This review page is supported in part by the sponsors whose ad banners are displayed below

With reduced wallet cramps, the Metrum Hex joins the ranks of posh NOS converters from 47lab, Zanden, Abbingdon Music Research, CAD, Audio Note UK, Vertex AQ and Concert Fidelity. Here it goes deliberately retro against machines like AURALiC's Vega which upsamples all incoming data to 1.5MHz/32-bit. Modern Sigma-Delta converters can exploit so-called apodizing filters to eliminate the pre-ringing otherwise endemic to their approach. NOS DACs don't suffer that behavior in the first place. If they omit digital filters, they do however retaliate with aliasing noise above half the sampling frequency (22.05kHz for Redbook).

Newly hatched DAC inside incubator: "First-stage test during the production process. Here we measure all the ins and outs before the board gets installed inside the chassis. This is called a 'fixture' which pushes sharp pins into the backside of the board to connect it without any wiring. It allows us to completely measure the board in one minute."

To paraphrase Russell Crowe's master and commander, there's much debate over which of the two is the lesser weevil, ringing or ultrasonic distortion. Here one mustn't forget recorded pre-ringing from anti-aliasing filters employed during mastering—which NOS DACs of course can't remove though they don't add to it either—nor this approach's slight treble roll-off. In the end we'll leave the theoretical pros and cons to the engineers both armchair and real and get on with shamelessly subjective listening impressions.

But first a few final spells on how to work this hex: "For USB we use a modified OEM M2Tech hiFace 1 for a few reasons.
In the past we had several complains of matching an XMOS device to either Mac or PC. From our point of view it thus was too early to implement such a module in our machine. On several occasions we asked people to return their USB interface and Octave DAC only to find that both were running satisfactorily on our own Mac and Windows systems. I am certain that the XMOS device wasn't to blame but rather setup problems on specific computers since I couldn't replicate these issues to establish true cause from the distance. That said, complaints about the original hiFace on either Windows or Mac machines after installing the proper hiFace driver have been exactly nil.

"But using this OEM module as a quasi hiFace wasn't our goal.
Marco Manunta relies on USB power. This can readily cause all kinds of issues by injecting HF noise into the DAC or forming ground loops which can pick up airborne frequencies from domestic equipment. So we disconnect USB power and feed the OEM module with our 15VA linear power supply which also has complete galvanic isolation from other parts of the Hex. As the data output from the OEM module is useless to the actual data format of our DAC, we add our glue-logic circuit for data 'translation' and buffering which fits below the USB module on the distribution board.

Modified hiFace module with Xilinx FPGA and dual clocks

"Before settling on this particular OEM module over competitors we also verified its digital transparency by feeding it well-known bit patterns from our analyzer. By measuring this pattern once it exited the OEM module we could ascertain that it indeed operated bit perfect. This became another reason to use it. The USB module is optional so people can use their own USB bridge if preferred. From a marketing perspective there's another reason to make it optional. We see two types of customers who might purchase our Hex DAC.

Music-loving hobbyists. They like to set up a computer as music server and spend much time tweaking it. They need a USB input to run their system. We think this audience makes up about 5%-10% of our total market.
2/ Music lovers. They spend money but aren't interested in configuring a system. This type of customer purchases a DAC to be plug & play. For them the S/PDIF connection is the only action required to make a system run. This is 90%-95% of our market. Here we don't need USB. And that also makes the Hex more affordable.

Mono DAC board with eight paralleled converter chips

"With the Hex we run two discrete converter boards in a master/slave configuration. The master board receives the data from the distribution board at the back of the unit and splits it into left/right-channel data. The right data is fed to the master board, the left data to the slave board. This creates risk for ground loops so we insert a few extremely fast optical isolators for full galvanic separation. Incidentally these particular devices are often used in data transmission systems where they might run 50MHz bandwidth.  

"Both DAC boards are programmable to perform several tasks:
1 - Left mono single ended
2 - Right mono single ended
3 - Left mono balanced
4 - Right mono balanced
5 - Stereo single ended

"This flexibility allows use of these boards in many applications and OEM products. Compared to the Octave we improved the max sampling rate to 192kHz. In applications like the Hex where the boards are programmed for mono balanced operation, no direct single-ended output is available.

"Enter the Lundahl LL1527XL output transformers. They create the required single-ended outputs. We selected this affordable transformer because many people already seemed bothered by our price tag for the Hex which obviously had to exceed that of the Octave. Especially in the USA the use of single-ended outputs appears less popular. There XLR becomes more important and people wouldn't want to pay extra for a more upscale output transformer they'd not use. For those who want the very best performance from our fully balanced DAC's SE outputs however, we recommend the upgraded Lundahl LL1588 parts. Hence those became our third à-la-carte option after USB and AES/EBU. That's also the transformers your review unit will have."

John Darko standing by in Australia for a second opinion had another question: "My experience with the Octave aligned with many others: that an Audiophilleo or similar USB bridge was essential because the direct optical connection from a Mac/PC was relatively weak. Is jitter/noise rejection on the Hex S/PDIFs superior to the Octave or the same design?"

About the direct optical or S/PDIF connection from a PC, I agree that this kind of output is worse than good CD or media players. It is like a Ferrari on a sandy road. There are limits to how much jitter our DAC can handle or we would have had to put in more data-buffering memory. As you probably know data handling in a PC is more or less determined by the processor. There are data-stream gaps at the S/PDIF output. It all depends on other computing actions where S/PDIF never has the highest priority. That said, the Hex jitter rejection is a bit better and <35psec while the Octave was less than 40psec.

"When you compare findings, keep in mind that your RCA outputs aren't the same. Your unit has LL1527XL transformers, Srajan's LL1588 with +10dB more headroom and lower distortion. As one of my distributors put it, the 1527 is more romantic and warm—it is also used in the Linn Klimax—the 1588 more real world and closer to the XLR outputs."