Digital Processor Reviews

I used the D33 while finalizing my choices for this issue's "Records 2 Die 4" feature. With the default Burr-Brown digital filter selected, the D33's sound was clean and clear, with excellent low-frequency definition and extension. But after a while, it became clear that the high frequencies had a little bit of "bite." The rosiny edge of the sound of Jacqueline du Pré's cello in Elgar's Cello Concerto, with Sir John Barbirolli and the London Symphony (24/96 Apple Lossless files from HDtracks, transcoded from FLACs), was slightly emphasized, and the soundstage was a little curtailed overall. While bass guitar had good weight and definition, well-recorded rock recordings were a bit too in-your-face. Jenny Scheinman's vibratoless violin on Bill Frisell's All We Are Saying . . . (CD, Savoy Jazz SVY17836) was harder to take than I remembered from listening to the track as decoded by the Devialet D-Premier D/A amplifier, which I reviewed last month.

It was obviously time to check out the D33's custom filters.

The FMJ D33's Filter 1 is described by Arcam as a "minimum-phase filter with a fast roll-off [that seeks] to remove the pre-ringing on transients found in normal digital filters. . . . [This] redistributes the energy from any pre-ringing until after the transient. Although this causes greater post-ringing than the standard digital filter, it results in a much more natural sound as there is no pre-ringing in live music."

And that was what I found when I compared the sound of Filter 1 with the Burr-Brown filter. The high frequencies became better integrated with the midrange and bass. The relationships between individual acoustic objects in the soundstage and the surrounding ambience became more evident, there being a better presentation of stage depth. Against these improvements, Filter 1 did sound more laid-back overall, which some listeners might feel detracts from the D33's sense of pace compared with the sharpness of the Burr-Brown's character. However, I did the bulk of my auditioning using Filter 1, because it reproduced my own recordings as I had intended them to sound.

For example, in 1998 I had recorded Stereophile's first jazz CD, Rendezvous (Stereophile STPH013-2), in the church acoustic of Chad Kassem's Blue Heaven Studios in Salina, Kansas. Although I used fairly distant miking on Billy Drummond's drums, the unequal loudness of the instruments—the main continuo instrument is not a piano but Steve Nelson's vibes—meant that I had to close-mike the vibes and take a direct feed from the pickup on Jerome Harris's acoustic bass guitar. Reconciling the different perspectives on the instruments in the mix was difficult (see ), but with the D33's Filter 1, the mix successfully jelled—my artifice didn't disturb the musicians' art. By contrast, the Burr-Brown filter focused too much on the sonic disparities.

Arcam describes Filter 2 as having a much slower rolloff than Filter 1, with no pre-ringing and minimal post-ringing. Filter 2 "would be the preferred listening position," according to Arcam's literature, and it appears that Filter 2 is much like the old Wadia Digimaster filter, trading off ultrasonic image rejection against improved time-domain performance. I listened again to the Elgar and Frisell recordings; switching to Filter 2 from Filter 1 endowed each of the instrumental images with more solidity, but also made the sound a little more robust, with less image depth.

For example, last November I recorded tenor sax player Ras Moshe duetting with double-bassist Shayna Dulberger, both virtuosi (footnote 1) at Bushwick's Goodbye Blue Monday club. I had put up an ORTF pair of DPA cardioids in front of the tiny stage, pressed Record on the Zoom H4n SD card recorder, and hoped for the best. When I listened the next day, as I finished writing this review, to the 24/96 WAV files decoded by the Arcam FMJ D33, Filter 1 was better at presenting the images of the musicians within the club's dry acoustic, and Filter 2 at presenting the leading edges of the instruments' sounds. But with either filter the D33 was a successful time machine, opening on the previous evening's superb music-making a rectangular, transparent, grain-free window.

I then changed to the D/A section of the Classé CP-800 preamplifier, which also has an asynchronous USB input. Our "Recording of the Month" for July 2003, Rachel Podger and Arte Dei Suonatori's recording of Vivaldi's La Stravaganza (24/96 Apple Lossless files transcoded from Linn Records FLACs; SACD/CD, Channel Classics CCS SA 19503), was reproduced with a less rich balance. With the CP-800, the focus was more on the higher frequencies, with a wealth of recorded detail but a thinner sound overall. With Rendezvous, the Classé sounded more delicate overall, with a less robust character to Jerome Harris's bass.

Comparisons
It's difficult to review a DAC without reference to its competition. The obvious candidate for comparison with the Arcam FMJ D33 is the similarly priced Bel Canto e.One 3.5VB ($3495), which Erick Lichte reviewed in June 2011 and January 2012, and which uses the same DAC chip as the D33. Unfortunately, our review sample had long since left my listening room, so I turned to the inexpensive Musical Fidelity M1DAC D/A processor ($749), which I reviewed in January 2013. This uses a Burr-Brown PCM1796 DAC chip, which in theory has about 6dB less dynamic range than the D33's PCM1792, according to the datasheets, but which I found had almost state-of-the-art measured resolution. All of the following comments were made with the levels of the two processors matched within 0.1dB at 1kHz.

Switching between the Musical Fidelity and the Arcam set to the Burr-Brown default filter, it was extraordinarily difficult to hear any differences between them—the Musical Fidelity is an excellent value at its price. But when I selected the Arcam's Filter 1, the M1DAC couldn't compete in the resolution of acoustic objects on the soundstage, sounding flatter and brighter.

The 24-bit remasterings on USB of some of the earlier Beatles albums sound rather hard. Beatles for Sale, for example, benefited from the Arcam's slightly softer sound, though Paul McCartney's walking bass line in "Eight Days a Week" had a tad better definition through the Musical Fidelity.

Following Wes Phillips's rave review of the Logitech Transporter in February 2007, I had purchased a sample to use for Internet radio and audio streaming and as my backup DAC. With the Transporter's volume control set to its maximum, the levels matched at 1kHz, and the D33 using my preferred Filter 1, the Logitech sounded less laid-back than the Arcam, a little harder and brighter overall. In fact, the Transporter sounded very similar to the Musical Fidelity M1DAC. The Arcam's minimum-phase filter puts it in a different class from the Transporter, which cost $2500 when last available.

The final comparisons were with my current standard for D/A conversion, the dCS Debussy, which Michael Fremer reviewed in January 2011. At $11,499, the Debussy is far more expensive than the Arcam FMJ D33, and also offers the choice of anti-imaging filters, including a minimum-phase, apodizing type very similar to the Arcam's Filter 1.

Some have commented on the Debussy's sounding too smooth, and it did indeed sound smoother than the Arcam. In fact, with levels matched at 1kHz, the more expensive processor sounded a touch quieter. But its smoothness meant that "Eight Days a Week" lost the last trace of hardness through the Debussy without the music's vibrancy being diminished. On the other side of the coin, the D33 reproduced the Smashing Pumpkins' "Bullet with Butterfly Wings" (256kpbs Amazon download) with a slightly greater sense of urgency. With the Debussy, the interplay between Ras Moshe and Shayna Dulberger was reproduced with the individual instruments slightly better differentiated in space. However, Dulberger's double bass lost some its body through the more expensive processor, as did Ray Brown's bass in "Exactly Like You," from the 24/192 DVD-A rip of Soular Energy. To draw a wine analogy: If the dCS is akin to a sophisticated Bordeaux, the Arcam is more like an excellent Côtes du Rhone.

Summing up
Other than its AES/EBU input's fussiness over what sources and sample rates it would accept, I enjoyed my time with Arcam's FMJ D33. Its high-speed USB2.0 input was a boon to musical enjoyment, and its three choices of digital filter make it possible to vary its sound to suit the needs of the system in which it is used and the tastes of its owner.

Assuming that that slight channel imbalance is sample specific, I can confidently recommend the Arcam FMJ D33. It offers a taste of what is possible from cost-no-object D/A processors at a competitive price.

---

Footnote 1: Listening to Dulberger play, I was reminded that the difference between a virtuoso and a journeyman bass player such as I is not that I can't play what she plays, but that I can't even conceive of what she plays.