Now I understand.

I find Balance armature has a very nice square. It must be very fast due to its very small size?

Hi RedBull, my friend!

A square wave by nature contains (almost) all frequencies, so it's significant for the whole frequency response. It's just that a 50-Hz square with its larger scale makes higher-frequency inaccuracies seem insignificant, whereas they become more obvious with the 500-Hz signal.

That's the 50-Hz signal, which is less telling. The 500-Hz signal is more significant:



But again the two curves look rather similar. They show that the HD 650 isn't really «slow» and that both have a relatively flat frequency response. But the HD 800 has higher similarity to the original square-wave signal, hence greater accuracy. That's why it is «faster», so to speak.

In the interest of comparability the squares of the different headphones are shifted against each other. So it happens that the two curves are about 165°/195° shifted in phase.


Balanced-armature IEMs are indeed some of the fastest and most accurate headphone drivers. They're just inaccurate when it comes to a natural spatial presentation.
.

You might want to read this paragraph, I think it explains the square wave good and simple:

This is an example between a balanced amateur driver from an IEM vs a dynamic driver from the ATH M50. As said in the paragraph above, these measurements tell you how good the driver is, not how it will sound.



NOTE: I have used different models (ER4P and S) due to how similar they are in the measurements and because they are easier to compare when the lines are close to each other).

Sorry.. Meant the DT48a..

What's shown in the graphs is the opposite of this. It shows that armatures have a very slow rise time. This reflects their small bandwidth: they cut off sharply at about 17khz. They also happen to have bad harmonic distortion. Though they have excellent low frequency extension and very quick decay (due to their tiny enclosure, not the drivers), "fast" they most certainly are not.

The attribute «fast» applies to the whole transient response, not just attack. A very short decay is at least as important for the impression of impulsivity. The admittedly lower high-frequency extension is a consequence of the acoustics within the drivers (canals!) and not of limited reaction speed. Nevertheless, 17 or 18 kHz is still beyond the upper limit of most adult ears. Listen to drum beats through an electrostat and though a balanced-armature IEM and tell me which has better attack!

As to harmonic distortion, according to HeadRoom's measurements balanced-armature drivers indeed seem to produce higher harmonic distortion than conventional dynamic drivers (but not higher HD compared to electrostatics). Harmonic distortion isn't coupled with he impression of «speed», though, and balanced armatures still sound explicitly transparent and clear to my ears.
.

I agree, they are exceptionally transparent in some respects, but this comes back to what you mentioned about decay. The very quick decay in this case is due to the tiny housing in the drivers are in, not due to the armature drivers themselves. The bandwidth though is a property of the drivers, which are by definition "slow" in that respect.

Since we're talking about something that's quite loosely defined here, I think it helps to know that different elements of the headphone play a part in what we commonly perceive as "speed". The bandwidth of the drivers plays a part, the frequency response of the headphone as a whole plays a part as does the driver enclosure.

I'm not sure what you mean by the tiny housings being responsible for the short decay. I think the tiny moving mass plays the biggest part, add to this the tight air cushion in the rear of the membrane when it comes to low frequencies.


Only if you just take rise time/impulse onset into consideration. In terms of motion control the high-frequency reproduction is still very «fast». But I would agree that high(est)-frequency resolution might be inferior to good dynamic and electrostatic headphones. So if you will we could call the high-frequency response relatively slow – from the bandwidth and resolution aspect.

However, it's important to know that transient speed (in the sense of accuracy) is never uniform from bottom to top, it varies considerably depending on frequency, and this applies to all driver principles. All things considered I think balanced-armature drivers are still fastest in the sense of accuracy, even though highest frequencies suffer a bit from bandwidth limitation.
.

Just to add my 2 cents worth to this old discussion

 

All drivers are the same "speed" - 5,000 cycles per second is the same 5,000/sec from an electrostatic as well as from a dynamic.  Also, the speed of sound is the same regardless of the type of driver used....

 

Typically what users perceive as "speed" is a mix of high-frequency response characteristics, dynamic compression (or lack thereof) and decay time.  But to to talk about "speed" is really an inaccurate syntax.

 

 

 

 

 

Edited by milosz - 8/25/11 at 3:44am

just my 2 cents

 

http://www.head-fi.org/t/514102/testing-for-speed-in-iem-s-please-participate-in-my-test-speed-metal-visual-kei-symphonic-metal-instrumental-electric-guitar-shredding/105

Electrostatic drivers are also light sounding, and not just fast. 

What's one of the "faster" electrostats? I'll be hearing a few at a headphone meet soon... just in case there's opinions on that...