Digital Processor Reviews

The PCM64 has a fairly high level of glitch in its output. DAC glitch is a spike of energy in the output signal when the converter changes state—as it does at every sample. To remove the PCM64's glitch, Krell designed a de-glitch circuit based on an AD841 op-amp. This is followed by another Analog Devices part, the AD846 op-amp used as a current-to-voltage (I/V) converter. The voltage output is then buffered by a PMI SSM2131 device. The de-emphasis circuit is built around the PMI output buffer. All this circuitry resides inside the aluminum DAC housing.

The Analog Domain
The analog signal from the DAC module is input to a discrete, direct-coupled (with a DC servo) output stage. The signal is first converted from single-ended to balanced with a phase splitter, with each polarity amplified independently. The single-ended output on the RCA jacks is taken from pin 2 (the non-inverting, or "hot" pin) of the XLR jack.

Note that the signal is balanced in the analog domain after the DAC, not in the digital domain before the DAC (as is done in Theta, Meitner, Mark Levinson, Sonic Frontiers, Kinergetics, and some Meridian processors). Converting a digital signal to balanced before the DAC requires four DACs, four I/V converters, and four analog output stages (for two channels), but has advantages over analog-domain balancing. First, digital-domain balancing doesn't require a phase splitter in the analog signal path. Second, any DAC artifacts common to both channels will cancel with true digital-domain balancing.

Frankly, I was disappointed that the Reference 64 wasn't balanced in the digital domain, considering its $14,000 price tag. This is, however, an expensive feature, and one that may not have been possible considering the cost of implementing a custom 64x-oversampling digital filter. One could also argue that a $14,000 processor shouldn't use an off-the-shelf NPC filter chip.

Finally, I must comment on the Reference 64's extraordinary build quality. Its gorgeous metalwork, rock-solid build, beefy power supply, and bulletproof appearance contribute to the impression of an all-out effort. Moreover, the inside, with its beautifully machined and finished DAC housings and input receiver block, looks just as finished as the outside.

Music
After sufficient warmup, my impressions of the Reference 64 were largely favorable. The 64's overall perspective was midway between the Mark Levinson No.30's laid-back ease and the Sonic Frontiers SFD-2's forwardness. On some music, the No.30's lack of incisiveness was a liability; on other music, the SFD-2's overly immediate perspective detracted from the musical experience. I found the Reference 64's incisive, yet not aggressive, portrayal a good match for a wide range of music. The Reference 64 sounded immediate, detailed, and "tight" without going over the top and sounding etched or analytical.

On the plus side, the Reference 64 had some spatial qualities I haven't heard from digital before. The processor excelled at presenting a cohesive, focused, and clearly delineated soundstage. The ability to hear exactly where an instrument was located in space was uncanny. Instruments that sounded somewhat amorphous through other processors—the acoustic bass on "Moonshine #2" from the first Robert Lucas CD (AudioQuest AQ-CD1001)—were perfectly focused and tangible through the Reference 64. Moreover, there was terrific front-to-rear depth, with layers of distance separated by space. The depth had many gradations of distance, rather than just a few layers. Soundstage transparency was similarly impressive, with a hear-through quality such that I could hear all the way back into the hall's farthest recesses. The Reference 64 maintained its superb soundstaging when the music got loud and complex. The soundstage stayed focused, deep, and transparent even when the processor was pushed with complex, full-scale signals.

The Reference 64 also kept individual instrumental threads separate from the whole. The music was never thick, homogenized, or blurred. Just as the soundstage stayed coherent and focused when the music got loud, the Reference 64 maintained its ability to differentiate individual images from the whole at high signal levels. During the exceedingly complex passages of the Liszt Piano Sonata in B-minor played by Robert Silverman (which just reached digital full-scale, 0dBFS, on the peaks), the Reference 64 kept the left- and right-hand lines perfectly delineated, rather than allowing them to degenerate into a roar.

Although the Reference 64's soundstaging was first-rate, the No.30 threw a broader, more enveloping feeling of width. The Reference 64 was tighter, narrower, and more focused. The Krell's soundstage, however, had a pinpoint precision not heard from the No.30. In addition, the Reference 64 had a unique quality that I greatly enjoyed: a feeling of palpability and immediacy from instrumental images, yet a sense of space and bloom. Many processors with good space and depth lack visceral immediacy. Others that are forward and incisive tend to be dry and lacking in air and space. The Reference 64 was stunning in its ability to throw a present, tangible image surrounded by air. This was particularly noticeable on the Robert Silverman recording: the Steinway was right there in the listening room, with the sound of the hall enveloping the image like a halo.

The Reference 64's bass balance was leaner than that from the somewhat fat SFD-2, but fuller than the No.30's tight rendering. Overall, the Reference 64's bass was excellent, with exceptional dynamics and drive. In particular, I was impressed by the feeling of transient attack in the lower registers. Roscoe Beck's bass on the Robben Ford CD Robben Ford and the Blue Line (Stretch Records STD 1102) had a terrific feeling of bounce and rhythm, created in part by the Reference 64's rendering of bass dynamics. I also had this impression listening to the magnificent Steinway on the original 20-bit Nagra master tapes of Robert Silverman. The left-hand lines of the Liszt Sonata were reproduced with tremendous impact, power, and suddenness. Throughout the auditioning, I kept noticing this quality as well as the forceful, visceral immediacy the Reference 64 imparted to the music. In terms of LF pitch resolution, the Reference 64 was good, but not quite up to the standards set by the Sonic Frontiers SFD-2. For example, the SFD-2 was slightly better at revealing the way the Steinway's low notes changed harmonically as they decayed during the rests.