Trying to address high frequency peaks in IEM measurements: A halfway decent solution

Discussion in 'Measurement Techniques Discussion' started by La Cenric, Apr 17, 2018.

  1. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    65,539
    Dislikes Received:
    108
    Trophy Points:
    113
    Location:
    Antarctica
    That's nonsense. That's like saying it's wrong for God or whatever to make human beings with different sized ear canals.

    If a measured peak changes position from 9kHz to 12kHz because of tube length, and many trained ears don't hear it, then it's simply not there. This is not to say that a good swath of that mid-treble region cannot be bumped already.

    As for B2 or other deep insertion IEM, I don't hear a peak shift with deep insertion. What I hear is an attenuation of the highs relative to the lows. I know the difference between a peak 8kHz and 12kHz. Shoving the B2 deeper didnt make the ssss into a zzzz, it just made it less bright.

    Similar with UERM (customs vs less deep universal), MEE P1, and ER4S. Never heard the measured peaks change frequency based on insertion depth, although I am convinced there is a treble peak on ER4S and upper mid peak on P1.

    Now this is absolute nonsense. I don't hear 7-9kHz peaks caused by my pinna and ear canal on flat speakers in a damped room. Never have and probably never will. Brain has inverse HTRF function that takes care of this.

    You are sure you are not hearing these peaks from your Voxactiv wide banders?
     
    Last edited: Apr 18, 2018
    ultrabike likes this.
  2. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    65,539
    Dislikes Received:
    108
    Trophy Points:
    113
    Location:
    Antarctica
    Finally, I do not wish to argue further. I will only say that @La Cenric is on to something that may yield interesting results, better results in some ways, and that any further arguments (especially of the autisic nitpicky sort which insists on seeing the trees instead of the forest) about the merits of IEC / GRAS vs El-Cheapo only serves the purposes of false authorities with nefarious agendas and discourages further experimentation and learning.

    Now you know exactly why I kept the details of my headphone measurement rig secret for years and refused to divulge the details. I didn't want to waste my time getting into arguments with people who believed that Tyll measurements with the head and ears were from God. Was and is my headphone measurement methodology perfect? Hell no, but it does certain things amazingly well.

    Next steps are data. More measurements and more data. I will probably put something together like @La Cenric's short length coupler to help with data collection.
     
    Last edited: Apr 18, 2018
    Priidik, Serious, Vansen and 2 others like this.
  3. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Friend MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    9,518
    Dislikes Received:
    19
    Trophy Points:
    113
    Location:
    Irvine CA
    Adding some links to measurements done in the past on IEMs discussed in this thread using old DIY rigs.

    TF10:

    http://www.changstar.com/www.changstar.com/index.php/topic,399.0.html

    Discussion includes effects of different tips. Most measurements are cut to 10 kHz.

    Carbo Tenore:

    http://www.changstar.com/www.changstar.com/index.php/topic,1663.msg44226.html#msg44226

    In some cases, measurements were compared against other sources (such as GE) and there is a good amount of discussion.

    Difficulty in measuring IEMs is acknowledged.
     
    Last edited: Apr 18, 2018
  4. La Cenric

    La Cenric Friend

    Staff Member Friend IEMW MZR
    Joined:
    Dec 12, 2016
    Likes Received:
    736
    Dislikes Received:
    0
    Trophy Points:
    93
    Location:
    SG
    Home Page:
    Just for my own reference, what were the rigs (current and old) that both of you used? My IMM6 + tubing system should not yield wildly different results from, say, a UMIK mic with tubing and minimal differences <7k against a Veritas (version 1, not sure about version 2).
     
  5. La Cenric

    La Cenric Friend

    Staff Member Friend IEMW MZR
    Joined:
    Dec 12, 2016
    Likes Received:
    736
    Dislikes Received:
    0
    Trophy Points:
    93
    Location:
    SG
    Home Page:
    FR (CM).png

    Think I'll use this format for future measurements. Any objections or nitpicks?
     
    Augmentin, Vansen, ultrabike and 3 others like this.
  6. Serious

    Serious Inquisitive Frequency Response Plot

    Friend BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    1,872
    Dislikes Received:
    0
    Trophy Points:
    113
    Location:
    near Munich, Germany
    Looks good. Just to confirm, this shows that the 7.5kHz spike is insertion depth related while the 11kHz one is real? So the real curve would be something in-between.
    So this confirms what they wrote on their website:
    "6-8kHz frequencies(i.e. /t/ sound)are attenuated for sweet and non fatique-ing vocals yet not affecting brightness and openess of female vocal presentations"
     
  7. La Cenric

    La Cenric Friend

    Staff Member Friend IEMW MZR
    Joined:
    Dec 12, 2016
    Likes Received:
    736
    Dislikes Received:
    0
    Trophy Points:
    93
    Location:
    SG
    Home Page:
    10k sounds real to me, yes. The CE-5 definitely has emphasis in the cymbals/shimmer region, this has been confirmed by my other audiophile friends who've recently heard it as well. I don't think the entire band of 6-10k is emphasised as per the IEC graph; probably the resonance pulling up surrounding frequencies on measurement. Probably.

    EDIT: Just found this on their website as well:

    ce5official.PNG

    Not sure what equipment they're using. But there we go.
     
  8. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    65,539
    Dislikes Received:
    108
    Trophy Points:
    113
    Location:
    Antarctica
    Just use the ONE plot that best shows what you hear.

    Simplicity is best. You are not presenting plots for a scholarly journal. There is way way too much worry in IEM circles about "proper" methodology, accuracy, history, credentials, proper references, etc. 98% of it is misplaced. Examples of this would be overworship of Sean Olive's craptastic FR targets / consumer research (derived from test subjects with crappy hearing), or objectivist idiots thinking that such FR targets can be applied to any random HATS.

    You know what you are doing. Take a stand. I personally don't want to see other crap in a simple FR plot, especially if it makes no sense. I'm sure @ultrabike and @zerodeefex would also agree.

    Chuck that IEC coupler into the garbage. Or at least get ready to.
     
    Last edited: Apr 19, 2018
    Vansen and ultrabike like this.
  9. Lyander

    Lyander Too sensitive for SBAF

    Friend
    Joined:
    Feb 25, 2017
    Likes Received:
    6,678
    Dislikes Received:
    25
    Trophy Points:
    113
    Location:
    Pasig, Philippines
    Prose, fat untrimmed:
    I've been following this thread with interest because I thought it'd be useful; about a year back I got some el cheapo IEMs to act as makeshift earplugs for school (near a busy road, noisy as hell), but independent measurements I've seen of them online just didn't jive with what I was hearing. Thought I was even more treble insensitive than I actually was, which scared the crap outta me.

    It was only relatively recently that I found out how much more trouble IEM measurements are than headphone measurements, or why some members go out of their way to stretch out the y axis to exaggerate variances in FR while headphones don't get the same consideration.

    Frankly, people go to Tyll's measurements over at IF and though they (usually) know to make some adjustments for his compensation curve, it's his subjective impressions that really make or break headphones, at least as far as I'm aware. He's got trusted ears, so even though his hearing isn't what it used to be people still keep close tabs on his WoF. Except perhaps TonyNewman, but that's something else entirely.

    @La Cenric, I think you have enough experience with IEM measurements and, more importantly, being able to describe what you hear independent of measurements, to be beyond suspect. This thread has been extremely engrossing due to the technical talk and minor drama, but beyond all that it's been educational. atomicbob's threads require quite a bit of prior reading, enough so that I still second guess myself a lot when reading his content and wait for people with listening impressions to chime in; I think most readers would be perfectly content to have you post graphs that align with what you hear— you've more than earned that, I feel.

    Prose, fat trimmed:

    One of the most popular mods on here is called KISS, for Keep It Simple, S'il vous plait. Think that speaks enough about the mentality here ;)

    Besides, most people who've taken the time to read enough about measurements and their shortcomings to know that they do lie sometimes, and trusted ears are far more useful in the long term.
     
  10. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Friend MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    9,518
    Dislikes Received:
    19
    Trophy Points:
    113
    Location:
    Irvine CA
    If you are referring to my measurements, I have only one rig for IEMs. It is not perfect and it's old, but I get the gist of it. I'm using a fake plastic ear (small) and a cheap ECM8000 mic. It seems to work well indeed below 7k. But again, not perfect.
     
    Last edited: Apr 19, 2018
  11. La Cenric

    La Cenric Friend

    Staff Member Friend IEMW MZR
    Joined:
    Dec 12, 2016
    Likes Received:
    736
    Dislikes Received:
    0
    Trophy Points:
    93
    Location:
    SG
    Home Page:
    WIth what I can do on ARTA now, seems like I can't really generate one single plot as requested. What I'm doing is basically aligning the small-volume coupler measurement with the existing IEC graph and using the "Set as overlay above cursor" function to create the grey plots you see above. Been thinking about how to go about erasing the "false peak" but came up with nothing. This is the best compromise for now, I believe.

    The whole process of alignment is turning out to be more art than science and is based off trying to create a smooth transition, subjective listening experience, avoiding exceeding SPL values (i.e. grey plots typically wouldn't go beyond the IEC curves when properly aligned) and a little sprinkle of gut feel. I can already feel some of the objectivists completely dismissing my data from now.

    Still playing around with the software to see what I can do. I'm not touching photoshop, that's for sure.
     
  12. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    65,539
    Dislikes Received:
    108
    Trophy Points:
    113
    Location:
    Antarctica
    You might need more data, but how close can the short length coupler get to IEC results with compensation? If it's close enough, just go with a one measurement workflow.
     
  13. Biodegraded

    Biodegraded Friend

    Friend
    Joined:
    May 28, 2017
    Likes Received:
    4,067
    Dislikes Received:
    1
    Trophy Points:
    113
    Location:
    Vancouver BC
    Hoping disturbing 4 months of silence doesn't qualify as necroposting and no kittens will be harmed...

    Thinking about HF resonance peaks: wouldn't a good way to confirm & illustrate their postions be with CSDs? Example: the 3 curves below are for 3 different configurations of the HiFi Walker A1 measured on my 2-plastic-tube Dayton mic coupler (a modification of @La Cenric 's single-tube one). Note the strong peak at 7k with the stock tip (red), the shift of this to 8k with the other (orange; maybe partly be due to insertion depth, but let's leave that for now), and the elimination of the peak with the other tip and a tape filter (purple):

    [​IMG]

    Now the CSDs corresponding to each. Note the persistence and sharpness of the ridge around 6.5k with the stock tip vs the bluntness and lack of persistence of the 8k peak with the other tip, and the appearance of a ~6.5k ridge that doesn't correspond to a peak on the taped version (note I dropped the floor on the last another 10 dB to compensate for the lower 'starting point'):

    [​IMG]

    [​IMG]

    [​IMG]

    The peak does appear low in the orange data too (mostly below the floor here, dropping it makes it more obvious).

    This behaviour seems specific to this point. I've checked with other earphones, and ones that have peaks at 5k or 7k, or 5k and 7k, don't exhibit it there. It seems to need at least a bit of a change in slope there to show up, but unfortunately I don't have another set with a ~6.5k peak to compare.

    Am I correct to interpret this as coupler resonance?
     
  14. La Cenric

    La Cenric Friend

    Staff Member Friend IEMW MZR
    Joined:
    Dec 12, 2016
    Likes Received:
    736
    Dislikes Received:
    0
    Trophy Points:
    93
    Location:
    SG
    Home Page:
    The way I see coupler resonance is that if a point shifts when insert depth is varied, it's resonance. The issue with that is that since it's so dependent on the volume between the microphone and the earphone, the data that the microphone spews out is only indicative of the average ear. So there's the problem when the layman comes in looking at my (or any other people's) graph, pointing to a particular spike and claiming that it's sharp, when in reality the spike only manifests itself if your ears follow the physical properties of the coupler to a T (entirely possible but unlikely). That and the fact that one's perceived resonance varies based on insert depth, eartips, individual canal volume, size, shape etc. etc.. If the graph shows a strong 8k peak, your ears may manifest this as a 10k peak at a much lower magnitude.

    In your case, the 6.5k/8k peak seems to be the resonance point in their respective measurements. CSD might show if a peak is resonance or not but my own experience with CSD stuff doesn't seem to show any strong link. Looking for peak variations with insert depth differences seems to be the most reliable way; that was what formed the basis of my small-volume coupler experiment.
     
    Last edited: Aug 22, 2018
    Lyander and Biodegraded like this.
  15. Biodegraded

    Biodegraded Friend

    Friend
    Joined:
    May 28, 2017
    Likes Received:
    4,067
    Dislikes Received:
    1
    Trophy Points:
    113
    Location:
    Vancouver BC
    Thanks for the feedback. Have to admit, I wasn't too careful about matching insertion depth between the stock and replacement tips (in fact, in general I tend to insert different tips to different depths on purpose to try to mirror whether they sit deep or shallow in my ears), so you're probably right that the difference in peak positions reflects that. Nonetheless, it's interesting how sharp it is at 6.5 and how it persists. But I haven't done enough background reading on IEM measurements; I need to go through this thread, and your measurements ones, more thoroughly.
     

Share This Page