Bricasti M1: Follow-Up, March 2012
In my coverage of Bricasti Design's $7795 M1 D/A converter in the August 2011 issue (the M1 also graced the cover of the February 2012 issue, which contained John Atkinson's impressions of it), I wrote:
"Violinist Arturo Delmoni's debut recording, Songs My Mother Taught Me, on the aluminum-CD edition remastered by Bob Ludwig (John Marks JMR 1), has never sounded bettersweet, yet realistically resiny. This was one of the albums I relied on as I scrolled through the M1's digital-filter options. I thought that Filter 0 (the off-the-shelf choice) was forgettableflat and uninvolving. It didn't take me long to decide that my favorite was Filter 4except in the rare case when a recording was too close-in, or tended toward a brittle sound. Then I preferred Filter 6, because it sounded more as if I were sitting in the audience, less as if I were onstage."
The meta-opinion I did not then express was that the only purpose I could imagine for including the "flat and uninvolving" Filter 0 was to prove beyond doubt that Bricasti's proprietary DSP engineering, as embodied in the other six filters, did in fact constitute better sound than the stock Analog Devices AD1955 digital chip. In other words, I couldn't imagine anyone's listening to Filter 0 for fun.
If what I surmised was indeed the thinking behind Bricasti's choice to include, as a credibility enhancer, access to the stock filter, which "comes free with the digital chip anyway," their thinking appears to have been not too far off the mark. Not long after the August issue was published, I received an e-mail from a reader suggesting that I evaluate and write about a certain $399 D/A processor from Manufacturer X, the DAC chip in that product was also based on the AD1955.
Um, sorry, no. I don't believe that it's possible to get a Pareto-optimal slice of the Bricasti M1's sound simply by shopping for a DAC based on the same OEM chip, which sounded flat and uninvolving on its ownunless, of course, that other DAC has the same kind of jitter measurements, proprietary digital filters, and blindingly fast dual-mono analog output stages, none of which is economically feasible in the budget-price range.
Be that as it may, the feedback that Bricasti received from its professional users carried two clear messages: No one liked Filter 0, and furthermore, the transition between Filters 4 and 6 was too abruptpeople wanted intermediate steps. (Somewhat confusingly, in the original filter set, Filters 4 and 6, which are based on the same algorithm but with different passbandsand different degrees of passband ripplewere separated by Filter 5, which was dissimilar, as described by JA in his review.)
In response, Bricasti has revised the M1's filters set and made the update available free for all registered owners. Using his laptop computer, Bricasti cofounder Brian Zolner updated the firmware of Stereophile's review sample at the annual holiday meeting of the Connecticut Audio Society last December.
Old Filter 0 (the stock AD1955 filter) gets the old heave-ho. Between old Filter 4's 20kHz passband and old Filter 6's 18kHz passband are now three new intermediate steps, at 19.5, 19, and 18.5kHz, these five similar filters now arranged as new Filters 15. So instead of seven filters, the M1 now has nine.
For a test recording I used a 48kHz remastering of the Ray Brown Trio's Soular Energy (DVD-Audio, Concord Jazz CHRM 2011), then rechecked with David Gray's White Ladder (CD, RCA 69351-2), both played from a Meridian Sooloos Control:15 music server. I did this because Zolner pointed out that new Filter 8 (old Filter 3) is optimized for 48kHz. I ran through all the new filters with both albums.
I expected that I would warm up to new Filter 3, in that it split the difference between old Filters 4 (my favorite) and 6 (JA's favorite). I liked it better than I did old 6, but still felt it a bit too polite. Of the new Filters 15, the one that gave the most satisfying amount of body and "bite" to the sound was new 1 (old 4).
But to keep all of this in context: My reactions are informed not only by my own musical tastes, but also by my room and the equipment I'm using just now: Vivid Audio K-1 loudspeakers, Parasound Halo JC 2 preamplifier and JC 1 monoblocks, and Cardas Audio Clear cables. In no case am I talking about night-and-day differences. As I said in August, there were times I preferred old Filter 6 of the set then on offer.
Apart from the similar Filters 15, the other new filters struck me as "good," but just didn't click into focus the way new Filter 1 (old 4) didwith the possible exception of new Filter 6 (old 5), which might be called the old-time rock'n'roll filter. It seemed not only a bit more laid-back than the others, it also seemed to have a bigger bottom end and, mysteriously (I may have imagined this), to impart a little bit of the sought-after benign optical/analog dynamic compression that gave golden-era rock recordings their sound.
At no time did the Bricasti M1 sound like anything other than itself. Trying to isolate and express in words the differences among these digital filters was not unlike trying to describe the differences among nine different brands of Dragonwell green tea; eg, at 48kHz I felt that new Filter 8 (old 3) sounded microscopically better than new Filter 0 (old 1). Microscopically.
Although Bricasti Design is to be applauded for making so many filters available, and for providing the new filter set free of charge, my own opinion is that what makes the M1 sound the way it does is neither the selectable high-resolution filter algorithms nor its phenomenal (and, for all I know, unmeasurable) jitter performance. I think the crux of the matter is the engineering that went into the analog output stages. I think what we are really hearing are things like power supplies, grounding and isolation and buffering schemes, and slew rate. I also think that, at least for the time being, Bricasti's M1 is the DAC to beat.John Marks