Rotel RCD-971 CD Player.

Manufacturer: Rotel of America, 54 Concord St., North Reading, MA 01864; 800/370-3741

Price: $699

Source: Manufacturer loan

Reviewer: Gregory Koster

New technologies offer great potential, but the performance of the early products tends to vary in unpredictable ways. Maturing technologies have identified the basic issues and a few very expensive masterpiece designs offer breakthrough performance. Mature technologies allow that breakthrough performance to filter down to the affordable level.

CD is a mature technology and the Rotel RCD-971 is that breakthrough affordable CD player.

The Rotel RCD-971's virtues start right at the faceplate with a handsome design, functional layout and solid feel. I miss the full number pad on the faceplate, which went out of style a few years ago, but at least Rotel has slimmed down the cabinet (only 2 7/8 in. high) to the room needed for the display and controls. The disc drawer is a typical plastic job, but thankfully its front edge is flush to the main assembly so there's little risk of scratching a disc. The display is white fluorescent with crisp letters (my favorite), and it's more intuitive than most (for example, it says "OPEN" when the drawer is open and "NO disc" when it's empty, as opposed to the Marantz CD-67SE which says "dISC" [sic] in both conditions). Controls are what you'd expect these days, with the exception of a red "HDCD" indicator that signals the presence of this bonus feature.

The back plane is a model of simplicity: hardwired power cord, coax digital out, and a pair of analog output jacks. All the jacks are gold plated, but this is neither mid-fi (no variable outputs with crappy volume controls) nor tweak heaven (no multiple digital outs or detachable power cord)--just good solid low-end high end, my favorite stomping ground. Rotel suggests that you use "high quality interconnect cables" and is nice enough to supply some (nothing esoteric, but heavy-gauge cable labeled "OFC").

The remote control is a little disappointing: 26 tiny identical buttons in a tight 3 x 9 grid (don't impugn my math--there's one empty space). Minimal color coding makes it usable in low light, but this is not an eyes-closed operation. Big brother RCD-991 has a much nicer remote, which I would have preferred, but this one is quite small, so maybe others will consider it a good trade-off.

Under the hood the RCD-971 is even more impressive. Rotel's claims start with the power supply ("multi-segmented" with a large toroidal transformer with separate windings for the digital and analog stages). The digital and interstage data transfer circuits were designed to minimize jitter, and the analog stage was designed to maintain precise interchannel timing relationships. That's all great stuff, but I can't tell if it's smoke or even if it's there (although the fact that Rotel centers its claims on items like these, instead of bells and whistles, is reassuring). What I can evaluate is the digital chip spec: dual Burr-Brown PCM63 20-bit Digital to Analog Converters (DACs), 8x oversampling Digital Filter, and the Pacific Microsonics HDCD Decoder are the right stuff for superior sound as CD technology enters the 21st Century.

The other thing I can evaluate is the sound, but the RCD-971 is so good that I'm almost left speechless. On the other hand, since they pay me by the word, let's take a stab at it.

The question that's been dogging my CD player reviews is whether 1-bit DACs are inherently inferior to multi-bit units, or whether I've just been unlucky (or hearing things). The four 1-bit units I've auditioned since buying Sony's last multi-bit player have all sounded tipped-up in the treble. There were always other variables that precluded a conclusion that it was the 1-bit DACs, and with multi-bit technology increasingly limited to super-high-end manufacturers such as Krell and Linn I was afraid that I'd never get to say for sure. That's why the Rotel RCD-971 is such a perfect test for me: it measures exactly the same on pink noise as the Marantz CD-67SE, one of our top recommendations and a 1-bit player. So let's duke it out head to head.

I went through my usual listening suite, starting with the Dvorak. The Rotel had a solid bass fiddle foundation, a very pleasant sound through the cool-sounding NHT Super One/ SW2Pi subwoofer combination, good balance among the four string choirs, realistic (not overemphasized) violin rosin bite--overall very natural and very involving. The Marantz, by contrast, made the violins sound screechy and too prominent, and had very little bass--for a thin, tipped-up sound.

The other selections all generated comments similar to the above: on a wide variety of musical styles, instrumental groups and recording techniques, the Rotel RCD-971 always sounded natural, balanced, dynamic and involving, while the Marantz always sounded tipped-up in the treble, deficient in the bass, and thin across the spread. The differences were subtle--far smaller than differences between speakers and even smaller than those between the Marantz and my old Sony--but they were audible and consistent. Only on the Reference Recordings Liszt did the Marantz produce a more interesting sound picture, with a piano whose high notes had overtones all the way out to Mars--but other RR recordings tell me that Prof. Johnson aims for the slightly rolled-off high end I heard on the Rotel.

So the clear and consistent result was that the Rotel RCD-971 produces exactly the kind of mature-technology sound that I'd hoped for: faithful to the original without indulging in the "ruthless exposition of faults" that makes so much high-end gear no fun with real music.

But wait, there's more! There's HDCD. I was afraid this would be the fly in the ointment, since HDCD sounds like another tweak technology. It requires the use of the proprietary Pacific Microsonics decoder chip, which might even ruin regular CD playback since it's in the circuit at all times--but no, the above results put that fear to rest. So it was with mixed emotions that I popped in Tutti!, the soundtrack from "Car 54 Where Are You?"--no, I mean the "Prof. Johnson 24-bit HDCD recording Orchestral Sampler (RR-906CD). Track 1 is Rimsky-Korsakov's Dance of the Tumblers, played by Eiji Oue and the Minnesota Orchestra. WOW! This is the most lifelike orchestral recording I've ever heard. Through the Rotel, that is. On the Marantz, without HDCD decoding, everything was veiled and less dynamic. Track 10, Malcolm Arnold's The Padstow Lifeboat by Jerry Junkin and the Dallas Wind Symphony, produced the same reaction.

Even the old HDCD Sampler CD (RR-S3CD), which I've never been able to use before, showed more immediacy with the Rotel's HDCD decoding than the Marantz's plain vanilla. A jazz cut on that CD even produced some detailed differences as HDCD and the players' basic personalities combined: the drummer's brush work sounded "real" on the Rotel but was unnaturally smeared on the Marantz, while the tenor sax was "in the room" on the Rotel and clearly a recording on the Marantz.

A final test ran both players (and the old Sony) through the CD-Check test CD (www.digital-recordings.com). The Sony showed its age, passing only track 1 (the "minimum SR/EC criteria" below which you need a service call). The Marantz passed track 1, failed track 2 ("average") and almost passed track 3 (although the instructions say that almost doesn't count). The Rotel sailed through tracks 1 through 3 ("good").

Conclusion: The Rotel RCD-971 is the best CD player I've ever heard. It plays regular CDs flawlessly and with great musicality, and HDCD adds immediacy and life that I didn't think was possible. Best of all, the price for this top-drawer performance is no longer in the stratosphere but down at the high end of the midrange. I'm sold, and I think that one listen is all it will take for nybody. Highly recommended. --GK

To supplement our coverage of CD players and DACs, we have invited David A. Rich, PhD, Technical Editor of The Audio Critic, to take a look at the units reviewed above. We requested schematics from all the manufacturers; some sent them, others did not. We hope you will enjoy David's findings and observations. --KWN

Below I present circuit analyses of some of the CD players and DACs under review in this issue. The Assemblage and Marantz units were evaluated only from the schematics. I would very much have liked to have seen the Rotel schematic, or the Rotel player, but neither was made available in time for this issue. (I have reviewed many Rotel units in the past for The Audio Critic. I have been favorably impressed with most Rotel components and have never found a badly designed one.)

In the design analyses below I make no attempt to review basic terms and techniques. This was all done in great detail in Issue 15 of The Audio Critic. (Back issues of The Audio Critic are available at $24, total, for any four numbers, postpaid; or $7.50 plus postage for a single issue. The Audio Critic, P.O. Box 978, Quakertown, PA 18951-0978. VISA/MasterCard: 215/538-9555, Fax: [phone].)

I must point out that because The Sensible Sound does not own high-performance test equipment, I am missing critical information on how well these units perform. Only through a careful verification of device specifications can one be sure that a unit is truly recommendable. Virtual (PC-based) instrument prices are dropping, however, thus it may thus be possible to afford a first-rate laboratory in the future. In evaluating CD players, an important test for transports is bit error rates and the ability of the transport to correct (not conceal) errors caused by CD defects. The measurement techniques were reported in previous issues of T$S and the results for the some of the units under test here are presented in the subjective reviews.

Because the audio world is at the threshold of transition to DVD-A, it really makes little sense to spend lots of money on a CD player or digital converter box without a clear upgrade path (we will define what we mean by that below). Moreover, when it comes to playing CDs, all-in-one CD players (as opposed to separate transports and DACs) are clearly the way to go, because in a one-box CD player, the problem of reconstructing the clock from the SPDIF signal does not exist. Jitter attenuation boxes are a waste of money. In general most of the clock jitter occurs with a bad SPDIF receiver, which will not benefit significantly if the SPDIF signal is retimed after the transport. A good SPDIF receiver will have excellent jitter rejection characteristics, making the issue of using a re-timing engine moot. Jitter attenuators which also separate the clock and data into a I2S data stream are more effective, but only a very limited number of DAC boxes take I2S input, and for best performance we would not like to see the oscillator clocking the converters so far away from the converters. An excellent design requires the VCXO (for a multiple-PLL-based SPDIF receiver) or crystal oscillator (for a sample-rate-converter-based system) to be close to the latches that drive the DACs.

Again, the best approach is to forget the whole problem by getting a single-box CD player. If you must go the external DAC route, be sure to get one with a good SPDIF receiver, or, even better, slave the CD transport from a clock generated in the converter box. However, this approach requires a CD transport that will lock to an external clock, and this is extremely rare.

If for some reason you simply must have something new today (for example, your old CD player just went up in smoke!), my advice would be to go for a good low-cost CD player to use while you wait for DVD-A. If you are into taping, now is the time to move up to CD-R (I will soon have lots of reel-to-reel decks for sale on eBay once I transcribe everything to CD-R): you get a CD player thrown in for free.

Another sensible approach is to find a CD player with a high-end pedigree but not a high-end price. Well, I can think of nothing better than the first unit whose schematics I looked at:

Marantz CD6000OSE: OK, let's talk about the good news first. This thing has an all-discrete analog stage. That is high-end in abundance! At this price we expect to find low-end opamps. In fact, even at 10 times this price, I have seen units sporting low-end little opamps that also are found in the Best Buys $100 special. Each discrete opamp in the Marantz is a single voltage gain stage folded-cascode design for wide bandwidth and fast settling time. FETs are used in the differential amplifier with, get this, a real current source. (I have seen resistors replacing this critical stage in systems that are only affordable if you mortgage the house.) The FETs are a good choice, for they have a wide open-loop linear range, which helps settling time.

The output stage employs depletion-mode MOSFET source followers with the gates tied together and a small amount of source degeneration. This is not the best way to ensure a constant bias current, given the lot-to-lot variation of MOS devices. To be generous, I will assume that some selection of devices occurs to keep the bias current somewhat stable between production units. A true active bias network would have been much better--but also more costly. Even in this simplified form, however, this output stage clearly has more drive current than almost all opamps. The output sees only one capacitor, a back-to-back 220ufd electrolytic. It's high end for less all the way in the analog stage of this unit.

The DACs are used two per channel in antiphase and each has balanced outputs. The first discrete stage does a balanced to single-ended conversion and some filtering. The next stage then combines the signals from the two DACs to cancel correlated distortion. All analog lives off a single +/- 12V supply but that is really generous at this price. The DACs get their own 5V supply.

At this point, all we would need to say is that this unit has one of the latest Philips DACs and we would be home free. But we cannot say that, because the circuit has instead an el cheapo NPC SM5872 combo DAC/filter chip. With an 87-tap filter, in-band ripple is an awful (for a digital filter) +/- 0.05dB and the stop-band attenuation is a low 53dB. The delta-sigma DAC is a 4th-order oversampling unit with 13-level output. The noise shaper runs at 32fs. That is four times slower than the best modern parts. The result is sub-par performance. Worst-case dynamic range is at the 15-bit level; THD is at a little less than the 15-bit level. Ughhhh! Both numbers should improve by 0.5 bit in the dual-DAC mode used by Marantz, but even then, these performance numbers are still well below the best.

What happened to the Philips TDA1307/ TDA1547 combo, which has true 16-bit performance? Philips is the parent of Marantz, after all, and the rest of the unit is filled with Philips parts (the transport and its electronics are Philips and very nice for the price, although one learns more in this case from measured error correction performance than from the schematic). I do not know the answer to that question, but I note that no longer are high-performance Philips DACs listed on the Philips semiconductor web site, which may indicate that the Philips consumer electronics group cannot use what Philips semiconductor no longer makes. If that is the case, we could still suggest lots of better units than the NPC unit they picked

Five hundred dollars appears to be a fair price for this. The analog stage is really something that is only seen in high-end units; the DAC is from the $200 CD player part bin, although it is used balanced here to get a little extra performance.

Assemblage 2.6: Nothing is wrong with this design but nothing is compelling about it. The SPDIF chip is a Crystal CS8414. Although this chip does its job acceptably, for excellent performance it needs an external circuit to reduce clock jitter. For example, a PLL based on a VCXO can be used after this chip, or a sample-rate conversion chip using a crystal reference could be used to follow the chip. Both are absent here. Assemblage makes an external jitter attenuation box, which has both options, and the literature on the web is very frank about the performance of the 8414. It is thus surprising they did not do a better job in this unit.

The digital filter is the HDCD-compatible PMD-100 from Pacific Microsonics. It gives decent performance (at least judging by the limited specifications I can get off the web) with excellent stop-band rejection, but its passband ripple is high relative to the best filters from Sony and NPC (but a lot lower than the low-end filters from NPC). I have no information on the PMD-100's multiplier structure, filter coefficient width, or methods of internal truncation and rounding. The NPC5843AP, for example, is a high-end digital filter that has excellent performance characteristics but lacks HDCD capability. I would rather see it used here, because its design is fully documented; it has been used in many high-end Japanese CD players with excellent results.

The DAC is a Burr-Brown PCM 1704. This is a unique device in that it does not use oversampling techniques. Oversampling approaches have lots and lots of downsides (see TAC issue 15) and only one upside--the chips are cheaper to make (the exception being the latest Analog Devices multi-bit delta-sigma, which gives the Burr-Brown multi-bit units a run for the money for the first time). The PCM 1704 uses weighted current sources in a proprietary configuration and generates no out of band noise (exclusive of thermal noise) other than what comes out of a theoretical sampled data system. 24-bit resolution is claimed for the PCM 1704, but these are marketing bits. In reality, it is one of the best performers on the market in its highest grade (like eggs, some DACs come in grades. The better their distortion and noise performance, the better grade they get) with true 18bit performance in K grade. You lose a bit in the grade supplied with this unit. The use of the lowest-grade version of the part is a surprise given the unit's price.

Unlike oversampling DACs that use noise shaping, no high-energy out-of-band noise comes from this converter's output. This makes anti-alias filtering much easier. With delta-sigma modulators, the downstream active stages can often be four levels deep (although the Marantz gets away with two, above) to get rid of all the noise. The active circuits also have to perform well in the presence of all that high-frequency noise. In the Assemblage we have just two active stages: a) a current-to-voltage converter, which is required because the PCM 1704 is a current-mode device, and b) a filter and buffer stage.

The current-to-voltage converter is the adequate OPA604, but one can do much better here with parts such as the Burr-Brown OPA627. This part offers lower settling time and lower distortion owing to a more advanced process technology. In the PCM 1704 demo board designed directly by Burr-Brown, they select the OPA627.

The output stage is a third-order filter made with a unity gain buffer as the active element. The Burr-Brown OPA134 is used here. This has good low impedance driving characteristics but I would like to see a more up-market device such as the Analog Devices AD797 or another OPA627 at this price point

This unit is DC-coupled throughout. A DC servo is around the current to voltage converter. They use an OPA604 here for reasons that are unclear to me. Typically, a low post ultra low offset opamp such as the OP27 is used to ensure minimum offset voltage.

The power supply is really nice, incorporating 1178XX/79XX regulators. Each channel runs on its own regulated rail at +/- 15V. The DACs get subregulated down to 5V. The digital supply is double-regulated, but the SPDIF shares a regulator with the digital filter.

Overall, it is hard to recommend this unit. The price is high for what you get. If you go for an outboard DAC you want a SPDIF with jitter reduction, top-grade DACs, and the best opamps or discrete circuit designs. If you will settle for less, get an all-in-one CD player. I note that The Parts Connection has a series of upgrade paths for this unit (by the way, the "signature upgrade" does not address the real problems in this unit, which are that the SPDIF decoder is not very good and the DAC is not K-grade). The upgrade path you really want is to the next unit in the product line.

Assemblage DAC-3: OK, so you must unload some money into audio equipment today, and you have your heart set on an outboard DAC. What do you do? What you do is purchase this thing.

I am first going to describe what is shown on the schematics supplied to me and then discuss upgrade paths.

The DAC-3 has four boards. The first is the power supply board, which has three--count them!--three transformers, one for each DAC/ reconstruction filter board (two for stereo) and one for the digital input/SPDIF decoder/digital filter board. Seven IC regulators are on the power supply board, and +/- 15V is generated for each analog board. From the third transformer, we get two regulated 12V supplies and one 5V supply.

On the analog board we start with an ISO 150 capacitive isolation device. What this thing does is to separate completely the DAC from the digital filter, including the grounds. Remember all those transformers? Fancy, fancy!

Next come not one but two PCM1704s. There are two because this thing is fully balanced out to the DAC. They run antiphase, with one getting an inverted digital input. If you want the K grade (and you WANT the K grade--, repeat after me, "we NEED the K grade"), it is an option on this unit. The DACs get powered by +/- 5V subregulators that are in place on the DAC board. Next comes the I/V converter (all the rest of the circuit is one per DAC). This time it is an LT1357 with the same DC servo used in the 2.6, but you can (and you WILL--this is an order!) upgrade to an OPA627. The I/V converter has half the trans-resistance of the 2.6, reducing distortion at this point. We get the gain back in the second active stage. A third-order passive RL double-terminated two-port filter follows to ensure that no out-of-band images make it out the back of this unit (the active filter topology used in the 2.6 will not work in a true balanced configuration). This filter is connected to the output stage, which is a composite of an OPA134, upgradable to a OPA627 (do it man, do it!), and a Burr-Brown BUF634 that can drive 600 ohms and not even feel it. This stage is needed even if resistive load is at the output because the feedback resistors are very small to keep thermal noise down--DO NOT REMOVE THE BUF634! Cross coupling at the inputs of the opamp for the stage provides common-mode rejection of distortion signals common to the two DACs and I/V converters. I would have liked to see a more clever circuit that would have eliminated the summing junction of the opamps from seeing the signal swing. The output of the pair of opamps that forms the balanced second stage comes together on the XLR output jack.

In my opinion, what we have just described is as good as it gets. I would not change a thing. Other ICs could be substituted, and discrete devices could replace the opamps, but in the end you will equal but not better this. Add in the fact that this is a four-layer board with complex ground and supply planes and nothing more needs to be said. Yes, it really needs to be measured to confirm that everything works as it should, but as we said going in, we have no Audio Precision or equivalent test instrument at The $ensible Sound. (At least The Parts Connections owns one, which is more than can be said of most micro companies, and the measurements they sent are goooooood!)

The final board is the digital front end, which is no fun at all in its present configuration, essentially a copy of the 2.6 except that it has more digital I/O, including I2S. It should come as no surprise that the company has a third box that takes SPDIF in and puts out jitter-reduced I2S. The Assemblage jitter-reduction box also upsamples, which is the new in thing in the digital box world now. Forget upsampling--the digital filter does the same thing for free. Indeed, forget this box and wait for the circuitry to come to the 3.0 digital board. I am told it is coming, along with the replacement of the HDCD digital filter chip with a Burr-Brown DSP 1704 digital filter, which will take 96kHz direct in, so the modified board can be connected to a DVD player that supports digital out at 96kHz (not all do). That lets you play the 10 or so DVD discs that support the DVD-Video (do not confuse this with DVD-A) 96kHz sampling rate.

More significant is the fact that the board can be replaced yet again when the DVD-A digital standard gets established, preventing this unit from becoming obsolete before any dust has a chance to collect on it. You cannot say that about any other DAC box. What distinguishes this unit is that the SPDIF decoder and digital filter are on a separate board that can be easily and cleanly upgraded with minimal cost. Most of the expensive stuff (i.e., the metal work, power supply, DACs, analog stages) does not need to be changed.

Bottom line? If you must have it, buy it. No other DAC box is a rational choice in this interim phase before the DVD-A spec is defined. No other DAC box known to me has a better converter/analog section. No other DAC box known to me has a more complex power supply (did I mention the 14--count theme--14 regulators in total distributed over the four boards?). So you can purchase the DAC-3 with no guilt--but do purchase it with the K-grade DACs and the OPA 627.

I will close with some brief comments about units for which I had neither schematics nor the opportunity to do any playing with:

AMC CD6b-MD5404as and CD6b Compact Disc Players: I have very little to say concerning these units. The literature makes some mention of a multi-level delta-sigma unit from Burr-Brown. It is not designed by Burr-Brown for the high end (for which they produce the PCM1704). Burr-Brown gives very little design information on this current unit, but AKM, Analog Devices, and Crystal all have similar parts, with their best delta-sigma parts having much better specs. The "Quad-DAC" configuration, to use AMC's term, is not novel. It is used in the Marantz, for example. Burr-Brown fights the good fight with their PCM 1704K multi-bit at the top of their line, and that is what should be in this unit if AMC wanted to use Burr-Brown parts.

Mach 1 D/A Converter: This is a modified Assemblage DAC-2.6. The original Assemblage design is OK at best (see my extended commentary on the 2.6 above), so upgrading the back end should accomplish little if anything in the way of solid improvement.

Rotel RCD-971 CD Player: This is likely the best of the CD players reviewed in this issue. The HDCD filter is good, better than the Marantz by a mile. The PCM63 DAC predates the PCM1704 (this must be an old design), but it is similar. I would guess they do not use the K grade at this price, but I really prefer these Burr-Brown units over the delta-sigmas used by everybody else (except for the latest high-end design from Analog devices, which is not in any product yet as of this writing). Moreover, I have never seen a Rotel analog section I did not like, and would not expect this unit to be any different. If the schematic ever does arrive, we will issue an update in a future issue.

--David A. Rich, Ph.D Technical Editor, The Audio Critic Guest Contributor