Schiit Gungnir MB Measurements

Discussion in 'Source Measurements' started by atomicbob, Oct 31, 2015.

  1. atomicbob

    atomicbob dScope Yoda

    Friend BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    13,267
    Dislikes Received:
    4
    Trophy Points:
    113
    Location:
    On planet
    0150926 UPDATE - last section added comparing 1 KHz -90 dBFS Balanced vs. Single Ended between Optical, SPDIF and USB inputs.

    Look carefully at the jitter graphs. Gungnir MB jitter is impressively low. Balanced output has an exceptionally low noise floor. Single ended jitter measurement is a little higher in residual noise floor as expected.

    Channel matching is excellent. Frequency response and square wave response left and right channels are so well matched as to overlay one another.

    The 1 KHz -90 dBFS sine Single Ended measurement appears a little noisier than I would have expected and would still entertain the notion of a setup problem, though many corrective actions were attempted.

    Balanced XLR measurements were at approximately 1 hr of power on time.
    Single Ended RCA measurements were at approximately 3 hrs of power on time.
    Balanced XLR through Radial Twin ISO (Jensen transformers) for conversion to single ended output measurements were at approximately 4 hrs of power on time.

    The Gungnir MB data presented were collected as follows:

    1. PrismSound dScope III, picoscope 5243B
    2. DAC balanced output XLR, unbalanced output RCA
    3. 100 Kohm load used for measurements
    4. 44 KHz sample rate, 24 bit depth
    5. USB input – Audioquest Forest
    6. Balanced cable Canare L-4E6S with Neutrik XLR
    7. Unbalance cable DH Labs Silver Sonic Air Matrix RCA
    8. Vaunix Lab Brick USB hub

    Balanced XLR output measurements part A

    Frequency Response
    20150925 Gungnir MB FR.PNG

    Frequency Response (y-axis zoom)
    20150925 Gungnir MB FR Y axis highly zoomed.PNG

    Dynamic Range
    20150925 Gungnir MB dynamic range.PNG

    THD+N
    20150925 Gungnir MB THD THD+N.PNG

    IMD
    20150925 Gungnir MB IMD.PNG

    Crosstalk
    20150925 Gungnir MB crosstalk.PNG

    Inferred Jitter
    20150925 Gungnir MB inferred jitter.PNG

    Inferred Jitter (x axis zoom)
    20150925 Gungnir MB inferred jitter zoomed.PNG
     
    trl, auri and money4me247 like this.
  2. atomicbob

    atomicbob dScope Yoda

    Friend BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    13,267
    Dislikes Received:
    4
    Trophy Points:
    113
    Location:
    On planet
    Balanced XLR output measurements part B

    1kHz @ -90dBFS
    20150925 Gungnir MB 1 KHz -90 dBFS.PNG

    20 Hz square wave 0dBFS 6Vpp 10ms

    20150925 Gungnir MB 20 Hz sqr 6 Vpp 10mS div.PNG

    20 Hz square wave 0dBFS 6Vpp 500us
    20150925 Gungnir MB 20 Hz sqr 6 Vpp 500uS div.PNG

    1 kHz gain linearity
    20150925 Gungnir MB 1 KHz gain linearity.PNG

    Imaging
    20150925 Gungnir MB imaging.PNG
     
    trl and auri like this.
  3. atomicbob

    atomicbob dScope Yoda

    Friend BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    13,267
    Dislikes Received:
    4
    Trophy Points:
    113
    Location:
    On planet
    Single Ended RCA output measurements part A
     

    Attached Files:

    Last edited by a moderator: Oct 31, 2015
    trl and auri like this.
  4. atomicbob

    atomicbob dScope Yoda

    Friend BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    13,267
    Dislikes Received:
    4
    Trophy Points:
    113
    Location:
    On planet
    Single Ended RCA output measurements part B

    1kHz @ -90dBFS: The noise has quieted somewhat after 27 hrs of power on time. See reply #6.
     

    Attached Files:

    Last edited by a moderator: Dec 1, 2015
    trl and auri like this.
  5. atomicbob

    atomicbob dScope Yoda

    Friend BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    13,267
    Dislikes Received:
    4
    Trophy Points:
    113
    Location:
    On planet
    Balanced XLR output through Radial Twin ISO (Jensen transformers) for single ended conversion measurements part A
     

    Attached Files:

    trl and auri like this.
  6. atomicbob

    atomicbob dScope Yoda

    Friend BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    13,267
    Dislikes Received:
    4
    Trophy Points:
    113
    Location:
    On planet
    Balanced XLR output through Radial Twin ISO (Jensen transformers) for single ended conversion measurements part B
     

    Attached Files:

    trl and auri like this.
  7. atomicbob

    atomicbob dScope Yoda

    Friend BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    13,267
    Dislikes Received:
    4
    Trophy Points:
    113
    Location:
    On planet
    In this section 1 KHz -90 dBFS comparisons between USB, Optical and SPDIF for both Balanced and Single Ended are presented.

    Due to concern about the results of Single Ended measurements for 1 KHz -90dBFS and measurement setup with the dScope,
    there is also a 1 KHz -90dBFS comparison using the P5243B oscilloscope.

    P5243 oscilloscope measurements were consistent with those obtained by the dScope.

    Gungnir MB power on time was 27 hrs at time of measurements.

    Bal 1 KHz -90 dBFS optical
    1-20150926 Gungnir MB Bal 1 KHz -90 dBFS optical.PNG

    Bal 1 KHz -90 dBFS spdif BNC
    2-20150926 Gungnir MB Bal 1 KHz -90 dBFS spdif BNC.PNG

    Bal 1 KHz -90 dBFS USB
    3-20150926 Gungnir MB Bal 1 KHz -90 dBFS USB.PNG

    SE 1 KHz -90 dBFS optical
    4-20150926 Gungnir MB SE 1 KHz -90 dBFS optical.PNG

    SE 1 KHz -90 dBFS spdif BNC
    5-20150926 Gungnir MB SE 1 KHz -90 dBFS spdif BNC.PNG

    SE 1 KHz -90 dBFS USB
    6-20150926 Gungnir MB SE 1 KHz -90 dBFS USB.PNG

    Bal 1 KHz -90 dBFS USB P5243B oscilloscope
    7-20150926 Gungnir MB Bal 1 KHz -90 dBFS USB p5243b 10mV range.PNG

    SE 1 KHz -90 dBFS USB P5243B oscilloscope
    8-20150926 Gungnir MB SE 1 KHz -90 dBFS USB p5243b 10mV range.PNG


     
    Last edited by a moderator: Nov 2, 2015
    trl and auri like this.
  8. atomicbob

    atomicbob dScope Yoda

    Friend BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    13,267
    Dislikes Received:
    4
    Trophy Points:
    113
    Location:
    On planet
    Picture of test setup.

    I should have used the hot lights but was in a hurry to finish before the next scheduled event on the list by my social program director.

    Gungnir MB - dScope - picoscope 5243b

    GungnirMB_dScope_P5243b.jpg

    Gungnir MB - dScope - MSD192 for SE measurement setup verification

    GungnirMB_dScope_MSD192.jpg
     
    auri and kapanak like this.
  9. Xen

    Xen Friend

    Friend
    Joined:
    Sep 30, 2015
    Likes Received:
    205
    Dislikes Received:
    1
    Trophy Points:
    43
    Location:
    Houston, TX
    Thanks for the all the hard work in collecting and displaying the data.

    1) Please comment on the difference in amplitude for the USB vs. Optical/SPDIF for 1 KHz Balanced out signal. Approximately twice the amplitude from USB input, but also about the same waveform.

    2) SE output measurements look pretty bad to this noob. They barely resemble a sinusoid. Disconcerting... Is the noise audible since it is about +/- 0.2 mV?

    3) Those PicoScope graphs are pure data pr0n. Just saying...
     
    Thomas Crown likes this.
  10. Thomas Crown

    Thomas Crown Rando

    Contributor
    Joined:
    Sep 29, 2015
    Likes Received:
    17
    Dislikes Received:
    7
    Trophy Points:
    8
    Awesome! Thanks for sharing those measurements, please when possible answer to Xen questions, I'm interested as well.
     
  11. atomicbob

    atomicbob dScope Yoda

    Friend BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    13,267
    Dislikes Received:
    4
    Trophy Points:
    113
    Location:
    On planet
    1a) the balanced output has 6 dB more level than the SE output
    1b) look carefully at the graphs y axis labeling - the dScope scripts I used autoscale the display axis

    2) I was apprehensive that this might cause interpretation issues - for the SE output that is +/- 95 uV referenced to +/- 3 V at 0 dBFS. Please note that even at this level of -95 uV, the L & R channels nearly overlay each other perfectly.

    3) the nearly perfect representation of a square wave should indeed be cause for celebration
     
    firev1 and Thomas Crown like this.
  12. Xen

    Xen Friend

    Friend
    Joined:
    Sep 30, 2015
    Likes Received:
    205
    Dislikes Received:
    1
    Trophy Points:
    43
    Location:
    Houston, TX
    1b) I am sorry, I did not say which plots I was referring to. I meant the P5243B oscilloscope plots. I doubled checked just now and the scales look the same. All 3 are centered around -80 DBFS. BNC and Optical both have crests at -55.59 and -104.41 DBFS. The USB plot has crests at -31.17 and -128.83 DBFS.
     
  13. atomicbob

    atomicbob dScope Yoda

    Friend BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    13,267
    Dislikes Received:
    4
    Trophy Points:
    113
    Location:
    On planet
    I'm confused. You mention P5243B oscilloscope plots but the values you reference are from the dScope multi-plots. Please note that the dScope displays colored information; eg. green axis labels go with green traces.
     
  14. johnjen

    johnjen Doesn’t want to be here but keeps posting anyways

    Friend
    Joined:
    Sep 29, 2015
    Likes Received:
    614
    Dislikes Received:
    15
    Trophy Points:
    93
    Location:
    Well Grounded
    So Bob, can the dScope measure rise time of those square waves?

    And those results are VERY impressive, especially the jitter 'noise floor'.
    In fact it looks like the indicated noise is residual measurement instrument noise, although it's probably a combination of the DUT and the dScope setup. Even so with ≈ -158dB for most of the plot, and that I don't see any sideband harmonics, and the narrowness of the rise above the 'noise floor' are mighty impressive showings for a $1250 dac.

    And the channel matching, no matter which measurement is viewed, is simply amazing.

    I suppose there are other dacs that can equal (or better) these measurements but at cost factors of like x10 or more.

    I think we are seeing the results of Professor MM's engineering "Happy place" indeed.

    JJ
     
  15. Xen

    Xen Friend

    Friend
    Joined:
    Sep 30, 2015
    Likes Received:
    205
    Dislikes Received:
    1
    Trophy Points:
    43
    Location:
    Houston, TX
    Sorry, I am not used to the presentation of so much data. I was confused when you mentioned that due to perceived potential issue with the SE plots, that you had used the P5243B. I thought the plots comparing Balanced and Single ended 1 KHz -90 dBFS plots had used the P5243B.

    So the scale I should be looking at the for the 1 KHz wave is centered around "INF"? If that is the case, then the USB still appears higher amplitude with a crest of -84.29 vs -90.31 for BNC and optical.

    Here are the screencaps I am looking at:
    [​IMG][​IMG]
     
  16. atomicbob

    atomicbob dScope Yoda

    Friend BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    13,267
    Dislikes Received:
    4
    Trophy Points:
    113
    Location:
    On planet
    there is a 6 dB difference between the USB vs. Optical and S/PDIF measurements. When the dScope script runs it does an auto-alignment for optimizing display of the -90 dBFS vs 0 dBFS reference point. If this is of great concern, I can investigate the auto-scaling more closely.
     
  17. atomicbob

    atomicbob dScope Yoda

    Friend BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    13,267
    Dislikes Received:
    4
    Trophy Points:
    113
    Location:
    On planet
    The rise times of the square waves will be better measured in the Picoscope. Expanding the timebase will reveal the rise and fall times. I'm not so sure about other DACs having as closely match channel measurements, including the distortion measurements.
     
  18. Xen

    Xen Friend

    Friend
    Joined:
    Sep 30, 2015
    Likes Received:
    205
    Dislikes Received:
    1
    Trophy Points:
    43
    Location:
    Houston, TX
    Thanks! So, is the 6 DB difference an artifact in measuring?

    I am trying to understand why there would a 6 DB difference in the balanced output if given a signal from USB versus the other inputs. I can understand the difference between balanced outputs compared to single-ended outputs. However, I just expected that the balanced output would be the same for all inputs because the output signal is generated by the DAC, almost from scratch.
     
  19. firev1

    firev1 Friend

    Friend
    Joined:
    Sep 27, 2015
    Likes Received:
    39
    Dislikes Received:
    0
    Trophy Points:
    18
    Location:
    Singapore
    Normal DAC square waves be like.
    1khz square.PNG
     
    SoupRKnowva likes this.
  20. T.Rainman

    T.Rainman Acquaintance

    Joined:
    Oct 12, 2015
    Likes Received:
    62
    Dislikes Received:
    0
    Trophy Points:
    18
    You can hardly compare Atomicbobs 0dBFS 20Hz squarewave signal to a -6dB 1kHz squarewave signal.

    What would a -6dB 1kHz squarewave look like on the Gungnir Multibit or Bifrost Multibit ?
    Would be nice to see that among Atomicbob's excellent work.
     

Share This Page