[DivineCurrent]

When looking at measurements of headphones with different driver types, I see many similarities between headphones with the same driver type, as to be expected. For example, we see dynamic driver headphones such as the HD650, DT880, and AKG K701 have a similar tilt in THD starting from about 100 Hz and rising as it approaches 20 Hz. There are of course exceptions, but most dynamic driver headphones will have this distortion in the bass, according to measurements all over the internet. Next, we have planar magnetic headphones, most of which have low THD, but many have rough frequency responses compared to dynamic driver headphones. For example, take the HD650 and compare to the Hifiman HE-560.
 
http://www.innerfidelity.com/images/SennheiserHD650.pdf
http://www.innerfidelity.com/images/HiFiMANHE5602014.pdf
 
Comparing the two, there are a lot of jagged edges and dips in the HE-560 across the range. HD650 is pretty smooth with only a few gradual peaks and dips. Electrostatic headphones also exhibit similar characteristics to planar magnetic measurements, with even lower distortion. 
Another driver type to mention is balanced armature for IEMs. I own a Klipsch X10, and playing sine wave tones at and around 1 kHz, I can clearly hear 3rd harmonic distortion. Graphs support this, as there is almost 1% 3rd harmonic distortion in that frequency range in the X10, as well as the other popular single balanced armature IEM, Etymotic ER4-S and their variants. You can test this out yourself by generating a 1 kHz tone in Audacity and generating another tone at 3 kHz, but at -46 dB (which is 0.5% THD). Play both together, and A/B the difference with and without the 3 kHz tone. There should be a clear difference. Most dynamic driver IEMs do not even have audible distortion, and yet balanced armature IEMs, which are supposedly higher quality than dynamic drivers, have much worse distortion.
 
All this got me thinking about driver types and their differences. Many claim that planar magnetics and electrostats don't have good soundstage and imaging compared to dynamics, but there's no measurements to confirm what soundstage or imaging actually looks like. Also, people mention multi BA driver IEMs have phase problems, but that may be based on speculation rather than measured evidence. It seems to me, based off the hundreds of measurement graphs of headphones I've looked at, dynamic drivers tend to win in the smooth frequency response area, while planars in general have better bass performance and lower distortion. And BA IEMs, while they have a more clear treble response, most have audible distortion, unless they have more than two drivers and the 3rd harmonic distortion is well under 0.5%. 
 
I realize I'm just sort of ranting and spewing out stuff right now, but is there any reason why planars don't measure as consistently and have as smooth frequency response as dynamic drivers, or single balanced armature drivers can't have as low distortion as dynamic IEMs? I'm sure there are physical limitations or each driver type, but I don't understand why even $1,000+ planar and electrostat headphones measure worse than under $400 dynamic ones. 

 

[spruce music]

I like your post.  It is sort of a stream of consciousness style that is how I sometimes reason out things myself when just beginning to ponder.
 
I'm not in a position to fill in much useful information unfortunately. 
 
I saw a discussion a couple years back about IM distortion.  Lots of people believed headphones had enough of this it made high sample rate recordings of some material sound different than CD.  Many more headphones have response to 40 khz than speakers. The idea being headphones would intermodulate down into the audible range.
 
Then someone pointed out some measures by different people showing most phones had tiny IMD above 10 khz.  I tried a 30 and 33 khz tone at max level.  The IMD product would have been at 3 khz where your ear is very sensitive.  I heard nothing, but silence.  Tried 22 khz and 25 khz.  Again even at what should have been an over driven level nothing, but silence.  This was with  DT880 and later Sony 7510 phones.  Later some Sennheiser 535 phones with the same silent result. 

 

[DivineCurrent]

I like your post.  It is sort of a stream of consciousness style that is how I sometimes reason out things myself when just beginning to ponder.

I'm not in a position to fill in much useful information unfortunately. 

I saw a discussion a couple years back about IM distortion.  Lots of people believed headphones had enough of this it made high sample rate recordings of some material sound different than CD.  Many more headphones have response to 40 khz than speakers. The idea being headphones would intermodulate down into the audible range.

Then someone pointed out some measures by different people showing most phones had tiny IMD above 10 khz.  I tried a 30 and 33 khz tone at max level.  The IMD product would have been at 3 khz where your ear is very sensitive.  I heard nothing, but silence.  Tried 22 khz and 25 khz.  Again even at what should have been an over driven level nothing, but silence.  This was with  DT880 and later Sony 7510 phones.  Later some Sennheiser 535 phones with the same silent result. 

That's interesting. I've never tested for IM distortion, but seems it's nothing to worry about if you can't even hear it at the most sensitive frequencies.

Right now, I'm comparing different driver types, but also considering how pads can change sound, whether bad or good. It makes sense that headphones with breathable pads like velour would have a bass rolloff, as the sound escapes from the driver cup and makes it harder to keep that bass inside. Measurements also show that bass extends farther on headphones with pleather or leather pads, regardless of driver type, like the new Focal headphones. Downside with pleather pads is it sometimes causes unwanted frequency reflections inside the cup, so you have more peaks and ringing than headphones that use velour. Most IEMs with a good seal already have good extended sub bass. The Hifiman RE-400 has a titanium dynamic driver, and has sub bass that goes flat all the way down to 25 Hz.

 

[pila405]

@achelgeson - we have almost the same setup (although I went with the Chord Mojo instead of ODAC\O2. I also noticed what you said and I think that the primary reason is that in this field there is little correlation between price and performance, and you see it time and time again. People pay more assuming they get a better product, but without actually properly testing for the manufacturers' claims, thus making their (manufacturers') lives very easy. Take Audeze for example - they release more and more expensive LCDs, yet the performance seems to get worse with each model (higher distortion, bass roll-off, etc'...). If more customers would demand measurements upon release and poses a more critical approach I guess it would change it and you wouldn't see as many expensive BA IEMs. I eagerly wait for measurements of the new iSine IEMs by Audeze.

Do you know of other dynamic IEMs with low distortion as the hifiman?

And about the stats - seems like many of the newer and much more expensive models also perform just as good or worse than previous ones (Stax 207 vs 009)

Some graphs:
Audeze:
LCD 4: http://www.innerfidelity.com/images/AudezeLCD4.pdf
LCD 2 (of a lucky owner): http://www.innerfidelity.com/images/AudezeLCD2SN5325928.pdf
 
Stax
009: http://www.innerfidelity.com/images/StaxSR009SNSZ92251KGSS.pdf
207: http://www.innerfidelity.com/images/StaxSR207SB2217.pdf

 

[DivineCurrent]

@achelgeson
 - we have almost the same setup (although I went with the Chord Mojo instead of ODAC\O2. I also noticed what you said and I think that the primary reason is that in this field there is little correlation between price and performance, and you see it time and time again. People pay more assuming they get a better product, but without actually properly testing for the manufacturers' claims, thus making their (manufacturers') lives very easy. Take Audeze for example - they release more and more expensive LCDs, yet the performance seems to get worse with each model (higher distortion, bass roll-off, etc'...). If more customers would demand measurements upon release and poses a more critical approach I guess it would change it and you wouldn't see as many expensive BA IEMs. I eagerly wait for measurements of the new iSine IEMs by Audeze.


Do you know of other dynamic IEMs with low distortion as the hifiman?


And about the stats - seems like many of the newer and much more expensive models also perform just as good or worse than previous ones (Stax 207 vs 009)


Some graphs:

Audeze:

LCD 4: http://www.innerfidelity.com/images/AudezeLCD4.pdf

LCD 2 (of a lucky owner): http://www.innerfidelity.com/images/AudezeLCD2SN5325928.pdf

Stax

009: http://www.innerfidelity.com/images/StaxSR009SNSZ92251KGSS.pdf

207: http://www.innerfidelity.com/images/StaxSR207SB2217.pdf

Yup, you are absolutely correct. I tried the HD800S and LCD-X a few months ago, and was not blown away. I thought the HD650 had a more smooth and realistic response than the two, supported by measurements. (Of course, the 800S had amazing soundstage) Only downside of the HD650 is the second harmonic distortion below 100 Hz, which is honestly only barley noticeable above normal listening levels.

The Hifiman RE-400 has extremely low distortion, much much lower than any BA driver IEM I've seen. Although most dynamic driver IEMs have inaudible distortion, the Etymotic MC5 had comparable measurements:
http://rinchoi.blogspot.com/2012/04/etymotic-research-mc5.html?m=1
Only problem is the drop off above 11 kHz.

 

[castleofargh]

aside from a huge spike in the midrange/trebles, I wouldn't even look at the distortion levels of IEMs. they really don't seem to relate to audibility, at least not consistently so.
 
about BA drivers supposed to be better, I have never thought that way. maybe it's another of those famous "it's more expensive to make so it must be better" situation?
 
 
 
on a side note, my MC5 is nowhere as low distortion as what Rin's measured. maybe he got lucky, maybe I got a lemon, IDK. I like it and I can EQ the low end way better than with other etymotics, but the shell is slightly too large for me when it should still go deep in the canal. and indeed it's not the IEM of trebles heads ^_^.
 
I bet the RE400 measures well thanks to its amazing cable(warning this comment is highly sarcastic and shouldn't be showed to sensitive pregnant women).

 

[briskly]

IMD isn't really separate from THD, but are just different methods to demonstrate the nature of the non-linearity of the system itself. You can determine modulation components from knowledge of the harmonic products . Using spruce-music's example, for 3kHz modulation product from 30+33kHz and 22+25kHz tones, the system would have to have sufficiently high 2nd order term to produce a noticeable artifact. As a note, listening for distortion by a sine wave is a worst case scenario. Listening to anything remotely like music will obscure new products: the more spectrally dense, the more effective the masking by high amplitude tones.
 
Balanced armature drivers are fundamentally non-linear. In contrast, the drive mechanism behind dynamic and the electrostatic drivers is linear. Practical concerns such as changing material stiffness and field strength tend to introduce non-linearity, but the basic principle is linear. As far as distortion is concerned, balanced armatures are handicapped from conception. Advantages of a balanced armature are in the small size, and their exceptional output and sensitivity. The stiff membrane has increasing output with increasing frequency into the ear canal load. This necessarily means output is also totally dependent on the load acoustic impedance, and that they are more sensitive to leak. This also goes for measurements with DIY couplers, and the treble droop they tend to exhibit compared to 711-style couplers is clearly evident. BAs will also be subject to multiple resonances in the audio range that need to be damped in some way.
 
Anecdotally, distortion is not the central problem for most balanced armatures.

 

[pila405]

  on a side note, my MC5 is nowhere as low distortion as what Rin's measured. maybe he got lucky, maybe I got a lemon, IDK. I like it and I can EQ the low end way better than with other etymotics, but the shell is slightly too large for me when it should still go deep in the canal. and indeed it's not the IEM of trebles heads ^_^.
 
I bet the RE400 measures well thanks to its amazing cable(warning this comment is highly sarcastic and shouldn't be showed to sensitive pregnant women).

Haha!
But maybe it's about time you move from the MC5 to RE400, because the correlation between when I see and what I hear is very high. The sound is very clean (though does require some EQ)
 

 

[castleofargh]

I have both. my opinion given the general fidelity of IEMs is that the signature makes most of how we feel about them. sadly it's only the rambling of a random guy as I have no evidence to back that up. I enjoy the sound of the RE400 because I find it really innocuous and that's often what I'm looking for, but I can't say I'm hearing anything special.

 

[DivineCurrent]

  Haha!
But maybe it's about time you move from the MC5 to RE400, because the correlation between when I see and what I hear is very high. The sound is very clean (though does require some EQ)
 

I do EQ the RE-400 a bit, mostly just to bring down 2 and 5 kHz down 3-4 dB. They really respond better to EQ than most headphones/IEMs I've tried. I don't really know the science behind how well drivers respond to EQ though. 

 

[pila405]

I also use EQ and noticed the same. But except dips around 2K\5-6K, I also think they need a little boost around 3Khz. Otherwise the vocals sound a bit far and unnatural.

 

[WoodyLuvr]

When looking at measurements of headphones with different driver types, I see many similarities between headphones with the same driver type, as to be expected. For example, we see dynamic driver headphones such as the HD650, DT880, and AKG K701 have a similar tilt in THD starting from about 100 Hz and rising as it approaches 20 Hz. There are of course exceptions, but most dynamic driver headphones will have this distortion in the bass, according to measurements all over the internet. Next, we have planar magnetic headphones, most of which have low THD, but many have rough frequency responses compared to dynamic driver headphones. For example, take the HD650 and compare to the Hifiman HE-560.

http://www.innerfidelity.com/images/SennheiserHD650.pdf
http://www.innerfidelity.com/images/HiFiMANHE5602014.pdf

Comparing the two, there are a lot of jagged edges and dips in the HE-560 across the range. HD650 is pretty smooth with only a few gradual peaks and dips. Electrostatic headphones also exhibit similar characteristics to planar magnetic measurements, with even lower distortion.
Another driver type to mention is balanced armature for IEMs. I own a Klipsch X10, and playing sine wave tones at and around 1 kHz, I can clearly hear 3rd harmonic distortion. Graphs support this, as there is almost 1% 3rd harmonic distortion in that frequency range in the X10, as well as the other popular single balanced armature IEM, Etymotic ER4-S and their variants. You can test this out yourself by generating a 1 kHz tone in Audacity and generating another tone at 3 kHz, but at -46 dB (which is 0.5% THD). Play both together, and A/B the difference with and without the 3 kHz tone. There should be a clear difference. Most dynamic driver IEMs do not even have audible distortion, and yet balanced armature IEMs, which are supposedly higher quality than dynamic drivers, have much worse distortion.

All this got me thinking about driver types and their differences. Many claim that planar magnetics and electrostats don't have good soundstage and imaging compared to dynamics, but there's no measurements to confirm what soundstage or imaging actually looks like. Also, people mention multi BA driver IEMs have phase problems, but that may be based on speculation rather than measured evidence. It seems to me, based off the hundreds of measurement graphs of headphones I've looked at, dynamic drivers tend to win in the smooth frequency response area, while planars in general have better bass performance and lower distortion. And BA IEMs, while they have a more clear treble response, most have audible distortion, unless they have more than two drivers and the 3rd harmonic distortion is well under 0.5%.

I realize I'm just sort of ranting and spewing out stuff right now, but is there any reason why planars don't measure as consistently and have as smooth frequency response as dynamic drivers, or single balanced armature drivers can't have as low distortion as dynamic IEMs? I'm sure there are physical limitations or each driver type, but I don't understand why even $1,000+ planar and electrostat headphones measure worse than under $400 dynamic ones.

Very interesting and thoughtful post.

I too have recently been investigating/researching this mysterious difference in headphone driver signatures in order to figure out how to improve my H6s via EQ'ing using a Harman Target Response Curve offset, as well as discovering why I so strongly preferred my H6s over my LCD-2s (recently gifted them away) and many other tier one audiophile cans (many of which are planars). Besides looking at the frequency response it is important to take square wave results into consideration and can explain why some frequency response plots look worse than they sound and vice versa... also, it is becoming evident that human auditory perception for some reason knows how to ignore (turn off) certain messy frequency responses that should be problematic.

 

[castleofargh]

many people prefer to show graphs with applied smoothing because they feel it better translates to how they will hear the frequency response. and IMO it does. the downside is that it makes the FR look way more linear than it really is and we may fail to see actual problems in the headphone.
but yes, a rough frequency change doesn't necessarily sound the way it looks. masking and other psychoacoustic tricks are of course responsible.

 

[WoodyLuvr]

I wonder how much human vs. individual ear resonance is a factor in smoothing out headphone frequency response naturally?

 

[castleofargh]

well the matter of having custom heads make it so that we ideally need custom sound on headphones. but I doubt it has a smoothing impact. aside of course from using a headphone with a 15db boost at a frequency where your ear already resonates less than most people. it might be the difference between a lot and too much in such a specific situation.
but I like to think that taste and personal obsessions over some kind of sound make a huge part of our specific tolerance that defines "too much". even more so for those who fell into the amateur audiophile rabbit hole. for example if my music contains any sort of audible noise or Celine Dion, it will make me mad in a way that is completely disproportionate to the acoustic variations. it's the "shaken not stirred" of James Bond, I make a mountain of something small, and then come to always see it as a mountain(or Celine Dion). ^_^
IMO that's a huge part of individual preferences. what we assume to be good, what we assume to be unacceptable. both can have roots outside of pure acoustic or individual bodies, we know it can even come from outside or reality.