Electrostatic Headphone Measurements

[arnaud]

I am not sure about the various waterfall graphs which have been posted here either, in part since I assume they must also be based on some sort of impulse signal which would presumably be subject to the same artifact noted above. I am have having some problem following some of this discussion as to what is an actual measured response, what is derived from existing data and what is some kind of hypthetical simulation.

I have now updated the recent CSD and FRF plots by applying Tyll's equalization referred as "Independent of Direction Equilization Parameters" in his spreadsheets. It doesn't seem to be a diffuse field equalization, maybe Tyll could clarify.

SR007 vs SR009:

Comparisons with HD800, LCD2r2:

PS: the normalization is a little "rough" because the frequencies for which I have the equalization values are not exactly matching those for which the CSDs are computed...

PPS: Ed, just to clarify the method: these CSDs are all using the same test data which is an impulse response measurement for 1 headphone position. I haven't compared my FRFs (i.e. the t=0 on the CSD graph) with those of Tyll but will shortly and report...

Edited September 24, 2011 by arnaud

[spritzer]

I think the SZ3 likes the dBs too. Any progress, oh busy one, on trying to modify SZ2s and SZ3s towards the 007? Might a thread be created once the quest is done? My 3s are still dead stock.

No progress I'm afraid. The SZ2's just need to have the arc's bent for a proper fit but the SZ3's are tricky...

[arnaud]

I am not sure about the various waterfall graphs which have been posted here either, in part since I assume they must also be based on some sort of impulse signal which would presumably be subject to the same artifact noted above. I am have having some problem following some of this discussion as to what is an actual measured response, what is derived from existing data and what is some kind of hypthetical simulation.

The Frequency response data seems interesting and uncontroversial. But I would mention some additional points of testing or discussion which it would be nice to see clarified..

First whether the measuring set-up is properly compensating for pinna and ear canal effects so as to measure "flatness." comparable to measurements of speakers. It is my understanding that the system does ini fact attempt to do this but exactly how I am not sure.

Secondly there is the issue of repeatability. To what extent you are getting positioning effects such that the measured response could change with a slight tweaking of position of the phones i.e. if the same phone is measured at different times is the response going to look the same.

Here is a confirmation that the FFTs I am doing in Excel are correct and that impulse response data I am using is in the range of the other measurements done by Tyll to generate the "spaced averaged (5 locations) and equalized (compensated for dummy head response) magnitude response plots:

Edited September 24, 2011 by arnaud

[Tyll Hertsens]

First whether the measuring set-up is properly compensating for pinna and ear canal effects so as to measure "flatness." comparable to measurements of speakers. It is my understanding that the system does ini fact attempt to do this but exactly how I am not sure.

Ideally, IMO, the best thing would be to have a compensation cirve that is derived from sound sources coming from 30 degrees off axis to either side --- modeling speakers --- but this compensation curve is not available. FF (free feild compensation curves are available, but are made for sources directly in fron of the head in an anechoic chamber, so are inappropriate. Diffuse field (DF) compensations are available, but are made with a source that is sound coming from all directions at once. Independant of direction (ID) compensation (which is what I use) is a compensation that is the average of sound coming from every direction. In other words, you have a sound source and move it into a position, take a measurement, then move it to another position, take a measurement, and so on until you've taken measurements at every position and then average them all. It's not the best, but it's better than FF or DF, and it's a legitimatly recognised standard so I use it.

Secondly there is the issue of repeatability. To what extent you are getting positioning effects such that the measured response could change with a slight tweaking of position of the phones i.e. if the same phone is measured at different times is the response going to look the same.

Arno coverd this, but I'll add that even though I'm careful, and measurements are spatially averaged over five positions, there's still easily noticable differences if I measure the same headphones twice. Nonethless, the measurements are consistant enough for their purpose of lay persons being able to compare basic performance measures.

There is the question of how repeatable the measurements are across other examples of the same model or whether there may not be individual differences between examples of the same model. One simple example could be that changes in stiffness and compression of ear pads could give different results. Then there is the possibility that the manufacturing process for phones does not give exactly the same result every time.

It's likely that variations in manufacture are smaller than the errors of repeatability. It's also pretty hard to get say 10 samples of the same headphone for a study of manufacture variences. Changes in earpad wear might indeed be big enough to measure; it's my experience that ear pads make a big difference.

On the subjective side, there is little doubt that a reasonably flat frequency response is desireable in a headphone simply because this indicates that the phone is neutral. However when you get up to the high frequency range you probably want some roll-off in part because there is a natural high frequency roll-off with distance from a sound source and most recordings are made with microphones closer to the source than one would normally hear. Plus there seems to be an awful lot of high-frequency equalizing boosting in recordings, especially rock-pop. It looks like most of the phones are in fact giving some high frequency roll-off.

10kHz @ 50% humidity attenuates 4dB over 30 meters, while 100Hz has essentially none; over 3 meters which is a more realistic listening room distance the difference is inconsiquential. So there is high frequency roll-off, but it's quite a small number. My understanding is that most of the percieved HF roll-off in listening rooms is from higher absorption of the highs in furnature or carpet in the first reflection and reverberant sound.

[arnaud]

Hey Tyll, not sure about your night vision skills but I have a hard time reading bright yellow over white lol

[arnaud]

10kHz @ 50% humidity attenuates 4dB over 30 meters, while 100Hz has essentially none; over 3 meters which is a more realistic listening room distance the difference is inconsiquential. So there is high frequency roll-off, but it's quite a small number. My understanding is that most of the percieved HF roll-off in listening rooms is from higher absorption of the highs in furnature or carpet in the first reflection and reverberant sound.

I guess Ed was referring to natural attenuation of HF with distance in a 100+ m concert hall in case the mics used for the recording are placed near the orchestra pit. Nevertheless, I agree with you the effective absorption in treated concert halls has much more to do with the walls / audience than actual attenuation in the air...

It's likely that variations in manufacture are smaller than the errors of repeatability. It's also pretty hard to get say 10 samples of the same headphone for a study of manufacture variences. Changes in earpad wear might indeed be big enough to measure; it's my experience that ear pads make a big difference.

This is an interesting topic and it's actually the foundation for the simulation work I do most days (Statistical Energy Analysis). Fundamentally, any systems that's made of a buildup of parts, each with its own complicated dynamics is bound to have large variability across a population of nominally identical samples WHEN you get into so called "high frequency regime" (e.g. where parts or all of the system has many modes, kind of like the diaphragm pictures in my 009 model).

It is so bad that, even if you were to crunch numbers with a supercomputer and try to model the response of the system with so-called deterministic tools (like the FEM simulations I showed in this thread) where each part is assumed perfect shape / perfectly connected to the next and what not, you won't be able to match the response of the actual system within even 20dB accuracy.... Simply because just measuring the system 100 times will give you 100 different curves with such kind of 40dB spread across the ensemble at any given frequency. The source of this is basically the very high sensitivity of the system to small perturbations (temperature, small mass, small change in shape, you name it) as the frequency increase (each local mode changes ever so slightly to make the transmission of energy vary from the previous time).

To illustrate the problem, here's a typical example for the SPL measured inside an SUV when a unit impact force is applied to one for the front shock towers. Top graph is 1 car with measurement performed 12 times, bottom graph is 1 measurement on 100 cars taken out of the assembly lines (export more variations with use). The bottom line is that in such scenario, it actually makes sense to focus on what is actually predictable, which is the overall energy of each region (like the roof, the windshield, the interior acoustic cabin, the engine bay...) and the energy transfer coefficients between regions. This is the world of SEA (Statistical Energy Analysis):

Now, what does that got to do with headphones? Well, it's a complex build up structure too so it is the subject to just the same physics... But, it's all very small scale so even 20kHz is basically "low frequency" behavior (not so many modes in the cavities, no modes in the structure, not so complicated joints between the parts...). The only exception is the diaphragm (zillions of modes...) but as mentioned by Bob earlier in the thread, pretty much only the first "pumping" mode seems to contribute efficiently to the response and it is only sensitive to the tensioning (and diaphragm thickness/properties but I assume this is well under QC). I guess, in the case of Stax, it means tensioning and diaphragm properties + ear pad shape are the main source of variability across samples. As Tyll pointed out, there are probably more repeatability issues with the measurement itself (e.g. how much clamping on the phone and actual position on the dummy).

[Robbo1802]

Hello arnaud

The only exception is the diaphragm (zillions of modes...) but as mentioned by Bob earlier in the thread, pretty much only the first "pumping" mode seems to contribute efficiently to the response and it is only sensitive to the tensioning (and diaphragm thickness/properties but I assume this is well under QC). I guess, in the case of Stax, it means tensioning and diaphragm properties + ear pad shape are the main source of variability across samples. As Tyll pointed out, there are probably more repeatability issues with the measurement itself (e.g. how much clamping on the phone and actual position on the dummy).

Hmmm,well I can't comment on modern Stax, my Lambdas ( as well as my SRXMk3 pros) came from the old Stax. In the Lambda case 5 of us got together and did a bulk deal OS. We had to wait and wait. All the phones were different most subtly but one pair was to our minds quite different but still resonably balanced. At the time we bought the Lambdas, it was said by people (who seemed to know) that the major problems Stax was having was low yield and the difficulty of matching the drivers well enough for a stereo pair. Not long after this the old Stax ceased to exist. I don't know but I suspect that part of the high price may still be tied to this problem - specifically diaphragms.

I note that there seems to be a similar issue with at least some of the planar magnetic types.

I have 4 sets of original 70s era ESS AMTs (yes they survived, like some other parts of my gear they were on loan) and even when the same type of diaphagm was used, they varied slightly which is amazing to my mind considering the principle of operation and construction.

Tyll,

Speaking of AMTs, after this electrostatic testing is done you will have covered about every type of transducer short of AMTs. I know the field is rather thin ... ahhh 1... but have you thought of trying to get a pair? Testing floats might be a bit of a challenge especially when it comes to interpretation but....

arnaud

Ohhh, I didn't know you did that sort of work (he says, figuratively inching away from the keyboard - hows the medication working out - sorry couldn't resist).

Regards,

Bob

[arnaud]

...it was said by people (who seemed to know) that the major problems Stax was having was low yield and the difficulty of matching the drivers well enough for a stereo pair.

...

arnaud

Ohhh, I didn't know you did that sort of work (he says, figuratively inching away from the keyboard - hows the medication working out - sorry couldn't resist).

Regards,

Bob

Hi Bob, there is an issue of low yield on the 009, but I reckon it's for the stator, not the diaphragm.

Sorry, I didn't figure out the last part of your post, you'll have to be less subtle for me to get it

[Nevod]

Different side to side diaphragm loading (or damping) should not create asymmetrical diaphragm movement and thus 2nd order harmonic distortion. The diaphragm 'sees' the load on both sides as the total load, and will respond to the total load ( again from both sides) independent of the direction of travel. Now if the behaviour of the damping material was asymmetrical, that is, there was more resistance to the air moving in one direction than the other, then it would produce asymmetrical movement of the diaphragm and thus even order harmonic distortion.

Ah, yes, forgot that the driver has not only to push air on one side, but pull it on the other. Well, that makes everything much simpler.

Yes the LCD-2s do seem to show the best measured performance, certainly the distortion plots are very good with an outstanding CSD. Indeed the CSD shows far less stored energy in the midrange than any others I have seen shown here, the bass is astonishing, okay there are a few small ridges but....

The slope on the square waves is mostly a function of the now airtight the ear cup is, the flat part of a square wave is after all DC - consider a square wave of 1hz, the air will leak out.

Considering the above moment about damping, this doesn't create a need for compromise. Thick, soft, well-sealing earpads - well, LCD's have it. Although I've always thought, why does no manufacturer line the insides of leather earpads with thick felt? That should reduce some reflections at least.

The stored energy in the mids moment is quite interesting. I've read that Sony once used an irregular pentagonal diaphragm in their electrostatic (or electret?) headphones, to reduce resonances on the diaphragm. Maybe, Audeze has used something similar.

It's been said there that no overshoot on the square wave may be due to either very good high frequency responce, or due to some overdamping. And looking on similar HF rollofs on all the high-end phones, I tend to think that this is indeed damping. Plus the fact that, while having not so different responces, the electrostats are still said to be much more detailed than magnetic planars arises some questions that I'll probably start a thread for.

When you say dip are you referring to the absolute 'phase' of the headphones, that is, a positive voltage spike creates a positive pressure spike? I will not get into the debate over absolute 'phase' except to ask: are you sure that Tyll's test rig has correct absolute 'phase' and doesn't invert the signal, perhaps it is LCD-2s that are inverted?

I've got this soon after I've written the post. Inversion doesn't bother me, it's simple to invert once more.

[Tyll Hertsens]

Holy Crap! That took some doing:

http://www.innerfidelity.com/content/comparing-world-class-headphones

[TMoney]

I'm honestly shocked how much my impressions of the headphones you used in the review matched yours, Tyll. I've never owned or spent extensive time with the HiFiMan gear so I won't comment on those, but I agree with you on almost everything else.

Lambda/Omega 1/O2 mk2 owners, time to get this man moar gears to measure!

Edited September 25, 2011 by TMoney

[screaming oranges]

Tyll, incredible work! That's probably the 1st review of that caliber ever. Talk about bringing out the artillery.... geez!

[HDen]

Great article Mr.Tyll, by the way, you have to love some of the (few) comments that have been posted

[Hopstretch]

Tyll, incredible work! That's probably the 1st review of that caliber ever. Talk about bringing out the artillery.... geez!

Echo this. Brilliant stuff, Tyll.

[Tyll Hertsens]

Great article Mr.Tyll, by the way, you have to love some of the (few) comments that have been posted

Fuck, I laughed and laughed at the first one: "LIKE A BOSS"

I get about 4000 YouTube vies a day now and have hundreds of subscribers (YouTube is a stunningly good PR tool). You've got to think most folks there will be dumbfounded by the collection of gear.

Fuck Dre.

[Dreadhead]

Nice job Tyll. I need to give a more detailed run through again but it sounds like I'm not going to get better than my GS-X/HD800 setup for what I like without getting into KGSSHV+009s which I'm just not going to do.

[deepak]

Very well written Tyll, oh lord I am going to end up buying the SR-009 aren't I

Curious what you thought on the soundstage depth of the O2 vs the LCD-2? IMO I felt the LCD-2 presents a flat stage with little depth, the opposite of the O2 with it's "ventriloquism" like effect.

[complin]

Fuck Dre.

What Dre.... Great stuff ay.. wouldn't consider anything else, world class (crap)

[screaming oranges]

One question: is this the end of the Stat measurement project?

I was hoping other stat cans would make it to Tyll. There's still plenty of time left, after all. Anyone else confirmed that other cans will be sent?

[ujamerstand]

I've sent some sr-404 and sr-507 to Tyll yesterday. Didn't send the sr-lambda because the BHSE doesn't have a normal bias output.

Would be nice if somebody sends in a pair of baby orpheus.

Edited September 25, 2011 by ujamerstand

[Currawong]

I don't listen at insane levels either but I do think the SR-007 needs to be pushed a bit to truly excel. Not to insane levels mind you but this "MOAR POWAH!!!" crap has some truth to it. When the SR-Omega (and to a lesser degree the SR-009) start to misbehave the SR-007 still just takes it all and asks for more. If any of you are brave enough, listen to the intro of Megadeth - Trust at silly levels and report back. The SR-009 doesn't like it one bit but the SR-007 is just fine with the BHSE maxed out... Still, I only try this for academic reasons.

I think that listening volume is overlooked very often, more so than the kind of music being listened to (which will affect the perceived frequency response of the headphones and al that), so this is a very good point IMO. I listen the quietest with the 009s at about 70dB average at last check, even with music that one might otherwise assume would be listened to loud, so this is definitely another factor worth exploring.

Holy Crap! That took some doing:

http://www.innerfide...lass-headphones

When I first started reading, I didn't think a point-by-point analysis was going to work, but the way you've done it I think is excellent and gives a very clear picture of what you heard, to me at least.

[K3cT]

Holy Crap! That took some doing:

http://www.innerfide...lass-headphones

Tyll, thanks for the very magnificent comparison bullet notes! I love this format, makes it easier to pinpoint on the differences between the headphones.

Reading that it makes me re-affirm my belief that I should stick with the LCD2 until I can afford the SR-009 in the far future. I just graduated from college last fall so I really should focus on paying for my car for now.

I'm surprised that you really like the HE500. I've heard it and it does sound pleasant and sweet but I have a problem with its bass 'bloom' which causes the bass notes to have a longer-than-normal decay and I'm sure this looseness causes the HE500 to have a less-than-optimal PRaT too which can be a problem with fast and complicated pieces.

[spritzer]

I think that listening volume is overlooked very often, more so than the kind of music being listened to (which will affect the perceived frequency response of the headphones and al that), so this is a very good point IMO. I listen the quietest with the 009s at about 70dB average at last check, even with music that one might otherwise assume would be listened to loud, so this is definitely another factor worth exploring.

It is indeed a huge factor and often overlooked. Even with superbly recorded stuff (SACD rip of Aja) they get quite a bit edgy once the volume is pushed much past 70dB.

[arnaud]

Interesting point on the volume. I think I'm at around 65dB in average so not what you'd call loud. I crank it up at times and crank it down too with the SR009 playing edgy recordings.

[arnaud]

You, sir, are insane .... insanely cool! Great stuff!

Tyll, I posted some new results in the simulation thread over at HF ( http://www.head-fi.org/t/572968/modeling-headphone-s-acoustic-performance#post_7782482 ) with some comparison against your test data. Not quite there yet with the model it seems