[purrin]

Just posting some preliminary measurements regarding characteristics of different driver types used in IEMs: balanced armature and dynamic. Unfortunately, I only have a few data points: UERM, Shure e2C, and ER-4S. I really need more, but there do seem to be some very distinct patterns going on. Again, more data is needed.
 
The measurements are consistent with what I know of how different orders of harmonic distortion sound like in my tests of speaker drivers and headphones. Many people have said that dynamic drivers reproduce texture better in the bass. My own experience tells me that higher odd order distortion (3rd, 5th) tends to make bass sound more "one-note" or indistinct than 2nd order distortion, which tends to make bass sound more "reinforced". Let see how these harmonic distortion measurements pan out.
 
The measurements were taken with the IEMs calibrated at 90db SPL/A to within 0.10db with white noise. D2 = 2nd order harmonic, D3 = 3rd order harmonic, etc. Finally, take this limited data for what it is. It's preliminary and refinements to the process may be required.
 
UERM

 
ER-4S

 
Shure E2C

 
 
It's interesting that the BA drivers of the UERM and ER-4S have higher odd order distortion 3rd, 5th in the lower frequencies compared to the dynamic driver. In fact, the dynamic driver has better distortion overall. The UERM does have less overall distortion than the ER-4S in the bass and most of the midrange, probably because a dedicated bass driver is used. This did come as a little bit of a surprise because the ER-4S and UERM sound much more agile, "faster", more much more detailed than the E2C.
 
It's important not to judge which is better or worse. This is just data and we are just trying to correlate the data with specific characteristics of what we hear. There are other sonic characteristics of BA driver which are better than dynamics, but they are not measured here. These particular measurements are not the end-all of everything. 

 

[billybob_jcv]

Very cool purrin - nice work!  What is the generally accepted limit of audibility for this type of test?  At 90 dB, 0.1% is 0.09 dB.  Are the higher orders actually significant for what we can hear?  I'm just speculating:  since this distortion is at the end of the chain, it will not be amplified, so does that mean it would also be valid to consider the audibility of the absolute magnitude of the distortion?

 

[purrin]

On the audibility of distortion, it depends upon who you ask, and where in the frequency range. These distortion measurements are single tones. For single tones in the bass, I feel that 1% is very audible, and that we start to hear things at 0.1%. I feel that distortion is harder to detect higher up in the frequency band. Real music is much more dynamic and complex, so the actual distortion with such may be much more evident than with single steady state tones.
 
I think Head-Fi'er ultrabike posted a doohickey web application which lets you test your ability to hear distortion (this test of course is limited by the inherent distortion of your headphones and equipment, which may mask the intentional distortion produced by the test).
 
For reference, the attenuation in db is related to the log of the distortion in percent:
 
0.1% = -60db
0.5% = -45db
1% = -40db
5% = -26db
 
(I think I did the math right. Someone please correct me if I can wrong.)
 
Finally, I do wonder more about the characteristics of BA drivers, with their higher levels of 3rd, 5th (both odd order) relative to conventional dynamic drivers. Square waves are composed of the the sum of the original fundamental sine wave and successively decreasing levels of odd order harmonics. While the levels of distortion with these BA drivers are not even nearly sufficient enough to turn them into square waves (LOL!), I do wonder about its effects, i.e. how the distortion may contribute to the "BA" sound.
 
Perhaps the "speed" or "edge" of BAs can be partially attributed to its distortion characteristics. I will look up the impulse responses too. Just postulating and no flames please. I've been pretty sick lately of extremists of either side of a debate. 

 

[billybob_jcv]

Thanks!  I forgot to consider we're talking about logarithmic functions... 

 

[frenchbat]

Thanks for the work Purrin, interesting as usual. According to that website http://www.sengpielaudio.com/calculator-thd.htm, your math is correct.
 
Does anyone know the address of the web application for distortion ? There's always the tool at klippel.de, but I'd like to try some other ones if possible.

 

[average_joe]

Great work.  I think it is also important to point out that balanced armatures reproduce sound due to different sources, Maxwell vs. Lorentz.  It is due to the different ways of recreating the sound that result in different performance of the harmonics.
 
The UERM and ER4S are more similar than different in note presentation to my ears and I am curious to see how other BAs such as the 5-way and To Go 334 because the bass sounds much closer to that of a dynamic driver than the UERM and ER4.  Also, I wonder how something like the Rooth products would measure against something like the JHA products, and see where the ES5 falls.
 
Also where would certain dynamics such as the FX700, a Merlin (or K3003), and GR07 perform.
 
Good stuff.