MiniDSP EARs measurement rig

Discussion in 'Measurement Techniques Discussion' started by dBel84, Nov 9, 2017.

Thread Status:
Not open for further replies.
  1. Serious

    Serious Inquisitive Frequency Response Plot

    Pyrate BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    2,594
    Trophy Points:
    113
    Location:
    near Munich, Germany
    @Hands I gotta say I preferred the results of your real ears. Treble generally looked smoother, especially on the LCD2C, although the bass is probably more accurate now. I guess you can use the LCD-2C to compensate your rig as that's probably almost perfectly flat to 10Hz. Also I would've compensated for the >10kHz rolloff on the in-ear results. And, as I mentioned somewhere else already, I would also lower the 5kHz region a bit (about 3db for me) - despite the resonance I don't think the area sticks out on the HD600 to me. I also don't hear a peak here with the modded HD800, which my measurements would show otherwise. YMMV.
    It's interesting how the peaks shift in frequency a bit. The ears are definitely different.

    Still, it's nice to see that the results are mostly comparable. I bet someone else's real ears would be even more similar to your ears than the dummy head ears. The EARS rig also has the added benefit of more consistent, more reproducible results, I bet.

    Right now the 2.5kHz region looks a bit high in both measurements to me, at least compared to your in-ear FR plots, btw. Nothing too major, though.

    @Biodegraded Probably not much different from his regular CSDs. Some treble razors from the ears that will ruin the IR in my own measurements.
     
    Last edited: Dec 30, 2017
  2. Hands

    Hands Overzealous Auto Flusher - Measurbator

    Staff Member Pyrate MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    12,285
    Trophy Points:
    113
    Location:
    Colorado
    Home Page:
    Oh, wow! That's interesting. I hadn't thought something like that might happen.

    Yes, the funky impulse and CSD results, with me playing around with a comp. curve meant to mimic my previous in-ear results, was when the EARs mic was all the way back at the end of the ear canal.

    As it stands now, with the mic flush at the opening of the ear canal, IR and CSDs look like this for the HD650 w/ rubber bands in place. Basically business as usual.

    HD650 IR Flush Mic.PNG HD650 Hands Mods Left CSD EARs w rubber bands.PNG


    @Serious I agree that some elements of my older in-ear measurements looked more accurate, mostly in mids and treble. Or, at least, they were accurate to what I heard. But the EARs rig is probably more accurate with bass in most cases. It also has lower distortion capabilities too, so there are trade offs either way.

    Still, I would place a bet that human ears will have just as much variance as we see compared this EARs rig to my own ears, both with mics flush in the ear canal. Then you factor in head size/shape, hair type...it's nice to have consistency with the EARs rig, ignoring mic differences (as is likely to be expected with cheap mics).

    Another upside to the EARs rig is the matching, well, ears. My ears are slightly different, even down to ear canal size. I actually can't fit my IEM mic inside my right ear these days, which is why almost all measurements I do are just on the left channel (unless pads aren't angled or can be flipped, and assuming all elements of headphone, including enclosure, are symmetrical - then I do both channels on my left ear). My two ears alone produce varying results that honestly aren't far off from comparing the EARs rig to my old results. I used to take measurements of both channels on both ears and then measure them to even this out, but I can't do that these days.

    Plus...My IEM method is a huge PITA. Putting the rig together is cumbersome. The mic slips out of my ears all the time. That and using the mics actually leave my ear canal in pain for a good day or two because the glue I had to use, after the original glue failed, is way too stiff and almost cuts into my canals. I'm at the point now where I'd have to rebuild the whole IEM portion, which would likely mean a new mic too. And I bet the results wouldn't be totally comparable to the old mic, nor can you find serial number specific comp. curves for the WM61a mics these days.

    Basically all a long winded way of saying I'm tired of my old rig, want something that is easier to use and can produce consistent results with dual channel measurements, and possibly standardized with folks that own the same rig, even if it means some accuracy is lost. I'm just hoping I can get things close enough on the EARs rig to make that a reality. :)
     
  3. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,771
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    For reference, V2 coupler. LCD2C
    [​IMG]

    MiniDSP with calibration files. LCD2C. Not happy that their calibration files didn't take into account the L+R microphone differences. However, I definitely see a lot of potential here.
    LCD2C miniDSP.png
     
  4. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,771
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Same as above, but raw uncompensated measurements from miniDSP EARS.
    LCD2C miniDSP uncomp.png
     
    Last edited: Aug 5, 2018
  5. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,771
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Focal Clear L (same unit)

    GREEN = MiniDSP EARS L calibration
    YELLOW = RAW uncompensated from GRAS45AC
    GREY = RAW uncompensated from MiniDSP EARS L
    vs gras.png

    We can see the GRAS45AC has completely different characteristics from the MiniDSP EARS. Looking back to the prior post, the MiniDSP EARS seems to have an inherent peak at 4.5kHz. Can this be compensated out?


    SBAF V2 for reference

    GREEN = Clear
    GREY = Elex
    Clear V2.png

    Jude's Elex
    Jude Elex.png
     
    Last edited: Dec 30, 2017
  6. cskippy

    cskippy Creamy warmpoo

    Pyrate MZR
    Joined:
    Oct 14, 2015
    Likes Received:
    3,999
    Trophy Points:
    113
    Location:
    Tempe, Arizona
    I think Hands is on the right track moving the mics flush to the ear opening. I'll see about doing this if it's reversible.
     
  7. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,771
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    I'm not so convinced yet. The MiniDSP EARS with microphones mounted flush have less 4kHz fuckery, but at the expense of more mid-high treble up-down fuckery.
     
  8. cskippy

    cskippy Creamy warmpoo

    Pyrate MZR
    Joined:
    Oct 14, 2015
    Likes Received:
    3,999
    Trophy Points:
    113
    Location:
    Tempe, Arizona
    How are you going to get past the induced distortion peak and ringing with the impulse resopnse? REQ Wizard doesn't allow compensation curves for the distortion graphs. Just curious.
     
  9. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,771
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    I don't seem to be getting the wild IR ringing.

    LCD2C IR.png
    CSD.png
    (miniDSP FR compensation applied above)

    [​IMG]
     
  10. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,771
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Another possibility is to apply an EQ to the FFT then do an iFFT to get correct the impulse response. That is provide a compensated impulse response, which is really what we should be doing if we are going to show an impulse response. Math is fun.
     
  11. cskippy

    cskippy Creamy warmpoo

    Pyrate MZR
    Joined:
    Oct 14, 2015
    Likes Received:
    3,999
    Trophy Points:
    113
    Location:
    Tempe, Arizona
    I stopped at Calc 2 for a reason...:( Actually it was the Trig that got me lol!
     
  12. Serious

    Serious Inquisitive Frequency Response Plot

    Pyrate BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    2,594
    Trophy Points:
    113
    Location:
    near Munich, Germany
    Whew, that looks pretty bad. I didn't expect it to be that different from the GRAS, but it looks like those guys know what they're doing. The MiniDSP guys... not so much. The raw response of the GRAS looks much better to me. Still, it's weird to see that Tyll's head has a 6 and 9kHz peak with the Clear, but the GRAS has a dip between 8 and 10kHz. Is this position sensitive?

    Also LOL at the channel matching. Oh well. It is what it is
     
  13. Hands

    Hands Overzealous Auto Flusher - Measurbator

    Staff Member Pyrate MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    12,285
    Trophy Points:
    113
    Location:
    Colorado
    Home Page:
    @Marvey Can you provide results on your end for HD650 or similar, raw and with MiniDSP's bad compensation, FR, IR, and CSD? May also need to verify I'm using ARTA to gather measurements the same way you are to see if my funky IR results are something I'm doing wrong. I can't rule out error and misunderstanding on my end. I'm just so familiar with the 650 or Senns in general that the comparison would be helpful.

    I don't personally think the treble results look all that out of sorts with the flush mic. Like on the HD650, that sort of upper-end squiggly nature I've seen even on Jude's GRAS results to a somewhat similar degree. I know that GRAS supposedly has a good ear canal setup. I can't speak for what the LCD-2C should really look like on any rig, but it certainly sounds kind of rough to me despite overall good tone. I'm not familiar enough with it either way. Again, not ruling out that I'm missing something or flat wrong. Your observation on the flush mic is valid and may indeed mean less accurate results there.

    I still have more headphones to go through for additional data points.
     
  14. bartzky

    bartzky Acquaintance

    Joined:
    Dec 22, 2016
    Likes Received:
    62
    Trophy Points:
    18
    Location:
    Germany
    Home Page:
    @Hands: Have you tried to place the mic about 8-9mm closer to the ear opening than default (if there's enough room)? Looking at Marvey's measurements and doing some quick calculations I think that could move the resonances to the correct places.
     
  15. Koth Ganesh

    Koth Ganesh Friend

    Pyrate
    Joined:
    Sep 26, 2015
    Likes Received:
    1,637
    Trophy Points:
    93
    Location:
    Chennai, India
    I tried measuring the 650 today. Mine is the first version of the 650 mod - the coin mod, removal of foam and dynamat in the appropriate places. In addition, @gurubhai inserted Arctic Cotton in then circular space where the original foam pieces were. I am attaching the file (L side only) but this seems to go against what @Hands, @Marvey, amongst others are getting.

    Like I said, I'm a total noob at this so I apologize if this throws the whole conversation off track. I am using the compensation files that REW provides. I am not 100% sure about the fit of the 650 on the device.
     

    Attached Files:

  16. Hands

    Hands Overzealous Auto Flusher - Measurbator

    Staff Member Pyrate MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    12,285
    Trophy Points:
    113
    Location:
    Colorado
    Home Page:
    @bartzky That's an interesting idea. Do you want more detailed dimension measurements of the coupler, mic, and canal first?

    @Koth Ganesh Is that mod meant to reduce mid-bass? Possibly looks like you're not getting enough clamping force on the rig, but otherwise looks like I'd expect.
     
  17. bartzky

    bartzky Acquaintance

    Joined:
    Dec 22, 2016
    Likes Received:
    62
    Trophy Points:
    18
    Location:
    Germany
    Home Page:
    That would help! I was doing a bit of guesswork based on the measured resonance frequencies and plausible dimensions.

    Divergent resonance frequencies can actually cause a lot of pain while trying to calibrate a coupler. Imagine having a Kemar and Ears side by side. It appears to be easy to calibrate the Ears by measuring the same headphone on both rigs. At least one could easily calculate a calibration curve from the difference of both measurements. Let's say the Kemar records a resonant peak at about 6 kHz, just like in Marvey's graph from above. The calibration file would take this into account and will put out the same (but now fake) resonance peak in the measurement taken with Ears. Place the headphone on the Ears and the FR would look exactly like the Kemar's. But if you measure a different headphone things will get complicated as the resonance frequencies could be different. The Ears would still put out an artificial resonance peak at 6 kHz due to their calibration even if there was no peak at all. Now as there are some peaks that should be there and some that are caused by the Ears and shouldn't be there, some peaks will artificially be created and some will be reduced due to the calibration. As soon as the frequencies change, the measurement will be totally off.
    IMHO number one priority must be to have the resonance frequencies correct.
     
    Last edited: Dec 31, 2017
  18. Tyll Hertsens

    Tyll Hertsens Grandpappy of the hobby - Special Friend

    Pyrate
    Joined:
    Sep 26, 2015
    Likes Received:
    2,805
    Trophy Points:
    93
    Location:
    Bozeman Montana
    Home Page:
    I'll be doing this sometime in the next few months. Measure headphones on both. Produce difference curves. See if the difference curves are the same or different, and in what way. It'll take me a bit but it'll happen.
     
  19. Hands

    Hands Overzealous Auto Flusher - Measurbator

    Staff Member Pyrate MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    12,285
    Trophy Points:
    113
    Location:
    Colorado
    Home Page:
    Quick check gives a straight canal about 8mm in diameter with what appears to be a slightly tapered opening at the ear. Towards the tragus side, it's around 9mm deep, and about 11 or 12mm deep on the concha side. Almost exactly 10mm deep when measured at the top and bottom of the entrance.

    Of course, I don't have precision measurement tools. It could be the entrance isn't tapered. But, all in all, it's roughly 10mm deep and roughly 8mm wide. The mic itself is about 4.5mm in diameter and surrounded by a plastic ring to firmly hold it in place in the canal.

    Here are a couple pics to better help visualize. You will notice the material they use attracts every possible piece of hair, lint, and/or fuzz within a 3 mile radius.
     

    Attached Files:

  20. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,771
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    If the behavior of the peaks is linear, then all should go well. Non-linear behavior where a compensation doesn't work consistently across different headphones is what I am scared about.

    BTW here is the raw HD6XX from the GRAS45AC. No 6kHz peak here, but similar 9kHz dip as with LCD2C.
    HD6XX GRAS45AC.png
     
Thread Status:
Not open for further replies.

Share This Page