Sony CDP-R1/DAS-R1 CD player
Consider for a moment their literal meanings. "No holds barred" means that if anything could have been done to make the product better, it would have been. "Cost no object" means no expense has been spared in producing this paragon of perfection which is destined to remain unsurpassed until the manufacturer's next incremental improvement comes along. Both terms make for great hype, but reality dictates a rather more practical approach.
To begin with, many of the things which are known to improve the sound of an audio product are open-ended. Take voltage regulation, for example. Improving this improves the sound, but improving it even more improves the sound even more. So where do you stop? Should a head-amp's power supply have 100 Farads of capacitor storage, a current capability of 30,000 amperes, and a signal-modulation ripple of 0.00000012%? Of course not; that's absurd. The product would be much too bulky and much too expensive for anyone to consider buying. But, taken literally, "no holds barred" and "cost no object" would mandate such design extremes. Idealism is all well and good, but one has to be reasonable about such things. You make things as nearly perfect as you can, up to the point where further improvement would not, in your opinion, justify the cost, and let it go at that.
The fact is that, protestations of perfection notwithstanding, even the priciest product designs must involve many design compromises. All manufacturers know this, even if they don't tell their advertising department about it. So, next time you read about a no-holds-barred, cost-no-object audio component, think of what such a product would entail and be appropriately skeptical. Having said that, we can now consider Sony's new CD player, which, although not claimed to be no-holds-barred or the state of the art, clearly aspires to be both.
Apart from any questions of actual value, the R1 looks as if it is worth every penny of its considerable price. All sides except the ends are of heavy-gauge aluminum with gold anodizing, the end panels are of what appears to be high-gloss rosewood (they're actually Tamo Ash), and each unit weighs enough to give the impression that there's a lot of hardware in it.
The CDP-R1 Transport
The player unit contains the laser-optical player mechanism, servo system, signal processing and error-correction circuitry, and a dedicated power supply. The simplified front panel shows only seven pushbuttons and a largish rectangular area for the disc drawer and fluorescent readout display. At the far left bottom corner is the AC switch, and to the right of the display are buttons for Door Open/Close, Play, Pause, and Stop. Above these are two smaller ones for Forward/Backward track change (Automatic Music Selector). All these controls, with the exception of AC power, are duplicated on the remote control unit, from which all other operations are controlled. And some of those other operations are unique to the R1.
Along with all the usual functions, including Shuffle Play (whose utility I still question), there's one called Custom File. This allows you to program-in data and playback instructions for each individual disc, then store them in a permanent memory for automatic recall each time the disc is played. You can store text, which is displayed on the DAS's LCD screen each time the disc is loaded; you can program the order in which tracks are played; and you can mark up to six of your own Index points on a disc for later call-up, just as though they had been originally encoded on the disc.
When you load a disc and close the drawer, the player scans the disc's "table of contents" (TOC), and a microprocessor notes the number of tracks and their precise playing times. If no Custom File programming is done, the TOC information is erased next time the drawer is opened, but if you store any Custom File data for that disc, both the data and the TOC information are stored in a nonvolatile memorymemory which is retained when the player is turned off and even, for about a month, if AC power is lost. There's room for storing data for up to 226 discs, and each time another disc is loaded, the microprocessor compares the TOC data with what it has in its memory. If they match, it automatically brings up the program for that disc. The identification is virtually infallible; the chances of two discs having identical numbers of tracks of identical playing times is zilch!
This echoes Philips' "Favorite Track Selection," and is well thought out and pretty much bullet-proof. I'm not convinced, however, that it will be of much value for the average audiophile. Some of the Sony's options include filing the name of the artist or record album, the name of the person who gave you the disc, or numerical data showing such things as equalizer settings (God forbid!) and the optimum volume control setting for that particular recording. This is fanciful stuff, considering that the LCD only displays 10 characters, and you can't have more than one display per disc. (Try abbreviating "Beethoven: Symphony No. 6" down to a meaningful 10 characters, and if you have two versions of the work, just forget about identifying one as Herbert von Karajan and the other as Felix von Weingartner. All this stuff is on the disc box anyway, and if you don't know what's in the player when you have the box in your hand, it's time you took an evening off to reorganize your collection or yourself. If the disc was a gift, the place to note that is on the album insert.)
Most consumers will probably only use Custom File's programming feature, while music educators, broadcasters, and people in film post-production work will appreciate the Custom Indexing feature and its ability to shift an index point forward or back by 0.15-second intervals. But I find it hard to see how Custom File's other uses can justify their inclusion. When the time comes that it can display up to 10 lines of text on a video screen, Custom File will be a worthwhile addition to CD's already impressive lineup of convenience features.
Sony's design engineers seem to have a phobia about mechanical resonance. If there was any possibility of anything in the R1 resonating, they damped it. The player frame is shock-mounted and made of an antiresonant mineral-loaded resin, the disc door has an acoustical seal around it which excludes sound waves when it is closed (to prevent them from vibrating the disc), and the player's feet are supposed to serve a damping function, although I fail to see how "extremely hard" ceramic feet with non-flexible rubber "isolators" can provide any isolation except at frequencies so high that environmental disturbance would probably not be a factor anyway. Even some of the discrete parts, like the larger electrolytic capacitors, are isolated from their circuit boards by soft rubber bushings to prevent resonance. How much all of this meticulous attention to minutiae yields improved sound quality can only be guessed at. I could only judge the end result, which is very impressive.
The transport's tracking servo uses two predictive techniques to minimize the time and the amount of servo current needed to "recover" from a blemish-induced signal dropout. One of the subcodes recorded along with the signal on a CD is a sequential time code, which allows (among other things) the player to find any spot on the disc according to its time from the start of the track. This information can also be used to get the optical head "back on track" after a surface scratch or blob of dirt has momentarily interrupted its "view" of the signal track. When an obscuring blemish has passed, the system looks for the time code closest to but higher than the last one it saw, and repositions the head over that track. The correction is almost instantaneous, but the sudden draw of extra servo-supply current can cause a momentary depletion of the player's power supply. The more positional correction needed, the more current the servo must draw momentarily from the power supply, and this can cause a brief drop in power supply voltage, leading to time jitter in the signal coming from laser pickup. With conventional players, it can also affect the DAC performance.
To minimize such voltage dips, Sony uses servo position prediction, which calculates the most likely future position of the optical head on the basis of what the servo was doing just before the system lost its view of the signal path, and correction-time prediction based on the rotational rate of the disc, in order to get the head as close as possible to where it should be before final correction is needed. This reduces the time required and the amount of power-supply depletion when final correction is performed. A very clever idea, but . . . The CDP-R1 is one of the few CD decks that hardly needs this kind of thing.
It contains no analog circuitry and no D/A convertershardly anything that would be affected by variations in power-supply voltage; all that stuff is in the DAS-R1. So the predictive correction would appear to be unnecessary. In fact, it does serve two purposes: It reduces the small thumps produced by counterforces in the tracking system when recovering from dropouts, and (in theory) it reduces distortion during the positional recovery period.
The DAS-R1 D/A Converter
The separate DAS-R1 is a multifunction digital/audio converter housing the sample-and-hold circuits, four time-staggered 16-bit DACs, 8x oversampling, 18-bit digital filters, the audio section, the master clock oscillator, and its own power supply. It has three digital inputs, individually selectable from a front-panel rotary switch, and the unit's sampling rate switches automatically (32, 44, or 48kHz) to match the sampling rate of the source.
Though the transport has its own internal clock (synchronization) generator, this is disconnected when the two-way fiberoptic coupler pair is used to carry signal data from the CDP to the DAS and the clock signal from the DAS to the CDP. This use of one clock is necessary, according to Sony, because of the extreme demands of the 8x oversampling on the timing accuracy of the clock signal controlling the DAC switching. The DAS-R1 also has two standard coaxial serial data inputs, selectable from the front-panel Source switch, and will handle any source that delivers a compatible digital bitstream. (For example, it will work with DAT, digitized satellite audio, and any CD player that has a bitstream output. It won't work with the Sony PCM-F1, even though that has the requisite 44.1Hz sampling frequency, because the digital data available from its output are organized into a format compatible with the video frame rate.)
A major problem with 16-bit D/A converters is that the binary digit representing the bottommost encodable signal-level change (the least-significant bit) often produces increments that are too large or too small, causing asymmetrical decoding of very-low-level signals. One solution to this, found in a lot of so-called 18-bit players, is to use a 16-bit DAC with a register-shifter, which moves all bits upward by two steps when signal level is low, and downward by two steps when signal level is high. The idea is to keep the system operating as much as possible above the LSB level, where the greatest nonlinearity is likely to occur, but this approach can result in switching glitches each time such register shifting takes place. The DAS-R1's use of time-staggered 16-bit DACs and an 8x oversampling filter results in 18-bit resolution, requiring no register shifting. It also allows for more accurate conversion from numbers to voltages. Here's how.
The number of discrete volume levels available from a digital system is equal to 2 to the power equal to the number of bits. Thus, 16 bits gives 2 to the power 16, or 65,536 different levels, while 18 bits gives 262,144 different levels. At 18 bits, there are four possible signal levels for every 1 in the original bitstream. For example, instead of counting (from the LSB) 0, 1, 2, and 3, 18 bits can count 0, 0.25, 0.5, 0.75, 1.0, 1.25, and so on. It is customary for 18-bit designs to round off the fractions by halves, assigning 0 to all values of 0.25 and 0, and assigning 1 to all values of 0.5 and 1. Any resulting errors can cause a grainy hiss that Sony has dubbed re-quantization noise.
In the DAS-R1, each sampled quantization level is treated individually; no numerical rounding-off occurs. If a binary 0.25 is delivered 4 times, the system outputs 0, 0, 0, and 1. There may still be errors, but now they occur at twice the frequency rate they would otherwise, placing the resulting re-quantization noise at four times the original 44kHz sampling rate instead of twice that frequency. Most of this is then closed out by the sample-and-hold circuit (the aperture chopper).