The Concert Fidelity DAC-040 Digital to Analog Converter
by Dick Olsher
Schoolhouse Rock Revisited: Zero is my hero!
Why a zero oversampling DAC? In May of 2006 I reviewed a quartet of DACs that challenged the assumption that an anti-imaging filter is necessary to remove ultrasonics generated during the digital to analog conversion process. Recall that the earliest players used 16-bit multi-bit DAC chips followed by analog "brick wall" filters designed to steeply attenuate image spectra above 22 kHz. While the specs looked good on paper, the emperor was soon discovered to have no clothes. A chorus of complains emerged echoing a common theme: bright, fatiguing sound that ultimately resulted in digititis - an allergic reaction to digital sound. Oversampling, also known as upsampling, and digital filters were invented as a means of easing the analog filter requirements. Instead of a 9th order or higher low-pass filter, a third-order was now possible. In the years that followed, the old Philips 16-bit DAC IC was replaced by cheaper, easier to calibrate, single-bit sigma-delta DACs which featured fancy noise shaping techniques to make them sonically palatable. Such noise shaping makes anti-imaging filtration mandatory. However, it does not appear to be a necessity in the case of a multi-bit DAC. It was Audio Note UK’s Peter Qvortrup who first opened Pandora’s Box. These Philips chips were not considered anything special in engineering terms either then or now. However, listening minus analog filters and what he refers to as “digital trickery,” Peter discovered a sound quality much more closely aligned to the vinyl experience. And that my friends is the zero oversampling advantage: smooth textures, solid image outlines, and the impression of a more believable 3D soundstage in terms of depth and width perspectives.
Today it is hard to find anything but 24-bit DAC chipsets in production. For example, my current reference, the Weiss Engineering Medea, sports the Analog Devices AD1853 chipset. This is a 24-bit/192 kHz multi-bit sigma-delta DAC introduced already 10 years ago around 1999. My favorite low-cost DAC, the Altmann Micro Machines Attraction uses the Philips TDA1543 dual 16-bit DAC IC introduced in 1991. And a favorite for DIY projects, the Philips TDA1541, was introduced in 1985. These examples highlight a minor design trend I have observed over the past several years of giving preference to older DAC chipsets. These Philips 16-bit DAC chipsets are still available as New Old Stock (NOS) and actually cost quite a bit more than current production 24-bit chipsets. Some of these DAC ICs have become collectible: select versions of TDA1541A (e.g., double crown markings) may fetch over $100!
Inside the DAC-040
Concert Fidelity’s designer, Masa Tsuda, asks a good question: Why is there a need for 18-bit or 20-bit DAC conversion in CD playback system, when there is only 16-bit worth of data on a CD to begin with? He says that in DAC-040 he undertook a radical approach given that the goal was to find out what 16-bit sound is really like and exploit its possibilities. Substantial listening tests and measurements were conducted during the development process to determine the audibility of filter circuits and DAC performance parameters. Masa was willing to trade measured specifications for audible results. That’s certainly a tradeoff music lovers should embrace as it meets the time-honored criterion of letting the human ear be the final arbiter in matters of sound quality. In the end, both the digital filter and analog low-pass filter circuits bit the dust. Towards that end, a NOS Philips multi-bit chip DAC was chosen not because its performance was judged to be superior to that of other NOS DACs but because it was felt that the best approach was to simply eliminate problems inherent in the architecture of today’s FIR type digital filters. The DAC IC chosen allows for compact mounting and short signal path – important priorities at Concert Fidelity. Masa believes that these considerations are even more important in digital than in analog circuits. By the way, in case you’re curious, it is not the TDA1541 because it apparently did not meet Concert Fidelity’s criteria for mounting due to its physical configuration. In fact, the specific identity of the DAC is kept confidential and is hidden from prying eyes in a black box. You should know, however, that each chipset undergoes a critical selection process; only about 10% of the entire population makes it into production.
Ultra-high precision resistors are used in the current-voltage (IV) converter circuit. The output stage is a vacuum tube gain stage (+6 dB) featuring the same circuit originally developed for the world-class CF-080 line preamplifier. As was the case with the CF-080, I substituted Mullard box anode 12AU7 twin triodes for the stock tube complement. The output stage provides response linearity within the audio bandwidth. However, its main task is to prevent generation of inter-modulation distortion caused by high-order harmonics, ultrasonics normally attenuated by analog low-pass filter, from impacting the next component in the system. Sound quality through simplification, I’m sure it pleases Mr. Tsuda to note that in the final analysis, sound quality of a DAC or CD player is largely determined by the analog section. Note that only a single S/PDIF input is provided and a BNC connector at that. A BNC to RCA adapter is included. That’s appreciated, but it would be best to replace the BNC with an RCA, something I’m told will happen in future production.
In recent years jitter reduction has been front and center as a critical aspect of digital component performance. It should therefore not surprise you to find out that my opinion of the CF-040 varied dramatically based on the associated transport. Since the Weiss Engineering Jason was on hand, the CF-040 benefitted from this coupling and what you’re about to read should be interpreted in this context. The Jason, Swiss-engineered and admittedly very expensive, is simply the best sounding transport I’ve ever auditioned. If you’re skeptical about the importance of a good transport, just give the Jason a listen. Unfortunately, the CF-040 wasn’t tolerant of lower pedigree (i.e., higher jitter) digital input. When using the PrimaLuna Model Eight’s digital output, the CF-040’s performance was just a shadow of its former self. By contrast, the Medea DAC (the Jason’s partner) proved more tolerant of the PrimaLuna Eight’s digital output. Don’t expect to make this puppy sing unless you’re willing and able to match it with a world-class transport.
What grabbed my attention pretty damn quickly was its sweet midrange, vibrant harmonic colors, and remarkable timbre accuracy. Female vocals were right on – as good as I’ve ever experienced on CD. Now, this is no mean feat, as I’m quite fussy when it comes to the vocal range, spanning a bandwidth of about 200 to 4,000 Hz. My annoyance factor for residual grain, grit, and gratuitous brightness is extreme. In fact, the CF-040 sounded quite the opposite, being smooth and refined. What a find! I’ve struggled for over twenty five years to find a DAC that captures the music’s heart and soul. I’ve had to put up with either bland, sterile sound that neither offends nor excites, or at the opposite end of the spectrum, with dynamically nuanced reproduction that failed to do justice to harmonic colors. The CF-040 seemed to combine microdynamic intensity with a textural palette that more closely approached my impression of the real thing.
The tube nature of its output stage was very much in play. Textures were slightly softer relative to the much more expensive Medea DAC, probably a consequence of a treble range that was a bit dark in character. Though not that far behind, the Medea with its solid-state output stage also trumped the CF-040 in terms of transient speed and low-level detail. On the other hand, the Concert Fidelity clearly asserted itself in terms of midrange suaveness, harmonic color saturation, and imaging. It became clear that the root cause of some of these sonic differences was nothing more that the age-old difference between tubes and transistors. This is not the sort of DAC you would want to mate with an overly romantic sounding line preamplifier lacking treble air and extension. Not surprisingly, Concert Fidelity’s own CF-080 line preamplifier provided the most neutral tonal balance with excellent bass control and definition. Other line preamplifiers were given room to impose their own sonic character, which leads me to conclude that the CF-040 is inherently neutral and free of any significant euphonic colorations.
Image outlines were clearly delineated, superbly focused, with plenty of palpability and analog-like stability. Soundstage dimensions unfolded with the sort of depth and width perspectives I’ve only heard once before from a DAC or CD player. I’m thinking here of the PrimaLuna Eight, which however lacks the textural purity and integrity of the Concert Fidelity DAC.
At the end of the day I have to confess that I’m unable to resist the CF-040’s siren call. Vivid harmonic colors, superb rendition of instrumental timbres, and an analog-like presentation that draws the listener into a spacious and solid soundstage. Zero oversampling truly is a gateway to a kinder, gentler, and ultimately more musical digital experience. It’s hard to believe that a NOS Philips multi-bit DAC chipset can be engineered to put current 196 kHz/24 bit Delta-Sigma DACs to shame, at least in musical terms. But then, as I’m sure that Mr. Tsuda would agree, it was never about the specifications.
2477 Paseo Circulo
Tustin, CA 92782
Contact: Hajime Sato
Digital inputs: 1-S/PDIF
Output: 1-Pair single ended (RCA)
D/A Converter: Philips Multi-Bit 16-bit DAC
Sampling Frequency: 44.1kHz
Tube Complement: 12AU7 (2) for gain (tube-swapping possible, NOS tubes preferable), 6CA4 (1) for rectification
Power: 110-120VAC / 60Hz
Dimensions: 450mm (W) X 100mm (H) X 310mm (D)
Net Weight: 6.5KG