If you spend any time at all on the internet audio forums, you know that there is a vocal contingent which maintains that all DACs sound the same. These are the same people who, if you disagree with them, demand that you provide proof of your position in the form of double-blind ABX tests, which is ironic, because, except in the rarest case, they don’t have any double-blind ABX tests to offer up in support of their position either. Their experience is also generally limited to inexpensive stand-alone DACs and the DACs built into mass-market audio video receivers. Don’t get me wrong. I think double-blind ABX tests are a great idea, but they are impractical, which is why everyone talks about them, but nobody does them. Except Harman; but they do not use them to establish difference, they use them to establish preference, acknowledging that differences exist in the products under test. They want to know which versions of potential products are preferred, so that they have the best chance of selling as many of them as possible, which is a perfectly rational thing to do if you want to stay in business, provided you have the financial wherewithal to build a dedicated double-blind listening test facility and the necessary associated equipment, like Harman’s million dollar speaker shuffler.
I don’t think double-blind ABX tests are necessary to prove that DACs can sound different. I have heard the differences with my own ears. However, if you want to dismiss my conclusions, for example, that the Wadia 121 and the Benchmark DAC1 HDR sound different than the DACs built into my Meridian G68ADV surround sound processor, or that the Wadia 121 sounds different than the Furutech ADL GT40 and the Halide Design DAC HD, based on expectation bias or other factors, that’s okay. Had I not heard the differences myself, I might even be inclined to agree with you. After all, bits are bits. Right? On the other hand, there are probably dozens of factors that could explain why some DACs sound different than others – different power supplies, different circuit topologies and layouts, different quality of components, different DAC chips, different amounts of jitter, different analog output stages – the list goes on. Among the most important of them is different digital reconstruction filters.
Different digital reconstruction filters do sound different. I found that out when I reviewed the Cambridge Audio Azur 751BD Blu-ray player. It has a button on the front which cycles through three different digital reconstruction filters, one minimum phase and two linear phase. The sound changes as you cycle through them – changes which are measurable if you have the equipment to do so. The differences are not dramatic, night and day differences, as some reviewers would have you believe. They are subtle, but clearly audible. Like Harman, you can decide which you prefer, which you believe gets you closer to the sound of live acoustic instruments, if any. Which brings me to the subject of this review – the Bricasti M1 DAC – which offers 15 different digital reconstruction filters. But I am getting ahead of myself.
I was not familiar with the Bricasti M1 until John Marks, who was intimately involved with the Ralston Listening Library and Archive project, brought it to Tam Carlson’s attention, who in turn brought it to my attention. Tam arranged for an evaluation unit for the Ralston listening facility, and I have had the good fortune to listen to it on any number of occasions in the last few months. Since Bricasti has made a big splash at the last couple of trade shows, including most recently at RMAF 2012, I thought I would share with you my listening impressions.
Like so many other components relating to the playback of computer audio files that have made their way onto audiophiles’ radar screens, Bricasti’s roots are in the professional audio arena. In addition to the M1 USB DAC (which started life without a USB input), it also manufactures the well-regarded M7 and M7M reverb units, and accompanying M10 remote. Bricasti also has a shared history with Lexicon and Mark Levinson. Co-founders Casey Dowdell and Brian Zolner were at Lexicon prior to Harman’s shuttering of its New England facilities. While Bricasti writes all of its own digital signal processing software, it subs out some of the hardware design to AeVee Labs, a company founded by Bob Gorry, former Chief Engineer at Madrigal Audio Labs. So even though it is a new company, Bricasti and its principals have a storied legacy.
Simplicity is the watchword in the design of the Bricasti M1. To paraphase Einstein, a design should be as simple as possible, but no simpler. The front panel contains a power (standby) button, buttons for selecting Input, Filter, Status, Level, Display and Enter, a rotary control for scrolling through selections, and a large, easy-to-read display. Input selects the digital input. Status has five levels which show input, sample rate, digital over errors, internal temperature, and phase inversion. Display adjusts to three levels of front panel brightness, and sets sleep or off mode. Filter will take you filter selection, which is done using the rotary control. Once selected, you press Enter to engage your selection. Level, as the name suggest, toggles between fixed and variable output. The Bricasti M1 can drive a power amplifier directly, eliminating the need for a preamplifier in an all digital source system. Around back, it has a combination power switch, IEC power inlet and fuse, a TRS jack for remote on/off trigger, four transformer isolated digital inputs for AES/EBU, coaxial S/PDIF, Toslink and asynchronous USB (which is galvanically isolated from the computer’s power supply) at up to 192/24, and both balanced and unbalanced analog outputs. Fit and finish are beyond reproach.
Inside things get a bit more complicated. There are four sections. In the center is the digital input section, which has its own digital switching power supply. This section contains an Analog Devices Sharc DSP that controls front panel operation, synchronizes the separate DDS clocks on each channel, and provides the 15 oversampled digital reconstruction filters, six of which are minimum phase (apodising) and nine of which are linear phase. The advantage to minimum phase filters is of course the absence of pre-ringing. Post-ringing is greater than with linear phase designs, but post-ringing is masked by the impulse itself. The analog output sections are a true dual mono design, with both the left and right channels having their own dedicated double-regulated linear power supply. The DAC chips are 24-bit delta sigma 8x oversampling Analog Devices 1955s, each coupled to its own clock connected by millimeter-length traces to minimize jitter. The analog output stage is a fully differential design using high slew rate analog op-amps. There are two discreet output buffer sections driving the balanced and unbalanced analog outs. Output gain is adjustable by setscrews from the back panel. The fourth section is the front panel controls and display. Circuit boards are made from an Arlon substrate with excellent high frequency impedance characteristics.
I first wanted to get a handle on how filter selection affects the sound. For the purposes of the review, I limited my examination to the six minimum phase filters which are labeled Minimum 0-5. I chose Mozart’s Piano Trio in Bb major, K. 254 performed by the Beaux Arts Trio on a Philips recording because you can tell a great deal about how a DAC performs by listening to both string tone, in this case violin and cello, and piano. I started with Minimum 0. In this recording, the piano is placed center rear, with the violin center left and the cello center right. The sound came from between the speakers rather than beyond them, though the soundstage was quite deep. The piano and cello were perfectly stable; the violin (and presumably the violinist) wandered a bit between center left and center. The tone of the piano was resonant rather than percussive. You could hear the wood of the soundboard. Leading edges were slightly dulled, which gave it a relaxed sound. String tone was terrific and, like the piano, was on the relaxed side. There was a sheen on the strings, but no glare. Were this the only filter selection, I would have been quite satisfied.
I next edged the filter up to Minimum 1 and 2. Doing so incrementally foreshortened the sound perspective. It was as if I had moved a few rows forward. The soundstage flattened slightly and also became marginally wider. This had the effect of bringing the piano forward. From a tonal standpoint, the presence region was elevated, making it somewhat brighter and more detailed, for both the strings and piano. The piano became more lively, changing the balance between the strings and piano. I don’t want to overstate the differences. The were subtle but clearly audible. Moving to Minimum 3 and 4 rendered similar changes, presenting a more dynamic soundscape. The changes between Minimum 4 and 5 followed suit.
These changes we not limited to the Mozart piano trio. I confirmed both tone and dynamics changes with Bach’s Cello Suite No. 1 in G performed by Yo-Yo Ma and Beethoven’s Piano Sonata No. 29 in Bb “Hammerlavier” performed by Richard Goode. At Minimum 0, Yo- Yo Ma’s cello sounded woodier and more resonant but less engaging. Dialing it up two notches to Minimum 2 yielded an agreeable increase in dynamic contrast that made the third movement Courante literally bounce along. I edged forward in my seat in anticipation. Moving to Minimum 4 was too much, as string tone became hard and the perspective flattened. But it was exciting. Similarly, at Minimum 0 Richard Goode’s piano was rich and reverberant, but a little on the subdued side. Moving to Minimum 2 brought you closer in, added pace, and increased the upper frequencies’ ringing quality somewhat, which was a welcome addition, in particular in the Scherzo movement. Moving to Minimum 4 had the right hand notes become too sharp and percussive, giving them a tinkly sound.
I also listened to Shostakovich’s Symphony No. 5 in D performed by the Leningrad Philharmonic under the baton of Evgeny Mravinsky to get a sense of how filter selection affected orchestral music. In Part 1 of the Moderato, at Minimum 0, the orchestra seemed a bit distant, the slower sections were somewhat ponderous, and where there was tension, it was slight. Bumping it up to Minimum 2 brought more life to the performance. Even the audience noises sounded more real, with coughs and shuffles having you turn your head. However, it wasn’t until I selected Minimum 4, that the symphony took on a dramatic tone, especially in the Largo sections. The solo winds were especially well reproduced, the horns had bite, and the percussion was right, all without the strings becoming zingy.
At this point you may be wondering which filter was the more accurate. I have no idea. I wasn’t present at the performances. Even if I were, my judgment would be affected by my location in the hall. In addition, as I explained in my essay The Sound of One Hand Clapping, there is no way to tell what is on the master tapes. Master tapes are chameleons, sounding like whatever system they are played back on. Fidelity to the master tapes, in the subjective sense, is a conceit. I can, however, tell you that my preference for the Mozart was the Minimum 0 filter, for the Bach and Beethoven, the Minimum 2 filter, and for the Shostakovich, the Minimum 4 filter. In each case, those filter selections gave me the soundstage and tone which most closely corresponded to what, in my mind, resembled actual performances. If I were limited to one filter, it would be Minimum 2. Of course, there are nine linear phase filters to try as well.
The Bricasti M1 USB DAC is a state-of-the-art design offering impeccable performance and a wide selection of digital reconstruction filters so that you can fine tune its performance to both your preference and the personality of your system. At $8495, while not inexpensive, it costs far less than the top-of-the-line offerings from many other well-known manufacturers. In that sense, it can be considered a bargain. I hope it becomes a permanent fixture in the Ralston Listening Room.
Update: Power Supply Upgrade
Bricasti has just released a linear power supply upgrade for the logic section of the Bricasti M1 DAC for the astonishing low price of $150.00. The previous power supply was of the switch-mode variety. Now all three power supplies in the M1 are linear.
The upgrade is relatively simple to install. First, you remove a dozen or so screws holding to top cover in place. Then, you disconnect the two wires connecting the existing power supply in the center of the chassis to the AC inlet on the rear panel and route them under the ribbon connector. Then you unscrew two cable clamps holding the distribution wires to the two existing linear power supplies in the front left and front right of the chassis, and unplug the cables. You will also need to remove the multi-pin connector between the power supply and the logic board. Next, unscrew the switch-mode power supply and distribution circuit board from the chassis and remove them. Finally, install the new linear power supply and reconnect the cables. Screw the top cover back on and you are good to go. It is a 5-10 minute process. Shown below is the interior of the M1 with the new power supply. Compare it to the photo above.
What can you expect to hear? I listened briefly to several CDs to get a feel for the new power supply’s impact on the sound. I began with the first movement of Tchaikovsky’s Symphony No. 5 on a DGG recording featuring Herbert von Karajan conducting the Berlin Philharmonic. The opening woodwinds had a more natural timbre and reedy quality than before. In addition, there was a greater three dimensionality to the instruments than previously portrayed. Dynamics remained superb. Moving to a collection of Verdi arias performed by Maria Callas on the EMI label, I immediately noticed a startling improvement in the representation of the cello to the right rear of Ms. Callas. The pluck of the C string followed by the strum of the G-D-A strings in a 1-2-3 rhythm was both better defined and more realistic than before. The cello occupied a distinct location in space whereas before it was only generally localized. Moreover, the slight hardness of Ms. Callas’s voice at the top of her range was ameliorated. String tone in Mozart’s String Quintet No. 4 in G minor, K. 516, as performed by the Smetana Quartet with Joseph Suk on viola on a Denon recording, was improved with the greater sense of natural sheen to the strings. Again, greater separation and three dimensionality of the instruments was evident. An incomparable DAC has just been made better. Who knew a power supply could have such a marvelous impact.
For details on the power supply upgrade, contact the DAC’s designer, Brian Zolner, at email@example.com
- Frank Berryman
Bricasti Design Ltd.
123 Fells Ave
Medford, MA USA 02155
Telephone: (781) 306-0420
The Ralston Listening Library and Archive:
Analog Sources: Ayre/Bauer DPS; VPI Classic 2, Koetsu Coralstone, Miyajima Shilabe and Mono BE; Ayre P-5xe
Digital Sources: Ayre DX-5; MacBook Pro; Weiss INT202; Bricasti M1, Sony XA-5400ES
Preamplifier: Ayre KX-R
Power Amplifier: Ayre MX-R
Loudspeakers: Wilson Alexandria XLF
Cables: Ayre Signature Series; Cardas Clear
Accessories: Grand Prix Audio Silverstone isolation system; EquiTech 10WQ balanced power system; Ayre L-5xe power conditioners