Audio-GD Singularity 19 DAC - Measurements

Discussion in 'Source Measurements' started by Hands, Mar 23, 2017.

  1. Hands

    Hands Overzealous Auto Flusher - Measurbator

    Staff Member Friend MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    11,051
    Dislikes Received:
    39
    Trophy Points:
    113
    Location:
    Colorado
    Home Page:
    First, a few notes. This was via SPDIF/BNC, from the Soundaware D100, which has been great with my other measurements. My measurement methods were consistent from the other DACs I've measured on SBAF. I had to turn the gain up a bit on my ADC, which can raise the noise floor, but to minor levels. Lastly, I tried all sorts of jumper/DSP variations. Any oddities seen in these measurements were generally shared across various jumper settings. Measurements were taken with default jumper setting (i.e. all of them removed).


    1KHz, -3dB, 16/48. Harmonic distortion results could look better. This is a bit high. However, that doesn't concern me. I've heard DACs with similar distortion that didn't seem to affect the sound subjectively. What concerns me is the skirting around the 1KHz area and additional noise in that area and in the lower end. Ideally, the noise floor would be flat, and the 1KHz would be a single line rising out of that, top to bottom.

    1K -3dB Linear.PNG


    Same as above, just extended to 48KHz.

    1K -3dB Linear 96K.PNG


    Same 1KHz signal, but showing the 1-1000Hz region. While the noise floor is fairly low overall, you can see some spikes in the area as well as weird artifacts around 1KHz. The spikes in the lower area are consistent in most measurements. Usually you want to see this be very clean in DACs.

    1K -3dB Low End.PNG


    1KHz down at -60dB. Again, harmonic distortion could be better, but this is a common 16-bit result, I suppose. Noise floor is about as low as the ADC will go. Cleaner looking area around 1KHz. Still, some noticeable noise in the lower end that gradually decreases until subsiding around 3KHz. This noise could affect the resolution of low-level signals, possibly.

    1K -60dB.PNG


    Something odd I noticed. This is the sine wave view of that 1KHz, -60dB signal. You might see the wave has some wiggle spots in it, some narrow and some broad. I've not seen this before on a -60dB signal. Weird.

    1K -60dB Sine.PNG
     
    Last edited: Mar 23, 2017
    pedalhead, Scott Kramer, Jeb and 10 others like this.
  2. Hands

    Hands Overzealous Auto Flusher - Measurbator

    Staff Member Friend MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    11,051
    Dislikes Received:
    39
    Trophy Points:
    113
    Location:
    Colorado
    Home Page:
    1KHz, but this time at -90.31dB. At this point, that noise really might get in the way. Harmonic distortion results of this magnitude are common and often expected with 16-bit signals at -90dB.

    1K -90dB.PNG


    This SHOULD look like three discrete steps (see my other measurements or Stereophile stuff for an example). Instead, we have something that looks like a sine wave, triangle wave, and noise all had a baby together. Effectively, this means the S19 isn't resolving a low-level, 16-bit signal in this case, whether it be due to DAC module limitations or the noise issues.

    1K -90dB Sine.PNG


    Here is the same test, but with a 24-bit signal. We do see the harmonic distortion drop, but not as much as some other DACs.

    1K -90dB 24bit.PNG


    This should look like a sine wave, perhaps a bit noisy. Instead, it looks a lot like a triangle wave. It's a bit cleaner than the 16-bit result. This DAC is definitely not resolving that 24-bit signal.

    1K -90dB 24bit Sine.PNG


    Here's something interesting. This is the 19KHz sine wave test, at -3dB. It looks like some futuristic landscape, with tall, skinny buildings and domes for space folk to live in. Of course, this great city has a pyramid-like, skyscraper monolith dominating the landscape.

    Seriously, though, that pyramid stuff is bizarre, as is the noise floor. Overall harmonic distortion results are OK beyond that.

    19K -3dB.PNG


    Same thing as above, just extended out to 48KHz.

    19K -3dB 96.PNG


    Let's zoom in on that pyramid. What the heck is going on here? (Serious question.)

    19K Pyramid.PNG
     
    Azteca, pedalhead, landroni and 6 others like this.
  3. Hands

    Hands Overzealous Auto Flusher - Measurbator

    Staff Member Friend MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    11,051
    Dislikes Received:
    39
    Trophy Points:
    113
    Location:
    Colorado
    Home Page:
    19KHz and 20KHz IMD test. 1-to-1 ratio at -3dB each. The IMD results aren't awful. Noise floor looks better than with a single 19KHz sine wave. Still got that weird pyramid action going on.

    13K 14K.PNG


    Same thing as above, but extended to 48KHz.

    19K 20K.PNG


    250Hz and 4KHz, 4-to-1 ratio. Based on the above, nothing too unexpected.

    250Hz 4K.PNG


    60Hz sine wave, at -3dB. Harmonic distortion is actually surprisingly high on this one.

    60Hz -3dB Zoom.PNG
     
    Azteca, landroni, Torq and 3 others like this.
  4. Hands

    Hands Overzealous Auto Flusher - Measurbator

    Staff Member Friend MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    11,051
    Dislikes Received:
    39
    Trophy Points:
    113
    Location:
    Colorado
    Home Page:
    Jitter test. J-test at 16/48. More of that weird pyramid stuff going on. Noise floor raises overall in this test. Some dirtiness, but otherwise none of the usual jitter spurs, which may or may not be a good thing. Usually you expect them with 16-bit tests, but maybe there would be a positive reason for them to not show? Remember, default jumper settings are NOT bypassing DSP, but do disable dither and PLL. 8X OS mode.


    Jitter 16bit.PNG


    Same test at 24/48. Very little change.

    Jitter 24bit.PNG


    In this test, I set it to non-oversampling mode and DID bypass the DSP. Dither and PLL still off. Jitter spurs pop up, which comforted me and will help me sleep tonight. Still, the spurs are at a fairly high level. Not outrageous, but should be better. Still got that pyramid thing going on.

    Jitter 16bit NOS Bypass.PNG


    This time, I had NOS mode, bypass on, dithering on, and PLL on. Not much change.

    Jitter 16bit NOS Bypass DSP PLL both on.PNG



    I tried a couple more jumper variations, and not once did that pyramid action stop.
     
    Azteca, landroni, Torq and 5 others like this.
  5. eastboundofnowhere

    eastboundofnowhere Facebook Friend

    Contributor
    Joined:
    Mar 28, 2016
    Likes Received:
    181
    Dislikes Received:
    1
    Trophy Points:
    43
    [​IMG]

    Sorry, was supposed to be the monolith from 2001. Will figure out images.
     
  6. Hands

    Hands Overzealous Auto Flusher - Measurbator

    Staff Member Friend MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    11,051
    Dislikes Received:
    39
    Trophy Points:
    113
    Location:
    Colorado
    Home Page:
    Overall, the Audio-GD Singularity 19 has some...odd measurement behaviors. The harmonic distortion can be a bit high at times, but usually that's not a deal breaker. What concerns me is the skirting and pyramid mess going on around signals. I don't know what that is exactly, what causes it, or even if it's that big of a deal, but it's odd. The noise spurs should be lower for a DAC, I think. The DAC can't seem to fully resolve a low-level 16-bit 1KHz signal, let alone 24-bit. I think some of Schiit's lower-end, multibit offerings may also suffer in that area, but I wouldn't be surprised if the S19 issues come down more to the DAC modules being used vs. the Schiit stuff.

    The S19 does pretty well in many measurement areas. But there are too many oddities all over the place, to the point it's concerning.
     
    Last edited: Mar 23, 2017
    landroni and Dino like this.
  7. Rockin_Zombie

    Rockin_Zombie Facebook Friend

    Joined:
    Jan 2, 2017
    Likes Received:
    129
    Dislikes Received:
    1
    Trophy Points:
    43
    Location:
    Toronto
    [​IMG]

    Very disappointing results. They push too many products out the door in too little time, also one guy doing all the design, perhaps actively building too. Was looking forward to an original chip.

    While we are dissecting the Singularity 19, they have release a plethora of new DACs with ES9038PRO.

    [​IMG]

    Hope you guys can return it, judging by your "crappy" review I wonder how the loaner trip will go.
     
  8. SSL

    SSL Friend

    Friend
    Joined:
    Nov 12, 2015
    Likes Received:
    1,159
    Dislikes Received:
    14
    Trophy Points:
    93
    Single-ended Gungnir Multibit, Bifrost Multibit, and Modi Multibit all have problems with resolving that -90dBFS signal. Doesn't seem too unusual.
     
  9. Hands

    Hands Overzealous Auto Flusher - Measurbator

    Staff Member Friend MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    11,051
    Dislikes Received:
    39
    Trophy Points:
    113
    Location:
    Colorado
    Home Page:
    In the Gungnir Multibit's case, I think this would mean that's on the output stage if it's fine via XLR, not the actual DAC portion. Not sure if same is true for Bifrost Multibit and Modi Multibit, but they have decent 16-bit chips, to the point I'd think the chips themselves would have no issues resolving the -90dB signal.

    In the S19's case, I'd be curious to rule out the output stage and see if that's limiting the resolution of this signal or if it's the DAC modules themselves.

    It's not my main area of concern. Lots of other weird stuff that has me scratching my head. That and the DAC sounds like garbage regardless of measurements, which is all that really matters. The Schiit stuff is actually good.
     
  10. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    64,224
    Dislikes Received:
    108
    Trophy Points:
    113
    Location:
    Antarctica
    Dude, you are doing the super polite act.

    The 1kHz -3db THD is complete shit for DAC. 0.11% - that's shit, even for some tube amps, lol, and even worse than some headphones; and what makes it all the more shittier is the emphasis on the odd order harmonics. Not only that, but the fact that each harmonic gets lit up all the way to 20kHz. I don't fuss about odd order harmonics being worse than even order (to me, both sound bad), but the fact that the harmonics don't die quickly after the 3rd or even 4th, but keep going on forever is of concern (it's a bad sign from other gear where I've seen even lower levels of this). The THD is essentially garbage.

    This distortion pattern is also seen in the lower level signals. BTW, the triangle stuff is expected from the distortion characteristics. Triangle waves are the main signal with a lot of odd order harmonics that get lower over time.

    Dunno about the sidebands - the pyramid stuff. Some kind of fuzzy jitter or maybe the result of a funky custom digital filter. Wouldn't want to speculate since I've never seen it before.

    Why I do get the feeling this is going to sound like garbage?
     
    Delayeed, Azteca, landroni and 3 others like this.
  11. Hands

    Hands Overzealous Auto Flusher - Measurbator

    Staff Member Friend MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    11,051
    Dislikes Received:
    39
    Trophy Points:
    113
    Location:
    Colorado
    Home Page:
    Oh, oops! Posted in a hurry and thought there was a zero after the decimal for distortion. 0.1% is pretty bad for a DAC indeed.
     
  12. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    64,224
    Dislikes Received:
    108
    Trophy Points:
    113
    Location:
    Antarctica
    I remember the same 1kHz distortion pattern on the Metrum Quad, but those spikes were a magnitude or two lower.
     
  13. Hrodulf

    Hrodulf Prohibited from acting as an MOT until year 2050

    Friend
    Joined:
    Sep 28, 2015
    Likes Received:
    3,298
    Dislikes Received:
    5
    Trophy Points:
    113
    This looks like a R2R ladder DAC, just not a signed magnitude one.
     
  14. Hands

    Hands Overzealous Auto Flusher - Measurbator

    Staff Member Friend MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    11,051
    Dislikes Received:
    39
    Trophy Points:
    113
    Location:
    Colorado
    Home Page:
    The old Metrum stuff was around 0.01% for the good stuff. Quad was a bit higher. Better harmonic distortion distribution. They didn't handle high level signals well or upper treble either.

    They also had a clean and low noise floor. Could resolve a low level 1k signal at least, albeit the actual chips and subjective characteristics were more like 13-bits. And when listening, they really suffered with precision and resolution. Not worth buying today, but kind of a unique vision a few years back that manifested into their new stuff.

    The S19 is just sloppy, shitty garbage. An actual enigma of what-the-fuckery.
     
    lm4der likes this.
  15. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    64,224
    Dislikes Received:
    108
    Trophy Points:
    113
    Location:
    Antarctica
    INL.png

    I think I know what the consistent squiggles are: severe INL (Integral Non-Linearity) issues. That they happen on the same spots of the sine waves would be expected for a ladder DAC. INL / DNL and DAC accuracy is not often talked about in rags like Stereophile. INL should not be confused with the linearity type measurements commonly seen on Stereophile.

    It has to do with resistors not being precise enough, so digital numbers don't get properly translated to the correct output levels. For example, 5 should translate into 0.5 volts, 6 into 0.6, 7 into 0.7, and so on. This doesn't happen in real life, and what happens is that 5 might translate into 0.49 volts, and 6 into 0.61, and 7 into 0.63.

    For us to even see this behavior is not a good sign, although being a -60db signal, it does make it easier to see, being 5-6 bits of signal (32 to 64 distinct values of voltage levels). This DAC will resolve like shit with a good amount of distortion. It's likely it may even sound distorted.
     
    Last edited: Mar 23, 2017
    Azteca, Psalmanazar and landroni like this.
  16. Hrodulf

    Hrodulf Prohibited from acting as an MOT until year 2050

    Friend
    Joined:
    Sep 28, 2015
    Likes Received:
    3,298
    Dislikes Received:
    5
    Trophy Points:
    113
    What's the warranty on this thing like? Obviously this isn't the product promised here.
     
    brencho likes this.
  17. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    64,224
    Dislikes Received:
    108
    Trophy Points:
    113
    Location:
    Antarctica
    They didn't send you the 220V version did they?
     
  18. Hands

    Hands Overzealous Auto Flusher - Measurbator

    Staff Member Friend MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    11,051
    Dislikes Received:
    39
    Trophy Points:
    113
    Location:
    Colorado
    Home Page:
    Don't think so. Will double check to be sure.
     
  19. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    64,224
    Dislikes Received:
    108
    Trophy Points:
    113
    Location:
    Antarctica
    Make sure the R-2R DA M1 modules have the QC stickers on them.
     
  20. Clemmaster

    Clemmaster Friend

    Friend
    Joined:
    Sep 28, 2015
    Likes Received:
    1,885
    Dislikes Received:
    4
    Trophy Points:
    113
    Nice pyramids.
     

Share This Page