Electrostatic Headphone Measurements

[Tyll Hertsens]

Hi Bob,

Thanks for your lengthy, and obviously well informed post.

In practical terms, my goal is to provide consumers and enthusiasts with a database of objective data to compare when shopping for headphones. You can find my current measurement set here. A couple of the guys have been playing with CSD plots, and I think they are visualizations of performance that may be useful for laypeople in evaluating cans. This thread is simply us trying to figure out how to possibly include them in the work that I'm doing at InnerFidelity.

I appreciate the more complex studies you highlight, but it is likely beyond my ability to pull off given I've also got review articles to write. The measurements you suggest may also be more complex than non-engineers can usefully interpret.

So, to some extent, I have to look at this adventure as simply getting a good basic set of measurements for folks to compare, and leave the complexities for some time later.

Still, talking about this stuff is good fun, and your post is interesting. Thanks for that, and feel free to stick around and join the fun. I'll caution you, however, that Head-Case is not your normal headphone forum. We all know each other quite well here. This place is really more a club house for a group of freinds that happen to have a common interest, than a headphone board in the normal sense. Your technical contributions are likely to be quite welcome, but it is your potential as a freind that's more important in the long run. I simply ask that you please be sensative to the fact you are entering a dialog where people know and love each other well.

Thanks mate, welcome.

Tyll

PS Sorry about your gear, dude. That sucks.

[spritzer]

BTW

Sorry about the rambling post but my english skills as well as abstract mathematics skills have abanded me these days. My first electrostatic - Micro Seiki (Stax in disguise), bought new circa 1975. From memory( 25+ years), mine seem to be different to the one sometimes shown on the internet. The diaphrapms (which eventually failed) were not tensioned until the cell was assembled. There was a ridge on the stator molding which pressed against the diaphragm locating it centrally in the gap while at the same time tensioning it, when dissassembled the diaphragms were quite loose. I haven't seem this technique mentioned elsewhere although it may have been something to do with the diaphragm material which did not appear to be mylar - it seemed more elastic. From memory, the bias voltages were also quite different from what you would expect, maybe someone can shed some light on this. Ohh yes, the Micro Seikis sounded quite good as long as you were a bass atheist, even then I has hooked on ESs (and Heil AMTs).

The Micro Seiki were designed with input from Stax but weren't OEM units like the Marantz, Realistic, Magnavox etc. etc. AFAIK there were two versions of the M-S sets, the MX-1 set which was an electrostatic with a 9V battery to generate the bias and the MX-5 which was an electret setup. The diaphragm material in the MX-1 was indeed not mylar but the plastic material shopping bags are made out of which will break down with time. 50 year old mylar is just as tight as the day it was stretched and heat treated if it has never been exposed to direct sunlight.

[arnaud]

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?

I am a bit curious as to what you are trying to achieve with these measurements. Is this an attempt to correlate possible audible effects to measured effects, i.e. resonances to colouration?

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.

You guys are going to quite a lot of effort to get together a large collection of ES phones but have you worked out exactly what tests you are going to perform and how you are going to do it? This is not criticism, I am genuinely curious and quite excited by the prospect.

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.

One of these would be to exploit the reciprocity of the electrostatic transducer and use it both as speaker and microphone.

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...).

One of the problems with this method is that it gives best results (from a designers point of view) when the transducer is radiating into an empty room - well at least one member of this forum seems to like pulling his expensive ES headphones apart. >;-)

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?

Other tests worth considering are the distortion tests, especially a swept Intermodular distortion test (IMD is sometimes referred to in loudspeakers as Doppler distortion)

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?

This test can often reveal unexpected diaphragm behaviour like the modal patterns shown by Streng in his papers (something often ignored by ES designers or their marketing departments - I notice that Wiki still implies but does not state the fallacy of pure planar behaviour).

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?

(many modes have little net output as the positive nodes and negative nodes tend to cancel each other out [see Olsen, Berenak etc] and may not show up very prominently on CSD plots).

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?

Unfortunately, I am not in a position to help. I am currently traveling around Australia in the a motorhome (year 7) and have none of my library on ESs with me.

Too bad as you seem to have a lot of interesting papers to share .

I should have brought them with me as a disaster last year has seen them and a great deal of other HiFi gear (which I had in storage) completely destroyed - some of it was practically irreplaceable - older, collectible stuff - often impressive to look at, some of it quite beautiful, and the odd bit here and there occasionally even sounded good.

What a shame... You seem to take it rather easy though, I would have gone probably nuts

There are copies of PDF copies of Borwick 3rd ed floating around on the internet so Peter Baxandall's work on ES is available but this seems to have changed considerably from what I remember of the 2nd ed. I have not been able to source the Walker, Streng, etal papers off the internet. The Streng ones originally published in the Philips Technical Review are most interesting as they led to an experimental segmented ES loudspeaker that was shaped (from memory) as an isosceles triangle.

Thank you for the hints, I will start from there...

[arnaud]

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 .

Hi Bob,

Thanks for your lengthy, and obviously well informed post.

In practical terms, my goal is to provide consumers and enthusiasts with a database of objective data to compare when shopping for headphones. You can find my current measurement set here. A couple of the guys have been playing with CSD plots, and I think they are visualizations of performance that may be useful for laypeople in evaluating cans. This thread is simply us trying to figure out how to possibly include them in the work that I'm doing at InnerFidelity.

I appreciate the more complex studies you highlight, but it is likely beyond my ability to pull off given I've also got review articles to write. The measurements you suggest may also be more complex than non-engineers can usefully interpret.

So, to some extent, I have to look at this adventure as simply getting a good basic set of measurements for folks to compare, and leave the complexities for some time later.

Still, talking about this stuff is good fun, and your post is interesting. Thanks for that, and feel free to stick around and join the fun. I'll caution you, however, that Head-Case is not your normal headphone forum. We all know each other quite well here. This place is really more a club house for a group of freinds that happen to have a common interest, than a headphone board in the normal sense. Your technical contributions are likely to be quite welcome, but it is your potential as a freind that's more important in the long run. I simply ask that you please be sensative to the fact you are entering a dialog where people know and love each other well.

Thanks mate, welcome.

Tyll

PS Sorry about your gear, dude. That sucks.

[purrin]

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?

I realized I didn't fully answer this aspect on the other site. IMD can very be significant because it's easily heard compared to harmonic distortion. If you look at the dual-tone (200Hz + 2200Hz) non-linear distortion graphs, see you will two side bands at 2000Hz and 2400Hz. For ESP950 and HD800, these bands are down 50+db. For the HP1000, it's only down a little more than 40db. This is still definitely hearable because 2000Hz and 2400Hz have nothing to do with the original intended signals (200Hz or 2200Hz) - or more precisely, they are not (low) integer multiples. (EDIT: 2000Hz and 2400Hz would be the 10th and 12th harmonics of 200Hz, but there's very little if anything contributing to that from the 200Hz test tone. You get the idea.)

Anyways, lowering IMD is one of the advantages for going to multiple drivers for speakers.

Edited September 19, 2011 by purrin

[Tyll Hertsens]

Hope this pleases the Stax Mafia: http://www.innerfide...ne-measurements

For the geeks out there, I've made the spreadsheet available. SR-009 and SR-007

[mikeymad]

thanks Tyll and thanks for the spreadsheets.

[ujamerstand]

Listening tests will be done with composer's music who's last name end in "i" or contain "cc" only.

Oh Tyll, you killed me.

[arnaud]

Hope this pleases the Stax Mafia: http://www.innerfide...ne-measurements For the geeks out there, I've made the spreadsheet available. SR-009 and SR-007

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:

Edited September 20, 2011 by arnaud

[Dreadhead]

Neat stuff. Tyll do you have a measurement spreadsheet for the HD800s? I'm going to be off for a few weeks and I may have some time for some math fun and if I have HD800 data from your setup I do some IRF inversion fun and see if I can model the SR009. Not that it's fun to many but since I spend most of my time wrestling with a giant CF software mess some elegant programming is fun at times.

[purrin]

Hope this pleases the Stax Mafia: http://www.innerfide...ne-measurements

For the geeks out there, I've made the spreadsheet available. SR-009 and SR-007

Awesome. The measurements look nice indeed. THD in bass is impressive.

[arnaud]

Tyll, I'd be interested in the hd800 data too...

Dreadhead: what kind of simulation? What is irf and cf?

Neat stuff. Tyll do you have a measurement spreadsheet for the HD800s? I'm going to be off for a few weeks and I may have some time for some math fun and if I have HD800 data from your setup I do some IRF inversion fun and see if I can model the SR009. Not that it's fun to many but since I spend most of my time wrestling with a giant CF software mess some elegant programming is fun at times.

[gurubhai]

Thank you Tyll, for finally providing some much needed measurements on the electrostats.

[padam]

Was it SR007 Mk1 or Mk2?

[Dreadhead]

IRF = impulse response function

CF = cluster fuck

I did a thread about the math a while back. I am on mobile and can't find the link.

[spritzer]

The big question is, how does Tyll like the 'stats.

Was it SR007 Mk1 or Mk2?

It's a late model Mk1.

[Tyll Hertsens]

Setting them up right now for my first listen ....

Also, I'm gonna remeasure the HD800 today and put the spreadsheet link here for you guys to toy with.

Edited September 21, 2011 by Tyll Hertsens

[arnaud]

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?)

IRF = impulse response function

CF = cluster fuck

I did a thread about the math a while back. I am on mobile and can't find the link.

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?
  

Edited September 24, 2011 by arnaud

[Robbo1802]

Hi all,

Thanks for the welcome.

Hopefully I will not make a fool of myself here or upset too many people. I don't normally use forums anymore - too much lost time on Usenet and Fidonet before it.

Now to the first point, I unequivocally revoke all mention of the term Doppler distortion as I was not really thinking when I added that bit, fortunately no has one called me on it yet.

Hi Tyll,

I followed your link, a most impressive set of tests. I can appreciate the work involved in both the testing and writing of reviews. I understand your desire to keep the testing confined to the most useful measurements. Nevertheless, the ex-engineer in me (yep, you picked it) can't help but wish for you to test every possible parameter while all those phones are available. >>;-)>

I downloaded the manual for your test equipment and did a quick scan, did you get the IMD option with it? I also noticed that the gear has a multitone test which also has some interesting possibilities. Being an ex-engineer I NEVER read the manual but just start playing with the knobs. Reading the manual for your gear makes my head hurt.

Using the impedance measuring test is not that difficult but it does require a small modification to the ES amplifier to allow the measurement of the output voltage and current across a small series resistor in the output leads as well as a switch to turn off the diaphragm supply. I did a crude version of this to assess the impulse response of DIY ES phones I made 20 odd years ago - these were initially based on the Pollock design but started to diverge after a while. Ahhh, the plans - like most plans, they were never fully realised.

Hi Spritzer,

They were the MX-1s. I have started to look at some of the photos you posted of your disassembled phones. I was especially intrigued by the photos of the "super rare SR-Omega driver ", do you have a closeup of the stator mesh? Is there a site that fully explores the design and construction of the various Stax headphone variations? It has been a while since I looked at ES speakers and headphones, there is just so much more stuff available on the internet than even a few years ago.

Hi Arnaud,

The blocked impedance is a little more complex (inductances and resistances) but you are essentially correct. This technique of assessing loudspeaker performance is very established and I am sure you could find a good account of it on the internet. Nevertheless I will see if I can find a good reference about it. The technique is especially good with ES speakers given their very high reciprocity.

The testing of the ES cells in free space is most useful for determining the sources of various resonances (as well as other problems) and assessing the results of various damping strategies. This information is useful in the design phase and of course is also useful for developing modifications.

As to the IMD part, this got a little confused in my original post. I will thus refer to harmonic distortion (THD) and IMD with all non-harmonic distortion (this includes pure IMD, as well a form of FM distortion caused by the movement of the diaphragm that creates sidebands - this distortion occurs in even perfectly linear transducers, and some low order AM distortions) lumped under the name IMD . If there is any non-linearity in the system, it will produce both harmonic and intermodular distortion, there is no escaping it - the question is: how does this distortion affect the sound? More on this later.

Unfortunately the diaphragms do not display the pure planar behaviour that we intuitively believe should occur. Streng proved it mathematically using Green's function (Memory?) and then validated the proof with measurements on a test speaker - I must really find a copy of his 4 papers. From memory the modal breakup was somewhat different to the rigid piston model and was entirely of the annular ring form, I vagely remember that the annular nodes actually swept across the diaphragm as the excitation frequency changed. The breakup resulted in localised 'movement of air’ parallel to the diaphragm from the 'in phase' part of the diaphragm to the 'out of phase' part of the diaphragm. One of the outcomes of the work was that a relatively small amount of damping material, close to the diaphragm, 'disrupted the air flow' and damped out the modal breakup - this may have consequences for people who like to remove damping in their ES headphones or who build ES speakers without damping.

Now, if the diaphragm does breakup, the deflection will increase, the inherant non-linearity will increase, the distortion will increase, thus my reference to using IMD (and the not mentioned - THD) to indicate possible membrane resonance. Here is a possible avenue to explore, if a CSD shows a problem resonance and a distortion plot shows a peak in distorion at the same frequency, does this indicate that there is a high probability that the resonance is from the diaphragm rather than elsewhere in the headphone - just a thought, could be wrong.

As to the loss of all that old gear, I am sad that some rare equipment is now even more rare, but there is a silver lining, my collection of broken (very expensive) phono cartridges is now gone. Every time I saw the box they were in I just got angry - that box was the reason I celebrated the introduction of the CD. I suppose the reason it doesn't drive me mad is that I don't really (not in my heart) believe that it is actually gone.

I will start looking for those papers in different places. I do have a lot of PDF books on HiFi and acoustics but not the ones I want. One of Streng's papers looked at charge distributions on the diaphragms, if memory serves, this paper also had a sting in the tail - must find Streng!

Hi Purrin,

I had a look at your distortion plots of the ESP950s. They are most interesting.

First, a look at the harmonic distortion measurements. Fortunately you show the spectral distribution of the distortion rather than just the THD. I find it interesting that the second harmonic distortion is higher than the third. The theory of push-pull ES consistently shows that the even harmonics should be very low, much lower than the third (actually, theoretically zero). There is a caveat, this cancellation of the even harmonics requires a high diaphragm resistance (the usual quoted value is 10exp9 per square or above, preferably 10exp11 per square). So either there is a problem with the testing <sorry>, Koss has got the diaphragm resistance wrong, or there is some other factor in play that I don't understand.

Tyll, Would it be possible for you to produce the plots for the different harmonic orders, at least for one channel and level?

Similarly, I believe the IMD is higher than it should be although I have no theory or data to back this up. I wonder if there are some IMD results for other ESs out there

Now, as to the importance of IMD to the sound. IMD has always been my favourite boogyman. I believe the fundamental problems I had in finding good phono cartridges was their poor IMD performance. <rant mode on> The cartridges I liked the most were always the fringe ones, okay I did like the occasional moving coil but I lived in a sub-tropical climate and all those cartridge manufacturers lived in cold climates, bloody suspensions packed it in on every one except the Ortofon (which I disliked intensely) within a year. <rant mode off>. I believe that the lack of clarity I hear in so much HiFi gear is caused by the hash imparted on the signal from IMD, there is plenty of it in recordings as well. I have proved this to myself many times by comparing live music to music played back via several rather expensive speakers. A live grand piano is just so sparse compared to the sound of the same one played back through most speakers - the speakers just add so much hash. I was in a position to test this many times (friends with large houses, grand pianos, Schoeps microphones, etc), although I only tested speakers, not headphones. Best result (for lack of hash) was Quad ESL57s but they had their own problems especially with midrange colouration and a soft top end. Horns were also good at low hash but again coloured, we could never get Maggies to sound right - must have been doing something wrong.

Oops, another long rambling post – things aren’t getting better.

Regards to all,

Bob

BTW

Just saw the Stax measurements – wow – much to think about

[edstrelow]

The big question is, how does Tyll like the 'stats.

It's a late model Mk1.

Theer is a link on the Inner Fidelity page to an "007 Mk2", which I thought was the same as an 007A.

http://www.stax.co.j...-brochure-s.pdf

Possibly Tyll can clarify.

There is a rather striking peak at the 9/10 kHz, which sounds like what I hear on my 007A. I am not sure if the Mk1 has similar peak since I haven't listened to it long enough.

[Tyll Hertsens]

Frankly, I'm a n00b re:Stax. I just linked to the closest thing I could find on the Stax site. I'll be happy to change the lnk if there's a better place to go.

[spritzer]

Hi Spritzer,

They were the MX-1s. I have started to look at some of the photos you posted of your disassembled phones. I was especially intrigued by the photos of the "super rare SR-Omega driver ", do you have a closeup of the stator mesh? Is there a site that fully explores the design and construction of the various Stax headphone variations? It has been a while since I looked at ES speakers and headphones, there is just so much more stuff available on the internet than even a few years ago.

No closeup that I could find and the drivers are packed away in storage. The stators resemble a chain link fence so the copper is woven together.

There is the unofficial Stax site in Japan but they rarely pull anything apart for diagnoses. It takes somebody as insane as me to rip this stuff apart...

Frankly, I'm a n00b re:Stax. I just linked to the closest thing I could find on the Stax site. I'll be happy to change the lnk if there's a better place to go.

Nothing wrong with linking to the nearest alternative but Stax has changed the internal design quite a bit. So the lineage is like this:

SR-007 Mk1 early version (different driver frame plus some minor material changes, older style carbon fiber box),

SR-007 Mk1 late version (the version tested here and the most common of the Mk1's, S/N either 7xxxx or SZ1-xxxx),

SR-007A/SR-007Mk2 (S/N SZ2-xxxx, Stax change the earpads, how they sit and introduce a port to the earcups. Sound is a mixed bag depending on fit but can be fixed to sound pretty much like the Mk1)

SR-007A/SR-007Mk2 (S/N SZ3-xxxx, often called the Mk2.5, Stax supposedly change the diaphragm which results in a much more forward sound signature)

Now you should point out that these are indeed SR-007 Mk1's just for the sake of clarity.

While we are on the subject, does anybody want to play "spot the difference" between the earliest and the latest SR-007 drivers?

[Tyll Hertsens]

So does everyone think the data is good to go on the Stax? I'm a bit troubled by the bump at 60Hz .... don't think it's a measurement glitch. Anybody have any ideas what's going on there?

Just measured the HD 800 again ... man, do they measure well. Here's the spreadsheet, have fun!

[nnotis]

Wow, this is absolutely the best headphone thread I've ever seen. You guys are the shit. Arnaud, your left/right averaged graphs are hugely helpful for looking at this data. I'd love to see four of them all using the same scale (to make comparisons visually intuitive.) You just did the SR-009 & SR-007. Can you post one with Tyll's new HD-800 data too? I don't recall if you ever made one for the LCD-2s either.

Lastly, did anyone send Tyll 4070s? I'd be really curious to see how those enclosures effect the measurements.

[limp]

Arnaud posted some HD800 plots on page 3 of this thread. Mighty impressive stuff.

I know it's as close as one can get to a sacriledge in this hobby to even consider this, but do the Stax measurements really live up to the expectations?

Edited September 21, 2011 by limp