Head-Fi.org › Forums › Head-Fi Special Forums › Head-Fi Bloggers › Jude › Headphone Measurements: Different Setups, Different Results
New Posts  All Forums:Forum Nav:

Headphone Measurements: Different Setups, Different Results

post #1 of 22
Thread Starter 

This is part of a post I made some time ago in another thread, but I thought I'd re-post it here. Take a look at some of the headphone measurement systems (below) that you've been checking out graphs from.

 

There's a lot of discussion about headphone measurements, and I think it's important, when looking at graphs/plots, to understand that there are many different headphone measurement systems and methods, and you're likely to get as much variation in measurements as there are different measurement rigs and people doing the measuring.

 

This is a topic I'll be discussing more and more in the coming months.

 


 

Quote:
Originally Posted by jerg View Post
 

You can't expect to derive anything out of comparing FR of two headphones from two separate measurement archives (different equipment/compensations)...

Quote:
Originally Posted by catscratch View Post

Purrin and Tyll use totally different measurement techniques and apply totally different compensation curves. It is absolutely pointless to cross-compare them...

 

This is an oft-overlooked discussion, the comparison of headphone measurements made by different people with different setups. As catscratch said, it can make measurements made with different systems rather challenging to compare.
 



For example, Tyll Hertsens (Editor-In-Chief of Innerfidelity), as he describes here, uses a measurement head (the Head Acoustics HMS II.3). It is one of the industry standard tools for measurement. From that link above, you can read about Tyll's complete measurement system, including the acoustic isolation chamber he custom-built, the Audio Precision System 2 Cascade audio tester, and other components of his setup (and his methodology). Here are photos of his setup that he posted in that article (click the images below to see them in full size):

 

 

 


 

In one of the more recent headphone studies presented at AES by Sean Olive (Director Acoustic Research at Harman International and President-Elect at AES) and Todd Welti (Research Acoustician at Harman International), measurements were made using a G.R.A.S. 43AG Ear and Cheek Simulator equipped with an ITU-T type 3.3 pinna. (More specific details can be found in the full paper--Convention Paper 8744*--on the AES website.) Again, the G.R.A.S. piece and pinna is another industry standard measurement setup. Here are photos of the G.R.A.S., both without and with a ear/cheek affixed (click the images below to see them in full size):

 

 

 

Keith Howard in Stereophile also mentions that he uses a G.R.A.S. ear and cheek simulator; Brent Butterworth of Sound And Vision also uses a G.R.A.S. device.

 

I've been to several headphone manufacturer facilities, and have seen gear similar to the above-mentioned equipment in them (heads, head-and-torsos, and devices like the G.R.A.S. devices).

 


 

In their AES Convention Paper 8994* (Listener Preferences for In-Room Loudspeaker and Headphone Target Responses), Sean Olive, Todd Welti, and Elisabeth McMullin used the G.R.A.S. 45CA. for their headphone measurements.

 

In AES Convention Paper 9118* (The Correlation Between Distortion Audibility and Listener Preference in Headphones), Steve Temme, Sean Olive, Steve Tatarunis, Todd Welti, and Elisabeth McMullin use the G.R.A.S. 45CA, as well as the Bruel & Kjaer Head and Torso simulator Type 4128.

 

Here are some photos of the G.R.A.S. 45CA:

 

(Photo from this page on rinchoi.blogspot.com.)

 

 

 

The G.R.A.S. 45CA is, in a way, like having two G.R.A.S. 43AG Ear and Cheek Simulators combined, so both channels of a stereo headphone can be measured at once. One key advantage of the 45CA (versus the 43AG, or even a pair of them) is that the 45CA's "head" housing has median adult human head height and width (with cheek angles), so headphone clamping force does not need to be measured separately and then dialed into the clamp. (The single-sided G.R.A.S. 43AG's clamp mechanism can also present some challenges with open-back earcups, because it clamps down on the center of the back of the headphone earcup.)

 

One key disadvantage of the 45CA versus the 43AG is that the 45CA is not nearly as portable. It is much larger, and much heavier than the 43AG.

 


 

On the more homegrown fronts, purrin and Rin Choi have created their own setups.

 

purrin has elected not to use a dummy head or, to the best of my knowledge, a pinna. I'm not sure that his measurement system was ever discussed in great detail. When asked about his measurement system on Head-Fi, purrin called his coupler "semi-secret." (link)  During a discussion at the Newport Beach audio show last year, when asked (on video) which coupler he uses (link), purrin (Marvin Chen on the video) responded: "I'm actually not using a dummy head. The reason is I actually did try a dummy head and there's some issues with that. With dummy heads you've got the ears, and the really good ones have got the tubes in them, and those create some specific resonances, that...don't necessarily represent what we hear. If I were to do a CSD measurement with one of those couplers, there would be resonances that are inherent to the human auditory system which I feel that the brain filters out. So what I actually did was I built a custom coupler. Essentially, what it does...it tries to minimize the effects of the ear, the ear canal, while at the same time trying to be realistic of what we actually hear."

 

Here are photos of purrin's custom coupler setup (click the images below to see them in full size):

 

   

 

2015-04-17 UPDATE: Marv (purrin) has updated his measurement rig to a second version. Here are photos (click photos to see larger versions):

 

  

 

He has described its basic structure as foam, plastic plate with hole (using a CD), and Creatology foam with hole cut for microphone. Here is Marv's description, in greater detail:
 

Quote:

Originally Posted by purrin
 

  1. In the middle of the foam later, a piece of felt about the size of an ear. Without this piece, I found measurements too ringy. Covered completely with felt, I found measurements too damped. The felt the size of an ear also serves another function. Proper measurement of supra-aurals that rest of the ear.
  2. I use a strip of felt to simulate hair and sometimes another strip to simulate imperfect seal of the nape of the neck. Knowing to get a properly simulated seal is crucial.
  3. The hard plastic plate (use a CD) is a good bone simulator.
     

I filled the inside with medium density open cell foam - to simulate brains. Use something squishy to avoid interactions between L and R.

 

Really, you could just put the felt / foam layer on top of a CD on top of a foam brick. I just made it look nicer and named it C.U.N.T. to establish "cred". We all know that stuff that looks nice, that is shiny white, and has a euro kind of acronym like G.R.A.S. or C.U.N.T. must be good.

​​​


 

Rin Choi (who goes by udauda on Head-Fi) created his own dummy head that he named EURI (photos below). It is covered in a latex skin to "effectively simulate the damping effect of a human skin & insulate any acoustic leakage between the pinna simulators & the head itself." Inside of EURI can be found IEC 60318-4 occluded ear simulators. On his website, it seems to me that Rin is shooting for using industry standards (in terms of methods and gear), and, with his own measurements, compares his EURI head to commercial alternatives.

 

Here are photos of Rin's EURI head and the 60318-4 occluded ear simulator (click the images below to see them in full size):

 

 

 


 

Again, the measurement setups and methodologies can differ substantially from measurer to measurer. Some use more industry standard equipment and methodologies; still others claim to sidestep industry standards for individual reasons and/or claims of more accurate measurement results. So, as was stated earlier, it can be tough to make meaningful comparisons of measurements between the results of the different providers of them.

 

* The direct-download links for the AES papers require AES membership with AES E-Library paid access.

 

2015-04-17 1211 EDT: Updated with v2 of purrin's measurement rig.

2015-04-24 0910 EDT: Updated with G.R.A.S. 45CA.

post #2 of 22

Difficult subject...Looking forward to your thoughts, Jude.

post #3 of 22

Why not use a head made out of ballistic gel? Density and damping-wise, it should be similar enough to human tissue. 

post #4 of 22

I would think that, with wear and tear, such a thing would fall apart fairly rapidly... require periodic re-investment at an untenable rate.


Home of the Liquid Carbon, Liquid Crimson, Liquid Glass, Liquid Gold and
Liquid Lightning headphone amplifiers... and the upcoming Liquid Spark!

Reply
post #5 of 22
Quote:
Originally Posted by Voidrunner View Post
 

Why not use a head made out of ballistic gel? Density and damping-wise, it should be similar enough to human tissue. 


The head would be comparatively cheap to make/discard once you have a good mould.  Lots of ballistic gelatine blocks are being made for ammo tests.  But the weakness is that the average density will correspond to tissue, but it's very homogeneously distributed and that could be a weak point, considering shock (sound) wave behaviour at interfaces of different densities, where you have various reflections etc.  In the end I guess the main thing is to pick one single method, use it consistently and only compare measurement results with others made in exactly the same fashion.

post #6 of 22

Yup, several of the current compounds can be re-melted and re-cast almost indefinitely. If you wanted to get really serious, you could include the skull. That would stand in well for a lot of those supposed tissue density interfaces. 

post #7 of 22

The people who make sorbothane made a sorb head for some of their work. Personally, I prefer to use sorb for damping the vibrations in headphone earcups.  See you at Canjam! 


Edited by edstrelow - 4/18/15 at 5:00pm
post #8 of 22

I use my own head FTW hahaha. nothing fancy like a acoustic isolation chambers because I'm not listening in a fancy room. it does lead to a 7kHz dip on all my measurements that I haven't found a way to compensate for, but I only use my measurements in a relativistic way compared to my other measurements, so it should be alright.

 

good post though and great reminder that it is difficult to compare measurements against completely different set-ups. always compare headphone measurements from the same source for best results I think.

post #9 of 22

Don't think it matters much as long as consistent in process. Obviously reviewers try to compensate (what they perceive) a flat sounding earphone to measure flat but there's a lot to consider and insertion depth of IEMs will vary by individual which will affect resonance and reinforcement.

 

I'm kind of a techie and appreciate all the efforts but there's nothing like a listen. :ksc75smile: There's also a lot more to sound than flat response and low distortion, even when adding overall phase response. I have some multi driver customs I love but I've listened to some others with low distortion and reasonably flat response that I find less musical than other single driver models with less bandwidth simply because the multis aren't as solid or coherent. Hard to know if everything is playing together on a spec sheet pic without measuring individual driver's time domain at overlaps.:smile: Don't get me wrong, I love the spec sheets and most devices are single driver anyway but I more appreciate the personal impressions of a reviewer I trust. :beerchug: 


Edited by goodvibes - 7/31/15 at 6:12am
post #10 of 22

Interesting subject, subscribed.

post #11 of 22

it's an interesting subject, but I'm not sure many people understand enough(me included for some stuff) to even fathom why 2 different sources fail to show the exact same measurements. in fact I still see many people getting fooled just by the unit scales, compensation curve applied or smoothing quantity involved. that's how uninformed people who think they can read a measurement often are. and that's not even scratching the surface, as yet no acoustic problems have been involved.

 

what I would love to learn about is how each person calibrated his gear(well I expect pro gear to be calibrated, but after some time what happens? are there some reference sound devices to use in a loop and recalibrate? do they need to send the stuff back to the manufacturer?). doing a pale imitation of that myself with cheap stuff on what's not even a proper desk, my only mean is to try and copy others results, but as you mentioned, they do all have at least some differences (mostly different resonance points, but it doesn't seem to stop there), so who should be the "better" model? does it really matter when it's so hard to get 2 drivers to actually output the same sound?

I found myself learning a lot in the process and can't recommend it enough to people who have a meaningful quantity of headphones/IEMs(if you have 2 pairs, getting involved in measurements is a waste of time). but I also became a lot more suspicious, not of the objectivity of measurements, but about how far we can go when interpreting them in relation to the same type of headphone we have at home that might not even sound really the same as the one measured. how my ear is not the measurement capsule, or how the averaging might show a curve that isn't really a good representation of a properly used headphone. or vice versa, an ideal measurement might not be a good representation of most people's way of using it. it's problems inside problems inside problems, and it seems people come to their own conclusions and ways to answer those questions, bringing even more reasons for differences in the end results. 

 

now, those are problems we see everywhere, not just in audio or just in the headphone world of course, but headphones seem to survive in a temporal crack in time where progress and science can't easily get through somehow. where you can still read fools telling people to be wary of science as if it was witchcraft, making a reenactment of the monty python's witch vs duck scene, except they really believe it, and instead of everybody laughing, a good deal of people seriously support the nonsense ideas.  that obviously cannot help promoting and making standards for objective measurements.

post #12 of 22

All these tools are very useful for technical engineers who designs audio stuff for the mass market.:deadhorse:


You probably can access to a better performance if you design your headphones with your own high definition audiograms
My CIEMS are far better now... concepts and brilliant technical analysis can't compete with real life measurements IMHO.

 


Edited by Silverprout - 10/1/15 at 10:35am
post #13 of 22

Some factors I think.......

Different signal source has different output impedance which will lead to get a different "look" of frequency response, especially IEM with BA driver.

Different setup with different couplers...   head and torso (dummy head), IEC318 coupler, IEC711 coupler. What should use when measuring IEM, earbud, hp? There's no standard.

Further more, the placement of hp onto the HATS, the shape of fixture to be used with the IEC318 coupler will affect the look of FR as well. IEM may be better as it usually measure with IEC711 coupler, just to make sure there's no air leakage.

Another factor is whether the DF compensation should be applied or not.

 

Just my 2 cents......:L3000: 

post #14 of 22
Quote:
Originally Posted by Music818 View Post
 

Some factors I think.......

Different signal source has different output impedance which will lead to get a different "look" of frequency response, especially IEM with BA driver.

Different setup with different couplers...   head and torso (dummy head), IEC318 coupler, IEC711 coupler. What should use when measuring IEM, earbud, hp? There's no standard.

Further more, the placement of hp onto the HATS, the shape of fixture to be used with the IEC318 coupler will affect the look of FR as well. IEM may be better as it usually measure with IEC711 coupler, just to make sure there's no air leakage.

Another factor is whether the DF compensation should be applied or not.

 

Just my 2 cents......:L3000: 


when I first looked up some measurement standards, I remember reading one where the source had to be something like 120ohm(I don't remember the exact number but it was pretty crazy when thinking about modern portable gears).

I wouldn't imagine anybody doing that when measuring IEMs. but yes different amps as source might be responsible for at least some of the variations.

post #15 of 22
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Jude
Head-Fi.org › Forums › Head-Fi Special Forums › Head-Fi Bloggers › Jude › Headphone Measurements: Different Setups, Different Results