Improving my IEM measurement rig.

Discussion in 'Measurement Setups, Systems, and Standards' started by Serious, May 31, 2020.

  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
    I think it does mean something since we already know it's not linear. If it was linear there would be no FR change for my 8 vs 10mm tube, just a drop in level. Also what I mentioned in my reply in the Andromeda thread. For full disclosure, here's the data (mind this is only 3 IEMs) I referred to.
    We would think the FR difference between a simple tube coupler and an IEC 711 coupler would be something like this:
    [​IMG]
    When averaging the three IEMs, I got this:
    MRO IEC 711 vs damped 8mm tube average from UERM RE400 RE272.jpg
    Which is somewhat close considering I didn't match the insertion depth and didn't use a microphone compensation. Also I think it's an expected result since the input impedance of the 711 coupler really only has that characteristic "step" from 1-2kHz in it, where the eardrum transitions from stiffness controlled to mass controlled (see above B&K study from G.R.).

    However looking at the individual IEMs we see this:
    MRO IEC 711 vs damped 8mm tube UERM RE400 RE272 individual difference.jpg
    I would disregard the data from 4kHz on since I didn't match the insertion depth. With three IEMs it's hard to say. I'm not really sure if we can see anything here. But the 711 coupler does seem to show slightly more bass with both DD IEMs relative to the BA UERM FWIW*. B&K 5128 likely even more so given the above measurements.
    I could include my ADDIEM measurement, but that's about it. You've probably measured more IEMs that we have 711 measurements of than I have.

    *Note that for generating the difference I only took Rin's bassy left channel UERM custom measurement. It seemed to me he didn't get a good seal with the right channel. Using an average of L&R or using his universal sample measurement, the difference would be much greater.
     
  2. 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
    Might be worth it for me to investigate. Any 711 kits on AliExpess I should look into?
     
  3. Serious

    Serious Inquisitive Frequency Response Plot

    Pyrate BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    2,594
    Trophy Points:
    113
    Location:
    near Munich, Germany
    @La Cenric which is the 711 clone you ended up buying?

    Somewhat lengthy, but I wanted to add my opinion to some of your comments:

    That looks concerning. Any idea why it's so wildly different?

    The point is to create resonators on the side to get a volume or input impedance that more closely matches the ear canal and middle ear. It's what the couplers do. See my above B&K link: https://www.bksv.com/media/doc/bn0221.pdf
    It's old research from before Gunnar Rasmussen left B&K to fund GRAS, including an early version of the (now) 711 before it became an IEC standard. At the reference plane the ear seems to have a volume of roughly 2cm³ in the lows, lowering to roughly 1cm³ between 1 and 2 kHz where the eardrum becomes mass controlled (instead of stiffness controlled).
    Reference plane IEC coupler.jpg
    Tube length is a different subject and IMO the only proper tube length is a realistic one, which is why I always try to match the insertion depth in my measurements to the insertion depth I use in my own ear.
    Exactly! This is why I was surprised you didn't find the 5128 measurement above more accurate.
    I'm not sure when I can give the new CFA IEMs a listen. Would really want to give them a listen.

    I think it's only problematic since the measurement equipment is so darn expensive. Many companies probably can't afford it. If we had a blueprint for DIYing a decent clone of such a system on the cheap, there'd be no authority left. (Ha, I wish!)
    Seems however that @La Cenric's cheap-ish rig is surprisingly close to industry standard equipment (reddit post I linked in post 17).

    Probably rather large variance. Some data from two people in the B&K paper I linked. We already know eardrum thickness varies quite a bit, effectively changing sensitivity and HF response. I wouldn't be surprised to see a difference in stiffness aswell. The eardrum/middle ear system can be seen as a bandpass centered around 1kHz, limited in sensitivity by stiffness for lower frequencies and by mass for higher frequencies, if I understand correctly.

    To be honest I doubt it somewhat. However I do think that our ears are less ringy than typical vinyl tube and microphone couplers.
    But there are peaks and dips specific to our concha (as seen in dummy head measurements and also in my own in-ear method measurements) that might be beneficial to reproduce with in-ears. For example I see a 12kHz dip with many high quality headphones. IEMs at the proper insertion depth may have a similar dip aswell and our brain might interpret it as a concha reflection, leading to good imaging. Or something.

    Oratory1990 seems to think so aswell: https://www.reddit.com/r/headphones/comments/9o2f5n/psa_oratory1990s_list_of_eq_presets/

    Heck, we might gain some insight from that, too.
     
    Last edited: Jun 7, 2020
  4. Serious

    Serious Inquisitive Frequency Response Plot

    Pyrate BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    2,594
    Trophy Points:
    113
    Location:
    near Munich, Germany
    Also regarding acoustical output Z... wouldn't there be no difference in FR (just a change in level) from adding a damper to the bore if it was linear? I think that makes a good point why the coupler should have the proper acoustical input impedance.

    http://rinchoi.blogspot.com/2012/08/the-effect-of-acoustic-dampers-on-iems.html

    So basically this means that the acoustical output Z is high at frequencies where there are peaks, right? And very low at low frequencies. That might support that DD IEMs (which IME are more broadly affected by such dampers) have a higher output Z at low frequencies. Solaris has more change from different acoustic input Z than Andromeda.
    Anyway, next post should be more data with moar different couplers I'll try.
     
  5. Biodegraded

    Biodegraded Friend

    Pyrate Contributor
    Joined:
    May 28, 2017
    Likes Received:
    7,975
    Trophy Points:
    113
    Location:
    Vancouver BC
    https://www.aliexpress.com/item/4000789796521.html

    I think that one is the same as @La Cenric 's, but we should wait for him to confirm. There are others, too:

    https://www.aliexpress.com/item/4000073069259.html

    No! Red is supposedly a GRAS unit, blue a Chinese B&K clone and green an unspecified Chinese 711. On a tube coupler identical to mine, 20Hz-1kHz is pretty close to blue. The earphone is a 4BA+1DD hybrid.
     
  6. Crinacle

    Crinacle Friend

    Staff Member Pyrate IEMW MZR
    Joined:
    Dec 12, 2016
    Likes Received:
    812
    Trophy Points:
    93
    Location:
    SG
    Home Page:
    Based off my talks with other measurement guys, the main issue of the 5128 right now is the leakage as you can tell from Head-Fi's small sample size above (and also some Rtings measurements I have in private, so take my word on that). I'm not really confident in using or reading 5128 graphs until they fix that issue, which is (IMO) even more unreadable than flatplate measurements as of now.

    Also take note that the 5128's occluded-ear simulator does not have the same resonant dampening structure as GRAS' RA040X couplers, so you can kind of extrapolate what an undampened version of Jude's GRAS rig would measure like by looking at his 5128 measurements past 5kHz. At any case, rather than trying to stroke some e-peen by jumping from rig to rig whenever some new shiny "state of the art" toy pops up in the HATS world, I wish Jude can just buckle down, stay consistent and just measure more things rather than his current once-a-year graph publishing schedule.

    I believe it was this one, but I cannot be sure since I got mine directly off Taobao and not AliEx. 3-ish years ago, too.

    But yes, it is basically identical to any IEC60318-4 coupler built by the "Big Three" (GRAS, B&K, Head Acoustics) which is also confirmed by data analysis from Jaakko Pasanen, so I don't see a need to change it up. That said, I might be getting a GRAS 43AG-7 (with the RA0402 coupler) just to shut up the naysayers, but we'll see.
     
    Last edited: Jun 8, 2020
  7. Serious

    Serious Inquisitive Frequency Response Plot

    Pyrate BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    2,594
    Trophy Points:
    113
    Location:
    near Munich, Germany
    You think it's leakage? oratory1990 mentioned the ear canals are too small to insert many IEMs somewhere, but both IEMs seem to get the same bass extension from both couplers here, just at different levels. Also they averaged 40(?) ear canals, so I would expect the size to be realistic.
    When there's leakage in my measurements I see a bit of a resonant bump and then a roughly 12dB/octave high pass behaviour.
    Yes, but it does seem more damped to me than a regular 711 coupler. Maybe it's the 1/12th octave smoothing Jude uses.
     
  8. Crinacle

    Crinacle Friend

    Staff Member Pyrate IEMW MZR
    Joined:
    Dec 12, 2016
    Likes Received:
    812
    Trophy Points:
    93
    Location:
    SG
    Home Page:
    It's not a full seal break but there's definitely something going on that's causing this "roll off". Also exists on 5128 headphone measurements, not just IEMs.

    At any case yeah, if I had the choice I'd rather just get a KB500X for a canal that's at least "easier" for measurements.
     
  9. Biodegraded

    Biodegraded Friend

    Pyrate Contributor
    Joined:
    May 28, 2017
    Likes Received:
    7,975
    Trophy Points:
    113
    Location:
    Vancouver BC
    Dunno whether this will be useful to anybody or not, but after farting around for so long in REW & Excel I felt the desire to inflict it on people :p

    These are difference curves between Crinacle's measurements of 13 single-DD IEMs using his B&K IEC711 imitation vs measurements by me and @Brause using identical plastic-tube couplers and Dayton IMM-6 microphones.

    [​IMG]

    We have a few more measurements in common, but I didn't include ones with bad channel imbalances in one database or the other, or ones for which measurements of multiple pairs suggest large unit-to-unit variations.

    All curves were aligned at 1 kHz, hence the singularity there. The thick red curve is an edit of the average that is zero up to 1 kHz (I don't feel the differences there are indicative of anything systematic beyond seal, which will be tip- vs tube-size dependent) and that attempts to minimize the the big diference due to the spike centred on 8 kHz where Crinacle's coupler resonance is often aligned.

    I was expecting some variation, but hoping for something less than 2-3 dB between any input and the average. In the area I'd most like to correct my tube, 2-10k (beyond the 12-13k resonance of my coupler I don't really care), some corrections give a good match, some are too high, and some are too low, and there's variability in where within that range the max difference lies. Sure, there's likely to be some unit to unit variation reflected in the two datasets; but if the difference between a damped IEC-compliant system and an undamped tube + small free-field mic is linear, I'd expect to see a more systematic variation vs frequency than shown by these 13.

    I'll see how many common BA measurements I can find and see if there's any systematic difference between them and the DDs that might reflect driver-type impedance differences as @purr1n suspected; but given the results so far I suspect any real difference will get lost in the inter-unit variability.

    So my conclusion so far is that simple static corrections between DIY tube couplers and couplers that attempt to better imitate real ears by incorporating acoustic impedance and its variations with frequency won't be valid.
     
  10. assassin10000

    assassin10000 New

    Joined:
    May 9, 2020
    Likes Received:
    20
    Trophy Points:
    8
    Location:
    CA
    This may be a step in that direction ( but I'm sure there is a better way). I just made my DIY coupler using a mix of 2 sizes of vinyl tubing and some latex tubing (1/4" i.d.). The latex tubing is much softer than anything else I could find easily at the hardware store.

    [​IMG]

    It would have been nice if a static compensation file could correct for the difference, but not completely surprising that it won't work.


    That said, did you see the post on HF where fluid applicator needles were used simulate the volumes of the IEC coupler?

    https://www.head-fi.org/threads/relatively-cheap-headphone-measuring-kit.664900/page-3#post-13944137
     
  11. Biodegraded

    Biodegraded Friend

    Pyrate Contributor
    Joined:
    May 28, 2017
    Likes Received:
    7,975
    Trophy Points:
    113
    Location:
    Vancouver BC
    I did see that, interesting idea. Because the biggest differences between my DIY coupler and the B&K clone measurements are in different parts of the range, though, I don't think it'd be a universal cure.

    How is your latex version working out?
     
  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
    I know it's been a while, but I just saw that Amir posted some data for the 5128 for a couple headphones. I have to say especially the HF response part looks very promising to me.
    I would be more concerned if the difference in the bass response wasn't consistent between IEMs and over-ears. At the moment it seems to me that it is consistent and that a simple curve which lowers the midrange relative to the bass could work well. In fact the DF response for the 5128 seems to have a slightly different contour sub 1kHz than Tyll's HATS did already.
    I just find it surprising that dummy heads in general do not seem to get a good seal with HD6X0 style earpads the way it seems. I got a 5dB bump at 100Hz for the HD650 on my head and no dummy head seems to measure such a strong midbass boost.

    You should be able to more or less copy the "professional" couplers by adding volumes through a tube. They don't really do much more than that, etiher. Just that it's in a much more professionally machined part:
    [​IMG]
    As I mentioned in Post 17 in this thread I was initially going to try that myself, however after listening to the phase shift from the multi-BA IEM crossovers I have to say that it's probably not necessary. See here.
    I will likely not bother making my own IEM coupler (at least not any better than it is) and I don't think I will generally post IEM measurements. Why bother when we already have the insane crinacle database covering most interesting IEMs. Also I think subjective impressions can be more accurate than measurements currently.
    I'll post some additional measurements for the IEMs I have. IMO the excess phase plot is critical and needs to be included as it does affect the frequency response perception.

    When taking the excess phase into account the difference between my current measurements and what I hear between my BA and DD IEMs really isn't as big anymore. It's still there, but the excess phase explains parts of what I subjectively thought were FR magnitude differences.
     
  13. 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
    Looks like it's all over the place for DDs at least. I can totally see why considering the tight seal cavity and the differences with DD implementations, with some being vented, variances in electrical impedance, sensitivity, etc. I'll cede exploration of this area to @La Cenric because he's got the best handle on it right now.
     
  14. assassin10000

    assassin10000 New

    Joined:
    May 9, 2020
    Likes Received:
    20
    Trophy Points:
    8
    Location:
    CA
    It works ok still tweaking it with different lengths.

    I did get pretty significant differences between silicone and foam tips. Partly because I don't have any large foam tips on hand, so I'm not getting near as good a seal as I should.

    Part of it could be due to the missing volumes as with the silicone tips there is a sharp rise between 1-2k, which that first volume dampens, if I understand correctly. The foam tips surprisingly completely smooth out that rise, its its soft surface acts as a damper.


    Ordered more tips, so further experimentation will have to wait.


    I'm considering trying it myself. Just need to figure out what to use, how to implement it and calculate (or verify) the volumes.
     
  15. Crinacle

    Crinacle Friend

    Staff Member Pyrate IEMW MZR
    Joined:
    Dec 12, 2016
    Likes Received:
    812
    Trophy Points:
    93
    Location:
    SG
    Home Page:
    Currently the biggest differences for the 5128 would be on the IEM side. Also, Amir's observation about IEMs being ridiculously hard to couple onto the 5128's canal matches with Sam's (Rtings) experience, so that might affect the bass response for a lot of IEMs.

    At any case, the biggest problem now with the 5128 is that it's essentially "reinventing the wheel" and going against IEC60318-4/-7, so whatever data's generated on it is not going to be immediately comparable to existing "industry-standard" measurements. Which is quite the uphill battle considering that majority of IEM measurements are now done on 711 couplers, and most mainstream headphone FR databases are also done on IEC60318-7 compliant systems (Rtings/InnerFidelity with the Head Acoustics HMS, and Oratory1990/0db.co.kr/ClarityFidelity with the GRAS 45CA).

    Probably not going to be anytime soon, unfortunately. Currently prepping for the headphones measurements database and all that, which would probably take up most of my time for the next year or so.

    I'll still continue to gather FR data for anyone who wants to take a stab at it though. Good luck to that fella!
     

Share This Page