Soekris dac2541 technical measurements

Discussion in 'Source Measurements' started by atomicbob, Feb 1, 2021.

  1. atomicbob

    atomicbob dScope Yoda

    Pyrate BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    18,652
    Trophy Points:
    113
    Location:
    On planet
    Here is an example FFT from the first warm-up run data collections, also highlighting the non-repeatable distortion spectrum for Loaner #13. This is one of the motivating factors prompting the second warm-up run.

    Loaner #13 dac2541 1 KHz distortion spectrum at 4 hours
    FFT_2_L1T12_1_B_13_r1_4hrs.png

    Loaner #13 dac2541 1 KHz distortion spectrum at 168 hours
    FFT_2_L1T12_1_B_13_r1_168hrs.png

    The two graphs above animated for visual differencing
    20210331 dac2541-13 THD+N r1 4 vs 168 hrs slow.gif

    While the above focused has been on anomalies, it is extremely important to keep in mind most other measurements, spectrums, graphs, etc. demonstrated very little difference through the series of data collections from either the Lab unit or Loaner dac2541. The two behaved more similar than different.

    Here is an animation of THD+N vs Amplitude for the Loaner during the second warm-up data collection.

    dac2541-13 THD+N vs Amplitude warm-up
    20210328 dac2541-13 THD+N vs Amp warmup.gif

    No significant changes in THD+N vs Amplitude over 264 hours of data collection.
    This is just one facet of the enormous amount of data collected.
     
    • Like Like x 2
    • Epic Epic x 1
    • List
    Last edited: Apr 2, 2021
  2. atomicbob

    atomicbob dScope Yoda

    Pyrate BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    18,652
    Trophy Points:
    113
    Location:
    On planet
    12 data collection reports for Lab dac2541 and 16 data collection reports for Loaner #13 are attached in the following posts. They are labeled by YYYYMMDD-Time of data collection.

    Lab unit: dac2541
    example:
    20210314-1230_dac2541_AES_44K_Bal_Lin_160dBFS.pdf

    Loaner #13: dac2541-13
    example:
    20210318-0900_dac2541-13_AES_44K_Bal_Lin_160dBFS.pdf


    Conclusion
    Soekris dac2541 may benefit from an initial burn-in of 96 to 168 hours. After the initial burn-in no notable measurement changes are observed after 1 hour of warm-up.
     

    Attached Files:

    • Epic Epic x 8
    • Like Like x 2
    • List
  3. atomicbob

    atomicbob dScope Yoda

    Pyrate BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    18,652
    Trophy Points:
    113
    Location:
    On planet
    Continuation of attached reports 2/3.
     
  4. atomicbob

    atomicbob dScope Yoda

    Pyrate BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    18,652
    Trophy Points:
    113
    Location:
    On planet
    Continuation of attached reports 3/3.
     
  5. Biodegraded

    Biodegraded Friend

    Pyrate Contributor
    Joined:
    May 28, 2017
    Likes Received:
    7,976
    Trophy Points:
    113
    Location:
    Vancouver BC
    @soekris am I correct to presume the slow min filter works essentially the same on your older dacs - so upsampling 44.1 kHz content to 88.2 before feeding it to my dac1321 will give the same benefits as illustrated above?

    @atomicbob I know you have other things to do - not least attempting to retire and listening to music - but if you have time and if it wouldn't just repeat what you've shown here, it'd be interesting to see the filter characteristics of the dac1421. It might require a firmware update to be comparable though.
     
  6. soekris

    soekris MOT - Soekris Engineering

    Pyrate
    Joined:
    Mar 25, 2016
    Likes Received:
    954
    Trophy Points:
    93
    Location:
    Denmark
    Yes, most of my DACs use the same filters, assuming newest 4K filter versions.
     
  7. atomicbob

    atomicbob dScope Yoda

    Pyrate BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    18,652
    Trophy Points:
    113
    Location:
    On planet
    I saw your request a few days back and began warming up the 1421. This dac, like the 1541 and 2541, presents a measurement problem for me ... it renders music in such an engaging way I forget about measurements and want to spend time enjoying the music. Similar issues with schiit MB dacs and Holo KTE Spring2 for me.

    Listening and measurement setup (dScope on bench above and not in this picture):
    DSC_0678_small.jpg
    The small combined footprint of dac1421 and SW51+ easily demonstrate high quality sound does not require cement mixer sized components.

    The filters in my dac1421 are the original that shipped with the unit when first released. No updates have occurred.
    Keep in mind the following FFT have bandwidth to 500 KHz, not the usual 20 KHz.

    dac1421 40 Hz sine FFT -20dBFS 44K spdif SE Slow
    20210801 dac1421 40 Hz sine FFT -20dBFS 44K spdif SE Slow.png
    Using a large number of averages, FFT indicates primary 44 KHz component is 49 dB lower than the 40 Hz stimulus.

    dac1421 40 Hz sine FFT -20dBFS 88K spdif SE Slow
    20210801 dac1421 40 Hz sine FFT -20dBFS 88K spdif SE Slow.png
    At 88KHz FFT indicates highest component is 67 dB lower than the 40 Hz stimulus, 18 dB lower than the highest spike when the dac is run at 44 KHz.

    dac1421 40 Hz sqr 100us div -20dBFS 44K spdif SE Slow
    20210801 dac1421 40 Hz sqr 100us div  -20dBFS 44K spdif SE Slow - no markers.png
    Some ripple is visible in the time domain at 44 KHz

    dac1421 40 Hz sqr 100us div -20dBFS 88K spdif SE Slow
    20210801 dac1421 40 Hz sqr 100us div  -20dBFS 88K spdif SE Slow - no markers.png
    At 88 KHz the ripple is reduced significantly and again the filter ringing is shortened compared to running the dac at 44 KHz.

    As many dacs behave similarly I have standardized on running my Dante network at 88 KHz.
     
  8. Biodegraded

    Biodegraded Friend

    Pyrate Contributor
    Joined:
    May 28, 2017
    Likes Received:
    7,976
    Trophy Points:
    113
    Location:
    Vancouver BC
    Thanks Bob. The soft min filter has been my preference on the two Soekris DACs I've heard (SBAF loaner dac1421 in early 2019 and dac1321 purchased from @Melvillian early this year) and >95% of my digital music files are 44.1, so Volumio on the Pi3B+Pi2AES is now set to resample to 88.2 kHz. Not sure I can convince myself I hear a difference, but it's satisfying to know the filter behaves better that way!
     
  9. Biodegraded

    Biodegraded Friend

    Pyrate Contributor
    Joined:
    May 28, 2017
    Likes Received:
    7,976
    Trophy Points:
    113
    Location:
    Vancouver BC
    Necro to briefly report: using PlayPcmWin shared mode +/- resampling -(USB)-> dac1321 -> PassDIY WHAMMY -> modded Yamaha YH-2s, I do hear a difference with oversampling vs not with good low frequency content. For example, the kick drum hits in the first few seconds of Slowdive's No Longer Making Time seem to be broadened (up as well as down, but down is most noticeable) and more rounded or perhaps dynamically reduced (I admit to often having trouble differentiating between transient speed and amplitude) when oversampled vs played at 44.1 kHz (FLAC ripped from CD).

    Whether this is consistent with the reduced HF artifacts and reduced filter ripple & ringing @atomicbob showed above I dunno; but my new default playback mode to the dac1321 is with 4x oversampling.
     

Share This Page