THE RIGHT ANGLE
Imagine that you're the president and sole owner of the Intergalactic Widget Corporation. You've just returned from your summer vacation; since you're so filthy rich, and you have a fondness for Monte Carlo, this particular vacation has lasted about 12 years. You immediately call a corporate meeting. Your faithful factory manager, Elroy, reviews the sales trend, presenting a chart that shows sales of the company's product rising almost exponentially. Nonetheless, you panic, frantically beseeching Smedlap, your faithful engineer, to start a crash research and development program to find a new product. Smedlap and Elroy look at each other, wide-eyed. Has the Chief gone mad?
Well, the Chief has always been eccentric, but nevertheless astute. The "Widget" in his company's name actually refers to prerecorded analog cassettes, and the sales curve parallels that medium's rise to dominance in the U.S. (Sales of LPs peaked in 1979 and were surpassed by cassettes in 1982.) So why is the Chief panicking? The answer, of course, is digital audio. Compact Disc sales are still only a fraction of either LP or cassette sales, but the CD's growth has been rapid, and is expected to double annually for the next several years. If today's trend stays on course, the CD will pose serious threats to the analog tape and disc industry within five years. And when record-once and erasable/recordable CDs hit the market, analog tape and disc might be doomed. Smedlap, ace engineer at Intergalactic Widget Corporation, therefore hurries to his workbench to contemplate new product development as well as IWC's new advertising slogan, "Where there's tape, there's hope." Sure, analog tape suffers from wear and tear, it loses high frequencies to friction from guides and heads, savage equalization is required, duplication necessarily doubles noise with each generation, distortion and hiss are inseparable companions, etc. Yet tape has an undeniable asset: Music is a sequential event, and a spooled length of tape lends itself to sequential storage. While a random-access disc might be convenient for locating a musical selection, the playing itself is always sequential. Tape is thus an inherently suitable medium for storing music information.
Analog tape's bandwidth is well matched to music's requirements, too. The tape must be able to accommodate the analog signal's entire range of frequencies. But an audio frequency range of 0 Hz to 20 kHz requires a tape bandwidth of not much more than 20 kHz-not too hard to achieve today, though it took several decades of tape progress to reach it.
And the principles of analog recording are simple. A plastic ribbon is evenly coated with a thin layer of acicular (needle-like) magnetic particles which behave as tiny magnets, with north and south poles. The particles are oriented along the length of the tape, but on a tape with no net magnetization, the north and south poles are randomly distributed. When an audio signal is applied (by passing the tape over a head which turns that signal into a varying magnetic field), !he poles begin to orient themselves according to the strength and direction of the magnetic field at the moment each particle passes the head. As signal amplitude increases, more and more particles' magnetic poles align themselves. This is called longitudinal recording.
The process has its limits. For one thing, when the signal amplitude rises to the point where all possible particles have aligned themselves with the signal, the tape becomes saturated. Any further increase in amplitude will cause severe distortion-clipping-as the signal flattens out against the tape's rigid limits, eventually turning sine waves into near-square waves.
However, given particles small enough to follow the period of a high frequency audio waveform, and enough of them to encode a sufficient range of amplitudes, analog tape recording is achieved.
This much Smedlap already knows. But, he wonders, would it be possible to fight digital with digital? The answer, as anyone who has ever encountered a computer storage medium (or a processional digital tape recorder) knows, is "yes." The digital tape recording process is essentially identical to that described above. However, instead of trying to record the subtle variations of an analog waveform, we are trying to record the more clear-cut, logical transitions between bits. Here, that bane of analog recording, saturation, works for us.
Its squaring of the waveform is ideal for the binary nature of digital data. Thus, instead of a continuously variable net orientation of particles, there are only two orientations. Each change in polarity is sensed at the read head as a binary digit. That makes our job much easier, in some ways: Tape noise and waveform distortion are irrelevant, as long as the polarity change can be read. (In fact, the waveform actually recorded on tape is far from a clean square wave; that would waste valuable bandwidth.) It is bandwidth that is the chief problem with the digital approach. In a digital recording, the signal waveform must be encoded as sampled data words, accompanied by additional data for synchronization, error correction and so on. The upshot is a considerable number of bits, each of which must be encoded as a level transition.
As a result, digitally recording a 20 kHz audio signal might require a tape bandwidth of 500 kHz. To achieve this, digital tape uses formulations different from those of analog tape, and mechanical tolerances (such as head-to tape contact) are more critical.
As Smedlap studies the problem, he becomes acutely aware of the magnitude of the medium's storage requirements. He observes that, with a 48-kHz sampling rate and a tape speed of 30 ips, a density of over 20 kilobits per inch is required. To achieve digital longitudinal recording, he must resort to particles with high magnetic energy levels (and hence higher packing densities), higher tape speeds, multiple recording tracks and esoteric head technology (such as thin-film heads, which are manufactured with methods akin to those used for integrated circuits). Even with these technological tricks, longitudinal recording has a finite recording density; as smaller and smaller particles are packed closer and closer together, self-demagnetization occurs--the poles neutralize each other, and signal output diminishes. The practical limit seems to be 25 kilobits per inch.
Smedlap is white-knuckled; a 30-ips digital cassette would be larger than most household pets. He ponders: Instead of aligning the particles lengthwise along the tape, how about lining them at a right angle to the tape surface? Such a method, called perpendicular (or vertical) recording, allows for greater particle density since it is the particle width that is the determining factor. Moreover, the thinner the particle, the greater its length-to-thickness ratio, and hence its magnetic strength. In other words, the denser the particle packing, the more robust the medium. The problem of self-demagnetization vanishes, so extremely short wavelengths can be recorded. A digitization system requiring storage for 800,000 bits per second would consume 40 inches per second of longitudinally recorded tape, but only 2 inches of perpendicular.
However, perpendicular media require costly manufacturing techniques; researchers have used alloys of chromium and cobalt in the form of hexagonal crystals. Furthermore, the alloy must be deposited onto the backing in a vacuum chamber. Clearly, more efficient methods will be required for mass production.
Suddenly Smedlap is struck by a brilliant notion. Why not take advantage of the fact that a magnetic layer can be magnetized in any direction? With a two-way technique called isotropic recording, the oxide layer may be magnetized in both longitudinal and perpendicular modes, simultaneously.
With special head configurations and tape formulations, the two recorded fields may be combined in phase to produce a strong output signal. The head is designed so that the longitudinal field penetrates the tape oxide deeply, and is recorded first. Then the perpendicular field is recorded over the longitudinal one, erasing the longitudinal field nearest the tape surface.
Thus, a perpendicular field lies over the longitudinal one. With such isotropic techniques, researchers have achieved recording densities of more than 250 kilobits per inch.
After many sleepless nights, Smedlap is convinced that digital audio tape recording can be achieved with available technology by using longitudinal, perpendicular, or isotropic recording methods. But would such a product be viable in the consumer marketplace? The answer to this question is largely dependent on such considerations as the digital cassette recorder's size, reliability, and cost. And that boils down to the recording mechanism itself--that is, the head and track format. Specifically, should a digital audio cassette employ a stationary or a rotating head system? Each offers certain advantages, as well as disadvantages. A rotating head yields great bandwidth but requires intricate mechanics; a stationary head is mechanically simple but would require many tracks to record the data load.
Meanwhile, the Chief is convinced that Intergalactic Widget Corporation is teetering on the brink of ruin. Smedlap racks his brains, balancing the pros and cons of rotating and stationary heads, not to mention playing time, cassette size, sampling rate, quantization word length .... Actually, Smedlap knows, his hands are tied. IWC can't sell prerecorded digital cassettes until there are digital cassette decks to play them on. And if manufacturers are sane, there will be no such decks until the major companies, at least, agree on tape and player standards--maybe one standard, maybe two (one each for rotary-head and stationary-head designs). Meanwhile, Smedlap must look into all the possibilities, but postpone action until the standards are set.
Will Smedlap find the ideal digital audio cassette system? Will the Chief agree with his engineer's decision? Or will Ted Turner buy IWC with junk bonds and restructure it as a bakery? Tune in next month!
(adapted from Audio magazine, Jan. 1986)
= = = =