Post by gaga on Feb 15, 2021 8:17:12 GMT
first, pardon for my english and for creating first post without introduction first.
for last few days I was doing some measurement on my headphones, an find something (possibly?) quite interesting.
I was testing my old superlux hd681 frequency response to see how my mods affected the frequency responses. I use panasonic wm61a mic inserted to my ear canal (just like earphone), and measure my headphone frequency response.
During measurements, I find the frequency response always has deep notch at +- 7~8khz. I was told that deep notch during measurement is caused by sound interferences (canceling?) or leakage during measurement. So, I tried to reseat my headphone over my ear to get rid the 7khz notch. But no matter how much I tried, there is always +- 7~8khz notch in measurement results.
Curiously, I tried to play sine wave at 7~8khz through my headphone, hear it, and compare to other surrounding frequency to hear if there is a deep attenuation at 7~8khz. Results? there is no big difference in ampilude/volume in +- 7~8khz area and other frequencies. Then I think, "well, this is must be my wm61a mic is broken or fake". Next I tried to measure it on flat surface (made from memory foam) to see the difference. Results is on flat foam surface there is NO 7~8khz deep notch!
Next, I test both of my ears to validate the 7~8khz notch. And the results is yes, both my ears has +- 7~8khz notch (slight different frequency and attenuation, probably because there is no symmetrically perfect human). But why my big ears has deep notch at 7~8khz?
After scratching my head few hours, and after a cup of coffee, I realize my measurements is ONLY over my ears (pinna, concha, ear flange), NOT include an ear canal. So I try to simulate an ear canal, I search on internet average human ear canal size and length, then made some couple rig from cable insulation 0.7mm diameter and 2.5-3.2 mm length, and insert it to flat surface foam test rig and run some sweeps.
The results is facinating, those "ear canal rig" has self resonance between 2.5~3khz (fundamental) and some odd harmonic at 7~10khz. Then done some research on internet and finds the calculation for Closed Air Column Resonance at hyperphysics.phy-astr.gsu.edu/hbase/Waves/opecol.html. The calculated results are same in 1st harmonic, but shifted at 3rd harmonic (possibly due to irregularity of human ear canal shape, e.g not a perfect tube like shape?).
Could it be that 7~8khz deep notch from pinna measurement is to "compensate" odd harmonic that produced by ear canal resonance? if it's true, then nature is weird and interesting is'n it?
Can anyone validate my findings? Is everyone else pinna has 7~8hz notch? since I only has a pair of ears 
(WARNING, this test require 6mm electret mic to be inserted in to ear canal as IEM does, I'm do not take responsibility for any hearing damage if you don't know what are you doing)
(WARNING, this test require 6mm electret mic to be inserted in to ear canal as IEM does, I'm do not take responsibility for any hearing damage if you don't know what are you doing)Here is some graph of my measurement results:
Blue: heavy modified old hd681 headphone (high pad mod), measured over ear with mic inserted in to ear canal (note the "missing ear canal resonances" at 2~3khz region and notch at 7.4khz)
_(edited: somehow the picture move to "attachment")
Green: tube inserted on to flat foam surface to simulate an ear canal resonance
Blue: headphone measured on flat surface, no tube to simulate ear canal, mic placed flat to foam surface (no notch at 7.4khz)
Orange: stock form hd681 with no modification (my ear slightly touch the driver filter), measured over my ear with mic inserted in to ear canal (has notch at 7khz region)
Orange: stock form hd681 with no modification measured over my ear with mic inserted in to ear canal
Purple: other generic no name headphone as comparison measured over my ear (also has notch but shifted to lower frequency, because my pinna touched and pressed on to driver filter due to shallow headphone pad)
Attachments:
