• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Measuring headphone amplifiers distortion

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
As we all know, headphones, similarly as speakers, have non-linear and frequency dependent impedance. This non-linear impedance creates a divider with headphone amplifier output impedance, so the distortion measured at headphone amplifier output (behind its intrinsic output impedance) would greatly differ from distortion measured into a resistor or linear RLC dummy load. Same applies to frequency response, but this is well known.

Let me demonstrate the distortion issue on example of virtually same headphone amplifier, once with 50 ohm output impedance and second with close to zero output impedance. The headphones used are Sennheiser HD598.

Headphone amplifier with 50 ohm output impedance

audiobuffer+hd598_fr.png

Frequency response and distortion into Sennheiser HD598

audiobuffer+hd598.png

Distortion in % into HD598


Output impedance close to zero

headamp_buf2+hd598_fr.png

Frequency response and distortion into Sennheiser HD598

headamp_buf2+hd598.png

Distortion in % into Sennheiser HD598


Conclusion

Not only frequency response is affected by headphone amplifier output impedance, but also distortion. The distortion depends on headphone impedance non-linearity, on headphone amplifier output impedance and on interaction of headphone non-linear impedance with the headphone amplifier output stage. Measurements performed into resistor or linear RLC load cannot reveal this behavior.
 

KeithPhantom

Addicted to Fun and Learning
Forum Donor
Joined
May 8, 2020
Messages
641
Likes
658
If what you say is true, this something that has not been explained by the major headphone measurement website to the best of my knowledge. Thank you for providing these insights.
 

KeithPhantom

Addicted to Fun and Learning
Forum Donor
Joined
May 8, 2020
Messages
641
Likes
658
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
This has been explained before.
What you are seeing in the measurements is the back EMF current from the driver measured across the 50 Ohm.

Yes, but to be more precise I would prefer to speak about nonlinearity and Ohms law. Clean voltage source divided by nonlinear impedance makes nonlinear current. This nonlinear current makes nonlinear voltage drop across the 50 ohm resistance and this is measured as a distortion. The conclusion should be that increased output impedance modulates not only the frequency response, but also the distortion. And it is good to measure it, as the prediction is uneasy from traditional measurements into resistive load.

The term “back EMF” is extremely confusing. RLC dummy impedance also creates the “back EMF”, but not the distortion! The distortion is created only in the nonlinear elements of the mechano-electric transducer of the headphones (or speakers).
 
Last edited:

KSTR

Major Contributor
Joined
Sep 6, 2018
Messages
2,690
Likes
6,013
Location
Berlin, Germany
Now the interesting thing about this is that with increased drive impedance the current distortion in the headphone will tend to be lower, and the measured acoustic distortion will follow that.
Low output impedance is good for flat FR (assuming that the headphones were designed for zero drive impedance which must not be always a given) but it does not help distortion, quite the contrary...
 

KSTR

Major Contributor
Joined
Sep 6, 2018
Messages
2,690
Likes
6,013
Location
Berlin, Germany
And I've said before that measuring any amplifier with real nonlinear loads does NOT expose increased distortions from the amp, rather it exposes nonlinear load current causing equally nonlinear voltage drop across any output resistor or R//L combination, which then might be mistaken for increased amplifier distortion which it is not.
 
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
Now the interesting thing about this is that with increased drive impedance the current distortion in the headphone will tend to be lower,

This is correct in almost all cases. Not true only in case that the output stage has had issues driving complex load. And this might happen, same as for power amplifiers and complex loads.
 

KSTR

Major Contributor
Joined
Sep 6, 2018
Messages
2,690
Likes
6,013
Location
Berlin, Germany
But to measure the amp's true distortion the complex load itself should be linear. I'd actually prefer an active load (sine current source fed back into the amp while it is outputting the same frequency, with a variable phase shift to massage all 4 output quadrants). Then it is true one will see the actual amp distortion.
 
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
But to measure the amp's true distortion the complex load itself should be linear.

I agree, but I would like to see the "real world" related measurements as well. This is my long term experience with power amplifier design and even in headphones we may find examples of very peculiar load impedance. So why not to try both.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
You can measure the amps performance but have to do this before the 50 Ohm.
After the 50 Ohm you are measuring 2 things.

With some amps, where the output R is caused by the amp and is not merely an added resistor it is a more difficult task.

I can show real world measurements but the problem is that the headphones and mics used in the room where I measure in exceeds that of the amplifier by a lot.

I quickly checked measurements with the passive N60NC at various output R's (0, 10, 32, 120) and because it has no impedance hump it shows a lower THD the higher the output R is. In this case it is because the excursion is lower.
With the 559 you see a slight increase in lowest frequencies and a lower distortion above 100Hz with increasing resistance.
Here too the culprit is the lower excursion.

Should repeat the measurements but with amplitude correction so distortion from the acoustical path is similar. Alas... way too busy doing that at the moment. Have an ATH SR9 incoming. May test that one this way while I am at it.
 

Custard

Member
Joined
Aug 18, 2019
Messages
18
Likes
12
This is all a bit of a headf**k for those of us (me, actually) who are slightly less technically inclined, so is it possible to generalise the effect of output impedance on distortion? Particularly as it applies to planars, because they're supposed to have essentially 'flat' impedance and intrinsically low distortion. And because, despite the kicking they've received on this forum, I really like the unhyped sound and sharp imaging of the LCD-X (after a bit of EQ) which I have intuitively supposed to be because of the low distortion.

Looking at PMA's graphs, the extra distortion would still seem to be quite low compared to that of a regular dynamic headphone anyway - is it still significant?

I really appreciate the topic, BTW, so thank you to all of the above contributors for the enlightenment.
 

maverickronin

Major Contributor
Forum Donor
Joined
Jul 19, 2018
Messages
2,527
Likes
3,308
Location
Midwest, USA
Yes, but to be more precise I would prefer to speak about nonlinearity and Ohms law. Clean voltage source divided by nonlinear impedance makes nonlinear current. This nonlinear current makes nonlinear voltage drop across the 50 ohm resistance and this is measured as a distortion. The conclusion should be that increased output impedance modulates not only the frequency response, but also the distortion. And it is good to measure it, as the prediction is uneasy from traditional measurements into resistive load.

The term “back EMF” is extremely confusing. RLC dummy impedance also creates the “back EMF”, but not the distortion! The distortion is created only in the nonlinear elements of the mechano-electric transducer of the headphones (or speakers).

@solderdude is completely right.

If you measure the distortion of the same headphone acoustically with near zero Z out you will see that the shapes of the distortion curve matches what you measure electrically with a high Z out almost perfectly.

It's just back EMF from the driver combined with a voltage divider between the driver and the amp's output impedance.
 
Top Bottom