 review by Andrew Ballew (Click here for DAC specifications) When the S.M.S.L D400 PRO DAC arrived here seemingly an eternity ago, I was quite excited to hear a different implementation of the latest state-of-the-art AKM dual chip AK4191+AK4499 DAC silicon. I had first encountered it in the Topping E70V Velvet DAC, and was intrigued by its potential. Something wasn’t quite right with the sound though. (Click here for that review.) Overall it sounded very, very good, but there was a slight graininess in the sound that didn’t sit well with me, that wasn’t present in other AKM implementations I have heard. At the time of review I chalked it up to being over-revealing. Also, the Topping (despite what some well-known online reviewers claimed) did NOT take advantage of the native DSD ‘Bypass’ mode. It simply had a fixed volume output mode. Bypass does more than skip the volume control; it bypasses the delta-sigma converter altogether, converting the 1 bit signal directly with a type of FIR output filter. It is a digital filter that can be realized directly by the analog output resistors, operating in unary code. More specifically a CIC type FIR filter without a decimation stage. Different chip iterations use slightly different implementations, some with more discrete analog parts, some with more pure digital logic but ultimately an analog end where the way they got there is really semantics. The results are identical. Previous AKM implementations of native ‘Bypass’ mode have sounded extremely good, using the combination of a digital CIC filter with equally weighted taps, each with a mathematical value of 1, outputting a digital unary code, with no more digital DSP than that. The digitally filtered signal was sent directly to the switched capacitor filters for final conversion to analog. I wanted the chance to hear it as implemented in their latest and greatest chip, which moved from switched capacitor filters to resistor-based digital to analog switches. Little did I know exactly what was in it for me- that seeming eternity of measuring and reviewing. (As an aside, I had assumed AKM along with the changing to resistors at the output also shifted the entire 'pure DSD' CIC filter to the end of the chain, using the resistors themselves as the 'only' taps, accomplishing the filtering and conversion to analog in an efficient, single logic, with the signal remaining entirely in bitstream until the 'end' of the chain. Based on further studies of the design of the new chipset after their datasheets became available for mass consumption, I believe they have stayed with what worked in the past. DSD is filtered digitally early in the process, in this case on the AK4191 chip. It is a CIC filter as before, with equally weighted taps with a value of 1 which will output a digital unary code, which is then converted to an equivalent binary code for transfer across to the AK4499EQ chip. The unary code is recovered fully from the binary code in a logic system, and is sent to the output resistors for conversion to analog. This is unlike other designs, whether discrete (Signalyst) or on chip (Burr-Brown) that keep the DSD signal as 1-bit bitstream until the final filtering, which while using similar digital operations, implement them in an analog manner with the taps of the digital filter being the very output elements themselves, directly creating the intended output voltage rather than outputting a digital 1 or 0 to be converted to analog later. Either method will lend the same results, therefore the discussion is a purely semantical curiosity.) Starting with the basics, the D400 PRO arrives as a well-built piece of kit. It is solidly put together, its display is quite crisp and of high quality. It is easy to use and the controls feel solid. The remote is the same generic remote I have seen with other S.M.S.L products. It does the job, and that is really all that can be or need be said about it. Some months earlier when I reviewed the S.M.S.L D300 I was stunned by how cheaply that DAC was constructed. (Click here for review.) No such issues with the D400 PRO. It is a properly built component as I would expect for its price. To evaluate the D400 PRO, I paired it with two different headphone amps, those being the iFi ZEN Can, and the S.M.S.L SP400. Headphones used in the evaluation were my trusty Sennheiser HD650, the Hifiman HE-560, and the Focal Clear MG Pro edition. Software for playback was Roon via standard windows PC. The DAC was galvanically isolated from the PC at all times via the Intona USB 3.0 SuperSpeed Isolator. The DAC was always evaluated via the balanced outputs, and all cabling from USB to XLR consisted of a variety, mostly from AudioQuest and WireWorld. I began my evaluation of the DAC with the Elgar Cello Concerto in E minor Op. 85 via TIDAL (16bit 44.1k PCM). Alisa Weilerstein, performing with Staatskapelle Berline under the baton of Daniel Barenboim, shows absolute mastery of the Adagio-Moderato opening movement, with breathtaking subtle musical movements of what might be the most beautiful Cello melody ever written. The incredible technical dynamic range of the S.M.S.L D400 PRO powered by the latest AKM chip really allows the finest of low level details to become present, offering up a stunningly ethereal, nearly live sounding performance. Fine details float in space both left and right, making me forget for a moment I am listening via Sennheiser HD650 headphones, briefly fooling my ears that I am listening via my reference Martin-Logan ESL speakers. Yet no crosstalk is turned on in the digital realm; neither is the iFi ZEN Can being used with any of its onboard analog signal processing active. Very impressive. Unfortunately as beautiful as the S.M.S.L resolution is a low levels, things begin to sound more flat and compressed at higher volume levels. All evaluation to this point had been with the ‘Low Dispersion’ filter. For the "fun of it" I listened to the same sample of music with every available PCM filter. In ‘Super Slow’ (non-oversampling), the cello tones and overtones sounded quite glorious. Even a bit more engaging than the ‘Low Dispersion’ filter. However, the background took on a bit of graininess and the deep pure black experienced in ‘Low Dispersion’ was no longer present. Once again, louder passages sound much less tonally controlled and retreat a bit from the strong stereo imaging. After much mind-numbing comparison, I decided the most balanced filter that provided excellent low-end control along with high-mid to treble sparkle and “freedom to sing” was the ‘Short-Slow’ filter, which is a low tap minimum phase filter. You will find this kind of filter used on MQA material. Note, I am not suggesting this is the MQA filter; only that its impulse response looks similar. Cursory looks can be deceiving. The bottom line is, by some margin this is the favorite D400 filter evaluated at Euphonic Review. Unfortunately, even the most optimal filtering cannot overcome the slightly flat, spatially indistinct sound at higher volume levels that seems to be characteristic of the DAC with PCM material, in spite of state-of-the-art measurements. Moving onto MQA, I “virtually” stayed in Berlin, but switched over to the Philharmonic, still under the direction of Daniel Barenboim, for a fine performance of the Allegretto movement in the A Major Beethoven Symphony, Op. 92. For the most part until one reaches the end, this is a low-level, highly detailed masterclass of classical music reproduction. The acoustics of the hall and stage are firmly entrenched and allow one to close their eyes and see in the mind each detail that must be taking place. The imaging is quite astonishing. An individual violin just ever so slightly stands out as out of conformity for a mere microsecond. I was surprised to hear that kind of resolution here. When the viola section begins to sing a melody in one accord, its richness of tone and musicality brought goosebumps. But, just as before, when things got a bit busy at higher output levels, the soundstage flattened, the imaging collapses somewhat, and one is pulled out of the illusion of ‘being there.’ I took a break and cued up “The Ayoub Sisters” self-titled album and was left with a large smile after the track, "Uptown Funk". The percussion here is so subtly detailed and nothing in the mix is missed. Each sister’s virtuosity with their respective instruments Cello and Violin was sharp in a good way and cut through the mix. This was also the first time I felt like the overall dynamics were not compromised. The 24/96 track was very nicely rendered by this DAC. At this time, I felt excited to check out the DAC’s DSD performance; specifically, the 'Bypass' mode that allows for native DSD conversion. As I stated quite clearly earlier in this review and in other places on this website, several review sites errantly equated the fixed volume output feature with this ‘Bypass’ mode in a different product using same chipset. Just to be on the safe side however, I personally confirmed with Topping, the maker of that specific piece of kit, there was no DSD bypass available on the product. The fact that the S.M.S.L DAC under review here does indeed provide access to this bypass mode was a big, big deal for me personally, as this is how I prefer DSD conversion. ‘Native’ DSD, which I would describe as meaning as little DSP and filtering as possible. In the menu there are three choices under ‘Pre-Mode’, those being ‘Variable’, ‘Fixed’, and ‘Fixed DSD Bypass’. I began my DSD playback reviewing via the 'Bypass' mode with the 'Wide' filter activated. In 'Bypass' mode both 'Narrow' and 'Wide' filters seem to cutoff at around 70khz on DSD64 material. The significant difference between the two is a much slower roll-off in 'Wide' mode. The SACD/DSD album I settled on was “Torme’, A New Album” re-released by Audio Fidelity in 2002 re-mastered by Steve Hoffman. Wow. This was DSD as its best. Just a stunning transfer from the original analog tapes to DSD64 with Hoffman’s legendary touch. On the opening track "Send in the Clowns", the brass cuts through with such a thick, natural wall of sound that one might be tricked into believing they were there during the recording. On the second track "New York State of Mind", Torme’s voice was reproduced with truly stunning harmonics that gave such an open, rich yet light tone that truly must be heard to understand any description. None of the “flatness” I heard with PCM files was present here. It was just an outstanding presentation, and the native DSD 'Bypass' was living up to its billing, sounding like other known native DSD DACs I have here in the lab such as those made by iFi (Burr-Brown), RME (AKM), and Signalyst. And right about here, is where things went crazy and why this review took way longer than usual. Going any higher than DSD64 in 'Bypass' mode could be a total disaster of white background noise; or it might not be. Taking measurements was maddeningly inconsistent. I couldn’t put my finger on the problem. And my finger still isn’t completely on the problem. But at least I know I am looking in the right place and these gremlins are on the run. There is a feature (non-defeatable) in the D400 called ‘ND Color’ in the menu structure and ‘Sound Color’ in the user manual. Sound Color is an AKM feature for which I can find no actual technical explanation, other than it is supposed to be some kind of tone control. ESS has something similar, so I suppose this is a case of “keeping up with the Jone’s”. After lots of frustrating experimentation, I came to a workaround. Changing the ‘ND/Sound Color’ option in the menu system from default to ANY of the other three choices is STEP ONE in eliminating the issue with native DSD, all the way to DSD512. But what about STEP 2?? I thought the problem was solved? Oh no, not just yet. Your choice of “Sound Color” will interact with your choice of either 'Narrow' or 'Wide' filter, for better or worse. Eventually with the right combinations the issue is 'rectified'. Of course not exactly, this is something that can only truly be rectified in a firmware update. Until then, however, one must do as one must do. What I WILL say about DSD via the ‘Bypass’ mode- once sorted out- sounds amazing. I will also say it is an absolute shame that anyone would have to go through this bug-ridden mess to find that amazing sound. It is some of the very best DSD sound I have ever heard, and coming from an approximately $650 product, is saying something. Again, that endorsement is for the 'Bypass' mode when you have sorted the gremlins. DSD via the other possible path- the DSP/remodulation route- sounds very much like PCM as I initially described-excellent low-level performance and detail, but a bit flat and lacking in high level dynamics. However, with DSD via 'Bypass', in its simplest conversion mode, the sound approaches if not matches the very best I have heard, no matter what the measurements say or will eventually say. Further thoughts? When all the bugs are worked around, DSD played back via the ‘Bypass’ mode sounds so good that for a time I considered placing the D400 in my primary headphone system so I could move my current reference headphone DAC, the iFi iDSD PRO to my new reference speaker system. But even still I can’t help but be so annoyed by the truly unacceptable bugs that had me pulling out whatever hair I have left. As for my recommendation? If you pay attention to the ‘ND Color/Sound Color’ modes and how they interact with the ‘Bypass’ mode filters, you should have a lovely experience with this DAC. Sounds like a true ringing endorsement. But truly, if you can live with the issues, DSD playback using the ‘Bypass’ mode is just outstanding and is as good as anything at the price and more. In summary there is no doubt the S.M.S.L D400 PRO measurements are state-of-the-art. Having a true native DSD Bypass mode is a major, major plus. If you have a large collection of DSD files as I do, there will be nothing at this price that sounds better. Final opinion? It reminds me of Italian supercars. When they work, which is rare, they are like nothing else. With ownership of the D400 PRO, you may find yourself in a similar stressful conundrum, or merely experience minor annoyances, all depending on your perspective. OUR RATING: (0-10 scale. 8 or higher numbers represent best of industry. For the final score calculation, each category is given a (undisclosed) weighting and is tallied for final score. Although perfect 'fairness' is unlikely, our system tries to be as unbiased as possible. sound quality: 9.3 build quality: 9.6 ease of use: 9.2 measurements: 10.0 value: 8 TOTAL RATING: 92.0 In the Euphonic Review rating system, how are things like 'bugs', annoyances, and intangibles accounted for? The place where that makes a difference in the rating is at the VALUE line. Considering the price of the D400 PRO, combined with the headaches, well, frequently expounded on in the review, lowers its value score. We end up with the cheaper S.M.S.L D300 DAC actually coming out ahead of the D400 PRO. Yes the D300 had its issues, to the point I couldn't recommend it unless one was using ASIO drivers. But on some reflection, I don't consider that as big an issue as the D400 PRO DSD problems. The D400 PRO issues will affect everyone; the D300 issues will not. Plus the value proposition gives the D300 a real significant boost, given it comes in considerably less expensive than the D400 PRO. MEASUREMENTS For now, I have included as many of the measurements I feel confident in displaying. Thanks to E1DA for all the work they do. We could not even begin to make measurements of this quality just a couple years ago, short of buying $20,000 Audio Precision gear. Not for us. Well, not Yet....  The test results were truly astonishing. Both in how crazy good they are, and that E1DA now has a suite of measurement devices that can actually resolve at this level. Dynamic Range = 128.9db THD = -132.2db or 0.000025% THD+N = -123.6db or 0.000066% IM Distortion 19khz+20khz = -134.1 db (this is nuts) Jitter = 75ps peak / 5ps RMS (this was slightly disappointing. see graph below) Linearity = essentially perfect to -115db, finally 1db of error at -120db    Sharp Filter Impulse   Short-Sharp Filter Impulse   Short-Slow Filter Impulse   Slow Filter Impulse   Low Dispersion Filter Impulse  Non Oversampling Filter  Super Slow Impulse. The visible ringing and artifacts are from the ARTA software, not part of the actual impulse   Narrow Filter DSD64 in Native 'BYPASS' mode  Wide Filter DSD64 in Native 'BYPASS' mode. The discontinuity around 96khz is an artifact of the test-tone itself.  NON-BYPASS mode DSD Filter. Both 'Narrow' and 'Wide' Settings measure identically
0 Comments
|