LAB NOTES
It's easy to think of the Compact Disc as a Japanese creation. Dozens of Japanese companies have been advertising for months as if they each had invented it, and most brands of CD players are Japanese.
In actuality, though, the basic system did not originate in Japan, but in the laboratories of Philips, in The Netherlands. ( Holland is actually just part of the country, but for brevity's sake, we'll call it that from here on.) A few months back, I and a group of other audio writers spent some time in Europe, looking into CD's roots, as guests of Philips and of Polygram.
Philips is a much bigger company than many of the Japanese firms whose names are household words in the U.S. That's partially because Philips doesn't sell much here under its own name, but through such subsidiaries as Magnavox, Sylvania, Philco, Norelco, Signetics and Amperex. An even more universal presence is the Compact Cassette which, like the Compact Disc, emerged from the Philips Laboratories in Eindhoven.
The Labs themselves are as big as many companies, employing 4,000 people, 240 of them working on patents alone. Funding comes directly from the corporate budget: The labs receive 1.5% of the company's turnover. That gives them a certain desirable independence, according to the lab's Director for Consumer Electronics, Dr. C. A. A. J. Greebe: "Between the product divisions and us, there is a low-pass filter. So, we can do research no operating division believes in yet, but which we feel will be needed in a few years. Hence, we were doing early research on digital error correction, back when you couldn't make the necessary electronics because they would have been too big.
"Because we are corporate, we have free exchange of manpower and technology with different divisions. Hence our work on CD with the Elcoma (ICs and other electronic components), audio, and pro audio divisions. Most of our people enter research work from college, then go to the product divisions at about age 35-we get people with original outlook, they get people with wide knowledge and experience." (It may also help the lab politically to have alumni in all the operating divisions.) "Aside from the Japanese labs, few labs have one discipline called 'consumer electronics.' Here, that includes research on perception (in conjunction with Eindhoven Technical University) and on computer architecture (with the Elcoma division).
"For speakers, you need a materials research department. The Japanese are heavy on this-it's common for them to put 300 students onto researching the table of elements and combinations. The small manufacturer is in grave danger here." The history of CD, according to Dr. P. Bagels, (Director of Development, Technology/Innovation, Audio) goes back to 1969, when Philips began work on what eventually became the LaserVision video disc. Even at that point, the system was considered as a potential audio medium too, and in 1974, Philips separated the audio and video development tracks.
The first try for an audio laser disc used low-density FM recording followed, a year later, by a higher density, and therefore smaller, digital disc. There wasn't much digitally recorded material at that point though. Denon was probably first to make digital recordings for eventual LP release, back in 1972; Soundstream began mastering for commercial record companies in 1978; 3M delivered its first digital recorder in 1979, and Sony introduced a digital recorder in 1976.
So digital source material was beginning to emerge.
In 1977, a 12-inch digital disc was demonstrated by Sony, Hitachi and Mitsubishi. JVC announced its VHD/ AHD 12-inch video/audio disc system in 1978.
By 1979, there were four contending home disc systems: An 11.5-cm (4 1/2-inch) system from Philips, tracked by lasers, and two stylus-tracked Teldec systems, the 13.5-cm (59/16-inch) mini-disc and 7.0-cm (2 3/4-inch) micro disc capacitive (Audio, June 1981), as well as the stylus-tracked AHD system. About that time, Philips and Sony were conferring on a mutual digital disc standard, eventually agreeing on a system using Philips modulation techniques and Sony error-correction systems, which was submitted to the Japanese MITI Digital Audio Disc committee in June 1980. In 1981, Matsushita adopted the system, even though JVC, developer of the competing AHD system, was a Matsushita subsidiary. With Matsushita and Sony joining hands (for once) on a common system, it was settled: CD would be the final system.
Somewhere along the way, the disc size increased from 11.5 to 12.0 cm (4.72 inch). According to an interview with Sony's new President, Norio Ohga, in Electronics, that was at his insistence, so the disc would have enough playing time-about 70 minutes-to hold Beethoven's Ninth Symphony or other similarly long works. (In that, CD echoes the LP's history, since the LP's groove spacing and speed were at least partially selected with an eye towards accommodating an entire symphony of average length.) The Compact Disc is now officially billed as holding 60 minutes of sound ("to make it acceptable to record firms," says Ohga), but is actually capable of holding 74 minutes.
It's also possible, says Dr. Bagels, to accommodate four sound channels, though with less playing time. Other improvements coming up will include improved lenses, the use of CD for computer data storage, and simpler, digital servo systems.
We saw one future development in progress at the Labs: A digital sound processor for reverberation, equalization, scratch and pop removal, and compression of the digital signal.
Philips sees the future hi-fi system as being digital almost all through.
The few remaining analog sources would go through a selector switch (with, presumably, a phono preamp for the turntable) and feed an analog-to-digital convertor. The output from the ND convertor, and from the system's purely digital sound sources, would feed a purely digital selector circuit and a digital sound processor.
Output would only be converted back to analog in the power amps and in the output to the analog tape deck (if any).
The digital processor would have many functions, including reverb, equalization, filtering, scratch and impulse noise removal, and compression. Doing all this digitally allows precise and flexible control as well as complex processing without loss of sound quality. For example, a digital signal can be delayed without quality loss, either for reverberation or to let an advance copy of the signal precede it as a control for the actual processing circuits.
The automated equalizer function of the processor was, like such existing automated equalizers as the dbx 2020, a simple, 10-band octave equalizer with 1-dB resolution-albeit with a signal-to-noise ratio of 124 dB! The demonstrator, a Dr. Vandenbulcke (Project Leader for Digital Audio), talked of such other possibilities as tone control, "physiological volume" (loudness?) control, and parametric bandpass filters.
I could not help but think of Acoustic Research's ADSP, which digitally analyzes and corrects room as well as system response, using variable bands sometimes as narrow as 1 Hz, a far more sophisticated system. But according to Vandenbulcke'(if I interpret my rapid notes correctly), "Based on perception studies, we feel it's best not to equalize the reverberation-it's best to leave the tail-off alone." I was more impressed by the demonstration of record-scratch removal. Comparing the signal from an old record before and after processing, I found the ticks and pops all gone. Some surface hiss remained ("We're working on that," Vandenbulcke said. "It isn't easy."), made even more audible by the absence of other distracting defects.
The system simply muted out impulse noises for about 10 mS, rather than interpolating to fill in the gaps.
Vandenbulcke says, "We are convinced that this is the best that will ever be needed for consumer use." Based on what I heard, he may be right. Digital processing was used here because this form of processing requires delay, which can be far quieter with digital than with analog circuits.
But why a compressor? That, Vandenbulcke explained, is because CD's dynamic range is too wide for many home listening situations-even though CD recordings don't use the medium's entire dynamic range. More on that, next month.
Billions of Bits
We try to check things carefully. So when Len Feldman wrote in the April issue that the Sony PCM-701's video monitor jack could "be connected to a TV monitor, should you wish to watch the billions of 'bits' form .. .
patterns on the screen," I wondered whether those bits really did mount up into the billions. So I checked it, with my calculator.
It turns out they do-but that it takes nearly half an hour per billion.
In the 701's 16-bit mode, it records 44,056 "words" of 16 bits apiece, each second, for a total of 704,896 bits of data per second. That's 42,293,760 bits per minute, and 2,537,600,000 per hour-or a billion bits every 23.64 minutes. Since a TV frame is on screen for just 1/30 second, the number of data bits you'd see on screen at any given moment would be just 6,830 bits.
I've been careful here to refer to "bits of data," because a PCM system also records additional bits used only for error-checking and correction. If you count those bits, the numbers above would become a little larger, and the times given would be a bit shorter. The number of bits actually on your screen would be smaller than calculated, too, because most TV sets and monitors "over-scan" the picture, projecting parts of the image beyond the visible portion of the screen.
As to 14-bit mode, the number of bits per unit time would be 7/8 as large, and the time required to output a given number would be 1-1/7 times larger. So a billion bits would take 27.02 minutes.
(adapted from Audio magazine, Sept. 1983)
= = = =