PS Audio DirectStream DAC, a chameleon in the high foothills, by Ted Brady
This has been the hardest review to write, for many reasons, not the least of which is that this DAC took me on a journey that seemed, at times, like a wild goose chase. But here we are, and if you, the reader, are like so many others, you'll go the end and find that this goose chase resulted in quite the golden goose. 
By now most of you know that this DAC is the brainchild of one Ted Smith, a longtime developer/designer/wunderkind who had been dabbling in this DAC design for over 10 years. Two major problems existed during that time: first, the technology didn't exist to do what he wanted; and second, he was but one guy in a lab, with no real ability to produce and distribute a DAC of this sort. The first problem was slowly solved by time....ingredients like the Xilinx Spartan 6 FPGA became available. The second problem was solved by one Paul McGowan of PS Audio fame. He offered Ted his help, and the result is the PS Audio DirectStream DAC, a major change in PS Audio's heretofore standard-chip PCM DAC evolution.
This unit has been reviewed by others, and I invite you to visit those reviews for a more in-depth technical discussion of what this thing is made up of, including dimensions, shipping weight and pictures of its innards. I am not the most technical and will gladly leave that up to the major publications. What I can bring to the reader is another audio lover's perspective of having listened to dozens and dozens of DACs over the past several years, albeit with my own personal listening biases (my system is included in my signature link, although this session includes my recently acquired Aerial Acoustics 20T speakers).
I first became excited to hear this DAC when, during a getaway to the Boulder, CO area with my wife, I was able to visit and chat with Paul at the PS Audio headquarters (thanks to friend and master cable maker Greg Graff of MG Audio Design for setting up the meeting and being our chauffeur). Paul took me on a tour of the facility, and it was clearly evident that this company knew what it was doing; it knew its strengths and had a long history of customer support and satisfaction. However, this DAC was new to all of them and was a "stretch" for PS Audio (my opinion, not theirs necessarily); they had never done a DSD product, and to my knowledge the DS was going to be their most expensive product to date, too. Paul was ready to go, though, and it was evident he was appreciative to have Ted Smith on this team.
As a DAC designer Ted was presented with the typical choices of off-the-shelf DAC chips, all of which are very capable at ultimately delivering toe-tapping musical output, given a white glove treatment after the chip output (very high quality analog stage of either solid state or tubed design, quiet powerful ac supplies, etc). Some chips, like the ubiquitous ESS Sabre 9018, have incredible sample rate choices, but trade those choices for the need to harness that power downstream lest one create what the audio lover has come to know as Sabre-allergy, the well-known easily-identifiable sound of a razor-sharp leading edge that can ruin the illusion of real musicians playing to one's whim. Folks like George Klissarov (exaSound) and Xuanqian Wang (Auralic) have figured out how to tame that beast better than others, and the results speak for themselves.
Ted, on the other hand, had a design idea that no off-the-shelf chip could handle. He was battling his love of DSD (and its inherent noise shaping baggage) with the very different, but noisy nonetheless, PCM brick wall filters. What if he could take either format and send it into deep space (oversampling to, say, 30Mhz, and 30 bits, where the many-multiples of all known PCM and DSD noise filters all come together), send the noise into the ether, then return to a high earth orbit of 5.6Mhz. At that height there is still very little noise to interfere, and then "simply" send to an industry-proven set of PS Audio power supplies and analog parts? Would it sound better?
Enter the field programmable gate array (aka FPGA, my audio acronym of the year). It is a sort of tabula rasa for DAC chips, a hardware/software platform chip that allows for almost infinite "roll your own" chip design. The code can be tens of thousands of lines, but the good news/bad news is that once a DAC manufacturer enters this hallowed world, it becomes a software company. Why is that good news? Because the consumer then buys into a product idea that has potential for significant improvement and change through downloadable firmware and software updates All audio equipment these days includes software, so why not invest in one where the code is owned and maintained by the equipment manufacturer, allowing for an evolving set of fixes, improvements and feature enhancements (within reason and scope). Companies like PS Audio, Chord and others are perfectly positioned for this new approach, as they have mature organizations and knowledge bases that can deliver updates and fixes in a reasonable process. Oh, the bad news? Not all audio companies can take this on, and not all audio units are designed for upgradeability. More to the point, this unit has already had two large FPGA and driver updates, and the resultant sound was improved both times.
The PS Audio DirectStream DAC will accept PCM and DSD music up to 24/352.8K PCM (aka DXD) and DSD128 (aka 2X DSD) and will accept it in a plethora of digital inputs; the back panel includes RCA S/PDIF, toslink (limited to 96k), USB, AES/EBU and the house favorite for PS Audio users, the somewhat proprietary (oxymoron I know) I2S input, housed via a pin-compatible HDMI connection. Each of these inputs eventually meet at an internal I2S connection, and then to the FPGA. There is also a smart card input for the aforementioned code updates (FPGA code, USB driver, etc). There is an ethernet connector for their own networked Bridge product, as well.
On the front panel the DS has a nice touch screen (an evolution from the look and feel of their earlier PerfectWave DAC units) which handles, along with the included remote, input selection, high/low gain selection, volume control (for direct to amp capability), balance, and software information (versions, etc), along with a running display of format and sample rate of the incoming signal. Note: I am not a huge fan of touch screens, given that my DAC is usually not within touch distance, nor can I read the screen from my listening position, but that's purely my own pet peeve (and poor eyesight). The screen display can be turned off remotely, and given PS Audio's long experience with them, they are likely not a huge cost issue nor repair issue. And it's a nice demo feature, too.
The DAC's chassis is a beautifully done silver or black copy of their own transport (the Perfect Wave Transport) and includes a hand polished top cover of black-lacquered automotive-finished lexan-looking MDF. Very nice. As it should be for $6000 ($4000 for PS audio owners who are in the Company's long-standing upgrade program, as they turn in their former PWD DACs for upgrade; a nice feature given the poor resale value of ever-evolving digital gear on the audio classified sections).
I was told (but forget by whom) that the DS would require significant break-in. Ted was not as adamant, but gave me a few hints on trying to accelerate that process (running on low gain will run more current through the transformer, etc). I have another good news/bad news item. The good news is that there seems to be nothing in this DAC that gets hot. The bad news, heat helps break in (eventually spoiling it, however) so my patience began to wear thin at about 800 hours. Why? Why not just listen and enjoy? Well, therein lies the theme of my first several months with the DirectStream.
I tried every known input source I had on hand (pc running JRiver/Jplay using high-end USB card and power conditioning, Auralic Aries renderer via USB/toslink/spdif/AES, Aurender Linux server running USB, yada yada). What evolved (there's that word again) after the 800 hours was akin to the most amazing new 4K video display you've ever seen, with details that boggled the mind, yet never looked artificial nor over-enhanced. Subtle details and image densities that I'd never heard before, on recordings that I was so familiar with that when I heard an even 30 db down nuance new to me I would jump out of my chair. I am a music lover, but sometimes this demo-quality playlist stuff can drive one insane. Why did I not want to venture on to my real fave selections? Getting back to the CES demo of that incredible 4K display.....lo and behold they had somehow turned the color and hue knobs all the way down. ? Ansel Adams in black and white is nice; Stardust Memories in black and white is still funny (but not as funny as his earlier stuff); but....my favorite music in black and white? I was bored. I tried everything. I put the DS on the shelf for a few weeks and then went back to it. I would try the bloomiest of music, or tube roll my preamp to the bloomiest/timbre-soaked tubes of Mullard variety, still no relief, only a black and white depiction with newly introduced veils. Argh!
I was stumped and ready to write a review that this DAC is the perfect fit, perfect antidote, for those systems that are overly colored, overly warm. Yes, a poorly hidden left-handed compliment. Not usable in my system, but maybe if you have one that already has plenty of color and tonality? (That is not to say my system is overly analytical, and therefore I felt bad that this review would be seen for what it was...throwing a bone to an otherwise good Company and good folks.)
A few weeks ago I got a call from a good audio buddy, Jesus R of Sonore fame. He wanted me to hear his flagship UPnP renderer, the Sonore Signature Rendu. It is an ethernet to SPDIF (or I2S-HDMI) converter/renderer. Sure, why not. A couple weeks ago it arrived and I hooked it up via BNC/SPDIF to my Chird Hugo. Hmmm...very nice indeed. (Note: Chris's review of said Signature Rendu is due out presently). The DirectStream, gathering a little dust while I decided how to write the review, was calling out to me "play me, play me".
Jesus is very proud of the work that went into the Signature Rendu, especially the very low jitter, very high quality I2S output (compatible with several pin-compliant I2S HDMI inputs on several different DACS, including the DirectStream.) As luck and irony would have it, the Rendu setup required me dusting off an old PS Audio P300 ac-generator to block some house-bound dc that caused the Rendu to drop off the LAN periodically (a fix that is coming soon, according to Jesus). After confirming that a P300-powered Sig Rendu was now rock solidly connected, I found a nice Nordost HDMI cable and entered the world of I2S (I can't tell you how many sources, format converters or DACs I've had in here that taunted me with this connection, but this was the first time I had input and output together in one room).
This began one Monday evening at 8pm. My first taste of this combination lasted, without wine-induced bathroom break or any other silly distraction, until about 3am in the morning. It was an audiophile's best dream, an audio lover's best attempt of heaven on earth, a contender for one of two or three best audio listening sessions of my life. Due to my well-heeled assumptions, I went from WTF to laughter (yes, I think I physically laughed out loud a few times; thank god no one heard me; they were all asleep) then back to WTF several times. The display analogy: not only had they turned the color and hue controls back on, the display became 3D (without glasses, thank you) and had the blackest blacks I'd ever seen. Again, sorry, but WTF! Why, with this input (Ted promised me all inputs were identical) would the DirectStream now explode into colors, tones, rhythms, clarity, absurdly accurate image density, and overall musicality that I've maybe never heard before. Why? Yes, the Signature Rendu is a clearly a special piece, but it needs a partner, and the DirectStream danced with it like they have been secretly practicing every dance style for 30 years.
I-squared-S is a native connection; by that I mean two things: 1) that both PCM and DSD can be natively streamed, without the need for DoP (DSD over PCM process, which for some odd paths requires wav or FLAC containerizing), so when using Minimserver one needs not do the Dopwav commands if your renderer formerly needed that; 2) it is an unencumbered signal path directly to the Xilinx FPGA Spartan 6. To this day I don't know what jitter sounds like, but as of that special night and going forward, I now think I know what no-jitter sounds like.
Whether it was demo tracks or any of my other regular rotation, this new combination threw a most musical, accurate (yes, those are not opposites) and detailed soundstage in front of me. On New Moon Daughter (Blue Note, 24/192) Cassandra Wilson's opening "Strange Fruit", a Billie Holiday classic, had all the wetness and organic vibe I'd ever heard from this tracks, with the cornet solo at the beginning occupying a perfect space at the rear of the left center soundstage. The decays sounded as natural as can be (dunno, I wasn't there when recorded). On Keith Richard's under-appreciated solo album Main Offender (Virgin, redbook) the capture of the amp buzz of those sessions is incredibly well done, and on the right systems the album sounds live and dynamic as heck. Yep, here too, in spades. On "Will But You Won't" you can actually here when the second guitar mic feed opens up, prior to first rip. Very wild indeed.
Those two examples represent PCM playback with a machine that eventually treats everything as DSD128. What about, say, DSD and DSD128 (neither of which get a free pass...they are both upsampled to deep space). Well, on Us (Geffen, DSD) Peter Gabriel's 6th studio album, the SACD mastering is a huge step forward and is one of my favorite test albums for pop/rock DSD playback. The DS did not disappoint; "Only Us"/"Washing of The Water" combination is a nice test for leading edge, percussive tonality and noise floor. Let me start with noise floor. Whether it is I2S or what, the DS portrays a noise floor of cavernous proportions, a trait that seems to be the most influential on adjectives like clarity, micro-detail, air and even dynamics (since you can now hear the quietest of the musical nuances). I sat there for 58 minutes (the length of the album) and almost gawked at the energy and mix of colors and tones I was hearing. Oh, and as a dyed-in-the-wool computer audiophile I almost NEVER listen to a full album anymore...which is a confession I am not always proud of. Well, when listener fatigue becomes non-existent (and I don't need to work the next morning) this setup tempts me to listen all the way through.
Other DSD recordings, including my favorite Mari Kodama Beethoven Piano sonatas collection (Pentatone, DSD) played back with energy, sizzle (when appropriate) and a very 3D soundstage. Was this the best DSD I'd ever heard? Maybe not (dynamics and air are 95% of the best DSD I've heard), but as a dual format DAC (PCM and DSD) it leads the pack, and at the end of the day, I am not thinking about other DACs (ok, that sounds too creepy I guess, sorry). When given DSD128, like the wonderful Eric Bibb download compilation of analog-to-DSD tape transfers, A Selection of Analogue (Opus 3, DSD128) his slightly gravely-but-soft voice comes through in great detail, without sacrificing tone and, well, analog(ue) naturalness. Whether it's the pedal organ and the Deacon choir on "Where the Green Grass Grows" or the steel guitar solos throughout, the relaxed feeling of Eric Bibb music is conveyed nicely, thank you.
This DAC is a difficult one to categorize. It is hyper-DSD internally, yet plays back PCM even better than it does DSD. It is supposed to be input-agnostic, yet I2S is easily the best input for me, followed by USB and then all the others relatively equally. By the way, it joins other open system DACs as being UAC2 compatible, meaning that no drivers are required for MAC or Linux (read: streamer) handshaking. That is a nice feature and shouldn't go un-noticed.
So...what does one do if he/she has no I2S connection (all my listening feedback here was via I2S)? I tried going back and listening to the other inputs (especially my highest quality USB, as noted above) , and although this evolution (damn it) has seemed to continue to open them up a bit (especially USB via the JCAT card and either JCAT cable or TotalDAC D1 cable), there is no turning back for me. As a dual format USB DAC it is very good and recommended on all $5K+ budget's short list. As an I2S DAC I can't think of why this wouldn't be your first choice. And as a Signature Rendu owner, sell your blood, sell something quickly and go buy this DAC (and yes, give it 800 hours or more, although I don't know how good it would sound cold....maybe not bad at all!!!).
Ted Brady
P.S. I welcome Ted Smith to please tell us why this last chapter (or more appropriately the first several bad ones) of listening happened, and why I am getting such different sonic experiences with different inputs.
Manufacturer's comment:
Thank you for your thorough and detailed appraisal. It's been a long journey to
bring Ted Smith's*design into production, from the first time Gus Skinas played Ted's
hand-built proof-of-concept*for us, until now.
Starting with*Ted's brilliant vision and code, our Director of Engineering, Dave Paananen,
our Chief Engineer, Bob Stadtherr*and their whole crew of designers and technicians have put in
thousands of hours to bring DirectStream to market.
The response to all the hard work has been gratifying; DirectStream has been well-received
all over the world. Perhaps the most exciting part of DirectStream's design is that, thanks to its
ability to be updated with new firmware, the best is yet to come!
Paul McGowan, founder and CEO, PS Audio
Product Information:
- Product - PS Audio DirectStream DAC
- Price - $5,999
- Product Page - Link
Recommended Comments