|
|
Every once in a while, it’s déjà vu all over again.
“What’s that?” she asks me.
“Lamborghini Diablo... 500 horsepower V12.. .does better than 200 mph.”
“How do you get in?” “Doors pivot up.” “Can we get one?” “Where would you put the groceries?”
Days later a messenger arrives, wheeling an instrument case. “Where do you want it?” he asks.
“Here will do.” Her piano stops. “What’s that?” “Sony’s new SACD player.” “What’s an SACD?” “A better CD.”
“What’s wrong with the old one?”
“Nothing, but this one’s better. It’s Super Audio.
“Where does the disc go?”
I point to a heavy plate on top. “In there.” “Oh, does it pivot up like that Lamborghini?”
“Lamborghini. I’m not sure; I think it slides to the side. Help me get this thing into the lab. It weighs a ton!”
The SCD- 1 is on the bench, plugged in, powered up, and I push “eject.” Sure enough, the plate rises slightly and glides to the side.
“Wow! Gorgeous! Looks like a space ship.”
“I’ll tell Sony you like it. Now stop fondling it and get out of here so I can get to work!”
“Can we get one?”
“That depends on how this re view turns out.”
The SCD-1 really is gorgeous, and it is heavy. Totes like a power amp, not a CD player. But then, it’s only incidentally a CD player—primarily, it’s the world’s first player for Super Audio CD (SACD), the Sony/Philips answer to DVD-Audio (see my feature article in this issue). The SCD-l reeks of class and no- holds-barred engineering, looks al most handmade, and costs 5 grand. I can’t imagine that Sony expects to sell tons of these machines, but what a way to launch a format!
The SCD-1 is stereo-only. When multichannel SACDs arrive, it will play just the two-channel mix-downs encoded on them. Meanwhile, the Sony can play conventional CDs, single- and dual-layer high-density (HD) SACDs, and hybrid SACDs. Hybrids are dual-layer discs whose outer, HD, layer carries Direct Stream Digital (DSD) data and whose inner layer carries conventional CD data. That inner layer, which enables hybrid discs to be read by ordinary CD players, is one of SACD’s major pluses.
--
Dimensions: 17 in. W x 574 in. H x 17¼ in. D (43 cm x 14.9 cm x 43.6 cm).
Weight 58.3 lbs. (26.5 kg).
Price: $5,000.
Company Address: 1 Sony Drive, Pork Ridge, NJ. 07656; 201-930-1000; www.sel.sony.com.
--
Except for hybrid discs, the SCD-1 detects what type of disc has been inserted and sets itself up accordingly. With hybrids, you can choose which layer to use, but you must decide before hitting the play but ton—no toggling on the fly. The SCD-1 has two lasers, one emitting light at a wave length of 780 nanometers for CD playback, the other emitting 650-nanometer light for HD layers. Sony also uses a dual-laser approach in its DVD players. But despite the similarity in density between the HD layers carrying DSD data and a DVD-Audio disc (DVD-A), the SCD-1 cannot recognize the latter.
Like my all-time favorite CD player, Sony's CDP-XA7ES, the SCD- 1 moves the disc over a stationary optical pickup and laser rather than the other way around. This is claimed to reduce variations in servo cur rent that can influence the data and, ultimately, the sound. Whatever. Like the CDP XA7ES, the SCD- 1 sounds great! And also like the XA7ES, it's slow as molasses when reading a disc's table of contents, skipping tracks, or doing anything else that requires lateral disc movement. I guess it's easier to pivot an optical head than to reposition a disc drive motor and all but for this kind of performance I'm willing to wait.
The SCD- drive motor is a new design.
Its bearing, which has a sapphire sleeve and a ruby thrust ball is said to be super durable and to reduce friction and motor servo current variation, thus avoiding data contamination from servo induced power supply noise The drive is built on a 6 millimeter (that s almost ¼-inch!) milled aluminum plate, with the spindle cutout reinforced by a second plate. (No wonder this beast is heavy.) The mechanism rests on four pillars and is isolated from the chassis by dampers to quell external vibration, while the full SCD-1 sits on "eccentric" isolation feet that use gel cushions sandwiched between dissimilar metals (brass and cast iron) to damp resonance, And instead of the usual automatic chuck that holds the disc in place, Sony includes a weight that you place on the disc by hand.
(Fail to do so, and you get your wrist slapped by a panel warning.) Nothing short of an earthquake upsets this player! Each optical sensor feeds an RF (radio frequency) amplifier, mounted close by to minimize RF path length. An RF processor chip extracts and synchronizes the clock and error-corrects and demodulates the data for both CD and SACD, but these data streams are so different that we're really talking about different circuits that share a single chip.
At this point, 16-bit PCM data from a CD or a hybrid disc's CD-compatible layer pass to a new VC24 (24-bit variable-coefficient) digital filter. This filter oversamples the data to 64 times the original sampling frequency and noise-shapes it to produce a 1-bit data stream at a clock rate of 2.8224 MHz, much like the DSD stream from an SACD.
An SACD's DSD data takes a different path. First, a DSD demodulator reads an "invisible watermark" pattern on the disc that confirms the disc is not a counterfeit; only if the watermark is found will the de modulator accept the bit stream. It then ex tracts such subcode data as track numbers and text, unpacks the audio, and rearranges it into a continuous stream.
The DSD data stream consists of pulses whose widths are proportional to signal amplitude at each point. So simply running those pulses through a low-pass filter would restore the data to the analog domain-if the circuitry were ideal. Real circuitry never is, however, so Sony came up with a clever dodge, Accurate Complementary Pulse Density Modulation (ACPDM), to squeeze out some extra performance. Even the fastest transistors take time to switch be tween the voltage levels that represent is and Os in a DSD signal, causing overshoot and ringing at every transition. In unmodified DSD bit streams, transitions occur only when data changes from 1 to 0 or from 0 to 1. There are no transitions between consecutive is or strings of Os, so the number of transitions, each with a characteristic ring, varies with the data pattern. This variation causes distortion.
Sony's ACPDM gets around this by forcing complementary transitions in each clock period. Instead of 1 remaining 1 for the full period, it is forced to 0 for a short time; similarly, Os are forced to 1 for a short time. This lowers the average signal level (no different from having chosen lower reference levels to begin with) but ensures that in every clock period there will be complementary 0-to-i and 1 -to-O transitions whose overshoots cancel each other out.
Thus, the data pulses reaching the D/A converter will have accurate amplitude.
Between the SCD-l's ACPDM system and its D/A converter is another Sony novelty, S-TACT (Synchronous Time Accuracy Controller), which is used for both CD and SACD playback. Power-supply noise caused by high-speed digital filtering can vary the axis crossings and affect pulse width. If allowed to occur, these variations, a form of jitter, can affect the audio. S-TACT is said to isolate the pulse generator and other digital operations from the D/A converter and en sure that each pulse's duration will be con trolled with quartz-crystal accuracy.
Whether it originated as DSD data or is an oversampled 1-bit representation of 16 bit PCM data from a CD, the signal at this point is a 1-bit pulse stream. The voltage pulses are converted by the SCD-l's Cur rent Pulse D/A converter into fixed packets of charge that are integrated by a capacitor to restore the analog signal. The final stage is the analog low-pass filter needed to sup press the quantization noise that was shifted to the ultrasonic region by the noise shaper and, in the case of DSD, not re moved by an interpolation filter.
= == = =
TEST RESULTS
Except where noted, results in the graphs are for the Sony SCD-1’s left channel (the right’s performance was essentially identical), unbalanced output, and the “Standard” settings of the output and interpolation filters.
As you’d expect, the effects of the SCD-1’s filter options are visible in the curves of frequency response. In Fig. 1A, made with Sony’s SACD test disc, response with the “Standard” output filter is flatter between 20 Hz and 20 kHz but is more extended with the “Custom” option: —2.94 dB at 60 kHz with “Custom” versus —2.72 dB at 50 kHz with “Standard.” Furthermore, output at 80 kHz is about 10 dB higher with the “Custom” setting. As usual with balanced outputs, treble response droops slightly. Figure 1B shows CD frequency response (using the CBS CD-1 disc) with the player’s “Standard” interpolation-filter setting and with settings #1, #2, and #4. (Response with filter setting #3 was almost identical to that of setting #4 but without its ±0.02-dB ripples.)
The Sony SACD test disc contains two “mute” tracks. “Analog Mute” keeps the A/D converters alive, as they would be in real-world operation. The other track, “Digital Mute,” is computer-generated “digital silence” (unrealistic, but the same nonsense used to get great S/N numbers from CD players). The noise spectra for the “Analog Mute” tracks (Fig. 2A) show how the SCD-l’s two output filter settings affect ultrasonic energy (the effect of noise-shaping used in recording). The noise energy peaks slightly above —60 dBFS at 65 kHz with the “Standard” filter setting and at nearly —50 dBFS at 80 kHz with the “Custom” setting. The difference doesn’t seem like much, but 10 dB is a tenfold increase in power, and the space between the two “Analog Mute” curves from 50 kHz to a few hundred kilohertz indicates there’s quite a bit more ultrasonic energy hitting the tweeter with the “Custom” setting. I understand Sony’s reluctance to have it used in systems not designed for it. In Fig. 2B is a spectral analysis of the SCD-1’s output with “digital silence” and with a 1-kHz, —60 dBFS signal. (Use caution when comparing Figs. 2A and 2B; their vertical scales differ.) There are no power-supply or hum components in the CD curves (Fig. 2B), but there is a trace of 120-Hz power-supply ripple in the SACD curves (Fig. 2A). Nonetheless, with “A” or CCIR weighting (which rank frequencies by their audibility), the signal-to-noise ratio for this player’s SACD system bettered its CD performance by almost 1.5 dB. Even more impressive, this also held true when the SCD-1 was playing the SACD “Analog Mute” track, which includes noise from the recording chain’s A/D conversion.
In Fig. 3A, the THD + N scale runs from —100 to —105 dBFS rather than my usual —80 to —100 dBFS; with a DSD test disc, the SCD-1’s THD + N at 1 kHz bottoms Out around —104 dBFS, more than an order of magnitude better than most CD players! Because the difference of about 0.6 dB between the top curve (made with a measurement bandwidth of 22 Hz to 22 kHz) and the bottom one (with a measurement bandwidth of 400 Hz to 22 kHz) probably stems from the 120-Hz ripple mentioned above, I consider the bottom curve to be the more meaningful.
Figure 3B, made with a test CD, shows the effects on THD + N of three of the SCD-1’s interpolation-filter settings. With the "Standard" setting, performance is outstanding. The curves taken with setting #1 and setting #3 were close enough to those with the "Standard" filter to warrant their omission. But with setting #2 and setting #4, all hell breaks loose over the upper 20 to 40 dB of dynamic range.
Figure 4A plots THD + N versus frequency for SACD and CD test discs, Both channels are shown for CD mode but only the left channel for SACD CD performance (taken with the "Standard" setting of the interpolation filter) is Performance with the DSD recording is even better-less than 0.002%, worst case. True, the SACD curve was taken at -3 dBFS (the recording level on the disc) and the CD curve at 0 dBFS, but I don't think that made much of a difference. Great, great either way! Figure 4B shows THD + N for a test CD and four settings of the SCD-l's interpolation filter, (Note that the vertical scale is grossly different.) Setting #3 isn't bad until you get almost to 20 kHz, where aliasing appears. But it's no match for the "Standard" setting, whose distortion heads down above 16 kHz. (Don't look for that curve in Fig. 4B; it's in Fig. 4A.) The curve with setting #4 reminds me of a rather punk CD player from days of yore, while the fold-down distortion permitted by setting #1 and setting #2 is just dreadful, Sure, those filter settings sound different; even golden ears can hear 10% to 40% intermodulation distortion!
Figure 5 compares linearity error in CD and SACD modes. The SACD test disc contains tracks for assessing linearity down to -160 dBFS, but I'm presenting results only to -120 dBFS because noise in my setup's tracking filter affected the measurements below that point. I'm convinced the SCD-1 produces the equivalent of 20 bit PCM performance on DSD recordings-true no bullshit 20 bits not some dumb-ass claim. Great, in my book!
Not that the SCD-1’s CD performance is an to sneeze at: With less than 0.1 dB of linearity error at -100 dBFS for a dithered signal, only DSD beats it! The fade-to-noise linearity error in CD mode (Fig. 6) is about the best I can recall. There s virtually no error at -110 dBFS, and the 3-dB noise-induced error point comes in below -118 dBFS, far better than one has any right to expect from a 16-bit PCM system. I couldn't run this test with a DSD recording, as there's no fade track on the SACD test disc; from other tests, I'd guesstimate the 3-dB noise-induced error point at close to -128 dBFS.
In CD mode, the choice of setting for the interpolation filter had no effect on linearity error or dynamic range and (except for setting #2, which was atrocious) had little impact on quantization noise. Results for quantization noise were the best I've logged from CD except for Sony's CDP XA7ES and Meridian s 508-24 With respect to DSD, whose data is not quantized to more than 1 bit, I'm not sure what (if any) significance a quantization noise measurement has. So all I m reporting is that it was substantially better than you can get from CD Channel separation was far better than anyone needs, especially via the SCD-1 s unbalanced outputs. Those jacks also have lower output impedance than the balanced connectors so I see no reason to use the balanced outputs unless you're running very long lines, experience a ground loop, or are just married to balanced connections. Channel levels matched perfectly, and the 0-dBFS output levels were the same for both CD and SACD playback.
Fig. 1A—Frequency response, SACD mode, for “Custom” output filter with unbalanced jacks and for “Standard” output filter with balanced and unbalanced jacks.
Fig. 1B—Frequency response, CD mode, for various settings of interpolation filter.
Fig. 2—Noise analysis for SACD mode (A) and CD mode (B).
Fig. 3-THD + N vs. level for SACD mode (A) and CD mode (B).
Fig. 4-THD + N vs. frequency for SACD mode and for CD mode with
"Standard" setting of interpolation filter (A) and for CD mode with other settings of interpolation filter (B).
Fig 5-Linearity vs level; Fig. 6-Fade-to-noise test, CD mode.
=====
MEASURED DATA
Line Output Level at 0 dBFS: Un balanced, 2.22 V; balanced, 4.4 V.
Line Output Impedance: Unbalanced, 33 ohms; balanced, 395 ohms.
S/N re 0 dBFS, with Digital Silence Track: CD mode, 106.3 dB, A- weighted, and 97.6 dB, CCIR weighted; SACD mode, 107.6 dB, A-weighted, and 99.1 dB, CCIR weighted.
S/N re 0 dBFS, with Analog Mute Track, SACD Mode: 107.2 dB A-weighted, and 98.6 dB CCIR-weighted.
Quantization Noise, CD Mode: Setting #2 of interpolation filter, —59.5 dB; all other settings, —96 dB or better.
Dynamic Range: CD mode, 96.7 dB unweighted, 99.4 dB A-weighted, and 89.6 dB CCIR-weighted; SACD mode, 103.5 dB unweighted, 107.3 dB A weighted,. and 98.4 d unweighted.
= = = =
Sony uses a generalized-impedance-converter (GIG) topology in the SCD- l’s Out put filter to avoid passing the analog signal through active circuitry prior to the output buffer. With GIG topology, the desired signal flows through passive components (resistors) while the undesired signals are shunted to ground through the active components. Thus, distortion in the active circuitry cannot affect the music. I buy the concept in principle, but although passing signals through resistors may not elicit distortion, it does increase thermal noise. (How much depends upon the value of the. resistors.) You pays your money, and you takes your choice.
When it comes to DSD, the output filter is by no means a trivial matter. A delicate balance exists among conflicting requirements: player bandwidth, flatness of frequency response, the amount of ultrasonic energy in the output, and filter group delay (phase distortion). Therefore, a designer must carefully select the filter’s characteristic (Butterworth, Bessel, Chebyshev, etc.), cutoff frequency, and slope. The idea be hind DSD is to avoid brick-wall filtering, analog or digital, but still dump ultrasonic noise.
Considering the attention devoted to the niceties of circuit topology, it goes without saying that Sony paid equal attention to circuit layout, grounding, choice of components, power supply design (using separate power trans formers for the analog and the servo and digital circuits), etc. This was clearly demonstrated on the test bench, where the Sony SCD-1 performed impressively.
Before I get into performance, a few words about this player’s features are in order. The SCD-1 has unbalanced and balanced analog outputs, via top-quality insulated gold-plated RCA jacks for the former, and XLR connectors for the latter. Dig ital output, available only for 16-bit PCM data from CDs or compatible hybrid layers, is through a gold-plated RCA coax jack and a Toslink optical interface. You can disable the balanced analog outputs via a rear-pan el slide switch and the digital outputs via a “Digital Out” pad on the front panel.
The SCD-1’s remote is stylish and intuitive; I liked it very much. The main difference between SACD and CD operation is that SACD supports 255 track numbers and doesn’t distinguish between track and index marks, so the remote’s “Index” keys function only with CDs. The remote can generate separate command codes for two Sony players, so it can independently control the SCD- 1 and a CD player or changer.
Because many more SACDs will probably carry embedded text than CDs traditionally have (a text mode was included in the CD standard but has never been widely supported), the SCD-1 has “Text” and “Language” buttons to display whatever info a disc includes. (SACDs can also carry limited video content, but the SCD- 1 does not support this.) Other than the “Text,” “Language,” and (maybe) the “Filter” pads, anyone familiar with using a CD player should have little difficulty operating the SCD-1 right out of the box.
“Filter” offers a choice of five coefficients for the VC24 digital interpolation filter used for 16-bit PCM (CD) data. They en able you to trade off resistance to high-frequency aliasing (intermodulation between high-frequency audio signals and the carrier that results in nonmusical in-band components) for a gentler rolloff slope. The gentle slope is thought to elicit less ringing on transients and therefore less “smearing” in the time domain. I have decried such filters in the past because I think the audible fold-down distortion they may cause is far worse than potential ringing at a frequency too high to be heard. Both my test instruments and my ears tell me that the difference people hear with gentle filters is, for the most part, simply distortion.
Although I remain unconvinced, I will say that some of Sony’s filter choices do less to make a travesty of the music than others I’ve heard. Setting #3 is particularly interesting. It is said to be a new characteristic. Using eight-times oversampling and combining original data with interpolated data to create a 224th-order filter, it is claimed to be flat to 17 kHz and to have 80-dB rejection above 26 kHz. This limits fold-down distortion to the extreme treble. With sloppy filters so much in vogue these days, I can’t fault Sony for including options on the SCD-1 as long as its “Standard” filter setting is really good—and it is. As good as it gets, in fact!
Since interpolation filters aren’t used with DSD (one of its main claims to fame), “Filter” has no effect when the SGD-1 is playing SACDs or the DSD layers of hybrid discs. However, as mentioned earlier, Sony offers a choice of characteristics for the low-pass output filter, which removes ultra sonic noise in DSD playback. You can change the filter setting with a rear slide switch that’s locked to “Standard” position by a removable plate. Its other position,
"Custom," extends frequency response when the SCD-1 is playing a DSD recording but leaves substantially more ultrasonic energy (shaped quantization noise) in the output. Because this could cause problems if your system's other components are not designed to handle it, the "Custom" setting is not recommended unless you're using wideband electronics and speakers.
When I moved the SCD- 1 into my listening room, the first thing I wanted to do was compare DSD and CD sound. However, that's nigh unto impossible with this player.
Although four of the five Super Audio discs Sony sent me were hybrids, I could not make AJB comparisons between any disc's DSD and CD layers. The SCD-1 took so long to stop, switch modes, and get back to the track I was listening to that I could read the disc's program notes in the interim, There were times when I actually got up and walked over to the player to be certain it had received the command from the re mote, only to be greeted by a blast of music as it finally finished setup and got on with things. I did convince myself that I heard a difference, but I'm well aware that expectation has a profound effect on conclusion.
The SCD-1 is such a bloody good CD player, and these discs (on the whole) are so well recorded, that it was hard to be sure, To put the interpolation bit to bed first, let me say that I always preferred the "Standard" setting to the others-oddly enough, for the very reason many criticize it: It had the best transient response! If you put distortion aside for the moment (a dangerous thing to do), some might claim filter setting #1 has more depth; I'd call it distance. Setting #2 made the performers sound closer but duller. Setting #3 could be worth toying with, but I still preferred
"Standard" and could not be dissuaded under double-blind conditions. Setting #4 focused the center-even when it shouldn't have. A pox on them all! Now, what about DSD? If I were to de scribe the difference I heard with DSD recordings, it would not be about what was added but what was missing. The veil-that thin gauzy tissue between listener and per former, so translucent you can neither identify it nor even be sure it's present until it no longer is-fell away when I played the DSD layer. I could tell you that I've never heard more realistic brass, hi-hat, and brushed cymbal than what Tom Jung captured on Tango (Stockholm Jazz Orchestra, dmp CD-05252), one of the SACDs Sony supplied for this review. I could also tell you that Michael Bishop captured pianist Julie Spangler's playing of "Ragtime" to perfection on the SACD version of On Broadway (Erich Kunzel, Cincinnati Pops Orchestra, Telarc CD-80498-SA). This disc sounded distant and insipid at first but grew on me as I discovered its often uncanny realism.
You may agree with me that its balance is often rather weird, but I couldn't help but like the sound of the English horn and the deeper woodwinds.
I don't think you'll have the same corn plaints with Jack Renner's recording of Kunzel and the Cincinnati Pops Big Band on Super Audio CD Sampler (Telarc SACDP 99-1): same conductor, same venue, but terrific presence and great muted trumpets.
(This recording is due out on CD in January as That Nelson Riddle Sound, CD 80532.) And if the "Deep Purple" cut on Telarc's sampler is any indication, the CD of Dave Brubeck's The 40th Anniversary Tour of the U.K. (Telarc CD-83440) is a must have (and is available now). Clearly Renner miked this cut rather closely-the British applause is veddy polite and quite distant- but he managed to capture (and DSD man aged to preserve) the ambience of London's Royal Festival Hall with extraordinary real ism. Ditto for the recording of Oscar Peter son, Ray Brown, and Milt Jackson (on CD as Oscar, Ray, and Milt: Live at the Blue Note, Telarc CD-83443), although I found the drum set at the opening of "Caravan" a bit heavy.
If you're into choral music, the Telarc sampler previews a great CD, Orlando and the Flemish Masters (CD-8052 1), which Renner recorded in the Netherlands. I prefer Renner's sound to Bishop's on Dvorak's Stabat Mater (Robert Shaw, Atlanta Sym phony Orchestra and Chorus, available as Telarc CD-80506), though one could argue that Bishop's is the more natural. The Bishop recording should be out now; Renner's is slated for an April 2000 release.
Another sampler, DMP Does DSD, contains an outstanding choral recording des tined to be a classic. It comes from my friend and almost neighbor, Tom Jung.
Tom's better known for his studio jazz recordings than big-scale choral stuff, but the tracks from the CD Sacred Feast (dmp CD-805), by a group called Gaudeamus, prove that his talents with choral music match those he's demonstrated with jazz.
Actually, every recording on the dmp sampler is a winner-tracks from Manfredo Fest's Just Jobim (dmp SACD-05) and Alto by Joe Beck and Mi Ryerson (dmp SACD
06), as well as the two mentioned above- but the ones that show DSD's potential to its fullest may be the tracks from Quality of Silence, by the Steve Davis Project (dmp SACD-04). I defy you to hear anything be tween you and the performers on this one.
That is what SACD is all about! But will SACD succeed? Of the major labels, only Sony Music has agreed to support SACD's launch; it intends to release 10 titles per month. The other majors are sitting on their you-know-whats, petrified about piracy. (Hey, guys, I dare say it's harder to pirate SACD than DVD-Audio.) That leaves the little guys-the Telarcs, the dmps, the AudioQuests labels for whom sound quality is of paramount importance and who don't care about getting ripped off or are willing to risk it to some degree in pursuit of that goal.
No question about it, you take a risk buying the Sony SCD-1. If SACD succeeds, you'll be limited to stereo while the format progresses to surround sound. But at least the SCD-1 won't become an albatross like the Elcaset: If SACD fails, you'll still have a world-class CD player that will be the envy of everyone who sees and hears it. The Sony SCD-1 strikes me as the Lamborghini of the audio world-of questionable practicality but absolutely incredible. And, yes, I'd like one! Who wouldn't?
Related: DSD and SACD -- Evolutionary or Revolutionary?
(Source: Audio magazine, Sept. 1999)