Weird AKG K1000 Isolation Measurements

Last week, when I posted about AKG K1000 measurements, many readers were perplexed about the apparent low frequency isolation with what ought to be about the most open headphone you could possibly have. How in the world could it be providing any isolation at all, much less in the low frequencies that are particularly had to shut out?

Frankly, I thought I must have just done something stupid and there was something wrong with the measurements (though I did double-check at the time). So, I remeasured again this weekend...and basically got the same result!

In the plot at the top of the page, the black trace is with no headphones on my measurements head with pink noise playing in the measurement chamber. The plots is somewhat noisy, but it's flat +/-5dB.

The red plot is with the AKG K1000 on the measurement head, and with the ear speakers in the closed, closest to the ear, position. As you can see there's a huge notch at 60Hz indicating about 25dB of isolation. With the ear speakers in ever more open positions, we see ever decreasing amounts of isolation in the notch.

My guess is that were seeing a natural resonance at 60Hz of the K1000 earspeaker, either in part or as a whole, which is somehow anti-resonant with the incoming sound and canceling it out at the ear. You can also see in the plot some very specific peak features shared by all the K1000 curves at about, 120Hz, 170Hz, 210Hz, 270Hz, 310Hz, and a few more. I'll take a shot and guess this may be the grills singing a bit...or, it's some sort of standing wave cancelation between the parallel grills. (Note the effect doesn't lessen as much with angle as the 60Hz notch, which indicates the artifacts are likely from different mechanisms.)

I have my doubts, that's for sure...I'd lay maybe even money on my guesses. Anybody else what to take a stab at it?

steaxauce's picture

Do you hear the isolation when you wear the K1000?

steaxauce's picture

Hey Tyll, I think it would be a great sanity check to try them out and see if you hear the same isolation effect that you measured, with, say, a 60Hz tone playing from your loudspeakers.

This would tell us if the isolation results from some serendipitous interaction in your specific dummy head/microphone/measurement chamber/speaker setup, or if it actually exists in use.

Anyone else with a K1000 and a subwoofer at home can feel free to try this out as well and let us know what you find.

kais's picture

The notch is close to the driver resonant frequency, 72 Hz on my own specimen.
So it's very likely it's the driver canceling the incoming sound with an antiphase-oscillation.
The distance effect is a strong evidence for this theory.
The spikes are more likely mechanical vibrations of the complete phone or certain parts of it (e.g. the metal arcs) hanging on the head.
The independence from the angling makes this very likely.
The frequencies are too low for a grill resonance, air or mechanical.

GNagus's picture

I think this is a good explanation. Are the drivers damped in the same way as normal headphones?

Dan Wiggins's picture

That would be my guess as well. When measuring any effects on any acoustical/transducer system, you mush short the terminals so you damp the transducer motion. Connect the sleeve to the ring and tip and I bet you'll have a different result!

Tyll Hertsens's picture
Yeah, I'm sure you're correct, the driver resonance itself, of course.
--------------'s picture

"I remeasured again this weekend...and basically got the same result!"

The results look nothing alike to me. In the original measurement, it shows a 12 dB dip centered around 100 Hz. In this measurement, the notch is 25 dB deep and centered around 60 Hz.

Tyll Hertsens's picture
In the original measurements the ear speaker was angled, more similar to the green trace.
MRC01's picture

If the black line is flat within 5 dB, does that mean the dummy head response is not raw, but already compensated for HRTF?
With a raw plot measured from mics inside a dummy head ear canals, listening to ambient pink noise in the room, I'd expect to see something like +15 dB @ 2700 Hz, among other variances.

Tyll Hertsens's picture
Yeah, I kind of side stepped going into it, but the chamber has a background curve subtracted. I did an initial test some time ago to measure the actual chamber response and then store it as a curve for background subtraction. So, isolation results are aways somewhat noisy, but they're a good general indicator of the isolation.
Serious's picture

The K1000, because of its open design, has to be tuned for a much stronger resonance than normal open backed headphones, so I guess we should see the same effect on every headphone, just to a much lesser degree. I haven't seen the effect on any of your open backed over ear headphone measurements, though. The other headphones most likely have a much stiffer suspension than the K1000.
I think it's best to think of it as some of the energy of the speakers going into the K1000 drivers, which then doesn't reach the ear. It sort of acts as a bass absorber at its resonant frequency. At least that's my guess.
Would be interesting to see impedance measurements of the K1000.

Hifihedgehog's picture

you might ever measure the (non-baby) HE-90's or the new Orpheus?

suatulanhsidebyside's picture

Whether you try the isolation of Z1R of Sony. Is it good? I need a review about this, i wanna buy one. :)
My website here

zobel's picture

When you established a 'correction curve' for the chamber as represented by the dummy head response curve restored to flat, wouldn't that curve be useful as a compensation curve to apply when measuring frequency response of cans?

It seems to me that it could represent the correction needed for measuring flat with headphones as well as the pink noise in the chamber, as long as the chamber was previously measured with a calibrated mic and a curve then applied to that measurement to remove chamber response from signal. Right?

zobel's picture

1) Play a precision sine sweep (or pink noise), in the closed chamber

2) Record the chamber response with a calibrated microphone.

2a) Derive correction curve for chamber
2b) correct signal to drive the chamber in flat response

3) Play flat (corrected) sweep (or pink noise) in chamber. The chamber is now calibrated to match the signal (flat)

4) Record the corrected sweep in chamber through the dummy head

5) Derive a difference curve from flat to dummy head response

6) Use this curve (5) to compensate the measured dummy head response from playing the original sweep through headphones into it.

kais's picture

Your proposal is the standard approach for diffuse field (or any field) correction.
It does not contain any new idea, sorry.
Unfortunately doing this inside a small chamber leads to unwanted results because of the very uneven sound field. To make things worse, the sound field is easily disturbed by any tiny change, like replacing the small measurement microphone with the large dummy head.
Furthermore even moving the microphone a few millimeters will yield different results, so there is no reference spot (or spots, you need two for both ears) to place the dummy head.
Sorry, but this leads to nowhere.
We had this discussion when Tyll tried to calibrate the dummy head to the Harman target response.
If you read through this you will find that even under best circumstances lots of questions are left unanswered.
Finally it seems Tyll did not decide to follow this idea any further.

Jim Tavegia's picture

You are worried about "isolation" from an open design, discontinues pair of headphones? I own many Grado's and a pair of AKG-K701's and I don't used them in a noisy environment, so I don't have any competing sounds
coming in the the back side. I use my Sennheiser HD-380's or my 2 pair of AKG-K271's for that. I also have two pair of Sony 7506's but they are a little long in the tooth now.

I must be missing something.

This this test for the noise that comes from the drivers that might re-enter the cans from the outside.

cspirou's picture

When I checked the measurements I noticed you didn't include an impedance curve like for other headphones. That might have given a clue about the resonance.

Also check to see if you get the same thing with a regular mic instead a binaural head.

david smith89's picture

thanks for sharing i like this and i learned a lot.....It's a good piece of information |

walmart savings catcher

nabila01's picture