Digital Processor Reviews

I've been running a 5.1 system for years. Recently, I expanded it to 5.3 with the addition of two more subwoofers. This system is manageable with one of my eight-channel DACs.

Even more recently, I dipped my toe into Dolby Atmos, which made it necessary to lash up at least four more channels. That was a big problem, since neither JRiver nor Roon can support and sync more than one output device at a time, and the multichannel DACs I already owned tap out at eight channels.

I turned to the Arvus H2-4D Renderer, which offers 16 channels of digital output over AES3 and 16 channels of balanced-analog output. This worked, but piping everything through the Arvus meant forswearing DSP, including DiracLive. I really needed a DAC with at least 12 channels.

The Merging+Hapi has been around for years (footnote 1). It's one of Merging Technology's professional audio converters, intended for studio monitoring, at recording sites and in mastering studios that employ Merging's Pyramix software, which supports DSD recording and playback up to DSD256. The MkII is the latest generation.

Merging Technology has first-class audio performance in its DNA. Its founder, Claude Cellier, formerly worked for famous Swiss audio maker Nagra Kudelski, where he played a key role in the design and development of the Nagra IV-S and T-Audio TC analog tape recorders. One need only scan the Hapi MkII's specifications (see Sidebar 1) to be impressed with their depth of detail and impressive performance of this unit.

In its base configuration, the Hapi MkII supports eight channels of input or output via AES3, ADAT, or stereo S/PDIF (footnote 2). Expansion slots allow you to add up to two plugin cards, each of which can support eight channels of D/A, A/D, or two-way A/D/A.

The DA8P board (above) provides eight channels of D/A conversion, with balanced analog output—so, two such boards support 16 channels of D/A conversion. It is capable of converting PCM up to 384kHz and DSD up to DSD256, with extremely low specified noise and distortion and wide dynamic range. A bonus is the high output voltage, useful in a system like mine where the DAC drives the amplifiers directly.

The Hapi MkII seemed to be the logical next step for me. Support for more than eight channels meant I could, when called upon, play back music in Dolby Atmos—though in these auditions, the music maxed out at eight channels. I was attracted to the idea that I would be using it to review other equipment with recordings that were recorded and mastered on the Hapi or its big brother, the Horus. As they used to say in the TV commercials, "I just had to have it!" So I bought it.

Don't worry be Hapi
As befits a serious professional device, the physical appearance of the Hapi MkII is severe, clean, and neat. On the front left are two blue-illuminated buttons shaped like nested pyramids. The larger button is the power switch; the smaller accesses the setup menu. On the extreme right are 6.35mm and 3.5mm (¼" and 1/8") headphone jacks. Adjacent to these is a knob that serves as both a rotary control for navigating the menu settings and as volume control. To the left of the knob is an OLED used to display menu settings during setup and the volume levels of the various channels. It is all perfectly functional and strikingly similar to its sibling, the NADAC+8 (footnote 3), which, in contrast with the Hapi MkII, is intended for audiophile use. However, the Hapi MkII lacks the graciously styled chassis and superior control ballistics of the NADAC. The displays are similar, but the Hapi must squeeze bars representing 16 channels onto the same OLED screen that the NADAC uses for eight. The Hapi MkII also lacks a dedicated remote control. All those issues can be resolved by accessing the Hapi's webpage with a smartphone or tablet.

Except for the headphone jacks, all the connectors are on the back. At the far right is an AC power inlet and a jack for an auxiliary power supply, intended mainly for backup. At far left are two optical jacks for ADAT or S/PDIF input/output, two BNC connectors for Word Clock input/output, a DB-25 jack for AES3 input/ output—eight channels of each—a DB15 connector for various sync functions, and a pair of RJ45 Ethernet connections. In between are the two expansion slots; on my Hapi, they are occupied by two DA8P boards, populated by more DB-25 connectors serving up 16 channels of analog output—eight on each board. The pinout of these is industry-standard AES59 (Tascam Analog). I used Mogami Gold DB-25-to-XLR snakes to connect these analog outputs to my power amplifiers.

The data source used with the Hapi MkII must be RAVENNA/AES67-compatible (footnote 4). In my system, this is achieved by running a MAD ASIO driver package supplied by Merging on my PC. The Arvus H2-4D Renderer mentioned above is another example of a RAVENNA-compatible data source. Any Dante (footnote 5) device is also compatible. Merging makes it RAVENNA-capable. The RAVENNA/ AES67 Virtual Audio Device (VAD) serves the same function for Mac OS.

Data arrives at the Hapi by Ethernet. At first I plugged the Hapi to a local Ethernet switch, shared by my PC/server and LAN. This worked, but I experienced frequent signal interruptions and excessive buffering. Merging says that RAVENNA operations do not demand anything from the network beyond 1Gbps capability; they add, however, that RAVENNA hogs (my word, not theirs) bandwidth, so they recommend keeping it on a separate LAN, apart from other domestic or studio operations. I have a managed switch on order. Until it arrives, I will connect the Hapi directly to my PC's second Ethernet jack with a CAT6a cable, creating a private domain. This worked.

Physically connecting the Hapi and server to the LAN is one thing; making the logical connections is another. This task has become much easier since my previous Merging experiences, with the NADAC+8 and Anubis. Partly this is because I have learned from those experiences, but most of the improvement is due to the fact that the Merging ASIO Device—"MAD" for short—is a greatly enhanced, more stable version of the earlier RAVENNA/AES67 ASIO Driver, and the Aneman tool, which allows the user to map specific channel connections among multiple RAVENNA devices, is more coherent, consistent, and better documented than before.

Still, the Hapi MkII is not "Plug-and-Play." The Aneman tool shows all the available devices. The user drags the selected ones into an active zone, shown in the upper left of Aneman's "Matrix View" (see image). The available inputs are stacked on the lower left, the available outputs are arrayed across the upper right, and the matrix of their intersections is at the lower right. The blue squares show the mapping of my PC server's source channels to the Hapi's output channels. The complete freedom to route the channels is especially useful for active multiamping of multiple multiway speakers or, as in my case, for the independent EQ of several subwoofers in a multichannel system. You can even route some channels to the AES3 or S/PDIF digital outputs rather than through the DAC, in case you're using active loudspeakers.

All this may seem intimidating, but it is manageable with a bit of attention and effort, and the setup is flexible and stable. If you need to reboot the server or the Hapi, this setup is preserved.

Up and running
Now that I set it up, as a 16-channel DAC with channels mapped, the Hapi serves as a 16-channel DAC—nothing more—so I have little direct interaction with it. JRiver or Roon serves as the user interface, just as before. Choose the album or track on the server and it plays without hiccups. Transitions between PCM and DSD are seamless. Once it's set up, it's easy to forget about the Hapi MkII.

That same, self-effacing character extends to the sound.

Let's begin with a recent recording of Robert Schumann's Kreisleriana, Op.16, and Humoreske, Op.20, performed by Jimin Oh-Havenith on an album bearing the title inSANE (24/96 5.1 FLAC download from Audite). The title probably alludes to the various mental disorders Schumann was diagnosed with during his short life, but no explanation is offered in the booklet. These are beautiful performances, presented in a close-up perspective but with an open acoustic ambiance. I was struck by the warmth, weight, and richness of the piano. Pictures in the booklet confirmed my suspicion that it was a Bösendorfer, recorded with a microphone array tightly clustered around the instrument. The recording venue was Konzerthaus Liebfrauen, a 500-seat concert hall converted from a Baroque church built in the middle of the 18th century, in Wernigerode, which is midway between Hanover and Leipzig in central Germany.

You can never know with certainty whether the playback of a recording is accurate unless you were present at the performance and possess eidetic acoustic memory, but via the Hapi, what I hear is a Bösendorfer right there where my center speaker sits and, by gum, it sounds as if I am hearing it in the hall I see in pictures.

---

Footnote 1: The name Hapi was originally used by legendary audio designer Bill Hegeman for several versions of his preamp.

Footnote 2: The base configuration would be adequate for users with eight or fewer channels and active, digital-input loudspeakers, such as the Kii Three, Dutch & Dutch 8C, KEF LS50 Wireless II or LS60 Wireless, or the recently reviewed JBL 4329P.

Footnote 3: NADAC is short for network-attached DAC; see Jon Iverson's NADAC report and my take here.

Footnote 4: A bit of pro-audio nomenclature: To interface with data sources, Hapi uses RAVENNA/AES67, an open-source audio-over-internet-protocol standard developed for the broadcast industry. RAVENNA is bidirectional and supports multicasting and proper synchronization: All devices on the network derive their own, locally generated media clocks from the network clock distributed with Precision Time Protocol (aka PTP).

Footnote 6: Dante is another audio-over-IP technology that is widely used in the pro-audio world.