Calibrating MiniDSP EARS Part 2

Discussion in 'Measurement Techniques Discussion' started by purr1n, Sep 27, 2018.

  1. Lyander

    Lyander Official SBAF Equitable Empathizer

    Pyrate Contributor
    Joined:
    Feb 25, 2017
    Likes Received:
    10,906
    Trophy Points:
    113
    Location:
    Philippines, The
    I don't know who needs to hear this, if anyone at all is silly enough to need to, but dampening the inside of the chassis, around the microphones, or on the metal discs that support the silicone slabs will not aid in CSD or impulse response results. Home has been noisy as per usual so some variation in distortion measurements etc., but for the most part anything susceptible to vibrations will still remain wonk AF. You'd be better off building a flat plate coupler on the cheap or saving up for less bad fake pinnae.

    Now if you'll excuse me I have some Dynamat Xtreme gunking up my hair.
     
    Last edited: Nov 15, 2019
  2. Magnetostatic_Tubephile

    Magnetostatic_Tubephile Friend

    Pyrate
    Joined:
    Dec 27, 2017
    Likes Received:
    957
    Trophy Points:
    93
    Location:
    GoogleChrome, EU
    From my experience:
    • It is recommended to measure a particular headphone many times with differing positioning on EARS, since some measurement takes may result in less or even (almost) unpronounced 50Hz and 4.5kHz artificial resonance peaks known to be caused by EARS. ( = As with distortion measurements, the more measurement takes you check the better understanding you potentially gain.)
    • If the measured headphone has some significant issues with decaying/resonance around particular mid/treble frequency, CSD graph would show the behavior anyway. ( = If you know the limitations of EARS well, it is perfectly fine even for CSD use.)
     
    Last edited: Nov 15, 2019
  3. Serious

    Serious Inquisitive Frequency Response Plot

    Pyrate BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    2,589
    Trophy Points:
    113
    Location:
    near Munich, Germany
    Did you try felt in the fake ear canal and/or around the mic? Obviously depending on the frequency of the resonances the felt needs to be quite thick (and the density has to be right). For in-ear measurements where the resonances are in the 6-15kHz region depending on insertion depth I find simply putting the mic in comply tips to be adequate for how I hear in-ears. I think ultrabike or Marvey might still prefer additional damping inside the tube.
    Since the resonance frequency is less than half (open tube, plus additional depth gained by not sticking in-ears into the ear canal) that for headphones the felt would have to be quite thick to have an effect around 3-5kHz. 10-20mm should be effective, but will obviously need lots of compensation.

    Just a thought. Maybe the resonance isn't entirely from the ear canal in which case that won't really help much. I don't have an EARS - the results were off-putting enough for me not to get one, haha.
     
  4. Magnetostatic_Tubephile

    Magnetostatic_Tubephile Friend

    Pyrate
    Joined:
    Dec 27, 2017
    Likes Received:
    957
    Trophy Points:
    93
    Location:
    GoogleChrome, EU
    Check the below comparison of two random measurement takes that only happened 22 seconds apart (ZMF Auteur, the same channel, different positioning on EARS):

    [​IMG]

    [​IMG]

    The difference is very obvious around 4.5 kHz. The difference seems to imply that the resonance issue is probably indeed not entirely about the ear canal.
     
  5. Lyander

    Lyander Official SBAF Equitable Empathizer

    Pyrate Contributor
    Joined:
    Feb 25, 2017
    Likes Received:
    10,906
    Trophy Points:
    113
    Location:
    Philippines, The
    @Magnetostatic_Tubephile Yup, it's easy to know what to look out for in CSD results especially when you have a bunch of well-known headphones in the database to check for consistent trends and to compare to other measurements, but I'm a vain bastard who wants nice looking graphs that can be read standalone; since there doesn't seem to be much to do to compensate time domain issues I just went with brutal mechanical mods, haha. As for FR, I do go for averages of multiple measurements, just that I want to keep single measurements relatively neat since I think you might lose some perspective doing averages considering how position-sensitive many headphones appear to be.

    @Serious I actually moved the mic to be flush with the opening of the ear canal because I didn't have to want to deal with the issues associated with identifying and neutralising resonances borne of that weird silicone cylinder. Oddly it didn't help with impulse response performance, loads of ringing before peak still as was the case before, I'm wondering whether my dicking around might have knocked something loose.

    No IEM measurements, but I wasn't too fixated on those to begin with really.

    There was white sealant or something affixing the mics to the plate, completely covering the rear of the the mic capsules (is that what they're called?) and securing the wires; I made sure to clean up as much of that as possible when extracting them. I replaced it with Dynamat Xtreme just now hoping that'd help with ringing in IR.

    I'm gonna investigate more when I have time. Bit too sleepy and might have the weekend packed, but I'll report back asap when I can be arsed to do more measurements. I have some SBAF-comped squiggles from before I completely eviscerated the EARS so I have a sanity check of sorts. Using RAW comp with the mics flush actually seems to be working out well for me, though it shows things as much too elevated 7~9kHz, and upper midrange looks strangely level.

    For now, sleep. Thanks for the advice! Also good call not getting one, hahaha :)
     
  6. Serious

    Serious Inquisitive Frequency Response Plot

    Pyrate BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    2,589
    Trophy Points:
    113
    Location:
    near Munich, Germany
    Interesting, I guess that means it's related to the ears, too. Maybe the material is more reflective than human flesh, but tbh I expected a real ear to be close to acoustically hard at frequencies in the 2-6kHz band. Maybe it's the geometry of the EARS or there's still some slight damping from human flesh. Ages ago I did some quick experiments that seemed to indicate that there's indeed some damping from the skin.

    I find the CSDs to look very good with my in-ear mic measurements up to 7-9kHz where there's a resonance with every headphone.
     
  7. Lyander

    Lyander Official SBAF Equitable Empathizer

    Pyrate Contributor
    Joined:
    Feb 25, 2017
    Likes Received:
    10,906
    Trophy Points:
    113
    Location:
    Philippines, The
    Screw sleep, it's the weekend. I churned out some quick squiggles of the Klipsch HP-3 and the Sennheiser HD600. Do note distortion sucks butts in terms of actual performance because there's a karinderia/cantina/place where people hang out to drink and talk in outdoor voices just a little bit outside my window so I'm not including it. It's why I usually wait until 4AM to generate measurements when I can be bothered, haha.

    The first two are with the mics flush with canal opening and backed with a congested nose's worth of Dynamat. RAW compensation file from MiniDSP too. Second set of HP-3 measurements is before I decided to disembowel the EARS, stock SBAF comp off Marv's spreadsheet.

    HD600:
    [​IMG]
    [​IMG]
    [​IMG]
    HP-3:
    [​IMG]
    [​IMG]
    [​IMG]
    HP-3 (before eviscerating EARS, SBAF comp):
    [​IMG]
    [​IMG]
    [​IMG]
     
  8. Hands

    Hands Overzealous Auto Flusher - Measurbator

    Staff Member Pyrate MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    12,284
    Trophy Points:
    113
    Location:
    Colorado
    Home Page:
    I've noticed the ZMF headphones in general are incredibly sensitive, during listening or measuring, to slight differences in placement, angle, clamp, seal, and so on.

    Just to make sure, had you recently placed the Auteurs on the measurement rig for the first results? My thinking is that if you had placed them on the rig, squeezed them a bit to help seal around the ears, measured, and then waited, the change could be due to the pads expanding a bit.

    Otherwise, I would still wager some level of pad compression/decompression occurring slowly over time, even without interaction. It could be enough to suddenly cause the slightest amount of leakage that affects CSDs.

    Or, given their relative weight, how they're built, and stability, they may very well have shifted slightly in some way without human interaction. Angle them in such a way that the center of gravity is thrown off, or if perhaps your rig is not on a totally level surface, that could contribute. (Again, the ZMF stuff is not easy to measure nor particularly stable on a head compared to other headphones.)

    Even if you had them on there for a long period of time before the first measurements, it could be that the headphones conveniently shifted on their own in that 22 second window. You know, it's like how double sided adhesive might seem to be holding something vertically without issue...until it all of a sudden doesn't.

    Either that or some sort of software processing variability. Dunno. I don't use REW.

    I think it would be more telling to put a fresh HD600/650 on there, let them sit for a few days, then do a few measurements with 30 second gaps in between.
     
  9. Magnetostatic_Tubephile

    Magnetostatic_Tubephile Friend

    Pyrate
    Joined:
    Dec 27, 2017
    Likes Received:
    957
    Trophy Points:
    93
    Location:
    GoogleChrome, EU
    Will try to answer in bullet points:
    • I never apply artificial pressure or use any other means to enhance seal unless explicitely stated. I just position headphones in varying ways (letting them "adhere" on its own), check for any visually visible gaps, and measure. I usually do at least 15 measurements per channel. Of course, slightly different pad compression is always possible given the fact I obviously have to manipulate with the headphone when repositioning (and position always a bit differently.
    • The posted "22 second apart" measurements were not first in the line - they were actually among the last ones from that particular measurement session. Those two graphs are not the only ones that show such difference from that particular measurement session.
    • Indeed, ZMF cans seem to be very sensitive for placement.
    • The demonstrated difference in terms of 4.5k region is something I observed to various degree with multiple other headphones e.g. DT880, HD660S, Verum 1. Not with HD600/HD650 though - those seem to very consistently show the 4.5k artificial resonance.
    I am inclined to agree with @Serious that the dummy outer ears (and the headphone pads surrounding them during measurements) do have at least something to do with this but this is obviously just guessing.
     
    Last edited: Nov 15, 2019
  10. Hands

    Hands Overzealous Auto Flusher - Measurbator

    Staff Member Pyrate MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    12,284
    Trophy Points:
    113
    Location:
    Colorado
    Home Page:
    Interesting...Observed on HD660S but not HD600/650? Now that seems odd to me. There is probably a technical explanation for it, but it may be incredibly complex.

    Well, it's not problematic enough to the point where it necessitates a ton of experimentation. But if you do play around with it more and start to dial in on a root cause, I think some of us would be interested in your findings!

    Otherwise, for any publicly shared measurements, it's a matter of showing averaged results or posting 2-3 examples displaying variance. "ZMF Auteur may or may not ring very slightly in this area, listener depending."

    And it could be rooted deep in the headphone/driver designs and materials themselves, stuff that we at SBAF may not have the tools, the intelligence, or the patience to investigate. :)

    Thanks for the clarification!
     
  11. Lyander

    Lyander Official SBAF Equitable Empathizer

    Pyrate Contributor
    Joined:
    Feb 25, 2017
    Likes Received:
    10,906
    Trophy Points:
    113
    Location:
    Philippines, The
    Sorry for the necro, requesting a reality check.

    I've gone back to pushing the mics flush with the ear canal opening and stuffing the back of the capsule with Dynamat Xtreme gunk (trying to undo this was incidentally how I killed the right channel). The difference this time around being that I've taken the screws out because besides possibly messing with seal I worry about the screw heads scratching up pads.

    These are all single-channel measurements, level set by measuring 100dBA full-range pink noise— I do like that the lack of massive resonances in the ~4kHz allows me to push sweeps a bit harder to lower the noise floor.

    I still have to buy a replacement mic set eventually (which sucks bums but is entirely my fault), for now I'm just trying to figure out whether this configuration is feasible. Bit poo that the rig will then be useless for IEM squiggles but I guess I could always save up for a cheapo mic and play with tubes.

    Personally I really like how the upper mids look now compared to stock but think that the treble between 6-9kHz is shelved up by about 4dB too much and that "air" region needs a bit of a push up.

    (Not-exactly-stock) HD600 with very worn out pads:
    [​IMG]
    [​IMG]
    [​IMG]
    [​IMG]

    Very-thoroughly-used Klipsch Heritage HP-3 (with a good seal):
    [​IMG]
    [​IMG]
    [​IMG]
    [​IMG]
     
  12. SickMind

    SickMind New

    Joined:
    Dec 3, 2020
    Likes Received:
    3
    Trophy Points:
    3
    Location:
    Saint-Petersburg, Russia
    Hi all, new here.

    Need some help with understanding, what am i looking at.

    Just received miniDSP, measured my Utopias, left and right channels, with miniDSP HEQ applied:

    [​IMG]

    Well, L and R mics inside miniDSP seem to be almost equal, near +- 0,3 dB difference.

    But when i'm looking at another Utopia measurements, they look diiferent from mine:

    [​IMG]

    [​IMG]

    My proposal: in last 2 graphs some ear compensation is applied.

    Something like inverted ear resonance function:

    [​IMG]

    If I am right, is this resonance measurements available anywhere for miniDSP for applying it to measurements?
     
  13. Lyander

    Lyander Official SBAF Equitable Empathizer

    Pyrate Contributor
    Joined:
    Feb 25, 2017
    Likes Received:
    10,906
    Trophy Points:
    113
    Location:
    Philippines, The
    HEQ is their take on the Harman target which mandates a massive bass boost and has controversial response from the upper mids onwards. Besides the MiniDSP possibly having poor tolerances as is currently being discussed elsewhere, there's the usual caveats about operator consistency (cuz measurements can be a pain to get right!). That said, your Utopia squiggles do seem to resemble a lot of the ones I've seen on EARS rigs with HEQ comps, so it appears representative.

    Reference Audio Analyzer has always had weird compensation applied to their graphs that some disagree with. Seems internally consistent at least, based on various headphones they've measured, but I don't take those at face value. @Crinacle has a GRAS 43AG-7 as mentioned in the watermark and that's more representative of what a human ear is actually like compared to what MiniDSP offers (and yup he has compensations of his own, though the graph you shared is of uncompensated squiggles). That isn't a shot against MiniDSP, they do a fair lot for the price. Still, there's only so much that chunky silicone pinnae and properly cylindrical ear canals are capable of.

    Yes it's possible to change the compensation to better reflect what you might get on other rigs, just need to edit the values in the .txt file supplied by MiniDSP for your specific unit. I recommend making a new one and keeping what you have on as a reference, though you could always just download it again off their website if you encounter any problems.
     
  14. SickMind

    SickMind New

    Joined:
    Dec 3, 2020
    Likes Received:
    3
    Trophy Points:
    3
    Location:
    Saint-Petersburg, Russia
    Thank you very much. More knowledege - more suffering. As always, however.

    I'll receive Ollo S4X headphones in 3-5 days. And Ollo can provide measuring data for specific pair of their headphones. They use G.R.A.S 45CC using IEC 60318-1 ear simulator.

    So i think, if i can use headohones + data from calibrated measurement rig to calibrate my own EARS...
     
  15. Lyander

    Lyander Official SBAF Equitable Empathizer

    Pyrate Contributor
    Joined:
    Feb 25, 2017
    Likes Received:
    10,906
    Trophy Points:
    113
    Location:
    Philippines, The
    Might be, yeah! Just be sure that you agree with their target curve because some manufacturers tend to want to share more flattering measurements of their own gear. Sennheiser's infamous "the HD800 is neutral" measurements come to mind, though that's an extreme case. A more recent example would be HiFiMAN's showing that their R10 ripoff measures better than the original on their rig:

    [​IMG]

    Best of luck :)
     
  16. SickMind

    SickMind New

    Joined:
    Dec 3, 2020
    Likes Received:
    3
    Trophy Points:
    3
    Location:
    Saint-Petersburg, Russia
    Looks like truth to me. 75 Hz - aberration, don't look at it.

    Green - IEM neutral (miniDSP EARS)
    Red - Focal Utopia (raw measurements without any calibration)

    [​IMG]

    Upd: final version:

    [​IMG]

    To compare with Crinacle's measurements:

    [​IMG]

    Meze Empyrean (ME - red, IEF neutral - brown):

    [​IMG]


    download TXT-file with IEM neutral target for miniDSP EARS
     
    Last edited: Dec 5, 2020
  17. SickMind

    SickMind New

    Joined:
    Dec 3, 2020
    Likes Received:
    3
    Trophy Points:
    3
    Location:
    Saint-Petersburg, Russia
    Please approve my new calibration curve, derived from IEF curve: https://yadi.sk/d/OHL9s-VW_PfUJA

    Neutral responce should be flat line at 60 Hz.

    Logitech G Pro X:
    [​IMG]

    Sennheiser GSP-670:
    [​IMG]

    Meze Empyrean:
    [​IMG]

    Focal Utopia:
    [​IMG]

    By the way: modded EARS to make it wider.

    [​IMG]
     
    Last edited: Dec 22, 2020

Share This Page