arnaud

I agree that there is no such thing as a perfect impulse response as it equals a headphone with unlimited bandwith. In the present case, the microphones may be rolling-off before the headphone and remember that all the measurement chain is lumped on the result... As for using too much damping, it's probably a bit difficult to answer when it is or not appropriate. I guess we all agree that sharp peaks in the magnitude response (and subsequent long ridges in the CSD plot) are not too desirable. On the other hand, maybe extreme damping on the surfaces of the earcup might actually sound unnatural, like giving too much of a feeling of a direct sound coming from the transducer (I hate to have the sensation that the driver is very near my ears, which does occur with some headphones). Another typical trick is actually using the resonances to enhances the tone (e.g. John Grado's headphones which certainly have a fanbase). The graphs Tyll shows - at least the FRFs - include results that are both averaged (across 5 or so headphone positions, averaging is done in dB directly so it smooths things out heavily) as well as compensated for the dummy head response. I am not clear yet at what angle of incidence (it's not diffuse field and referred as "Independent of Direction Equilization Parameters", maybe Tyll can clarify). As for the erratic behavior at high frequency, this is inherent to the physics in place. In particular, imagine that the acoustic wavelength (=340 / frequency) is ~1.5cm at 20kHz... That means that the sound pressure inside the earcup - if it was undamped - would have 40dB swings when you move the microphone of barely 8mm (half a wavelength at 20kHz). Similarly, just pushing the earpads by say 3mm would have also a huge effect on the results. To illustrate the point, I have updated the simulation results (I said I would not talk about this in this thread anymore but you're asking for explanation and there's nothing better than an image here with SPL from microphones placed ~1cm apart. In the animation plots, you can see it's fairly homgoneous SPL in the base of the earcup (mics location) at 3.5kHz but starts to go a little more wild in the 6kHz one. I can't do these simulations easily up to 20kHz but you'd essentially get even more complex pressure field. Original post: FRF: Animation 3.5kHz: http://www.youtube.com/watch?v=_cV_0vsZX38 Animation 6kHz: http://www.youtube.com/watch?v=HrcnUQ-DXmQ

@Marv: I wouldn't mind using your raw data for my model... It's with anechoic plate, isn't it? @Ed: Here's a reason why I shouldn't have posted my simulation result in this thread. I will create a new thread and post there (did the same on HF http://www.head-fi.org/t/572968/modeling-headphone-s-acoustic-performance#post_7774290 , it'd be better to have it all in 1 place though). There are no assumptions in the processing of test data. Only thing is that my CSDs and FRF comparisons are for a single position and not equalized (yet... Expect some update today on this). For the data acquisition, there is maybe something going on with artificial smoothing to get the signal to die within 3ms but this still needs to be investigated (difficult to compare directly the results with those of Marv due to entirely different recording process).

I must be the exception to the rule then. Kind of itching though with the strength of the yen atm...

Thank you, insanely tired too, those 4 hour nights are starting to be felt . Hi Bob, the diaphragm is modeled as a circular membrane and edges are assumed pinned. In regards to the animation, actually I should have mentionned: it is NOT to scale! It is just an animation of one of the "coupled" modes around 3.7kHz. Velocity range is actually visible for the 2 peaks where I show the contour plot. But again, properties I used for the membrane are bogus so really the diaphragm velocity isn't too trustable... BTW, I assumed 10% damping for the diaphragm and not actually modeling the perforated stators. Good question dreadhead... The acoustic boundary condition is simply rho *c (acoustic impedance in free field). But I was indeed concerned about how large the box should be around the headphone to get reasonable results when I did the same kind of simulation for the HD800. I have been looking at another type of simulation (using boundary element method referred above) where the headphone would actually be modeled along with the whole radiation space but not quite satisfied yet with the HD800 results. I may try it with the SR009 geometry...

Using the 'Headphone Data' sheet, column AH for the time vector, columns AK and AL for the impulse responses of Left/Right ears respectively. Let me know, I could certainly have goofed of somewhere.

Edit: 9/24, different mic positions and cleaner pics / animations Here is an initial attempt at understanding what makes the SR-009 measure how it does. It's all very very fresh and largely guessed properties so I could be totally off the map with these predictions, but at least I have some pretty pictures and animations to share . Here's the simplified geometry of the model (note that the earpads are currently modeled as rigid surfaces. Since there's a thick leather cover, it shouldn't that far off (little absorption): Here is the resulting SPL at some locations ~1cm above the "ear surface" (currently assumed to be a rigid surface, I may be able to add actual dummy head geometry later on). The microphones at above 1cm apart so you can see typical variations as the headphone is moved around: Then, some illustration of the pressure response around the headphone at the 3.6kHz resonance as well as 6.6kHz where you see lobes in some of the microphones response: Finally, an animation of coupled structural/acoustic dynamics at the the same frequencies. Note that the diaphragm material is bogus as I have absolutely no idea of the tensioning and resulting natural frequencies. I just used a 1.5 micro meter nylon with increased stiffness to simulate the tension and adjusted that to get the first diaphragm "piston" mode at about 50Hz. This turns into over a thousand wiggling modes in the diaphragm by 10kHz...: http://www.youtube.com/watch?v=_cV_0vsZX38 http://www.youtube.com/watch?v=HrcnUQ-DXmQ

Thanks for looking these up Bob, I now remember I also have a reference on loudspeaker design (Van Dickason or some similar name) which may have some info. Will look these up. Green's function is used to solve boundary integral problems so indeed this may have been used to simulate the radiation of baffled plate. The "acoustic near field" is defined as ka<<1 with a the source dimension and k the acoustic wavenumber (2 pi / acoustic wavelength). So, actually, near and far field is a function of frequency. Another thing is that the paper you reference most like looked at a baffled source radiating in free field which again is very different from an small ear cup. But, anyhow, I have been proved to be very wrong several times when guessing about headphone behavior so best is to simply simulate it, and I have the luxury to do that. The software I use has a so called "Boundary Element Method" formulation which is essentially making use of acoustic's green function for solving a boundary integral problem (radiation of surface / diffraction in an acoustics domain), albeit numerically not analytically (so any shape can be solved not just simplified ones). It turns out it's much more efficient to use traditional Finite Element Method though but I am investigating both ways at the moment ... Humm, not totally relevant discussion for this thread, sooner or later hell is gonna fall on us arnaud

Hi Rob, Thank you for your response. I need to think more about this IMD stuff, I guess my work skews my vision as I am systematically modeling linear systems . In regards to diaphragm "break-up" modes, it's actually Bessel functions of first kind in the case of circular membranes ( http://en.wikipedia.org/wiki/Vibrations_of_a_circular_drum ). Green's function is a completely different wolf and I won't get to it (partly because I barely understand it lol). Probably break-up naming is not appropriate here because there's no rigid body motion of the diaphragm in the case of tensioned membrane vibration (as opposed to the piston motion of an electrodynamic cone before so-called breakup modes catch on). As for wave cancellation for higher order modes, again it's not that simple when you are looking at headphones because these modes actually radiate energy in the near acoustic field (it just happens to not propagate efficiently to the far field due to the +/- cancellation effect). Anyhow, I am about to create a simplified numerical model of the SR-009 to investigate this as well as the effect of spatially distributed forcing on the diaphragm and possibly to estimate acoustic resonances in the ear-cup. If only someone had some measurement of only the electrode to estimate the natural frequencies of the tensioned diaphragm ......

Not sure it explains everything but there is quite a bit of variability with the fit and - as far as I understand Tyll's methodology - the decay graphs come from a single position measurement (MLS test) of the impulse response. See below a reprocess of Tyll's 009 FRF data:

I'm confused, the shape of the frame tells me your left pic is the 007A, yet the gold color of the electrode tells me the right pic is the 007A. Here's a pic of my 007A (late SZ2):

Edit 9/24: CSD and FRF graphs are now equalized to compensate for the dummy head response (not sure for which heading, does not appear to be diffuse field, Tyll?) Hehe, I agree that interpreting these results appropriately is no easy feast... The data varies also widely across test setups, maybe because I am processing non-equalized data. As Purrin said earlier, it really only makes sense to compare curves from the same test setup like the most recent 007/009 comparisons and the following... @nnotis: I have Audeze LC2r2 data from Tyll as well so I could generate the comparisons. It gets a bit crowded when comparing all 4 together though... All 4 headphones: HD800 with SR009: HD800 with LCD2r2:

Edit 9/24: CSD and FRF graphs are now equalized to compensate for the dummy head response (not sure for which heading, does not appear to be diffuse field, Tyll?) Lol, I guessed the F but not the C . BTW, the impulse response is already in the spreadsheets (MLS test) and that's what I am using to generate the CSD. Tyll, maybe you like these a bit more? Results until now were 1/12 Octave Band averages but I now feel these narrow band results are more informative to really guage the sharpness of the ridges and just simply nicer looking (smoother) toward higher frequencies as we keep the same resolution. Only drawback is that I can't make the frequency axis as a log scale... You'll notice the Inner Fidelity background . These results are now the average impulse response of Left and Right ears Finally, I added a comparison of the basic FRF but normalized by the highest SPL which is around 3-4kHz for both 007 and 009. Since this happens to be where our hearing is most sensitivite (well as long as one's ears are not shot ...), I assume this is more representative of what people will feel when comparing both headphones at equal loudness impression. Nonsense?

Tyll, I'd be interested in the hd800 data too... Dreadhead: what kind of simulation? What is irf and cf?

Hi Tyll, brilliant! The geek in me could not resist and generated the graphs (I love this Excel automation stuff ). Did not check in detail, but first look is that the 009 does have some resonances in the midrange (1-4kHz) but they're better damped than the 007 peak at 8-9kHz:

The man himself posted before I hit the post button . Forgive me if I did not introduce your work as nicely as you did. BTW, I'd still be happy to help if you get a chance to do some measurement with longer time window. Looks like you're very busy listening to the 009 at the moment though, which is understandable .

Dear Bob, Welcome and thank you for your informative post, I would like to read up on the reference you posted. You're referring to people who published at some previous AES conference or? In regards to CSD, I guess the one and only point is to get an idea of how damped the various resonances are, with the idea at any sharp / long ridge in the time axis should be quite audible as a coloration. I am not sure how familiar you are with Tyll Hertsen's recent venture and he will explain better for himself but the idea is simply to reproduce the same bench tests as he does for all the headphones he gets his hands on. As I know, these are "standard" direct dummy head measurements of headphone response with various input signals / post-processing. CSD would be a new result if Tyll can get to it. This is interesting. In the case of ES driver, isn't the blocked impedance just the capacitance of the driver and resistance of the cable? I agree that due to strong electro-acoustic coupling, any acoustic resonances / diaphragm resonances will show up on the electrical impedance curves but it sounds like the measurement of blocked impedance is only practical for an ES driver? I like the idea of extracting a radiation impedance from these two measurements though, would like to read more on this. Note that the prime objective of the current tests is simply to get the SPL at the entrance of ear canal for a "standard" dummy head, possibly normalize it based on the heads calibration HRTFs, and use the result as part of a database to compare different headphones (including electro-dynamics and other ortho types...). You're referring to ES speakers with unwanted room resonances here no? I don't see how headphone response without the proper acoustic load (e.g. dummy or someone's head) can be of any use if just to see some diaphragm resonances? This is interesting too. I believe I replied in the CSD thread I did not understand the purpose of IMD test for headphones (or passive speakers for that matter) because I can't imagine how these distortions can be significant in passive devices? You're referring to "modal patterns". While I understand you could excite higher order modes at a given frequency, these are still responding at the driving frequency, not their natural frequency. So, there won't be any IMD to speak of, where am I getting mislead here? Do you mean here that those thin tensioned diaphragms have loads of "wobbling" modes and responding nowhere near uniformly? This I can't imagine at this time since the ES force is spatially distributed over the surface and I assume (?) inertial forces of the diaphragm are minimal compared to this ES force? Again, this is true for speakers but here's we are measuring in the acoustic near field of the drivers, which is a very different business. While it's true that all these higher order modes don't propagate in the far field (just sloshing air around), aren't they going to be picked up by the nearby microphone in dummy head test? Too bad as you seem to have a lot of interesting papers to share . What a shame... You seem to take it rather easy though, I would have gone probably nuts Thank you for the hints, I will start from there...

Similar thing for me. I was mesmerized at the beginning since I was using select recordings. Then was the phase of playing random tunes in my library and trouble started. IMO, there's no free lunch and both Omega 2 and 009 are basically complementary for this type of issue. I went back to doing headphone simulation as it had worked pretty well for the HD800 (to identity the 6kHz peak for instance). I'd like to give it a shot with the 009 and see if the new shape of the enclosure has any bearing with the new forward acoustic signature. But per your message spritzer, you feel it's more of an effect of the newer generation diaphragms (507, O2 SZ3)?

ouch, that hurts.

Sounds good Tyll. BTW, I know I am not "supposed to" but I paste her a discussion with Purrin's from the HF thread on CSD measurements, in case you're not visiting there. Bottom line is that, after thinking about Marv's setup, I believe there's very little chance for your and his CSD results to have much resemblance: ------------------------------------------------------------------------------------------------ Marv, I reckon you started doing these tests on regular dummy head data and then found out you got cleaner results with your current anechoic termination test rig. While I thought this was a great idea at first, I am not sure of the meaning after thinking a bit about it: There does not seem to be a fundamental issue with using a standard dummy head (see my CSD post-processing of Tyll's data) Headphones are (for the most part, grant you the K-1000 is a different beast) designed to feel an acoustic load forward of the driver (e.g. the acoustic chamber between the driver the ear due to earpad spacing). I see several major issues with the anechoic termination test: As you pointed out, the low frequency part of your CSD plot has little to do with actual performance. (that's regardless of open or closed design, being in the acoustic near field of the driver, e.g. it's almost like doing an IEM test without a proper seal) The anechoic termination completely changes the acoustic resonances of the enclosure (from a sort of rigid boundary condition to anechoic one, acoustic resonances will shift down in frequency VERY significantly, by a factor of 2 for example). This will be seen up to very high frequency because such acoustic resonances occur above 2kHz. The anechoic termination adds significant acoustic absorption in the front chamber (ear to driver, bounded by the earpad) so the acoustic resonances of this region will not only be shifted in frequencies but also significantly damped so you pretty much won't see them in the CSD plot. Have you thought about going back to standard dummy head measurement now that you got more experience with CSD testing? I understand you might want to isolated the driver resonances from the acoustic chamber but in this case, extracting the driver (like you did with the SR80?) or better yet, doing a laser vibrometer measurement on the diaphragm (can you get your hands on that ? ) would be suitable. ------------------------------------------------------------------------------------------------

That explains... I always wondered how much the typical issues with 007A's bass overbloom were coming from improper seal. At least for me, I felt the proper seal made very significant difference in how clean / tight the bass sounds - regardless of using a non-linear amplifier . Good news with the 009 is that I don't experience such sensitivity to fit and it basically fits like a glove...

I thought 007A was the naming for domestic market (mkII overseas)? In any case, I have no issue sending mine as another point of reference (late SZ2 series, SZ2-16xx). Shouldn't cost more than 50USD to ship to, problem is shipping back...

Yeah, this is a reminder of the worst years of Microsoft when they used to cram all sorts of options in some sub-menus. It's getting much better recently though... As for the user manual at hand, there is one place I found which may be a potential culprit (see exponential settling in page 431 as part of the sweep testing settings). Maybe it's set to perform some exp. windowing in the last ms or so? Tyll, I will need you to clarify how you are exporting / generating the MLS result as this interface is an utter confusing mess (I now understand why you had to run away with your bike for a week lol ). My (poor) understanding of how this thing is setup goes as follows: The MLS sequence (output of your front end, feeding the headphone amp) is generated by a DSP as opposed to traditional analog signals so you must assign it a D/A sample rate and it must match that of the signal analyzer (you're using 65kHz) The microphone data is recorded back and can by viewed as is in analog domain (going through a 10Hz-22kHz bandpass filter though) However for MLS processing, the system actually needs to perform FFTs so it's digitizing the analog inputs, in which case you must set the sample rate to match that of the MLS signal generation (not exactly sure why, but I guess it's to guarantee that the MLS sequence does correspond to the acquisition window (as it's not quite fully random signal). In your case, you don't only want to see the result on the screen but also record the data so it must be digitized regardless of the post-processing or not. For some reason, you're then having to setup a time sweep recording in order to perform the MLS acquisition. In that panel, you're actually overwriting all previous sampling settings above as you're specifying the time step (=1/sample rate) and start / stop time (acquisition period = 1/ frequency resolution of the resulting fft). From trying to decipher the manual, what's happening in the background is that the thing is actually decimating the data and sampling at much higher rate than you're specifying in the various menus. As a result of this mumbo-jumbo, it's coming up with a 3ms impulse response with an apparent 170kHz sample rate. The only recommendation I have after reading the manual is no different than the first one - you should try to modify the sweep panel settings which really is the only thing that controls the estimated system impulse response from MLS: Keep the MLS sequence to 28k length, 65kHz D/A Keep the digital analyzer to 65kHz A/D Get at the very least 6ms of data in the time sweep panel, which is sufficient to see the ringing of most headphones (this will give a resolution of 165Hz (1/0.006) but this can circumvented by zero padding the data). I would go with 12ms (we can always zero out room reflections later if need be). You could probably reduce the number of steps as the current settings are a bit overkill (apparent 170kHz sample rate, I was using 1 in 5 samples to generate the CSD graphs). But since we're better safe than sorry, I'd say go with 2047 steps for 12ms which will be the same rate as currently (for a fast headphone like HD800, it might be nice to have that kind of bandwidth). BTW, my current spreadsheet reads 2048 samples. I doubt there will be issues with reflections from the walls of the test cell, you're using pretty thick liner I believe? In any case, as mentioned above these reflections can be removed from the impulse response afterwards (except if they're swamped in the middle of headphone reflections, in which case you gotta accept this or move to larger test cell). hope this will help you...

Wow, I've seen a few data acquisition front ends in my life, but this one is a winner in the confusion department!! Looks like the person behind the GUI was not a transpose from Apple's human interface team, lol . Maybe it's because it a full featured analyzer which caters to all kinds of crowds, but still allowing the user to input different sample rates in various parts of the GUI, while at the same time specifying in the user guide that one should not do that is pretty crazy! End of rant, going back to trying to decipher this but I am a feeling this is beyond my limited IQ...

I just started looking at the manual but clearly, some things are fishy just from the graph you posted: - The standard measurement specifies a 160 micro sec. delay (for subtracting from the phase response) and you see in the impulse response that the signal ramps up at around that time - Yet your 2nd graph at higher sample rate with 128k sequence shows an impulse response starting at 750micro seconds or so. - Yet again, the resulting impulse responses in the spreadsheets your sent are virtually identical and effectively ramp up at 160 micro seconds

Hi Tyll, have a nice ride . In regards to the " there's not much going on out there signalize", there's more than meets the eye I'm afraid, especially if you're referring to the lin scale impulse response. We want to see at least 30dB worth of decay and Purrin's data has shown that many headphones take more than 4ms to decay by that amount. I am not sure yet why I don't see it with your data but I suspect one of these menus forces the data to zero toward the end of the time window (exponential decay), even though it doesn't say so. Surely, could do with more comparisons, please send me if you have!