Manufacturer's Specifications:
Sampling Frequencies: 48, 44.1, and 32 kHz; see text.
Frequency Response: Standard, 2 Hz to 22 kHz, ± 0.5 dB; long play, 2
S/N: Standard, more than 93 dB; long play, more than 92 dB.
Dynamic Range: Standard, more than 93 dB; long play, more than 92 dB.
THD At 1 kHz: Standard, less than 0.004%; long play, less than 0.08%.
Rated Input Level: Line in, -4 dBm; coaxial digital input, 0.5 V peak to peak ± 20%.
Rated Output Level: Line output, -4 dBm; coaxial digital output, 0.5 V peak to peak, ± 20%; headphone, 0.6 mW into 32 ohms; optical digital output wavelength, 600 nm.
Maximum Recording Time: Standard, 120 minutes; long-play mode, 240 minutes.
Power Requirements: 120 V a.c., 60 Hz, 32 W.
Dimensions: 18 1/2 in. W x 4 1/2 in. H x 13 in. D (47.0 cm x 11.5 cm x 33 cm).
Weight: 18 lbs. 5 oz. (8.3 kg).
Price: $950.
Company Address: Sony Dr., Park Ridge, N.J. 07656, USA.
It's been a long time coming, but DAT recorders are finally available in the United States. The first company to introduce DAT through authorized distribution channels in the U.S. was Sony Corporation of America, who, in late June of this year, provided DAT units to dealers handling their high-end "ES" line. A few weeks later, the company introduced a somewhat lower-priced unit to other, mainstream dealers. Having owned home and portable DAT units for nearly two years (obtained during trips to Japan although these models were also available through so-called gray market dealers in the U.S.), I was particularly interested in how these late-generation recorders from Sony differed from earlier models. The most outstanding difference, of course, was the price. First-generation players, even in Japan, sold for $2,000 or more in early 1987, when they were first introduced. By introducing the DTC-75ES at under $1,000, Sony has taken a major step toward ensuring the success of this new technology.
Another major difference between the earlier and later models is the inclusion of SCMS on the DTC-75ES. This is the so-called compromise that permits direct taping from sources such as CD via the digital channels but limits digital copying from the resulting tapes. It does this by adding a code to any recording made via the deck's digital inputs so that other decks with SCMS will refuse to record the digital bit-stream from those recordings. The exception is if the original source material contained a code that turned the SCMS encoding off. Taping via the analog channels is unrestricted.
A final difference between this and earlier Sony DAT recorders is its ability to record in the long-play mode. By moving the tape at half speed (4.075 mm/S as against the standard 8.15 mm/S), using a lower sampling rate (32 kHz instead of 48 or 44.1 kHz) and using nonlinear 12-bit 'instead of linear 16-bit encoding, maximum recording time is doubled, to yield up to four hours per tape. The trade-offs for the lower sampling rate and word length are, respectively, poorer frequency response (flat to just under 15 kHz) and slightly poorer signal-to-noise ratio and dynamic range. This slow speed, it seems to me, would be ideal for recording FM programs, where frequency response is also limited to 15 kHz and dynamic range falls far short of that attainable on DAT even in the long-play mode.
Both the D/A and A/D converters use Sony's High Density Linear Converter system, a variation of the much-publicized one-bit conversion method. The HDLC system also uses a 45-bit, noise-shaping digital filter and has a direct digital sync stage to reduce time-base errors known as jitter. Other new technology includes a four-stage feedforward circuit that improves error-correction capability and a new digital servo IC that reduces servo control variations caused by component aging and temperature drift. A four-motor transport results in both quick loading and fast track-to-track access time.
As for user convenience features, the Sony DTC-75ES offers high-speed music search (200 times normal playing speed), music scan, three-way repeat, 10-key direct-access track selection, and 60-track programmability. In addition, various subcode interactive features built into the DAT standard can be utilized, including "Start ID," "Skip ID," and "End ID" functions. A digital fader lets you make professional fade-outs and fade-ins.
Start ID codes, which mark the beginning of each selection, can be entered automatically or manually. Skip and end IDs are entered manually. When the deck encounters a skip ID, it stops playing and then fast-forwards to the next start-ID code; end IDs, at which the deck stops automatically, can be used to mark the next recording point on a partially used tape, or to mark material at the end of the tape that you don't want to play. Program numbers can be "written" onto a tape during recording or later, in a renumbering process. Since all these subcodes are written onto the tape separately from the recorded digital audio data, they do not affect audio quality.
Control Layout
The switches for power, external timer operation, and recording mode ("Long" or "Standard") are located at the far left of the all-black front panel. The DAT cassette loading compartment is nearby; unlike the tape trays of other DAT recorders I have seen, it has a small window through which the user can see the tape when it is loaded into position. The counter reset and counter mode buttons are immediately to the right of the tape compartment. Absolute time, elapsed time of the current selection, and total remaining time on a tape can be selectively displayed. Clustered beneath the display are the buttons that control the tape transport, including the functions already familiar from analog cassette decks, an "Open/Close" button for the tape compartment, and forward and reverse "AMS" (Automatic Music Search) buttons used to locate the beginnings of selections during playback. The buttons that initiate fast forward and rewind when the tape is stopped can be used in playback mode for audible searching. Number buttons, at the right of the display, are used to call up a selection on the tape by number.
Also here are the "Clear" button, used for cancelling selections, and a "Music Scan" button that plays the first few seconds of each selection on a tape. The "Fader," "Repeat" mode, "Skip Play," and "Margin Reset" buttons are found just below the number buttons.
The "Margin" indicator on the display is extremely useful when recording via the analog inputs. It constantly monitors and displays the loudest level being fed to the recorder, expressing it in dB below maximum recommended recording level. (When recording via digital inputs, levels automatically match those of the incoming signal.) In digital recording, exceeding maximum record level causes horrendously high distortion, so it's vital to be able to monitor levels this way even if instantaneous peak reading level meters are also provided. The "Margin" display always shows the highest level seen at the analog inputs, and updates that reading when a higher signal level is encountered.
Buttons for writing or erasing start, skip, and end IDs, and for renumbering start IDs are all to the right of the tape transport buttons. Further to the right are dual concentric record level controls (used only when recording via the analog inputs), an input selector for the analog or the optical or coaxial digital inputs, and a stereo headphone jack and level control.
The large display, which shows every conceivable aspect of DAT operation, also provides level indications during recording and playback. Among the more unusual displays is the "Rehearsal" indicator, which is used when manually writing a start ID during playback of a tape. This rehearsal mode facilitates more precise entry of the start ID than the record mode because the precise point at which the ID is written can be shifted backward or forward in increments of 0.3 seconds while the current start point is played over and over allowing you to judge the direction in which it should be shifted for the final written start ID. Other visual displays include sampling rate, the three time indications mentioned earlier, the type of input employed (optical, coaxial, or analog), program numbers, recording mode (standard or long play), and the previously described "Margin" indication in dB. The supplied remote control has a button to turn off this display, though Sony doesn't say whether that's to eliminate any possible noise from its digital circuitry or to make it less distracting. Other functions found only on the remote include an "A-B" repeat button, buttons for entering and checking sequence of programmed selections, "CD Synchro" buttons to synchronize tape motion with a Sony CD player when copying, and buttons to control Sony CD players. The remote also duplicates many of the controls on the front panel including transport controls, the numeric keypad, and the various ID write and erase buttons.
The rear panel is equipped with two types of digital inputs and outputs (coaxial and optical) as well as with analog pairs of input and output RCA-type jacks.
Lab Measurements
A true evaluation and test of any DAT recorder must include lab measurements via the digital as well as the analog inputs. Furthermore, since the digital inputs will accept digital signals at sampling rates of either 44.1 kHz (as from a CD) or 48 kHz (as from another DAT), I felt it necessary to make at least some tests at both sampling frequencies. Finally, some measurements had to be made in the long-play mode via the analog inputs only.
Fig. 1A--Frequency response for recording in standard mode with 48-kHz sampling
via analog inputs (top) and digital inputs (middle), and via analog inputs
with 32-kHz sampling in long-play mode (bottom)
Figure 1A shows the frequency response obtained at the analog outputs for signals recorded in the standard mode (linear 16-bit with 48-kHz sampling) from the analog and digital inputs and in long-play mode (nonlinear 12-bit with 32-kHz sampling). Since recordings made via the analog inputs are automatically digitized at a 48-kHz sampling rate, you can see that flat frequency response extended slightly beyond 20 kHz, to around 22 kHz, before dropping off quickly due to the low-pass filter action required in any digital system. The slight rise in amplitude noted at the low end was probably the result of something in the analog stages and appears in all four response measurements. It amounted to no more than +0.15 dB at any point.
Similar results were obtained when I recorded a digital frequency sweep, with a 48-kHz sampling frequency, using my Audio Precision System One. I did not program this sweep to go beyond 22 kHz, avoiding the sharp drop-off visible in the previous curve.
Next, I recorded a frequency sweep signal via the analog inputs, this time in long-play recording mode. This mode's 32-kHz sampling rate limits the possible frequency response to a top end of around 15 kHz, as shown.
Fig. 1B--Frequency response, via digital inputs, for recording in standard mode
at 44.1 kHz.
Figure 1B shows the response obtained via the digital input, with 44.1-kHz sampling, using my CBS CD-1 test disc as a source. The disc was played on a CD player equipped with an optical digital output connected to the DTC-75ES' optical input via a fiber-optic cable. Results were similar to those of Fig. 1A except that response was now limited to very slightly more than 20 kHz.
Fig. 2--THD + N vs. frequency for (top to bottom) analog input in long-play
mode, analog Ok 20k input in standard mode (48 kHz), and digital input (44.1
kHz).
Figure 2 shows distortion versus frequency for signals at maximum level. When signals were applied via the analog inputs (middle curve), THD + N at all audio frequencies was less than the 0.004% that Sony quotes for 1 kHz. When the test was repeated via the optical digital inputs (bottom curve), THD + N was considerably lower, measuring less than 0.002% over most of the audio range. The top curve in Fig. 2 was produced in the long-play mode, with its lower sampling rate and lower bit count.
Incidentally, the block diagram for the DTC-75ES implies that one would get the same readings going through the A/D encoder and D/A decoder as when actually recording signals and playing them back, and this is true for the standard recording modes. In long-play mode, however, just going through the A/D and D/A circuits gave me distortion readings virtually as low as they had been for standard modes. I knew something was wrong, as Sony would not have listed a THD rating of 0.08% for this mode if it was in fact only half that much. Sure enough, when I actually recorded and played back my test signals in the long-play mode, my results were very close to Sony's. This may be because the long-play mode uses nonlinear 12-bit coding rather than the linear 16-bit coding of the standard modes, but it's also possible that long-play recording involves some element not shown on the block diagram.
Figure 3A shows how THD + N of a 1 kHz signal varied with amplitude, using the analog inputs. Interestingly, "0 dB" as indicated on the level meters (and as read by the numeric "Margin" display) really does not correspond to maximum digital level. The steep rise in THD that occurs when applying too great an input signal to a digital recording system did not occur until my input level was 5 dB greater than indicated on the meters. Obviously, Sony calibrated their metering system this way because they were aware of the fact that most recording enthusiasts accustomed to analog cassette or open-reel recorders tend to allow signals to push meters "into the red." In analog recorders, slight overload results in a gentle increase in distortion. In digital recording, when you "run out of binary ones," distortion rises to unacceptably high levels immediately. If you should purchase this superb DAT recorder, my advice would still be to keep levels low enough that you never go above the arbitrary "zero" dB level established by Sony.
I used my CD-1 test disc once more to couple a digital optical signal to the digital input and to show how THD + N varied with amplitude for a 44.1-kHz digitally sampled input.
Results are shown in Fig. 3B. Since the CD-1 test disc has no signals above "maximum recording level," the distortion shown in Fig. 3B never rises above normal levels. Note, however, that when using the digital input this way, overall distortion level remained about 95 dB below maximum recording level-an improvement of about 5 dB over results obtained in Fig. 3A.
Fig. 3A--THD + N vs. signal amplitude, referred to 0-dB meter level, for recording
via analog input. Fig. 3B--THD + N vs. signal amplitude, referred to absolute
0-dB level, for recording via digital input.
Using the analog inputs once again, I measured the actual harmonics generated when recording and playing back a 1-kHz test tone. Figure 4A shows results using the analog input. With an indicated recording level of 0 dB, no harmonic component exceeded 0.002% in amplitude. When this measurement was repeated at a-60 dB level, the distortion components increased and the noise floor rose, as expected in a digital system; for example, a major harmonic component at 16 kHz measured 0.3% of the (now reduced) signal amplitude. Figure 4B is similar to Fig. 4A, except that this time a digital signal, at the 44.1-kHz sampling rate and maximum recording amplitude, was applied to the optical digital input. While harmonics were still negligibly low (not much more than 0.003% for any one of them), the only reason I can think of as to why these components were higher than those obtained in Fig. 4A is, as already stated, because "0 dB" as shown by the metering system when using the analog inputs is really lower than maximum permissible recording level. On the other hand, "0 dB" digital signals as derived from playback of my CD-1 test disc are truly maximum-level signals.
Fig. 4A--Spectrum analysis of 1-kHz signal, recorded via analog inputs at 0 dB
and -60 dB indicated level. Fig. 4B-Spectrum analysis of 1-kHz signal recorded
via digital input at 0 dB actual level.
The benefits of digital-to-digital recording, already obvious from some of my earlier measurements, became even clearer when I began to measure signal-to-noise ratios under various conditions. Using the analog inputs (with no signal applied and the input jacks shorted), A-weighted S/N at the analog outputs was 92.7 dB for the left channel and 92.8 for the right channel. These results were referred to 0 dB as indicated by the level meters and the "Margin" display. Recall, however, that actual maximum recording level is as much as 5 dB higher, so one could maintain that, relative to actual maximum recorded level-via the analog inputs-S/N was closer to 98 dB. In fact, when I played the "no signal" track of my CD-1 test disc, again using the optical digital connection from the CD player to the DTC-75ES, S/N measured 103.5 dB for one channel and 103.6 dB for the other channel. Using a 48-kHz sampling rate generated by my Audio Precision system, but with no audio modulation, S/N measured 99.5 dB on one channel and 98.0 dB on the other. In the long-play mode, S/N decreased, but only to just over 92 dB. Those familiar with digital audio theory may, at first, be surprised at that figure. S/N ratio is supposed to be about 6 dB/bit and, this being a 12-bit system, theoretical S/N should be about 72 dB. The figure is higher because this is a nonlinear 12-bit system that involves some companding.
Figure 5 shows third-octave spectrum analysis of residual noise, measured at the analog outputs, for both the analog and digital inputs. With the digital inputs, residual noise was about 10 dB lower at any given frequency, and noise peaks attributable to the power-supply frequency and its harmonics were far less prominent.
Figure 6 shows the excellent linearity from input to playback output of the DTC-75ES. Using the analog inputs and a sampling rate of 48 kHz, deviation from perfect linearity at an indicated-90 dB was only 0.45 dB for the channel shown, and about +0.48 dB for the other channel. (Since the meter's 0-dB point is actually about-5 dB, these readings actually show performance at about-95 dB.) Using the digital inputs and 44.1-kHz sampling, deviation from perfect linearity at an actual-90 dB was only 0.32 dB for the left channel and 0.79 dB for the right channel.
Stereo separation, as measured via the analog inputs and outputs, is shown in Fig. 7. At low and middle frequencies, separation exceeded 80 dB, decreasing to between 71 and 75 dB at 20 kHz. The cecrease in separation at higher frequencies was no doubt caused by capacitive coupling, either internally or between the external cables connected to the analog outputs of the recorder.
Fig. 5--Spectrum analysis of residual noise with no audio signal applied,
for analog and digital inputs.
Fig. 6--Linearity (output vs. input) for analog and digital inputs; see text.
Fig. 7--Stereo separation vs. frequency, analog inputs.
Use and Listening Tests
I have owned a DAT recorder for some time so perhaps my familiarity with such products may color my evaluation of how easy these products are to use. Still, I think even a newcomer to the world of digital audio tape recording will be delighted with the features that have been built into the R-DAT format-just about all of which have been implemented in the Sony DTC-75ES. Even before I began measuring the performance of this unit in my lab, I couldn't resist the temptation to record a couple of my most recent CD acquisitions onto DAT. As with most DAT recorders, start IDs and program numbers are automatically written at the start of each track, triggered either by a CD's own start codes when copying via the digital inputs or, when recording via the analog inputs, by the pauses that normally occur between tracks; on some earlier machines, one had to use the "Renumber" function (which is automatic but takes several minutes) to apply program numbers at each start ID point. I also appreciated the fact that the DTC-75ES can display absolute time indications based on time codes recorded on the tape; some machines I have dealt with in the past have simply approximated elapsed and remaining time based on motion of the spindles that drive the DAT cassette hubs.
The fade-in/fade-out feature, on the other hand, is not yet common. It lets you fade recordings in and out even when recording through the digital inputs, which are not affected by the Sony's record level control. If you press the "Fader" button while the deck is in record-pause mode, recording starts and the signal fades in; if you press the button during a recording, the signal level is gradually reduced to zero, after which the deck goes into record-pause mode. For playback, pressing the button when in pause mode starts the tape and fades the sound in, while pressing it during play fades the sound out and puts the deck in pause. (In either recording or playback, the fader does not affect the signals at the digital outputs.) An especially pleasant surprise was the variable fade duration from as short as 0.2 S to as long as 15 S. I tried recording a couple of CDs two ways: First using analog outputs from my reference CD player into the analog inputs of the DTC-75ES, and then using a digital-to-digital interface to record the same program material. While both recordings sounded great, I could distinguish some very slight differences between the original CD and the played back analog-to-analog recording. This was especially true of some of the organ music contained in a new Delos CD (Organ Music of Vierne and Reger, Delos DE3096); less so in comparing my analog-to-analog recording of a Denon CD featuring a Mozart serenade for violin and orchestra (Denon CO-73676). There was no distinguishable difference between the original CDs and the digital DAT equivalents. In making these comparisons, I was particularly careful to ensure that levels were set identically as I switched back and forth between the original CD and its DAT "copy." I also recorded some FM programs, using the long-play mode and the lower, 32-kHz, sampling rate. Of course, here I had to trust memory, since I could not compare the FM program with the recording of it, but it seemed to me that the quality of sound was in no way compromised through the use of this long-play mode. I would not, however, recommend using the long-play mode for transcribing CDs to tape or for any application where widest frequency response and lowest distortion are of paramount importance.
Obviously, the nearly four years during which political considerations have delayed the introduction of DAT recorders to the U.S. have been well spent by Sony. The DTC75ES, Sony's latest DAT recorder, not only outperforms its earlier models in every way, but can be bought for less than half the cost of that first-generation unit. That being the case, I can't imagine how any serious amateur tape recordist will be able to resist owning a DTC-75ES. Rejoice! The long wait for DAT is over.
-Leonard Feldman
(Audio magazine, Nov. 1990)
Also see:
Sony DTC-2000ES DAT Recorder (Jan. 1995)
Harris XD-001 UH DAT Recorder (June 1989)
Sony TCD-D3 DAT Walkman Recorder (Jan. 1991)
Akai AD-93 (DAT) Digital Audio Tape Recorder (Jun. 1988)
DECIDING ON DAT: Witness of the Persecution (Mar. 1988)
Nakamichi 1000 DAT Recording System (Nov. 1989)
= = = =