Massdrop x Grace SDAC Measurements

Discussion in 'Source Measurements' started by purr1n, Sep 12, 2017.

  1. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    54,321
    Dislikes Received:
    92
    Trophy Points:
    113
    Location:
    Antarctica
    Figured I'd post these since Jude is bragging about his AP555. Noise floor is higher here because I'm using a balanced probe with a /10 divider (less spurious garbage, but at the expense of moving the noise floor up 10db). I wouldn't worry about it because really, anything under 100db from the main signal... yeah, only John Atkinson cares to interpret stuff like that. It seems like a good night tonight for measurements (that's how HARD they are to do and get consistently - so many environmental factors can screw things up). I'm also running it through a Wyrd, because the Wyrd provides cleaner power and has the current capacity of the USB spec, which is something computer makers don't always meet.

    BTW 1kHz measurement here. Something not available at HF.

    SDAC 1kHz.png
    SDAC 20kHz.png
     
    Last edited: Sep 12, 2017
  2. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    54,321
    Dislikes Received:
    92
    Trophy Points:
    113
    Location:
    Antarctica
    Zoom in on 1kHz and below. Fairly clean with no 60/120Hz spikes, although I'm seen crap at 60/120Hz come and go with other DACs. Again, getting good environmental conditions tonight.

    SDAC 1kHz and below.png
     
  3. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    54,321
    Dislikes Received:
    92
    Trophy Points:
    113
    Location:
    Antarctica
    19kHz, including wideband measurement to 50kHz

    SDAC 19kHz wide.png
    SDAC 19kHz.png
     
  4. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    54,321
    Dislikes Received:
    92
    Trophy Points:
    113
    Location:
    Antarctica
    19kHz and 20kHz
    SDAC 19kHz and 20kHz.png
    [​IMG]
     
  5. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    54,321
    Dislikes Received:
    92
    Trophy Points:
    113
    Location:
    Antarctica
    -90.31db 1kHz tone. 16-bit data.
    SDAC 90db 16bit.png

    Yup, looks right to me with the steps, since -90db is a toggle of the LSB. Is there some dithering going on? (made sure it was turned off in software).
     
    Last edited: Sep 12, 2017
  6. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    54,321
    Dislikes Received:
    92
    Trophy Points:
    113
    Location:
    Antarctica
    Here's the same signal, but using 24-bit data.
    SDAC 90db 24bit.png

    This is consistent with the measurement on HF.
     
  7. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    54,321
    Dislikes Received:
    92
    Trophy Points:
    113
    Location:
    Antarctica
    Still scratching my head on the one above because the Modi 2 24-bit -90.31db signal looks like this:
    Modi 2 90db 24bit.png
    Modi 2 24-bit -90.31db 1kHz

    Wondering if this is because of something going on with the SDAC: XMOS, XMOS Windows drivers, dithering, or that the AKM4452 isn't quite as good as the AKM4490 (Modi 2). Either way, I wouldn't stress about it, unless you have a lot of hires content. And even if you do have hires content, it could be mastered like shit and sound MUCH worse than 16/44 - see here: http://www.superbestaudiofriends.or...ul-simon-diamonds-on-the-music-analysis.4050/

    Tells you how fricking deaf most audiophiles are.
     
    Last edited: Sep 12, 2017
    Lyander likes this.
  8. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    54,321
    Dislikes Received:
    92
    Trophy Points:
    113
    Location:
    Antarctica
    Lastly, here is 1kHz using the high precision Focusrite 2i2 (balanced inputs) which has a 25db lower FFT noise floor than the QA400 and /10 balanced probe. For those of you who can hear harmonics 110 to 130db below the fundamental.

    SDAC 1kHz Focusrite.png
     
    Last edited: Sep 12, 2017
    Lyander, ultrabike and Cspirou like this.
  9. atomicbob

    atomicbob dScope Yoda

    Friend BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    11,717
    Dislikes Received:
    4
    Trophy Points:
    113
    Location:
    On planet
    Given the flap about the infamous sdac 10KHz measurement I thought I'd take a minor whack at this one particular measurement. Here are my observations.

    sdac 10KHz sine measured with QA401
    20180126-01A sdac 10KHz -1dBFS 8dBu 65K FFT 11Avg QA401 WDM 96K sync gen 192KHz Analyzer.png
    M2 is an anomaly for which I don't have a good explanation yet. A hint of it may be seen in Marv's 10KHz meaurement.

    Multiway comparison QA400 (Marv) - APx555 (Jude) - dScope (atomicbob)
    20180127 Marv - Jude - atomicbob comparison.png
    sdac driver MS Windows WDM
    sdac sample rate 96 KHz
    dScope analyzer sample rate 192 KHz
    0 dBr is referenced to -1 dBFS = 7.996 dBu for dScope measurement
    FFT size 65536
    Averages 11
    Window Prism-7

    Some interesting things to note:

    1. all three measurements depict second harmonic distortion at approximately -91 ~ -93 dBr to fundamental. Guess that QA400 and QA401 are rather useful after all.

    2. QA400 and dScope measurements have relatively flat FFT noise floors. Jude's measurement contains shallow broad lobes at 5 and 17 KHz and a rather large lobe at 21.5 KHz. Hmmm.

    3. my measurement has the lowest FFT residual noise floor. Do I win the unicorn (or unreasonable facsimile)?

    If there is enough interest, I'll create a tutorial explaining the differences at a later date. PM me with expression of interest, please don't clutter this thread. I'll tally for a week or so.
     
    Last edited: Jan 29, 2018
    Psalmanazar, Lyander, trl and 5 others like this.
  10. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Friend MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,936
    Dislikes Received:
    17
    Trophy Points:
    113
    Location:
    Irvine CA
    Great job Bob! It's great to see measurements complementing each other.
     
  11. trl

    trl Acquaintance

    Joined:
    Apr 23, 2017
    Likes Received:
    40
    Dislikes Received:
    10
    Trophy Points:
    18
    Location:
    Iasi, Romania
    Seems that levels are quite off the scale: SDAC has +/-5mV and Modi2 has +/-12mV, so probably equalising the levels for both DACs might show similar graphs.

    BTW, could you please give me some thoughts about Focusrite 2i2 (or mk1 or mk2) as ADC? Can't find any measurements here on SBAF and I'm looking to get a cheap ADC for myself. Thanks a lot and sorry for the OT!
     
  12. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Friend MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,936
    Dislikes Received:
    17
    Trophy Points:
    113
    Location:
    Irvine CA
    Adding SDAC THD+N vs Vout (compared to 2i2 Line out both balanced and single ended).

    SDAC_vs_2i2_THD+N_vs_Vout.PNG

    The SDAC reaches about 0.0005% (-106 dB) THD+N around 1 Vrms output.

    I could not get the 2i2 to output about 2 Vrms SE and 4 Vrms balanced. Instead I get 0.75 Vrms and about 1.5 Vrms respectively. This might be due to their latest ASIO driver, or perhaps my computer not being able to supply enough power to the interface. The SDAC outputs 2.2 Vrms though.

    The SDAC obviously is superior to the 2i2. The 2i2 might be held back by input noise at the line driver/buffer, which may explain why it's single ended output exhibits lower THD+N than it's balanced output.

    Both systems are set to 96 kHz / 24 bits for this sweeps. Both systems are using ASIO drivers. In all cases, sweep stops at the DAC's internally defined 0 dBFS which corresponds to the last Vrms output data point in the x-axis.

    For folks bitching about the SR1 being difficult to manage and inferior to other modern analyzers in terms of ease of operation because it's not USB interfaced and so forth, bare in mind I remote login to it (ethernet), and fully control it through my POS laptop. It is a big system though.

    Note the SDAC has 4 measurements, 2 per channel. They were taken at different times and they do show one channel distorts a little more than the other above 1 Vrms output (for this particular unit). It is of little concern, since below 0 dBFS the unit exhibits exactly 0.0012% (left channel) and 0.0014% (right channel) distortion. But do note that the balanced output of the 2i2 approaches this number. It would therefore be misleading to provide 0 dBFS numbers as a final performance metric, because it would make it seem as if the 2i2 balanced output is as good as the SDAC SE output, where in fact the SDAC is a good 10 dB better across most of the operating range.
     
    Last edited: Dec 11, 2019
    Psalmanazar, RobS, purr1n and 4 others like this.
  13. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    54,321
    Dislikes Received:
    92
    Trophy Points:
    113
    Location:
    Antarctica
    Strange on the Focusrite. The specs say 15.5dBU output, which I assumed is balanced. That should be well over over 4Vrms. I am assuming you ran up to 0dbFS.

    I wonder how many facets of the 2i2 are even up to spec. I seem to clip the inputs far quicker than what the specs indicate, although I haven't confirm it with measurements.

    Yes, I prefer plots like this as opposed to the single SINAD number. Some DACs will maximize their THD+N just a bit before 0dbFS. This gives a much more comprehensive view, given how properly mastered music won't bit hitting 0dbFS regularly, with much lower average and mean levels.
     
    Psalmanazar and ultrabike like this.
  14. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Friend MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,936
    Dislikes Received:
    17
    Trophy Points:
    113
    Location:
    Irvine CA
    For the 2i2, this was the Gen 1. I suspect it is my laptop that is holding it back from hitting those numbers. The ASIO driver may define 0 dBFS as a function of how much power the USB port can give somehow. I might try another laptop to drive it.

    The reason I feel this way is that the SDAC ASIO driver through REW will also define 0 dBFS as 0.8 Vrms when using one of my usb ports and 2.2 Vrms when using another one. With the 2i2, both ports gave 0.75 Vrms as 0 dBFS.

    I could try a powered USB hub as well next.

    This does seem to indicate that performance may not be completely isolated from the computer (and port) used w the DACs.

    EDIT: Range on the 2i2 seems to be limited by the latest driver.
     
    Last edited: Feb 12, 2020
    Psalmanazar likes this.

Share This Page