Back in the mid-1990s, I believed that the design of D/A processors was fundamentally a solved problem. The resistor-ladder, multi-bit DAC chips of the 1980s, with their linearity errors, had been replaced by sigma-delta types that had minimal linearity error down to the lowest signal levels. All that remained for the designers of PCM D/A chips was to increase resolution and dynamic range to the theoretical limits, and to improve the mathematical precision of oversampling digital filters to match the performance of the 20- and 24-bit recordings that had just begun to be released.
Twenty-five years later, nothing seems to have been resolved. The world of D/A processors has split into multiple warring factions. While sigma-delta DAC chips have indeed improved to an extraordinary degree, they are rejected by some hard-core audiophiles for failing to reproduce the force and attack of live music. Conversely, the digital processors featuring resistor-ladder DACs—which, their proponents argue, excel in these areas—can have linearity errors and levels of distortion that raise the eyebrows of conventional audio engineers, especially when those DACs are used without any reconstruction filter (such DACs are usually called “non-oversampling,” or “NOS,” footnote 1).
And even when the processor uses a digital reconstruction filter, arguments rage about whether that filter should be linear-phase or minimum-phase, and whether it should be short (have a very small number of coefficients or taps) or as long as possible. And that’s without considering newer forms of lossless digital conversion. Single-bit DSD encoding, featured in the Super Audio CD medium, came out of left field at the end of 1990s. And even within PCM, there is the MQA format, which rejects Claude Shannon’s 1948 thesis that the initial analog/digital conversion should be performed by eliminating all spectral content above half the sample rate, regardless of its level.
I recently spent some time with Aqua Acoustic Quality’s Formula xHD, from Italy, which Jason Victor Serinus reviewed in June 2018, and in June 2017 I reviewed the DAVE processor, from British company Chord Electronics. Both are expensive—the DAVE costs $12,488, the Aqua $17,000—and each represents one side of the DAC debate: the Aqua is a NOS resistor-ladder design, while the Chord uses the longest, most complex digital filter available.
As you can read in our December 2018 issue, while I felt the Aqua DAC didn’t commit any egregious sins, its sound quality was not totally to my taste (footnote 2). By contrast, Chord’s DAVE was one of the best-sounding DACs I’ve had in my system. I miss it still.
So when Chord Electronics announced a new DAC at a much lower price, the Qutest ($1895), featuring technology trickled down from the DAVE and identical to that used in Chord’s Hugo2 ($2395, footnote 3), I didn’t need my arm twisted to agree to review it.
Describing
The Qutest is a small, black device wider than it is deep (6.3″ vs 2.85″) and housed in a case of machined aluminum. A circular glass window in the top panel illuminates to indicate the sample rate of the incoming data: red for 44.1kHz PCM data, through orange, yellow, green, blue, and violet as the sample rate increases to 768kHz; and white for DSD data up to DSD256 encoded as DoP.
Two spherical, polycarbonate buttons are inset in the front panel. The left button selects the reconstruction filter chosen, of which four are available: Incisive Neutral (button illuminates white), Incisive Neutral HF roll-off (green), Warm (orange), and Warm HF roll-off (red). The right button is used to select the source. On the rear panel are four digital inputs: two S/PDIF on BNC jacks (yellow and red button illumination, respectively), one S/PDIF optical on a TosLink jack (green), and a USB port (white). The USB port handles PCM data sampled at up to 768kHz, and DoP-formatted DSD data up to DSD256. No USB driver is needed for Macs; PCs require an ASIO driver, but can send native DSD data up to DSD512 (not that I’m aware of any music files available at that resolution). Unusually, the two coaxial S/PDIF inputs can be operated in parallel to cope with 32-bit PCM data sampled up to 768kHz from Chord Electronics’ Blu Mk.2 transport.
The Qutest has a single pair of unbalanced analog outputs on RCA jacks. The maximum output level is fixed, but if the Filter and Source buttons are pressed simultaneously when the DAC is powered up, the level can be set to 3V, 2V, or 1V.
The Qutest is powered by an external 5V, 2A supply, this connected by a Micro USB port. The DAC is Chord’s proprietary, 10-element Pulse Array Design type, designed by Rob Watts and implemented in a Field Programmable Gate Array (FPGA) chip. The oversampling digital reconstruction filter uses 45 208MHz DSP cores in the FPGA and operates at 16Fs. In keeping with Watts’s philosophy (see my 2017 DAVE review for extracts from my interview with him) is very long, with 49,152 taps compared to the typical 128 or thereabouts.
Setting up
I connected one of the USB ports on my Roon Nucleus+ server to the Chord Qutest with a 6′ AudioQuest Coffee USB link. After the Roon 1.5 app on my iPad mini had recognized the Qutest, I enabled the DAC as an audio player and set Roon to send it DSD data without first converting it to PCM. I used my Ayre Acoustics C-5xeMP player as a CD transport, sending the Qutest S/PDIF data over an Esperanto coaxial link. (As the Ayre has only an AES/EBU digital output, I pressed into service a Z-Systems RDP-1 digital preamplifier, set to its bypass mode, to act as a format converter.) The only unbalanced interconnects I had that were long enough to reach the amplifiers in use were the 3m Canares I usually use in the test lab. As the Chord lacks a volume control, with the Lamm M1.2 Reference monoblocks I used a homemade passive volume control based on a stereo 10k ohm Alps potentiometer.
Listening
I started my auditioning with the Qutest’s Incisive Neutral filter. In my review of Chord’s DAVE, I’d commented on the clarity with which it reproduced recorded detail. With the Incisive Neutral filter, I was strongly reminded of that description. In pianist Angela Hewitt’s performance of Mozart’s Piano Concert in c (16/44.1 ALAC file, Hyperion CA 68049), every instrument in the National Arts Center Orchestra was stably and clearly positioned behind the piano. Perhaps there wasn’t quite the soundstage depth that I remember from the DAVE, but that was of minor concern, given the excellent sense of forward motion the Qutest lent to this music. Similarly, with Rachel Podger’s performance of Vivaldi’s La Stravaganza Concerto 6 in g for violin, strings, and continuo (24/96 FLAC file, Channel Classics CCS SA 19503), the baroque orchestra was unambiguously placed in space, with the theorbo, cello, and harpsichord continuo in the slow movement sounding particularly well-focused.
Footnote 1: See the relevant section of my 2011 Audio Engineering Society lecture here.
Footnote 2: See Herb Reichert’s Follow-Up in this issue (p.117).
Footnote 3: See here and here.