Bel Canto REFStream Asynchronous Ethernet Renderer

Device Type: Network Player
Input: Ethernet
Output: AES/EBU, Coax S/PDIF (BNC), STFiber
Dimensions: 8.5” W x12.5” D x 3.5” H (216mm x 318mm x 88mm)
Weight: 14lbs (6.5kg)
Availability: Authorized Dealers
Price: $2,495.00
Website: www.belcantodesign.com

Rendered
Server, renderer, control point. Those are the ingredients of a UPnP/DLNA playback system. The server stores your music as well as the server software, the renderer is responsible for requesting your music from the server and passing it along to your attached DAC, while the control point is your interface to make this happen. Today, we're looking at and listening to the Bel Canto REFStream Asynchronous Ethernet Renderer to see how well it performs this deceptively simple task.

The REFStream supports FLAC, ALAC, AIFF, WAV, and DSD file formats in PCM resolutions up to 24-bit/192kHz. Meaning, while you can play back your DSD64 files, they will be converted by the REFStream to 24/176.4 PCM data before being sent to your DAC (religious warriors have at it). The REFStream supports gapless playback and the company recommends using JRiver Media Center and JRemote as the control point. Since the REFStream is UPnP/DNLA compatible, you can use any 3rd party control point you'd like if the idea of running JRiver on a computer strikes you as adding an unwanted layer of hardware into the network player picture.

Bel Canto has employed Ultra Low Noise Femtosecond Master Clocks, a Hybrid Distributed Power Supply which "ensures that the Ultra Low Noise Clocks perform to their potential with power supply isolation and local low noise regulation right where it is needed", and two-stage galvanic isolation. I think it's safe to say that Bel Canto has spent some time and energy dealing with noise and ensuring the digital data sent to your DAC is as clean and jitter free as possible.

The REFStream fits into Bel Canto's e.One series so it's housed in a full e.One chassis. You can opt for a silver or black aluminum faceplate which is adorned with the embossed Bel Canto logo and a two-color LED status light; Red = Standby, Green = Go. The unit's backside houses all of the ins and outs listed above plus an on/off switch and IEC inlet for the included power cord. The only unusual thing going on here is the inclusion of an STFiber output which fits very nicely with the Bel Canto DAC's STFiber input thus eliminating the possibility of any electrical noise being passed to the DAC.

Bel Canto cites "High performance asynchronous ultra-low phase noise audio renderer" as one attribute of the REFStream which is intriguing so I asked my contact at the company for more info. Here's the response from Jon Stronzer Bel Canto's owner and product designer:

"We have found that providing the lowest possible jitter audio stream into the DAC is critical to achieving the best sound quality. Our ultra-low noise clocks remove layers of digital grunge and low level noise that impinges on the musicality of the DAC. To achieve these low noise levels it is necessary to use free running oscillators that are designed for extremely low phase noise. We specify our master clock oscillators with noise floors that represent jitter levels on the order of 70 Femtoseconds or less when measured from 100Hz to 1Mhz. The only way to profit from oscillators with this kind of specification is to use an asynchronous transport mechanism where the local master clock is free running, and controls the incoming streamed data. This insures that the retimed data output has the lowest possible noise. We have also found that using the Ethernet interface results in a more transparent audio path, yielding better sonic performance than USB alternatives. I suspect that may be the result of the Ethernet interface not needing any audio software drivers, as opposed to the USB interface."
I find it interesting that the more people talk about dealing with noise, even with digital data or should we say especially with digital data, the resulting sound quality moves toward clarity, improved spatial representation, and overall a more natural sound. Let's see.

Clarity, Improved Spatial Representation, and Overall A More Natural Sound
Yup this stuff works. Of course there's no way for me to know exactly what's what, but in addition to delaing with noise, re-clocking the signal before handing the data off to the DAC sure seems smart to me. I connected the REFStream to the Auralic Vega via AES/EBU while the Vega was attached to my Pass INT-30A via XLRs. Speakers were my trusty DeVore Fidelity The Nines. Bel Canto recommends using JRiver Media Center [footnote 1] and JRemote to control the REFStream which is what I did, controlling playback mainly from the listening seat with JRemote running on my iPad.

I listened to this system for a while, weeks on end, before jumping into comparison mode. The reason being I was simply enjoying my music so much that I figured I'd see if this enjoyment would stick over time. And it did. With the REFStream there are a few qualities that stand out pretty much from the get go. These qualities include a wonderful sense of clarity as if the sound image is emerging from a very quiet and undisturbed place. Springing from this, music sounds very natural with great levels of detail, dynamics, and focus. It is stunningly easy to know where the various players that make up the whole resided in the recorded space which is portrayed in speaker defying largeness.

Think clean, clear, and precise while also getting a heavy helping of natural ease. By comparison my MacBook Pro connected to the Vega DAC via USB provided a fatter, sloppier sound. The overall sound image also condensed and shallow, making things muddier and more difficult to define in spatial terms. While the MacBook/Vega get a lot of things right, the REFStream/Vega gets more things righter and the difference is as clear as day.

A more relevant comparison comes in the form of the Simaudio Moon MiND (see review). I know this sounds too easy but it was—the MiND is certainly an improvement over the MacBook Pro but it doesn't equal the REFStream in terms of that wonderful sense of clarity and natural ease. Another aspect of the MiND as compared to the REFStream that had me searching for words was a sense of a tonal shift with the MiND where it seemed to emphasis a heftier sound. The same violin from Elvis Costello and The Brodsky Quartet's The Juliet Letters sounded darker and nearly a touch tonally lower as compared to the REFStream which sounded fleeter of foot and lighter of weight. Go figure. The Moon MiND comes with its own app which I find to be a plus since I don't need to run JRiver on a computer as is the case if you want to use it to control playback with the REFStream.

Next up was the Auralic Aries (see review) connected to my network via WiFi and to the Auralic Vega via USB since I prefer the Aries' USB output over AES. The Aries takes us a step closer to the REFStream and removes the weight I heard with the MiND but there are a few areas where I feel the REFStream betters the Aries; the overall size of the sound image, dynamics and natural ease. The Aries sounds a bit less refined than the REFStream and a bit softer. Violins don't have as much pluck and massed strings sound a bit more homogenized where the REFStream keeps its cool and maintains a great sense of differentiation and detail. You could say that the Bel Canto sounds more natural and I wouldn't argue with you one bit.

If I had access to a Bel Canto REFStream wish, list number one on that list with a bullet would be the addition of Roon. Our review of the Roon software is in the works but I will say that a Roon-enabled device gives it a head and shoulders and the clouds above advantage over any other control software I've seen. It is in a word freaking fantastic. I could go on but I won't. I also had a few JRiver crashes but I'd imagine that was an objection to all of the A/B'ing. Other than that, the Bel Canto delivers the sonic goods to your DAC of choice in time and in great musical shape.

Closer To Music
If you want to get closer to your music and you're using a MacBook Pro as source, or more than likely any off-the-shelf computer, the Bel Canto REFStream gets you there. Its way with music delivers a lovely sense of ease combined with great attention to detail all wrapped up in a nearly super-natural sense of space. Class A all the way.


Footnote 1. JRiver does not support Tidal which is really too bad but Bel Canto also recommends Bubble UPnP running on an Android tablet for those looking to control their music library and Tidal from within the same app and without the need for a computer.


Associated Equipment

Also in-use during the REFStream review: Auralic Aries, Simaudio Moon MiND

COMMENTS
DH's picture

Did you try it? How did you think it sounded?

audiofool's picture

Nice review of how JRiver sounds using Bel Canto vs Auralic streamer. Unfortunately JRiver is not the best streaming sound source. Can you come back at us with a Kinsky controlled NAS using Minimserver comparison please vs a review limited by JRiver.

Michael Lavorgna's picture
But it will take a few weeks.
Patatorz's picture

What makes you say that ?

BR

Patatorz's picture

Thanks for this review. I own the refstream after the Auralic Aries, totaldac d1 server, 3dlab nano etc...the refstream is definitely the best streamer I had in hands : more dynamic, more sound stage, music less fuzzy than the Nano and more precise on timbres than the Aries.

One comment, the refstream can perfectly stream DSF files to my DAC (Totaldac D1 dual) in DoP mode (DSF transcoding in minimstreamer), no conversion to PCM 176kHZ

Patatorz's picture

Hello. Just a confirmation from belcanto and my tests, native DSD64 are converted to PCM 24/176kHz.
The refstream can push DSD format to the DAC as soon as the DSD is encapsulated in PCM (DoP) and the DAC is able to decode DSD on AES for example.

Best regards

DavidL's picture

Any chance of a comparison of the Bel Canto REFStream renderer with the Sonore Signature Rendu or the original Rendu?

I did a comparison of renderers in January 2014 auditioning the Krell Connect, the Naim ND5 XS and the Cyrus Stream X2. I thought the Krell sounded the best, slightly better than the Naim with the Cyrus some way behind. I bought a Sonore Rendu unheard, purely on the basis of reviews, as I was deterred by the huge box and price of the Krell. The sound quality from the Rendu is excellent but I did not have the opportunity to compare it directly with the Krell Connect. Memory of sound quality is notoriously difficult but I tend to think the Krell Connect may have had the edge over the Rendu.

Your very positive review suggests to me that the sound quality of the REFStream is in the same class as that of the Sonore Signature Rendu reviewed on Computer Audiophile.

Any thoughts?

David