I am fascinated by DACs and the shifting sands of today’s digital-audio marketplace. This month, I am reporting on two more DACs, both made by Denafrips: the $4498 Terminator, until recently their flagship DAC, and the $768 Ares II, the company’s least expensive model. Like the HoloAudio May DAC I described last month, both Denafrips converters employ R-2R conversion schemes, and both render recordings into direct, unprocessed sound.
The Denafrips Terminator is not new. It has been around since March 2017 and has been reviewed and discussed (but not in Stereophile). When I checked, Stereophile‘s Recommended Components listed a baker’s dozen sigma-delta DACs, five of which use the same ESS9038Pro chip; two use the ESS9028 chip. Five others employ FPGAs. Inexplicably, there was not a single R-2R ladder DAC. Especially glaring was the absence of popular audiophile brands like AudioNote, Lampizator, Totaldac, Aqua, Audio GD, HoloAudio, and Denafrips—all of which specialize in ladder DACs. (MSB does, too, but they say theirs aren’t R-2R.) In my usual Herb-being-Herb-the-contrarian way, I have taken it upon myself to correct these omissions.
I like reviewing DACs
When I review DACs, I feel especially close to my readers. I mean, we all have a DAC, right? I assume also that most of you stream music from somewhere: Tidal or Qobuz or Spotify or Amazon HD or Apple Music. When I describe how a track sounds on my system, you can try it on yours. This is important because in these Denafrips explorations, I am going to be more specific than usual about what I am listening for and how I assess today’s digital products by their ability to recover all the ambient and reverberant information stored in recordings.
All non-anechoic spaces pulse with audible reverberant energy. The complex time and phase relationships of these interacting reverberations are what we use to locate ourselves in our environment and assess our immediate safety. With our eyes closed, audible reflected energy tells us we are in the bathroom not the garden. Most importantly, these intricate patterns of reverberant energy assure us—perhaps deceive us is a better word—into thinking that what we are experiencing is real.
Sound recordings have carefully calculated amounts of naturally occurring and electronically manufactured reverberation. Record producers use reverb to thicken up vocals and locate singers and instruments in acoustic space. Unfortunately, adding reverb is a dangerous proposition. Too much, and the singer loses density and moves too far back in the recording’s spatial illusion. Too little, and the singer sounds fake, dry, and disembodied. With the right amount of reverb, the singer displays a lifelike, intimate density. (I am deliberately not mentioning microphone proximity or the grossly audible effects of dynamic compression.)
The biggest challenge for audiophile audio is the proper recovery of all of the reverb captured on a recording—but no extra! I have found that only the finest systems can do this. The less-than-finest systems either subtract reverberant information, as a result sounding dry, gray, and hard; or they add reverb-like distortions (such as second harmonics, phase shift, or microphonics), which make recordings sound wetter or more atmospheric than they actually are.
It’s hard to say where recorded reverb ends and system-added “reverb” begins because no one really knows what reverb is on the recording. I think the best way to tell is: If the reverb you hear from your system sounds “whole and a part of” the resonant fingerprint of the recording as a whole, it is likely on the recording. Labels like ECM and 2L make this job easier by employing recognizable house-sound reverb. If the reverb you hear from your system seems different from the whole in form or texture, it is probably added distortion. For example, second-harmonic distortion fills in what would normally be empty spaces and clouds detail like a fog. It sounds limp and de-energized compared to what recording engineers use.
To put the importance of reverberant information in a larger biological context: I believe reverberation is the auditory equivalent of shadows in visual perception. Exactly like reverberant sound, the complex interactions of shadow-matrices supply the primary data that tells our brain where we are and that what we are experiencing is real. In concert with our other senses, shadows and reverberations are the dominant “facts” of our perceived reality.
Once I realized this, my empiricist mind became fascinated by recorded sounds of quiet spaces of varying dimensions. Hence, I have come to believe that the stereo system that recovers the most complete reverberant airmass from a recording of an empty church, a deserted subway, or a tent in the wilderness is the highest in fidelity. In like manner, the system that most accurately maps the perimeters of those reverberant spaces is the most resolving.
The Terminator
The Terminator DAC is designed by Denafrips’s chief engineer, Mr. Zhao. It is a chipless, discrete-resistor, 26-bit R-2R converter for PCM decoding with a 6-bit processor for DSD decoding. The only major change since its introduction was the addition of a DSD module in 2019.
The Terminator is built and assembled in Guangzhou, China, at Denafrips’s own facility and sold directly by global sales agent Alvin Chee at Vinshine Audio in Singapore (footnote 1).
According to the Denafrips website, the Terminator DAC is a fully balanced, dual-mono design. Its core conversion modules are constructed with 1000 (!) high-precision (0.005%!), low–thermal-effect resistors. It uses Crystek’s top-of-the-line CCHD-957 ultralow–phase noise femto clocks, with FIFO technology to “ensure the DAC clock is independent to the input signal.” An FPGA controls the switching of the resistor ladders.
The Terminator uses a beefy linear power supply with two O-core power transformers (one for analog, one for digital) situated in a fully shielded metal enclosure just below the DAC board. This two-compartment box measures 17″ (430mm) × 15″ (380mm) × 4.1″ (105mm) and weighs 42lb (19kg).
The Terminator has nine digital inputs: three S/PDIF (one RCA, one BNC, one TosLink), two AES/EBU (XLR, supporting dual L/R AES/ EBU), three I2S (one over HDMI, two over RJ45), and one USB. It supports data rates up to DSD1024 and PCM1536 (!) on the USB and I2S inputs. It decodes DSD natively on those inputs using those 0.005% precision resistors; the other inputs accept DSD64 via DoP and PCM data up to 24/192. As for outputs, there is one single-ended RCA pair with a fullscale output voltage of 2.3Vrms and a specified output impedance of 625 ohms and one balanced (XLR) pair with exactly twice that output voltage and impedance.
Vinshine Audio has created a set of quality-assurance standards to ensure that each DAC leaves the factory in “flawless operating condition,” as Chee put it in an email. Tests include 100 hours of burn-in followed by a comprehensive examination with an Audio Precision APx525 audio analyzer to ensure that all measurements are up to spec. This is followed by a listening test at all supported sampling rates via all the inputs. Shipping is by DHL, UPS, or FedEx. The warranty is three years, transferrable.
Reverb and tangibility
If you stream Qobuz, I entreat you to test your system right now by playing Vladimir Horowitz Rehearsal at Carnegie Hall, April 14, 1965 (Remastered) (24/96 FLAC, Sony Classical/Qobuz). This is not an uber-transparent, I-can-see-the-microphones recording, but the way it was recorded makes it an effective reviewing tool. I found it a perfect test of my reverb theory. As the master strikes the keys, it is easy to hear the start, middle, and finish of individual notes and chords. It is easy to “watch” the notes expand into the air surrounding the piano. The pedaling is also obvious.
The Denafrips Terminator in non-oversampling mode—it can also oversample—driving the Rogue RP-7 preamp driving the Elekit TU-8600 single-ended amplifier (with Takatsuki 300B output tubes) driving the Klipsch RP-600M loudspeakers (connected with Cardas Clear Cygnus wires) exposed the unique forms and harmonic spectra of the individual notes.
I noticed how each note’s harmonic expansion activated the space between the piano and the microphones.
In this same system, the similarly priced HoloAudio May DAC sounded similar to the Denafrips in most ways. Both were open, effortless, un-digital, and smooth. The May recovered a larger, darker, denser volume of room air, while the Terminator (still in non-oversampling mode) delivered a more brilliantly lit, less-shadowy soundstage.
Interestingly, both DACs presented Vladimir’s piano at the same size and weight, located in precisely the same place. (When a reviewer compares two audio components using the same recording, they may declare that component A puts the listener “mid-hall” while component B puts the listener in “the second row.” What they are really saying is that the two components recover different amounts of apparent reverb—and also, of course, where the microphones were placed and how they were mixed—footnote 2.)
Footnote 1: Denafrips. Web: denafrips.com. Global sales agent: Alvin Chee, Vinshine Audio PTE. LTD., Singapore. Web: vinshineaudio.com
Footnote 2: The listening room also adds its own sonic signature, but I listen in the nearfield, and anyway, to me, room reverberation is easy to separate out and ignore.
NEXT: Page 2 »
ARTICLE CONTENTS
Page 1
Page 2
Measurements
Click Here: penrith panthers team jersey