Sony CDP-X77ES CD-player
It is much more difficult - I am tempted to say impossible-to keep up with Compact Disc player development these days. Sony is only one of very many manufacturers in the field but here I am, delving into a machine which is the direct successor of a model, the CDP-X7ESD. However, in this case the justification is simple, everything considered this CDP-X77ES rates as the best CD player I have tested to date. Looking back, if you care to, you will find that I was less than enthusiastic about that CDP-X7ESD, which struck me as a hurried product, lacking in the finer points of design we seek in the ES range, not at all sure about its market direction, possessed of a number of 'sillies' and above all expensive. Fortunately Sony and I remained on speaking terms and although a number of those sillies are carried through to this model, because a large part of the earlier design has not been changed significantly, other important matters have received due attention-not least, I am sure you will agree, a drop of £300 in the price!
The outward appearance of the X77 differs from its predecessor by just one, almost unnoticeable, addition to the controls: a tiny button to switch the fluorescent display on and off. This addition also features amongst the 48 buttons on the hand held infra-red remote commander but unfortunately no one at Sony has yet thought fit to reinstate that most useful facility, the ability to control volume from one's seat. Instead, all their attention over the past twelve months has been directed to the implementation and inclusion in this machine of the highest form of Sony's gestation of the 'one-bit' digital-to-analogue conversion process. With the typical Japanese love of high-flown terminology, this is called "Sony High Density Linear Converter System" and uses PLM (Pulse Length Modulation-for reference Philips use Pulse Density Modulation and MASH-primarily Technics - Pulse Width Modulation). All these related and very similar techniques are aimed at overcoming some of the limitations of the existing methods of conversion by removing fundamental sources of inaccuracy rather than pursuing a 'tidying up' approach which, for all practical purposes, probably peaked some four or five years ago with converters like the Philips TDA1541 series.
In essence those DACs consist of a group of 16 electronic switches, each of which controls a specific current set by a very accurate internal resistance in the ratio of 1,2,4,8,16 etc. to 32,768 so that by permutation they can provide 65,536 possible total output currents. The 16 switches are turned on and off by digital information derived from a short sample of the original sound, processed and captured on the disc. There are 44.100 of these samples every second, each one of which has to reset those 16 switches simultaneously. It is easy to see that minute differences in the timing of these switches, particularly the ones controlling the higher values, can produce some unfortunate spikes in the resulting final current for that sample and that equally minute departures from 100% accuracy in the resistors which decide the current, can upset linearity as they are used both singly and in combination. Even though the resulting rapid succession of changing currents is smoothed off by filters which restrict it to the audible 20kHz frequency range, it is no mean achievement to reproduce the original sound with the accuracy now prevailing.
The attractive alternative to the process outlined above is a single switch and a single resistor which would operate on one bit at a time instead of all 16 bits simultaneously. At a stroke this would remove most of the sources of error in recovering the waveform but would have to operate at speeds which are quite impossible now or in the immediate future (44,100 x 216 = 2,890,137,600 operations per second). What we are actually saying here is that provided the original signal does not exceed the 20kHz audio band, if we check on it often enough it won't have had time to change by more than one of those 65,536 steps-if at all. But do we have to examine all those steps so often? All we really need to know to establish an absolute value is if, from time to time, the signal is increasing or decreasing and how fast.
This is the basis of all the 'one-bit' systems, although the steps to achieving it, whilst running on parallel paths, differ in degree and in the arithmetic employed. All are based on the now familiar technique of oversampling and a process called noise shaping which uses a sort of digital negative feedback to reduce the 16-bit data dramatically to only a few bits, any resulting error being returned to the input over and over again until it results in a definite decision plus the generation of quantizing noise; the latter is spread over a large frequency area, most of which is outside the audible range and is subsequently discarded. In a practical system this is accomplished in stages, the proportions of which vary between proponents, and results in data which can be converted at speeds realizable by their various chip manufacturing facilities, Sony being the fastest so far, working at over 45MHz, 1024 times the 44.1 kHz sampling frequency of CD. I must stress that this is a very simplified summary of the operation and there are a number of clever tricks incorporated which are too complex to deal with here and properly belong to a future textbook. In the present case all this is accomplished in a new Sony integrated circuit DAC, type CXD2552, which now appears in all their new range of CD players and in the X77 is duplicated in a complementary arrangement for even further enhancement of the sound quality.
As I hinted earlier this player has provided yet another benchmark on CD's progress (the previous one, as far as I am concerned, was their 1987 CDP-555ES) and I am still in shock after some of the measurements it threw up. Not only that, but some of my test gear threatens to go on strike if pushed to such extremes again. In fact it jibbed at the distortion check which was less than its reliable threshold of 0.001%. Weighted signal-to-noise ratio was about 120dB but the wideband noise was spoilt by considerably mucky RF (Radio Frequency) leakage at quite high frequencies, including some 30mV peak-to-peak spikes at the 45.158MHz clocking rate of the new chip. As none of the controls caused this to vary it must be concluded that its presence is due to some unwanted coupling in the internal layout.
Frequency response was flat to better than 0.05dB from the fixed phono outputs over a range of 5Hz to 20kHz, although the signal voltage was over the top at 2.52 Volts, 2.2dB high, at a source impedance a shade over 200 Ohms. The nominally 600 Ohms balanced outputs (actually 660 Ohms) when correctly loaded showed a drop of about 0.6dB at both ends of the scale; this is a good result for a transformer coupling. This facility with its driver stages and XLR type sockets must add considerably to the price, perhaps as much as £100; but I wonder how many will use it? Linearity was excellent right down to the - 90db level and distortion at these low levels was close to the theoretical minimum. Crosstalk between channels, even at 20kHz, was virtually non-existent and it was just possible to see the effect of switching on and off the fluorescent display on the silent channel, its presence producing a small damped wave-train showing up in the noise spectrum on the oscilloscope trace. I have had several machines through my hands lately which presented a fine set of measurements but this Sony has managed to top them all with considerable ease.
Coming down to earth for a moment, the X77 incorporates both Custom Edit and Custom File, the former, together with Fade In and Fade Out, help the operator to construct a composite recording on to a cassette. The latter lets you select preferred tracks or playing order from about 180 discs and put this choice into operation every time you return to that particular disc. Of course someone else may have played it in the meantime and banked a different selection-this can lead to domestic harmonic distortion! All the usual features of today's players are also present: Shuffle play, Single play, Repeat play, A to B repeat, Slow and Fast Music Search and Direct Track Access. Nearly all of these are only available via the remote control, only a minimum of operations being available on the uncluttered fascia of the machine itself. The fluorescent display immediately below the soft rubber sealed disc drawer is similarly rather restricted.
The quality of construction and finish are well in keeping with the ES flagship standard and in particular the disc handling mechanism and the laser pod, both making use of precision die castings, are quite in the scientific instrument class. The latter, driven by Sony's established linear motor, moves with a speed and accuracy which makes the rack and worm driven pinion which was the norm only a few years ago seem primitive indeed; the latter is now only used for the drawer loading mechanism. The disc is rotated by a small direct-drive brushless multi-coil motor of a type which Sony has also perfected over the years and as a result the mechanical noise is inaudible in the quietest room. Completely separate power supplies with individual mains transformers are provided for the digital and analogue sections of this player and the internal copper-plated steel screening divisions rigidly isolate the two. A futher sign of this 'keep 'em in their place' discipline is the heavy gauge two-metre mains power lead, reducing any common impedance right back to the wall socket. Optical and digital coaxial output connectors appear on the rear face together with two pairs of gold-plated phono sockets, one for the fixed 200 Ohms source outputs and the other from the front panel volume control (which also sets the level at the adjacent headphones socket) where the source impedance can rise to over 5k Ohms around the mid point of its travel.
How it performed
The first few seconds of my listening tests were enough to convince me that with this X77 Sony had returned to the top of the league. In no time at all I was dredging through my CD collection for remembered passages in a number of discs which have spelt trouble for lesser models and this went on and on until the realization struck me that I had not yet sat down and listened to a whole piece of music so that the X77's talents could register in depth. Because they were due to play it the next night at a Promenade Concert my hand fell on the excellent Rite of Spring which Simon Rattle and the CBSO recorded for EMI at the Warwick Art Centre a couple of years ago (CDC7 49636-2; 11/89) but then-such is the beauty of my filing system!-right next to it was one of the first CDs and one that I still regard as amongst the best orchestral recordings of all time: The Firebird by the Concertgebouw under Colin Davis. Analogue it may be and old it certainly is, but surpassed it is not (400 074-2, 3/83-Philips at the height of their powers). In the end I played through both and what a revelation it was. Here is a sound which is certainly the equal, maybe just the musical master of my long term reference, being blest with a beguiling subtlety which pulls you into the minds of conductor and composer. The obvious improvement in stereo definition may have had something to do with it - it is this rather than any change in the quality of sound which first attracts one to this machine: everything has firmer and better defined outlines, a sharper photograph of the scene if you like, taken with a small aperture so that depth and distance are registered in great detail. Unlike all the other 'onebit' machines I have met with so far this one puts out a clean, fresh rendition of the original sound without a trace of the busy action which one senses is taking place behind the scenes as each note is put together.
I know that other people, perhaps carried away by the technological arguments in favour of this new system, have heaped praise upon it but failed 10 take in this shifting background activity, a sort of signal related fuzz, similar in some respects to analogue tape modulation noise but much more subtle. Sony has certainly put paid to that one. An interesting example cropped up later when playing a disc of the Jacques Loussier Trio made in Paris in 1986, their first to be digitally recorded I believe (Start CD SCD2). Here it was easy to hear as never before the much exaggerated placing and multi-tracking of a number of microphones on the drum kit-no one has arms that long! However, in several of the Bach pieces done in the Loussier style, percussion plays no part and these mikes were obviously shaded off, leaving a very much more realistic piano and plucked bass to carry on. But just at one point Arpino, sitting there waiting, brushes in hand, can't resist a pass at the cymbal and it is real.
Let me end with a cautionary tale. I took this X77 along to a friend who has a first issue 1983 CD player and other quite respectable equipment, which I won't name, of equal vintage; he had asked to hear a 'state-of-the-art' machine. Regrettably, the X77, although different, was little better and certainly nothing like it can be. Remembering Paul Miller's AES lecture about the effects of RF on some equipment I returned a few days later with a couple of 2.2nF capacitors across the variable outputs (taking advantage of the variable source impedance) and this experimental rolloff did the trick. The friend was then as much impressed by my 'magic' as he was by the X77!
Tempted though I am, I shall not be replacing my old faithful 555ES yet; Sony have some more work to do first. That amount of RF leakage is careless, the omission of the remote control of volume strikes me as perverse and the expensive balanced output is unnecessary; but the sound itself, now that is something else.