Feasibility of measuring headphones down to 0.01% THD

[fineMen]

Hello. I have been planning an audition for a high-end electrostatic headphone purchase, namely a choice between the Warwick Acoustics Bravura, Audeze CRBN, and Stax SR-X9000, whereby I am hoping to obtain comparative distortion measurements (among others) to bolster my decision, and of course provide the community with this data.

Not tightly related, but do you know about Dr. Griesinger's assessment of individual headphone equalization? Due to physiological differences between people e/q cannot be generalized. E/q will alter the distortion figures significantly.

 

[solderdude]

E/q will alter the distortion figures significantly.

That is not consistent with my findings. Of course the equalizer circuit could add distortion.
Also when EQing low bass in and reproducing that many dB's louder with a headphone that already exhibits higher distortion in that area the distortion will rise (also IM distortion)

 

[fineMen]

That is not consistent with my findings. Of course the equalizer circuit could add distortion.
Also when EQing low bass in and reproducing that many dB's louder with a headphone that already exhibits higher distortion in that area the distortion will rise (also IM distortion)

Yep, sloppy language on my side. What I'm saying is that the need for individual e/q implies that the individual's sensitivity over frequency is individual. First to know is that the frequency response of a headphone is pre-equalized to cope with a resonance of the standard ear canal and maybe a standardized outer ear. But we are not standard individuals ;-)

Best I give an example:
one person has a longer ear canal resonating at 3kHz
the other a shorter one resonating at 4,5kHz

Some distortion generator within the headphone would sit on 1,5kHz. For person (1) the HD2 @3kHz gets pronounced, for person (2) the HD3 @4,5kHz.

Conclusively, the distortion figures have to be calculated taking the various resonances within the receiving device, the ear namely, into account. Only because the o/p speaks of 0,01 percent of an accuracy.

We cope with this subtlety when assessing speakers in that we say, distortion must be low in the 1..2kHz range. Because the most prominent HD2 (2kHz..4kHz) and HD3(3kHz..6kHz) will hit the most sensitive range of the human ear.

But with headphones I tend to say that speaker/ear became a tightly coupled system so that an excuse like the above rule isn't appropriate anymore.

 

[solderdude]

That still does not mean distortion is increased.
It only means for some instruments and tones the harmonics may sound a bit different than originally intended for that person.
The distortion itself does not increase.

 

[fineMen]

That still does not mean distortion is increased.
It only means for some instruments and tones the harmonics may sound a bit different than originally intended for that person.
The distortion itself does not increase.

I eddited my post as to indicate that the topic is a matter of perspective. No offense.

 

[Mr. Haelscheir]

I was considering measuring distortion figures after normalizing the headphones' FRs toward some calibrated at-shop speaker reference.

 

[fineMen]

I was considering measuring distortion figures after normalizing the headphones' FRs toward some calibrated at-shop speaker reference.

Please do so.

 

[DanWiggins]

Hello. I have been planning an audition for a high-end electrostatic headphone purchase, namely a choice between the Warwick Acoustics Bravura, Audeze CRBN, and Stax SR-X9000, whereby I am hoping to obtain comparative distortion measurements (among others) to bolster my decision, and of course provide the community with this data.

As far as I could tell, to measure down to 0.01% THD (a benchmark level apparently attained by the Sennheiser HE-1), if these headphones actually reach such performance (the DCA Stealth per Jude's measurements on Head-Fi reaches such in the treble), I would need to at the shop provide a portable acoustic enclosure ensuring a noise floor of 20 dB or less while measuring 100 dB signals at the mic. The mic's self-noise would have to be at that level, and its own distortion must not exceed that noise floor at said 100 dB recorded level.

The miniDSP EARS would be the most convenient setup, but does anyone know its self-noise and distortion specs (at least of the supposed UMIK-1 capsules used)? At the minimum per the link in https://audiosciencereview.com/foru...tting-thd-in-the-perspective.9085/post-232599, it attains 1% THD at a venerable 133 dB max SPL. https://www.avsforum.com/threads/the-great-waveguide-shootout.1475674/page-12 mentions issues with second order harmonics. But https://audiosciencereview.com/foru...l-world-performance-review.33918/post-1186487 claims the ability to "measure up to ~120 dB SPL without compression or added harmonic distortion."

If the miniDSP EARS is known to be insufficient for measuring down to 0.01% THD, how feasible would it be to DIY a rig within a $1k budget (where I am not so concerned about ear gain simulation)?

We built two of these for kickaround use in the lab/warehouse/production area. Plenty good to give you 30-40 dB of attenuation and are quite portable. When I need absolute quiet, I'll use one of these inside my chamber (11 dBA noise floor) and get down to the limits of the microphones and AP gear.

Build a Portable Acoustic Test Box

 

[Mr. Haelscheir]

Welp, I called up the local shop (I originally wanted to wait until around the time I was actually ready), and despite the online advertised availability of pickup, they do not actually have those models of estats available for demo. There is one other shop that only has the CRBN available for demo. So in practice, it seems like the only stuff I would be able to measure are stuff I've bought.

Anyways, at least the Audeze CRBN's distortion performance is known, and the X9000's could be extrapolated from the 009S's. I will still be receiving in-ear mics with a headtracking kit (Earfish) and might be able to use those for some time domain measurements.

PS: When it came to typewriters, curiosity regarding unique mechanisms and typing feels was easily sated with purchases yielding lasting and objectively unique joy, but it's a lot harder to justify owning a high-end headphone collection that I intend to EQ to the same speaker target just in the name of gathering those detailed measurements.

 

[greatestprateek]

Even if you could reliably measure well below 0.01% (noise floor, wideband) it is a level that is far below any audible thresholds so not very important.
Above 100dB many headphones start to 'compress' a bit already and above 120dB it will already be an audible mess (many many percent).

Averaging techniques can be used to measure below noise levels as well and one can only 'look' in very narrow bands which can also improve things.
For simple measurements (with sweeps) you will indeed run into practical noise floors, acoustical as well as electrical.

Hey can you tell what % distortion is audible. I am a high spl listener and the many many headphones I have heard sound very bad/harsh/(distorted?) to me when pushed to 90-95db and above. At 70-75db even cheap things sound good to me. What could be the reason? The only headphone that I remember not distorting at high spl was the Susvara. I didn’t measure its db but I confidently remember it was super loud. That headphone has magical clarity. Is it also born out in THD measurements? Comparatively two other headphones, one that gets a lot of praise Edition XS and one thats a classic HD650 really sound trash to me at any higher than moderate volume. I would stay away from calling these headphones good even or worth buying. Normal speakers are so much better and more enjoyable sounding louder cleanly and in a refined manner. It just doesn’t make any sense. Speakers cannot be always played loud because of disturbance to others, so we turn to headphones but if headphones distort at a perceived loudness lower than even speakers then those headphones are nothing but useless.
Edition XS is really really poor sounding harsh and edgy even at low volume. HD650 sounds refined upto a satisfactory point but push it and it quickly spits its dummy. HD800S was much much better but it too had its limit. Abyss Diana TC was garbage. Surprisingly Crinacle TruthEar Zero Red was quite refined too but it doesn’t have any midrange to be considered useful.

 

[fineMen]

Hey can you tell what % distortion is audible. I am a high spl listener and the many many headphones I have heard sound very bad/harsh/(distorted?) to me when pushed to 90-95db and above.

Well, I'm not the person addressed by you. But did you had a peak at Dr. Griesinger's explications on individual headphone equalization? Cans are tuned to fit the common ear, but your's may differ enough to end up with some elevated frequency response in the most sensitive area.

Second to that, especially when exposing the hearing to loud sounds often, one might develop a (subliminal?) hyperacusis. Better see a doctor on this, no alarm raised, though!

Could it be the amp, eventually?

 

[greatestprateek]

Well, I'm not the person addressed by you. But did you had a peak at Dr. Griesinger's explications on individual headphone equalization? Cans are tuned to fit the common ear, but your's may differ enough to end up with some elevated frequency response in the most sensitive area.

Second to that, especially when exposing the hearing to loud sounds often, one might develop a (subliminal?) hyperacusis. Better see a doctor on this, no alarm raised, though!

Could it be the amp, eventually?

Hyperacusis? Whoops, I hope not. There are times if I am mentally exhausted that I feel irritated by ambient noises around me though.

As for Amplifier, I have used multiple and its the same with all.

But like I said the main thing is that some headphones sound better than others. So I am wondering if its indeed distortion/lack of quality. Would it be that Susvara and to some extent the HD800S are designed in a way that suits my ears. If I am reading FR graphs of Crinacle properly the Edition XS has very close matching to Susvara. But they sound nothing alike to me. Its bonkers to pay $6000 for a headphone for sure, even if great like Susvara, but Edition XS isn’t worth a penny either.

 

[fineMen]

Hyperacusis? Whoops, I hope not. There are times if I am mentally exhausted that I feel irritated by ambient noises around me though.

As for Amplifier, I have used multiple and its the same with all.

But like I said the main thing is that some headphones sound better than others. So I am wondering if its indeed distortion/lack of quality. Would it be that Susvara and to some extent the HD800S are designed in a way that suits my ears. If I am reading FR graphs of Crinacle properly the Edition XS has very close matching to Susvara. But they sound nothing alike to me. Its bonkers to pay $6000 for a headphone for sure, even if great like Susvara, but Edition XS isn’t worth a penny either.

Reiteraded, please do not feel alarmed. I suffer once in a while from hyperacusis. You know, test-listening new speakers at way to high volumes. All fine, but the day after it sounds horrible - hyperacusis.
My Sennheiser and the AKG perform great in regard to HD. I measured it ok. So much so that I used these for investigations of my personal THD limits. You can do that too. Be very (!) careful with single sines. They can burn into the ear like staring at the sun.
Btw, the ear generates HD by itself, especially at higher volumes (sic!). From the perspective of physics it is a limited mechanical apparatus in the end, right?

 

[Mr. Haelscheir]

Regarding hyperacusis, I have long had tinnitus situated around the 7 kHz region, such coinciding with my hearing a rather prominent and stable peak through the Jabra Elite 85h, HiFiMan Arya Stealth, and Meze Elite which ramps up from 4 kHz to a culmination at 7 kHz despite what you would expect from the measured frequency responses and the equal loudness contour. My recent hearing test showed a quite even hearing acuity around 5 dB, so I don't think any actual hyperacusis (and thankfully no hearing loss) was detected in that region; I would still love to be able to hear true silence. One theory I have relates to how page 8 of https://www.phon.ucl.ac.uk/courses/spsci/AUDL4007/12.pdf (and https://www.researchgate.net/figure...t-column-and-blocked-ear-canal_fig1_344532774) shows said 7 kHz peak being raised when sound is directly incident from the sides, but I am surprised that this side-incidence HRTF effect doesn't show on HATS measurements, unless my ears somehow exaggerate that peak despite headphones being designed to tame such.

Anyways, the point is that I came to find that while listening to sine sweeps and pink noise, I needed to EQ that 6 kHz to 7 kHz peak down (likewise a 3 kHz prominence and in the case of the Arya, some peaks in the top octave) so that the sine sweep sounded reasonably even if not decreasing in volume. This substantially cleaned up the sound (particularly eliminating unnaturally amplified noise floors) and improved clarity along with some other tweaks like EQing in a full ear gain region and taming the upper bass and the midrange closer to Harman. With that, the opening "explosion" of Mahler 5 can sound reasonably in control while blasted at 95 dB, even through the Jabra. But I am still curious as to whether it could sound yet better, hence my looking into in-ear mics, a Genelec EQ reference (which I can't be blasting at 3 AM), and then the Stax SR-X9000. I simply want "complete peace of mind".

I document my EQing journey in https://www.head-fi.org/threads/mez...official-thread.959445/page-311#post-17549413 (post #4,665).

 

[solderdude]

Hey can you tell what % distortion is audible.

This depends on the frequency and swing of the driver. The latter you can gauge from distortion measurements of lower frequencies. Particularly higher order (3rd and 5th) are indications of compressions. When this happens the amplitude of higher frequencies is also modulated (creates side-bands).
So while LF distortion may not be very audible high SPL bass + music can also affect the rest of the sound.

I am a high spl listener and the many many headphones I have heard sound very bad/harsh/(distorted?) to me when pushed to 90-95db and above.

Have you determined whether 90-95dB is 'average' SPL and how do you know ?
95dBA average with low frequency peaks can easily reach 115-120dB. At these levels most headphones already distort.
You will need planar magnetics, and the less sensitive ones at that (+ a very powerful amp) if you want clean sub-bass + clean mids.

At 70-75db even cheap things sound good to me. What could be the reason?

Compression... drivers simply stop being linear to the applied voltage above a certain SPL. Some already start to compress above 75dB in the bass. Note that this is not the same as 75dBA or music with 75dBA average levels. 75dB SPL at 30Hz = 40 Phon and not is quite soft. Yet the also modulated mids and treble will already be starting to get compromised at the peaks of those bass-notes.
For this reason it is useful to check Amirs measurements at 114dB and we see other measurebators doing the same thing now. Unfortunately I can only measure up to 100dB but even that has shown that many headphones already compress at 100dB and below. People always think... a well 100dB damages ears and is very loud and point at tables that state average SPL (dBA) for long term exposure of noise levels. This is not the same as peak SPL in music, certainly not in the bass.

So yes, you are not crazy when you hear headphones turning 'coarser' sounding at above sensible levels.

The only headphone that I remember not distorting at high spl was the Susvara.

Yep... one of the insensitive planar headphones around and why some folks like this headphone. Plays very loud and above all clean. Screamingly expensive for mere mortals.

Edition XS is really really poor sounding harsh and edgy even at low volume.

This could also be caused by the +7dB treble peaks at 7kHz and 12-14kHz.

It bothered me as well, certainly at louder levels. Can be addressed. It is low distortion and shows no compression up to 95dB.

HD650 sounds refined upto a satisfactory point but push it and it quickly spits its dummy. HD800S was much much better but it too had its limit.

These headphones are fine up to medium loud levels and somewhat loud levels.

Surprisingly Crinacle TruthEar Zero Red was quite refined too but it doesn’t have any midrange to be considered useful.

I have no opinion of IEM's. I don't like the feel, I don't want to shove ear-wax that slowly makes its way to the outer ear being shoved back in and possibly damaging hairs in the ear canal either. Regardless how great they might sound.

 

[numbfreq]

Not that it helps but I have measured a few headphones to below 0.01% using my Gras fixture and Audio Precision analyze without any chamber. So it can br done.

Distortion measurements that low seem to say the headphone itself was providing isolation from ambient at certain frequencies, closed back and/or internally damped. With very open headphones such as estats or planers all bets are off for reliable measurements without isolation from ambient.

 

[DanWiggins]

Hey can you tell what % distortion is audible. I am a high spl listener and the many many headphones I have heard sound very bad/harsh/(distorted?) to me when pushed to 90-95db and above. At 70-75db even cheap things sound good to me. What could be the reason? The only headphone that I remember not distorting at high spl was the Susvara. I didn’t measure its db but I confidently remember it was super loud. That headphone has magical clarity. Is it also born out in THD measurements? Comparatively two other headphones, one that gets a lot of praise Edition XS and one thats a classic HD650 really sound trash to me at any higher than moderate volume. I would stay away from calling these headphones good even or worth buying. Normal speakers are so much better and more enjoyable sounding louder cleanly and in a refined manner. It just doesn’t make any sense. Speakers cannot be always played loud because of disturbance to others, so we turn to headphones but if headphones distort at a perceived loudness lower than even speakers then those headphones are nothing but useless.
Edition XS is really really poor sounding harsh and edgy even at low volume. HD650 sounds refined upto a satisfactory point but push it and it quickly spits its dummy. HD800S was much much better but it too had its limit. Abyss Diana TC was garbage. Surprisingly Crinacle TruthEar Zero Red was quite refined too but it doesn’t have any midrange to be considered useful.

THD above 1% is shown to be audible in controlled situations; above 2-3% and it's audible in almost all situations. When it really starts to sound bad, you're probably at 10% or more.

NOTE: This is for THD only, not the oft-plotted THD+N. Both are useful measurements, but IMHO you need both to get a good assessment of the performance of a headphone, especially an active one.

Fit of over-ear headphones is quite problematic, especially if you have glasses. When we measure over-ear headphones (we do a lot of OEM development), we'll measure in 5 positions (headband rotation about the lateral axis) because people wear their headphones with the headband all the way from "on forehead" to "on top" to "down near the neck", and as you rotate the headband the clamping force changes the way the earpads engage around the ear.

We'll do 3 measurements at each position, removing the headphone all together, setting it down, and having someone else put it on - yes, 3 different people mount the headphone on the ear, in each position. So we can cover "range of use".

And then we repeat with thin frame glasses, and thick frame glasses.

We'll plot the entire dataset for each case (no glasses, thin frame glasses, thick frame glasses) at all positions, the full data set for each individual position, the average of each position, and the average of the entire dataset. So that way you can see what might be construed as "normal" (average), and also see the range of spread of seal based on rotation.

Not surprisingly, non-round earpads tend to have wider spread in these measurements, as they have the greatest variance in seal around the periphery of the earpad and thus are most prone to leaks. Leaks will dramatically change the bass reinforcement you have from your headphone, and can easily blow 10+ dB SPL out from the "ideal" situation of the headphone - and when you "crank it up" to compensate for the loss in output, you're adding 10+ dB more THD versus the sealed situation.

And remember: SPL is Sound PRESSURE Level. It's a measure of the displacement of the transducer (or transducers) relative to the volume of space in which the pressure is received. A large, 90mm diaphragm only has to move 0.1mm to displace 6.4cc of air (a LOT when you think about the volume inside the earpad). A 40mm diaphragm has to move 5 TIMES as far to get the same pressure. And as any transducer guy will tell you - THD correlates pretty well with motion. The more your diaphragm moves, the more distortion tends to increase.

(What about IEMS! Well, for example ours is just 10mm in diameter and is quite linear to 0.2mm excursion, meaning it only displaces 0.16cc of air. But we're energizing just 1.2cc of air, so about an 7.5:1 ratio; that's on-par with a 90mm transducer and a typical 14mm thick over-ear circumaural pad).

Some 40-50mm transducers have zero linear travel, meaning with any measurable movement (as little as 0.1mm on the Klippel) you're already pushing 3-4% or more THD. They're relying upon tens of microns of motion to reach "acceptable" levels. And if you're playing per EN50332 (EU directive about how loud headphones can be), well, 100 dB for Bluetooth headphones is the limit and with a 40mm transducer in a smallish over-ear cavity, 20-30 microns of motion will get you 100 dB SPL. If you want more, though - you're SOL with that solution.

 

[DanWiggins]

Hyperacusis? Whoops, I hope not. There are times if I am mentally exhausted that I feel irritated by ambient noises around me though.

As for Amplifier, I have used multiple and its the same with all.

But like I said the main thing is that some headphones sound better than others. So I am wondering if its indeed distortion/lack of quality. Would it be that Susvara and to some extent the HD800S are designed in a way that suits my ears. If I am reading FR graphs of Crinacle properly the Edition XS has very close matching to Susvara. But they sound nothing alike to me. Its bonkers to pay $6000 for a headphone for sure, even if great like Susvara, but Edition XS isn’t worth a penny either.

It could just be preference. Personally, I hate "flat bass" headphones. When I listen to music - live or via speakers - I also FEEL the midbass/bass. We all do. The chest and nasal cavities add a lot of perception of sound, and that's completely missing with headphones. So if you reproduce just flat, IMHO you're tossing out 6-8 dB of perceived midbass/bass output. I'd rather dial that back in to the frequency response, and end up with something that might not "measure flat" as people want - but sounds/feels "right" to most people.

 

[greatestprateek]

THD above 1% is shown to be audible in controlled situations; above 2-3% and it's audible in almost all situations. When it really starts to sound bad, you're probably at 10% or more.

NOTE: This is for THD only, not the oft-plotted THD+N. Both are useful measurements, but IMHO you need both to get a good assessment of the performance of a headphone, especially an active one.

Fit of over-ear headphones is quite problematic, especially if you have glasses. When we measure over-ear headphones (we do a lot of OEM development), we'll measure in 5 positions (headband rotation about the lateral axis) because people wear their headphones with the headband all the way from "on forehead" to "on top" to "down near the neck", and as you rotate the headband the clamping force changes the way the earpads engage around the ear.

We'll do 3 measurements at each position, removing the headphone all together, setting it down, and having someone else put it on - yes, 3 different people mount the headphone on the ear, in each position. So we can cover "range of use".

And then we repeat with thin frame glasses, and thick frame glasses.

We'll plot the entire dataset for each case (no glasses, thin frame glasses, thick frame glasses) at all positions, the full data set for each individual position, the average of each position, and the average of the entire dataset. So that way you can see what might be construed as "normal" (average), and also see the range of spread of seal based on rotation.

Not surprisingly, non-round earpads tend to have wider spread in these measurements, as they have the greatest variance in seal around the periphery of the earpad and thus are most prone to leaks. Leaks will dramatically change the bass reinforcement you have from your headphone, and can easily blow 10+ dB SPL out from the "ideal" situation of the headphone - and when you "crank it up" to compensate for the loss in output, you're adding 10+ dB more THD versus the sealed situation.

And remember: SPL is Sound PRESSURE Level. It's a measure of the displacement of the transducer (or transducers) relative to the volume of space in which the pressure is received. A large, 90mm diaphragm only has to move 0.1mm to displace 6.4cc of air (a LOT when you think about the volume inside the earpad). A 40mm diaphragm has to move 5 TIMES as far to get the same pressure. And as any transducer guy will tell you - THD correlates pretty well with motion. The more your diaphragm moves, the more distortion tends to increase.

(What about IEMS! Well, for example ours is just 10mm in diameter and is quite linear to 0.2mm excursion, meaning it only displaces 0.16cc of air. But we're energizing just 1.2cc of air, so about an 7.5:1 ratio; that's on-par with a 90mm transducer and a typical 14mm thick over-ear circumaural pad).

Some 40-50mm transducers have zero linear travel, meaning with any measurable movement (as little as 0.1mm on the Klippel) you're already pushing 3-4% or more THD. They're relying upon tens of microns of motion to reach "acceptable" levels. And if you're playing per EN50332 (EU directive about how loud headphones can be), well, 100 dB for Bluetooth headphones is the limit and with a 40mm transducer in a smallish over-ear cavity, 20-30 microns of motion will get you 100 dB SPL. If you want more, though - you're SOL with

Thank you for this comprehensive information!

 

[Mr. Haelscheir]

I had forgotten to share that I had recently been pleased to find that I could measure down to 0.05% THD or less with my in-ear mics: https://www.head-fi.org/threads/mez...eadphone-official-thread.959445/post-17743506. Interestingly, the microphones' self-noise could decrease by 10 dB just from resting a hand on my DAC/amp's chassis. And this was before I had learned that you could decrease the noise floor with longer sweeps. I don't yet have an audio interface suitable for measurement averaging. Does anyone know if the Focusrite Scarlett 2i2 maintains sample synchronization?