Search This Blog

Wednesday, May 29, 2013

Etymotic Research CUSTOM•FIT ear sleeves: An inconvenient truth?


Two upper rows: made by DST SWISS
Two lower rows: made by ACS UK
Dr. Mead C. Killion, the founder of Etymotic Research, used to say: "Adding custom earmoulds to high-fidelity earphones could compromise fidelity." Then he also added: "but we found a partner in ACS that allows us to maintain the Etymotic sound signature while providing an extra level of customisation."

Thus, with Dr. Killion's quality verification, I had to give a shot, which turned out to be extremely unsuccessful  in terms retaining accuracy of the stock configuration. After four trials of re-molding efforts gone to no avail, these eight CUSTOM•FIT sleeves I ended up with were completely abandoned: Not only bore lengths did not match between each channel, but also the insertion depth of these sleeves was not secured as deep as recommended by the manufacturer.

Even though Dr. Craig Kasper of Etymotic ResearchACS tried the best he could to accommodate my concerns, it seemed the technicians simply did not get what the fundamental problem was with their approach.


Here are the first two pairs, which were manufactured by DST SWISS, analyzed at the reference plane. Even if they are situated at the manufacturer-recommended insertion depth, they still exhibit sharp peaks below 10 kHz, indicating a sign of miscalculation with the bore size / insertion depth.

The peaks are located at 7.4 kHz and 8.5 kHz respectively, which means, if the bore lengths were decreased by 4.6 mm on the left and 3 mm on the right(with the same insertion depth), or the insertion depth was increased by 10 mm on the left and 7 mm on the right would've pushed the peaks to the frequency range above 10 kHz, and simply solved the problem.

And here are the last two pairs, which were made by ACS UK, analyzed at the reference plane. Their accuracy is actually worse than that of the DST SWISS pairs, as bore lengths are further elongated without offsetting the insertion depth, pushing away the quarterwave resonance of the IEM downward.

There is, however, a way to compensate the insertion-depth induced resonance shift, according to Mr. Don Wilson, the mastermind behind the design of ER-4 (provided by Mark, who is a regular visitor of my blog. Thank you!):
"You can remove the filter from the ER-4 and place it 3mm from the end of the earmold and the response will be good. As long as the hole from the end of the driver to the end of the earmold is 2mm the filter should fit good. Part of the purpose of the filter is to terminate the tube removing any resonances."
Yet, above comparison indicates there is definitely a lot more to than what meets the eye as far as the location of a damper goes, unlike what Mr. Wilson stated.

According to Dr. Harvey Dillon,
"...Dampers are most effective if they are placed at locations where the resonance causes the largest flow of air particles. For wavelength resonances, particle velocity is least at the end of the tube that joins the receiver..." 
What we must understand here is that, such notion of acoustic damping is indeed applicable in hearing aid acoustics, of which the required bandwidth is quite limited, but is not applicable in high-fidelity IEM acoustics; The result of Mr. Wilson's damper location adjustment simply cuts away the bandwidth above 8 kHz, leaving ER-4 further crippled in fidelity.


References

H. Dillon, Hearing Aids, Boomerang Press, Turramurra, Australia, 2001.

35 comments:

  1. So this is the correction I came up with using your graphs.
    http://i.imgur.com/0k4hZeG.png
    No idea if this is a pointess effort but I'll try it out on my favorite recordings, in conjunction with the Oliver-Welti correction, and report back when I get more time.

    ReplyDelete
    Replies
    1. That is great, Mark. I hope other people also take advantage of my data as much as possible. Makes my effort worthwhile! :)

      Delete
    2. I'm struggling with finding a recording where the above correction filter sounds worse with than without, right now. Pretty promising for at filter that you wouldn't expect to be one-size-fits-all.

      Delete
  2. I have worked at Etymotic Research for a long time, but have no idea who Dr Craig Kasper is. I can assure you he doesn't work at Etymotic Research.

    How are your measurements made? Are they at the eardrum with a probe microphone, or in a simulated ear. If in a simulated ear, what is your insertion depth. Is is at the reference plane? If the earmolds fit deeper than the reference plane as many do, you need to insert your molds deeper into the coupler.
    Did you remove the damper from the end of the ER-4 and place it in the end of the earmold, or add another damper?
    I don't know why your stock curve looks abnormal. There shouldn't be a second peak @ 4kHz. Is the stock curve with an eartip?

    ReplyDelete
    Replies
    1. It is such an honor to have Mr. Don Wilson visiting my blog! Dr. Craig Kasper is affiliated with ACS, not ER. I was directed to him when I made a phone call to ER for inquiries on CUSTOM FIT.

      I use an IEC 60318-4(formerly 711) simulator, which is roughly equivalent to ER's Knowles DB-100 Zwislocki coupler in terms of accuracy. All of the measurements have been made right at the reference plane, according to the IEC standard. The Earmold can't go any deeper, since a metal mesh is blocking the way.

      Yes, I removed the damper from ER-4 and placed it at the end of the earmold first, and continuously made a 2 mm adjustment for each measurement. You should refer to the gray curve in the first graph for the stock ER-4P response, as other data are all of custom sleeves. The cyan graph is ER-4P stock, with CUSTOM FIT equipped.

      Meanwhile, please refer to here for the accuracy of my simulator: http://rinchoi.blogspot.com/2012/09/evaluating-accuracy-of-my-new-simulator.html

      Please feel free to let me know whatever's in your mind, as I am totally open to criticism and in-depth discussion.

      Delete
  3. Which of the IEC couplers are you using? We have a bunch of the GRAS couplers and the screen is 4-5 mm below the reference plane. That would not account for the response issue with your stock ER-4, nor would it be easy to simulate the actual in ear response of the earmolds if they fit deeper than the 12-13mm depth of the reference plane. I will have to think about why your stock response is not normal.
    You are correct that the IEC 60318-4 couplers are very close to the older ANSI db 100 couplers we use on this earphone. We plan to switch to a newer test system after it is fully documented and validated. The coupler starts looking like an acoustical short above 16kHz due to the diameter of the coupler/ average ear canal. Once the 7.5mm diameter is larger than the wavelength you start to have problems. That's why your Sennheiser IEM didn't work past 20kHz.

    The custom fit earmolds work well for most people who have comfort, or seal issues with eartips. You can find response issues with people who insert the eartips to different depths, use different eartips, or have different ear canals. The curves you posted with the dampers moved look pretty good. The best comparison would be against the diffuse field response, not strictly comparing it to the stock eartip response.

    ReplyDelete
    Replies
    1. That sounds like my 60318-4 coupler alright, LOL! (I believe mine is from ACO) I have total 3 units, with only one for IEM measurement. The other pair is implemented in my KEMAR equivalent dummyhead, EURI.

      Yes, compared to Zwislocki coupler, 711 coupler is missing two Helmholtz resonators- but I won't try to judge which simulator does better job in representing the "average" ear canal, as the goal of Mr. Rasmussen, the inventor of 711, was to come up with a simple robust design which can yield great repeatability, of which Zwislocki coupler ultimately lacked.

      The cyan "stock" curve is not actually of a stock; Rather, it indicates the response of CUSTOM-FIT measured with the damper situating at the tip of the nozzle("stock" damper location). Here's my ER-4P with a triple flange sleeve, red and green, superimposed on top of ER's calibration chart: http://2.bp.blogspot.com/-2aRGhrkFWOI/UEcwxur_fNI/AAAAAAAABZE/2AssjcrcFSA/s1600/image01.png

      Delete
  4. Since the Zwislocki is no longer in production, and the 711 is the ISO standard and produced by at least 3 companies we will eventually move all of our production equipment to them. The actual differences are very small. Your curve (in the link) shows more of a response difference than I thought there was just between the two couplers. I am glad you have such a nicely matched pair. I looked up the data from your serial numbers on our server.

    ReplyDelete
    Replies
    1. I talked to Dr.Zwislocki a few years ago, and he wasn't aware that Knowles discontinued DB-100 & sold the entire division to GRAS at all. He was slightly disappointed.. and so was I.

      Thanks to you and Etymotic Research, I was able to fine-calibrate my measurement setup nicely. I send my respect to Dr.Killion and especially you, Mr. Don Wilson for all the contribution you made within this snake-oil driven industry of in-ear headphones, and am deeply honored to have you on my blog. I think I sent you a LinkedIn invitation before! Would you kindly consider accepting my as one of your connections? :)

      Delete
    2. One of these days I will go through the hassle of resetting my password so can get back on Linkedin. I have rarely used it.

      I don't think Knowles sold many of the Zwislocki couplers, and after they got rid of all of the engineers who could make them, its fate was sealed. We considered taking it over at one time, but decided we might be the only user of them. We work closely with GRAS, and I think they are a good fit to take over the KEMAR, DB-100 line.

      Delete
    3. I'll be waiting till that day comes! It is really easy to reset the password, trust me haha.

      Thanks for sharing such a precious piece of history. With most of the headphone manufacturers are actually using occluded ear simulators and HATS these days, it would've been rather nice had Etymotic Research acquired the franchise. There is no 'what-if' in history, but the Zwislocki coupler and the relevant ANSI standard might have had a chance against the IEC-711.

      Indeed! GRAS is definitely a suitable contender as the new manufacturer of KEMAR. Dr.Rasmussen is such an important figure of the industry.. It's rather a disappointment though, that some solution providers consider B&K 4128C to be more robust than KEMAR when it comes down to quality control related manufacturing variance. (Or perhaps they recommend B&K as selling the brand yields more profit for them? I am not sure..)

      Delete
  5. I'm curious as to the difference between ACD and Microsonics custom tips? I had a problem with my first two tip's sound quality via Microsonics. Rick Carlson at Etymotic has been helping and he's been excellent. I just had my ears remolded and I'm waiting for my third set to arrive. I will try moving the filters to the end of the molds if necessary as Rick also recommended, but the first two were noticeably different than any of the stock tips.

    Should I expect a reasonably similar frequency response if everything is done well? Or is it expected that they will be different enough from stock as to be noticeable? I love the accuracy of the er4p, but seal is my problem, so I thought the cutstoms would help that. However, I "can" get a good seal with the stock tips, but at the cost of discomfort. So, I may sound crazy, but if the sound is noticeably different, then I'd rather stick with the discomfort... :-o I'll have to see how this last pair works out.

    ReplyDelete
    Replies
    1. First, measure the bore length of your sleeves in mm.
      Second, run a log sweep from 3 kHz to 10 kHz on each ear, and see if there's any distinguishable harshness present at certain frequency. If there is, then write down the frequency.
      Third, I may be able to calculate what's going on with your custom sleeves, from above numbers.

      Delete
    2. The point with Custom fit, is to get a consistent, and comfortable seal. There are some slight frequency response trade offs, but for the most part they are not enough to worry about. You can find big changes in the insertion depth, and sometimes even the eartip used. The early ER-4 foam eartips had a horn due to the resources I had available to design the insert.

      Delete
    3. Custom fit is definitely great for comfort & a good seal, and can be further improved acoustically, if the technicians at ACS become more aware about the basic principles of in-ear acoustics, IMHO. (of which will end up with a price increase LOL)

      An ear sleeve with a horn, that is exactly what Shure's olive foam sleeve is: http://www.amazon.com/dp/B000NDHIKY A pair of orange 3300 Ohm dampers + ER-4B + Olive foams attached does the job for me!

      Delete
    4. Thank you both for replying. Unfortunately, after two separate molding processes (one without bite block and one with) the process has had pretty bad results for me in every way. First, they aren't even that comfortable for me. The audiologist checked the fit and said they fit well, but they don't feel like they seat well. They seal, but don't block much noise. Probably about half the stock tips. I simply can't ignore them and focus on the music when they're in my ears for two reasons.

      First, the comfort just isn't there for some reason. Second, the frequency response is pretty far from stock for me. I didn't expect perfection, but I also didn't expect large spikes in the treble. 

The frequency response for my tips is much more noticeably different than any stock tip at any depth. Even using a stock tip not inserted deeply sounds better overall. The customs are very fatiguing and still sounds like there are spikes in the treble even after the third set of remakes.

      I get the impression nothing can be done. With each successive tip, the molds have been made deeper as well as seating the er4 nozzle deeper in the tip. This seems to have 'slightly' improved the response, but it is still pretty far from being neutral in any sense. :-/ 

The last, deepest tips are so deep, the channel they drilled into the tips is practically touching the wall of the tips, so I'm not sure they could go much deeper, or if that would even help. The filters placed in the tips act like a low pass filter, so that doesn't really help either.

      So, unfortunately, I've ended up with an uncomfortable, inaccurate, expensive set of tips. :-P 

Microsonics doesn't seem to keen on helping me either. My last phone conversation about my options wasn't productive, and they haven't called me in a week, despite telling me they'd call back the next day. Luckily, I've been getting more comfortable with the stock foam tips, and I prefer them most. So, now I just feel like I spent $150 on pretty silicon paper weights. But they aren't even heavy enough to hold down paper! haha ;-)

      I'm not sure if anything can be done, but I just have that sinking feeling in my stomach from the cost/value of the experience. :-/

      Delete
    5. :/ I am sorry to hear that.. The fit may be improved if you can find the exact downward angle of which your jaw is joined together. My jaw is tilted about 15° towards right, so I had to slightly open up my jaw accordingly when the mold was made.

      My best recommendation is to utilize the damper location adjustment suggested by Mr. Don Wilson here. Take out the green filters from your ER4, and place it inside of the bore close to the tip of the nozzle. This will certainly get rid of most of the harshness.

      Delete
    6. It gets rid of the harshness, but the accuracy is far from stock for me. And that doesn't make them any more comfortable. If I knew it was so far from stock I wouldn't have gone anywhere near the customs. I asked two people at etymotic beforehand, and no one would specifically say whether they were better or worse. They said they made them more comfortable, but anytime I asked about quality they said they might improve if the tips improve your seal, thereby giving you better bass, but they may not. No one ever mentioned the beautiful er4s treble would be ruined. :-/

      I think these things should be made more clear upfront. These seem like common issues with a basic web search or headfi search. Always the same thing, treble spikes, put the filters in the tips, etc. I realize etymotic doesn't want to deter people from buying them, and some people might actually see an improvement, so it's an optional bonus they offer. However, I had assumed at a minimum they would be close to stock sound. Obviously they wouldn't be identical, as they are a different material, different shape, etc. But I didn't expect such a change in treble. That's what makes the er4s the er4s! haha. the amazing treble. :-P They are nowhere near close enough to stock for me to ever use them. I prefer my pfe112 over the er4s with custom tips, and those have peaky treble... :-/

      Perhaps I'm just venting, but it wouldn't be so bad if they didn't cost so much. I suppose I feel like a just got out of a casino. ;) hahaha I gambled and lost everything. ha! At least I still have stock tips. :-P I want to make it clear though, I've had nothing but excellent service from etymotic. Rick and Maureen have been nothing but helpful and accommodating. And Don comments on blogs! haha I can't say the same for microsonics so far unfortunately. I just think the expectations should be realistic before someone goes into these things, that's all. My impression from advertising, conversations, etc. was that at a minimum they would be very close to stock sound and more comfortable. At best they would improve sound quality if you weren't originally getting a good seal.

      Delete
    7. Thanks for sharing your experience with us, Sean. Yup, it is indeed quite frustrating if the end result is quite different from what's been claimed. Unfortunately, that's as far as the current custom in-ear technology would go. All of custom IEM manufacturers suffer from such limitations, and I mean every single one of them.

      "I think these things should be made more clear upfront"
      That's exactly what I am trying to do here, so that there's no more of our hard-earned money wasted on some pseudo-scientific bullshit. Some companies may find my effort uneasy, but I don't give a damn.

      Delete
    8. Ha. Thanks Rin. That comment made me feel good after what just happened...

      I keep all of my er4s tips in a little baggy. Triple flange (both sizes), modded triple flange (both types) and of course my custom tips. I couldn't find them all day. I found them last night. They went through the washing machine in my shorts. Ugh. All of the tips are ruined except the customs. Surprisingly they seem perfectly fine. That does me no good though. ha. But now all I have are my worn out foam tips on the phones. :-/ Oh well. If anyone wants tips that hold up to a washing machine I highly recommend the microsonics custom tips. :-P

      Interestingly, the triple flange tips actually shrunk! They look like they have been compressed vertically into short fatter flanges. ha. Weird.

      Delete
    9. Just thought I'd check back in. No luck. Had my fourth remold. The sound has gotten better, but not great. They actually don't seal well or feel stable in my ear. I'm going to contact them yet again. They don't guarantee they're products, so I'm going to keep having them try different things in hopes i get one that works. Otherwise I have a $150 piece of silicon. :-P

      For what it's worth, etymotic was awesome and sent me all new tips for free. I didn't even ask them to. I asked if they sold a "kit" with all the stock tips because of what happened and they just sent them to me. :-P Woo!

      Delete
    10. I meant they made a fourth set of tips for me...

      Delete
    11. I've given up on them long time ago. :/ Hopefully they could get at least one pair right..

      Delete
    12. You never responded to my head-fi message. :-P can you message me your email on head-fi or something. I have a few questions and things to discuss if you don't mind...

      Delete
    13. Oh, my bad. Lemme check it out!

      Delete
  6. So you recommend EQ-correcting the peaks with the dampers in the stock position rather than moved out in to the molds? BTW regarding the filter I made from your graphs, I realize I overlooked that you measured your molds at the reference plane, mine don't reach that far. I'm getting a resonance at around 9k. I went ahead and generated another set of filters from your previous HF5 and ER4B insertion depth graphs, the HF5 6mm correction + damper correction almost drives it home. What's your advice?

    ReplyDelete
    Replies
    1. Is it at 9 kHz without the damper adjustment? That is definitely better than mine, but yeah falling little short from perfect for sure. Still, even if insertion depth is little off, that damper location adjustment should compensate the ill effect of it. Your approach should be sound.

      Delete
    2. Nevermind, these molds are going back to the audiologist for an unrelated issue.

      Delete
  7. For earphone less sensitive to insertion depth issues, let's say, UE900. Custom ear sleeves should not be too big a problem, right?

    ReplyDelete
    Replies
    1. All IEMs are prone to acoustic interaction caused by bore acoustics. So the answer is most likely no.

      However, one well-vented dynamic IEM is surprisingly inapt to it: LG Quadbeat. Its acoustic output impedance is superbly low, LOL

      Delete
  8. Is there a way to private message you rin?

    ReplyDelete
    Replies
    1. PM me on Head-Fi.org. My screen name is udauda.

      Delete
    2. pm me on head-fi, my name is luisdent

      Delete
  9. I'm on my fifth attempt soon working with the owner of ACS now. She is going to look at my molds and see what the remaining options are. Is there anything you think I should ask them or recommend? They've already lengthened the tips to go further in my canal and tried to seat the er4s deeper in the tips as well. They still stick out of my ears a lot though. Do you think this is a technical impossibility to achieve near ety quality with customs? Or should I ask for anything specific to be done?

    ReplyDelete