony resulted in three times more bit errors than the M2 method proposed by Philips. It was also reported as being less adequate in terms of dropout, fingerprints and sensitivity to asymmetrical pit shape. Sony had predicted this situation and had developed 3PM modulation, a new and improved method to replace FT modulation. Consequently, Sony agreed to abandon FT modulation immediately and propose a new “ASAP” method as soon as possible. The improved 3PM modulation method was later named eight to fourteen modulation (EFM) and became the mainstay of basic CD technology.
In terms of error correction, the proposals by both Sony and Philips were reported as having no issues for ordinary discs, although Sony’s method proved to be better in the low-error range and Philips’ method proved to be better in the high-error range. However, from a technical point of view, the reports indicated that there was room for improvement in error correction capabilities for scratches and fingerprints on the disc surface and where signal reading tracking or focus was out.
Despite the tireless efforts of the engineers, time went marching on. The directive from management to gather measurement data in 1979 and reach a final agreement in March 1980 was impossible to achieve. However, the engineers at Sony and Philips admirably did all they could to put together the main specifications in time for the DAD Committee scheduled in March.
To achieve this, they did anything they could and used any resource they could find. For error correction in particular, they separated the factors causing random error and the factors for burst error caused by physical anomalies in the disc and implemented a very thorough computer simulation. At this time, Sony only had one IBM 1340 and it was shared by the business division. Commandeering this IBM, used to calculate the payroll, to run simulations overnight meant that the payroll could not be calculated, resulting in delayed salary payments and angry scolding from the managers in general affairs. At the beginning of March, an ASAP modulation method was proposed to Philips, with this muddled month having widened the window margin. The error correction method proposed by Sony was cross interleaved Reed-Solomon code (CIRC) error correction, developed by converting adjoining code into Reed-Solomon code with the idea that it would make the hardware configuration easier.
Fourth meeting: March 17-19, 1980, Tokyo
The objective of this meeting was to have a summary of basic CD technology ready in time for the DAD Committee meeting scheduled for the end of March 1980. The main specifications agreed on were: (1) a sampling frequency of 44.1kHz; (2) 16-bit quantization; (3) Sony’s proposed error correction method of converting adjacent code into Reed-Solomon code; (4) disc diameter and playback time. Although Philips insisted on 115mm/60 min and Sony insisted on 127mm/75 min, they agreed that this was a question of marketing and better left to the judgment of senior company executives in both companies. Unfortunately, the modulation method was still being experimented with and was left out of the discussion.
The standards for professional and consumer digital audio at the time included several standard frequencies. One was 50Hz, set as the studio standard in 1978 by the Audio Engineering Society. Another was 44.1kHz, used by Sony in the PCM-1 and PCM-1600 (derived from (525-30)×3×30 NTSC video signal), and its integral multiple (8/7) 50.4kHz. A third was 48kHz, an integer multiple of the 8kHz frequency used for digital telephone communication being considered for use by broadcasters and record companies inEurope. Amidst all this, Sony withdrew its initial insistence on 50.4kHz in favor of 44.1kHz, as this was the frequency it used in the PCM-1600 master recorder released in March 1978. Sony had decided that this sampling frequency was still theoretically able to produce audio playback at 20kHz. At the same time, the greatest advantage to digital audio was no loss of sound quality by having the same sampling frequency from recording to editing, cutting and disc production. This was necessary to be able to easily use the many master sources already recorded on the PCM-1600. This revolutionary decision meant that the same sampling frequency would be used for professional use and consumer use alike.
Quantization bit number
Philips had pegged the CD as an easy listening medium for in-car audio and insisted that a quantization bit number of 14 could theoretically produce a dynamic range of 86db and would be adequate for consumer use. In fact, most analog cassette tapes and LP records to date have been around 70db, so the 86db achieved with 14-bit quantization would have provided far superior sound quality. It is also a fact that many of the key 16-bit devices at the time were only used for military or measuring equipment. Using these for consumer use would have added two extra digits to the cost, which actually was a preposterous proposal. Despite this, Sony continued to insist on the importance of 16-bit audio even for consumer use, as it would ensure digital audio discs were a high quality sound system capable of classical or jazz.
This situation did not sit well with the claim that “the mission of the CD is to provide households with a far superior recorded sound than existing LP records and cassette tapes”. At the time, Sony had its sights on the forthcoming development of a 1-chip 16-bit computer. Most computers processed in 8-bit units, but the technology for integral multiple processing was soon to emerge. In his book “CD Ōdio Dangi [Discussion of CD Audio]”, Nakajima is of the opinion that “as a system for the 21st century to come, improving by a mere 10db was not enough; we had to ensure a dynamic range of over 90db for the future – we had to do 16-bit.”
The 14-bit/16-bit debate dragged on with no focal point. As part of the debate, the task force compared recordings of the sound of a triangle, the last test carried out by the engineers. Heemskerk (Philips chief engineer) recalled that as they all listened to the metallic clang reverberate, he thought, “There seems to be no difference, but maybe there is”. However, the Philips team were also engineers and had no objection to making the sound quality as high as the technology would allow. After this experiment, the decision was finally made to have a quantization bit number of 16.
Error correction method
The task force had differences of opinion regarding burst error. Generally speaking, it is impossible to reduce optical disc defects to zero, regardless of what process is used to manufacture it. Discs can also be scratched or get fingerprints or dust on them. As this can cause abnormal reading, it was clear that steps had to be taken to somehow detect errors and restore the signal back to normal. Most disc defects are errors called random bursts (short bursts) that last for around 1-2 bits. As these can occur as much as several dozen time per second, Philips proposed an efficient error correction system using an experiment-based, mathematical approach. Meanwhile, Sony proposed a design with a practical simulation that effectively corrected long burst errors, which do not occur frequently, but can last for dozens of bits as a result of mistracking, or from scratches or fingerprints on the surface. The differences between these two underlying approaches to error correction were preventing the companies from reaching an agreement.
As a final step, the task force decided to test error correction performance by open experiment. The first round was a comparison of ordinary high quality discs. The second round was a comparison of discs covered in dust and smeared with gunk. There was no difference between the two methods for these two rounds. The third round was a comparison of discs that had deliberately been scratched with sand on a tabletop, and this time Sony’s method was clearly better. In response, Philips confused the meeting by offering to add improvements to its own method to make it perform as well as Sony’s method, such as by having a longer interleaving length.
However, Philips noted that the improved version suggested by Sony improved redundancy by up to 17% and cooperated in making changes, including making simulation estimates for the LSI proposed by Sony to prevent concerns about rising costs. The two companies finally came to a basic agreement to use Sony’s new proposed method. They also agreed on converting adjacent code to Reed-Solomon code, and Sony agreed to present the final version of the error correction to Philips at the next meeting. The method was named cross interleaved Reed-Solomon code (CIRC) and performed as well as they had hoped. As luck would have it, having error correction in byte units, and later having signal modulation in byte units as well, made the CD more compatible with computers, which led to the rapid development of the CD-ROM.
No agreement was reached on modulation method at this meeting. There was discussion comparing the Philips M3 method proposed at the third meeting with the new method proposed by Sony. Sony’s method was suited to a higher density than Philips’ method, but there were still questions about its stability in relation to fingerprints, scratches and other noise. At this meeting, the parties agreed that if Sony could increase the density by a further 20% and confirm that the impact of bit asymmetry was within a tolerable range, they would proceed with Sony’s proposed method, otherwise they would use the M3 method. However, after the meeting, Philips started developing a modulation method similar to the one proposed by Sony. This method, with further efforts to reduce the low frequency components, was the BES method that formed the prototype for the EFM method that would become the modulation method used in the CD.
Up until the fourth meeting, the Sony/Philips specification negotiations had been reaching basic agreements and getting basic conditions prepared in time for the DAD Committee submission deadline. However, none of the DAD Committee members knew that Sony and Philips were developing jointly. Of course, there was no issue with Sony or Philips making proposals independently, but it must be explained why Sony kept the DAD Committee in the dark about going into partnership with Philips when the two companies started working on joint development. Although there is no problem at all with having an agreement or alliance with a particular company as a management strategy, in Japanese society it was unthinkable. Nakajima later recollected having to weather a storm of anger from most of the members. Members Watanabe Shū (Hitachi) and Kamio Kenzō (Matsushita Electric) recollected that “although everyone at the time hated it that Sony had teamed up with Philips to develop the CD, it is normal to do it now. Looking at the fuss over the DVD after that, we have to say that Sony’s ability to take action on the CD at that time was remarkable” (JAS Journal Proceedings).
Fifth meeting (May 13-15, 1980, Eindhoven)
Based on the agreement from the previous meeting, Sony’s amended error correction was agreed upon and the official name designated as cross interleaved Reed-Solomon code (CIRC).
At the previous meeting, the decision on disc diameter and thickness had been left to senior management. A diameter of 120mm and thickness of 1.2mm were confirmed, according to the specifications agreed on in early April between Sony vice president Ōga Norio and Philips audio division general manager Joop van Tilburg. Philips had previously proposed 115mm with a playback time of 60 minutes. This proposal had been strongly underpinned by Philips’ subsidiary PolyGram, asserting that the cost of the disc was not the cost of manufacturing the disc, but the cost of the software on the disc. By this rationale, the longer the playing time, the greater the cost of the disc. It claimed that around 90% of all music, including classical but not opera, was no longer than 60 minutes. Sony’s vice president invoked the famous argument: the Japanese love Beethoven’s 9th Symphony. Being able to fit that piece of music on one disc would give meaning to the new media. Ōga insisted that 75 minutes would allow enough room for the piece to be conducted at any tempo. van Tilburg reluctantly agreed on the condition that PolyGram would not be further aggravated. (The published CD standard playback time remains “60 minutes”, not the maximum 75 minutes.)
Disc diameter, as previously mentioned, had started out with a product plan for in-car players, in keeping with the DIN standards. Sony engineers had concluded that the maximum possible diameter was 130mm. van Tilburg responded that he “wouldn’t concede 130mm, but 120mm would be OK”. Ōga said that would be fine, and the diameter was promptly set at 120mm. Incidentally, the 15mm inner bore diameter of the disc was the size of the Dutch 10 cent coin. By adjusting the drive motor and the optical pickup in the design, the playback start position was set at 25mm. Calculated from this, 10mm or so was a suitable inner diameter. This coin, playfully nicknamed by the Dutch as the “dubbeltje”, is an unforgettable symbol as the only coin in the world that can fit through the center hole in a CD.A Dutch 10 cent coin 'dubbeltje' inside the center hole of a compact disc
The later DVD and Blu-ray discs also had 15mm center holes. While this is partially for compatibility, allowing different types of discs to record and play back on the same player or recorder, in another sense it is a “fun fact” that the “DNA” of this ten cent coin, the smallest in the world, has been successively passed down to other types of optical discs after the CD.
The main topic at the fifth meeting was the modulation method, the most pivotal point in the negotiation.
The M3 (modified Miller modulation) method proposed by Philips was DC free, but lacked in recording density. The method proposed by Sony had around 10db greater low frequency component, but lacked in system reliability, leading to concerns over the recording density level that could be achieved practically. Accordingly, Sony proposed an amendment that reduced the low frequency component, while Philips proposed a new modulation method suited to high density recording. There was no great difference between the two proposals in terms of performance. Philips pointed out that Sony’s method would transmit errors, but Sony insisted that error transmission was not highly important. The method proposed by Philips had the issue of hardware cost. The debate continued in two parallel lines with no signs of convergence. The suggestion came up of discarding the agreed-upon CIRC error correction method, and the Japanese team started preparing to go back to Japan. At this point, the two companies had lost all the achievements they had built up together, and the decision was left to management. Bögels suggested extending the meeting for one more day and arranged a friendly dinner party. He affirmed his appreciation for the negotiations between the two companies and his respect for technology. In the end, Nakajima accepted Bögels’ proposal and a de facto agreement was reached on the modulation method. The details of the agreement were that Philips would further investigate the servo circuit synchronization recovery time and the effects of bit asymmetry on the BES method, while Sony would agree to the BES method if the scale of hardware could be brought within a reasonable range. As if to soften the antagonism between the two companies, the following well-known memo was appended.
Philips and Sony agreed that contribution to modulation as well as error correction from Sony and Philips is equal.
H. Nakajima, Tr. P. W. Bögels
The patents for the agreed technologies were filed in the names of engineers from both companies. In that sense, both companies contributed equally to the compact disc system.
Sixth meeting: June 17-18, 1980, Tokyo
The BES method was adopted as the modulation method, but the official name was up for review. It was later changed to eight to fourteen modulation (EFM).
The official name of the system was designated as Digital Compact Disc at this meeting, although the official name was changed to Compact Disc Digital Audio due to trademark concerns. It was also confirmed at the time that a Sony/Philips alliance would present the CD system to the DAD Committee.
With the above final agreement in place, the Sony/Philips CD standardization task, begun in August 1979, reached the first stage of completion in June 1980 and could finally be revealed to the world. At this point, with much still to decide on, Sony and Philips continued on into the second stage of negotiations. Fig. 2-4-1 shows a commemorative photograph taken at the end of the sixth meeting.On the last day, June 18 1980, at the end of the six meetings, a photograph was taken in Tokyo. It shows a happy smiling team. From left to right: 2n row: Heemskerk, Harada, Miyaoka, Vries, Nijboer, Tsurushima, Doi, Ogawa, Naruse, Odaka. Front row: Sinjou, Bögels, Nakajima, Mizushima.
A year later, in April 1981, Sony and Philips held a grand CD demonstration in Salzburg, backed by Karajan, for the music and record industry. This kicked off a world debut, with an official announcement was made in New York on June 27 and an exhibit at the All-Japan Audio Fair in the fall. Fig. 2-4-2 shows a photograph of a Philips prototype compact disc player. However, at this point there were as yet no prospects for any of the key devices, such as semiconductor lasers, optical pickups or signal processing LSI.From left to right: Joop Sinjou (Philips), Herbert von Karajan, Akio Morita (Sony), April 1981 Salzburg
The following is a summary of the CD-DA specifications agreed by Sony and Philips.
- Performance time: 74 minutes 42 seconds
- Disc diameter: 120mm
- Sampling frequency: 44.1kHz
- Quantization number: 16bit
- Error correction code: CIRC
- Modulation method: EFM