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

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

  1. roshambo123

    roshambo123 Friend

    Friend
    Joined:
    May 26, 2018
    Likes Received:
    639
    Trophy Points:
    93
    Location:
    Los Angeles
    Slap me if I'm talking rot, I am not an EE, but is the 37hz SINAD test using a 300 ohm resistor vs. HD650-as-load an apples-to-apples comparison given the variable impedance of a dynamic headphone at different frequencies? Like an HD650 at 37 hz is closer to 430 ohms, so therefore the fixed resistor measurement should in my mind also be at 430 ohms. Of course this speaks to the narrowness of the ASR approach. If a SINAD measurement is taken at a single impedance in my mind that measurement is only valid for something like a planar headphone with a flat impedance curve assuming both the fixed resistor and the planar headphone had the same impedance. If you're applying that measurement to the HD650 you'd only have select bands where things were close to to 300 ohms, with the peak at 85 hz being over 500 ohms.

    HD650. Source: Innerfidelity
    upload_2021-7-5_16-43-14.png
     
    • Like / Agree Like / Agree x 1
    • List
  2. Marvey

    Marvey Super Friend

    Staff Member Friend BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,664
    Trophy Points:
    93
    I could not quite replicate the results. For 1kHz yes, it's close enough. For the others, not so close. Perhaps I have a defective unit where it's not just one channel slightly worse than the other, but both channels are out of spec? I'd be happy to accept a cherry picked unit from John Yang at Topping to reconfirm.
    upload_2021-7-5_19-26-19.png

    Here I've left the switches for 30 and 300 ohms flipped up so they are not in the circuit. One phono jack on the left goes back the AP. The other next to it is meant for an external load which is the HD650 here. The HD650 is sealed on the miniDSP EARs.
    PXL_20210705_235614687.jpg
     
    • Like / Agree Like / Agree x 2
    • List
    Last edited: Jul 5, 2021
  3. Marvey

    Marvey Super Friend

    Staff Member Friend BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,664
    Trophy Points:
    93
    To answer your question, it's not the same - that impedance at 37Hz that is - not 300 ohms.

    However what we are seeing here is a different phenomena. So far balanced or transformer coupled amps (per @atomicbob) don't exhibit this "back-EMF" behavior as much where the distortion of the headphone ends up reflected in the measurements. And again, this only shows up with dynamic headphones.

    With orthos it doesn't. The could be that orthos are almost purely resistive loads or because they have minimal excursion therefore less back-EMF. This is all conjecture at this point.

    The debate is whether these measurements are artifacts or not. Whether they reflect the performance of the amp, the headphone, or maybe another possibility: both. I think it's too easy to dismiss this could also be the amp. We are measuring the system, the amp and the headphone after all, and shouldn't this count? The problem is that we really can't settle this by measuring amps which exhibit this phenomena at the headphone because the distortion of all these amps are still too low!

    And really, in the end at any reasonable SPL, the distortion we are seeing here isn't enough to make a dent in headphone measurements: https://www.superbestaudiofriends.o...-sinad-can-suck-my-s.11175/page-4#post-352748 An -80db THD amp into a headphone with -30db THD at 37Hz is still a headphone with -30db THD at 37Hz.
     
    • Like / Agree Like / Agree x 5
    • List
    Last edited: Jul 5, 2021
  4. Elnrik

    Elnrik Super Friendly

    Friend
    Joined:
    Jan 10, 2017
    Likes Received:
    8,706
    Trophy Points:
    113
    Location:
    Denver CO
    Home Page:
    So, John's little rebuttal post has the headphone and what I can only assume is the measurement device wired in parallel. Resistance in parallel is Rt = R¹+R²+R³... Yeah? Isn't he measuring a device that now has the resistance of the headphones + measurement device? All he is using is a Y splitter. Does it matter?

    Also, Marv... I hate to seem stupid, but I'm going to ask anyway. Your measurements are taken from the headphone via measurement dummy head + mic? I think so, but I'm just not sure I kept up on the process you've got going on there, or how your little switch is wired up.

    Sorry in advance if this is a pants-on-head stupid question.
     
    • Like / Agree Like / Agree x 4
    • List
  5. Roget

    Roget Acquaintance

    Joined:
    Nov 17, 2018
    Likes Received:
    39
    Trophy Points:
    18
    • Like / Agree Like / Agree x 2
    • List
  6. Marvey

    Marvey Super Friend

    Staff Member Friend BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,664
    Trophy Points:
    93
    The dummy head is only for ensuring that there is a seal on the headphones. The test point is here. With the resistor switches up, the board acts as a Y splitter. The measurement device, the APx555, has selections for load. I used 100,000 kohms.
    img2.jpg

    Now what may be interesting for me to investigate is using different cables. I do have an extra breakout cable from the DUT (amp) to the board.
     
    • Like / Agree Like / Agree x 3
    • List
  7. Marvey

    Marvey Super Friend

    Staff Member Friend BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,664
    Trophy Points:
    93
    That's an interesting article.

    If Benchmark's amps, including those built-in on his DAC do not exhibit this performance, then I don't see how this phenomena is purely reflective of the headphone. I don't see how it's not reflective of an amp's performance when different amps are behaving differently. That's like tossing information aside because it doesn't fit with what you believe. It's doesn't make sense to isolate systems which are meant to interact.

    As I mentioned earlier, I wish we would put this to rest by measuring at the headphone, but either headphone distortion tends to be too high or the phenomena doesn't doesn't occur at lower voltage levels where the headphone distortion is low enough.
     
    • Like / Agree Like / Agree x 4
    • List
  8. Roget

    Roget Acquaintance

    Joined:
    Nov 17, 2018
    Likes Received:
    39
    Trophy Points:
    18
  9. Marvey

    Marvey Super Friend

    Staff Member Friend BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,664
    Trophy Points:
    93
    Well as I mentioned, it's up for debate. Everyone is going to have an opinion. Obviously John Siau's opinion is different from Amir's and Solderdude's. I'm strongly tending toward Mr. Siau's side, but I'm on the fence. I would like more data, more information.

    I 100% agree with Amir that one doesn't measure at the input of a headphone. That one should measure at the output of a headphone.

    However, this is mere double-speak nonsense from Amir again because if one measured at the headphone output, any amp with SINAD > 60db would be sufficient given normal listening volumes and ambient SPL levels.

    Which goes back to the topic of this thread: SINAD can suck it.
     
    • Like / Agree Like / Agree x 12
    • Epic Epic x 1
    • List
    Last edited: Jul 5, 2021
  10. Roget

    Roget Acquaintance

    Joined:
    Nov 17, 2018
    Likes Received:
    39
    Trophy Points:
    18
    You are both right. Amir wants dummy load to isolate pure SINAD of amp. He is right. You want real load to get THD of amp with headphones. You are right. Different ways.
     
    • Like / Agree Like / Agree x 2
    • List
  11. Gazny

    Gazny Friend

    Friend
    Joined:
    May 11, 2020
    Likes Received:
    731
    Trophy Points:
    93
    Location:
    SoCal
    Great work marv, I have been thinking about this for quite a while since your first post.
    while many will disagree saying a headphone is not a real load for a headphone amplifier, I disagree.

    what got me wondering about the point brought up previously, what is this effect and is there a way we can hear the difference?

    My question is, while some single ended amps exabit this behavior, do they all?
    Akin to the removal of the veil? We see the Jot1 does this, but my question is can a single ended amplifier do it as well?

    Terms I look into, EMI/EMC SUPPRESSION, Current Return Path design, and return path impedance. With that, does a cable make a difference? I know some people have braided cables, and their formations may reduce the effect. would be funny if cables made a difference, wouldn't it?
     
    • Like / Agree Like / Agree x 1
    • List
  12. Marvey

    Marvey Super Friend

    Staff Member Friend BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,664
    Trophy Points:
    93
    A lot more questions than answers. We need data, more measurements from different amps. Does this always happen on SE amps? Are all balanced amps immune (less affected)? And when it happens, do different amps exhibit different behaviors? Do all amps start to fall apart and perform similarly after reaching a certain output because the back-EMF is so bad? Or is it variable?

    Even then, I don't think it invalidates the results when we take the closed system (amp and headphone) into effect. Maybe past a certain point, amps no matter how good their THD or SINAD, can't control a what a headphone wants to do. I'm in agreement with @Roget. Different methods here accomplishing different goals.
     
    • Like / Agree Like / Agree x 4
    • List
  13. Marvey

    Marvey Super Friend

    Staff Member Friend BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,664
    Trophy Points:
    93
    Dual tone 37Hz and 130Hz at 0.5Vrms. Two different amps using HD650 as load. Measured at amp output. it could be that after a certain output, it doesn't matter and that below a certain output level, things still do matter.

    upload_2021-7-5_21-0-43.png

    upload_2021-7-5_21-3-23.png
     
    • Like / Agree Like / Agree x 4
    • List
  14. Marvey

    Marvey Super Friend

    Staff Member Friend BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,664
    Trophy Points:
    93
    I'm not trying to poop on Topping here so I don't think Yang has to prove anything. The L30 just happened to be the amp that I was currently testing when it was suggested that I use a headphone as the load. With respect to the 37Hz signal and HD650, the higher SINAD amps like Magni 3+ do just as poorly, while still being slightly worse with respect to THD at lower output levels and higher frequencies. If you believe that SINAD explains everything, and again I don't find fault with this approach, then I would recommend the Topping L30. Except...

    Now if you want to hear me "poop" on the Topping L30 amp and my concerns with respect to ASR receiving cherry-picked units, I suggest we go here: https://www.superbestaudiofriends.o...iew-and-measurements.11164/page-4#post-352845
     
    • Like / Agree Like / Agree x 2
    • List
    Last edited: Jul 5, 2021
  15. mocenigo

    mocenigo Acquaintance

    Joined:
    Sep 26, 2017
    Likes Received:
    26
    Trophy Points:
    18
    Location:
    Munich
    "One doesn't listen to the input of a headphone but the output" – which is also what Marvey is saying – the resulting THD is swamped anyway by the THD of the headphone. But some amps have this behaviour, some do not, and it would be interesting to properly understand why.

    The key argument here is consistency. One cannot rank equipment by SINAD - pushing people to prefer a device with 120Db of SINAD over one with 110Db - and then also say, oh, well, in that particular case the measurement is 80Db? Who cares, it is inaudible anyway. This does not strike me as consistent.
     
    • Like / Agree Like / Agree x 5
    • Epic Epic x 5
    • List
  16. Merrick

    Merrick A lidless ear

    Friend
    Joined:
    Jan 6, 2016
    Likes Received:
    8,495
    Trophy Points:
    113
    Location:
    Portland, OR
    That’s the game they’ve been playing this whole time. When it comes to recommending gear, it’s got to be the best measuring EVAR or it’s garbage. When it comes to defending faults in said gear, well it’s inaudible anyway.
     
    • Like / Agree Like / Agree x 4
    • List
  17. Brad358

    Brad358 Rando

    Joined:
    Jan 14, 2020
    Likes Received:
    73
    Trophy Points:
    13
    Location:
    England
    Having a cake and eating it. At a certain point what is called objective audio evaluation becomes just as subjective as the sort of florid prose people obsessed with measurement hate. Once a piece of equipment is below audible thresholds for metrics like SINAD then any further improvement is only of academic interest and buying gear based on such further improvement is a subjective preference. Now I get why people do it and if they enjoy then why not, but don't then sneer at people who admit to valuing other subjective aspects.
     
    • Like / Agree Like / Agree x 4
    • List
  18. JohnM

    JohnM Author of REW - Rando

    Joined:
    Dec 13, 2015
    Likes Received:
    127
    Trophy Points:
    33
    It may be worth trying a much shorter cable from the amp to the breakout board, the impedance of that connection will make a difference to the results. Or replicate the JY connections with a splitter at the amp output.
     
    • Like / Agree Like / Agree x 1
    • List
  19. Marvey

    Marvey Super Friend

    Staff Member Friend BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,664
    Trophy Points:
    93
    I've ordered a headphone splitter. Indeed I do have a longer run of cables, one from the test board to the AP (this one is short) and another from the amp to the test board. Curious to isolate the reasons from the discrepancy.
     
    • Like / Agree Like / Agree x 1
    • List
    Last edited: Jul 6, 2021
  20. mocenigo

    mocenigo Acquaintance

    Joined:
    Sep 26, 2017
    Likes Received:
    26
    Trophy Points:
    18
    Location:
    Munich
    True, but if there is EMF from the headphones, then the parallel measurement will show it.

    Whereas if it is a series measurement then you are actually measuring the signal through the headphone?

    I would say that to determine whether EMF plays a role, only a parallel measurement give an answer.
     

Share This Page