Schiit Modi MB firmware v2 update technical measurements

Discussion in 'Source Measurements' started by atomicbob, Jul 9, 2018.

  1. atomicbob

    atomicbob dScope Yoda

    Pyrate BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    18,653
    Trophy Points:
    113
    Location:
    On planet
    If you are unfamiliar with audio measurements please use a search engine with the query:
    "audio measurements" or "audio measurement handbook"
    Look for publications by Richard C. Cabot and also by Bob Metzler, both from Audio Precision. There are other useful publications as well. These will provide basic knowledge.
    Interpretation of the following measurements is beyond the scope of this post.

    The data presented were acquired as follows:

    1. PrismSound dScope III, picoscope 5243B
    2. DAC balanced (Bal) output XLR and unbalanced (SE) output RCA
    3. 100 Kohm load used for measurements
    4. dScope analyzer sample rate 48 KHz unless otherwise noted
    5. DAC 44.1 KHz sample rate, 24 bit depth unless otherwise noted
    6. USB cables – Schiit Pyst and Audioquest Forest
    7. SPDIF input - Tecnec 75 ohm SMPTE 259M/292M Digital Broadcast Cable
    8. Unbalance RCA cable - Worlds Best Cable Gotham GAC-2 or DH Labs Silver Sonic Air Matrix
    9. Balanced XLR cable - Canare L-4E6S starquad with Neutrik XLR connectors
    10. Vaunix Lab Brick USB hub
    11. Shielded 14AWG and 16AWG power cables


    Schiit Modi MB firmware v2 (updated from v1)
    Modi MB - AD5547CRUZ resistor multibit 16 bit resolution DAC
    Measurements commenced after 100 hours of warmup post installing upgrade firmware IC.
    Measurements were performed over a period of 8 hours.

    Modi MB v1 firmware measurements may be found here

    Notable highlights:
    The glitch in V1 firmware previously visible at -70 dBFS is gone in V2. -90 dBFS now renders a sine.
    While V1 Gain Linearity was previously quite good for a 16 bit multiplying multibit architecture DAC,
    with V2 Schiit has set another new standard for this type of 16 bit DAC. Well done @schiit !


    WARNING - PERSONAL OPINION FOLLOWS:
    Whether either the glitch at -70 dBFS or gain linearity deviation below -70 dBFS in V1 was audible to most of the listening population is highly debatable in my opinion. However, the improvement in blackground is welcome.

    Setup picture:
    Modi MB v2 firmware small.jpg

    Index
    Post 1 - measurement setup description, highlights
    Post 1 - SPDIF input - SE outputs part A
    Post 2 - SPDIF input - SE outputs part B
    Post 3 - SPDIF input - SE outputs 1 KHz resolution series, J-Test, residual noise


    SPDIF input unbalanced output measurements part A

    THD and THD+N Spectrum
    20180708-01 Modi MB SE THD THD+N spectrum - spdif.PNG

    THD and THD+N vs Frequency
    20180708-02 Modi MB SE THD THD+N vs frequency - spdif.png

    THD and THD+N vs amplitude at 1 KHz
    20180708-03 Modi MB SE THD THD+N vs amplitude - spdif.PNG

    IMD spectrum
    20180708-04 Modi MB SE IMD spectrum - spdif.PNG

    IMD vs ampitude
    20180708-05 Modi MB SE IMD vs amplitude - spdif.PNG

    Frequency Response
    20180708-06 Modi MB SE FR - spdif.PNG

    Frequency Response (y-axis zoom)
    20180708-07 Modi MB SE FR Y axis highly zoomed - spdif.PNG

    Inferred Jitter
    20180708-09 Modi MB SE inferred jitter - 7 KHz BW - spdif.PNG

    1 KHz gain linearity
    20180708-12 Modi MB SE 1 KHz gain linearity + bandpass - spdif.png
    Very impressive Gain Linearity for a 16 bit multiplying multibit DAC architecture. Well done Schiit!

     
    Last edited: Jul 9, 2018
  2. atomicbob

    atomicbob dScope Yoda

    Pyrate BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    18,653
    Trophy Points:
    113
    Location:
    On planet
    SPDIF input unbalanced output measurements part B

    Imaging
    20180708-13 Modi MB SE imaging - spdif.PNG

    Dynamic Range
    20180708-14 Modi MB SE dynamic range - spdif.PNG

    Crosstalk
    20180708-15 Modi MB SE crosstalk - spdif.PNG

    1 KHz 0 dBu THD+N THD and spectrum
    20180708-26 Modi MB SE 1 KHz THD+N THD FFT 0dBu.png

    50 Hz + 7000 Hz dual tone
    20180708-16 Modi MB SE 50+7000Hz dual tone 3 dBu - spdif.png

    600 Hz + 1700 Hz dual tone
    20180708-17 Modi MB SE 600+1700Hz dual tone 3 dBu - spdif.png


    1kHz @ -90dBFS
    20180708-18 Modi MB SE 1 KHz -90 dBFS - spdif.PNG

    1kHz @ -70dBFS
    20180708-19 Modi MB SE 1 KHz -70 dBFS - spdif.PNG

    20 Hz square wave -3 dBFS 4Vpp 10mS/div
    20180708-23 Modi MB SE 20 Hz sqr -3 dBFS 4 Vpp 10mS div - spdif.PNG

    20 Hz square wave -3 dBFS 4Vpp 100uS/div

    20180708-25 Modi MB SE 20 Hz sqr -3 dBFS 4 Vpp 100uS div - spdif.PNG
     
    Last edited: Jul 9, 2018
  3. atomicbob

    atomicbob dScope Yoda

    Pyrate BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    18,653
    Trophy Points:
    113
    Location:
    On planet
    SPDIF input unbalanced outputs 1 KHz resolution - DAC equivalent of the limbo - how low can you go

    -100 dBFS
    20180708-27 Modi MB SE 1 KHz -100 dBFS - spdif.PNG

    -120 dBFS
    20180708-28 Modi MB SE 1 KHz -120 dBFS - spdif.PNG

    -130 dBFS
    20180708-29 Modi MB SE 1 KHz -130 dBFS - spdif.PNG

    -140 dBFS
    20180708-30a Modi MB SE 1 KHz -140 dBFS - spdif.PNG
    Remember, this is a 16 bit multiplying multibit DAC. Damn, that's impressive. Well done Schiit!

    J-Test
    20180708-41 Modi MB SE J-Test - 7 KHz BW - spdif.PNG
    Not quite the noise floor of Gungnir MB and Yggdrasil but still very impressive for a 16 bit DAC.

    Residual noise and interference
    20180709-60 Modi MB SE noise and interference - spdif.PNG
    The highest component is at the lowest frequency, 60 Hz mains. This plays to the Fletcher-Munson curve response of the Human Auditory System. Very good design trade-off. Outside of the power mains components, the remainder of the spectrum is clean and even.

     
    Last edited: Jul 9, 2018
  4. atomicbob

    atomicbob dScope Yoda

    Pyrate BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    18,653
    Trophy Points:
    113
    Location:
    On planet
    reserved 4/4.
     
  5. Elnrik

    Elnrik Super Friendly

    Pyrate
    Joined:
    Jan 10, 2017
    Likes Received:
    8,973
    Trophy Points:
    113
    Location:
    Denver CO
    Home Page:
    Thanks for your work @atomicbob!

    Are you aware of when V2 firmware started shipping? I ask because I don't know if this is a new development, or something that happened a while back.

    Thanks again!
     
  6. atomicbob

    atomicbob dScope Yoda

    Pyrate BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    18,653
    Trophy Points:
    113
    Location:
    On planet
    I don't know. I received an 8-pin DIP a while back and Jason suggested I might enjoy the results of the firmware update. I didn't get around to it for several months for reasons that can be found in my profile posts.
     
  7. SVChucko

    SVChucko New

    Joined:
    Mar 12, 2018
    Likes Received:
    11
    Trophy Points:
    3
    Location:
    Silly Valley
    Home Page:
    Wow, nice. I bought my Modi Multibit last November, wonder if it has this FW?

    I would love to see the results of USB input testing, if you find time.
     
  8. atomicbob

    atomicbob dScope Yoda

    Pyrate BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    18,653
    Trophy Points:
    113
    Location:
    On planet
    The USB input measurements are substantially similar to those found here with exception that the -70 dBFS , -90 dBFS sine and Gain Linearity look like the spdif input above. To obtain best results use an Eitr for USB to SPDIF DDC to SPDIF input on Modi MB with v2 firmware.
     
  9. Kernel Kurtz

    Kernel Kurtz Friend

    Pyrate Contributor
    Joined:
    May 19, 2018
    Likes Received:
    1,680
    Trophy Points:
    93
    Location:
    Winnipeg, Canada
    Just wondering if the firmware DIP is socketed or soldered (asking is easier than opening mine to look). I know the Modi is not marketed as being upgradable, and I'm sure that will remain a selling feature limited to their higher price point devices, but I'm curious if it is "plug and play" regardless.
     
  10. Cspirou

    Cspirou They call me Sparky

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    8,200
    Trophy Points:
    113
    Location:
    Northwest France
    Internal pics at schitt.com show that it's socketed
     
  11. Kernel Kurtz

    Kernel Kurtz Friend

    Pyrate Contributor
    Joined:
    May 19, 2018
    Likes Received:
    1,680
    Trophy Points:
    93
    Location:
    Winnipeg, Canada
    Thanks!
     
  12. Hands

    Hands Overzealous Auto Flusher - Measurbator

    Staff Member Pyrate MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    12,285
    Trophy Points:
    113
    Location:
    Colorado
    Home Page:
    I also wanted to share measurement results I got from my own Modi Multibit. I wanted to see if I had a V2 unit, which I confirmed I do, but figured I'd get a whole suite of measurements done and post here. Of course, my hardware and methods pale in comparison to what @atomicbob has and can do, yet I found it interesting and enlightening nonetheless!

    Some notes:

    - All measurements utilized the Modi Multibit's SPDIF input, with the source being an Allo DigiOne Player.
    - If in any measurements you see the noise floor creeping up as you approach the lowest octaves, that is a limitation of my ADC and not inherent to the Modi MB.

    We'll start with some basic 1KHz sines, 16-bit/48KHz, at values of -1, -3, -10, -60, -70, -80, and -90dB. This is more than I'd usually do, but I was curious to see how the Modi MB performed as the signal levels got lower and lower.

    -1dB results, with my rough THD calculations coming out to about 0.0057%.

    Modi Multibit 1KHz -1dB 16 48.png

    0Hz-1KHz noise floor result with a 1KHz, 0dB signal. Here I used both the SB1240 and 2i2 to get the best look I could with my modest hardware. PSU 60Hz spike is at a low -125dB, with the 120Hz harmonic at -130dB. Very clean results in general. Nice! I didn't post the FFT of the 0dB sine itself because my ADC didn't care for it even with gain adjusted.

    Modi Multibit 1KHz 0dB Low End Noise Floor.png

    -3dB results with a rough THD calculation of around 0.0053%.

    Modi Multibit 1KHz -3dB 16 48.png


    -10dB results with a rough THD calculation of around 0.0076%.

    Modi Multibit 1KHz -10dB 16 48.png

    FFTs of -60, -70, -80, and -90dB 1KHz sines, 16/48.

    Modi Multibit 1KHz -60dB 16 48.png
    Modi Multibit 1KHz -70dB 16 48.png
    Modi Multibit 1KHz -80dB 16 48.png
    Modi Multibit 1KHz -90dB 16 48.png
     
  13. Hands

    Hands Overzealous Auto Flusher - Measurbator

    Staff Member Pyrate MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    12,285
    Trophy Points:
    113
    Location:
    Colorado
    Home Page:
    Here's a look at the actual sine wave from some of the results above.

    16/48, 1KHz, -60dB looking fairly good here.

    Modi Multibit 1KHz -60dB Sine 16 48.png


    -70dB still looking pretty clean given the 16-bit, multibit nature of this (budget) DAC. No glitch seen here, thanks to revisions.

    Modi Multibit 1KHz -70dB Sine 16 48.png

    -80dB still more or less looks like a sine wave, with some expected noise and the like.

    Modi Multibit 1KHz -80dB Sine 16 48.png

    -90dB shows the quantization steps we'd hope to see at this level. It's a little noisy but still very clear.

    Modi Multibit 1KHz -90dB Sine 16 48.png

    Seeing as this is a 16-bit DAC, I wasn't personally expecting a 24-bit signal to make any measurable differences. And yet we can clearly see a cleaner output in both the -60dB and -90dB results.

    Modi Multibit 1KHz -60dB 24 48.png
    Modi Multibit 1KHz -90dB 24 48.png

    The -60dB sine wave itself does look a little cleaner with a 24-bit signal.

    Modi Multibit 1KHz -60dB Sine 24 48.png

    And here's where I'm stumped. This is a 24-bit, 48KHz, 1KHz, -90dB signal. The quantization steps you'd expect from a 16-bit DAC are gone, and you can see a relatively normal looking sine wave. @atomicbob Am I misunderstanding something? Because I'm not really sure how you can squeeze this sort of resolution out of a 16-bit DAC.

    Modi Multibit 1KHz -90dB Sine 24 48.png

    For kicks, here's the output of a 24-bit, 96KHz, -100dB, 1KHz sine. This is about as resolving as my ADC will get. You can still see some semblance of a sine wave here! Color me impressed...

    Modi Multibit 1KHz -100dB Sine 24 96.png
     
    Last edited: Sep 7, 2018
  14. Hands

    Hands Overzealous Auto Flusher - Measurbator

    Staff Member Pyrate MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    12,285
    Trophy Points:
    113
    Location:
    Colorado
    Home Page:
    50Hz, -3dB, 16/48 sine. Low harmonic content overall.

    Modi Multibit 50Hz -3dB 16 48.png

    19KHz and 21KHz sines at -3dB. Given the Modi MB is a 4X oversampling DAC, we shouldn't expect full alias suppression, so both results look pretty good given that and the budget nature of the DAC. Everything in the audible range is kept well under -90dB, if not mostly -100dB. (For those of you that might not be wild about full blown, 8X oversampling, the Modi and Bifrost MB offer a nice middle-ground with their 4X OS.)

    Modi Multibit 19KHz -3dB 16 48.png

    Modi Multibit 21KHz -3dB 16 48.png

    Results for dual 13/14KHz and 19/20KHz sines, -3dB. Like above, results look pretty darn good.

    Modi Multibit 13KHz and 14KHz -3dB 16 48.png
    Modi Multibit 19KHz and 20KHz -3dB 16 48.png

    Another kind of IMD test, this one being a 4:1, 250Hz and 8KHz dual sine. Results at -1dB and -3dB. Modi MB handles these without issue as well. Little to say.

    Modi Multibit 250Hz and 8KHz -1dB 16 48.png
    Modi Multibit 250Hz and 8KHz -3dB 16 48.png
     
    Last edited: Sep 7, 2018
  15. Hands

    Hands Overzealous Auto Flusher - Measurbator

    Staff Member Pyrate MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    12,285
    Trophy Points:
    113
    Location:
    Colorado
    Home Page:
    Last but not least, jitter results in the form of the classic J-Test.

    Overall, looks pretty good with the 16/48 test. Sidebands are low, noise is minimal, etc. There are a couple stray elements in there, but, again, everything as a whole is low and clean. Little to no skirting around the 12KHz line. Very clean overall.

    Modi Multibit Jitter 16 48.png

    Seeing as ARTA may or may not be able to actually pump out a real 24-bit J-Test file, I found a random one on the internet. So, this may or may not be valid.

    But, hey, most of the sidebands disappeared. As a whole, this looks even cleaner. The spike right above 10KHz may just be something related to my source. We do see an appearance of some larger sidebands closer to that 12KHz mark, which is curious but probably a non-issue.

    Overall, jitter results look pretty darn good with a solid SPDIF source.

    Modi Multibit Jitter 24 48.png


    Conclusion and findings:

    I was a bit conservative in my measurement commentary throughout. The Modi MB V2 actually measures really nicely even ignoring its multibit, 16-bit, budget price-point nature. These are genuinely solid results.

    Now, there are some areas for improvement in an absolute sense. 8X oversampling would help in some cases, but you might prefer the slightly easier going sound of 4X oversampling. The noise floor is low and clean, but you can get lower results out of a DAC, even on my modest ADC. And you could probably get "better" results overall with what better, pricier hardware brings to the table, especially with the nicer multibit chips Schiit uses in the Gungnir Multibit or Yggdrasil. But that's obvious and not a ding on the Modi MB. I think this is about as good as it gets in this category (and bests a lot of pricier DACs, objectively and subjectively).

    These results are really good in the grand scheme of things, and really, really good for a 16-bit, multibit DAC that only costs $250. Schiit really knocked it out of the park with their Modi MB revisions!

    And I'm still not sure how it seems to actually be resolving more with a 24-bit signal, given the 16-bit DAC...
     
    Last edited: Sep 7, 2018
  16. Skyline

    Skyline Double-blindly done with this hobby

    Pyrate
    Joined:
    Sep 28, 2015
    Likes Received:
    3,426
    Trophy Points:
    113
    The question in all of this is: could someone blind test a v.1 unit and a v.2 unit and tell a difference?

    I only ask because people are flipping their lids over this at the other site and begging Schiit to upgrade their v1 units (which they won't do).

    I have a v1 and think it sounds just fine, but my nervosa threshold is much higher than most others in this hobby.
     
  17. Cspirou

    Cspirou They call me Sparky

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    8,200
    Trophy Points:
    113
    Location:
    Northwest France
    Thanks for the measurements @Hands, especially the 16-bit ones since I usually only see the 24-bit measurements.
     
  18. atomicbob

    atomicbob dScope Yoda

    Pyrate BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    18,653
    Trophy Points:
    113
    Location:
    On planet
    Nice work @Hands

    Thank-you for proving that good results may be obtained with modest equipment and some knowledge of what is being measured.

    The Focusrite 2i2 is a very good ADC / DAC combination. It is worthy such that I have an HpW-works profile for it in addition to the ADI-2 Pro often used. The 2i2 is more than adequate to obtain valid results for almost every headphone amp on the market and only is challenged when measuring the best DACs and ADCs out there. Also, scripts in the dScope, APx5x5, etc make it easier to obtain amplitude and frequency sweep data, such as gain linearity and THD THD+N vs frequency. But with enough manual effort useful data may be acquired with 2i2 and graphed with a plotting program such as PSI-Plot.

    Modi MB v2 firmware is a monster. Here is what may be achieved in the way of jitter when fed a really clean signal:
    04 20180727-03 Modi MB SE inferred jitter - 7 KHz BW - MC-3+ + LiveClock + FS725.PNG

    Have a look here to see how it was done.

    20221007 edit to fix link broken by a defective person that inflicted damage on SBAF awhile back.
     
    Last edited: Oct 7, 2022
  19. Hands

    Hands Overzealous Auto Flusher - Measurbator

    Staff Member Pyrate MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    12,285
    Trophy Points:
    113
    Location:
    Colorado
    Home Page:
    Thanks! I mostly use the Creative SB1240 as the ADC, but the 2i2's noise floor doesn't creep up as much below 200Hz. That, and the SB1240 clips more easily even with software gain control. 2i2 can adapt to many more DACs with its own gain control. But when overloading the SB1240 is not an issue, it usually outperforms the 2i2. Still, combining the two helps rule out each other's weaknesses at times and gives a clearer glimpse into the true nature of whatever you're measuring, I've found.

    Any thoughts on how the Modi MB is seemingly gaining performance from 24-bit signals? I just wouldn't have thought that possible given the 16-bit nature of the DAC chip.
     
  20. Cspirou

    Cspirou They call me Sparky

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    8,200
    Trophy Points:
    113
    Location:
    Northwest France
    WOW WOW WOW!
     

Share This Page