This review page is supported in part by the sponsors whose ad is displayed above

Reviewer: Stephen Marsh
Digital Source: Vecteur D-2 CD Transport, Audio Note DAC Kit 1.2 with upgrades (PS choke, tantalum resistors, Black Gate caps, copper grounding bars on digital chips wired to central ground, VTV silver foil/oil output coupling caps), Toshiba laptop with Empirical Audio Off-Ramp I²S and modified Benchmark DAC-1 [under review]
Analog Source: Nottingham Analogue Mentor turntable with 10" Anna tone arm, Roksan Shiraz cartridge
Preamp: Hovland HP-100 MC tube preamplifier, fully updated
Power Amp: Red Rose Model 2A Silver Signature tube amplifier (Mullard xf1 EL34s, Tung Sol black plate 12AT7s and RCA black plate 12AT7WAs)
Speakers: Audio Physic Anniversary (SLE) Step speakers, Audio Physic Luna active subwoofer
Cables: Music Metre Fidelis digital interconnect, Harmony Audio interconnect, Stealth CWS interconnect, homemade twisted pair of mil-spec silver cladded multi-strand speaker cables
Power Cords: Analysis Plus Power Oval (amp), PS Audio Mini Lab (preamp)
Equipment Rack: Michael Green Justarack
Sundry Accessories: steel shot loading in Audio Physic Step stands, Audio Point coupling discs under Step stands
Power Line Conditioning: PS Audio P300
Room Size: 29' long x 16' wide x 10' high (sunken living room with open floor plan, listening across width of room)
Review Component Retail: $2,645 - $3,895 depending on options

Are you feeling down because the megabuck digital front end you bought just last year is already looking a little hoary? Are you overwhelmed at the thought of trying to keep up with the advancements in digital audio playback? You're not alone. The array of new products and the concomitant depreciation on last year's digital components can be mind-numbing.

That is why I have elected to spend conservatively on my digital playback equipment while keeping an eye peeled for any really significant developments. My Vecteur D-2 transport and tweaked-out Audio Note DAC 1.2 have served me well and I am still enjoying their sound greatly. However, during my recent review of the Bastanis Prometheus Mk. II speakers, I was loaned an Empirical Audio-modified Perpetual Audio P-3A DAC. While I did not find the time to do a direct comparison with my Vecteur/AN rig in my downstairs system (the speakers were the focus), I was mightily impressed with the performance of this DAC in the Prometheus-based system.

Despite a penchant for tinkering with vintage audio, I also make every effort to keep abreast of developments in the modern audio marketplace. In particular, the rapid developments in the digital music server/computer audio sector have caught my eye. While none of my friends have gone this route yet, many are touting it as the inevitable future of audio. So, I decided it was time to take the plunge and find out what computer audio was all about.

Since this is my first venture into computer audio, those of you who are more advanced might find some of my explanations and revelations elementary. On the other hand, there is a large sector of the audiophile community who seem to have little knowledge, but much curiosity, about this relatively new method of music playback.

Audiophile-quality computer audio generally means using either a USB (or Firewire) DAC, which can be directly connected to the computer via a USB cable, or various combinations of a Wi-Fi wireless connection between the computer and a converter. Computer audio competitors include the high-end USB DACs from Wavelength Audio, Audio Note, Hagerman Technologies and Bel Canto, plus lower-priced options such as the Scott Nixon USB DAC, Trends Audio UD-10, MHdt Laboratory USB Constantine, Apogee Mini-DAC, APL HiFi NWO-DAC1 and others. You can also purchase an intermediate device such as the Waveterminal U24 to convert your computer's USB output to S/PDIF Coaxial & Optical digital I/O and use your computer with just about any DAC. Notable Wi-Fi wireless devices include Olive Opus and Musica, Slim Devices Squeezebox and the newer audiophile Transporter, Apple Airport Express and Sonos ZD80.

Steve Nugent, the proprietor of Empirical Audio (a one-man operation), is adding another dimension to this market, the I²S (I-squared-S) DAC. Steve, an emigrant from the Silicon Valley electronics industry (Intel), started his own audio business designing and producing a myriad of audio products. His special passion though is advancing the state of the art in computer audio. His prior expertise was in digital electronics so he is well suited to this task.

The assault on Mount Jitter
While Empirical Audio offers USB DACs and a vast array of options/upgrades for computer audio playback equipment, Steve Nugent's all-out assault on high performance computer audio playback is largely about the conversion of the computer's USB data output to a data transfer format called I²S.

To fully appreciate why Steve Nugent believes that I²S computer audio is the future of digital audio playback, readers should refer to his article in Positive Feedback, issue 22, Nov./Dec. 2005. Quoting from this article regarding I²S transmission: "...I²S or I-squared-S is an interface found on some transports, such as Audio Alchemy and on some DACs, such as the Perpetual Technologies P-3A and the Northstar 24/192, but is not a common interface. This interface replaces the "digital coax" or S/PDIF interface that is common for most CD and DVD digital audio. The I²S is a 4-signal or 4-wire interface, consisting of two clocks, a L/R channel select and a serial data signal. There are several advantages to this interface, one of which is that unlike S/PDIF, it includes a bit-clock, which eliminates jitter contributor 7 [referring to another part of the article which lists different sources of jitter - Steve Marsh]. Another advantage is that it is the "native" interface for most DAC chips. This means that no translation of the data stream need take place in order to drive a DAC chip. This inherently reduces the amount of hardware that the clock and data signals must pass through. As in most audio equipment, simpler is better and this is no exception. The clock jitter is minimized but not eliminated in systems that use the I²S interface. It is still possible to have significant jitter even with this interface if it is not well implemented. An interface that converts USB, Firewire or Wi-Fi from a computer to I²S driving a DAC chip directly has the potential to outperform all other current techniques if implemented well..."

Implementing it well is what Steve Nugent's background in digital electronics allows him to do. Even for manufacturers that include I²S input on their DACs, Steve says they have design issues that prevent them from performing optimally. If you are a bit rusty on the sources and evils of jitter, the 1993 archives of Stereophile are a good place to start. The linked article points out how the AES/EBU or S/PDIF interface between the typical CD transport and external DAC can cause jitter.

Steve Nugent's article gives a more current and comprehensive list of jitter sources, making his case for I²S even stronger. An email he sent me, commenting on the Stereophile article, said: "With I²S, the shorter the cable is, the lower the jitter. This is primarily because at very short cable lengths, it does not behave like a transmission-line but more like a lumped capacitance. Therefore, it is the HF roll-off that results in jitter due to the cable capacitance. However, it is not the HF roll-off of the cable alone that causes the jitter; it is the detection of the signal transitions. One thing that is not mentioned, which is very important and relevant to jitter, is the effect that power system noise and thermal noise has on the detection of signal transitions by the receiving chip. If the signal has a slow rise-time, then more jitter will be caused simply because the receiving chip cannot decide when the transition actually took place. There is more ambiguity due to the changing "switching threshold". This is the voltage at which the receiving chip decides that a signal transition has occurred. This is why both quiet power systems and fast rise times are essential to achieving low jitter."

Empirical Audio's computer audio solution to jitter converts the USB output signal from the computer into I²S format by means of an outboard box is called the Off-Ramp I²S. This is not a simple design exercise and care must be taken to interface the I²S properly with the DAC (more about this later). In practice, one end of the USB cable plugs into the computer and the other end into the input of the Off-Ramp I²S. Then, the I²S cable exits the back of the Off-Ramp and plugs into the DAC (modified to accept I²S).

The other important anti-jitter technology Steve incorporates in the Off-Ramp box is the Superclock (version 4 now). This is a circuit developed and produced by Audiocom in order to "minimize the effect of jitter". It is a master clock oscillator and is important for reclocking the data after conversion from USB to I²S format and transmission to the DAC chip. (The Superclock has been a somewhat popular mod that audiophile tweakers get installed into their CD transports.) As you can see, Steve Nugent is dead set on becoming the anti-jitter superhero. Or, using another metaphor, he is trying to climb the Mt. Everest of computer audio playback.

The tactical gear
Steve Nugent was until recently modifying Perpetual Technologies P-3A DACs (and others) for I²S input with computer audio but he has now switched his primary efforts to the latest Benchmark DAC-1 because he feels its new AD1853 DAC chip is superior. As a result, Steve sent me the following sizeable list of hardware and software for the review:
  • Toshiba laptop computer, $1,000 from Best Buy
  • Off-Ramp I²S with Superclock 4, $950
  • Benchmark DAC-1 ($945) with I²S input mod ($500) and minimum mods ($250, see details on website)
  • 10 foot USB cable (connects computer to Off-Ramp I²S)
  • 1 foot I²S cable (connects Off-Ramp I²S to DAC)
  • 12VDC wall-wart power supply, plugs into Off-Ramp I²S CD-ROM w/ M-Audio driver software*, ASIO plug-in and SRC** 24/96 up-sampler plug-ins for Foobar2000

* purchased from M-Audio
** SRC licensed from Mega Nerd Pty Ltd., Australia

The total cost for the Off-Ramp I²S/minimum mods Benchmark DAC-1 combination package described above is $2,645. You would still need to supply your own computer. This is the combination that Steve Nugent feels is his "most affordable computer solution" that he believes makes world-class sound.

The I²S mod for the Benchmark DAC-1, while obviously installing the RJ-49 female jack for the I²S cable, also eliminates the upsampling clock and upsampling chip to become a NOS or Non-Over-Sampling DAC. Instead, upsampling is done in the computer and fed to the modified Benchmark.

The Toshiba laptop computer supplied to me for this review thankfully came loaded with the necessary software:
1. Foobar 2000 version 0.8.3 - this software manages your playlists and contains all of the typical control panel functions for playing selections. It is open on your desktop when playing music. Steve Nugent told me that there is a newer Foobar version but it does not sound as good.
2. ASIO driver - this software supplants the native Windows ASIO as well as bypasses the Windows resampling/DSP software, KMIXER. It is sonically superior and "bit-perfect".
3. M-Audio Transit Driver - allows selection of playback resolution level to enable transfer of high-resolution 24-bit/96kHz files through any USB-compatible computer.
4. SRC Resampler - 24/96 upsampler plug-in software for use with Foobar 2000.
(For Empirical Audio customers, Steve supplies this software on CD along with installation instructions.)

Base camp - Superclock 3 vs. 4
I put the Toshiba laptop on a spare Target single-component equipment stand I had on hand and placed it right next to my equipment rack. The Off-Ramp I²S and the modified Benchmark DAC-1 fit easily on one shelf of my Room Tunes equipment rack. Steve Nugent also sent me a one-meter pair of Empirical Audio interconnects to use between the Benchmark and my Hovland preamp so I dutifully used them in place of my Harmony Audio interconnects in this position.

The first order of business in the not inconsiderable list of listening comparisons ahead of me was to compare the Off-Ramp I²S with the brand-new Superclock 4 (and not broken in) to the Off-Ramp I²S with Superclock 3 (sent along for comparison only). The version 4 became available only shortly before Steve shipped me the gear so he did not have time to break it in.

I started listening with Mussorgsky's Pictures at an Exhibition, St. Louis Symphony under Leonard Slatkin [Mobile Fidelity SACD UDSACD 4004] from the Toshiba's supplied 16/44.1kHz playlist. When playing these files, the user needs to be sure that SRC Resampler is moved into the left field (active) in the Foobar Preferences DSP Manager. This activates the upsampling software to output 24/96kHz.

The improvements wrought by the Superclock 4 over the Superclock 3 were not subtle. The version 4 provided this classical workhorse with greater dynamic authority throughout the frequency range, combined with more delicacy in the softer passages and a greater sense of presence. The improved bass power was particularly noteworthy. If you are currently using Superclock 3, do not hesitate for a minute to upgrade to Superclock 4. I think it ranks as one of the most cost-effective sonic improvements you can buy.

The Ascent - The Two Benchmark Mod Levels
Hot on the heels of the initial equipment delivery, Steve shipped me his more extensively modified Benchmark DAC-1. This model is $3745 including the companion Off-Ramp I²S. The unit supplied to me also had modified S/PDIF BNC input circuits, which added $150. The mods at this level (which I will call level 2 for the purposes of this review) include the minimum mods plus the Turbomod, dual op-amp upgrade and I²S interface. Please see the website for full details. (There is still a mod level above this, which has 18 OPA627 op-amps on adapters. That package is $4295 complete.)

Another of the classical pieces from the supplied 44.1kHz playlist, Ravel's Bolero (recording source unknown), served to illustrate the differences between these Benchmark mod levels. The opening movement's incessant snare drum and tympani rhythm had more weight and drive with the level 2 Benchmark. Since this is such an integral part of this piece of music, it resulted in a much more involving presentation. Other improvements noted were a wider and deeper soundstage, more information and better bloom of the acoustic envelope into the room. While the minimum mod Benchmark gave a very respectable showing with a clean and tonally balanced presentation, for my taste and system level, I would definitely spend the extra money for the level 2 mod Benchmark. In fact, I had trouble going back to the minimum mod Benchmark once I had heard the level 2 version. From this point forward, I used the level 2 modified Benchmark although I did go back to the minimum mod version toward the end of the review to confirm my impressions.

Track 3 of the Bolero selection had me reaching to lower the volume control. The tympani percussion was truly thunderous and caused my amp to clip. I'd never experienced this level of sheer dynamic explosiveness from my system. It got me thinking more seriously about switching to a solid-state amp for my Audio Physic Step SLE speakers. After all, they are 84dB/4-ohm speakers and they would probably do better with more than 40wpc of tube power even though they are small speakers and the Red Rose is known for its drive capability.

From what I've read on the Empirical Audio-sponsored discussion forum on Audio Circle, the first buyer of the Off-Ramp I²S with Benchmark DAC-1 also commented on the significantly improved bass. While the most notable improvement is in bass power and dynamics, it is not at the cost of sloppiness or blurring. My amp/speaker combination was clearly taking a beating, which it had not seen the likes of before.

At this point, I began to sample other selections from the supplied playlist. Steve initially told me that the Off-Ramp I²S would need about four days of constant play for break-in but even after this period, I noticed a considerably unpleasant level of harshness in the upper mids and treble on some of the rougher recordings in the playlist. I talked to Steve about this and he then said that I really needed to give the Off-Ramp a month of constant play for near complete break-in. It turns out that there are a number of Black Gate caps in the Off-Ramp, which are notorious for an odiously long break-in period. So I set the Foobar to Repeat mode and periodically did a sound check. The Led Zeppelin II tracks from the 10-CD box set release [Atlantic 7 82526-2] on the 16/44.1 playlist proved to be the acid test for checking on the break-in progress. The break-in month went slowly and with intermittent swearing at the Black Gate caps, they finally relented.

Camp One - Vecteur D-2 with Benchmark DAC-1 via S/PDIF
Not one to pass up new music being served to me on a silver platter - er, hard disk, I set out to listen to all 164 of the 16/44.1 selections in Steve Nugent's supplied playlist, with the SRC Resampler doing the upsampling. All listening from this point on was done with the level 2 mod Benchmark DAC-1.

After listening to all of the selections several times over, I was able to establish what could be considered the characteristic sound of the entire rig. It was one of remarkable clarity, unsurpassed levels of information, powerful bass, reference level dynamic performance and excellent transparency. The only criticism I had was that on lesser recordings, there could be some stridency in the upper midrange and treble. This was noticeable on the playlist's violin recording by Friedman, as well as the cymbal crashes in the Led Zeppelin cuts and others.

I decided to try to ferret out which of the components in the chain was responsible for this treble quality. Connecting my Vecteur D-2 transport to the BNC input (using an RCA/BNC adaptor), I listened to a variety of my workhorse CDs. It quickly became apparent that the Benchmark DAC-1 showed the same slightly tipped-up treble as it had when using the computer as the front end. Perusal of Audio Asylum turned up several comments agreeing with my observation while others disagreed. I would venture that this is a matter of taste analogous to the solid-state vs. tube camps. This gets into the whole sticky business of arguing whether the Benchmark is simply allowing you to accurately hear what is on the recording, or whether you feel it is the blame of the Benchmark and would prefer the tonal balance of another DAC.

I should add that playback via the Vecteur was definitely inferior to the computer and Off-Ramp I²S, with reduced information, dynamics and focus.

Sherpa guide needed
I would be remiss not to mention the most vexing annoyance that can be experienced with computer audio playback via the USB port. That is the problem of pops, ticks and hiccups during playback. Pops and ticks are not a problem with computer audio via wireless transmission but wireless can have other problems such as bandwidth limitations or interference from microwaves, wireless phones and other wireless networks. These pops and ticks sound like the music just disappears for a fraction of a second and that is exactly what is happening. I find it to be more annoying than a pop on a record where the noise is usually superimposed on the music. These seem to occur almost at random but can be as prominent as two ticks in an average length track. Conquering this problem can be a pain and the PC-Audio discussion forum on Audio Asylum is replete with testimony about how to solve this problem. It is certainly quite possible to eliminate the pops but it can require some patience and guidance.

Steve Nugent explained the reason for these pops to me. It has to do with the software (i.e. M-Audio Transit) driving the audio through the I/O (input/output) port. With large latency drivers (i.e. large buffering or storage capacity), there can be delays in delivering the information from the hard disc to the memory and out the I/O port. These delays cause the temporary dropouts. Steve is trying to get driver software with less latency.

For now, there are a number of steps one can take in order to eliminate these pops:
1. Disable virus protection, instant messaging, screen saver and all non-essential software.
2. In the Foobar DSP Manager, Resampler SRC, switch to "slow mode".
3. In the Foobar Preferences, Output, ASIOdII, double or triple the buffer
4. In Foobar DSP Manager, Playback, try increasing the full file buffering from 0 to 2, 6, 10, etc. until successful.

I'm sure by now many of you are saying "sheesh, I don't want to go through all of that to play my music." I can relate and must state that this method of computer audio playback not only has a considerable learning curve, it can also require a lot of fiddling at the beginning. Steve Nugent assured me that the Toshiba was playing music fine without pops at his home before he sent it. While I greatly reduced the problem with Steve's help, I never did get rid of all of the pops and ticks and just lived with some. Even with the hassles of dealing with the pops and ticks, Steve summarizes the situation with USB output as follows:

"The advantage of USB is that it is a dedicated connection that will always have the bandwidth required. The problem with ticks on the USB interfaces is more of a computer resource issue and the efficiency of the USB drivers. With a fast-enough computer or a low-latency driver, you will never have ticks with USB, even with the current Synchronous-Adaptive protocols."

Apparently, the real solution to this problem is for engineers to develop a new type of driver software that feeds the data out of the USB bus in packets rather than a continuous flow. This new type of USB driver would be called asynchronous as opposed to the current synchronous protocol. (If you really want to dig deeply into this topic, go to the Audio Asylum PC-Audio forum and type in "synchronous" in the search window.) Steve's comment on the prospect of developing asynchronous USB protocol, excerpted from Audio Asylum is:

"If you read the blow-by-blow of the trouble that the TI (Texas Instruments) engineers had to even get the synchronous audio USB working properly, you will see that it is probably extremely hard. I understand from this article that the main problem with asynchronous is that it is not isochronous by definition. In other words, there is no bandwidth or channel non-interruption guarantee. I believe this is why they went with synchronous."

Venture capital, anyone?