[Silentanarchy]

On Inner fidelity i'm trying to read some graphs.http://www.innerfidelity.com/images/HiFiMANHE5602014.pdf 
 I understand how to read the first two but I need help learning how to read the following.
 
%THD+noise @ 90dB and 100dB
 
Electrical Impedance and Phase
 
Impulse Response
 
and also what the significance of the two square wave tests are.
 
If you could tell me where to go to learn more advanced audio engineering for free. it would be appreciated

 

[bigshot]

Well, I'll start with the THD+noise...
 
On the right, the vertical scale is distortion measured in a percent of the signal. In general 1% THD is the threshold of audibility, so anything under that is good.
 
Along the horizontal axis we have frequencies. Distortion can vary from frequency to frequency. The most important range there is from 1kHz to 4kHz, because that is the range where our ears are most sensitive to noise and distortion.
 
The various colored lines represent the amount of distortion and noise at two volume levels (90dB and 100dB) in two channels (left and right). Both of those volume levels are quite loud, but 100dB is probably louder than would be comfortable to listen to.
 
So looking at the chart, we can see that aside from one tiny narrow spike between 100Hz and 1kHz on the right channel (which looks to me like just manufacturing variation), nothing gets up into the audible range. (And even that little spike is so narrow, it wouldn't be audible.)
 
Hope this helps. Someone else can jump in on the rest.

 

[SilverEars]

I'll add about impedance and phase.
 
When you look at the impedance graph, you will notice it states magnitude.  Impedance is a sum of real and imaginary value.  Z = R +jX.  R is the real value, and X is the value of the imaginary part of the impedance Z.  Impedance is a general term so if X = 0, impedance is real valued or purely resistive or resistance. If R is zero, it's purely reactive(capacitive or inductive with Z being either -jX or jX)
 
The magnitude value of Z is like the hypotenuse of the triangle of the impedance or the root of R^2 + X^2 which is the value at specific frequencies on the impedance response graph.  
 

 
When there is a complex component X to the impedance Z, there is a phase.  Or what we can visualize as current and voltage waveform not being synchronized or timed together.  Since according to Ohm's law V = I*Z,.  Z=V/I and can get the phase angle from converting the rectangular form of Z to phasor form using Euler's identity. With some trig you get the phase angle, and positive phase is voltage leading current.
 

 
So with this said, if the magnitude increase as frequency increases since if inductive, Z=jwL with w being frequency, and capacitance Z=1/jwC.  
 
So, based on this, when you look at the increase in magnitude of impedance as frequency gets higher, you know the complex part of Z  is inductive, and if getting lower, complex part is capacitive.
 
Also if complex part X of Z is negative valued, that means it's capacitive.  Since for capacitor, Z=1/jwC = -j/wC
 
This also means that the phase will be a negative value or current leading voltage.
 
So, all this explains why when the magnitude of impedance increases, impedance has inductive part, and capacitive when magnitude is decreasing.
 
  Also phase is negative when there is capacitance, and positive when there is inductance, and purely resistive when phase is zero.  You can see all this on the graph below.
 

 
And with all this info, you can guess what it means when the graph looks like this.
 

 

[castleofargh]

you're asking a lot ^_^.
 
about thd+noise, the idea of having both is to show all the sounds that end up in your ears but were never on the record. I guess that's sums it up.
and to get an idea of how loud is 1%, a signal is sent, you can see with a signal at 90db, and the same measurement at 100db on innerfidelity. so you actually have 2 separate measurements on the same graph, I guess the point is to see if the headphone becomes a mess when pushed a little loud by comparing the evolution at both loudnesses.
10% is 20db below the original signal
1% is 40db below
0.1% is -60db below
etc
 
meaning that if you end up with THD+noise at 1% with a signal at 90db, what you will hear is the signal at 90db and something that isn't the original signal up to 50db(40db below the original signal).
strangely enough we use to say that anything below 1% is ok for headphones, but we cry if an amp has 1% distortion, and nobody would buy a DAC with even 0.1% ^_^.  it's just that headphones are so bad as a reproduction tool that we lower our expectations.
and to make things more fun, speakers have even bigger distortion values.
 
 
but maybe we could just complain to Tyll for never startnishing this http://www.innerfidelity.com/headphone-measurements-explained and this http://www.innerfidelity.com/headphone-measurement-procedures
Tyll Y U no do?

 
"it would take ages and nobody will read it" is not a valid answer anymore, @Silentanarchy want to read it all 3times.

 

[bigshot]

  strangely enough we use to say that anything below 1% is ok for headphones, but we cry if an amp has 1% distortion, and nobody would buy a DAC with even 0.1% ^_^.

 
That's because of the number rabbit hole that people fall into when they believe better numbers always mean better sound. The truth is that 1% is fine. It's so far down below the signal, it doesn't make any difference when you're listening to music. Better specs are fine and dandy if they are easy and inexpensive to achieve, but it won't make better sound for human ears.

 

[Silentanarchy]

  Well, I'll start with the THD+noise...
 
On the right, the vertical scale is distortion measured in a percent of the signal. In general 1% THD is the threshold of audibility, so anything under that is good.
 
Along the horizontal axis we have frequencies. Distortion can vary from frequency to frequency. The most important range there is from 1kHz to 4kHz, because that is the range where our ears are most sensitive to noise and distortion.
 
The various colored lines represent the amount of distortion and noise at two volume levels (90dB and 100dB) in two channels (left and right). Both of those volume levels are quite loud, but 100dB is probably louder than would be comfortable to listen to.
 
So looking at the chart, we can see that aside from one tiny narrow spike between 100Hz and 1kHz on the right channel (which looks to me like just manufacturing variation), nothing gets up into the audible range. (And even that little spike is so narrow, it wouldn't be audible.)
 
Hope this helps. Someone else can jump in on the rest.

 
 

  I'll add about impedance and phase.
 
When you look at the impedance graph, you will notice it states magnitude.  Impedance is a sum of real and imaginary value.  Z = R +jX.  R is the real value, and X is the value of the imaginary part of the impedance Z.  Impedance is a general term so if X = 0, impedance is real valued or purely resistive or resistance. If R is zero, it's purely reactive(capacitive or inductive with Z being either -jX or jX)
 
The magnitude value of Z is like the hypotenuse of the triangle of the impedance or the root of R^2 + X^2 which is the value at specific frequencies on the impedance response graph.  
 

 
When there is a complex component X to the impedance Z, there is a phase.  Or what we can visualize as current and voltage waveform not being synchronized or timed together.  Since according to Ohm's law V = I*Z,.  Z=V/I and can get the phase angle from converting the rectangular form of Z to phasor form using Euler's identity. With some trig you get the phase angle, and positive phase is voltage leading current.
 

 
So with this said, if the magnitude increase as frequency increases since if inductive, Z=jwL with w being frequency, and capacitance Z=1/jwC.  
 
So, based on this, when you look at the increase in magnitude of impedance as frequency gets higher, you know the complex part of Z  is inductive, and if getting lower, complex part is capacitive.
 
Also if complex part X of Z is negative valued, that means it's capacitive.  Since for capacitor, Z=1/jwC = -j/wC
 
This also means that the phase will be a negative value or current leading voltage.
 
So, all this explains why when the magnitude of impedance increases, impedance has inductive part, and capacitive when magnitude is decreasing.
 
  Also phase is negative when there is capacitance, and positive when there is inductance, and purely resistive when phase is zero.  You can see all this on the graph below.
 

 
And with all this info, you can guess what it means when the graph looks like this.
 

 
 

  you're asking a lot ^_^.
 
about thd+noise, the idea of having both is to show all the sounds that end up in your ears but were never on the record. I guess that's sums it up.
and to get an idea of how loud is 1%, a signal is sent, you can see with a signal at 90db, and the same measurement at 100db on innerfidelity. so you actually have 2 separate measurements on the same graph, I guess the point is to see if the headphone becomes a mess when pushed a little loud by comparing the evolution at both loudnesses.
10% is 20db below the original signal
1% is 40db below
0.1% is -60db below
etc
 
meaning that if you end up with THD+noise at 1% with a signal at 90db, what you will hear is the signal at 90db and something that isn't the original signal up to 50db(40db below the original signal).
strangely enough we use to say that anything below 1% is ok for headphones, but we cry if an amp has 1% distortion, and nobody would buy a DAC with even 0.1% ^_^.  it's just that headphones are so bad as a reproduction tool that we lower our expectations.
and to make things more fun, speakers have even bigger distortion values.
 
 
but maybe we could just complain to Tyll for never startnishing this http://www.innerfidelity.com/headphone-measurements-explained and this http://www.innerfidelity.com/headphone-measurement-procedures
Tyll Y U no do?

 
"it would take ages and nobody will read it" is not a valid answer anymore, @Silentanarchy want to read it all 3times.

 
Thanks for such wonderful explanations. I love the community here, very blunt and informative.

 

[cjl]

   
That's because of the number rabbit hole that people fall into when they believe better numbers always mean better sound. The truth is that 1% is fine. It's so far down below the signal, it doesn't make any difference when you're listening to music. Better specs are fine and dandy if they are easy and inexpensive to achieve, but it won't make better sound for human ears.

1% can be audible sometimes, and besides which, there's really no excuse for any modern electronic equipment to be above 0.1% anyways (it's really easy to achieve).