Halide Design S/PDIF Bridge USB-S/PDIF converter

The weekly Vote! Page is one of the most popular features of the Stereophile website, and the August 22 question, "Should Stereophile review more or fewer computer-audio products?", generated a record number of responses. No less than 88% of those responding asked for more coverage of products that allow a computer to be a legitimate source of music in a high-end context. Just 7% of readers wanted less coverage.

We have been trying to meet this demand for some time. I wrote a primer on the subject a few years ago (footnote 1), and more recently have been auditioning relatively inexpensive products that take audio data from the PC's humble USB 1.1 output port and transform it into a conventional biphase-encoded S/PDIF datastream to feed to a high-end D/A processor. I wrote about the Bel Canto USB Link 24/96 in May 2009, and the Stello U2 from April Music and the Lindemann USB-DDC 24/96 in May 2010. These three converters operate in what the USB specification describes as "adaptive isochronous mode," in which the host computer controls the flow of data from the USB port. A PC is not optimized for uninterrupted streaming; while the sample rate of the output data, adjusted every millisecond and averaged over a longish period, will indeed be the specified 44.1 or 48kHz, there will be short-term fluctuations, or jitter.

Depending on the design of the USB receiver and its clock generator, some of that jitter will make it through into the S/PDIF output—a wise old engineer once told me that you can never eliminate jitter, only low-pass–filter it—so these products need to be used with D/A processors that offer effective jitter rejection. However, it is also possible to operate the USB interface in what is called "asynchronous mode," which lets the DAC control the flow of data from the PC, clocking the output data with a constant-frequency, high-precision crystal oscillator. In theory, asynchronous USB operation (not to be confused with the asynchronous sample-rate conversion used in some D/A converters) reduces jitter to unmeasurable levels. However, only a very few products currently available—from Ayre Acoustics, dCS, and Wavelength Audio—have featured this mode, and these have used it as the front-end circuit for a full-function D/A processor.

The Halide Design S/PDIF Bridge ($450) is the first USB–S/PDIF converter of which I am aware that operates in asynchronous mode, and that does not require that a driver program be installed on the host PC.

From the outside, the Halide Bridge is about as utilitarian as it gets. A 6' USB cable is terminated with a 3"-long black aluminum tube half an inch in diameter, with either a 75 ohm BNC plug or an Eichmann Silver Bullet RCA plug on its other end. The complexity is on the inside. The Bridge gets its 5V power from the USB bus, but then low-pass–filters it with a Pi filter (a series inductor between two shunt capacitors) that has a corner frequency of 3kHz. The filtered voltage is then regulated to 3.3V for the digital electronics, and separately to 3V for the clock oscillators.

The USB datastream is fed to a Texas Instruments TAS1020B receiver chip, which converts the audio data to two-channel I2S format. The TAS1020 includes an embedded microprocessor that runs the proprietary Streamlength code licensed from Wavelength's Gordon Rankin. This code allows the Halide Bridge to be operated in asynchronous mode without the host computer having to run a proprietary driver program. Streamlength works with audio data having sample rates of 44.1, 48, 88.2, or 96kHz.

The output from the TAS1020B is converted to S/PDIF with a transceiver chip. To minimize any logic-induced jitter, the output from this device is clocked by the system's original master clock with a D-type flip-flop. This is coupled to the outside world with a small pulse transformer, which isolates the output ground from that of the PC. The Halide S/PDIF Bridge can be plugged into the coaxial S/PDIF input jack of any D/A processor.

I hooked up the BNC version of the Halide Bridge to the 2006 Mac mini that I use as a music server; the Mac's USB Prober utility revealed that the converter identified itself as "SPDIF Bridge" from "Halide Design" and, instead of the usual serial-number string, displayed the text "(C) 2010 Wavelength Audio Ltd." I plugged the Bridge into the coaxial input of my longtime reference Benchmark DAC1, plugged the balanced outputs of the DAC1 into the Simaudio Moon Evolution P-8 preamp, fired up iTunes and Pure Music in Memory Play mode on the Mac, and sat back to listen.

Hmm. The Benchmark sounded as I'd expected: a little dry and a little forward. But it was also easier on the ear than I'd anticipated. The asynchronous sample-rate conversion used by the Benchmark upstream from its DAC chips is very good at rejecting jitter. So in principle, using a source with lower levels of jitter shouldn't make much of a difference.

Except that it did. I could hear deep into the imaginative mix of 10cc's "Old Wild Men" (256kbps MP3 Amazon download, from Sheet Music, UK Records), and the contrast between the similarly pitched tenor voices of Eric Stewart, who sings the first verse, and Kevin Godley, who sings the second, was readily apparent despite the lossy coding of the Amazon download. The deliciously supportive acoustic of the Church of St. Catherine in Vilnius surrounding the solo baroque flute of Vytautas Sriubikis in his performance of Bach's Partita in A Minor (24/96 FLAC, converted to Apple Lossless with Max, free download from LessLoss) was superbly well defined—as was the distant traffic.

Footnote 1: A basic guide to the various strategies for getting the best sound from a computer can be found here. A discussion of audio file formats can be found here.
Halide Design/Devilsound Labs
(858) 224-3551

rashumon's picture

Hi There,
You mentioned in your review that only a few DACs use USB asynchronous mode and mentioned those that you thought did. but you forgot to mention the HRT MS product which also works in asynch USB mode and at a price very similar to the Halide design reviewed here.

It was reviewed by Stereophile in Nov 2009 with positive results...

Would be good to see your opinion of how these compare to each other...

stereodudeiii's picture

Suppose your USB port is 2.0 instead of the older 1.1 - makes any difference (I understand that from a purely computer perspective, USB 2.0 is [supposedly] backwards-compatible with 1.1 - the question is from an audiophile perspective)?

John Atkinson's picture
"the HRT MS product which also works in asynch USB mode and at a price very similar to the Halide design reviewed here. It was reviewed by Stereophile in Nov 2009 with positive results..."

The first-generation HRT Music Streamers did not operate in asynchronous mode. The Mk.II versions do, but were not available when I wrote this Halide review.

"Suppose your USB port is 2.0 instead of the older 1.1 - makes any difference?"

No. As you wrote, USB2.0 is indeed backward compatible with 1.1. But the common USB-serial audio chips available are based on USB1.1.

John Atkinson
Editor, Stereophile