Digital Processor Reviews

Although I've been a happy owner of an Ayre Acoustics CX-7e^MP CD player for five years, I decided last fall that it was time to upgrade my CD playback. My choice was a PS Audio combination of DirectStream DAC ($5999) and DirectWave Transport. What I found particularly appealing about the DirectStream (DS) was that its data processing is performed by field-programmable gate arrays (FPGAs) rather than by a dedicated chip, and can be upgraded with downloadable firmware that's free to all DS owners.

When I got the DS last fall, it had firmware v.1.1.4, but a few weeks later saw the release of v.1.2.1. According to Paul McGowan, PS Audio's founder and CEO, the new firmware, the work of designer Ted Smith, fixed some technical problems of v.1.1.4 that John Atkinson had identified in his measurements, and represented an overall improvement in sound quality. Having installed firmware v.1.2.1, I noted a deeper, wider soundstage, more three-dimensional images, cleaner bass, and better-defined transients (see my Follow-Up in the February 2015 issue, p.129). In his measurement Follow-Up of the v.1.2.1-equipped DS in the March 2015 issue (p.115), John Atkinson found that, compared to v.1.1.4, both the noise floor and the already low-level power-supply products had dropped by 3–5dB; there were significant improvements in low-level linearity and IM distortion; and there was a "dramatic" drop in low-frequency harmonic distortion.

With v.1.2.1 installed, the performance of the DS was such that I wondered if, like Kansas City, they'd gone about as far as they can go. If there was to be another update, I figured it wouldn't be for at least another year.

But less than four months after the release of v.1.2.1, there's new firmware. Paul McGowan has been thinking about what these updates involve, and decided that, rather than referring to them as just updates of the "firmware," they would be more accurately described as updates of the DAC's operating system (OS), analogous to a new version of Windows. For this latest firmware—excuse me, OS—PS Audio abandoned the numerical designation, naming the new OS Pikes Peak. (PS Audio is based in Boulder, Colorado; I guess they're entitled to use the name of this famous Colorado mountain.)

Downloading and installing Pikes Peak caught me in the middle of a speaker review: a Follow-Up of the Fujitsu Ten Eclipse TD712z, now in Mk.2 designation. Fortunately, I'd been listening to the Eclipse TD712z Mk.2 for several weeks, so I had a pretty good handle on its sound.

Or so I thought. Installing the Pikes Peak OS seemed to transform the Eclipse's sound: more dynamic, better bass, more extended treble, even better soundstaging (already one of this speaker's great strengths). I won't say much more about the Fujitsu—I have to leave something for its review—but keep in mind that, like the original Eclipse TD712z, the Mk.2 has only a single 120mm driver—it's not capable of the sort of volume you can get from something like the GoldenEar Triton One. But with the DS updated with the Pikes Peak OS, the sound was decidedly more dynamic, almost as if the volume had been turned up.

I measured the output of my McIntosh MC275LE power amplifier at the speaker terminals, using the 1kHz tone on the first Stereophile Test CD, and compared the Pikes Peak OS with the v.1.2.1: the voltmeter readings were identical to the third decimal place. The DS with Pikes Peak OS somehow allowed the speakers to sound subjectively louder and more dynamic without increasing the gain.

The first CD I played through Pikes Peak was Simon and Garfunkel's Bridge Over Troubled Water, and while it had the enhanced dynamics and soundstaging that I found most appealing with v.1.2.1. Art Garfunkel's sibilants were more prominent than I'd been used to. I played a bunch of familiar records, trying to assess whether what I was hearing represented just higher resolution (definitely a characteristic of Pikes Peak), or higher resolution combined with some treble emphasis. As I continued listening, it seemed that whatever treble emphasis had been present at first was gradually disappearing. When I played Bridge Over Troubled Water two days later, the sibilant emphasis was gone, leaving behind an extended but smooth treble—apparently, a result of break-in. Switching back to v.1.2.1 (and allowing for suitable warmup), the sound was a bit softer, more laid-back, not quite as dynamic. I very much preferred the sound through Pikes Peak, but I can imagine that some people whose systems are on the borderline of being too bright might prefer v.1.2.1. Both are available from PS Audio's website.

What, precisely, are the technical differences between Pikes Peak and v.1.2.1? According to McGowan, they involve: a) a wider acceptance window for incoming data to improve compatibility with sources, including changes in the input architecture; b) redistribution of FPGA resources; and c) improvements in layout and data paths within the FPGA. McGowan continued: "The DirectStream operating system consists of two main bits of architecture: the FPGA and the Microcontroller. Each subsystem is written first in high-level code, then compiled into machine language. The compiling process has an impact on the sonic qualities of the final operating system, and both compilations must be matched to each other for best sound. Orchestrating this process, which sometimes requires each of the programmers to rewrite aspects of their code, is Arnie Nudell's task, a job taking several weeks of listening to pull off."

That explanation is way over my head. What I take away from it is that refinements of the DS operating system involve thorough technical understanding, hard work, and attention to detail, with listening providing the final arbitration. My system does not include DSD or other high-resolution sources, so I can't say what Pikes Peak might do for them—but CDs played through the PS Audio DirectStream DAC equipped with the Pikes Peak OS sound so good that I'm not about to rush into hi-rez.—Robert Deutsch