Calibrating MiniDSP EARS Part 2

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

  1. james444

    james444 Mad IEM modding wizard level 99

    Pyrate Flathead IEMW
    Joined:
    Jun 3, 2016
    Likes Received:
    2,100
    Trophy Points:
    93
    Location:
    Vienna, Austria
    Well, there has been a new development since that post, because I "modded" my miniDSP EARS to make it more IEM-friendly.

    Basically, I used part of a 10mm plastic drinking straw to extend the ear canal. Makes it easier now to get a seal with medium sized silicone tips, no more need to cram them in like crazy. This, and the distance from nozzle exit to microphone (as well as trapped air volume) are closer to that in a human ear canal.

    [​IMG]
     
  2. Walderstorn

    Walderstorn Friend

    Pyrate
    Joined:
    Mar 20, 2016
    Likes Received:
    1,895
    Trophy Points:
    113
    Location:
    Europe
    Hahahaha a straw, that's kinda genius. Will try to replicate.
     
  3. Biodegraded

    Biodegraded Friend

    Pyrate Contributor
    Joined:
    May 28, 2017
    Likes Received:
    7,952
    Trophy Points:
    113
    Location:
    Vancouver BC
    Is there much difference in your measurements before and after? I'm wondering how different uncompensated measurements using your modified version would be from plastic-tube type couplers...
     
  4. james444

    james444 Mad IEM modding wizard level 99

    Pyrate Flathead IEMW
    Joined:
    Jun 3, 2016
    Likes Received:
    2,100
    Trophy Points:
    93
    Location:
    Vienna, Austria
    I didn't notice much of a difference, apart from shift of resonant peaks.

    The stock artificial ear canal is way too shallow for IEMs, so you usually end up with just 3-5mm distance between nozzle and microphone, because you have to cram them in for a secure seal. In a human ear canal, even with a deep fit, the distance between nozzle and ear drum is more like half an inch.

    Which is what I'm trying to get closer to with the extension, with the added benefit of easier seal.
     
  5. Walderstorn

    Walderstorn Friend

    Pyrate
    Joined:
    Mar 20, 2016
    Likes Received:
    1,895
    Trophy Points:
    113
    Location:
    Europe
    Can someone help me with step 8 of the manual? It says i should increase the volume until the dBFS meter reads the SPL calibration target. Which number is the SPL calibration target? Want to make sure i dont blow up anything.

    Am i to assume it should be -40?
     
    Last edited: Mar 18, 2019
  6. Josh83

    Josh83 Friend

    Pyrate
    Joined:
    Mar 1, 2017
    Likes Received:
    1,372
    Trophy Points:
    93
    Location:
    Charlottesville
    It will be slightly different from -40 (for headphones). You add the sensitivity factor from the sensitive channel in your calibration file to -40. It’ll end up being close to -42. See 2.6.1 (page 9) in the manual. So for each pair of headphones you measure, you’ll raise the volume to approximately -42, then hit the calibrate button.
     
  7. Walderstorn

    Walderstorn Friend

    Pyrate
    Joined:
    Mar 20, 2016
    Likes Received:
    1,895
    Trophy Points:
    113
    Location:
    Europe
    Thanks i will try this.

    I have a few questions:

    1) How do i use @purr1n proposed compensations?
    2 My REW software is very different from the manual that comes with the minidsp. For example it has loopback input and time reference output, what should i do with those?
    3) Which of the compensation files should i use? In the manual they use both and so i got confused.
    4) How do you guys get consistent seal in order to get consistent measures? If i move my headphones 1 mm it changes a lot, specially with headphones like the Verum 1.


    I tried to measure my Verum 1 and i got these results, first with flat eq compensation:


    [​IMG]


    With non-flat EQ compensation:

    [​IMG]

    This can't be right, can it?!

    Not to mention sometimes i get things like this just by rotating the pads on the Verum1 :

    [​IMG]
     
    Last edited: Apr 4, 2019
  8. JustAnotherRando

    JustAnotherRando My other bike is a Ferrari

    Contributor
    Joined:
    Aug 15, 2017
    Likes Received:
    1,373
    Trophy Points:
    93
    Location:
    Hong Kong
    I'd echo this, Verum measurements vary wildly on my rig.

    @Walderstorn what smoothing are you using? Even with the variance you're experiencing, your graphs are looking far more usable than mine.
     
  9. Vtory

    Vtory Audiophile™

    Pyrate MZR
    Joined:
    Sep 12, 2016
    Likes Received:
    10,826
    Trophy Points:
    113
    Location:
    East Coast
    1. Go to the following thread: https://www.superbestaudiofriends.o...g-sbaf-compensations-from-minidsp-files.7067/

    On top, you should understand a certain set of assumptions
    • Cruel truth: All EARS can have product to product variation.
    • Nobody guarantees whether Marv's SBAF compensation curve works for a specific EARS other than his. One possible way is to apply "differences" (e.g., sbaf compensation - flat eq compensation) -- this presumes that MiniDSP-provided mic compensations almost adjust product variation (kinda strong assumption).

    2. Manual relys on the old version of REW but newer version has good compatibility with legacy versions. You can simply ignore discrepancy.. but the best way is to go through REW online help carefully: https://www.roomeqwizard.com/help/help_en-GB/html/index.html

    3. Different compensations are derived based on different set of thoughts. I believe there is NO single right compensation as of late. You should understand why headphone frequency responses might not be easy. Here is a good introductory article to start from: https://www.innerfidelity.com/content/headphone-measurements-explained-frequency-response-part-one

    4. Some headphones are more sensitive to position changes, due to internal space around ears, sounding structure, and/or possible interactions among all such things. Your results perfectly demonstrate why proper seal is important to measure planars (and estats by extension). Measuring people have various beliefs regarding this issue. For example, I prefer to get averaged results from multiple different positioning.
     
  10. Walderstorn

    Walderstorn Friend

    Pyrate
    Joined:
    Mar 20, 2016
    Likes Received:
    1,895
    Trophy Points:
    113
    Location:
    Europe
    1/12th because i am just trying to follow the manual for now.

    Thank you very much mate, i think that averaged results will be the way to go. The biggest question, for me, though is which compensation should i use the stock for flat eq or the other one? The flat eq one gave me weird results so i used the other one the most. I will, for curiosity try his compensations and see the difference and check that innerfidelity article.
     
    Last edited: Apr 5, 2019
  11. Josh83

    Josh83 Friend

    Pyrate
    Joined:
    Mar 1, 2017
    Likes Received:
    1,372
    Trophy Points:
    93
    Location:
    Charlottesville
    In terms of consistent seal, you just need to play around. Your measurements look good. It’s common for there to be some variation because of seal and positioning.

    In order to use Marv’s SBAF compensation, you need to start from the HEQ compensation provided with the EARS.
     
  12. giovampi

    giovampi New

    Joined:
    Jul 31, 2019
    Likes Received:
    4
    Trophy Points:
    3
    Location:
    Norway
    Hello everybody,
    "long time listener, first time caller" here, I have been hovering around the MiniDSP EARS for a while now but I have read very conflicting opinions on the validity.
    I would like to, in the near future, either buy a EARS or build my own flat bed jig cloning the one used by diyaudioheaven.

    My use case is mainly to do comparisons between the same headphone, for instance to check what happens when I replace pads on vintage units with newer replacements. For example, I have a mint HD540, with the original pads looking good as new, I would like to find a good replacement before they start to crumble and it's too late to measure.

    I also would like to be able to compare the measurements with others from headphones I don not how. For example, I recently put together a Revox synergy (same housing of the old DT990) with DT 1990 drivers and pads, and I would want to see how close is it to the real 1990.

    What do you think would be the best choice ?

    Thanks
    Giovanni
     
  13. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,621
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    On the validity issue: well obviously you are here soliciting advice. Do you expect me to say that the EARS results are invalid? Haha. I will tell you this (keeping in mind that I have access to a fancy GRAS and various other head simulators):
    • Don't believe anything you hear from Jude / Head-Fi. He likes to e-peen about his expensive measurement gear and assert himself as the one true authority over all others. Ultimately, these are just tools. The humans doing the work matter more than the tools.
    • As all ear simulators are different, as just all human ears (pinna, concha, canal) are different. These differences will result in different measurement results. The GRAS will do certain things better. I won't go into specifics, but it's easier developing a compensation curve for the GRAS.
    • Let the results speak for themselves. All this talk about whether measurements from different systems are invalid or not need to be tested against what we hear. How well do the results correlate with what you hear? This is what matters most.
    • I would argue that the compensations are more important than the head simulator. We've seen Olive curves (often mistaken for perceptive neutral) recommend +5db bass boosts to headphones such as Audezes. Go figure on validity. However, I'd say the Olive curves (whatever iteration) might be valid for a "consumer" desired frequency response which tends to be V-shaped with more emphasis on the lows than in the highs. It's all context.
    FLATE PLATE COUPLER

    For comparing results of mods on the same headphone, we what we are looking for is relative differences. I'd say a flat plate coupler (no ear) does the job here. It's quick and cheap. The advantages of a flat plate coupler are as follows:
    1. Easily repeatable and consistent measurements.
    2. Moderate to high immunity from measurement differences in the highs caused by minor variances in positioning.
    3. High immunity from measurement differences in the lows caused by variances in seal. It's very easy to get a consistent seal.
    The major downside is the lack of ear (pinna, concha, etc.). We can miss out on a few things because some headphones (RAAL SR1a, HD800) will be more greatly influenced by the ear than other headphones (Verum, Audeze, Grado, etc.)

    Here are the layers of such a flat plate coupler. The top pad is a piece of felt about the size of an ear which helps absorb some internal reflections, Behind this piece of felt is a circular cutout of Creatology high-density foam. This foam really seals well with earpads. This foam is then supported by a CD. You can use any other solid material as this just serves as support. A cheap UMM-6 USB microphone is shoved through a block of medium density foam which serves as the pedestal for the above layers

    IMG_20190731_205034.jpg
    IMG_20190731_205018.jpg
    IMG_20190731_205010.jpg
     
    Last edited: Jul 31, 2019
  14. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,621
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    FOAM COUPLER

    In the last photo (repeated here below), we see what I would term a foam coupler (as opposed to flat plate).
    [​IMG]

    The foam coupler sucks for bass measurement. Some headphones with certain characteristics will be shown with better bass extension. Others will fair poorer. In other words, don't expect bass measurements below 500-1000Hz to be reasonably reflective of what is actually going on.

    However, the foam coupler is useful in certain situations. It has a crucial advantage: High immunity from measurement differences in the highs caused by minor variances in positioning.

    The low-medium density of the foam either absorbs or lets pass soundwaves to minimize the interactions of the head and ear with the cups. It's not that these interactions are inaccurate, but they can result in significantly different measurement results if the headphone is moved even a slight bit on dummy heads with ears and even with the flat plate coupler.

    This foam coupler is really good at seeing relative differences in the highs of headphones.

    Finally, it's worth noting that since a foam coupler will lack seal and a pressure cavity, therefore some types of headphones that depend upon a sealed cavity (orthos or fully closed dynamics) may exhibit sharper peaks where peaks exist.
     
    Last edited: Jul 31, 2019
  15. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,621
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    In this case, you are still using the same housing. A flat plate coupler (or foam coupler for the highs) would suffice, and would actually be ideal since you are looking for relative differences.

    Now if you want to compare measurements between different types of headphones, that is headphones with different structures in how they interact with the ear, it's probably better to get an EARS, GRAS, or another type of dummy head and ear.

    The bottom line is to use the best tool for the job.
     
    Last edited: Aug 1, 2019
  16. giovampi

    giovampi New

    Joined:
    Jul 31, 2019
    Likes Received:
    4
    Trophy Points:
    3
    Location:
    Norway
    It's been a wile, I'm sorry, but your post was very helpful.
    I gathered some material and made my own flat plate coupler.
    I couldn't easily get the UMM-6 here in Norway so I had to settle for a miniDSP UMIK-1.
    for the head I used two yoga blocks and I will try to add the felt and the cd to see if that makes a difference.[​IMG] [​IMG] [​IMG]
     
  17. giovampi

    giovampi New

    Joined:
    Jul 31, 2019
    Likes Received:
    4
    Trophy Points:
    3
    Location:
    Norway
    I'm comparing my results with what I can find on innerfidelity (the raw data) and diyaudioheaven, as far as I can see the measurements are good but I need to understand why is there such a rolloff after 12khz, it might be the microphone's dac.
    It might be off topic is so just let me know and I will remove the post, but here there are some samples of the measurements I'm getting:

    AKG K501

    [​IMG]

    Beyerdynamic DT 880 M (old one from 1984 with the original diffuse field equaliser )

    [​IMG]

    Beyerdynamic DT 880 (old one from 2003)

    [​IMG]

    Beyerdynamic DT 990 (old one from 1985)

    [​IMG]

    Sennheiser HD 580 (original one, not the Massdrop version)

    [​IMG]

    What do you think?
    and thanks again for the help
     
  18. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,621
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Rolloff could be microphone response. AD conversion even at 48kHz sampling rate shouldn't do that.
     
  19. giovampi

    giovampi New

    Joined:
    Jul 31, 2019
    Likes Received:
    4
    Trophy Points:
    3
    Location:
    Norway
    The advertised frequency response is 20 Hz - 20kHz +/-1dB and I'm using the calibration file they provide... I'm not sure, I will try to contact them.
     
  20. JohnM

    JohnM Author of REW - Rando

    Joined:
    Dec 13, 2015
    Likes Received:
    134
    Trophy Points:
    33
    I provided some support for EARS in the latest REW build, including allowing different left and right cal files and automatically calibrating the SPL readings from the cal file sensitivity figures, EARS gain setting and OS input volume. REW extracts the gain figure from the device name. Note that on Windows the name shown in the Windows Sound controls and in the mixer selections remains the one containing the gain setting when the device was first connected, even if it is changed afterwards. REW extracts the 'real' name containing the gain being used from the Windows registry. On macOS and Linux the name updates when the gain changes, but in all cases it is necessary to disconnect and reconnect EARS to pick up the new gain. The recommended process is to exit REW, disconnect EARS, change the gain, reconnect EARS and restart REW. The build is here.
     

Share This Page