EC Designs MOS16 DAC Measurements

Discussion in 'Source Measurements' started by purr1n, Mar 12, 2019.

  1. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    50,968
    Dislikes Received:
    84
    Trophy Points:
    113
    Location:
    Antarctica
    These are not comprehensive. I just don't have too much time. Going to take a basic set.

    0dbFS 1kHz tone
    upload_2019-3-12_19-54-29.png

    -60dFS 1kHz tone, note the -60dbFS signal is aligned to 0dbFS on the graph
    upload_2019-3-12_19-56-39.png

    -90dbFS 1kHz
    upload_2019-3-12_19-57-51.png
     
    dark_energy and monacelli like this.
  2. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    50,968
    Dislikes Received:
    84
    Trophy Points:
    113
    Location:
    Antarctica
    -100dbFS to 0dbFS sweep 31 steps ar 1kHz THD+N in db
    upload_2019-3-12_20-1-31.png

    OK, thinking out loud for the geeks. SINAD is derived from (signal+THD+N) / (THD+N). However, what is interesting is whether a "SINAD" figure with the signal at 0db is always appropriate. It might be interesting to analyze music to get a "histogram" of sorts for levels. Real music content is not always blaring with signals at 0db. Not to mention that with complex signals, it's impossible for many simultaneous signals hit 0db on an individually.

    Either way, it's good enough when we consider that 60db is the difference between a jet flyover at 1000ft (100db) and a quiet library (40db).
     
    ultrabike and monacelli like this.
  3. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    50,968
    Dislikes Received:
    84
    Trophy Points:
    113
    Location:
    Antarctica
    20-20khz sweep @ -10dbFS THD+N in db
    upload_2019-3-12_20-14-31.png

    20-20khz sweep @ -60dbFS THD+N in db
    upload_2019-3-12_20-17-26.png
     
    ultrabike and monacelli like this.
  4. Vtory

    Vtory Illogical Spock

    Friend MZR
    Joined:
    Sep 12, 2016
    Likes Received:
    3,983
    Dislikes Received:
    37
    Trophy Points:
    113
    Location:
    TX
    [​IMG]
    [​IMG]
    [​IMG]

    The three above are from Marv's airist r2r measurements. To my eyes, 0dbfs result of mos16 seems similar with that of airist. Not sure whether or not it's due to discrete ladders or cheaper multibit implementations. -90dbfs is way cleaner with airist.

    https://www.superbestaudiofriends.org/index.php?attachments/Yggdrasil A2-1khz-se-1-jpg.13121/

    Also it's interesting to see that Yggy2 result (from everlab) is definitely cleaner than the two at 0dbfs. Can't remember if (similarly priced) modi mb has been measured with the same rig.

    And surely as expected power-related low freq harmonics are impressively clean with mos16, likely due to its battery-based power. At -90dbfs, battery and switching power seem to have a certain trade-off between each other: battery having less harmonics but higher (elevated) noise floor up to 1k or so. I couldn't expect this though. Is downward-sloping noise floor a typical character of battery-power?
     
    Last edited: Mar 12, 2019
  5. monacelli

    monacelli Friend

    Friend
    Joined:
    Jul 23, 2017
    Likes Received:
    848
    Dislikes Received:
    0
    Trophy Points:
    93
    Location:
    PNW
    Cool plots, Marv. What's going on in the bass in these amplitude spectra? There is a downward sloping nature in the response as a function of frequency that I don't think I've seen with these types of measurements before. Is there some kind of low-level broad-based noise below 1 kHz that is slowly getting more and more revealed as the test tone decreases in volume? Seems possible since the slope of the response becomes more pronounced as the volume goes down. Kind of a moot point if it's not audible, and not totally surprising considering the architecture. Am I basically interpreting these correctly?
     
  6. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    50,968
    Dislikes Received:
    84
    Trophy Points:
    113
    Location:
    Antarctica
    About 10db higher noise floor at 20Hz compared to 500Hz. No idea of the cause. It does happen. I don't really see this as a problem considering the noise floor for a 0dbFS signal is 115-125db below that. Unless you can hear a pin drop on an aircraft carrier deck. I love science when it is understood properly.

    Note no 60Hz or harmonics of that from the power line.

    I would have liked to see the distortion a bit lower, but it's still 60db below the fundamental for a 0dbFS signal, which is my book is "good enough". 80db down would have been better in the case of "just in case". This is home audio for enjoyment. We are not scanning for tachyon particles emanating from time rifts or getting transporter locks were higher SNR or 0.00001% distortion would be required.
     
    Last edited: Mar 12, 2019
    yotacowboy and monacelli like this.
  7. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    50,968
    Dislikes Received:
    84
    Trophy Points:
    113
    Location:
    Antarctica
    Biodegraded and Vtory like this.
  8. Clemmaster

    Clemmaster Friend

    Friend
    Joined:
    Sep 28, 2015
    Likes Received:
    1,142
    Dislikes Received:
    1
    Trophy Points:
    93
    I wouldn't expect any measurement using frequency amplitude to show info about soundstage.

    Temporal or phase measurement might?
     
  9. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    50,968
    Dislikes Received:
    84
    Trophy Points:
    113
    Location:
    Antarctica
    1. The measurement zealots disagree with you. SINAD for a 0dbFS 1kHz sine explains everything or at the very lesst should be a primary factor.
    2. More seriously: possibly with phase, but the funky phase curves of multidriver speakers are so much worse than the slight phase shifts of DACs.
    3. Might be in the time domain, the waveforms, but what do we look for? Someone needs to invent a math transform.
     
    Last edited: Mar 13, 2019

Share This Page