Sony DTC-1000ES R-DAT recorder & Sony PCM-1630 A/D converter

With the furor over the launch of DAT (digital audio tape), it is worth remembering that commercial digital recorders have been with us for some time now: it is now nearly a decade since studio PCM converters were first successfully used for audio recording in conjunction with a VCR (which uses a helical spiral recording via a spinning drum to achieve the high writing speed).

Such systems evolved into such relatively popular units as the Sony PCM-F1, used with the matching SL-2000 VCR for many amateur and professional recordings, and Technics even introduced a PCM console recorder with a built-in VHS transport. Nowadays, however, with professional multitrack digital recorders becoming commonplace, the VCR method has been largely superseded by linear tape recording. Multitrack digital recorders, however, are very expensive, and considerable scope remains for the PCM "adaptor" type of recorder, which exploits the high information-storage density of established inexpensive video recording systems.

When a true domestic DAT system was proposed, two systems emerged. One was based on the use of a stationary multitrack head for which thin-film technology was proposed, 20 or so parallel digital tracks being required for the data and error-protection data for a stereo audio channel. For commercial reasons, the Japanese industry wanted to get DAT onto the domestic market and apparently could not wait for the thin-film technology to mature. Rotary-head designs could deliver the required information-storage performance, but conventional VCR mechanisms were thought to be too big and noisy for serious use in a domestic audio recorder.

The development of the compact 8mm video medium provided the potential for very small cassettes and tape widths. From this it was but a short step to define a unique miniature "video" type cassette, specify the complementary rotary-scan video-type recording method, and dedicate it to consumer audio recording, resulting in the R-DAT format. The design includes a synchronizing timecode to allow quite rapid access to any part of the tape and the modest head-wrap angle of 90° allows for the use of easy-loading mechanisms, as well as very fast wind speeds, at present 200 times the play speed.

The commercial result is a range of first-generation domestic machines, looking not unlike large CD players, with a similar drawer-load tray and a very similar control keyboard and numeric display. Automatic track programming is a standard feature of the R-DAT designs: once a tape has been recorded, play modes and track selection operate just like a CD player. Compared with an older PCM/VCR system, or even the professional PCM-1630/U-matic combination, the R-DAT recorder is a joy to use, with fast-access quick rewind of 2.4s/minute (50s for a two-hour tape), and the mechanism is also reasonably quick by domestic CD standards.

Potentially, R-DAT provides the consumer with a digital recorder of near-professional quality, allowing, in theory at least, very high-quality recordings—better than large open-reel sound but with cassette convenience.

For comparison, Sony's two-track professional machine is the PCM-1630, a unit in the $23,000 class when bought with a matching U-matic recorder. The 1630 is well established in the industry and is specified by Polygram for CD mastering. As its lab performance will show, it is something of a reference machine. I was fortunate enough to have access to a 1630 for extensive lab testing, and the results give some idea of the advances which have been made with regard to PCM recording technology. Against this background, it was instructive to subject the budget domestic equivalent, in the form of Sony's DTC-1000ES R-DAT recorder, to virtually the same battery of tests (footnote 1). How would the domestic equivalent compare with the professional model at over 10 times the cost?

Sony PCM-1630
Taking the 1630 first, a look at the manual gives some insight into its internal organization. As with many professional units, the design is founded on a large, rack-mount enclosure incorporating an internal backplane (an array of vertical multiway sockets). The electronics are carried on a number of circuit-board assemblies which slide into the enclosure from the front and plug into the backplane sockets. The enclosure constitutes the mainframe, and the backplane wiring, often quite complex, routes signals and power supplies from board to board.

The two balanced input signals enter the 1630's back via XLR sockets and semi-preset level controls, and are routed to the A/D board via 20k ohm series resistors; the first stage is a low-noise 5532 op-amp (a dual 5534) with differential input clamp diodes. Up to 10V RMS may be applied to this input as it is followed by a subsidiary level control, with maximum attenuation limited to 36dB. Another op-amp follows, whose frequency response is determined by an FET-switched capacitor, this the recording pre-emphasis option. Two more biFET op-amps drive the thick-film, brickwall, anti-alias filter, which chops frequencies above 21kHz; more op-amps add feedback/phase correction to the filter. The filter stage for each channel is buffered by yet another 5534 op-amp, followed by the 16-bit linear A/D convertor, a Sony CX-20018, driven by three peripheral IC op-amps; in this case, 356s. The digital data for both channels are then conditioned and processed on the modulator board to constitute an NTSC standard video signal ready for recording.

The 16-bit linear DAC board is stereo, and has a considerable amount of circuitry, with Sony CX 20152 chips used for D/A conversion. Industry-standard IC regulators—7815 and 7915 types rather than the lower-impedance 317/337 types—supply the ±15V op-amp rails, one set per channel. Many small L/C filters are present throughout the power supplies and lines in order to filter and isolate stages from each other, suppressing the digital pulses and their harmonics.

The interesting part of the 1630, as compared with earlier professional PCM machines, is the D/A section, which employs 2x oversampling with digital filtering. It appears to owe much to the design technology of Sony's DAS-702es domestic CD decoder (see Stereophile Vol.9 No.8 for the '702 and Vol.10 No.4 for the similar '703 decoder). Quite a lot happens to the two stereo channels after the oversampling/digital filtering and D/A conversion. The output passes through LF356 op-amps, via the logic switch deglitcher, to a "2525" IC filter stage, using inductor and feedback capacitor methods. This buffer drives the main, thick-film technology, AF LI/201, low-pass filter, this labeled identically to the record input filter. Four more ICs are present, mainly 5534s, for additional filtering, phase compensation, and de-emphasis. The final op-amp drives two class-A/B discrete output stages per channel, one for the "hot" and one for the "cold" halves of the balanced differential output.

I have included this circuit description because I feel that it might give some insight into the sound quality of this processor. It should be noted that both the record and the replay level chains are essentially DC-coupled throughout, so questions of coupling capacitor quality do not arise. However, given that every active stage plays a part in the sound-quality chain, it is worrying to find 13 of them in the record/replay path, excluding the A/D and D/A converters. It is known that, under critical domestic listening conditions, we can hear the loss imposed by a single 5534 op-amp in a preamp line stage. Considering the recording chain as a whole, if the audio has passed through a mixer console it may at best have encountered another dozen or so 5532/4 stages, or, even worse, through stages with less well-considered ICs, as well as through a number of coupling transformers. It strikes me that a lot needs to be done to improve the design practice of recording systems.

Sony DTC-1000ES
I was not able to perform an accurate head count on this R-DAT machine but from a quick look at the boards, I would estimate the number of active stages to be one third that of the 1630, while the physical signal-path length is probably a quarter, and without the need for a backplane or balanced input/outputs. Assuming a similar digital-section performance, one might expect the DTC-1000 to sound more transparent purely as a consequence of its relative simplicity.

As befits its "ES" suffix, the DTC is carefully designed in the Sony Esprit tradition. The chassis is copper-plated to combat induced currents circulating in the metalwork, and the case, like that of the PCM-F1, is damped to reduce acoustic vibration. The analog and digital power supplies are separated—a clear change from the first-generation Sony CD players—and it is clear that some considerable care has been taken over the choice of components: ELNA Cerafine capacitors, for example, are used in both reservoir and decoupling applications.

As with the 1630, the record section is dual-channel, using two of the same Sony CX-20018 ADC chips, non-oversampled. There is one input op-amp leading to the good-quality Soshin anti-alias filter, a thick-film encapsulated model. Two further op-amp stages lead to the A/D. Once in digital form, the coded audio is multiplexed and routed to the main digital processor, the two R-DAT chips, CXD-1008Q and CXD-1009Q. These, in conjunction with two blocks of 64k memory, handle the error-protected transfer of data to the rotary-head recorder section and back.

Footnote 1: J. Gordon Holt reviewed Sony's DTC-2000ES DAT recorder in November 1994.—John Atkinson
Sony Consumer Products
Sony Drive
Park Ridge, NJ 07656
Product not yet released in the USA (1987)

Axiom05's picture

Maybe this is a dumb question but why do we see no pre-ringing from the anti-aliasing filter?

John Atkinson's picture
Axiom05 wrote:
why do we see no pre-ringing from the anti-aliasing filter?

It's a minimum-phase filter, presumably operating in the analog domain.

John Atkinson
Technical Editor, Stereophile

bdiament's picture

We sure have come a long way from the 1630 and DAT, thank goodness!
While Apogee's retrofit filters provided an appreciable increase in 1630 performance, the limitations of the 16-bit system, particularly at low levels, could not be overcome.

And though DAT certainly seemed convenient, I found the shelf life to be rather limited. After a few months, many tapes developed sections that played back sounding much like a fax transmission! I wouldn't want to archive to such a format.

Happily, we now have systems that can deliver a superbly clean 24-bit capture at 192 kHz.
When I first heard digital back in early 1983, my feeling was that it took vinyl 100 years to get to where it was, and that in another 100 years, digital will have arrived too. ;-}
I feel most fortunate that we didn't have to wait anywhere near that long.

Best regards,