Master Complaint Thread About Headphone Measurements

[moosso]

rtings.com has scroes of imaging, passive soundstage.

It's calculated from frequency response and group delay thus I didn't include them, and that formula is quite questionable.

Crosstalk is cable, source and headphone impedance dependent.

Polarity: Is included in the phase measurement or squarewave I measure it and report when deviating from the standard.
Tyll measured polarity (seen in impulse measurement).

Phase response: GD, excess group delay and phase response are not the same thing but do have a relation.

Man you did a great job for providing such uncommon measurements like linearity and leak test. Personnally I think frequency response consistency is as important as frequency response measured on standard rig, the Stealth has almost perfect measured FR but many people doesn't give positive feedback probably due to poor consistency.
Crosstalk is more of a QC issue with the stock cable, I'm not sure acoustical crosstalk is a problem just included it in case. Although some headphone introduce crosstalk on purpose like the CROSSZONE CZ series.

 

[Deleted member 16543]

Do you mean lower in value (so below -30dB) or lower in the sense of -20dB for instance ?

Lower numbers, as in -20 dB or even lower.

Sometimes when measuring headphones and sometimes a contact of say the R channel is not there but L is but the right channel is measuring what only the left channel is doing then, depending on the frequency, there is somewhere in -20 to -50 of signal there but this is just 'mechanical' coupling of the fixture (in my case wood) so measuring any acoustical crosstalk would probably say more about the fixture than the actual crosstalk of a headphone.

Yes. One would need a binaural mic with skin-like material to come close to the real listening scenario.

But before even getting into all this.. can you hear a difference between the two cables in your previous example? I know it's not easy to ABX it, but maybe one could create several tracks with those cross talk numbers added in the DAW, and then ABX the files. Even -24 dB, as bad as it looks, is quite difficult to pick up with our ears.

 

[Robbo99999]

It's calculated from frequency response and group delay thus I didn't include them, and that formula is quite questionable.

Man you did a great job for providing such uncommon measurements like linearity and leak test. Personnally I think frequency response consistency is as important as frequency response measured on standard rig, the Stealth has almost perfect measured FR but many people doesn't give positive feedback probably due to poor consistency.
Crosstalk is more of a QC issue with the stock cable, I'm not sure acoustical crosstalk is a problem just included it in case. Although some headphone introduce crosstalk on purpose like the CROSSZONE CZ series.

I agree there with your point about consistency, that is important. Yep, it's not included in the tests here. I've developed some knowledge re my own various headphones and how consistent they are, as I've measured them on my miniDSP EARS rig and some headphones are very repeatable between reseats on the rig and others are less so, with my NAD HP50 closed back being the least consistent - which makes sense because they're small earcups and are closed back. Open backed headphones are the way to go for consistency, and large earcups.

 

[solderdude]

Lower numbers, as in -20 dB or even lower.

Where would -20dB crosstalk come from in any 'normal' open over-ear headphone ?
Do you have any evidence of this ? Any well known examples with 20dB crosstalk that you know of ? What is the crosstalk path (if not electrical) ?
I have never seen -20dB appear on my test fixture when one channel was not working and this is with a fixture that is coupled pretty tight mechanically.

The next time I measure an open headphone (hopefully OLLO S5X or HD660S2, both should be arriving) I will run a test to see what the spectrum is.

can you hear a difference between the two cables in your previous example?

It was not mine and only had it briefly. Would never buy an AQ headphone myself though and would replace the cables with lower resistance ones.
I did no listening tests but the owner mentioned he could tell differences and it is why I measured the cables. -24dB crosstalk is not that great. Even vinyl wasn't that bad unless the cartridge was not aligned properly.

 

[Deleted member 16543]

Where would -20dB crosstalk come from in any 'normal' open over-ear headphone ?

Sound escaping from the cup, wrapping around the head and entering the other cup. Sound waves doesn't behave like rays. You don't need reflections form a bathroom, like you mentioned earlier, to get acoustical cross talk.

What is the crosstalk path (if not electrical) ?

I have never seen -20dB appear on my test fixture when one channel was not working and this is with a fixture that is coupled pretty tight mechanically.

As I was saying, you would need a binaural microphone with skin-like head material

 

[solderdude]

Sound escaping from the cup, wrapping around the head and entering the other cup. Sound waves doesn't behave like rays. You don't need reflections form a bathroom, like you mentioned earlier, to get acoustical cross talk.

Do you have any measurements showing this actually happens and at the levels you claim ?
A HD800 is quite open. When I play music loud (I would guess 85dB average) on one cup only (cable pulled on the other side) and plug my ear where the sound is coming from, and stuff that cup on the inside so very little comes out of it into my already plugged ear and the sound is clearly coming out on the back.
I literally hear nothing on the side where there is no cup. When plugging that ear or not plugged I hear nothing in that ear.
How can that difference be 20dB ? I would be able to hear sound from that ear (finger or no finger in that ear did not make a difference).

I highly doubt sounds (only treble coming out) wraps around the head and is 20dB softer than a driver firing away in your ear.

 

[Deleted member 16543]

Do you have any measurements showing this actually happens and at the levels you claim ?
A HD800 is quite open. When I play music loud (I would guess 85dB average) on one cup only (cable pulled on the other side) and plug my ear where the sound is coming from, and stuff that cup on the inside so very little comes out of it into my already plugged ear and the sound is clearly coming out on the back.
I literally hear nothing on the side where there is no cup. When plugging that ear or not plugged I hear nothing in that ear.
How can that difference be 20dB ? I would be able to hear sound from that ear (finger or no finger in that ear did not make a difference).

I highly doubt sounds (only treble coming out) wraps around the head and is 20dB softer than a driver firing away in your ear.

The only way to know for sure is by measuring it.
I have done it, but with speakers (still in very near field, though). Comparing the L and R spectrums I was expecting a lot more difference, and I was surprised to see how much cross talk there actually is.
My -20 dB estimate stems from those measurements, accounting for the inverse square rule for a source (cup) that's closer to the ear than my speakers are.
Again, the only way to know more precisely is by measuring it, with the appropriate rig.

 

[solderdude]

A driver that produces 85dB SPL at 2cm distance, in free-air, will produce 59dB at a distance of 40cm.
The biggest part of the sound will be treble will usually is 10-20dB lower in level so 39dB SPL.

I can't hear the left driver playing with my right ear at all even when attenuation on the left ear is -30dB so I think even the 39dB SPL of the treble from the rear of a headphone driver is already very optimistic as the rear of the driver isn't produce the same SPL as the front and there is some extra damping because of the headphone.

Just measured sound from the rear of the HD800 (on a test fixture) with music reaching average level 'peaks' of 97dB and then measured using the same SPL meter at the backside of the HD800 and using the same music got 52dB peaks. So -45dB and only upper mids and treble.

The -20dB or even -30dB is not happening with this simple test.

 

[Deleted member 16543]

A driver that produces 85dB SPL at 2cm distance, in free-air, will produce 59dB at a distance of 40cm.
The biggest part of the sound will be treble will usually is 10-20dB lower in level so 39dB SPL.

I can't hear the left driver playing with my right ear at all even when attenuation on the left ear is -30dB so I think even the 39dB SPL of the treble from the rear of a headphone driver is already very optimistic as the rear of the driver isn't produce the same SPL as the front and there is some extra damping because of the headphone.

Just measured sound from the rear of the HD800 (on a test fixture) with music reaching average level 'peaks' of 97dB and then measured using the same SPL meter at the backside of the HD800 and using the same music got 52dB peaks. So -45dB and only upper mids and treble.

The -20dB or even -30dB is not happening with this simple test.

What you can or cannot hear doesn't make for a proper measurement.
You need to use a proper rig. And you don't have one.
Also.. why 40 cm? The head is ~6 inch wide. That makes it ~9.5 inch (or 24 cm) from ear to ear, over its perimeter.

Anyway, you proposed a worst case scenario of -24 dB electric cross talk. I think a normal case scenario of acoustical cross talk is right in line with that, or even more.
Definitely not lower enough to warrant your "don't worry about it" attitude.
And if we're talking about balanced headphones (with or without a balanced output) we are orders of magnitude in difference. So what's your point?

 

[solderdude]

You need to use a proper rig. And you don't have one.

You don't either and if you do have access please do the measurements.
Well at least I made some crude attempts to get in the ballpark using actual headphones(and not nearfield monitors) and they do not get me anywhere near your -20dB.
Even a poor tuner or misaligned vinyl rig aren't that small and have shown measurements that shows a practical example of a cable crosstalk (which differs from acoustical).

Also.. why 40 cm?

tape measure from left to right cup of a headphone. So an objective measurement. The sound needs to travel this distance.

Anyway, you proposed a worst case scenario of -24 dB electric cross talk. I think a normal case scenario of acoustical cross talk is right in line with that, or even more.

I did my part and made some quick measurements using a calibrated SPL meter (C weighting) in free air and only used my fixture to place the HD800 on. You should not theorize/suspect but show me the actual numbers using a proper rig and show that acoustical crosstalk with open headphones is in that ballpark.

 

[Deleted member 16543]

You don't either

We talked about this already. I'm light years closer to a proper rig than you are..

and if you do have access please do the measurements.

.. and I don't even care about measuring headphones.
If you want to flex your headphones measuring muscles, do what Crinacle just did: buy a 5128

Well at least I made some crude attempts to get in the ballpark using actual headphones(and not nearfield monitors) and they do not get me anywhere near your -20dB.

You didn't measure it properly, sorry. See above.

Even a poor tuner or misaligned vinyl rig aren't that small and have shown measurements that shows a practical example of a cable crosstalk (which differs from acoustical).

Yes. But it's a worst case scenario. Single ended output, high resistance cable, very low impedance headphones.. I call that cherry picking

tape measure from left to right cup of a headphone. So an objective measurement. The sound needs to travel this distance.

Measure it again. 40 cm is too long a distance.

I did my part and made some quick measurements using a calibrated SPL meter (C weighting) in free air and only used my fixture to place the HD800 on. You should not theorize/suspect but show me the actual numbers using a proper rig and show that acoustical crosstalk with open headphones is in that ballpark.

I'm not going to measure it because, as I said, I simply don't care enough.
But just think about this.. if you walk into the same room where somebody is playing music on open headphones and you are wearing open headphones yourself, but not playing anything on them, how close do you have to get for you to hear that they're playing music on them?
Have them play just the right channel and approach them from the left, or vice versa.

 

[solderdude]

Well ... this is a 'Master Complaint Thread About Headphone Measurements' and it seems you have a lot to complain about...

 

[Deleted member 16543]

That's a mature reply.

To think I even edited it..

You comment on my (simple) findings. Make claims but provide no proof,

Not true. I measured the cross talk with speakers (again, in very near field), and accounted for the difference in relative distances to come up with the -20 dB ballpark number. But even if it were -30 dB (which I really doubt it is), it's still high enough to be at least comparable to the (cherry picked) electrical cross talk figure you shared.
Let's put it this way. If acoustical cross talk is negligible, electrical cross talk is even more negligible, under normal circumstances.

which would be easy to do with your vastly superior gear but 'just don't care'.

Then it must not be that easy, don't you think? My binaural mic is buried under a pile of crap since I last moved. Why would you put the onus on me to make a proper measurement? At least attempt to make a decent one first.

State my rig is crap (I know it is and openly admit it) and boast how great yours is.

Didn't state your rig is crap. Not recently, at least. But yes, it is crap.
And not boasting. Just stating a fact, once, when it was integral to the conversation.

The funny thing is I did not even use it to make my point other than use it as a stand.

Which adds to the high inaccuracy of your measurement.

Why should I even consider buying a 5128 ?

Because it's the most accurate way to measure headphones and, among other things, acoustical cross talk.
Your tests are worth nothing, the way you describe them. Sorry, they just are.
Buy a 5128 and measure it properly.
Or do the simple test I suggested and see it for yourself. You can at least do that, can't you?

 

[solderdude]

Neat.... You make a claim and want others to prove it for you.

 

[Deleted member 16543]

Neat.... You make a claim and want others to prove it for you.

No man. Here is how things really went down.
I claimed (or rather implied) that acoustical cross talk is generally speaking higher than electrical cross talk.
I already tested this, as far as I am willing (and currently able) to do.
You don't believe me, and wanted to disprove my claim.
To do so, you concocted a test which proves absolutely nothing. No offense. Sorry to call it as I see it.
I suggested you buy a 5128 to make sure, or try a simple test that requires 2 min to do (obviously, nobody expects you to actually buy a $40k mic. But if you want to do things right it is the only way, unfortunately. Besides, you're the one clogging the internet with meaningless headphone measurements, so I thought you might want to at last get serious about it).
You're not willing to do either of the two.
That's fine. I'll still sleep at night.

 

[solderdude]

Besides, you're the one clogging the internet with meaningless headphone measurements

So there is the pain.... clogging... right... your complaint is noted ... and ignored.

 

[markanini]

It's been a year I haven't commented here.
I created a table about the possible metrics of headphone measurements and how much we obtained, these are based on my understanding which is unprofessional and imprecise.

Metric	To Measurement Rig	To Human
Frequency Response	Many reviewers measured this using GRAS, 5128 and some taobao clone 711 rig, it's the most useful and most publicly available data.	Although rare, some people and research company (Harman) measure this with pin microphones such as SP-TFB.
Frequency Response Consistency	For headphone, solderdude and rtings did the seal broken test. For IEM, I haven't seen anybody measure it (imperfect seal, different angle).	Dr. Sean Olive published a paper shows that open backs headphone has more consistent response than closed backs to real human [1], but he doesn't include the model names. This metric deserve more research.
Distortion	amirm, solderdude, rtings, sonarworks and many third party reviewer measured this, it's not useful as frequency response since most of them below audible threshold.	No one measure and publish it AFAIK. Can a headphone has higher distortion to real human than a rig due to imperfect seal, ear shape, hair?
Phase Response (Group Delay)	amirm, solderdude, rtings measured this. Some people said it's useless since headphone is minimal phase device.	No one measure and publish it AFAIK. Can a headphone has audible phase distortion in real human ear even it's minimal phase to the measurement rig?
Linearity	solderdude measured this partialy (60dB, 70dB, 80dB, 90dB and 97dB). Surprisingly some devices show nonlinear result.	No one measure it AFAIK.
Intermodulation Distortion (IMD)	No third party reviewer measure it AFAIK.	No one measure it AFAIK.
Crosstalk	No third party reviewer measure it AFAIK.	No one measure it AFAIK. Only matter to open backs if the cable isn't faulty.
Polarity	Some reviewer will mention it if the polarity is inverted, not widely tested.	-
Pads And Eartips Difference	Some reviewer measured it when they were interested.	No one measure it AFAIK.
Bone Conduction Effect	No third party reviewer measure it AFAIK. Some measurement rigs support this but they are for hearing aids. I have never seen anyone measured BC earphone or IEM with BC driver (e.g. UM MEST).	Not measurable at this moment.
Body Vibration	No measurement rig support this AFAIK. This is the reason why headphone need more bass than speaker on the graph [2], not sure all headphones generate the same amount of body vibration (e.g. 70mm vs 35mm driver, dynamic vs planar).	Not measurable at this moment.
Base Air Pressure	No third party reviewer measure it AFAIK. This is more an IEM issue, IEM isolates well may have higher base air pressure. Some audio company provides gimmick to reduce it [3].	No one measure it AFAIK.

[1] https://www.audiosciencereview.com/...ooking-for-opinions.41124/page-2#post-1454536
[2] https://www.audiosciencereview.com/...peakers-in-a-treated-room.41407/#post-1463320
[3] https://www.64audio.com/pages/apex

Please share you thoughts, and let me know if there is any metric not in this table.

This site is still active measuring distortion in IEMs. It was quite enlightening comparing figures between much hyped planar driver products and common dynamic driver. One measured as much as 5% distortion in the midrange, guess which one.

네이버 블로그

당신의 모든 기록을 담는 공간

m.blog.naver.com

 

[moosso]

This site is still active measuring distortion in IEMs. It was quite enlightening comparing figures between much hyped planar driver products and common dynamic driver. One measured as much as 5% distortion in the midrange, guess which one.

네이버 블로그

당신의 모든 기록을 담는 공간

m.blog.naver.com

The 7hz one right? It also has pretty high channel imbalance, I guess bad design always exist.

Add a note about acoustical crosstalk, I found headphonetestlab did this test though it's on a dummy head.
https://headphonetestlab.co.uk/test-results-manufacturers-q-t-sennheiser-hd-800-s-full-test
https://headphonetestlab.co.uk/test-results-manufacturers-e-h-hifiman-arya
https://headphonetestlab.co.uk/test-details-how-the-tests-are-performed-acoustical-crosstalk

 

[Deleted member 16543]

The 7hz one right? It also has pretty high channel imbalance, I guess bad design always exist.

Add a note about acoustical crosstalk, I found headphonetestlab did this test though it's on a dummy head.
https://headphonetestlab.co.uk/test-results-manufacturers-q-t-sennheiser-hd-800-s-full-test
https://headphonetestlab.co.uk/test-results-manufacturers-e-h-hifiman-arya
https://headphonetestlab.co.uk/test-details-how-the-tests-are-performed-acoustical-crosstalk

Aside the fact that that is not a proper binaural microphone and measurements are inaccurate (but at least the rig is a step in the right direction), I did expect the acoustical cross talk to be higher than that.
Maybe other open headphones would measure higher.
In any case, if one measured the electrical cross talk, these being high impedance balanced headphones, they would find it is still lower than the acoustical one.

 

[solderdude]

Add a note about acoustical crosstalk, I found headphonetestlab did this test though it's on a dummy head.

Interesting. I have never seen measurements. It was designed to see how sounds are conducted through the headband and the air. The large open area of the hifiman is shure to emit and receive more sound.

I also made an attempt today but different setup as I did not want sound from the construction or test fixture to influence the measurement.
Headphone on the fixture (OLLO S5X which is open but not as open as HD800) and the second time only one channel, headphone on my head only one channel active and then with my head about 10cm away from the fixture and measuring the SPL there.

Of course this is just a 'crude attempt' to get an idea of how much the acoustical crosstalk this particular headphone has.