Headphones Frequency Response: Challenges & Solutions

Discussion in 'Measurement Techniques Discussion' started by samvafaei, Jun 19, 2017.

  1. TomHP

    TomHP Facebook Friend

    Joined:
    Nov 24, 2016
    Likes Received:
    153
    Trophy Points:
    43
    Location:
    Taiwan
    lol, that's a ridiculous statement, but nevertheless...

    (1) influence of position on flat surface coupler with 711 coupler and ear (GRAS 45AG).
    3 different measurements, in 3 different locations
    The seal was lost on 2/3 measurements, hence the loss in bass response, but that doesn't influence the area of interest.

    [​IMG]

    Can easily see that the big dip at ~3kHz resulting from a standing wave in the longitudinal axis of the cup is measured differently depending on the positioning on the flat coupler.

    (2) http://www.aes.org/e-lib/browse.cfm?elib=17698 --> paper from Harman on multi-mic headphone measurements
     
  2. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    I understand that moving the headphone around a flat coupler will yield different results. Specially if that flat coupler is a GRAS 45AG and has an ear on it.

    However, how about a coupler with no ear and a little bit of sound absorption material that is capable of maintaining seal? So far I have seen a lot less positional variation and perhaps less issues with standing waves.

    I don't think >this rig used by Philips (scroll down to the 8th photograph)< have sound absorption material, but it is claimed to have much less positional variation relative to something like a GRAS 45AG.

    I have used something similar with different materials, and can say the results are indeed somewhat positional invariant. Though taking the headphone to the edge can make a difference depending on headphone.
     
  3. TomHP

    TomHP Facebook Friend

    Joined:
    Nov 24, 2016
    Likes Received:
    153
    Trophy Points:
    43
    Location:
    Taiwan
    haha, funny you quote the Philips rig. I used to work in that exact lab. :) The rig is indeed less positioning dependent than the HATS used in the same lab. But if a headphone has a severe cup resonance problem (like the one I showed, an extreme example admittedly), a flat jig will not solve the issue. That jig in particular is just a wooden jig with no additional damping.

    EDIT: spelling
     
  4. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    Yes. It's not perfect.

    I brought my POS rig to a San Diego meet, and I still remember MrSpeakers asking me to move his cans around to see if any of the measurements would make his cans look good. It only got worse. The cans depended on seal (full size closed) and the pads were semi-stiff. Still comfortable though.

    The pads on those cans tended to deform and the headband was too loose, which was a problem considering that they seemed heavily seal dependent. Seal might have been inconsistent across products of the same model. Possible he fixed that on rev 682.28841 of his cans.

    That said, I don't see as many issues with open headphones, and if the cans are somewhat centered around the mic.

    What other approach would you recommend to reduce standing waves and position dependence?
     
  5. TomHP

    TomHP Facebook Friend

    Joined:
    Nov 24, 2016
    Likes Received:
    153
    Trophy Points:
    43
    Location:
    Taiwan
    Well, from an R&D point of view you don't really want to reduce those standing waves by adding damping. You want them to show up on your rig. E.g. that's how I found it was a standing wave problem and not a driver break-up mode in the example I posted.

    To get a better idea of how things will "sound", averaging using measurements made in different locations makes sense.

    Your point about open cans is kind off logical though. If the pads have acoustically semi-transparent wrapping (e.g. velour or other fabrics - HD6xx for example or others) a huge amount of damping is added to the front cup, particularly on the edges where the pressure maxima would be for standing waves.
     
  6. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    From an R&D point of view I agree it makes sense if one wants to narrow down the problem to the cup or the driver. But would that same logic apply when you are no longer troubleshooting, and doing an overall characterization?

    What approach do you feel would yield results most representative of what we actually hear?

    Averaging from different locations might make sense, but note that in some cases, if there is a standing wave problem moving the cans around may just result in a frequency shift of the null. And the question would still remain about whether or not we perceive such a null. I think we do based on tone sweeps I've done in the past with headphones exhibiting this behavior.

    I think the wavelength of a 10 kHz sound signal is like 1.3 or so inches. At what frequency does the volume starts/stops behaving like a pressure chamber?

    From what I'm understanding of your comment, the pressure field inside the cup is not uniform for high frequency sound. If you were to put a mic inside, which location would yield the SPL that is likely the most representative of the sound we actually hear?
     
  7. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    BTW @TomHP I'm not trying to put anybody on the spot or anything. I understand you work on this area, so you likely know some of this stuff better than the average dude. If there is something that you feel is proprietary and not at liberty to share, I can understand.

    This is somewhat also a site to kick back and relax. Sort of hobby like. No need to make it a second job either.
     
  8. TomHP

    TomHP Facebook Friend

    Joined:
    Nov 24, 2016
    Likes Received:
    153
    Trophy Points:
    43
    Location:
    Taiwan
    I've seen people mention on-head measurements before. It can certainly help in cases where bad fit on a HATS is just not workable, but only up to a certain frequency point. Otherwise, something that includes an ear + earcanal (e.g. B&K HATS or GRAS KEMAR or other GRAS jigs) are the best tool available to us right now. There is no ideal and one of the things that takes time and experience in this field is knowing your tools and how to work around the imperfections. So far from what I've seen in the industry is that companies typically will have a target defined based on a particular measurement jig but will always strongly rely on additional subjective testing (thank god!).

    Yup, you can see the null move around in my measurement as well. In general, people are more perceptible to peaks than nulls (I forgot who/where this was researched). So if you're applying additional EQ on a set of headphones, you're better off trying to correct unwanted peaks.

    Depends on the cup size and driver diameter, there's no easy definition, at least not that I'm aware off. In literature (Poldy) and from experience you'll see simulation of headphones with the pressure chamber assumption start to fall apart around 2-3 kHz. Insert earphones on the other hand can be modeled relatively easily up to much higher frequencies. I've had models match the 711 measurement to within 1-2 dB up to the couplers working point (~10 kHz).

    Again, something that has an ear, a canal and a microphone at the end. Since you don't know exactly how people will wear the headphones, spatial averaging helps. Of course all HATS are supposed to represent some statistical average human head/ear, so it may not match your response that well, but it's best we have so far.
     
  9. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    Makes sense.

    @Marvey used to say this all the time. Something about sins of commission and omission or something like that.

    Makes sense.

    I'm more familiar with flat coupler + absorption material or the likes. Though I have used a fake ear too. Others use their own heads. And others maybe a sex doll. Seems like an art.
     
  10. TomHP

    TomHP Facebook Friend

    Joined:
    Nov 24, 2016
    Likes Received:
    153
    Trophy Points:
    43
    Location:
    Taiwan
    That's alright man, but thanks for the heads up. I can't and won't disclose anything that could come back and bite me in the ass :D

    I'm always surprised to see how deep some people in this hobby get into this without having it as a day job. I've picked up very useful bits of info from this and other online sources actually. The only thing I find annoying to read sometimes is the statement that "measurements are overrated". It's such bullshit. Without measurements you can't achieve excellence. Whoever says otherwise is just not well versed enough in the engineering involved in headphones.
     
  11. samvafaei

    samvafaei New

    Joined:
    Jul 15, 2016
    Likes Received:
    35
    Trophy Points:
    13
    @Serious

    I double-checked my human measurements from the first page. They are correct. They have a bit more sub-bass below 40Hz compared to what we have on the website, but I did those measurements very quickly for you guys, so I may have them on a bit too tight or something. But the relative difference among human subjects in the treble remains.
     
  12. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    @samvafaei, have you done any measurements of flat speakers using your HATS on anechoic chamber conditions (or semy-anechoic)? I know some dudes have and perhaps you could use that for compensation. Or at least compare that to the compensations you have at hand. Just a suggestion that came to mind.

    If you have, it might be interesting to see how much that compensation deviates from other compensations.
     
  13. samvafaei

    samvafaei New

    Joined:
    Jul 15, 2016
    Likes Received:
    35
    Trophy Points:
    13
    One of the things I didn't include in the video was that I did try to make my own compensation curve with a pair of speakers in a stereo setup (30 degrees), in a simulated anechoic environment. I did the test twice, with two different speakers and in two different rooms. They both gave very similar results, but the diffuse field treble still sounded better to us. So either I did the test wrong, or that diffuse is actually better for treble.

    For bass this test wouldn't work, since it was a simulated anechoic environment. Regardless the response of the HMS below 500Hz is basically identical to a measurement mic.
     
  14. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    Too bright? Too veiled? Too boring? Too lively? Just weird?
     
  15. samvafaei

    samvafaei New

    Joined:
    Jul 15, 2016
    Likes Received:
    35
    Trophy Points:
    13
    I'd say not too dissimilar to Tyll's measurements at Harman. Also, this is the only graph I could find at the moment, and I can't remember if this was a test run or the actual one. The fact that I took this picture with my phone tells me that this was not the final run. So don't quote me on this, just posting it for your information.

    [​IMG]
     
    Last edited: Jun 26, 2017
  16. Bill-P

    Bill-P Level 42 Mad Wizard

    Pyrate
    Joined:
    Sep 25, 2015
    Likes Received:
    4,801
    Trophy Points:
    113
    Location:
    San Jose, CA
    Measurements are overrated! :p

    No, but seriously, I would say that measurements should be taken with a very healthy grain of salt and extra precautions like one must understand the measurement system involved, as well as the preferences and habits and experience of the person taking said measurement.

    Excellence "can" in fact be achieved with regards to headphones without measurements. It's pretty easy, actually: just make a headphone that subjectively, everybody finds pleasing, without ever looking at its measurements. The funny thing is: when such a thing is achieved, it typically measures well, but the inverse is not true. (Good headphones will measure quite well, but headphones that measure well aren't necessarily good sounding)

    Yes, without measurements to see problems and stuffs, it's hard to tweak and tune it toward the "perfect" target response, but both you and I know there is no such thing as a "perfect target response" yet. There is only "an average of what some think is right." And relying too much on measurements has proven not to be a good idea sometimes, too (look at some of Mr. Speakers' earlier prototypes, look at Sonoma Acoustics' overuse of DSP in their new headphone).

    So with that said, I do actually believe measurements are overrated. Until someone can prove to me that a headphone that measures well almost always sounds good.
     
  17. TomHP

    TomHP Facebook Friend

    Joined:
    Nov 24, 2016
    Likes Received:
    153
    Trophy Points:
    43
    Location:
    Taiwan
    I said similar things. All about knowing the limits of the tool and gaining experience in using the tools at hand.

    I would be happy to hear about headphones that are indeed made (1) without any measurements involved and (2) that subjectively everybody finds pleasing. I cannot imagine how cumbersome and long that trial and error process must have been. Not to forget that quality control on the manufacturing line must have been non existent, or are you saying that can be done without measurements as well?


    My statements are not directed at any particular target response and I agree that there is no "perfect" target response because we are simply not measuring the headphones correctly now for there to BE a correct target response. Regardless... using a target derived from research I still find a better approach than aimlessly tuning away at a headphone. Also, I'm not talking about some niche part of the industry, I'm talking about headphones designed to actually sell in any meaningful volumes.
     
  18. Bill-P

    Bill-P Level 42 Mad Wizard

    Pyrate
    Joined:
    Sep 25, 2015
    Likes Received:
    4,801
    Trophy Points:
    113
    Location:
    San Jose, CA
    Agreed.

    1. A lot of people modify and make tuning changes to their headphones these days. I'm sure you will find that not many of them have a measurement system to verify the results objectively, but that of course some of them have achieved amazing stuffs. This is beside the usual suspects here (me, Luis, Alex, etc...) of course. Trial and error does exist, and while I agree that it's a tedious process, it's not impossible. I think all that is necessary here is for a person to have good taste and good ears (and probably some amount of perseverance), and I do firmly believe such individuals exist. (and no, I'm not one of those folks, but I can name a few others who are ;) )

    2. Manufacturing is beyond the point (I kind of skirted around the idea and avoided mentioning the word), but just in case, I'm sure I can point out one or two headphones made in Japan in "good volume" that measure like crap. Disregarding their manufacturing variations, of course. In fact, sometimes there isn't even a clear "target" to those headphones (and yeah, I'm subtly hinting at Audio Technica here), and not only do they measure wildly different, they also sound wildly different. This is a real thing. And yet it's pretty clear they have decent quality control. Am I saying that manufacturing can be done without measurements as well? Yes! I would argue that that's how certain cheap earbuds and headphones are created. They sell in large volumes and at such low a price point that I suspect adding in measurements (of any kind, meaning "tighter quality control" in this sense) will cause an exponential rise in price for those things.

    Oh no, I agree that research is very necessary. However, I think I have seen (and done) enough measurements up to this point to realize that:

    1. Not all of the researched methods of measurements work well for everything. There are things I don't believe anybody has figured out how to measure yet. I may be wrong but I have yet to see any measurement that can sufficiently point out to me what "soundstage" even looks like.

    2. Ideally, having a target for tuning is indeed better than nothing, I agree, but... from my perspective, there still exist a fair number of things on the market that clearly do not have a tuning target in mind. I'll even stop picking on Audio Technica now to bring in... oBravo. Those who have heard their full-size headphone(s) can probably immediately recognize what I'm talking about. :bow: (@Marvey ?)

    3. Ideally, I'd hope that all manufacturers and companies interested in doing audio products should start measuring their headphones more, and taking into account more than just this "house sound" that every single one of them seems to have (again, with the exception of a few that are all over the map). But I don't think we're quite "there" yet in 2017. Basically, I believe what you're saying is probably true but not of all manufacturers in the world right now.

    It is precisely because some manufacturers still don't quite "get it" that I think individual measurements have arisen in order to point out the flaws that got ignored during the engineering phase. It is also precisely because measurement as an art in itself is incomplete that we are discussing this right now, too. Perhaps at some point in the future, this may change, but I don't think we're at that point in time yet.
     
  19. TomHP

    TomHP Facebook Friend

    Joined:
    Nov 24, 2016
    Likes Received:
    153
    Trophy Points:
    43
    Location:
    Taiwan
    Trial and error is fine for modders, but in a tightly time controlled product development, there us no way that kind of approach would fly.

    Why would manufacturing be beside the point? If you don't include it in the discussion, than we are purely talking about modding exercises.

    Measuring like crap does not mean you can't have controlled tolerances during MP. You typically look at relative measurements: DUT FR - Golden Sample FR = +/- 3dB tolerances. The FR can be totally ridiculous, but you would be able to deliver that crappy performance consistently :D

    I've seen and been involved in manufacturing of earphones/headphones from $5 to $500. No testing of FR on individual units is very very rare indeed. The cost is minimal of having a setup like that on your line vs. total project cost.

    I've said the same. We can't measure everything yet, but that doesn't mean we shouldn't measure at all...

    Of course there's a lot of crap on the market. I saw the oBravo in-ear measurements and laughed.

    Yes. Don't forget that BT headphones have overtaken wired headphones in value last year. BT headphones (in most cases) give you EQ, which will result in tighter convergence to a "house sound" for the brands that care.


    Passive headphones are always riddled with compromises. If you actually look at the maths that govern the FR of headphones (closed ones in particular), the problem is of such high order that to understand all parameters and their influence on the FR is simply not fully understood or modeled yet. The brands that put the R&D effort in (Sennheiser, AKG, Philips, Bose, ...) will always attempt to get the best performance for the price as possible, but there are always compromises and unknown FR issues that cannot be solved.
     
  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
    It's not a ridiculous statement in that we need specifics. We already know that coupler with an ear will show nulls and dips depending upon position of measurement. (A flat plate coupler without an ear is actually quite immune to this effect).

    My concerns are "random" placement of headphones on the coupler, in positions that would not reflect actual use by humans. For example, I might see people wear headphones low and forward, but never high up (this is uncomfortable). I know that I wear headphones in a certain way, probably adjusted within 0.5mm every time.

    Also, I'd like to see data on different types of headphones and measurements linked to diagrams of specific positions. Overlaying a bunch of plots on top of each other doesn't tell me much, other than different positions will change peaks and nulls, especially from 2-3kHz up where wavelengths are shorter. For example, what happens to HD800 when it's worn on centered? Then happens when the HD800 is 2mm shifted forward and low, which I do myself, or see at meets. Same thing with HD600. And SUSVARA, etc.
     
    Last edited: Jun 27, 2017

Share This Page