Hacking Schiit

Discussion in 'Modifications and Tweaks' started by Marvey, Jul 22, 2021.

  1. ergopower

    ergopower Friend

    Friend
    Joined:
    Mar 21, 2018
    Likes Received:
    452
    Trophy Points:
    63
    Location:
    South Central PA
    Haven't done any DIY in a while, and for some reason the opamp swap intrigues me, since I have a Modi MB V2.
    Nobody has stock of the 1656, but Mouser has 2156.
    I ordered 10 since I just couldn't bring myself to pay $8 shipping for one $3 chip.

    So here's the deal - any US Friend that wants to do a swap, let me know and I'll mail you one.
    Let's say 8 available, I'll hang on to an extra. I'll do it in order of PMs received.
     
    • Like / Agree Like / Agree x 6
    • Epic Epic x 1
    • List
  2. Marvey

    Marvey Super Friend

    Staff Member Friend BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,679
    Trophy Points:
    93
    Fun stuff. Will move to a Hacking Schiit thread tonight and include the usual disclaimers: you may void your warranty, don't melt the parts and send them back, etc.

    I have many stories. There was one time where an EC ZD came back with carbon resistors (not used in the design)... but on one channel.

    Will take measurements of OPA1656 replacements of the DACs. Will also circle points on Magni 3+ board where it can be used an I-V/output stage for NOS DIY DAC jobs.
     
    • Like / Agree Like / Agree x 9
    • List
  3. spoony

    spoony Spooky

    Friend
    Joined:
    Sep 28, 2015
    Likes Received:
    502
    Trophy Points:
    93
    Anyone knows the difference between OPA1656IDR and OPA1656ID?, I can only get the latter here.
     
  4. Hands

    Hands Overzealous Auto Flusher - Measurbator

    Staff Member Friend MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    12,053
    Trophy Points:
    113
    Location:
    Colorado
    Home Page:
    I personally use the 2156 in my Modi/Bifrost 1 MB because of its tighter noise and offset specs, faster slew rate (probably doesn't matter), and because it seemed to measure ever so slightly better than the 1656. Slight enough to be within margin of error. They more or less sounded the same.

    Ultimately, I felt better going for the more precision-oriented part rather than the one marketed for audio. I mean, the DAC IC itself is marketed as a precision DAC, not one for audio.

    But either should work essentially the same in this use case.
     
    • Like / Agree Like / Agree x 2
    • List
  5. Hands

    Hands Overzealous Auto Flusher - Measurbator

    Staff Member Friend MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    12,053
    Trophy Points:
    113
    Location:
    Colorado
    Home Page:
    Datasheet makes me think it's about the quantity/packaging in which they sell them to distributors.
     
    • Like / Agree Like / Agree x 1
    • List
  6. ergopower

    ergopower Friend

    Friend
    Joined:
    Mar 21, 2018
    Likes Received:
    452
    Trophy Points:
    63
    Location:
    South Central PA
    I forget which is which, but they come in either tape, tube or reel. If you're just ordering one, it won't matter.
     
    • Like / Agree Like / Agree x 1
    • List
  7. spoony

    spoony Spooky

    Friend
    Joined:
    Sep 28, 2015
    Likes Received:
    502
    Trophy Points:
    93
    Thanks!
     
  8. Marvey

    Marvey Super Friend

    Staff Member Friend BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,679
    Trophy Points:
    93
    One thing about the hair dryers. If you need to get to a part jammed next to other parts (e.g. Modi MB is packed dense on both sides), then a hair dryer is going to blow off more than the part that you intended. Sometimes it's easier to wick off solder (get super wick) and pry off. If you like to live dangerously, you can also apply a big solder blob to the part if it is small, stick it to the iron, and fling it over your head. (You don't want to know where I learned this shit).

    Flux and drag solder with the iron also works. I haven't needed to use any flux, but I am sure it would make things even easier - just need to be careful on boards packed densely.

    Bottom line is that everybody develops their own bag of tricks, does stuff differently, and practice makes perfect.
     
    • Like / Agree Like / Agree x 4
    • List
  9. Marvey

    Marvey Super Friend

    Staff Member Friend BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,679
    Trophy Points:
    93
    Ding ding ding.

    The OPA1656 will sound like shit (or do weird shit) if you use it in the wrong place: for example, the AKM 4493 DACs (seems as we go up the scale, they sound brighter). I can't imagine the OPA1656 with an ESS unless some special whisperer secrets were applied. I pick opamps specifically based on their sound from an subjective point of view, ther datasheets. And of course measurements will confirm that the part worked.

    I had a vision in mind with the Modi 3+ and the OPA1656 happened to work and achieve this vision. If one happens to like the sound of the Modi 3+ as it is, it's probably they will detest the swap to the OPA1656. In this case, it's really more different rather than better.
     
    • Like / Agree Like / Agree x 2
    • List
  10. Martigane

    Martigane Rando

    Joined:
    Mar 5, 2019
    Likes Received:
    51
    Trophy Points:
    13
    Location:
    Belgium
    Thanks Marvey.
    Live and learn, I guess.

    I have tried 2156 in my DIY headphone amp in a non-inverting voltage gain stage.
    Subjectively, I found it a bit sterile but technically good. It was just not really involving. Not as harsh/cold/strident as LME49720 though.
    I preferred the OPA1642, and even more so the more V-shaped (but also a lot more expensive) OPA2107.
    I heard many like the OPA1612, but I've never tried it.

    Interestingly, this OPA2156 was the only Op Amp to show an odd behavior at higher playback levels of 10kHz square wave (checked on scope, really zoomed in). All other 8 Op Amps tested did not have it:
    https://imgshare.io/image/opa2156.p0NzPl


    Not sure which Op Amp is the better fit for the Modi3 application, but tweaking is fun :)
     
  11. Marvey

    Marvey Super Friend

    Staff Member Friend BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,679
    Trophy Points:
    93
    As promised, here are some Modi MB measurements with the OPA1656. Unfortunately I didn't take measurements beforehand. It would appear that I have a Modi MB v1 or maybe 1.1 or 0.9 (there are always silent revisions). Note that the v1 does not measure as good as the v2. The v2 uses some tricks and fancy math to get better measurements. Funny how the same data can measure much better! Lots of things we can do, offset, noise-shape, dither, trade frequency for bits, etc.

    Anyway, I believe this DAC is most comparable to the v1. I think @atomicbob has some Modi MB v1 measurements on SBAF for comparison. Word of warning: this is a ancient MB DAC using a lot of NOT state-of-the-art parts. If you like SINAD do not buy this DAC.

    Personally, I'm glad that I grabbed this and put it back on my computer desk after "acclimating" to various other DACs I had available, all delta-sigma. (Lots of random stuff I'm finding as I slowly unpack from the move to TX). More and more I an disliking delta-sigma DACs. We can read about @k4rstar rail against delta-sigma DACs here: https://www.superbestaudiofriends.org/index.php?threads/delta-sigma-an-inconvenient-truth.9480/. Although @k4rstar favors NOS DACs and I do not, I think we can both agree that delta-sigma sucks.

    I think Schiit is certainly due for an update on a sub $300 multibit DAC - but at least this hack, as I mentioned earlier, breathes new life into it. All my prior subjective notes stand and @Hands has offered his two cents.

    Schiit Modi Multibit v1
    OPA1656 in I/V Hack
    0dbFS 24-bit 48kHz
    upload_2021-7-31_8-38-51.png
    Note that V2 gets minimizes the higher harmonic spikes, but trades them off with rising noise. FWIW, I felt the opamp swap made a much bigger difference sonically compared to the V2, which from I understand was mostly a firmware trick to do different maths on the signal.

    Schiit Modi Multibit v1
    OPA1656 in I/V Hack
    -70dbFS 24-bit 48kHz
    upload_2021-7-31_8-44-5.png
    upload_2021-7-31_8-45-8.png

    Schiit Modi Multibit v1
    OPA1656 in I/V Hack
    -90dbFS 24-bit 48kHz
    upload_2021-7-31_8-54-2.png
    upload_2021-7-31_8-51-18.png

    Now we know what those harmonics are - they represent the steps for a multibit implementation. Remember, this is a 16-bit part! Oh, not a perfect sine wave huh? When was the last time you listened to a hires 24/96 track on Qobuz and heard differences from the 16/44? I have to yet hear any difference between HiRes and CD quality - probably because the they were sourced from the same 16/44 master. Hahaha. Seriously, Daft Punk Random Access Memories in 24/96 is a waste of bandwidth.
     
    • Like / Agree Like / Agree x 7
    • List
    Last edited: Jul 31, 2021
  12. Marvey

    Marvey Super Friend

    Staff Member Friend BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,679
    Trophy Points:
    93
    OK, now let's do a trick:

    Instead of a 24-bit input, let's use a 16-bit signal for input, but dither it!

    Schiit Modi Multibit v1
    OPA1656 in I/V Hack
    -90dbFS 16-bit 48kHz dithered
    upload_2021-7-31_8-56-33.png

    Wow, look ma! What happened to those harmonic spikes?

    Audio scientists are getting wet! Note that the v2 takes a similar approach, but the noise shaping may be less drastic and more advanced - shaped so it rises in frequency (I'm just guessing what the Schiit engineers would do). Note that the noise floor has jumped from below -140db to just below -120db, but the harmonic spikes (some of which hit -110db -100db) are now totally gone. Math is fun!

    What we've done is traded a form of multibit related distortion for noise. Does it really matter? Maybe. Maybe not. I'm not even sure we can hear this. To put things in respective, CD quality has a limitation of about 96db of dynamic range (approximately 6db for every bit). So if we think about things this way, this is how we should look at it:

    Schiit Modi Multibit v1
    OPA1656 in I/V Hack
    -90dbFS 16-bit 48kHz dithered
    upload_2021-7-31_9-3-45.png
     
    • Like / Agree Like / Agree x 4
    • Epic Epic x 1
    • List
  13. Lyander

    Lyander Too sensitive for SBAF

    Friend
    Joined:
    Feb 25, 2017
    Likes Received:
    8,234
    Trophy Points:
    113
    Location:
    Pasig, Philippines
    Minor typo I'm guessing. Also, thank you for saying high res isn't entirely worth all the hype. I have a few sampler freebies as reference tracks "just in case" but for the most part I continue to stan Redbook.

    Just out of curiosity has there been a DAC in recent years that actually had too high a noise floor that it caused proper issues in the listening experience?

    Dumb question but how do the harmonic spikes correlate with the "shape" of the reproduced sime wave? Far out of practise using my brain for this sorta thing.
     
  14. Marvey

    Marvey Super Friend

    Staff Member Friend BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,679
    Trophy Points:
    93
    I have a few good tracks where hires does sound marginally betterer. For the most part, the stuff on Qobuz or Tidal, I have heard zero difference. For there to be a difference, I think someone would have to find the original tapes and run them through a good ADC and maybe remaster them again. Basically a boutique job - a labor of love. The problem is that of lot of these ancient tapes have degraded. And if we are stuck with bad early digital, it's just bad digital. I don't hires matters with modern sound production because the stuff is so processed and mixed, often in Pro Tools. There no money in music anymore, unless you are big name.
     
    • Like / Agree Like / Agree x 1
    • List
  15. M3NTAL

    M3NTAL Friend

    Friend
    Joined:
    Sep 27, 2015
    Likes Received:
    992
    Trophy Points:
    93
    Location:
    Arizona
    I have to remember that the IEMagni is specific for my BA's - "don't get any ideas mental"

    Do you prefer the more neutral sound with the face tweeters also?
     
  16. spoony

    spoony Spooky

    Friend
    Joined:
    Sep 28, 2015
    Likes Received:
    502
    Trophy Points:
    93
    Any tricks to make the Jot 2 'edginess' go away?
     
  17. Senorx12562

    Senorx12562 Case of the mondays

    Friend
    Joined:
    Oct 26, 2015
    Likes Received:
    3,259
    Trophy Points:
    113
    Location:
    Bird-watcher's paradise
    I spent 4 years listening to the og, then bought the J2. Sounds pretty smooth to me.
     
    • Respectively Disagree Respectively Disagree x 2
    • Like / Agree Like / Agree x 1
    • List
  18. spoony

    spoony Spooky

    Friend
    Joined:
    Sep 28, 2015
    Likes Received:
    502
    Trophy Points:
    93
    Well, good for you. However, a bunch of people shared this impression during the loaner tour.
     
  19. aamefford

    aamefford Nothing like chamberpot coffee

    Friend
    Joined:
    Nov 26, 2015
    Likes Received:
    1,047
    Trophy Points:
    93
    Location:
    Between SF and Sac
    My “tricks” in general order of application. Typically for late night listening:
    1) switch to SE input. I run both RCA and XLR cables from BF2 to Jot2
    2) Low gain
    3) plug into SE (1/4”) HP jack.

    no actual technical hacks.
     
    • Like / Agree Like / Agree x 1
    • Epic Epic x 1
    • List
    Last edited: Jul 31, 2021
  20. Biodegraded

    Biodegraded Friend

    Friend
    Joined:
    May 28, 2017
    Likes Received:
    5,803
    Trophy Points:
    113
    Location:
    Vancouver BC
    Modi Multibit:
    To confirm - that's at U17 & U18, right? Underside of the board:

    [​IMG]

    What's U20? TL082C is the legend.
     
    • Like / Agree Like / Agree x 1
    • List

Share This Page