Note: for the full suite of measurements from the SoundStage! Audio-Electronics Lab, click this link.
Recently I’ve been reviewing a lot of DACs, and I’ve noticed that two of the biggest differentiators among them are the type and quality of digital filter used. I have yet to hear a DAC from, say, Chord or dCS, so this is only a hypothesis, but I believe that the digital filter is at least as important as most other technical aspects of a DAC.
An interesting test of this hypothesis showed up at my door in the form of the Denafrips Ares II R-2R resistor-ladder DAC, which costs $1098 in Singapore dollars (ca. $810 USD at time of writing; all prices USD except as noted), as it’s made in China and marketed through Singapore. The Ares II offers the user a choice of one of two linear-phase digital reconstruction filters, or, in its non-oversampling (NOS) mode, no filter at all. I’ve been interested in resistor-ladder DACs for a while; e.g., my own Schiit Audio Modi Multibit.
Denafrips’s Ares II is a roughly half-rack-size DAC measuring 8.5ʺW x 1.85ʺH x 9.1ʺD and weighing 7.7 pounds, most of that weight down to its large toroidal transformer, thick metal faceplate, and heavier-than-usual-gauge metal casework. Sitting on four feet, each a truncated cone, it feels strong enough to survive being run over with a car (but don’t—that ain’t covered by the three-year warranty).
The front panel—available only in black—is a classic case of function following form: Its looks and layout are aesthetically pleasing, but I found it an ergonomic challenge. The Ares II looks quite different from all other Denafrips DACs, which have a more or less uniform design and a user interface that looks easier to navigate. At left, under the Denafrips logo and the model name, is a large Standby button. Most of the central portion of the faceplate is blank; then, at right, is a row of seven inset buttons, flanked above and below with corresponding rows of tiny red LEDs, each of the 14 LEDs labeled. The upper row is dedicated to the input selected: USB, Coaxial 1 and 2, Optical 1 and 2, followed by Phase and Mute. The lower row indicates the oversampling rate or DSD: 44.1kHz, 48kHz, 1x, 2x, 4x, 8x, and DSD. Unfortunately, the labels are quite small, and the tiny LEDs are hard to see from an angle. I wish they’d used all that empty space for bigger labels and LEDs.
On the rear panel are one USB Type-B and four S/PDIF inputs: two coaxial (RCA) and two optical (TosLink). There are also pairs of single-ended (RCA) and balanced (XLR) outputs on gold-plated jacks. At far right are the main power rocker, the fuse bay, and the IEC power inlet.
The Ares II supports 24-bit PCM data at sample rates of up to 1536kHz, native DSD up to DSD1024 via its USB input, and 24/192 PCM and DSD64 via all other inputs. PCM and DSD signals are sent to separate hardware decoders. One uncommon feature that I appreciated is that the Ares II reclocks its S/PDIF inputs as well as its USB inputs. I did most of my listening via the USB input, as that seems to be the most popular method these days. The dedicated USB driver required for use with a Windows-based digital source can be downloaded from the Denafrips website.
What most intrigued me about the Ares II was its non-oversampling mode. I felt a bit like the guys in The Matrix, looking at the raw code flowing down their computer monitors. Because the Denafrips ended up spending four months in my system, I was able to compare its sound with those of several DACs using different chipsets and filters.
There are very good reasons to use oversampling: moving the Nyquist frequency further out of the audioband, allowing use of digital reconstruction filters, preventing aliasing, sculpting the data back into something more closely resembling the original analog voltage signal, etc. The Nyquist frequency is half the sample rate, and the highest frequency that can be recorded with at least one sample each for the positive and negative halves of the waveform—the minimum amount of information needed to reconstruct an audio waveform. (This is why 44.1kHz was chosen as the sampling rate for the Compact Disc: Its Nyquist frequency of 22.05kHz includes the entire audioband of 20Hz-20kHz.) Raising the Nyquist frequency also relaxes the requirements for the analog low-pass filter, reducing its negative effects on the audioband.
But there’s no such thing as a free lunch—oversampling has its downsides. Chief among them are phase shifts and “ringing” from the signal processing. Taking the filter out of circuit logically flips the advantage/disadvantage calculus. If you were to output a sinewave from a NOS DAC, you might be reminded of a ziggurat, or an old video game (or Minecraft) in which all lines are either horizontal or vertical. To me, at least, it looks far worse than it sounds. The advantage is, essentially, perfect impulse response with no ringing. The main disadvantage of a NOS DAC is that it will not remove signal aliases above the Nyquist frequency without a very steep analog low-pass filter. Also, because of the blocky nature of the output waveform and the small number of samples (data points), at high frequencies NOS DACs have a natural high-frequency rolloff (it’s a calculus thing: the area under a curve, and all that).
On their website Denafrips talks a lot about the musical superiority of their non-oversampling mode, and nothing much about their oversampling modes. A lot of this is, of course, marketing, but it’s clear they intend the Ares II to be used primarily in NOS mode. I tried the two OS filters, Fast and Slow, and while they improved NOS mode’s treble rolloff, they still seemed to thin and dry out the sound. Because Denafrips stresses bypassing them, I suspect the OS filters were not the highest priority in the design of the Ares II. I did all of my critical listening in NOS mode because I found the sound, overall, more pleasant.
I used a variety of gear during my time with the Denafrips Ares II, and found in some of those pairings that its high output impedance affected the sound. My reference Tandberg 3012 integrated amplifier has one input specified at 13k ohms that, as in Kinki Studio’s THR-1 headphone amp, is unbuffered and sent right to the volume control. The other line-level inputs in the Tandberg have a nice, high input impedance of 100k ohms. I tried both, because the Ares II produced some interesting results into the Kinki when the input impedance dropped. Long story short, and as an electrical engineer would expect, the Ares II sounded much better going into the much-higher-impedance inputs of my Tandberg, so that’s how I used it.
Also used with the Ares II were my reference system’s trusty Boston Acoustics A70 Series II speakers, and two headphone amps that are part of my desktop system: a Schiit Asgard 3 and the Kinki THR-1 (reviewed on SoundStage! Solo). That desktop system also included pairs of HiFiMan HE1000 V2 headphones and Emotiva Airmotiv 4 active speakers. Comparison DACs were primarily my Schiit Modi Multibit and a Cambridge Audio DacMagic 200M I recently reviewed. Sources were mostly Tidal streams, along with some CDs played on my trusty ol’ Sony DVP-NS55P DVD player, used as a transport.
Sound and use
The first thing I noticed when I plugged in the Ares II was the ease with which music came out of it in NOS mode. In a word, it sounded natural. In some ways this was counterintuitive—what comes out of the Ares II in NOS mode most definitely did not match the input signal before analog/digital conversion. The same is true for OS mode, but for different reasons (mainly filter ringing).
Fortunately, our brains are pretty adept computers that can adapt to a lot of things. One thing I adjusted to was the high-frequency rolloff of the Ares II’s NOS mode. Some of this is inherent to the NOS output but it rolls off more than calculus alone can explain. The Ares II will require some careful system matching in NOS mode. The rolloff was never really bothersome, but in the few instances where an already dark-sounding recording demanded it, I was able to add an appropriate amount of HF boost via my Tandberg integrated: +2dB at about 6kHz did the trick. (The Tandberg has a selectable turnover frequency for both its high- and low-shelving EQ.) I wouldn’t recommend the Ares II for use in a system whose sound is already tonally on the dark side, but I think it would mate well with any system that’s on the brighter side of flat.
I’m a big fan of Paul Weller’s first two solo albums: the eponymous Paul Weller, which draws heavily from late-1960s music; and Wild Wood (both albums, 16-bit/44.1kHz FLAC, Go! Discs Ltd./Tidal), which has an early-’70s vibe. Wild Wood sounds relentlessly analog, with a ripe lower midrange—some might say it sounds a bit “woody.” In the first track, “Sunflower,” the Denafrips accentuated this analog quality without adding any mud (or fine wood paneling) that wasn’t already on the recording. Every track on this album, which I highly recommend for the music, sounds a little rounded-off, as analog tape tends to; that sound was retained by the Ares II without any sharpening or smoothing.
I felt I would be remiss if I didn’t listen to Adele’s mega-hit, 21 (16/44.1 FLAC, XL/Tidal), with its sharp treble known for its ability to shave the cilia from your inner ears. Most people attribute this album’s bright sound to a lot of compression and high-frequency equalization, but the cause is more complex. The Ares II let it be heard without much problem while providing a welcome taming of that blinding brightness—Ray Bans for your ears. Andrew Scheps, who mixed 21 as well as Metallica’s Death Magnetic and many other albums, is known for using a lot of compression in the mixing stage. Like running compressors in series together to remove dynamics. All of them. Most of the high-frequency problems audible in 21 arise, in my opinion at least, from excessive conversions between analog and digital, plus HF EQ to compensate for the effect of compression on HF balance. Compressors are triggered by the level of the signal passing through, and in most pop and rock there’s more energy in the bass and midrange than in the treble. This action tends to increase the level of the low frequencies in relation to the treble, and the signal is often equalized to compensate for this—though that’s not the only way to do it. All of this is on display in Adele’s “Rolling in the Deep,” including overcompression of the vocals in the chorus. If you listen carefully to the loudest vocal parts in the chorus, you can hear Adele’s voice suddenly lose some HF content and slightly recede, a compression artifact known colloquially as “too much.” This was clearly audible through the Ares II.
“Trip Like I Do,” from the Crystal Method’s Vegas (16/44.1 FLAC, Tiny e/Tidal), had more and juicier texture through the Ares II than I’ve heard from any other DAC. To say that I haven’t heard every DAC out there is an understatement, but through the Ares II this track combined both a great deal of corduroy and faux-fur texture with the moist atmosphere of a packed dance club on a hot August night. The Ares II is by no means a neutral DAC, nor do I think it was intended to be. What it did best was reproduce music in a fluid and enjoyable way. I really liked that.
The differences between the Denafrips Ares II and the Cambridge Audio DacMagic 200M ($499) were stark. The Denafrips costs about 50% more, but both easily fall in the range of “under-$1000 DACs.” The Ares II gave me a more pleasurable listening experience and far more three-dimensional soundstages. It didn’t have quite the DacMagic 200M’s retrieval of detail, but there will always be trade-offs in this price sector. With the Crystal Method’s “Trip Like I Do,” the Ares II exhibited a little less drive but a much bigger soundstage, and with none of the Cambridge’s rounding of attacks. The DacMagic 200M threw a fairly static soundstage with little height and width; the Ares II added not only more depth, but more movement on the soundstage.
With this track and in general, the Schiit Modi Multibit fell somewhere between the Cambridge and Denafrips in terms of texture and soundstaging, while revealing less detail and more noise than either. Like the Cambridge, the Schiit doesn’t roll off the highs as the Denafrips does. Schiit’s single digital filter sounds better to me than any of Cambridge’s three digital-filter choices, but the Schiit has the unfair advantage of being designed by Mike Moffat of Theta Digital fame, as well as a smidgen of low-mid boost to warm up its sound.
Listening through the Denafrips to “I’ve Never Been to Memphis,” from Lyle Lovett’s Joshua Judges Ruth (16/44.1 FLAC, MCA/Tidal), was a stark contrast to hearing the same track through the DacMagic 200M. There was some loss of air due to the Ares II’s rolloff of the highs, but all the texture I expect from this recording, which was engineered and produced by George Massenburg, was there. The bass poured in from the right side of the soundstage, as I expect it to, again unlike through the DacMagic, which couldn’t reproduce aural images as well as the Ares II in NOS mode. That bass didn’t hit with quite the impact it has through my Schiit Modi Multibit, but they were texturally similar. Again, the Ares II was quieter than the Schiit—noise was removed like winnowing chaff from wheat.
Compared to all the DACs I’ve listened to this past year—Clarus Coda ($300), Cambridge Audio DacMagic 200M ($449), Schiit Audio Modi Multibit ($249)—the Denafrips Ares II ($810) in NOS mode sounded consistently darker in tonality while managing to sound more natural and right. I attribute a lot of this “rightness” of sound to the lack of a digital reconstruction filter and its inevitable ringing. Is this essentially perfect impulse response really that important? To me, it very well may be, though the scientist in me (I’m a biochemist) needs more data. I did notice that, with a very small subset of original-issue, digitally mixed ADD or DDD CDs from the 1980s, the Denafrips Ares II all-too-convincingly transported me back in time to that era of wonky-sounding early digital, the latter no doubt due to the quality of the digital recorders used in studios. Digital didn’t sound that great in the 1980s, even in the professional world, where mastering for the pressing plants was done on a system comprising a Sony PCM 1610/1630 video recorder and 3/4ʺ U-matic video tape—yup, basically a big ol’ VCR. Modern oversampling converters can significantly improve the sound of CDs from back then, but the Ares II in NOS mode reproduced a few of them—e.g., The Cure’s Pornography (CD, Elektra 9 60785-2)—with a bit too much honesty, ending up sounding similar to those old converters.
I very much enjoyed the Denafrips Ares II DAC’s long visit in my systems. It has some old-fashioned technology and some technical shortcomings, but the limitations of its user interface did not extend to its sound. It reproduced music in a consistently relaxed and invariably enjoyable manner—even from 35-year-old CDs. If your main concern is simply enjoying music from digital sources but you think digital sounds inherently too sterile or clinical, then the Denafrips Ares II might scratch an itch you didn’t know you had. If you’re not sure, listen to another resistor-ladder DAC with non-oversampling mode—there are quite a few out there these days. What once was old is new again.
. . . Mark Phillips
Note: for the full suite of measurements from the SoundStage! Audio-Electronics Lab, click this link.
- Source: Sony DVP-NS55P DVD/CD player (as transport), Dell Alienware desktop and Toshiba notebook computers (via USB)
- Speakers: Boston Acoustics A70 Series II, Emotiva Airmotiv 4
- DACs: Cambridge Audio DacMagic 200M, Clarus Coda, Schiit Audio Modi Multibit
- Integrated amplifier: Tandberg TIA 3012 (restored by Soundsmith)
- Headphones: HiFiMan HE1000 V2
- Headphone amplifiers: Kinki Studio Vision THR-1, Schiit Audio Asgard 3
Denafrips Ares II Digital-to-Analog Converter
Price: $1098 SGD (ca. $810 USD).
Warranty: Three years parts and labor.
Denafrips Co. Ltd.
33 Ubi Avenue 3