|
You are reading the older HTML site
Positive Feedback
ISSUE
25
Something Old, Something New: The Monarchy M24
DAC & Linestage
By now you may have noticed the high-resolution digital movement has lost momentum, and older media—LPs and standard Red Book CDs—are making a comeback. And contrary to what we may have thought a few years ago, the playback of the older media continues to improve. This is most noticeable with Red Book CD playback, where the best DVD-A and SACD players have only just held even with the best CD-only players. This is more than a little counter-intuitive, and points to some unsolved problems in the high-resolution formats. I suspect jitter is getting in places it doesn't belong, and there may be theoretical issues with upsampling, noise-shaping algorithms, and delta-sigma D/A converters. It may seem a bit retrograde, but I now feel that high-res and Red Book really deserve their own players, each optimized for their own media. This applies to the transport as well; against expectations, DVD players have proven to be some of highest-jitter transports out there, The most likely explanation is the transports are optimized for high data transfer rates in and out of computers (48x or higher), as well as retrieval of the MPEG2 video datastream. The demands for leisurely low-jitter 1X playback of Red Book are very different than extracting a MPEG2 datastream or the latest copy of World of Warcraft on to your PC or Mac. Jitter has little effect on heavily compressed video, and no effect at all on computer data getting loaded to a hard drive. Even though the majority of high- resolution digital players are one-box solutions, they all must use DVD drives (SACD is an audio-only system using DVD technology as the disc carrier—CD lasers can't even see the SACD layer, which is entirely separate from the CD layer of a hybrid disc.) Not only are the requirements for CD and DVD/SACD transports different, but I'm beginning to suspect the requirements for the DAC are different as well. By now, we've had several years of upsampling 96/24 and 192/24 players, and I'm not convinced it's an improvement. More spacious, yes, sometimes more open too, but something about the tonality has been lost. The sound has lost some of its drama and vividness—and when you read about how upsampling and delta-sigma DACs work (http://www.iet.ntnu.no/~ivarlo/files/School/PhD/Report_audiodac.pdf), there's an awful lot of fast footwork going on behind the scenes. (This report is highly recommended for those who'd like to better understand the design principles of upsampling, noise shaping, and delta-sigma DACs.) When a digital engineer at the Ph.D. level describes the underlying mathematics of upsampling, noise shaping, and delta-sigma converters as "non-trivial", he's using a term of art to describe mathematical problems that are not just difficult, but potentially insoluble. Looking carefully at this report, there are some very odd things happening to the low-level spectra in these converters—in essence, the problems of the older ladder converters have been exchanged for newer, harder-to-measure, and subtler problems. It is more than a little surprising that high-resolution digital media—both DVD-A and SACD—have genuine problems that remain to be solved, even at the level of theory. As of the summer of 2006, they are still a work in progress. Since we all have substantial Red Book CD libraries—ranging from hundreds to thousands of discs—we shouldn't just walk away from the best possible Red Book playback, in the hope that exotic signal processing can make these sound like DVD-A or SACD. I think the signal processing alters the sound all right, but something is added, and something is taken away as well. Your ears are not fooling you—with modern players, your CD collection does sound different, a bit more like the high-res formats, but some of the excitement is gone. Some of the reluctance to listen to Red Book could be an appreciation of the merits of high-res, but it could also be that the new players don't play Red Book as well, either. It was a growing disenchantment with the sound of high-res digital that led me to look for Red Book players with an older architecture. There's been a major flat-earth movement to return not just to the previous generation of ladder converters, but to no oversampling as well—the converter runs at the native Red Book rate and resolution, 44.1/16. These are the DACs based on the Philips TDA 154X chipset, using resistive I/V conversion, and typically a tube, transformer-based, or minimalist discrete transistor analog section. The best-known examples are the 47 Labs DAC, Audio Note, Wavelength Audio, and various European kit projects. (There are excellent discussions of design techniques required for straight-ahead "classical" CD playback at: http://www.sakurasystems.com/articles/pitracermemo.html and http://www.sakurasystems.com/articles/Kusunoki.html.) I fully understand the appeal—this approach gives an "analog" balance and vivid tone colors, thanks mostly to the absence of digital signal processing (reducing jitter at the DAC chip)—but I'm not a fan of the antique TDA 154X architecture. Yes, there are 16 bits in the chip—which matches what's on the disc—but the unfortunate fact is that ladder converters don't actually deliver 16 bits of accuracy, due to errors in the values of the resistors on the chip. Without laser trimming, an ordinary group of resistors on a chip only has 0.1% accuracy, which is equivalent to 10-bits. In practice, a standard TDA 154X is about 12 to 13 bits accurate—the higher-grade "Crown" version buys more accuracy and a closer approach to the 16-bit ideal. On a subjective level, I've never liked the sound of the TDA 154X—I'm sorry, it's just too crude and old-fashioned for my tastes. Red Book is low-resolution enough without the converter eating up another 2 or 3 bits. That may not sound like much, but remember, 16-bits is only 65,536 possible levels—with nothing in-between—and 13 bits are a shockingly crude 8192 possible levels. Seriously, you really don't want to go much lower than 16 bits—it's pretty much the lower bound of high-fidelity. The flat-earthers make a big deal about the virtues of not oversampling, but I don't buy it. As long as you stick to multiples of the original 44.1 kHz rate (very important!) and avoid introducing jitter (also important), no additional distortion or spectral information is created—and you can run the converter faster, which is a good thing, not bad, since it can linearize the converter and get the last few bits out of it. It was Karna who reminded me a few days ago how great our system sounded in the mid-Nineties. Her word was the "magic" way sound came into the room when we listened to CDs—and she was right, it had been a long time since I had heard anything like that. Why? The Amity amplifier has a similar sonic character to the Karna amplifier I use now, I use the same Ariel speakers, hmm, what's different? Ah, it was the Monarchy M-33 DAC/Linestage, which uses the Burr-Brown PCM-63K converter and extremely fast (2500V/uSec) analog sections. Well, my original unit was long gone, but a quick Google search showed that Monarchy is still around! This is a very low-profile company—I think it's just Mr. CC Poon and his wife—but the product line looked very similar to what I remember, complete with the M-33 DAC/Linestage. Just what the doctor ordered. Since Mr. Poon is such a nice guy—I remember that from my original conversations many years ago—I called him on the phone to check that the M-33 still has the same goodies in it that it had ten years ago. Yes, he said, it's essentially the same, but why not try the new M24 DAC/Linestage instead? Well, honestly, I was a little reluctant. The M-33 is one of the best Red Book DACs ever made, and my tweaks (replacing most of the analog electronics) took it even further. Interestingly, the digital side of the M24 was much the same, especially the part that counted the most: the Burr-Brown PCM-63K laser-trimmed true 20-bit converter, which I feel is probably one of the best converters ever made—maybe the best, although the laser-trimmed 24-bit PCM-1704 has its partisans too. The S/PDIF receiver and following digital filter were improved and updated, allowing the M24 to be used for DVD-V/DAD 96/24 playback—a nice plus as long as your transport supports 96/24 without downsampling (the Pioneers pass through 96/24 on the S/PDIF connection). The analog side is a SRPP 6DJ8, frankly, not my first choice for circuit or tube. But one thing I had to remember as I talked to the ever-gracious Mr. Poon; this man has taste and good ears. He was smart enough to design the original M-33, which bears direct comparison with the Levinson DACs. He's also a straight arrow, has no axe to grind, and if he says the M24 is much better than the M33, I have to take him seriously. Designers usually know their own products pretty well. So I went ahead and ordered the M24, and it arrived in a very big and heavy box a few days later.
Even though this is a made-in-China unit, the build quality is first- rate. It also has a number of different parts and circuitry than the Chinese original—a Lite 50 DAC if my Google searching is accurate. The power transformers are different, there's a linestage that's not in the other Lite DACs, and a lot of the circuitry and overall build quality is different as well.
I think Mr. Poon has been to the factory in China and had a special run made for him; for lack of a better expression, the M24 has a much more "American" feel to it than the usual run of Chinese products. This extends to build quality, a number of parts inside I remember from the M-33 (which is still built in the Bay Area of California), and what must be a thorough QC in San Francisco. In short, it doesn't look or feel Chinese at all, but like a high-quality small-production- run specialist made-in-the-USA product. That kind of thing can only happen with close collaboration with the overseas factory. The sound? Well, I'm keeping it! Mr. Poon was right; it is much better than the M-33. The DAC is the best I've heard to date, and the kicker is, the linestage is really good too! I've auditioned the M24 in Thom Mackris' (http://www.galibierdesign.com/) system, with its $27,000 front end and $15,000 Azzolino speakers, and the linestage works beautifully with this exotic all-analog system. It also sounded so good on CD that the little Red Book monsters kept up quite well with Thom's analog front end—staggering considering this analog front end consisted of a Dynavector XV-1s, Tri-Planar & Schroeder Reference tone arms, the top-of-the-line Galibier turntable, and the Artemis Labs phono preamp. (Which by the way is the best record player I've heard to date.) This is especially surprising since an analog front end of this quality normally means that Red Book listening sessions are quite short—the quality difference is so appalling that Red Book just about always sounds like a car stereo in comparison. This is the first time I've ever heard Red Book in the same quality league as a world-class LP front end. The M24 also has an important attribute of the best Red Book players—it "rescues" the rough-sounding digital recordings made in the Eighties. This was an attribute I remember from the M-33—a quality of emotional engagement from the performer that came through—regardless of recording quality. Thom has some terrible-sounding Grateful Dead live recordings, made at the nadir of digital, with the most execrable mix-board work I've ever heard—overloaded transistor mike preamps, levels all wrong and changing all the time, outright wacko equalization, muffled and distorted singing—but you know what? As terrible as the sound quality was in an audiophile sense, the performance put you right there in the audience. This was the opposite of the oh-we're-so-refined hifi show experience; the sound itself was terrible, but you were there. This was called "presence" back in the 1950s—well, it still means the same thing today. Very few high-end systems have it—there's plenty of space, air, dimensionality, socko dynamics, subsonic bass, but not much sense of actually being there. When you hear that you-are-there quality in a high-end system, that's rare. Most high-end systems, especially what you hear at shows, can really only be used with audiophile-grade recordings, or they'll drive you out of the room with the harsh and shrill sound. This isn't "accuracy", by the way—it's a type of electronic distortion that emphasizes and makes more intolerable any defects in the original recording. Systems that are honest enough to let the original sound come through—without wisecracking and editorializing about it—are not common. How does the M24 sound with audiophile recordings? Well, really good, of course, all of the audiophile virtues in abundance. The most noticeable trait is the vividness of tone colors and dynamics that are nothing less than stunning. When you've been listening to the delta-sigma DACs for a while, the sheer push-you-back-in-the-chair impact of ladder-DACs can take you by surprise. There's also an emotional quality to the experience that has nothing to do with audiophile adjectives at all—if music moves you, this is the one to get. Yes, I'm talking to you, triode, electrostat, and horn fans. This is the DAC that will make you fall in love with your CD collection, possibly for the first time. The tube stage, both in the DAC and Linestage, to their credit, do NOT sound either like an SRPP or a 6DJ8. I don't know how CC Poon did it; on a direct bypass test, the Linestage had no coloration I could hear—and this was on a very high-resolution system playing LPs. Maybe an exotic NOS direct-heated-triode with extensive anti-vibration measures would have been a little bit better—maybe—but this was really good. The power supply, tube type, circuit, and output cap have been very skillfully balanced. Is there a weak point to the M24? Yes. It's the built-in volume control, which knocks down the quality of the linestage quite a bit. Thom and I were using the Dave Slagle autoformer volume control, which is in the same sonic league as the Bent Audio TX102 control, or other high-quality autoformer or transformer volume controls. If you just use the DAC portion of the M24, you'll never hear it. But if you listen to the linestage, which as mentioned above, is superb quality—the difference between the external autoformer volume control feeding the linestage, vs. the internal volume control—was the difference between a fairly decent $1000 tube linestage and a top-rank $10,000 tube linestage. Of course, what's a little unfair about this is autoformers or transformer volume controls usually cost as much as the entire M24! I quite seriously recommend using the M24 with an external top-quality control—and chassis mods are not necessary, since the necessary inputs and outputs are already there on the rear chassis. If you use the Linestage with the built-in control, it'll sound noticeably more closed-in, less dynamic, and less open than it really is, with just a bit of grain at the top. With an external AVC or TVC feeding the linestage, the linestage actually sounds better than the AVC or TVC by itself, and the sound quality leaps to the top of the tube-linestage league. The real giveaway is the grain—a good tubestage should never have any grain, at least with a competent power supply. Get rid of the control, the grain completely disappears, the soundstage width and depth doubles, the bass deepens, and dynamics really open up. It's not a small difference. I should add the M24 is not one of those worst-of-all-worlds players that "smoothes out" the harsh and metallic sound of op-amps with a following tube stage, giving neither the resolution of an all- solid-state player nor the smoothness of tubes. No. That's just wrong. By the time a low-quality op-amp has done its damage, there's no way to retrieve it. If a player is solid-state, the analog electronics have to be very fast—a minimum of 800V/uSec for Red Book playback—and also operate in Class A for all combinations of signals. These two requirements are NOT optional for high-quality reproduction from digital media. Throwing a bunch of Black Gate caps at a player won't unslew the op-amps, or remove Class AB distortion. The M24 sidesteps the slewing problem by using a passive I/V converter—which cannot slew—followed by a ferrite RF toroid combined with the Miller capacitance of the 6DJ8 SRPP circuit. This two-pole low pass filter quiets down most of the RF coming from the converter, and it doesn't hurt the 6DJ8 was originally designed as a preamp tube for FM and TV front ends. (Don't try to get RF through an op-amp, they're not designed for that.) The zero-feedback tube sections are always in Class A, of course, and have hundreds of volts of headroom. No wussy 15 volt op-amp supply rails here. The signal path, from the PCM-63K switched-resistor array to the output connector, is nothing more than circuit traces, triodes and capacitors. No op-amps or solid-state to slew or exhibit Class AB distortion. (By the way, the 25-year 5532 op-amp, which is no faster than 13V/uSec, is quite commonly seen in $5,000 audiophile SACD and DVD-A players, and left in place by the less competent modifiers. Yes, there are better op-amps, far better, but almost all operate in Class AB, which does no favors to 16-bit digital, which needs all the low-level resolution it can get.) Oh yes, turn-on thumps. CC Poon warned me not to get worried when I first turned on the M24. There is a remarkably long two-minute muting sequence, and the typical in-a-rush audiophile might think it's broken. It's not. Just wait, and sound will pour forth in a little while. What about the usual interminable wait for the DAC to sound at its best? With solid-state units, this is so long—many days—the best strategy is to simply leave them on all the time. With tube units, that's not always a good idea—you don't want to wear out the tubes for no good purpose. I can say that right out of the box, unlike many DACs, the M24 sounded very listenable—quite good, really. Out of the box, the sound is already in the $1000 league—the M24 starts making friends right away. The true high-end sound started about an hour in, and over the course of several days of burn-in, the unit opened up even more, reaching the elevated plane of the exotic players. But it didn't have the usual "bad attitude" of all-solid-state units, which sound nasty for several hours after they've been turned off for a day or more. In practice, it seems like 20 to 30 minutes is plenty of warm-up, very much like a good tube amp. Unlike a lot of high-end audio, the M24 is quite enjoyable during the warm-up interval, with beautiful tonal balance, and very inviting sound. You're not suffering any, I can tell you. What happens afterward is that it opens up, gets 3-dimensional, and starts to get that "magical" quality that Karna describes. What was her take on the M24? "Magical, even better than the old unit, you did right getting this one." She also said I'm not allowed to modify this one—leave it just as it is! CC Poon also made me promise to not even try any tube-rolling until I've listened to it for a month. I can tell you, though, with an external autoformer or transformer volume control, well, there's no need to tamper with it. It's that good. If the urge to improve just can't resisted, play with the transport and assorted jitter-reducing boxes instead—the M24 is so transparent the best CD-optimized transports are a very good idea. Additional Reading The modern professional-grade approaches to upsampling, asynchronous sample rate conversion (and its many problems), noise-shaping, and delta-sigma DACs are at: http://www.iet.ntnu.no/~ivarlo/files/School/PhD/Report_audiodac.pdf http://www.proaudioreview.com/october99/Weiss-web.shtml http://www.weiss.ch/pow-r/images/brochurePOW-R.PDF http://www.apogeedigital.com/products/uv22hr.php The completely opposite view (which I describe as flat-earth, taking the cue from the contrarian Linn/Naim school in the UK) is at: http://www.sakurasystems.com/articles/Kusunoki.html http://www.sakurasystems.com/articles/pitracermemo.html A once-over-lightly on the sound of transports—not just a matter of jitter specs, which leave out crucial information on whether the jitter is random (OK), correlated with mechanical noise in the player (not so good), or signal-correlated (worst of all, this is real distortion). Spectrum analysis of jitter shows you the spectrum, but not where it came from—the analyzer has no awareness of correlation to noise or music, but the listener does. http://www.hi-fiworld.co.uk/hfw/email1.html System I auditioned at Thom Mackris/Galibier Design (www.galibierdesign.com/):
As Thom's neighbor in Northern Colorado, I've spent a lot of time listening to this system as it's evolved over the last year, as well as getting acquainted with the other "Colorado Boyz" with their way-out-there audio systems. Are prewar Telefunken DHT driver tubes, Eimac 75TL transmitter tubes, and custom-designed Dave Slagle transformers extreme enough? It's not just skiing, mountain biking, and hanging off the side of mountains that are radical around here.
Monarchy Audio
To read Constantine Soo's interview with Mr. CC
Poon, see:
Lynn Olson
|
