Digital Domain (Feb. 1986, By Ken Pohlmann)

Home | Audio Magazine | Stereo Review magazine | Good Sound | Troubleshooting


Departments | Features | ADs | Equipment | Music/Recordings | History


HEAD LINES


As you may recall from last month's column, you are the Chief and sole owner of Intergalactic Widget Corporation. Upon your return, tanned and relaxed, from your 12-year vacation in Monte Carlo, you asked your faithful factory manager for a review of your company's sales of analog cassette tapes. You panicked, despite seeing an apparently bullish trend, and beseeched your faithful engineer, Smedlap, to start a crash pro gram to develop a digital audio cassette system. Clearly you had seen the bits on the wall. After pondering the limitations of magnetic recording, Smedlap responded with two possible solutions: A stationary-head system and a rotary-head one. When we left off, you were deep in thought, contemplating the advantages and disadvantages of each. . . .

At first glance, a stationary-head de sign is quite appealing. Analog tape recording has long used stationary proved them selves in terms of mechanical simplicity, low cost, long life, rugged and consolidated construction, and ease of editing. Indeed, in the professional re cording world, a stationary-head design is the only choice, particularly for multi-track recorders.

Recording engineers have always labored to accommodate the needs of artists, coming up with rather sophisticated production techniques and the hardware to back them up. One very cost-effective piece of hardware is the razor blade; tape recorded on a stationary-head machine may be cut and spliced to achieve much better technical and artistic results.

Overdubbing is supremely important in popular music. Instead of recording all parts of a group together, in real time, as is usually done in classical recording, the artists record basic tracks on a multi-track machine, then record additional tracks while listening to the old. The new tracks must be performed and recorded synchronously with the old ones. This requires a feature called "sync," common on professional recording decks, which allows the old tracks to be played back from the recording head rather than the normal playback head. (Since the record and playback heads on open-reel decks are about an inch apart, new tracks laid down while listening to the playback head would lag about an inch behind the other tracks on the tape, making them nearly 0.1 S out of sync.) The fidelity of the signal delivered to the artists' headphones in the sync mode won't be quite as good as if the playback head were being used, but the fidelity on the tape itself is unaffected.

Punch-in and punch-out-the ability to switch instantly from playback to re cord, and back, while the tape rolls-is also an essential creative tool. An otherwise acceptable track might be marred by a momentary flub-fingers caught in the guitar strings or a hiccup on a vocal track. Rather than re-re cording the entire track, the engineer can replay the track, and the deck can be punched in and out of the record mode.

With overdubbing, punch-in/punch out, and razor-blade editing, a song is thus recorded layer by layer, with only the best performances retained, then edited into a final product. With a stationary-head recorder, these corner stone operations are easily accomplished. Although it might be possible to devise a system with multiple synchronous rotary heads, the complexity and cost would be considerable, and precise punch-in and-out would be extremely tricky. Besides, tape splicing would never be practical on a rotary-head machine because of its helical-scan track format (Fig. 1); the razor would have to cut precisely on the same diagonal as the tracks, cut just on the boundary between tracks, and cut where an "odd" field would follow or precede an "even" one-all with the tracks of microscopic size and utterly invisible. So an electronic editing sys tem would be mandatory. For a stationary-head digital recorder to accommodate overdubbing, punch-in/punch-out and razor-blade editing, more sophisticated hardware is required than for the analog equivalent, but those obstacles have already been overcome. In general, any multi-track professional recorder, analog or digital, seems des tined to be a stationary-head design.

But as Chief of Intergalactic Widget, your main target is the profit from the mass consumer market, not the prestige of the smaller professional market.

If you rule out multi-tracks, that leaves only stereo consumer models. Are stationary-head recorders still advantageous? Yes, but to a lesser extent.

Without the advantages of editing, synchronous recording, and punch-in/punch-out capability, a stationary-head machine is still mechanically simpler and more rugged.

However, the data itself presents problems. Any digital recorder must be able to distinguish one data field from another. This is fairly easy in a rotary-head design, where the signal is divided into thousands of short, diagonal tracks. However, with a stationary-head recorder, special synchronization data words must be inserted as place markers on the tracks, which run the whole length of the tape. This adds to the data overhead, which brings us to the biggest problem facing a stationary-head design-the fundamental problem of data storage-bandwidth.

As we have seen, the bandwidth requirements for digital audio are large indeed; because of the sheer number of bits in a data stream, the frequencies to be recorded could be as high as 1 MHz. Thus, a stationary-head tape recorder has its work cut out for it; the problem can only be solved by high tape speeds or the use of multiple tracks to distribute the data load over a greater tape area. But high tape speeds lead to long tape lengths and large cassette sizes. To achieve the required data densities at a tape speed slow enough to afford a small cassette size, a stationary-head digital cassette recorder might have to record 20 data tracks in each tape direction.

Those tracks would have to be tiny, so tiny that strict alignment would be both vital and hard to achieve. Servo alignment systems would therefore be needed, adding both mechanical and electronic complexity. In addition, the high data density of a stationary-head digital cassette system would dictate the use of a special, thin-film head-a tough piece of hardware to manufacture. Suddenly, the apparently simple stationary-head design isn't so simple after all. . . .

A rotary-head cassette recorder al ready has precedents in the digital audio field, and with good reason. Video signals demand a bandwidth of more than 2 MHz; it was therefore logical to adapt rotary-head video recorders, which already had such bandwidth, to digital audio recording. To fully under stand the operation of a rotating-head video recorder, a look at a video signal might help. And to get a good look at a video signal, just flatten your nose to a television screen. You'll see the pixels--the triple pinpoints of primary colors which comprise the overall picture--and perhaps you'll also see the pixels arranged across the screen in slightly diagonal lines. In fact, your television has 525 such scan lines; using received video information, the electron beam illuminates each pixel as it scans across the line. To reduce flicker, all odd lines are scanned to make one "field," then all even lines are scanned to make another field; the two fields are interlaced on screen to make one complete picture ("frame") In total, 30 complete frames are scanned each second.

A video recorder uses a tape format quite similar in nature to the picture itself. Because of the high bandwidth requirements, a high tape speed is needed. This problem is neatly solved with a rotating head; the tape speed itself is slow, but the head moves across the tape quickly, thus yielding an effectively high tape speed. With helical scanning, two heads are mounted opposite each other across a drum, which is rotated at 1,800 rpm in a direction opposite that of the tape.

Since the drum is tilted across the tape path, the heads sweep diagonally across the tape. Thus, instead of a continuous tape path, a series of short diagonal segments is recorded, as shown in Fig. 1. Each segment is a video picture field, and so the tape is recorded as a series of odd and even sets of video information. Sync pulses are placed in the video information to separate video fields, and frequency modulation is used to overcome the tape's bandwidth limitations.


Fig. 1--Recorded pattern of videocassette recorder.

Of course, a video recorder will re cord any kind of program, as long as the data is properly formatted. For ex ample, we can rather easily take digital audio data, put it in a video format (with sync pulses, etc.), then record this "pseudo-video" audio data on a video recorder. This is how PCM processors work, converting analog audio to digital, then formatting it for use on a VCR. Indeed, this is the method used for all Compact Disc mastering; the standard processor used to encode the digital information onto videotape prior to making the disc master is the Sony 1610, which requires professional, 3/4-inch videotape cassettes rather than the 1/2-inch tapes used in home VCRs.

Of course, any rotary-head format designed for the consumer market would take the next logical step: The video format would be discarded, and an audio-only format would be specifically developed. This would result in greater cost efficiency because the format could be designed strictly ac cording to audio needs. But the essential advantages of the rotating head high recording density and low tape consumption--would be retained. The electronics required for such a system are relatively straightforward; since both audio channels are multiplexed to a single recorded track, redundancy of recording circuitry is avoided. In addition, data synchronization is easily achieved because the synchronization pulses that delineate even and odd fields are an inherent part of the for mat. The only major drawbacks are the more complex mechanical assemblies required for the rotating head, and a shorter head life than that of a stationary head.

The rotary-head design looks pretty good, but just how compact can the complex mechanism of a rotary-head tape player or recorder be? Small enough for the all-important portable and automobile markets? Also, though the stationary-head design is more complex electronically, the rotary-head design is more complex mechanically.

Through the years, manufacturing costs of electronics have decreased, while those of mechanics often have not... .

Widget's top engineer, Smedlap, nervously reviews the balance sheet for the Chief. Either format, stationary-head digital audio tape (S-DAT) or rotary-head digital audio tape (R-DAT), is entirely feasible from an engineering standpoint. With 16-bit quantization and 44.1- or 48-kHz sampling, either tape would be about half the size of an analog cassette, with a playing time of 45 minutes per side on S-DAT and 120 minutes (one side only) for R-DAT.

Smedlap ends his presentation, nervously awaiting the Chief's decision. Will it be S-DAT or R-DAT? The fate of Intergalactic Widget hangs in the balance. Suddenly the door to the board room bursts open and Ms. Meyerbeer, Widget's ace marketing analyst, storms into the room! She furiously pulls a Compact Disc recording of the Verdi "Requiem" from her purse and flings it across the room, neatly separating the Chief's toupee from his perspiring brow. Everyone's mouth drops wide open.

To be continued.

(adapted from Audio magazine, Feb. 1986)

= = = =

Prev. | Next

Top of Page    Home

Updated: Tuesday, 2018-09-18 8:06 PST