Oh shit! SINAD can suck my ****s

Discussion in 'Measurement Techniques Discussion' started by Marvey, Jul 2, 2021.

  1. Baten

    Baten Friend

    Pyrate
    Joined:
    Mar 18, 2018
    Likes Received:
    1,130
    Trophy Points:
    93
    Location:
    EU
    Might be a dumb question, but considering it is mostly low-end distortion breaking up, AFAIK the HD6-- series haven't got the best performance in that range anyway. Right? So maybe not even an audible issue, unless you are boosting the bass way up? From my own listening, using amps like heresy/IEMagni or this L30 I don't personally hear the troublesome areas exposed here. But I don't listen extremely loud, and I don't EQ to harman dum-dum levels.

    But all of this is extremely interesting none the less, especially so with Jotunheim not exhibiting any of these problems.
     
  2. Ox Cart

    Ox Cart Facebook Friend

    Joined:
    Mar 13, 2018
    Likes Received:
    202
    Trophy Points:
    33
    Location:
    Austin, TX
    Good question. The levels that @Marvey tested would be slightly over 100dB for an HD650. That may seem loud, but as far as I understand it, bass levels (especially transients) can exceed averages by quite a bit. Also, the typical standard for loudness (A-weighted SPL) is heavily weighted towards 1 KHz +.

    Based on the below spectrum of 30 seconds from a live recording of Phronesis' "Eight Hours", listening at what I'd consider a comfortable 70 dB SPL(A), bass levels will actually be closer to 100 db SPL(A). For critical listening some might turn up the volume higher than this.

    Also, I don't know how transients play into this (i.e. the music might have transients in the bass that are even louder).

    @Marvey would you mind getting a measurement of the HD650 at a few higher volumes? I'm curious if the distortion gets exponentially worse as it does on the transducer itself.

    [​IMG]
     
  3. Michael Kelly

    Michael Kelly MOT: Pi 2 Design

    Pyrate
    Joined:
    Apr 22, 2016
    Likes Received:
    4,112
    Trophy Points:
    113
    Location:
    Rhode Island, USA
    Home Page:
    Perhaps I am being overdramatic, but I believe that this thread has exposed the weakness in pretty much any headphone review that has ever been done! Without knowing the interaction between the headphone and the chosen amplifier we cannot tell if the results are at all translatable to other amplifiers.

    There has clearly been a large body of anecdotal evidence of some headphones being well received across multiple reviewers, with others being more consistently poorly received. But it is a rare headphone reviewer who will, or even can, test the headphones with multiple amplifiers in order to see how well it does.

    This thread and @Marvey in particular is doing an incredible service to the Headphone listening audience.

    as a manufacturer who is preparing to release some analog oriented products including those with headphone amplifiers, I will now have to adjust my testing procedures to take into account these possible variances. What would be very useful to me, quite selfishly, is a list of headphones that seem to be either well-behaved across multiple reviews or poorly behaved yet somehow still popular!

    P. S. It is now more incumbent upon us as the consumers of various headphone reviews to pay much more attention to the signal chain especially the amplifier being used.
     
  4. Lyander

    Lyander Official SBAF Equitable Empathizer

    Pyrate Contributor
    Joined:
    Feb 25, 2017
    Likes Received:
    10,961
    Trophy Points:
    113
    Location:
    Philippines, The
    Would rather have informed conjecture from a genuine geek than "objective truths" from a self-serving, self-proclaimed messiah figure. A lot of this conversation regarding back-EMF is going over my head but that's at least in part due to my lacking mental bandwidth for very much beyond eat, sleep, work haha. This visualisation helps though, and evidences the whole "if you can't explain something simply you don't actually understand it" adage.

    This is why reviews that don't at the very least have upstream components disclosed are much less credible IMO, it's long been consensus here I think that it's not so much headphones and amps that get recommendations but headphone and amp pairings. The virtues of system synergy and relative gear "scalability" are behind a lot of discussions here it seems like, and that isn't something I've seen much evidence of in many other communities.
     
  5. Marvey

    Marvey Super Friend

    Staff Member Pyrate BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,712
    Trophy Points:
    93
    Theoretically, the lowered THD performance (electrical measurement) is not an issue with the HD650. We're still talking -80db or -90db THD which is really good by normal standards. The problem is hacks like Amir who have misappropriated SINAD and used them for FPS like benchmarks.

    Let's say the HD650 driver has 3% THD for a 37Hz tone. This is translates into -30db. If the amp is maintaining -80db (bad by ASR standards, "good enough" by the standards here), we are talking about a voltage ratio of 0.0001. Yes, that's 1/10000!

    That HD650 with the -30db THD at 37Hz is still going to have -30db THD (actually -29.97db) with the -80db THD amp compared to a perfect zero distortion amp!

    This is why chasing those SINAD was nonsense in the first place! Unless those THD+N numbers start to get really bad, that is close to that of transducers.

    Now the caveat is that there could be other factors. Could these "discoveries" be indicators of other things (yet to be measured or not yet measurable) that we hear subjectively. It's possible. There's just so much that we do not know and I will be the first to admit that. Saying that you worked as a Microsoft VP with a direct line to Bill Gates email (probably only for Gates to yell at you or tell you that you are dumb) doesn't make you know any more.
     
    • Like Like x 12
    • Epic Epic x 2
    • List
    Last edited: Jul 5, 2021
  6. Marvey

    Marvey Super Friend

    Staff Member Pyrate BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,712
    Trophy Points:
    93
    Distortion always gets worse at higher SPL. The questions are how much worse, how much quicker, and when does it start.

    HD650
    upload_2021-7-5_9-30-41.png
     
    • Like Like x 3
    • Epic Epic x 1
    • List
  7. jnak00

    jnak00 Friend

    Pyrate
    Joined:
    Feb 9, 2017
    Likes Received:
    608
    Trophy Points:
    93
    Location:
    Canada
    Many of the reviews and commentary on this site have pointed to this issue, even if indirectly. System synergy, often scoffed by the likes of ASR, has always been a point of discussion here. Sometimes it's more of a "don't pair warm amp with warm headphones" type of comment, but it's clear there is recognition of interaction of the various components.
     
  8. Marvey

    Marvey Super Friend

    Staff Member Pyrate BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,712
    Trophy Points:
    93
    This is why we challenge: "What's your (component) chain?". I'm a bit more loath to do this these days because such questions can seem off-putting to newbies. Makes it seem like we are elitist snobs when what we are trying to do is gather information. Component chains make a difference. Component synergies from a electrical / measurement point of view (impedance, load, gain, distortion pattern, etc.) and also from a subjective angle.

    All of this is information. The subjective angle is spotty, and I agree with that. But if 24 out of 25 people say the Asgard 3 is a warm sounding amp even though the THD at 150 mVrms is -120db, then it probably is. The distortion pattern also happens to have a small H2 spike at -120db - nothing else*. Both these pieces of information will tell me a lot about the tonality of the amp. I'll take a double blind test to prove I can hear this too.

    * And yes, a H2 by itself without other harmonics will sound different from descending harmonics. My observations thus far are that distortion pattern seems to matter more than THD, unless THD gets really bad, like -55db. THD+N (SINAD) is also both good and bad. If the N component is low, it's really not in play. Thus the N pollutes the result. THD by itself can be better.
     
    Last edited: Jul 5, 2021
  9. Beefy

    Beefy Friend

    Pyrate
    Joined:
    Apr 10, 2021
    Likes Received:
    1,738
    Trophy Points:
    93
    Location:
    Canada
    I don't think this changes anything for you.

    You can continue to do what good designers have always done - employ solid design principles, good engineering, make sure it measures well enough not to openly oscillate, and listen to the end result to check that it still sounds pleasing to your ears.
     
  10. Marvey

    Marvey Super Friend

    Staff Member Pyrate BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,712
    Trophy Points:
    93
    Again, same deal. 2Vrms from output. Adjusted via volume knob to get 0.777Vrms at the output. HD650 and Elex as the load. Unity gain. 90kHz bandwidth.

    Geshelli Erish balanced in/out only amp
    BLU = HD650 as the load
    MAG = Elex
    PUR = 300-ohm resistor load
    upload_2021-7-5_10-16-12.png

    Much better performance than L30 here:
    [​IMG]

    Do balanced topologies have an advantage? Note SWT51 @atomicbob tested is inherently balanced because of the output transformers (unknown if secondaries are grounded or if that makes a difference).
     
    Last edited: Jul 5, 2021
  11. Marvey

    Marvey Super Friend

    Staff Member Pyrate BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,712
    Trophy Points:
    93
    Also, 85db is still very good. THD or THD+N isn't a rat-race where those in the 3rd or 4th quartile suck. Don't be controlled by Sauron's Rings. Just as we shouldn't be controlled by MQA.
     
  12. Michael Kelly

    Michael Kelly MOT: Pi 2 Design

    Pyrate
    Joined:
    Apr 22, 2016
    Likes Received:
    4,112
    Trophy Points:
    113
    Location:
    Rhode Island, USA
    Home Page:
    Unfortunately, as a manufacture I cannot rely on my ears, or frankly anyone’s ears for that matter. But I am starting to get a glimmer of a test plan that includes gathering a handful of headphones and measuring the THD as @Marvey has done and see that the amplifier I am creating is as resistant as possible to the headphone load.

    More work, but rewarding!
     
  13. Beefy

    Beefy Friend

    Pyrate
    Joined:
    Apr 10, 2021
    Likes Received:
    1,738
    Trophy Points:
    93
    Location:
    Canada
    At the very least, balanced is dumping back EMF from the driver into an active load on both wires. Single ended, back EMF is also being dumped into signal ground.

    Seeing the Grado balanced versus single ended would be really neat. I'd also be interested to see what happens into an active vs passive ground in a B22. I've always thought that if there is any genuine benefit to an active ground, then preventing ground pollution from back EMF is indeed the ticket.
     
  14. Marvey

    Marvey Super Friend

    Staff Member Pyrate BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,712
    Trophy Points:
    93
    Anyway, I'm going to call it off for today and go back to product review. However some good ideas have been formulated for testing of amps - the aim is to be as real as possible.
    1. Instead of dummy loads, use a good set of real headphone as loads: HD650, Grado SR325X, Elex, HE400SE, LCD2 (I need to get an Audeze)
    2. Use an input voltage of 2Vrms into the amp from the analyzer, but utilize the volume pot to arrive at the desired output.
    3. The desired output to arrive at THD should not be based on voltage, but for SPL (100db at 1kHz) for each of the headphone loads. This volume knob setting will be especially loud for pop music, but actually works well for high dynamic range classical. I suspect the THD differential will be less extreme at lower levels.
    4. I need a fancy name for this method / visualization, e.g. "RealTHD", etc.
    Good point on the ground. I need to buy some 4-pin XLRs to re-terminate the Grados. I can't find my stash. It's in a box somewhere.
     
    Last edited: Jul 5, 2021
  15. Biodegraded

    Biodegraded Friend

    Pyrate Contributor
    Joined:
    May 28, 2017
    Likes Received:
    7,985
    Trophy Points:
    113
    Location:
    Vancouver BC
    DRL - Distortion with Real Load. Not to be confused with Darth Real Life.
     
  16. atomicbob

    atomicbob dScope Yoda

    Pyrate BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    18,653
    Trophy Points:
    113
    Location:
    On planet
    The SW51+ headphone output is a TRS jack. One side of each output transformer is connected to the shield of the TRS jack. The shield is connected to amplifier case and also the ground of IEC 320 C14 power inlet.
     
  17. Brad358

    Brad358 New

    Joined:
    Jan 14, 2020
    Likes Received:
    74
    Trophy Points:
    13
    Location:
    England
    I'm not an audio nor an electronics engineer, but in my own field of engineering it is well understood that measuring performance is about a lot more than buying a fancy tool and looking for a simple more/less = better metric (although, unfortunately even in my field marketing types are always trying to reduce it to just such a magic numbers game). And the dangers of over optimizing around such simplistic metrics are well understood. For measurement to be useful you really have to understand what you are measuring, how the measured parameters relate to performance and how the measurements are taken and under what conditions. Unfortunately in audio a certain website has done a splendid job convincing an audience that SINAD is the be all and end all of audio evaluation. Thanks for this thread, it's a thread that anyone who has an interest in audio measurement should read.
     
  18. JohnM

    JohnM Author of REW - Rando

    Joined:
    Dec 13, 2015
    Likes Received:
    134
    Trophy Points:
    33
  19. Elnrik

    Elnrik Super Friendly

    Pyrate
    Joined:
    Jan 10, 2017
    Likes Received:
    8,973
    Trophy Points:
    113
    Location:
    Denver CO
    Home Page:
    Yeah you know, that string of Christmas tree lights wired in parallel continue to work on a bulb burns out. The one wired in series though...
     
  20. Marvey

    Marvey Super Friend

    Staff Member Pyrate BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,712
    Trophy Points:
    93
    Just be to clear, I wasn't picking on Topping. This behavior occurs with Schiit Magni 3+ too. 0.777Vrms here.
    upload_2021-7-5_18-38-48.png
     
    Last edited: Jul 5, 2021

Share This Page