InnerFidelity June 2015 Update

Headphone Measurements
Wow! A pretty good haul this month. Averaged more than one headphone measurement a day! Big thanks to all who have sent headphones in. I apologize in advance for not mentioning people's names—some may wish to remain anonymous, but the main reason I'm not going to mention names is that I've not done a good job keeping the names related to the cans measured and I've not been able to find the names of a few folks. Rather than leaving someone out randomly, I'm just not going to do it at all. But thank you all who've sent cans in, it's very much appreciated.

If you'd like to participate with the headphone measurement program by sending in cans that aren't already measured and displayed on the measurement datasheets page, simply go to this forum page, read and follow the instructions, and we'll get it done.

As usual I've updated the AllGraphs.pdf of course, and added the cans to the measurement datasheets page. I've also linked to the measurement sheets to the headphone name in paragraphs below. New additions include:

HiFiMAN HE-1000 Pre-Production - First I have to say I do enjoy the sound of these headphones—they're fairly neutral, without any harshness, and very inoffensive. But they're also strangely soft in their attack. Not sure where that shows up on the measurements, but I will point out the strangely bumpy frequency response between 200Hz and 2kHz—quite unusual. Also, the somewhat high distortion at 100dBspl, and corresponding (though small and difficult to see) bumps in the impedance response. I'll investigate further when I do the review—and it's worth noting these are one of the pre-production units that are going around right now.

Philips Fidelio M2L - This is the worlds first headphone to run off the Apple Lightning connector. Which is cool and all, but it means I don't have a way to do my normal measurements. So, I put some square waves and pink noise on my iPad and did some measurements. This is a good sounding little headphone with a bit too much bass. Personally, I kind of like it for movie watching, but it's a little heavy handed for music.

Grado SR325e - First of the "e" series measurements. Looks like it's quite a bit more efficient than the 325i or plain 325. Frequency response looks quite similar to the 325i, but distortion seems to have fallen quite a bit. I'd say it's a change for the better.

Audio Zenith PMx2 - Heard this headphone at Newport last week. I quite liked the sound—definitely seemed quite a bit quicker than the PM1 or PM2 although it doesn't measure that much different. Alex will be sending a review unit along soon.

Koss Pro4AA 1975 - My very first headphone. I clearly remember listening to "Dark Side of the Moon" with them and my Fisher receiver and Radio Shack 8-Track deck. Obviously a well matched system. (Ha!) What's interesting to me is the 1975 version, while not very good, is better than the currently manufactured Pro4AA.

Noontec Hammo S - Lots of interest in this cheap headphone, and one look at the measurements and you'll see why—quite like the Harman target curve but for the big notch between 3kHz and 10kHz. I did remember the request to try measuring with the inner fabric padding removed. I like the headphone well enough to review it, and well enough to do a little video of how to put the pads back's almost impossible if you don't know the trick.

oBravo HAMT1 - I've been checking out the oBravo cans every time I see them at a shows for the past couple of years. They've always sounded uneven to me, but it also seemed they'd gotten better over time. Evidently they've still got a way to go.

Soul Headphones - Out of the blue, Soul contacted me to see if I might have an interest in reviewing their cans. I, of course, said I'm interested in reviewing anything that sounds good. They sent their Combat and Jet Pro (ANC on and ANC off) models. I guess I shouldn't be so surprised when people don't understand what "sounds good" means.

Audeze LCD-1 - What a treat to have the chance to hear and measure this rather rare headphone. And a bigger treat yet to find it actually does quite well—a little anemic down low, but other than that quite good.

150608_Update_Photo_YamahaYH5MYamaha YH-5M - Okay, just have to show you a picture of this thing. The YH-5M is a planar magnetic in-ear headphone...yes, you heard that right. The ear pieces are fairly small and they're mounted to the headband through a swiveling mechanism that allows you a great deal of freedom to adjust the drivers so that the ear tips are properly positioned at the entrance of the ear canal. Measurements are just "meh", but these sure are interesting gadgets.

And without further comment, the remainder are: AKG K240DF; AKG K240 Monitor; AKG K702; AKG K702 65th Anniversary Edition; Aurex HR-V9; Beyerdynamic DT 100 2x2kOhm; Fostex T50; Koss Pro4AAA; MyST IzoPhones 60; Ocharaku Flat4-Kuro Type II; Oppo PM2 DIY Modified by hans030390; Oppo PM3 Sample C; Sansui SS100; Sennheiser HD 414 and HD 424; Sennheiser HD 565 Ovation; Sennheiser PX100; Sony MDR-D77 Eggo; Stax SR207 with leather pads; Yamaha HP-1 and HP-3.

Seth195208's picture

..have the flattest response I have ever seen! Heh heh..

GNagus's picture

Full Sized Open headphones seem to be the easiest to get right.

Also, lets not judge a headphone just on its FR graph. They may sound much better than measured

Seth195208's picture


GNagus's picture

Good response for such an old design

Takato14's picture

Curse you, dummy head --!! Looks like the Sansui SS-100 didn't seal either. But then again, the T50 has similar low bass rolloff... very odd. Maybe damping does more to these than I thought. Will investigate.

I'd like to point out something: while the 4AA measures quite bad, take a look at the 4AAA. Now: for those who don't know, these headphones use completely identical drivers. This is very illustrative of how important implementation is when it comes to headphone design.

It's a real treat to see all these old cans measured, Tyll. That Aurex HR-V9 looks really yummy, and now I want to try an HP-1 again. Thanks a lot!

Seth195208's picture

..a few weeks ago. Pretty amazing sound for the early 70's. No wonder they sold so many.

Long time listener's picture

No in-ear headphone measurements?

Rillion's picture

Now there are two more that I own that Tyll has measured! I noticed the 3 kHz notch on the original PX100 (great headphone by the way) using my in-ear mics, but the notch is so narrow that I find it barely audible in music. I also have the K702 and have been wondering how it compares to the K701--I had been assuming they are pretty much the same but there does seem to be some differences.

andrewdiaz's picture

What does the image mean? Is the headphones designed for rock music? Great creativity.

Hjelmevold's picture

How many kidneys do I have to sell, and what deal must I strike with the devil in order to acquire one of the 25 LCD-1s that were produced?

" actually does quite well" is an understatement! Nevermind the Harman Target Curve, the LCD-1 measures like an open version of the Etymotic ER4PT, and for that I would happily sell both my kidneys and my firstborn child for the next three generations. Oh, Audeze, why did you stray from this wonderful path when designing the current models?

Takato14's picture

... it sounds absolutely nothing like the ER-4. The Etys have better bass, better bass extension, and the LCD-1 sounds pitched up and etched by comparison. The detail is much better on the ER-4 as well.

They "strayed from the path" as you put it because the driver was too close to the ear and the excursion was simply too much. It makes them sound much brighter than they measure.

(The LCD-1 was my pair, ICYDK)

Beagle's picture

...has been my "reference" in my main system(s)more or less since 1977. Others are "better", tons are worse, but none so consistent..

Charles Hansen's picture

Hi Tyll,

Yep, that was my first headphone as well. Pretty good sound, but an hour in them was like putting your head in a vise. Heavy, sweaty, and tight-fitting.

The other misleading claim made about them was that they gave such a good seal because the earpads were "fluid-filled". After a few failures I realized that the "fluid" was air. Yes, technically air is a "fluid", but the lay person is thinking "fluid = liquid"...

Then came the revolution -- the Sennheiser HD-414. It was the world's first "open-air" headphone. Light, comfortable, and half the price of the Koss. They had little bass, but were much clearer sounding than the Koss.

The only one I kept for many years were a pair of Stax SRM-Mk II's, but that was later. The Koss electrostatic came out around the same time, but cost a full month's salary and weighed twice as much as the Pro4aa. Good times!

Keep up the good work.

Charles Hansen's picture

Missed the footnote with the Yamaha HP-1 and HP-3, which were the first planar-magnetics that I knew of. Not bad, but not close to those Stax. The Yamahas definitely had a coloration. Spent many, many, many hours laying in bed listening to the Stax into the wee hours.

Thanks for the memories.

Charles Hansen's picture

Hi Tyll,

One poster commented that the LCD-1 measured "quite well". I have to disagree -- vehemently.

Look at the 300 Hz square wave response. The overshoot is ~6x as high as the signal itself! 6x = 15 dB. That means there is a 15 dB peak at the ringing frequency.

Now look at the Impulse Response graph. Here we see the exact same ringing as at each transition of the 300 Hz square wave. There is only one main frequency of ringing present and it is clearly at ~4 kHz.

Finally look at the uncorrected frequency response. The peak is 15 dB at ~4 kHz. This leaves us with two possibilities:

1) Your ear coupler has a 15 dB Helmoltz- or pipe-resonance at ~4 kHz which is corrected for in the frequency response, but still damages the other measurements.

2) The LCD-1 has a 15 dB resonance at ~4 kHz and the "correction" is inappropriately applied.

Which do you think is correct?

Seth195208's picture

Say what? Charles?

Charles Hansen's picture

Hi Seth,
All of the information in Tyll's acoustic measurements can be computed from the impulse response of the headphone -- the frequency response and the square waves both. This is essentially what a Fourier Transform does -- converts between the time domain (time on the horizontal axis, as in the impulse response and square wave response) and the frequency domain (frequency response graph, with frequency on the horizontal axis).

When you see a huge aberration in one graph, it will appear in all of the graphs. You are just looking at it from different vantage points.

There is clearly a huge problem in the headphone/coupler set-up with the LCD-1 that seems much larger than most other headphones. One possibility is that the Audeze is much more accurately showing the artifacts of the artificial ear-canal (coupler) that Tyll uses. Another possibility is that the Audeze has a resonance in the same frequency range that compounds the problem.

Measuring headphones is a bitch. With speakers and modern test equipment, one can do well above 1 kHz quite easily. Below that it gets harder and harder to get accurate results, let alone anything that correlates with what happens in an actual listening room. With earphones it is almost the opposite. Low frequencies are easy, but oonce you go above 1 kHz there are so many resonances in the coupler that it becomes difficult to separate the artifacts of the coupler from the defects of the headphone.

briskly's picture

I believe it to be both.

The square wave is uncompensated, square waves after the frequency correction is applied may be found here. When corrected for, headphones that measure flatter produce waves that look more square.,496.msg8491.html#msg8491

As for the LCD-1, the measured response is that of the coupled system: the headphone, pinnae, and canals. An ideally accurate headphone measured by microphone at eardrum is going to have a very strong emphasis around 2-4kHz, as would a speaker in nice room.

Charles Hansen's picture

This makes sense. What puzzles me is why the LCD-1 had massively greater overshoot than other high-quality headphones. It looks like the resonance is largely that of the "canal", but there shouldn't be a variation from headphone-to-headphone...

Both the Audeze and Yamaha HP-1 are planar magnetics. Would not expect resonances at 4 kHz with either design, yet the Audeze has a much larger resonance. After compensation, the Yamaha looks *depressed* in the presence range...

Hmmm. Still scratching my head on this one.

Tyll Hertsens's picture
There's a lot of complications with headphone measurements so there's no doubt a combination of things happening here. Bottom line, in the raw measurements there should be a roughly 15dB peak at around 3.5kHz (4kHz in this case) that results from both ear canal resonance and the focusing action of the concha bowl around the entrance to the ear canal.

In the case of these headphone I would say the initial spike is higher than it should be and will likely result in a somewhat piercing sound. But I would also say it's not terribly excessive, nor is it noisy with extraneous crap. So it's a pretty good measurement actually. Especially so as it was Audeze's first headphone.

Charles Hansen's picture

Hi Tyll,

Thanks for the explanation. If there is supposed to be a 15 dB peak in the raw measurements with the coupler, then it appears that the Audeze is doing exactly what it is supposed to be doing -- accurately exposing the resonances of the coupler.

If the ~4 kHz ringing in the Impulse Response and Square Waves are subtracted out, the response of the headphones is nearly perfect. I would not expect it to sound "piercing" in the least in this case.

The best question is "How can we confidently and accurately know exactly what aberrations are generated by the ear coupler?" Once you know that, it should be trivial to subtract the errors out.

The other approach is to perform a two-part measurement, as is commonly done with loudspeakers. One measurement is done with the coupler, and one without. Splicing the two together at a specific frequency would possibly give the most accurate view into the headphone's actual performance.

physics's picture

Hi Tyll,

I know that you're busy with a lot of headphone measurements and other stuff, but I was just wondering if you could give us information on when we might see an evaluation of these phones.

Thanks for all the hard work that you do for the audiophile community.


Tyll Hertsens's picture
I've asked AT numerous times for the headphone and have not been able to get it as yet.
SunByrne's picture

Bummer that AT hasn't been forthcoming. I hated the M70x (no bass), but I love the R70x and can't wait to see those measured.

Would also love to see the K553 Pro measured as well!

Krzysiek Maj's picture

I have question about Phase/Impedance measurement. What does it mean that it was measured with 600 Ohm output impedance? Is it amplifier out impedance? If so, 600Ohm is pretty high, since some have much lesser impedance around 1 or even 0.5 Ohm. Asking, because measured like this almost all cans have certain peak around 80-100Hz. Does that mean that they are difficult to drive for bass region?

sgrossklass's picture

Impedance measurement makes use of an effect that otherwise is often undesired - nonzero output impedance plus a load impedance that is frequency-dependent will result in frequency response being warped (complex voltage divider). So here it's basically being done on purpose, with the frequency-dependent output level loss being measured and used to calculate impedance levels.

I actually have a little online calculator for a similar job, which is converting between frequency response deviation and output impedance when load impedance response is known. Basically the same, just a different set of unknowns.

Actually spots with high impedance tend to be easier to drive, demanding less current from the output stage. If you look at frequency reponse measurements (done at low output impedance), you will find that sound output tends to remain at the level of surrounding areas.

Historically there were some speakers with very low-impedance spots of decidedly nonzero phase in the low midrange, which promptly became known as "amp killers".

sgrossklass's picture

That oBravo frequency response is somewhere between WTF and LOL. Easily the most messed-up mids I've seen on an open headphone in ages.

Good to see some more classics in there. K240Ms don't look bad at all, but then again neither did the old Sextetts (pretty amazing for 1976-ish dynamic cans). I strongly suspect DT150s weren't always as good as this (possibly closer to the 2x 2 kOhm DT100s), but I'm not the one complaining about good cans.

Now would someone mind sending Tyll some HD590s, and/or K240S and family (Mk II, 242 HD)? Pretty please? Speaking of AKGs, there's no K501 either (heresy!), let alone K401, K500, K400, or "poor man's DT880", the K301Xtra / K530. Old (like '90s-vintage) Beyer DT990s could be quite interesting, too...

I'll second the R70x, got one recently (first new cans in ages) and I could see its somewhat warm and bathtubby response become quite popular among the general public. Handily beats my trusty HD580s for deep bass, too, though I guess that's expected these days. Comfort is a bit of an issue though, my large-ish ears just about fit but are not terribly amused in the long run.

HD414 vs. HD424 results are not quite what I expected. (Maybe one is newer than the other and/or age took its toll, the foam behind the drivers is quite likely to be disintegrated now if it was never replaced.)

Anyway, the All Graphs PDF has accumulated a number of bugs over time:

  • AKG K240M, K240 Sextett and Beyer T90 are filed under Closed Full-Size, but are in fact (semi-)open.
  • There are a few more dubious candidates, like Superlux HD668B. That Teknmotion thing isolates extremely little, too. (Same for DT48 S 5 ohm, looks like no seal though.) By contrast, the Philips L1/L2 isolate more than the nominally "closed" AH-Dx000. Pesky grey areas.
  • There's a second K702 Anniv. page under Closed Full-size (p.115).
  • HD239 is nowhere to be found, should go after HD238. Same goes for K240DF. Both are available separately.
  • I assume the "Sennheiser HD575" listed actually is an HD555 (early 150-ohm sample) rather than the real HD575, which is quite exotic. Old Headroom measurements looked similar, anyway.

There also is a bunch of numbers on the Focal Spirit One S page (p.192).

richard3d's picture

Worth mentioning since I had to look at the post for the full list, while this came up first on Google search .. this from the full test archive listing is the vintage 4AAA, not the re-issue with the titanium drivers, best called PRO-4AAAt