An Interview with John
Stronczer of Bel Canto
When my review of the Bel Canto 3.5VB DAC/preamp (see elsewhere on PFO) was nearly finished, I forwarded several questions to John Stronczer, the founder of Bel Canto Design.
Robert Learner: How have Bel Canto DACs evolved? Or perhaps more pointedly, how has John's thinking on DACs evolved? Looks to me like there's been a honing of technologies to reduce noise and jitter, as well as some adjustments and advancements with digital filtering? John?
John Stronczer: Starting with our first DAC product in 1991—the Aida—all our DAC designs have focused on understanding and reducing errors that influence the D/A conversion process. Primary focus has been on reducing analog errors. As the core dynamic range performance of DACs has improved to greater than 120dB the influence of external error sources on its performance has increased. The latest generation DAC1.5, DAC2.5 and DAC3.5VB have definitely evolved up a level with improvements to jitter isolation, power isolation and noise and output-stage layout and performance.
We dealt with power supply noise by developing our VB—Virtual Battery—technology. The DAC3.5VB combined with the VBS1 provides an extremely effective level of isolation from AC power noise. There's over 100dB of filtering by 100Hz that isolates and protects the power supplies from noise modulation coming in through the power line. There's over 1 Farad of filter capacitance and energy storage, along with several stages of inductive filtering for a power supply noise floor that's lower than many analog preamplifier stages! We use switching power supplies in the front end of the VBS1 to lower audio-band noise and move low frequency noise up to very high frequencies where it can be easily filtered. VB technology can be found in the power supplies of both the DAC2.5 and DAC1.5 as well.
The next critical analog error source we address is clock jitter noise. Clock jitter behaves much like power supply noise and causes intermodulation of the original signal that's easily perceived as a loss of dynamics, inner detail and resolution. We handle jitter errors with two-stage filtering that works like to our VB technology but reducing jitter from digital sources. The second stage uses a digital PLL and filter that starts at 2Hz providing 10,000 times the reduction of incoming jitter by 100Hz! Effective jitter filtration in conjunction with our local low-jitter clock (less then 2pS RMS jitter) provides a pristine environment to take advantage of the 126dB dynamic range of the DAC3.5VB. You'll find this jitter filtration and low-jitter clock—the Master Reference Ultra Clock™—in all of our new DAC products.
Output stage refinement consists of improved PCB layout to better take advantage of the differential operation of the 24/192 DACs and selected audio grade 0.1% resistors used in all of our DAC products.
RL: More narrowly, what is the difference going from the x.0 series to the x.5 series, particularly the 3.0VB to the 3.5VB?
JS: Refinements to the power supplies brought about by our VB technology, jitter rejection, board layout, and local supply decoupling devices are the primary improvements in the DAC3.5VB. The core DAC and analog stages are essentially the same as the DAC3 but the isolation from outside interference through the power supply plus high jitter filtration allows the original DAC3 core to perform at a significantly higher level. The DAC1.5 and DAC2.5 designs grew from the DAC3.5VB development and share little with the original DAC1 and DAC2 designs.
RL: What do you mean by: "with our proprietary low noise digital to analog converter core" on the description page of the 3.5BVB?
JS: It refers to the PCM1792 in the DAC3.5VB and the PCM1796 in the DAC2.5 and DAC1.5. Also the specific amplifier devices used, resistors, analog supplies. The entire analog section relies on a single gain-stage folded cascode biased to class-A, and direct-coupled throughout. And our DAC cores, as we call them, are capable of class-leading 122dB to 126dB dynamic range!
RL: How did the present chassis design come about?
JS: The e.One chassis was originally inspired by our ability to put a tremendous amount of technology and performance into a compact chassis by using advanced switching power supplies and amplifier technologies. The first e.One product was the REF1000 monoblock amplifier, and a complete line followed from this original form factor. This chassis design has numerous benefits that weren't completely obvious when first chosen. Their compact size also speaks to efficiency, material usage, shipping costs and "livability" of the designs.
The compact size imposes a focused design esthetic where the chassis is stripped down to its essentials to achieve our performance and function goals. The compact, strong, rigid chassis also brought about benefits in mechanical performance, reducing the resonant profile of the chassis without resorting to expensive and massive construction. At one point we tried aluminum and found that the sonic signature changed with notable upper mid-range prominence. We traced it to mechanical resonance from the lighter and less rigid aluminum. We went back to the original steel construction and immediately heard our signature sound balance restored!
RL: How did Bel Canto SET become Bel Canto digital lifestyle-sized components?
JS: The e.One product line is really a logical step in design. I've always worked with forward-looking, leading-edge technologies. The beginning of my engineering career included more than 10 years in government research in advanced technologies. The R&D environment was instilled into me from the start and it's in my DNA... I've checked! I still do work for fiber-optic communications and keep a hand in advanced semiconductor circuit design work.
Tremendous amounts of government and industry R&D money are focused on improving efficiency. The need to do more with less is fundamental to progress. The original SET amplifier was a sideways approach to this concept, realizing an entire amplifier with just two stages and a single output device was my "physics experiment". I designed my first 845 amp back in 1986. SET design was efficient regarding the number and use of amplifier devices and signal stages but horrendously inefficient in every other way. After living with SET amplification for 12 plus years I decided to look ahead to class-D technology.
In fact, class-D architecture actually captures some of the elegance and simplicity of SET amps. A class-D amplifier uses only 2 output devices to drive the load and does so with great efficiency. The evolution of class-D and switching power supplies has been very exciting. The latest architectures are giving traditional amplifiers a real run for their money and I'm convinced they'll eclipse older technologies in absolute performance terms.
The notion of efficiency drives all our product development. For example the new DAC3.5VB with VBS1 uses only 6 or 7 watts of power while the original DAC3 used 12 to 13 watts. We lowered the power usage and greatly improved performance with our VB technology. That's what I call a win-win!
Bel Canto products focus on system architecture and pare it down to a DAC driving a class-D amplifier and getting closer to the original recording in the process. Drawing from advances in switching power supplies and amplifiers, DAC and ADC technologies, Digital Signal Processing, system architecture and analog design I can better hear what the original recording sounds like whatever the source material.
RL: It seems that some DACs are concentrating on the math these days with, for example, apodizing filters like Meridian and Ayre. They're focusing on digital signal processing, and this is where they feel the most improvement in DACs is to be found. What is Bel Canto's take on this?
JS: I've looked into apodizing filters, they're interesting and will no doubt provide improvements in future. However, as with any technology, application is critical. Before applying advanced filters and DSP you've got to get everything else just right. Low-jitter from all sources, low power supply noise, isolation from power line noise, optimized analog output stages, active and passive analog component selection, even PCB layout, they all profoundly influence the ultimate performance of the DAC. There's still much to get right before adding DSP and new filter algorithms.
RL: What would you say is the common "DNA" of Bel Canto products in terms of design, sound, philosophy and execution? In other words, what makes a Bel Canto a Bel Canto?
JS: Cliché or not, Bel Canto components get out of the way of the original recording. I've done recording work myself over the past couple of years and that's informed some of my design decisions. I also play jazz piano as a committed amateur in a student ensemble so I recalibrate my ears on a routine basis! And I'd say my commitment to the sound of live music has influenced my designs as well.
After all, it's easy to change the sound of a given system in some way. It's much more difficult to make changes that consistently result in better performance. Strong engineering, adherence to scientific principles, and critical listening with a view toward improved clarity, reduced noise, distortion and signal modulation of any type yields the most meaningful improvements.
I'm not going for a specific house sound in our products; they should all be transparent to the source. As you move up our range of products the fundamental sound changes little, but with higher specification power supplies and other refinements you can hear into the original recording for a more dynamic and engaging experience.
A design that obscures or colors the sound is a step backwards in my view. Total transparency, strong dynamics, and an "out of the speaker" listening experience are simply more engaging and compelling. This philosophy extends from component level to the system's architecture; my system preference, for example, is very direct: DAC, amplifier and speakers with the digital processing running at 24-bits.
The quality of music is deeper and more personal when the system communicates more information more coherently, it's as simple (and as difficult) as that! Reproducing the full dynamic range of the original recording is fundamental, and we don't offer an editorialized version of the music. Instead I design so that you experience the language, as it were, of the original performance.