Effect of output impedance on headphone with a flat impedance-freq curve

[jkp]

Hi everyone,

I'm relatively new here so I'm not too sure about how things work, but I was curious about the effects of output impedance on headphones that have flat impedance-frequency curves.

I recently acquired a 2011 LCD 2 and am thinking about picking up a headphone amp for it. I'm currently using an Onkyo TX-NR509 Speaker amp's headphone jack and there seems to be limitless power and pretty clear sound.

I've read that an amps output impedance can colour a headphone's sound depending on how the impedance of the headphone changes over frequencies. In the case of the LCD 2 (and many other planars) the impedance is basically constant, as can be seen in this graph from innerfidelity:

Am I correct in assuming that my amp shouldn't colour the sound of my headphones from its (probably) high output impedance?

and would be any other audible differences in me getting an amp like the JDS atom as compared to my current set-up. Perhaps a slight increase in clarity and decrease in noise? Or are there other things that I am not considering?

 

[JohnYang1997]

If you measure the distortion across the headphone, distortion increases along the output impedance. No matter how flat it is as measured, there is still magnetic field and counter force of the diaphragm, causing loss of impact and cleaness to the sound.

 

[jkp]

That makes a lot of sense, thanks for the help!

 

[solderdude]

Whether sound is colored or not due to output impedance depends on its impedance.
I have written an explanation here
It's all about voltage division.
The LDC 2 will not change it's sound but will just play softer.

 

[JohnYang1997]

Whether sound is colored or not due to output impedance depends on its impedance.
I have written an explanation here
It's all about voltage division.
The LDC 2 will not change it's sound but will just play softer.

The sound will change. Tonality or fr probably not. But there is difference in texture and dynamic.

 

[JohnYang1997]

Not only output impedance. If you measure the distortion at the output jack of amplifier and at the headphone end of cable, the distortion will swing 10 times or more, especially at lower frequency range and resonant frequency.
Another way to think of it is the feedback sensing point. The output impedance is essentially a isolation from the feedback point and the driver. A lot of things can happen in the resistor and cable. So best scenario is to have minimal output impedance , good amount of feedback , low impedance cable.

 

[solderdude]

Nope, there is a difference in level which is easily confused with difference in dynamic, texture and even FR.

I tested this (with a planar) by adding output resistance and with the same switch remove the exact attenuation at the input of the amp to correct for the attenuation.

Of course, there are lots of headphones that DO change their tonal balance and as usually this means elevated mid bass the sound becomes 'bloated' and 'fat' which is easily perceived as 'lack of control' in the bass.

What is measured at the headphone is back EMF coming from it that drops across the series resistor and as that is a deviation from the original signal it measures as 'distortion'.

You are totally free though to have another explanation for it.

 

[JohnYang1997]

Nope, there is a difference in level which is easily confused with difference in dynamic, texture and even FR.

I tested this (with a planar) by adding output resistance and with the same switch remove the exact attenuation at the input of the amp to correct for the attenuation.

Of course, there are lots of headphones that DO change their tonal balance and as usually this means elevated mid bass the sound becomes 'bloated' and 'fat' which is easily perceived as 'lack of control' in the bass.

What is measured at the headphone is back EMF coming from it that drops across the series resistor and as that is a deviation from the original signal it measures as 'distortion'.

I said: no change in fr.
But distortion increases.
In perception it's impact and cleaness of sound. I tested er2se recently also sundara before. There are headphones that has impedance looks flat but actually not. There will always be a resonant frequency, probably at 4k or somewhere. And low frequencies because the driver needs to move air the emf it generates will cause distortion. But if you see in time domain, it just means it can't go that high at the top. The way to solve it is to use feedback to push it up. This is still in electrical sense, not talking about servo. But there is difference.

 

[solderdude]

I did not hear any difference when level corrected.
It is very easy to hear differences when not level corrected and attribute them to the wrong things.
Also I see no differences in decay (CSD and spectrum plot) nor scope shots measured with a microphone with planars.

 

[JohnYang1997]

I did not hear any difference when level corrected.
It is very easy to hear differences when not level corrected and attribute them to the wrong things.
Also I see no differences in decay (CSD and spectrum plot) nor scope shots measured with a microphone with planars.

I hear difference with level matched. It's not as big like fr change of course. But very audible.
It's very easy to measure the distortion across the output of amplifier and the driver. There will be distortion rise at lower frequencies and resonant frequency.

 

[JohnYang1997]

My procedure was to use a dac/dongle db magix ac3 which has opa1612 + a resistor at the output.
Two same amplifiers I made with -125db distortion across the frequency range. 0.6uv noise from 20-20khz. 0.03ohm output impedance.
Chain1: ac3 - er2se
Chain2: ac3 - amp 0db gain - er2se
Chian3: ac3 - amp + gain - etymotic PtoS adapter - er2se
All volume matched

 

[jkp]

If I'm understanding everything correctly I think the biggest change in my case would be the effect of the damping factor between a proper headphone amp and the one I am currently using.

As you mention in your article, damping factors above 10 have quickly diminishing returns; however, in the case of using a speaker amp's output, there is a good chance that my damping factor could be well under 10 (I don't have a source for this, but I have read around that many speaker amps can have ridiculously high output impedances.)

Would you think that I am correct in my reasoning?

Whether sound is colored or not due to output impedance depends on its impedance.
I have written an explanation here
It's all about voltage division.
The LDC 2 will not change it's sound but will just play softer.

 

[solderdude]

Speaker amps themselves don't have rediculously high output resistances.
The headphone sockets of said amplifiers often DO have output resistances that may even be as high as 100 to 600 Ohm .
So when you connect to speaker amp terminals you basically have a very high power and very low output R amplifier but chances are you can only use a very small part of the volume control (with associated poor channel balance in that region) and may hear noise.
You won't be blowing up LCD-2 easily though.
What one can do is use an output attenuator for speaker amps.
(This answers your original question)
Another option is to use an amp like the Atom or other amp that can drive the LCD2 well (60 Ohm is a very friendly load for most amps) and has an R out below 10 Ohm.

Damping of planar membranes is mostly mechanical/acoustical.
The fact that the impedance does not vary is evidence of this so damping factor is of no issue here.
When the impedance rises this is caused by back EMF which lowers the current.
As the voltage remains the same and the current is lower (generated by the driver) the impedance changes.
As the impedance does not change there is no back EMF current generated of any significance.
Of cousre some small current is generated (after all you can use it as a microphone) but is very, very small.

For speakers with heavy cones it can be an issue but the DF is usually already worsened by the large inductors and cables already.
Another reason to go active with speakers ..

 

[Theo]

It would be nice to have some measurements to be sure we're talking about the same thing. A distortion spectrum measured at the amp output when loaded by the headphones, maybe?

 

[jkp]

I might just have to build one of those! Would make for a fun project!

Can the fact that the impedance does not vary be used to assume that the resonant frequency of the drivers themselves is outside the typical frequency listening range and therefore not add any distortion?
Or is that that just due to the large amount of mechanical/acoustic damping that allows any resonance of the drivers to be neglected without the electric damping?


Also, I'm intrigued by your last sentence, do active speakers really have that much of an advantage over passive ones? I had always assumed that as long as the cables were properly sized and amps chosen well that there wasn't much difference between the two types.

Speaker amps themselves don't have rediculously high output resistances.
The headphone sockets of said amplifiers often DO have output resistances that may even be as high as 100 to 600 Ohm .
So when you connect to speaker amp terminals you basically have a very high power and very low output R amplifier but chances are you can only use a very small part of the volume control (with associated poor channel balance in that region) and may hear noise.
You won't be blowing up LCD-2 easily though.
What one can do is use an output attenuator for speaker amps.
(This answers your original question)
Another option is to use an amp like the Atom or other amp that can drive the LCD2 well (60 Ohm is a very friendly load for most amps) and has an R out below 10 Ohm.

Damping of planar membranes is mostly mechanical/acoustical.
The fact that the impedance does not vary is evidence of this so damping factor is of no issue here.
When the impedance rises this is caused by back EMF which lowers the current.
As the voltage remains the same and the current is lower (generated by the driver) the impedance changes.
As the impedance does not change there is no back EMF current generated of any significance.
Of cousre some small current is generated (after all you can use it as a microphone) but is very, very small.

For speakers with heavy cones it can be an issue but the DF is usually already worsened by the large inductors and cables already.
Another reason to go active with speakers ..

 

[solderdude]

It would be nice to have some measurements to be sure we're talking about the same thing. A distortion spectrum measured at the amp output when loaded by the headphones, maybe?

You will measure a higher THD for certain when a headphone is connected via a resistor. John Siau has written an article with proof of that.
The higher distortion NUMBER and plots is not really distortion but the back EMF voltage that drops across the resistor and none of it across the amp output (which is low impedance)
Damping, however, is done by current. That current is generated by the driver which has a resistance. That resistance and generated (unloaded) back EMF voltage determines the max damping current.
The added resistance lowers the current . There thus is no difference in current when an output R (in this case) is 0 Ohm or 6 Ohm (60 vs 66 Ohm)
As the generated EMF voltage is ver low in a planer and resistance is 60 Ohm there will be nothing substantial to 'damp'.

 

[solderdude]

Also, I'm intrigued by your last sentence, do active speakers really have that much of an advantage over passive ones? I had always assumed that as long as the cables were properly sized and amps chosen well that there wasn't much difference between the two types.

Woofers can generate quite a lot of back EMF. The impedance of that source is low (4 to 8 Ohm). As one can see the impedance of woofers (most woofers without special countermeasures) rises substantially which means the generated current is high.

Passive speakers have inductors in its path. Most manufacturers don't like to buy thick copper wire inductors and resort to iron cores etc. costs need to be low. So the resistance of them is of importance and when filtered near a resonance frequency the impedance is higher of that inductor and can change the sound.
When active the speaker is connected with small thick wires and no inductance in series is arguably and measurably better. A small increase in series resistance lowers damping currents substantially in speaker systems and that's where high damping currents exist and matter.

 

[KSTR]

Oh well, a lot of misinformation here....

Generally, headphones are no different that speaker drivers, they are current-operated devices (except ESLs). Measured acoustic distortions alway correlates more to current, not voltage. This is reflected in the fact that when driven from high source impedance the terminal voltage gets distorted by the back-EMF (BL is nonlinear and that applies both for the translation from current into motion as well as to motion into current), measured distortion (acoustic and current) is typcally lower, however. And vice versa, when driven with zero source impedance the measured distortion in current is highest (as is acoustic distortion).

Like with speaker drivers, it turns out that there may be a little electrical damping needed around the (low frequency) system resonance but generally high source impedance is best and should be preferred. It depends on the system Q how much, if any, electrical damping is required. The lower Q and the more stable (vs operating conditions like seal etc) it is, the less damping is required.

Response must be EQ'd according to the impedance curve (if it isn't flat), to get the same transfer as with voltage drive which is the condition the phones were voiced with (modern phones, at least. Older models may have been voiced with somthing like 100Ohms source resistance as that was the typical headphone output impedance for decades).

Note that distortion in good headphone drivers is already quite low so the improvement from current drive isn't very relevant. Current drive complicates matters because of the model-specific EQ needed, that's why we don't see in practice.

 

[solderdude]

Note that distortion in good headphone drivers is already quite low so the improvement from current drive isn't very relevant. Current drive complicates matters because of the model-specific EQ needed, that's why we don't see in practice.

That's also why planars sound the same on the Bakoon (current drive HP amp) as on any 'normal' amp but other headphones like HD595 etc sound substantially different from the Bakoon as it drives with a current instead of a voltage and the output voltage peaks where the impedance peaks.
Since most headphones are designed to be voltage driven (low output R), that is with the odd exceptions not counted, they should not be driven from a current source nor higher output R voltage source (which kind of acts a bit as a current source)

 

[Sharur]

This is reflected in the fact that when driven from high source impedance the terminal voltage gets distorted by the back-EMF (BL is nonlinear and that applies both for the translation from current into motion as well as to motion into current), measured distortion (acoustic and current) is typcally lower, however. And vice versa, when driven with zero source impedance the measured distortion in current is highest (as is acoustic distortion).

I currently use my DT880 600 Ωs from an integrated amp with an output impedance of 68 Ω. I've been considering trying them from the speaker terminals and was intrigued by what you said. If I drive them with close to zero source impedance, there will be more distortion?