Halcro Logic MC20 power amplifier
In that one stroke, Candy became the unluckiest designer on earth, in the same sense that the great San Francisco band Moby Grape ran out of luck when their record label, Columbia, released eight songs from their debut album as four different singles—on the same day. What could you possibly do for an encore?
In the case of Moby Grape, while the merits of their subsequent albums are open to debate (I rather like Wow), there's little question that their wad had in fact been shot. On the other hand, Bruce Candy focused his talents in a new direction, on a very different and altogether less expensive sort of product.
The result of his recent work, the Halcro Logic series of amplifiers, has at least one thing in common with the company's more expensive Reference series: Both lines are the intended embodiments of Candy's complete and utter loathing of distortion, and his desire to shrink and then kill it, Grover Norquist style. Stereophile's readers have already heard how he did that the first time around; today, some five years after the release of the groundbreaking dm58, Candy is making something very different and very new: a class-D amplifier.
You say distortion like it's a bad thing
Well, class-D isn't entirely new, having been toyed with as early as the 1950s. Nor does the D in class-D stand for digital: After classes A through C had been described and created, D was simply the next letter up for grabs (footnote 1).
Even so, and notwithstanding the fact that a class-D amp doesn't necessarily have a digitized throughput, it does behave in a digital manner: Its output devices, typically in complementary pairs, are used purely as switching devices, and so their output signal, if left unmodulated (I'll come back to that), is either a 1 or a 0—meaning the device is either fully on or fully off. Think of it this way: Driven by the raw output of a class-D amp, a loudspeaker driver would either come flying out of the enclosure toward the listener's nose, or struggle equally hard to escape its moorings in the other direction, with nothing in between.
Coaxing the driver into moving in and out is what you want the amplifier to do, of course, but preferably with more in the way of finesse—and, ideally, in mimicry of the continuum that is sound. So you have to put something in front of the class-D output section to control, with near-infinite speed and accuracy, the ratio of time that those output devices spend between being on and being off—which is to say, you have to modulate the output. Most designers of class-D amps do that with an analog technique that has a digital-sounding name: pulse-width modulation.
That's easy: Add a fixed, very-high-frequency carrier signal from a triangular wave oscillator to the input, alongside the audio signal, and make an active circuit—you could even use a tube, I suppose—to compare the two and to then send the difference between them to the switching output devices. Let a higher-amplitude portion of the audio signal equal a wider, positive-going pulse, a low-amplitude part of the wave equal a wider, negative-going pulse, and so forth. Now you've given the output devices an analog blueprint to follow, and their own output is amplified sound—or at least it will be, once you perform the final trick of filtering out that triangular wave's harmonics and a few other nasties.
It's an efficient amplifier, because those switching devices don't require any bias current. It's also a compact amplifier, for the same reason and because MOSFETs that are up to the task can be made very small. But is it a clean amplifier? I'm sure the boys at Stereo Review would have thought so, but let me remind you: Bruce Halcro Candy hates, and I mean really hates, distortion. And in addition to the switching noise that you'd imagine is part and parcel of a system such as this (think of a class-D amp as an audio-frequency switch-mode power supply), he's identified a distortion that, in his words, is mathematically intrinsic to a class-D amplifier with pulse-width modulation: The shape of the triangular carrier wave becomes distorted—is actually rendered asymmetrical—while the amplifier is in use, resulting in a nonlinear phase advance in the final output. Which ain't good.
So Candy, in his quest to cancel out every last iota of distortion, developed a circuit that actively modulates the carrier wave and corrects its symmetry during amplification. The Halcro company has appended the trade name Lyrus to this distortion-canceling circuitry, but you could also call the technique carrier-symmetry modulation: That, in fact, is what Bruce Candy called it in a paper he delivered to the AES convention in 2004, a copy of which I've been trying, with only limited success, to fathom.
MC, phone home
The Logic MC20 ($4990) is Halcro's entry-level Lyrus amp; in fact, it's Halcro's least expensive amplifier, period. In addition to the unique class-D circuitry described above, the 400Wpc MC20 offers a choice between unbalanced and true balanced inputs, an apparently sophisticated switch-mode power supply, and current-sensing and thermal-protection circuits that keep the MC20 from going postal on itself or the user's loudspeakers.
In the unlikely event that a glitch does occur, the MC20 and its owner can fall back on Plan B, which is called HRAS—for Halcro Reliability Assurance Service. Every Halcro Logic component has RS-232 and Ethernet sockets on its rear panel and is supplied with a CD of HRAS software, complete with high-level user interface, ready to be installed on the Windows XP-based PC of choice. With the amp connected to the computer, the HRAS software loaded, and the computer logged on to the Internet, a malfunctioning MC20 can literally call the nearest Halcro distributor, report the fault, and have a new board or perhaps a whole new amplifier delivered by air freight before the owner knows there's anything wrong. As a parent, I can't help being reminded of the time my daughter had a tiff with a playmate at the other child's house, called me using their phone, and cried, "Daddy, come get me now!" (footnote 2) (And as someone who has recently had to jump through flaming hoops of bureaucracy to get his Sony SACD player through the front door of a factory service facility, I can't help being very impressed.)
Overall, in a technical sense, I'm hard-pressed to think of anything that differs more from my low-output, tube-powered, single-ended, class-A monoblock amplifiers (the Lamm ML2.1s) than this high-output, solid-state, push-pull, class-D stereo amplifier. More to the point, Halcro's US high-end representative, the thoroughly genial Philip O'Hanlon, was well aware of that fact when he suggested that I give the Halcro a spin in my system and write this review. Talk about courage.
Footnote 1: It's a good thing audio engineers don't use the Etruscan alphabet for naming their operating classes; otherwise, the fourth class of operation would have to be called Class F, which would elicit much snickering from the back of the room.
Footnote 2: But this is a slippery slope: Could an HRAS-enabled amp harvest the user's credit-card number from an unsecure page and buy a new preamp for itself? Could it notify the federal government if the user plays politically unacceptable music (I'm thinking Phil Ochs here), or call the local drug squad if it senses that "Interstellar Overdrive" has been repeated too many times in a single listening session? The implications are staggering.