Manufacturer's Specifications
Tape Recorder
Sampling Frequencies: 48, 44.1, and 32 kHz.
D/A Conversion: 16-bit linear.
Frequency Response: 2 Hz to 22 kHz, ±0.5 dB.
Dynamic Range: Greater than 90 dB.
THD: Less than 0.05% at 1 kHz, 4 dBs input.
Wow & Flutter: Below measurable limits.
Power Requirements: 100, 120, 220, or 240 V a.c.; 50/60 Hz; 37 watts.
Dimensions: 17 in. W x 4 in. H x 16 3/8 in. D (43 cm x 10 cm x 42 cm).
Weight: 27 lbs. (12.2 kg).
Interface Unit
Rated Analog Input Level: +4 dBs (0 dBs = 0.775 V rms).
Maximum Analog Input: +24 dBs.
Rated Analog Output Level: +4 dBs.
Maximum Analog Output Level:
Digital Input and Output Levels and Impedances: Sony/Philips Digital Interface Format, 0.5 V peak to peak, 75 ohms; AES/EBU Format, RS-422, 110 ohms; Sony Digital Interface Format-2, TTL level, 75 ohms.
Sync Output for Sony Digital Interface Format-2: TTL level, 75 ohms; accuracy at 25° C (±2°), ± 30 parts per million; accuracy from 10° to + 60° C, ± 60 ppm.
Maximum Cable Length for AES/ EBU Format: 300 meters.
Power Requirements: 110, 120, 220, or 240 V a.c.; 50/60 Hz; 12 watts.
Dimensions: 17 in. W x 4 in. H x 15 3/8 in. D (43 cm x 10 cm x 39 cm).
Weight: 13.7 lbs. (6.2 kg). Price: $3,200.
Company Address: Sony Communications Products, 1600 Queen Anne Rd., Teaneck, N.J. 07666.
Although consumer versions of R-DAT machines remain unavailable in the U.S., pro R-DAT units are being eagerly gobbled up by small and large recording studios and other professional sound facilities. These R-DAT recorders generally cost quite a bit more than the consumer machines, but then again, they can do so much more.
Any Audio reader who absolutely, positively must have a DAT recorder right now can follow two courses of action. You can take a quick trip to Japan (or one of several European countries) and buy one; there is no prohibition against bringing R-DAT recorders through U.S. customs.
Alternatively, you can search for a local dealer who has made the journey to Japan. Doing either will, naturally, substantially increase the unit's final price. If you follow the first method, you'll be paying for the trip as well as customs duty. If you follow the second, you'll be paying a substantial increase over the retail price in Japan-and, because manufacturers do not usually recognize "gray market" dealers, you are not likely to be given a factory warranty.
Recording engineers have a third option. They can walk into a pro dealer and obtain an imported R-DAT recorder that is guaranteed by its manufacturer. The price will still be high, but the pro machine, as mentioned, has much greater capabilities than a consumer model.
The Sony PCM-2500 is such a unit. It consists of two physically matching components which can be screwed together; a pair of side panels is supplied for this purpose.
Cables are also supplied to connect the R-DAT recorder itself with the companion interface unit. But what makes the PCM-2500 a professional product? For one thing, this combination is able to record, via its analog or digital inputs, at either the 44.1or 48-kHz sampling rate. Consumer R-DAT units can only record at a sampling rate of 48 kHz via their digital or analog inputs--although they can play back prerecorded DAT cassettes that have been recorded at 44.1 kHz.
This restriction was voluntarily assumed by the makers of consumer R-DAT machines in an effort to placate the major record companies. Because professional R-DAT models like the PCM-2500 can record at 44.1 kHz, they can make recordings that can readily be transferred to digital editing equipment or used to make CDs. In addition, recordings can be made at a sampling rate of 32 kHz via the digital inputs; this sampling rate has become the standard for satellite transmissions in Japan and parts of Europe.
The PCM-2500, being a pro unit, does permit multiple sampling rates in recording as well as playback. I must stress that it does not permit dubbing a CD in the digital-to digital mode if that disc has been encoded with the copy inhibit flag. In addition to sampling-rate differences, the PCM-2500 interface unit permits the user to add pre-emphasis at will. It is equipped with XLR balanced-line connectors for compatibility with professional analog audio equipment. It is also fully compatible with three recognized digital interface standards that have evolved over the past few years; these are the AES/EBU, Sony/Philips, and Sony Digital Interface Format-2. (Incidentally, SDIF-2 is used in the Sony PCM-1630/1610 digital audio processor installed in many professional recording studios.) The R-DAT recorder of Vie PCM-2500 strongly resembles the Sony DTC-1000ES, the one-box consumer model sold overseas, which I reviewed in the July 1987 issue. Although their rear panels differ, in that the PCM-2500 incorporates the various connectors needed for linking it with the interface box, the front-panel features are identical. Specifically, three types of subcodes (start ID, program number, and skip ID) can be "written' onto the tape. Start IDs permit each selection to be located at high speed, and program numbers allow any desired selection to be searched for directly. Skip IDs mark portions of the tape which the user wants the deck to zip past during playback. All of these subcodes can be "written" to a tape even after it has been recorded, without erasing any of the program material itself.
As is true of the Sony DTC-1000ES, four-times oversampling, digital filtering, and separate D/A converters in each channel are used.
Recorder Control Layout
A power switch is at the left end of the panel, adjacent to the DAT cassette tray. Below this are a timer switch and a headphone jack. The tape-position indicator (which can show elapsed time within the current program, estimated time elapsed since the tape's start, or remaining tape time) is part of a large display area to the right of the tape tray. To the right of the position indicator are displays of program and "AMS" (Automatic Music Search) numbers. A pair of peak-level bar-graph meters, calibrated from below -50 dB up to 0 dB (followed by a red "Over" warning), take up the bottom of the display. Other indicators in the display panel are for tape-counter function, sampling frequency, presence of copy-inhibit code in digital input signals or in playback or its insertion in recording, pre-emphasis detection or use, skip and start ID (which light when IDs are written, erased, or detected during playback), and search mode actuation. Finally a "Caution" light glows when condensation is present on the tape heads.
Below the display are buttons to open and close the tape tray and to start and stop play, plus rockers for forward and reverse fast-winding and program search. To the right of these transport controls are smaller buttons for record, pause, and record mute. Still further to the right are numeric buttons for accessing programs directly and the large, dual concentric recording level knobs. At the lower right of the panel are a "Skip" switch (set to the "On" position when you want skip IDs to be recognized during playback), an "Input Select" button (for choosing analog or digital input recording), and a small headphone level control.
Interface Control Layout
There are far fewer controls and switches on the interface unit of the PCM-2500 than on the main recorder. At the left are a power switch and a switch to select wired or wireless remote control (both of which are supplied) or neither. The wired remote is essential when the recorder is not in the operator's line of sight, as often happens in recording-studio control rooms. A jack for the wired remote is below the switch. At the extreme right of the panel are recessed, screwdriver-adjustable record and playback level controls.
The record level controls adjust the interface unit's output level to the recorder within a range of -20 to -2 dB. The playback level controls adjust the output level at the analog output connectors within a range of -6 to + 12 dB. Several small toggle switches are at the lower right corner of the panel. The first selects one of the three available digital interface formats. Next come on/off toggles for "Emphasis" and "Master Safe." (When the latter is on, recording cannot take place, regardless of other switch settings.) The final toggle switches are "Copy Prohibit (Write/Off)" and "Sampling Frequency (44.1 kHz/48 kHz)." The "Sampling Frequency" switch is only needed when recording from the analog inputs. If the digital inputs are used, the PCM-2500 automatically sets its sampling rate to 48, 44.1, or 32 kHz to match the incoming signal. (The 32-kHz sampling rate is only available through the AES/EBU and Sony/Philips digital inputs.) Once the cables have been connected between the RDAT recorder and the interface unit (the supplied cables cannot be connected incorrectly since the plugs are all different), all signals are routed to and from the PCM-2500 via the interface unit.
The interface's rear panel has connectors for use with the three standard digital formats, a "Word Sync Out" connector for use with SDIF-2 signals, a ground terminal, a "Voltage Selector," and stereo pairs of balanced-line XLR analog input and output connectors. The XLR connections are wired with pin 1 as ground, pin 2 as the hot lead, and pin 3 as the return or cold lead. A line-voltage selector completes the rear-panel layout.
Measurements
The multilingual owner's manual incorporates some excellent diagrams that illustrate how digital and analog interconnections should be made between the PCM-2500's units and between the PCM-2500 and other equipment. It took me only a few moments to make the necessary connections between the PCM-2500 and the input and output terminals of my Audio Precision System One test equipment, after which the DAT recorder was up and running.
In preparing to test the PCM-2500, I thought it necessary to record several definitive test tones (via the analog inputs) and to play them back and measure the results via the complete record/play loop. This is the same procedure I would follow for any tape recorder. But the PCM-2500 is also a digital tape player, so I wanted to know how well it could play back the tape I have been using to test car DAT players. (This tape is a dub of the CD-1 test disc I use when evaluating CD players.)
Fig. 1-Record/play frequency response at 48and 44.1-kHz sampling rates.
Fig. 2-THD + N vs. frequency at 0 dB (maximum) recording level, for record/play
at 44.1 kHz. Dashed curve is right channel.
Fig. 3-THD + N vs. recording level for record/playback of a 1-kHz signal
via the digital inputs. Dashed curve is right channel.
Fig. 4--Residual noise vs. frequency for playback of "quiet" track
of test tape.
Fig. 5--Interchannel separation. Even after full record/play cycle, separation
was virtually identical from left to right and right to left, so only one
curve is shown.
Fig. 6A--Deviation from linearity for direct digital record/play of undithered
signals at 44.1-kHz sampling rate.
Fig. 6B--Same as Fig. 6A but recorded via analog inputs. Note that there
was less deviation from linearity in this mode.
Fig. 7A--Deviation from linearity at low signal levels, using dithered signals
played back from digitally recorded test tape at 44.1 kHz. Dashed curve is
right channel.
Fig. 7B--Same as Fig. 7A but for full record/play cycle of one channel.
Since the copy-inhibit flag prevents me from making further digital-to-digital dubs from my CD-1 disc, I also wanted to record the contents of my CD-1 dub using the PCM-2500 in the digital-to-digital mode, with my Sony DTC-1000ES DAT recorder as the playback device and its digital output fed to the digital input of the PCM-2500. After making comparison measurements for a few of the tests in this report, I eventually abandoned this latter approach, since I could see no difference in performance between playback of the copy I made on the PCM-2500 from the CD-1 dub and playback of the CD-1 dub itself. Therefore, though some of the test results that I will present were obtained by recording via the PCM-2500's analog inputs and some by recording via the digital connections, others were obtained by playing back my existing CD-1 dub. This combination of tests, I feel, presents a thorough analysis of this professional R-DAT recorder's capabilities.
First, I applied an analog frequency sweep signal from my Audio Precision test system to the analog inputs of the PCM2500. The recording was made at the 48-kHz sampling rate and was played back to produce the response curve shown in Fig. 1. Other than a very slight rise around 7 to 9 kHz, amounting to no more than about +0.2 dB, frequency response was virtually flat from 20 Hz to 20 kHz for this complete record/play cycle. The procedure was repeated, this time using a sampling rate of 44.1 kHz. These results, also shown in Fig. 1, were substantially the same (if anything, the slight rise was a bit smaller), so the curve has therefore been displaced by 1 dB for clarity. Again, the signal was supplied by my DAT copy of the sweep signal from the CD-1 test disc.
Figure 2 is a plot of THD + N versus frequency, using a signal level of 0 dB (maximum recording level). The results were obtained by copying my CD-1 dub, using the analog outputs of my DTC-1000ES and the analog inputs of the PCM-2500, and then playing back via the PCM-2500's analog outputs. For much of the frequency range, THD + N measured 0.007% for the left channel and 0.008% for the right. A rise in THD + N can be seen above 10 kHz, but even at 20 kHz, THD + N was still below 0.1%. Using direct digital connections for recording yielded somewhat better results. Figure 3 shows how THD + N varied as a function of recorded level for a 1-kHz signal. As usual, this graph depicts THD + N in dB referred to maximum recorded level, and at all levels, it was at least-92 dB. If one wanted to translate this to a percentage, THD + N would be less than 0.0025% at 0 dB and would be even better at lower levels. Since the distortion shown in Fig. 2 is 0.007% at 1 kHz and 0 dB, it's obvious that some quantization distortion is added when going through the entire record/play cycle via the analog inputs and outputs of the machine. The additional THD + N is by no means audibly significant, but it is there, nevertheless. Although Sony's literature does not mention the methodology used in their A/D conversion (during recording), I have ascertained that it does not involve oversampling nor does it employ digital filtration. This may possibly account for the added quantization distortion.
Overall, A-weighted signal-to-noise ratio of the PCM2500, after recording and playing the "silent" track of my CD-1 through the deck's analog inputs, measured 90.93 dB for the left channel and 91.72 dB for the right. By contrast, when I played back the "silent" track from the same tape on the PCM-2500 and plotted the residual noise (Fig. 4), at no frequency did the noise level exceed-105 dB relative to maximum recorded level. Again, this shows that some noise is contributed by the PCM-2500's analog input circuitry.
Figure 5 shows how separation varied with frequency. This is a worst-case result, having been derived from a test tape that was put through the entire record/play cycle. At 10 kHz, separation in either direction was still around 88.9 dB, decreasing slightly to around 88 dB at 16 kHz. At 1 kHz, separation was more than 110 dB between channels.
The results of the two linearity tests represented by Figs. 6A and 6B are especially interesting. In Fig. 6A, a series of signals, each decreasing in amplitude, was recorded from my CD-1 tape copy onto a new tape in the PCM-2500. The 44.1-kHz sampling rate was used, as was the digital-to digital recording mode from my own DAT recorder to the PCM-2500. Deviation from perfect linearity was certainly not excessive--only 3 dB at -90 dB recorded level. However, when the same signal was recorded from the analog outputs of my consumer DAT deck into the analog inputs of the PCM-2500, playback resulted in even less deviation from linearity (Fig. 6B). The only explanation I can offer is that the combination of the D/A converters in my unit and the ND converters in the PCM-2500 may have produced a positive going deviation from perfect linearity which nearly cancelled out the negative-going deviation seen in Fig. 6A. The plots in Fig. 7A were obtained when I used the PCM2500 to play back the low-level, digitally recorded dithered signals of my CD-1 dub. Maximum deviation from perfect linearity was -4 dB at -100 dB recorded level. Going through the entire record/play cycle via the analog outputs and inputs resulted in a somewhat smaller deviation from linearity (Fig. 7B). These results further confirm my theory regarding the cancelling effects of ND and D/A errors in the two recorders. I had no way to determine whether the compensation was being afforded by circuits in the DTC1000ES consumer DAT recorder or in the A/D converter section of the PCM-2500. In any case, these differences are so slight that, from an audibility standpoint, they are rather academic. Then, too, there's no predicting whether the D A and ND converters in any other PCM-2500 will exhibit exactly the same linearity characteristics as my sample.
Interesting results were also obtained when I employed the fade-to-noise signals from my digital tape copy of the CD-1 test disc. Initially, I simply played back this track of the tape in the PCM-2500. Results are shown in Fig. 8A. Next, I recorded and played back the same signal, using the analog outputs of my own DAT machine and the analog inputs of the PCM-2500 (Fig. 8B). Note how much more noise is evident at the extremely low levels, from around 100 dB downward, in Fig. 8B. Clearly, the additional noise must be coming from the D/A converters of my own DAT machine, the A/D conversion process of the PCM-2500, or both.
Finally, I recorded the monotonicity test signals of my CD test dub. A slight amount of jitter during playback made the signal difficult to "lock" or synchronize on my oscilloscope, but Fig. 9 is still sufficiently clear to illustrate the nearly perfect uniformity and symmetry of the signal's positive going and negative-going steps.
Fig. 8A--Fade-to-noise test, for playback of digitally recorded test tape
at 44.1 kHz.
Fig. 8B--Same as Fig. 8A for full record/play cycle via analog inputs and
outputs. Predictably, the noise level is higher than in digital-to-digital
recording.
Fig. 9--Monotonicity test.
Use and Listening Tests
After using my own DAT machine, which has only 48-kHz digital recording capability and does not let me add emphasis to my recordings, it was a joy to be able to use the full range of technology built into the R-DAT Standard. I did not have access to any of the equipment that utilizes the special professional digital interface modes, but even being limited to DAT digital-to-digital recording, I became increasingly angry that this technology is not granted to us "ordinary" consumers.
It's worth mentioning again how quickly a DAT machine can access a given numbered selection. High-speed search occurs at something like 200 times normal tape speed! Being able to number selections, either while they are being recorded or during post-production work, is also a most convenient feature, as are the numerous helpful displays made possible by the R-DAT subcodes.
In addition to test signals, I recorded my voice using an old pair of Beyer 500 microphones and an equally old little TEAC mixer. Of course, neither the mikes nor the mixer were up to the quality of the PCM-2500, but I was still impressed with the noise- and distortion-free sound in playback. In the past, whenever I have heard analog recordings of my voice, I have tended to say that they didn't really sound like me.
I've always thought that this was a normal reaction, because we don't hear ourselves as we sound to others. Surprisingly, when I played back my voice recordings made on the PCM2500, I could honestly say that the voice I heard did sound like Len Feldman! I transcribed several CDs, too, though I had to copy them through the analog outputs of my CD player and the analog inputs of the PCM-2500. In some cases, I deliberately switched from the 44.1to the 48-kHz sampling rate in the midst of recording. During playback, I could detect only a slight click at the point where the switching had been done. There was no difference in perceived sound quality or tonal balance.
As I testified many times during the fight against Copy Code, DAT represents a new and better type of professional and consumer tape recording technology. It should be available, in appropriate form, to both professionals and serious audio enthusiasts who are willing to pay for it. Until it is, I suspect that those of us who want this technology at any cost and who can't make a pilgrimage to Japan or Europe will simply have to find some way to purchase-and afford-the professional PCM-2500. In so doing, the buyer will get not only a superb tape recorder, but the very facility the recording company giants hoped to deny consumers-the ability to do digital-to-digital taping.
-Leonard Feldman
(Adapted from: Audio magazine, Feb. 1989)
Also see:
Sony PCM-1 Audio Unit (Mar. 1980)
Sony TCD-D3 DAT Walkman Recorder (Jan. 1991)
Hitachi PCM-V300 Digital Audio Recorder (Jun. 1982)
Sony Model TC-366 Three-Head Stereo Tape Deck (Aug. 1970)
= = = =