Mr. Zhao on Denafrips. "I guess it was in my DNA. My father and uncle were electronically trained. I was born into a family with electronic components where transformers filled all the places in our house. Their presence nurtured me from early childhood. I liked playing with electronics that could reproduce sounds like radios, tuners, Walkmen and such. I lost count of the number of devices I ruined when I was young. Later I acquired my Bachelor in Electronics at the university. After graduation, I worked for one of the largest audio OEM in China. For 15 years, my role was always product design so I accumulated some valuable design and development experience."


On discrete R2R: "R2R DACs aren't novel technology but the very first D/A conversion approach. Strictly speaking, using discrete R2R is no technological innovation but regression. It's the D/A integrated chip which is the embodiment of technological progress. From an audio perspective, it's simply not true that technical advances must automatically produce better sound. Sound is a simple yet complex matter. Sound is simple because it uses the most basic electronic knowledge. It is complicated because it also involves endless knowledge from outside the field of electronics where electronics suddenly become the most basic requirement. A good designer must also understand music cultivation, the recording processing, replay in a room dealing with specific acoustics and more. Good sound is about all of these intangibles. To reproduce high-quality music is not an easy task. To design good equipment, you must first know whether the sound is right or wrong; and then why it so. Playback must get as close to the recording as possible. So the designer must be able to recognize sonic differences and their poor aspects and possess the technical know-how to overcome them by electronic means.


"In effect, any converter IC that's been designed by a semiconductor firm is a black box. Its ultimate performance is already written to its hardware. In most cases, semiconductor engineers are not specialized audio engineers who understand music playback intimately. They won't perfectly solve certain sonic problems while they design their chips. Once professional audio designers work with them, they cannot overwrite their built-in parameters. Many of their own design ideas can't be exploited fully. They must always design around the limitations of their chosen silicon. This is precisely why we opted to revisit the most fundamental and early approach of the discrete resistor array. With it, we enjoy perfect freedom to design our very best solutions outside the usual constraints and boundaries which are imposed by the integrated converter chips.


"Just so, we cannot claim that a discrete R2R DAC is more advanced than one using ICs. In fact, discrete R2R is retro tech. Whether good or bad depends entirely on the designer's ability. If a designer cannot maximize and get the best from an integrated chip, there is no point to design a discrete R2R DAC; or vice versa. One must never presume that the discrete R2R approach is more advanced. It's simply an unconstrained platform from which a designer can fully implement his own ideas."


On DSD: "DSD or direct bit stream is a pulse-density modulation format. It can be decoded with an analog low-pass filter to remove out-of-band HF noise. That's the DSD hardware solution. It shares similarities with the core principle of digital amplifiers. It's simply that the analog filter won't ever be as steep as a digital filter slope so out-of-band noise is never completely filtered out. Thus properly done, the hardware approach to DSD conversion is far from simple. What we apply is a technique called FIR filter-core circuit. The famous DSD1700 chip used eight levels of FIR filtering. We use 32. In that sense we are more advanced than that chip was. But whether conversion to PCM or via simulated FIR filter, neither approach to DSD guarantees good results. Hardware decoding isn't innately superior to software conversion. Again it boils down to a designer's skill sets and his in-depth appreciation of the pros and cons of either method."

The power supply sits inside its own box which additionally sits on a central steel divider plate segregating the chassis into an upper and lower storey. Note the massive PCB traces.

On multi-paralleled capacitors. "The principle of an electrolytic capacitor is based on two metal plates with an in-between dielectric which creates a relatively large inductive value. The consequence of high inductance is non-ideal filtering of the AC power with very low and high-frequency ripple. By using arrays of many small caps in parallel to replace the large electrolytic capacitor, we avoid these drawbacks. Greatly reduced inductance improves the power supply's ripple suppression and quietness."