EARS FR Measurement Dump (ZMF, Focal, Audeze, Extras)

Discussion in 'Audio Science' started by E_Schaaf, Jul 22, 2019.

  1. E_Schaaf

    E_Schaaf MOT: Upscale Audio (Thread Judge Code-6 District)

    Friend BWC
    Joined:
    Mar 28, 2017
    Likes Received:
    2,307
    Dislikes Received:
    5
    Trophy Points:
    93
    Location:
    Pasadena, CA
    Home Page:
    Looking for feedback on my compensation curve - I used the SBAF-approved tweaks to the supplied MiniDSP curve and made a few adjustments.

    In general I'm seeing some weird artifacts on some headphones but not others (like a 5-7k recession on the Focals and a 4k push on the HD800S and HE6). Bass below 100Hz seems to read a bit lower than it should too. Not sure if this is just EARS variation or a problem with measurement technique, even though I did at least 5 repositions on each headphone (always maintaining the best seal I could) and captured the measurement that seemed least erroneous.

    Anyway, here's the dump -

    Focal:

    Utopia -
    [​IMG]

    Clear -
    [​IMG]

    Elear -
    [​IMG]

    Elear w/ Clear Pads (AKA Elex) -
    [​IMG]

    Stellia -
    [​IMG]

    ZMF:

    Stock pads (meaning the pads shipped pre-fitted to the headphones) used on all.

    Verite -
    [​IMG]

    Auteur -
    [​IMG]

    Aeolus -
    [​IMG]

    Eikon -
    [​IMG]

    Atticus -
    [​IMG]

    Audeze:

    LCD-3F (2015) -
    [​IMG]

    LCD-4 (200 Ohm, 2016) -
    [​IMG]

    Miscellaneous:

    Klipsch HP3 -
    [​IMG]

    Abyss Diana v1.0 (IMPOSSIBLE to get a good seal on this one) -
    [​IMG]

    Sennheiser HD800S (this one looks F'd) -
    [​IMG]

    Audio Technica ADX5000 -
    [​IMG]

    LFF Code-6 (universe suede pads, HE6 6-screw) (this one and the ESX6 were measured last night - same compensation curve, hadn't calibrated the Y-axis level yet but taken at the same actual SPL // will re-measure and edit ASAP) -
    [​IMG]

    ESX6 (universe suede pads, HE6 4-screw) -
    [​IMG]
     
    Last edited: Jul 22, 2019
    purr1n, sealand1, WHO23 and 6 others like this.
  2. Lyander

    Lyander Too sensitive for SBAF

    Friend
    Joined:
    Feb 25, 2017
    Likes Received:
    4,361
    Dislikes Received:
    23
    Trophy Points:
    113
    Location:
    Pasig, Philippines
    Does your EARS rig still have the screws in? I definitely notice that those might screw (HAHA) up the seal on some headphones. I take a similar approach, getting as good a seal as I can manage so that bass is always best/worst case scenario (depending on taste), the loss below 100Hz might be due to that?

    The lower treble area is definitely weird, that's what broke me and made me give up tweaking the stock curve for now. There's only so much that can be done there I think, because of how weird the fake pinnae are.

    Also, this might just be a preference thing but I think it'd be cooler if you decompressed the y axis a bit, make it easier to see differences between headphones and channels, if you feel like doing channel matching tests so you can inadvertently figure out that it was the headphones that were bollixed and not your own ears (cough)
     
    E_Schaaf likes this.
  3. E_Schaaf

    E_Schaaf MOT: Upscale Audio (Thread Judge Code-6 District)

    Friend BWC
    Joined:
    Mar 28, 2017
    Likes Received:
    2,307
    Dislikes Received:
    5
    Trophy Points:
    93
    Location:
    Pasadena, CA
    Home Page:
    @Lyer25 Yes, the screws are still in. From what Marv said in the other EARS threads it seems taking them out isn't quite a cure-all so I simply didn't bother. I'm sure I'll end up trying that anyway though. Weird how the loss of bass seems most apparent on Auteur and the HE6s, which should be almost or completely ruler-flat (respectively) like the Audeze ones I measured. Maybe because the pads are thinner?

    I wouldn't be so worried about the artifacts if they showed themselves consistently - but they don't. Why there's a huge suckout or bump on one headphone and not another (and not simply due to the actual FR of the headphones, since we've seen these in the flat-plate and other EARS measurements) completely baffles me.

    I guess at the end of the day, taking measurements is really just for my own use - as long as I can establish some semblance of a pattern or reference in my head, that's good enough for seeing relative differences with mods.

    From here on out I'll take the measurements with 60dB of range on the Y-axis vs 100dB if that'll make it easier to read.
     
    Last edited: Jul 22, 2019
    Lyander likes this.
  4. Lyander

    Lyander Too sensitive for SBAF

    Friend
    Joined:
    Feb 25, 2017
    Likes Received:
    4,361
    Dislikes Received:
    23
    Trophy Points:
    113
    Location:
    Pasig, Philippines
    To add to that, a somewhat lossy seal would also be representative of most people's use cases yeah? Fair point, my screws are still in too, I plan to remedy the too-low sub-100Hz (feel the same way about my squiggles) via compensation.

    It does make sense to me that thinner pads might mean the mic will be more easily affected by driver motion, though I've nothing to test that theory with just now. Haha, any pads on your end that are similar in every way but thickness?

    I KNOW. IT'S FRUSTRATING. The only hypothesis I have is that the orientation of drivers in some headphones interact with the couplers more than others. I just take a ton of measurements and either average them or pick one set that looks right, it's less troublesome in the long term. Definitely plan to point out any weird artefacts I come across, which kinda stresses the importance of taking multiple measurements.

    Agreed, going entirely by measurements and discounting the importance of subjective impressions goes against the forum philosophy and conventional wisdom, haha. They'd be useful in certain contexts like modding and in substantiating comments on one headphone v another, but definitely don't pull a me and lose the forest for the trees :))
     
    E_Schaaf likes this.
  5. Jinxy245

    Jinxy245 Vegan Puss

    Friend
    Joined:
    Nov 17, 2016
    Likes Received:
    3,276
    Dislikes Received:
    0
    Trophy Points:
    113
    Location:
    Montgomery, New York
    But they are loose
    :p
     
    Lyander and Syzygy like this.
  6. Lyander

    Lyander Too sensitive for SBAF

    Friend
    Joined:
    Feb 25, 2017
    Likes Received:
    4,361
    Dislikes Received:
    23
    Trophy Points:
    113
    Location:
    Pasig, Philippines
    Writing this bit last: reading the below, I'd like to reiterate that I'm aware the MiniDSP EARS isn't made for this sorta thing and that I very likely know far less about how headphones work than most others on here. Please do feel free to let me know if I'm overlooking something ridiculously simple or am being ridiculously simple :p


    Bumping this thread to ask whether you've had any luck with impulse response measurements, @E_Schaaf? I've got three headphones lying around and it seems to me some headphones are weirdly more likely to wobble for decades after initial pulse than others. Not enough headphones on hand to determine whether it's related to some headphones interacting more with the pinnae and relying on a seal to perform optimally.

    Noisy right outside with loads of birds chirping constantly and some motorcycles gunning their engines (though I abort measurements when I hear any of the latter), but same conditions for at least the Beyers and the Klipsches. Mezes are on loan to a friend just now. I have taken the Beyers apart and it looks to me like the insides of the cup are pretty well-stuffed (https://i.imgur.com/3GcCzVF.jpg), and FR measurements show a less pronounced loss of bass than the Klipsch when the seal is broken (-14dB at 30Hz for the HP-3, -9dB for teh DT880), making me think the Beyers aren't as reliant on getting a good seal (Klipsch is red +green, Beyer is blue + orange: https://imgur.com/a/LbGj1ha)

    Makes sense considering they ship with velours I guess.


    KLIPSCH HP-3:
    [​IMG]

    MEZE 99 CLASSIC:
    [​IMG]

    BEYERDYNAMIC DT880:
    [​IMG]

    I wondered whether it made sense that some headphones's IR performance relied more on getting a decent seal (cushioning effect and all that) than others so broke the coupling with something I just had lying on my desk, a Star Wars cable organiser thing (https://imgur.com/a/oJELo4K) and retook impulse measurements of the Klipsches and Beyers. The DT880s' IR was transposed a few μs, while the HP-3s' just goes bonkers crazy, increasing in ringing and looking even more like soggy spaghetti.

    DT880 busted seal:
    [​IMG]

    HP-3 busted seal:
    [​IMG]

    What's interesting to me is how features on some headphones like the way the Klipsch seems to second-guess itself on the second peak before coming back for a stronger third peak remain constant, making me think that has to do with stuff behind the driver. @cskippy's IR measurements of the Klipsch (flat plate?) don't necessarily show that "hesitation" in second peak, but does have a stronger third peak than second as well.
     
  7. Biodegraded

    Biodegraded Friend

    Friend
    Joined:
    May 28, 2017
    Likes Received:
    2,452
    Dislikes Received:
    0
    Trophy Points:
    93
    Location:
    Vancouver BC
    Something is off with these measurements. Up to 400 usec of pre-ring and 3 ms or more of post-ring are way too much, and the curves look too symmetrical about the y-axis to be 'real'. Also, note how the two DT880 measurements appear to have opposite polarity at the zero timing reference (I'm guessing you have 'set t=0 at IR peak', which lets REW pick the tallest, selected in the prefs).

    In the 'Measure' dialog, are you using multiple sweeps? What sweep length? I have different ADCs (cheap USB dongles) that screw up IRs to different degrees as sweep length increases. @Serious has found this as well. If you try generating CSDs from these IR data, they'll probably look pretty different at 0 decay time from your FR measurements. @purr1n mentioned elsewhere he was having trouble generating meaningful CSDs with the EARS. Maybe it's inherent to the ADCs in these mics?
     
    Lyander likes this.
  8. Hands

    Hands Overzealous Auto Flusher - Measurbator

    Staff Member Friend MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    8,914
    Dislikes Received:
    29
    Trophy Points:
    113
    Location:
    Colorado
    Home Page:
    It’s the fake ear canal. It doesn’t offer a load like a real one, so it causes some sort of resonance.
     
    Lyander, purr1n and Biodegraded like this.
  9. Biodegraded

    Biodegraded Friend

    Friend
    Joined:
    May 28, 2017
    Likes Received:
    2,452
    Dislikes Received:
    0
    Trophy Points:
    93
    Location:
    Vancouver BC
    Ah, right. That would explain it.
     
  10. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    49,536
    Dislikes Received:
    81
    Trophy Points:
    113
    Location:
    Antarctica
    It's the fake ear canal with the impulse response. GRAS has similar same behavior with it's canal, just not as bad. It's probably best to do a free air or foam coupler if we want to see impulse behavior of the driver, otherwise the impulse response will more reflective of the ear and related structures.

    As for the measured low-end drop on, I wouldn't worry about it on the HE-6. Most of the HFMs and some Audezes and other plans do that below 30-35Hz. Could be seal or a tiny gap between pad and cup on the HFMs (note the Audeze pads are glued). You don't notice on 99% of music because 99% of music has very little content below 35Hz. My JBL speakers with the 18" woofers start to roll off below 37Hz. People would not notice this with music.

    The hard part of these measurements is that they are just as much about measuring the ear, the canal, concha, pinna, as the headphone. It's both an art and a science of take the ear aspect part away from the measurement.
     
    Last edited: Sep 22, 2019
    Lyander and Biodegraded like this.
  11. Lyander

    Lyander Too sensitive for SBAF

    Friend
    Joined:
    Feb 25, 2017
    Likes Received:
    4,361
    Dislikes Received:
    23
    Trophy Points:
    113
    Location:
    Pasig, Philippines
    @Biodegraded for the record yup, t=0 at IR peak, and I do a single 1M sweep (multiple sweeps just don't work on this), which seems to work well for generating FR but if greater sweep length does bollix IR results then I might try comparing with shorter ones. Of course if the silicone does just make it impossible to get anything decent out of IR then I might as well save up for a decent mic and a cardboard box for dedicated CSD etc eventually :p

    I did catch in the older threads about how CSDs and IRs are pretty much non-starters with the EARS, but just based on some cans I've been able to make cruddy squiggles for I was curious why some seemed to trigger resonances in the fake pinnae more than others.

    I also noticed the polarity weirdness on the same headphones. I figured at first it might have had something to do with physical mods (e.g. front filter or rear dampening would inverse polarity), but I realised after a while that just changing placement or adjusting clamp force sometimes had the same effect, which for a second made me wonder wtf was going on. More EARS weirdness, I can believe XD
     
    Biodegraded likes this.
  12. E_Schaaf

    E_Schaaf MOT: Upscale Audio (Thread Judge Code-6 District)

    Friend BWC
    Joined:
    Mar 28, 2017
    Likes Received:
    2,307
    Dislikes Received:
    5
    Trophy Points:
    93
    Location:
    Pasadena, CA
    Home Page:
    Here's stock measurements of @sheldaze HE5SE. Haven't changed my comp curves from before.

    The headband on these things is a PITA - had to carefully apply pressure to get a good seal. They measured consistently with subtle movements and re-positioning though.

    [​IMG]

    Here are Marv's measurements also with EARS + SBAF comp curve of the re-issue HE5 (non-SE) for comparison.

    [​IMG]

    My re-issue HE5 measured more closely to Marv's results back before they were completely disassembled. Not sure if the HE5SE difference is simply due to different pads and housing or if the drivers are different as well. Might be different drivers - the re-issue HE5 has always shown that 700hz dip.

    My goal with the mod is to bring down that 3.5k area to be somewhere in between the response up to 1k and the suckout at 1.5k, and shelf down / smooth out everything above that point. And find a way to tighten up the dynamic response - these feel slightly soft and loose, compared to the HFM of yore. Definitely not the same drivers as OG HE5.

    Edit - the magnetic assembly is clearly different.

    OG HE5 (4-screw, not re-issue). this driver is symmetrical earside and outside. The 2017 HE5 reissue drivers did look identical to this, aside from having 6 screwholes.
    [​IMG]

    HE5SE Earside
    [​IMG]

    5SE Outside
    [​IMG]

    The magnets are clearly asymmetrical (across the entire surface on the backside but with a central hole on the earside). I'm also a little worried about the 'floating' of the stock felt on the 5SE. It's cleanly adhered to the driver's face on the OG. I do appreciate the metal mounting rings on the 5SE (not pictured) in addition to the six screwholes - it does make it easier to hold all of the materials in place without leakage than the original plastic rings, which warp and bend. I might pick up some metal mounting rings for my older HFMs too, hopefully the screwholes line up...

    @zerodeefex maybe we can move this post to one of the HE5 / re-issue threads? I'm realizing another place may be better for this post.

    EDIT - One pad swap and 7 materials later...

    [​IMG]
     
    Last edited: Nov 7, 2019
    Vtory, Serious, sheldaze and 5 others like this.

Share This Page