NwAvGuy's Dream Cums Tru: Schiit Magni 3 Heresy

Discussion in 'Headphone Amplifier Measurements' started by purr1n, Nov 22, 2019.

  1. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    52,314
    Dislikes Received:
    89
    Trophy Points:
    113
    Location:
    Antarctica
    https://www.schiit.com/products/magni-1

    Many eons ago, there was this guy named Doug Self who loved distortion measurements. He was the original measurebator and objectivist. Fast forward to around 2010, and this dude NwAvGuy happens to unearth's Doug Self's tomb. Inspired by Self's writings and religious philosophy, NwAvGuy goes off on a crazy insane rant on the dangers of high-end audio in an Internet blog, complete with clip art pictures of wine, sports cars, the Edison phonograph, for use in lame analogies. Douglas Self's objectivism was now back in vogue, at least for a few years; but then everybody seemed to suddenly forget about it and returned to their old ways of buying amps with 0.01% distortion instead of 0.0003% distortion. The horror, the horror! Then about last year, or two years ago, ASR came online, and the cycle repeated - the third coming. I think by now toward the end of 2019, heading into 2020, the neo-objectivist trend is waning again, like all other times before it since the 70s.

    The view that almost perfectly (or as close to perfect with today's technology) measuring amps equates to full transparency and highest performance is an anathema to SBAF. I don't think that most SBAF members think 0.0003% amps suck, but it would seem that most experienced members gravitate toward amps that run more in open loop rather than closed loop. Heck, even audio God Nelson Pass believes such nonsense! There's a school of thought that open loop sounds better, more immediate, more real, despite not having distortion measurements so good that only TOTL modern instrumentation can detect. SBAF members here have performed various sighted, blind, double blind tests to determine for themselves what they like. Ultimately, it's our ears, not AP-555s that decide.

    What I liked about NwAvGuy is that he was for stuff that not only measured great, but that could be done with a bargain barrel approach. This was the best aspect of NwAvGuy - he wasn't all evil. Note that in 2019, there's plenty of great stuff out there that measures extremely well (e.g., Drop THX AAA 789, Schiit Jotenheim, and lately Neurochrome HP-2). But all these lack NwAvGuy's refreshing approach in that it can and should be on the super cheap.

    Regardless of how I feel about the limitations (from a subjective point of view) of these kinds of amps, one particular MoT did approach me on reviewing this "objectivist" amp. It's technically still under press embargo until next week, so I cannot reveal much about it. However, the MoT did give me permission to share any measurement results.
     
    Last edited: Dec 3, 2019
  2. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    52,314
    Dislikes Received:
    89
    Trophy Points:
    113
    Location:
    Antarctica
    The following measurements were taken with the pot maxxed out. The load is 300-ohms.

    1kHz -0.76dbU, which is about 0.71Vrms
    The 1kHz is normalized to 0db on the y-axis
    0dbU.png
    The second harmonic is about -117db. That's 0.0001413% distortion. There's a nice pattern where each successive harmonic is lower than the one before one. We don't see suppression of even harmonics relative to odd harmonics.

    1kHz 3.24dbU, which is about 1.25Vrms, a little bit higher level than the one before
    The 1kHz is normalized to 0db on the y-axis
    4dbU.png
    We turned up the volume a bit here. Second harmonic is -115db. That's 0.0001778% distortion.
     
    Last edited: Nov 23, 2019
    atomicbob, Sqveak, Clemmaster and 4 others like this.
  3. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    52,314
    Dislikes Received:
    89
    Trophy Points:
    113
    Location:
    Antarctica
    Here's a voltage level that would be more reasonable with normal use on 300-ohm headphones such as HD650s.

    1kHz -4.76dbU, which is about 0.45Vrms. 300-ohm load.
    The 1kHz is normalized to 0db on the y-axis
    -4dbU.png
    Second order is down to -120db or 0.0001%. Third order is a bit higher though.
     
    Last edited: Nov 23, 2019
    atomicbob, ultrabike and pure5152 like this.
  4. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    52,314
    Dislikes Received:
    89
    Trophy Points:
    113
    Location:
    Antarctica
    Now this measurement is with the amp almost maxxed out near 19.24dBu, which is 7.1Vrms. I'll play with the amp a bit more to see where it clips exactly. I'm pretty sure that in another volt or two, the amp will start to clip. 7.1Vrms is guaranteed to explode your Sennheisers, if not immediately, soon thereafter.

    1kHz 19.24dbU (probably just a bit before clipping) 300-ohm load.
    The 1kHz is normalized to 0db on the y-axis
    20dbU.png
    Second order is -110db or 0.0003162% distortion. Oh, this is so horrible.
     
    Last edited: Nov 23, 2019
    atomicbob, pure5152 and ultrabike like this.
  5. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    52,314
    Dislikes Received:
    89
    Trophy Points:
    113
    Location:
    Antarctica
    Again, I didn't push the amp to clipping (AverLAB has limited output to 20dbU).

    THD+N vs Power in mW. Analyzer set to 22kHz bandwidth (per custom)
    upload_2019-11-22_19-14-24.png

    Overall distortion is low, but if one cared about ultimate numbers, the sweet spot of this amp is wide (for say HD650): anything past 0.5mW to exploding the headphones. Mind that these are THD+N figures, which will be higher than THD.
     
    Last edited: Nov 23, 2019
  6. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    52,314
    Dislikes Received:
    89
    Trophy Points:
    113
    Location:
    Antarctica
    I'll try to gather some apples-apples measurements and compare with THX AAA 789 and Jotenheim. The ones I'm thinking would be the 2Vrms into 330-ohms.

    MEASUREMENTS ARE HARD!

    In the meantime, here is some fun stuff. Have you every wondered why @atomicbob, who is our resident professional measurer (I'm just a hack) takes photos of this gear and lists accessories down to the T? It's because when it comes to 0.00000 type measurements, all the little things matter.

    Take this for example. This is the exact same measurement, but using a variety of cables. Some high quality. Some garbage from Chinese vendors selling on Amazon. Some longer. Some shorter. Some thicker. Some with BNC to RCA adapters. The fact is, things like cables and connectors matter. Connectors seem to matter a ton, and the best performing stuff is soldered with as few connectors as possible.

    variance.png

    The moral of the story is when we are in 0.00 territory, it's very easy to make a mistake, be lazy, be inconsistent. Heck, there's even no guarantee that I can replicate the results exactly the next day. So beware. It's much easier to bork measurements than objectivists with expensive toys will have you believe. This is the kind of stuff that seems so rarely talked about, but is absolutely crucial, especially if you are the kind of guy that likes to say that one car beat another in the 1/4 quarter-mile by 2/10s of a second, or that one DAC beat another by 2db. I'd argue that those with bias against a brand are more apt to "unintentionally" bork measurements of products from that brand via lack of attentiveness.

    Note that as the power goes up, the effects of the cables and connectors get less.

    Anyway, the way I look at measurements is like this: If two cars are within a few tenths of a second in the quarter-mile, we say it's a driver's race. Too close to call. Too many other factors.
     
    Last edited: Nov 23, 2019
  7. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    52,314
    Dislikes Received:
    89
    Trophy Points:
    113
    Location:
    Antarctica
    MEASUREMENTS ARE HARD! PART 2.

    Here is another example of how measurements can easily be borked. This set is one of my favorites because I've noticed ASR getting different results from others in terms of AC mains noise on the same gear. Again, if we are in rush at a friend's house, are not attentive, use the wrong drivers, or the wrong USB power plug, or don't care much because we dislike a certain brand, we are more apt to miss things like this.

    Examples below of exact same amp, same measurement. One measurement has "horrible" AC mains noise. Well, actually, it's not horrible. I'll gave $100 to the first person who can hear tell the difference 8 of of 10 times this AC mains noise using an HD650 listening to music in blind test.

    image1.png image2.png

    You guys wanna know what I did different to get the "bad" result? (And really, it's not that bad to be of any concern - last time I checked, I couldn't hear -110db down or 0.0003% distortion). I used plugged the AC wall wart into a power strip that was closer to the AverLAB and kept the power cord tied up with a wire-tie. Note that it would be easy for me to intentionally bork the measurement of a vendor that I disliked.
     
    Last edited: Dec 3, 2019
  8. Clemmaster

    Clemmaster Friend

    Friend
    Joined:
    Sep 28, 2015
    Likes Received:
    1,196
    Dislikes Received:
    1
    Trophy Points:
    93
    Nice DAC!
     
    Lyander, Josh83, elmoe and 2 others like this.
  9. wadec22

    wadec22 Acquaintance

    Joined:
    Sep 28, 2018
    Likes Received:
    48
    Dislikes Received:
    0
    Trophy Points:
    13
    Location:
    USA
    one of the reasons I enjoy your posts is that you give good points of reference and context for why particular measurements may not be as meaningful as some believe they are. I like reading ASR reviews, as they are one of the few places to get points of reference between two pieces of equipment. I think most of us want that comparison when spending more than $100 or so.

    I think the "third coming" going too far is probably due to the fact that many folks on head-fi give such emotionally charged responses that have nothing backing them up but monetized justification/confirmation bias.

    This site has a lot of members who give great context that is founded on experience, I enjoy and appreciate that very much. I also like to see it carry over into posts about real world scenarios like your recent post about career advice (a lot of it hit home for me).

    I imagine being around a long time and seeing different things pop up like annual weeds is exhausting, but helping educate others is priceless. I know I very much appreciate it and I do not think I am alone. There are just so many options in the audio world and many with imposing price tags. People want help protecting their hard earned jack.
     
  10. Hands

    Hands Overzealous Auto Flusher - Measurbator

    Staff Member Friend MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    9,624
    Dislikes Received:
    32
    Trophy Points:
    113
    Location:
    Colorado
    Home Page:
    OK, @purr1n, spill the beans. Which cables do I need to buy now?

    Though, how do I know you didn't mess up those tests?

    Also, multitone is the current "Aha!" measurement that most closely correlates to what we hear with music. I doubt it is ever to fall out of favor and am certain we have finally found something to squash all subjective concerns. Given that, can you please provide multitone measurements? I must see how much grass I must consume.
     
  11. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    52,314
    Dislikes Received:
    89
    Trophy Points:
    113
    Location:
    Antarctica
    You don't. There are ways to obtain some assurance however:
    1. Overall experience and anal retentiveness of the person conducting various measurements of any kind. (I still have nightmares of my freshman year in engineering class where we had to take repeated measurements, trying to ensure that they were consistent. Some teams tended to suck at it. Haha.) "I went to a buddy's house, but didn't have much time to set up..." that kind of information should be disclosed at the start and may very likely affect the quality of the measurements.
    2. Earnestness - willing to admit biases, fallibility, potential for error. Lots of buttons, switches, parameters, etc. to set up. Guys who were around during the Changstar days would question the headphone measurements, pointing out areas of possible human error. It's easy to make a mistake. I would go back and correct things that were done wrong. I still make mistakes.
    3. Peer review - especially for the 0.0000 -115db kind of tests where little things like cables and connectors matter. One cannot rest of their laurels of say, having overseen HDCD, HDDVD, Zune, Head-Fi, Cloud security architectures, having expensive instrumentation, etc. as sufficient credentials for infallibility. Taking measurements is a skillset in itself and requires practice. Any good scientist, will welcome peer review.
    4. The tester being willing to repeat tests if others point out areas where they may be suspect.
    Touche! Multitone Tests are here.

    Unfortunately, my AverLAB with its current software can't pull off FFTs with 16 million taps. I've only got 32768 to work with. Therefore fine resolution in the low frequencies is obscured and the FFT floor is higher. (On FFT "noise floor" it's important to understand FFT process gain: https://www.superbestaudiofriends.org/index.php?threads/judegate-3-shut-the-f-up-jude.5674/) I'm also using the Modi 3 as a signal generator, since the AverLAB's software can't do multitone tests yet.

    upload_2019-11-23_11-20-11.png
     
    Last edited: Nov 23, 2019
    Joshvar, Lyander and LetMeBeFrank like this.
  12. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    52,314
    Dislikes Received:
    89
    Trophy Points:
    113
    Location:
    Antarctica
    Here's another one, trying to get more data.
    upload_2019-11-23_11-48-11.png
     
  13. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    52,314
    Dislikes Received:
    89
    Trophy Points:
    113
    Location:
    Antarctica
    Last multitone
    upload_2019-11-23_11-50-24.png
     
  14. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    52,314
    Dislikes Received:
    89
    Trophy Points:
    113
    Location:
    Antarctica
    LOL.

    An important thing to point out is that we are at the limits of the signal generator for the AverLAB.

    The distortion in measurements presented could be just as likely to be from the AverLAB as from this "NwAvGuy / SBAF-heretical" amp.

    From general observation and use, the AverLAB's analyzer is slightly better than the AP. The AP has a better generator, and the AP software is miles ahead.
     
    Last edited: Nov 23, 2019
    Psalmanazar and Lyander like this.
  15. zerodeefex

    zerodeefex Grumpiest admin

    Staff Member Friend BWC
    Joined:
    Sep 25, 2015
    Likes Received:
    10,339
    Dislikes Received:
    43
    Trophy Points:
    113
    [​IMG]

    Graph for the Neurochrome HP-2 for comparison. If it really is a fraction of the price, this is a pretty awesome feat.
     
  16. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    52,314
    Dislikes Received:
    89
    Trophy Points:
    113
    Location:
    Antarctica
    Apples to Apples comparison
    to THX AAA and Jotunheim
    8.23dbU into 330-ohms (330 here because that's what I used as the load for the above amps)

    "neo-NwAvGuy" msytery amp
    1kHz 8.23dbU into 330-ohms
    upload_2019-11-23_12-40-50.png

    Drop THX AAA
    1kHz 8.23dbU into 330-ohms
    [​IMG]

    Schiit Jotunheim
    1kHz 8.23dbU into 330-ohms
    [​IMG]
     
    Lyander, pure5152 and Baten like this.
  17. Roget

    Roget Rando

    Joined:
    Nov 17, 2018
    Likes Received:
    15
    Dislikes Received:
    24
    Trophy Points:
    3
    [​IMG]
    [​IMG]

    Neurochome HP-2 looks better THD at lower power. Def interested in this amp if you say its what it is.
     
  18. zerodeefex

    zerodeefex Grumpiest admin

    Staff Member Friend BWC
    Joined:
    Sep 25, 2015
    Likes Received:
    10,339
    Dislikes Received:
    43
    Trophy Points:
    113
    No it doesn’t. The graph for the neurochrome starts at 1mW while mystery amp starts at .01mW.
     
    Xecuter and Baten like this.
  19. IUONA

    IUONA Friend

    Friend
    Joined:
    Jan 8, 2016
    Likes Received:
    10
    Dislikes Received:
    0
    Trophy Points:
    3
    That is exactly why I read his blog (well, the entries that didn’t seem stupid.) I’ve occupied the strange space of being (ehhh, audiophile sounds too Mr. Monopoly for me, so I’ll go with...) in a hobby where, unless one has a fair amount of money to burn, or lives in a wealthy city and/or happens to be lucky enough to live near a solid hi-fi store that demos a lot of headphone gear, one is limited to cheap gear, DIY’ing, or traveling long distances. So NwAvGuy’s PCB & plans gave me my first decent amp. (My tin can CMOY didn’t cut it.) His attitude & monomaniacal objective hard-on was off-putting, but the amp itself (I’ve made a couple modifications to it) was the best I’d heard at the time, and it only cost me like $100 and some time in the garage while I was in high school. I still use it, sometimes.

    I’m still on board with the idea that good sounding gear doesn’t need to cost a lot of money. Schiit, Drop collaborations & a few others have done that, but I’d really like to see more companies offering legitimately good amps or DIY plans & DACs (that, at least, don’t measure like complete shit, and sound really good) under the $200 mark.

    Whatever this amp is, I’m intrigued...
     
    wadec22 likes this.
  20. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    52,314
    Dislikes Received:
    89
    Trophy Points:
    113
    Location:
    Antarctica
    NwAvGuy wasn't that bad. The thing that undermined him was his attitude, you know, "the how to make friends and influence people" aspect. He got a in tussle with AMB for reasons I never understood, but NwAvGuy let his emotions take over, and as a result inappropriately used the AMB mini^3 for comparison. Chalk it up to youth.

    The problem? The AMB mini^3 is actually portable (pocketable), and because it was so, had to rely on a smaller lower voltage battery. Power supply is almost everything when it comes to amps. So the comparison wasn't quite fair or apples-apples.

    Also NwAvGuy dismissed valid engineering criticisms regarding putting the pot after the gain stage. Engineering is about trade-offs and idiot-proofing, so the few db SNR gained wasn't going to be worth the users (who did not read the "manual") who would end up clipping their amps. I felt like I was part-time technical support for NwAvGuy for two months when I was active on HF, helping HF'ers resolve clipping issues.

    Other than those things, a horrible PCB layout where a 'lytic cap was placed right next to a hot regulator, and the super lame pictorial analogies, NwAvGuy was actually OK. The Neo-objectivists today are off their rocker - read their criticisms of Schiit's PCB coaster on ASR. Seriously, these gets are getting upset at a fricking gimmicky coaster (that can happened to be turned in to Vali 1), that is being given away for free at the Schiitr.
     
    Last edited: Dec 3, 2019

Share This Page