Matrix X-Sabre Pro DAC Review - Stream of Consciousness

Discussion in 'Digital: DACs, USB converters, decrapifiers' started by purr1n, Mar 19, 2019.

  1. neogeosnk

    neogeosnk Friend

    Pyrate
    Joined:
    Sep 28, 2015
    Likes Received:
    1,882
    Trophy Points:
    93
    Location:
    Los Angeles, CA
    X-Sabre MQA uses the xmos xu216 chipset for usb, which is the latest and greatest (Native dsd 1024!) and only one other dac is using it. I actually prefer the usb over the Allo Signature transport. Did need a long time to burn in, approx. 400 hours to settle to it's final state. Unlike the R2R dacs, there is no warm-up time for it to sound optimal. IMO the Matrix MQA bests the Benchmark & Mytek Manhattan 2 dacs I've tried. It's a great second dac that compliments the Yggdrasil A2 nicely.

    Talking with Arthur of Matrix, he did mention that the older Pro version sounded pretty much the same minus the MQA encoding and older u208 xmos. Those have been going for about 1k used.
     
  2. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,778
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    ^ Everything here. I'm most happy when I don't have to type much.

    To add, I can see the perspective that the Convert-2 would just wear people out. I would surmise that the X-Sabre Pro is like this, but not to that extent. Still, it's a DAC that keeps you on your toes. This ain't polite and forgiving (overly) Classic R2R sound like below:
    upload_2019-3-19_16-56-27.png
     
  3. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,778
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    I can confirm no warm-up required. Sounds the same from 60 seconds on to 24 hours on.

    But but but, the does the X-Sabre handle intersample overs correctly like the Benchmark DACs according to the Benchmark whitepapers? My seriously, I'll put this on the test bench.
     
  4. ChaChaRealSmooth

    ChaChaRealSmooth SBAF's Mr. Bean

    Staff Member Pyrate Gearmaster
    Joined:
    Dec 6, 2018
    Likes Received:
    10,827
    Trophy Points:
    113
    Location:
    The Complex
    I should add that when I had the X-Sabre Pro, I listened through a Liquid Platinum and either a LCD-3 or HD 6XX. Both headphones are stock.

    I have a feeling that the Liquid Platinum, as much as I love it (and as good as it is; just look at the SBAF thread + @atomicbob's measurements), is actually the weak link. I think the LP might err on the side of polite rather than athletic, and to use a strange analogy, it felt like the LP was hanging on for dear life on the space truck that is the X-Sabre Pro. Seems to be confirmed by Marv's thoughts on the sound, though how much of it is down to the transports is unknown to me.

    Still was a great listen through both headphones.
     
  5. rlow

    rlow A happy woofer

    Pyrate Contributor
    Joined:
    Jul 18, 2017
    Likes Received:
    7,738
    Trophy Points:
    113
    Location:
    Canada
    I’m definitely not into “relaxed” I don’t believe - and I don’t think Yggdrasil would classify as that - the Schiit DACs to me are fairly well balanced on this front, with some being a bit more laid back. But not like some of the Woofson or Citrus DACs I’ve heard. Something with a bit more jump and bite would be interesting I think, although I do love to have a bit of meat on the bones and tonal density/body through the mids. Will be interested to see what final combo of settings @purr1n manages to best match up with those aspects of Gungnir Multibit.
     
    Last edited: Mar 19, 2019
  6. ChaChaRealSmooth

    ChaChaRealSmooth SBAF's Mr. Bean

    Staff Member Pyrate Gearmaster
    Joined:
    Dec 6, 2018
    Likes Received:
    10,827
    Trophy Points:
    113
    Location:
    The Complex
    He did post about his settings earlier. Although it seems like he's back and forward between modes 3 and 4 (completely understandable given that I was much the same way).

    Disclosure of my settings (which I should have mentioned sooner):
    1. Jitter: Off
    2. Dither: On (I actually wasn't too sure which I preferred, so I just left it default on)
    3. Filter: Mode 3
    4. ASYNC
    Keep in mind that all my listening was done via USB straight from my PC. The optical on my PC is okay, but was worse than the USB (despite USB 3.1; darn these gamer motherboards). I was not able to test the AES or other inputs.

    In my experience, modes 3 and 4 are similar with subtle differences in presentation. I preferred mode 3 because I felt the music had more energy, but on certain songs it can sound a little high-strung and does have a slight edge in the treble (not exactly glarey, not sure how to describe). More fatiguing than mode 4 maybe? Fatiguing might be too strong of a word. Regardless, I didn't like the other non-linear modes.

    To all others, yes I can confirm there will be a LIMITED loaner on this DAC. Details to be ironed out.
     
  7. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,778
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    ^ holy shit. Are you sure we are not the same person? (only difference being ASYNC, but understandable as I would probably be using ASYNC too from a PC optical).
     
  8. ChaChaRealSmooth

    ChaChaRealSmooth SBAF's Mr. Bean

    Staff Member Pyrate Gearmaster
    Joined:
    Dec 6, 2018
    Likes Received:
    10,827
    Trophy Points:
    113
    Location:
    The Complex
    ^Marv multiple accounts confirmed.
    Illuminati also confirmed.
    And pigs can fly as well.

    Jokes aside, there's a distinct possibility we may hear similarly. If this is true this will make my life a lot easier reading listening impressions on this forum.

    That being said, I used to own a Gungnir MB and I really liked it, so I also tried to use the filters to get closer to that sound (grated, the X-Sabre will NEVER sound like a Gungnir MB, but it definitely sounds really good in its own right with synergistic settings).
     
    Last edited: Mar 20, 2019
  9. jexby

    jexby Posole Prince

    Staff Member Pyrate Contributor
    Joined:
    Sep 28, 2015
    Likes Received:
    8,145
    Trophy Points:
    113
    Location:
    Exit stage left....
    having Mod privs and reviewing source IPs of these posts, it's clear that @ChaChaRealSmooth is actually @purr1n 's wife in Real Life. the shilling of sigma delta DACs must stop.
    will be locking this thread once I get @zerodeefex to confirm the low-level access logs.
     
  10. gixxerwimp

    gixxerwimp Professional tricycle rider

    Pyrate
    Joined:
    Sep 30, 2015
    Likes Received:
    5,773
    Trophy Points:
    113
    Location:
    small island claimed by China
  11. Metro

    Metro Friend

    Pyrate
    Joined:
    Dec 27, 2016
    Likes Received:
    1,590
    Trophy Points:
    93
    Location:
    San Francisco
  12. ChaChaRealSmooth

    ChaChaRealSmooth SBAF's Mr. Bean

    Staff Member Pyrate Gearmaster
    Joined:
    Dec 6, 2018
    Likes Received:
    10,827
    Trophy Points:
    113
    Location:
    The Complex
    I can also confirm the non-MQA version is still available (got word back from Arthur).

    The $300 premium also includes XU216 USB interface and an updated SPDIF interface in comparison to the non-MQA X-Sabre Pro, which has a U6 USB.

    Edit: Was mistaken about the non-MQA unit. That has U6 (confirmed by Arthur). The reason for the upgrades was partly to make MQA processing possible. I'm not entirely sure the value of such upgrades, but it's nice to know that the $300 is not just merely for the sake of MQA.
     
    Last edited: Mar 20, 2019
  13. Vtory

    Vtory Audiophile™

    Pyrate MZR
    Joined:
    Sep 12, 2016
    Likes Received:
    10,831
    Trophy Points:
    113
    Location:
    East Coast
    [​IMG]

    So, finally the loaner matrix x-sabre pro (XSP) in the house.

    It was much smaller than what I thought. Note that X26 is already a little smaller than DM Convert-2 or Schiit Gungnir.
    Weight and product dimension are 8 lbs and 11.8*1.5*8.2 (inches, W*H*D). Very futureistically cute. It made X26 automatically butt ugly lol.

    [​IMG]

    The fist 5-min hearing motivated me to explore option menu. Honestly, UX/UI was very sub optimal with XSP -- (1) it cannot be done on the fly; (2) there's no way to see how each option was set unless entering option menu; and (3) unnecessarily many taps (touches) needed even for simple changes. By contrast, X26 was much more straightforward, although it didn't look fancy.

    Anyway, after several trials and errors, I could settle sound-related parameters like the following

    Option 2 = Dither on
    Option 3 = PCM Filter MOD7 (apodizing linear filter)
    Option 6 = Jitter reducer off
    Option 7 = Dac mode async

    Speaking of PCM filter, XSP seems to use pre-defined filter sets of ES9038 (guessing from the same set of filters). Other filters sound relatively too smooth (slow rolloff ones) or touch my nerve a bit (fast rolloffs and brickwall). Apodizing filter hits hardest without serious side effects.

    ES9038's internal jitter reducing feature made everything slightly muddier. Async was more positive though. By the way, XSP's usb input was very good. It slightly and marginally outperformed external matrix usb converter (matrix XSPDIF 2).

    I will use those options for the rest of loaner period.

    With those settings, XSP sounded good in its own way. Couldn't imagine such polished and refined sounding from a Sabre dac chip. However, to me there were some gives and takes .. will elaborate them after more tests.

    Like posted in another thread, I am currently in searching for a new dac that can replace convert-2. Yes, surely the two dacs in the photo are qualified to stay in my final short list.

    PS. As an aside, hearing quite a few sabres recently, now I can comfortably argue a couple of things
    • Definitely avoid Sabre 9018 and 9028 if you don't love traditional sabre-ish-ness.
    • Also avoid 9038QM and single 9038 pro if possible -- some improvements over 9018/28 but what really improve seems to be dual-mono configuration (possibly by using multiple but differently-phased outputs to cancel out fuckeries?)
    Anyway it looks a bit strange that only Chinese dac makers utilize 9038 pro as of late, particularly considering ESS tech is a US(CA) company. Hmm..

    EDIT: After spending several hours, turned pcm filter back to mod 4 (linear slow roll-off), because to my ears it felt optimal for XSP (mod 7 is not bad but seems to lose some cohesiveness).
     
    Last edited: Apr 5, 2019
  14. Vtory

    Vtory Audiophile™

    Pyrate MZR
    Joined:
    Sep 12, 2016
    Likes Received:
    10,831
    Trophy Points:
    113
    Location:
    East Coast
    Since I got the loaner unit yesterday, literally I’ve been hearing this dac nearly all the time (surely more than >20 hours) -- because I need to make a decision for my next dac by Monday.

    Listening were done with Black Widow + Ether 2 unless specified. Mainly compared to the same dual-9038pro dac, Gustard DAC-X26, with -1 db of digital attenuation (to match XSP and X26 output levels -- possibly unfair for x26 though). Convert-2 was occasionally used when I need to “(re)calibrate” my perception.

    Also for convenience, I used coaxial spdif input for XSP because by doing so I can switch between XSP and X26 (toslink optical input used) relatively quickly. Matrix X-SPDIF 2 converted usb signal to spdif and toslink.

    In short, XSP dac is hyper impressive in that it is by far the most refined and smoothest among all ESS-based dacs. In this regard, XSP even outdoes Convert-2. Hypes were legit. So, is XSP a world-best perfection? Maybe... or maybe not. My take is XSP needs VERY GOOD digital transport to pull most from it -- will detail this more later.

    While I was very enthusiastic about X26, it still has a little digital-ness, artificialness, and roughness -- only evident when compared the two best sounding dacs on the planet (convert-2 and yggdrasil), XSP has none of them. No. Nothing. Never. In terms of smoothness, XSP nearly reaches to the level of top leaguers. Maybe due to crazy and obsessive chassis structure worked? Idk.

    Tonality is very neutral too. No hint on tipped up/down on both extreme. Ruler-flat from bottom to top. By comparison, X26 sounds very slightly tipped up on the top octaves. Timbre is superb, too. Convert-2 level accuracy and yggdrasil level tone were nicely blended.

    Staging is good for both headphones and speakers. Particularly with my Lore speakers (amped by cap-added JLH) XSP could render very holographic staging and layering. X26 is close but the main difference is that unlike x26, XSP could reproduce meaty and proper feeling of existence for each instrument, as opposed to paper-like thin sheets of x26. To me Yggdrasil A2 is the king in this criterion but XSP doesn’t lose much.

    Above are remarkable strengths of XSP. Now, let me dig into shortcomings.

    I am a foobar-first audiophile. Manage and hear digital musics primarily with Win10 pcs or laptops as digital transport. It seems that my transports induce unignorable jitters. I didn’t believe it but Convert-2 and X26’s jitter controlling features convinced me. Bypassing relevant features, my perception of jittered sound is characterized as one or combinations of (1) grayer background, (2) more haziness, (3) blurred sonic focus, (4) lacking crispness, and (5) less involving. Note that jitter perception can vary across different electronic surroundings, different lisners, and different gears. I don’t recommend to generalize by all means.

    Anyway, the thing is that XSP does NOT proactively handle such timing problems. It has two features: asynchronous clocking and and jitter eliminator (as claimed by ESS tech). Both are built in the 9038 chip, and matrix closely followed such presets. I have no idea how effective those are in measurements. But to my ears, jitter eliminator do more harms than goods -- meaningfully decrease “good signal” (energy and soul). I strongly preferred to turn off jitter eliminator. Async was not that bad. It did some job but simply not enough.

    Thus, to get a satisfying result, XSP requires users take care of transport jitters by themselves. This ultimately implies that good or great digital transport is needed for full potential. High quality cd transport and/or audiophile-grade digital stream solutions such as Allo DigiOne can be easily considered. Unfortunately I don’t own/plan either. XSPDIF (or Eitr by extension) is not enough to this job imho.

    Don’t confuse this with crappy usb inputs of some dacs like X26. Compared usb-input and xspdif-converted spdif for XSP. XSP’s usb input was marginally better, possibly due to a better converting chip of xu216?

    On top of this issue, to me XSP sometimes over-smoothen and oversimplify some kind of plankton which I believe to link to sonic texture. This is surely track-dependent. But with certain recordings, this issue was more discernible to me. Using convert-2 as a golden reference, X26 and Convert-2 showed much higher consistency than XSP-CV2.

    As an aside, during this intensive listening session, I am surprised again how effortlessly convert-2 nailed all the tests I gave to XSP and X26 (except staging).


    PS. My use of "jitter" is very naive and subjective in this post, rather than a scientifically defined/measured term. To me it's more like an alias for a set of perceptions, as listed in (1)-(5).
     
    Last edited: Apr 6, 2019
  15. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,778
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    I prefer the timbre of the Convert-2 and Yggdrasil A2 over the X-Sabre Pro, but only by a bit. X-Sabre Pro just a bit thinner, a bit grainier, less liquid. It's discernable, but not something that would haunt me as I found myself getting accustomed to it. It comes down to synergy. The titanium diaphragms of the buttcheek tweeters can be difficult with the wrong source or amp.

    On the jitter reduction feature: it sounds like shit. Reminds me of the ASRC chips of the era before. Great way to kill the life of the music. It seems like the universal recommendation is to turn this off.

    On async: use it only if your source sucks, but ultimately it's a stopgap.

    On resolution: IMO, a top-notch transport (which means CD Transport or RedNet SPDIF / AES) is necessary to get the most of the X-Sabre Pro. The X-Sabre Pro scales more than the Convert-2 in terms of plankton when using good SPDIF source - none of the smoothing / deadening effect as you mentioned. I found the XMOS USB very good, very competent. But either one of my CD transports has crushed every USB solution which I have tried, including expensive doodads such as the Berkeley Alpha and the Off-Ramp 5.

    Stage: agree with you here. Yggdrasil A2 is placed just as deep and is more holographic. X-Sabre kind of fakes being one of the Schiit A2 DACs, but hey, at least it can fake it where all others just fail.

    In terms of the classic vs. modern sound which I described in my blog, the X-Sabre has a more modern and square sound than the Convert-2, but I think overall (which still means some things better, some worse), the execution of that vision is just slightly better than that of the Convert-2.
     
    Last edited: Apr 6, 2019
  16. Highfilter

    Highfilter New

    Joined:
    Mar 31, 2018
    Likes Received:
    15
    Trophy Points:
    13
    Location:
    Vancouver, Canada
    Any impressions of using the X-Sabre Pro in PRE mode vs DAC mode? Interested in the quality of the preamp mode and how much of a compromise it is (or isn't) vs. external passive/active preamps.

    Thanks for the impression so far. Might be the next DAC I give a spin.
     
  17. Vtory

    Vtory Audiophile™

    Pyrate MZR
    Joined:
    Sep 12, 2016
    Likes Received:
    10,831
    Trophy Points:
    113
    Location:
    East Coast
    My understanding is that companies tend to call variable output with adjustable digital attenuation as "dac/pre", which is nothing to do with impedance matching, pre-gain stages, or other tasks that dedicated preamps can do.

    I never believe in hyped "transparency" of removing pre amps in the rig -- honestly never saw dac-pre worked as companies claim.
     
  18. GoodEnoughGear

    GoodEnoughGear Evil Dr. Shultz‎

    Pyrate
    Joined:
    Oct 25, 2015
    Likes Received:
    3,070
    Trophy Points:
    113
    Location:
    Cape Town, South Africa
    I think there are a few people here with active monitors who are curious what the quality of the digital attenuation is, not necessarily looking for full preamp capability with gain etc.
     
  19. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,778
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Yeah. Test quickly? Is it along the along the lines of the Convert-2's volume control which is "avoid use unless absolutely necessary" or Crane Song Solaris which is "not too shabby down to 20-bits attenuation".
     
  20. Vtory

    Vtory Audiophile™

    Pyrate MZR
    Joined:
    Sep 12, 2016
    Likes Received:
    10,831
    Trophy Points:
    113
    Location:
    East Coast
    This makes more sense.

    I briefly tested like the below:
    • -X (X={-3, -6, -20}) db of digital attenuation in XSP
    • Comparison 1: +X db of boost from amp sides so that overall SPLs remain the same (UMIK and EARS used for matching)
    • Comparison 2: -X db digital attenuation in XSP versus the same amount of attenuation in DSP vst plugin (software)

    Findings and thoughts
    • For down to -6db, digital attenuation of XSP didn't lose much. Only marginal (likely placebo) degradation compared to no attenuation. If you need, use it without too much worrying. In other words, XSP is better than Convert-2 for smaller attenuation.
    • With a decent amount of attenuation (-20 db), loss became more evident and meaningful. RME ADI-2 did this best among I tried (regardless of its sonic shortcomings, it had the best implementation of digital attenuation imho). Perceptively, attenuated (and compensated from amp side) music sounds drier and less involving, let alone a bit of micro details lost.
    • Overall, not bad. But I will probably avoid it if possible.
     

Share This Page