• 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!

An attempt to measure headphone (non) linearity

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
Based mostly on discussion in this thread
https://www.audiosciencereview.com/...be-amplifier-distortion-vs-solid-state.18832/
I thought about alternative methods to measure headphone linearity. I tried the X-Y plot where on X-axis there is a driving voltage and on Y-axis there is a transfer ratio in dBV/V of acoustical output/drive voltage. I tried my Sennheiser HD598 at several frequencies from 100Hz to 10kHz and voltage range from -80dBV to +6.85dBV. HD598 have sensibilty of 112dB/1V/1kHz. The measurement was calibrated at 1kHz to give 0dBV/V approx. The area between -80dBV and -60dBV is affected by microphone noise at 100Hz and 200Hz plots. It seems that at 100Hz and 200Hz the output SPL is stressed above 0.1V driving voltage. On the other hand, higher harmonics make the 1kHz and 2kHz plot to rise above 1V driving voltage. I would especially appreciate if @solderdude could put a comment.

HD598_linearity_s.png
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
IME most headphones are fully linear below 70dB SPL.
The 'noise' below -50dB is bound to be misinterpreted by plot gazers.

Perhaps, when using colors adhere to a single color scheme (rainbow order)
Because 5 and 10kHz are very small in amplitude perhaps concentrate on lower frequencies as that is where we can expect non linearities.

Maybe 30Hz, 70Hz, 150Hz, 400Hz, 1kHz and 6kHz ?
Most headphones have problems between 5 and 8kHz, or at least a peak there.

When possible use Volt (log scale) as most people prefer that over dBu or dBV which is more of a studio thing (because of the dB)

A worthwhile but time consuming ? method.

I sweep at 3 or 4 levels and then overlay when I find a headphone with issues (mostly seen in high 3rd order distortion at 90dB)

linearity.png


and overlaid:
linearity-overlayed.png

(note the dB scale)

Below the very cheap HD681:
lin-hd681-overlay-1.png

(don't mind the 50,100 and 200Hz wiggles :facepalm:
 
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
Loop or industrial hum?

-20dBV makes 92dB SPL, so yes. Nonlinearity is above this. I will be lazy to change the axes, for the simple reason. Those who know they know. For those who don’t the change in labels would not help.
 
Last edited:

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
cheap ass WM61A microphone on a home made mic amp. Was designed for battery usage but now on a small linear PS. Hum did not increase.
It's picking up mains hum, probably in the mic capsule as everything is well screened and in metal boxes and screened cables.
When I put an iMM6 mic on the same mic pre-amp the hum is gone.
Would need to make a different calibration file for it. The WM61A, however, is much better extended in the lows and highs... so a trade off. Hum vs flatness.
Still planning to properly address it one day.
The SRH1840 plot was measured with the iMM6 (which I use only for this type of measurements) the HD681 was done with the WM61A capsule

Interestingly the upwards tilt at 1 and 2kHz is also observed in the SRH1840.
 
Last edited:
  • Like
Reactions: pma
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
Understood, I found the same issue when playing with electret capsule.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
Just for fun you can also sweep at 50, 60 and 70dB SPL but I found that even the worst linearity headphones do this about equally well.
Perhaps measure down to 20 Hz ?

The advantage of this method is you see the effect for the entire FR range.

Interestingly the effect one gets is in opposite direction of that of the Phon curves.
However, also 3rd harmonics increase (very soft clipping effect) and for high SPL that may reach audible levels (> several percent)

There are plenty of headphones that do not show 'compression'. Of course some might above 100dB (I can't measure higher).

K371:
linearity-overlay-95-65.png


Edition XX:
lin-ed-xx-overlay.png
 
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
I still think that the idea shown in post #1 is worth considering. Below the measurement of Vivanco SR750 linearity at 100, 200Hz and 1kHz.

SR750_linearity_s.png


and again HD598
HD598_linearity2_s.png
 
Last edited:
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
Yes, its worth considering but it leaves you guessing about other frequencies.

Then a combination of the two methods would make it. One leaves you guessing about other frequencies, the other one about other levels ;). One is continuous in frequencies, the other one in levels.
 

bigjacko

Addicted to Fun and Learning
Joined
Sep 18, 2019
Messages
721
Likes
359
Nice discussion and measurement here.:) What makes headphone nonlinear with voltage? Compression of port or suspension issue? Now Peter got even more work to do....:facepalm:
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
Can be caused by overdamping of the driver, can be caused by the maximum excursion of the voicecoil where the magnetic force is going outside that of the magnet, can be caused by ports or a combination of those factors. Fact is some do this at higher SPL levels. I have never seen one doing this at sensible listening levels.
The lower the frequency is the higher the excursion is.
 
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
How does it do at say 60Hz @ 2mW ?

Badly. However, my setup is primitive and I am not confident how good is the measurement below say 200Hz. I will try to make the setup improvement, in 2021 ;). My first New Year resolution ;). Anyway, PF2021!!

HD598_ac_H2H3_thdfreq_s.png
 
OP
pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
I would like to show what I use as a "headphone amplifier" for my tests. It is a small, linear amplifier with output impedance of 0.06 ohm. It is able to drive very difficult real life headphone loads easily. To demonstrate this feature, I have made a couple of measurements with JBL Control1 Pro connected as a load, instead of the headphones.

First, JBL Control1 Pro impedance

JBL_impedance.png


THD and THD+N at 1kHz/1V into JBL Control 1 Pro (about 125mW)
headamp+JBL1.png

Please note complete absence of the (linear) power supply hum components


THD vs. frequency at 1V into JBL Control1 Pro
BUF2+JBL_Control1_distortion.png


With this headphone amplifier, that is able to drive the 2-way real speaker, I am quite confident to make headphone measurements, but even so I always check the electrical output distortion.

BUF2_headamp.JPG
 
Last edited:
Top Bottom