• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

AXPONA 2017: JansZen Electrostatic Headphones

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
About the main advantage of electrostatics, while the lack of breakup is the key aspect for you, the lack of HD and IMD are also very noticeable. Amplifier companies make a big deal about low distortion, but most dynamic drivers have orders of magnitude more distortion than most amplifiers, on the order of 0.2% - 1.0% and sometimes much more, whereas a good electrostatic can have THD as low as 0.01%, making low distortion in the signal from the source electronics worthwhile.

I'd be happy to accept this if I had measurements to support it, but I can't find any. And if we look at the measurements I posted above of the NADs, we see THD in the range of 0.03% @ 100dB @ 500Hz. These are not an expensive pair of headphones so I assume this level of THD is not SOTA for dynamic headphone drivers.

Also, your figure of 0.2-1% is correct for loudspeaker dynamic transducers driven to 90-100dB at 1m, but I'm not sure this can be translated to headphone dynamic transducers measured @ a few centimetres.

It would be good to see some detailed measurements of electrostatics that demonstrate all these claims better than what I've been able to find :)

EDIT: it might be more like 0.05%-0.1% on a second look at the graph, with the highest harmonic at about -75dB (about 0.02%)...
 
Last edited:

SIY

Grand Contributor
Technical Expert
Joined
Apr 6, 2018
Messages
10,386
Likes
24,749
Location
Alfred, NY
The artificial ear makes no sense because the pinnae are effectively rendered non-functional with headphones -- the in-your-head sound is proof of that.

Measuring headphones with different pinnae shows enormous frequency response differences with changes in the pinnae shape. So I wouldn't agree with "non-functional."
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,395
Location
Seattle Area
IMO, standard headphone measurements are fatally flawed.

It seems to me that placing a microphone at the bottom of a tube that's supposed to represent the ear canal and surrounding the exit of the tube with an artificial ear is a wrongheaded convention, so to speak, that sells expensive test equipment but makes no sense.
I am with you on this. Even though I bought the gear and read all the papers and looked at other people's work, it is the most worthless set of professional measurements we make in audio! Such incredible variability, no way to compare two sets of measurements performed by others, what it all means for everyone's ear, etc.

It is so bad that at this point I just like to see some anechoic measurements of the transducers than this mess of sticking them on some artificial "ear."
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
Gotta agree with you @ztatic and @amirm about headphone measurements. Even if you wanted to keep something like the Harman curve, you could transpose it to match raw anechoic responses and then use that (or something similar) as your benchmark.
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
Zero experience with the JansZens, but my subjective impression of Stax ESL was that they were ultra clean and detailed. THD doesn't tell the whole tale- I'd be interested in comparisons to dynamics using things like Geddes-Lee metrics, POLQA, ABC-MRT, and multitone, all of which I'm equipped for.

FWIW, the result of my running the Soundstage distortion figures for the NADs through a Gedlee calc using sagemath.com was: 0.097244543757

Geddes and Lee's study found that a Gm of 0.2 was the minimum for reliable detection in the group tested.

This score is based on my eyeballing of the Soundstage graphs, and is valid only at 500Hz (if at all), although their THD sweeps showed THD+N to be very consistent across the audible spectrum apart from the very low bass (it rose to 2% @ 20Hz @ 100dB).

I should restate that I don't think the Gm is the best metric for distortion audibility as it is a pretty crude instrument, and fails to take into account Fletcher-Munson curves, listening level, and various other factors. But it's the only one I have access to a calc for.

Also, of course, measuring headphones, including for harmonic distortion, is fraught for the reasons @ztatic mentioned, so the result should doubly be taken with a large pinch of salt. But I do think it makes plausible the question: do these very affordable dynamic headphones generate any audible distortion at 100dB SPL?

I can post further details of how I ran these calcs for anyone interested.
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
Here are some interesting measurements from a Japanese blog of the $60k Sennheiser HE-1 electrostatics, which do show it to have significantly lower distortion (taken at 94dB and under different conditions, of course) than the $180 NADs.

EDIT: and here are someone's measurements of the previous Sennheiser HE90, which show it to measure slightly better than Sennheiser's higher-end dynamic models in terms of non-linear distortion, and similarly in terms of ringing/resonance (but with very scant info in terms of measurement conditions).
 
Last edited:

ztatic

Member
Technical Expert
Joined
Jul 13, 2018
Messages
23
Likes
15
Very interesting, Andreas. Wow, yeah, -75 dB for any harmonic is indeed inaudible, even when it's one of the nastier ones.

I noticed that the author of the Soundstage article wrote that the pronounced resonances in the NAD's will be inaudible because they're narrow. Actually, the narrower they are, the more they stand out, partly because when the Q is high, the decay time is longer, so when excited, their amplitude builds up above the excitation amplitude and also persists a bit. These peaks are the embodiment of the term peaky.

Hit those phones with pink noise and listen, and I'll bet those peaks will sound nearly like tones, and since 1.8 and 2.9 kHz aren't harmonically related, not the cleanest sounding pair of tones. With music, any time a note has a harmonic at one of those frequencies, and lots of notes will, that harmonic will be exaggerated and persist briefly after the fundamental has died out.

BTW, despite my opinion of standard headphone measurements, I will probably submit mine for just such measurements, probably by Jude at Head-Fi, who's within driving distance and has a very nice rig. On the other hand, I'll make my own comparison measurements on a number of high grade cans using my method.
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
Very interesting, Andreas. Wow, yeah, -75 dB for any harmonic is indeed inaudible, even when it's one of the nastier ones.

I noticed that the author of the Soundstage article wrote that the pronounced resonances in the NAD's will be inaudible because they're narrow. Actually, the narrower they are, the more they stand out, partly because when the Q is high, the decay time is longer, so when excited, their amplitude builds up above the excitation amplitude and also persists a bit. These peaks are the embodiment of the term peaky.

Hit those phones with pink noise and listen, and I'll bet those peaks will sound nearly like tones, and since 1.8 and 2.9 kHz aren't harmonically related, not the cleanest sounding pair of tones. With music, any time a note has a harmonic at one of those frequencies, and lots of notes will, that harmonic will be exaggerated and persist briefly after the fundamental has died out.

BTW, despite my opinion of standard headphone measurements, I will probably submit mine for just such measurements, probably by Jude at Head-Fi, who's within driving distance and has a very nice rig. On the other hand, I'll make my own comparison measurements on a number of high grade cans using my method.

It's great that you're planning to submit yours for measurement :) And I think using a dynamic driver for LF and an electrostatic for HF is a really interesting idea - I'm sure they'll measure and sound extremely good.

I tend to disagree with you about narrow-band resonances, on the basis of Toole and Olive's research at Harman. Their conclusion was actually the opposite. Although their research was based on music, not pink noise.
 
Last edited:

ztatic

Member
Technical Expert
Joined
Jul 13, 2018
Messages
23
Likes
15
It's great that you're planning to submit yours for measurement :) And I think using a dynamic driver for LF and an electrostatic for HF is a really interesting idea - I'm sure they'll measure and sound extremely good.

I tend to disagree with you about narrow-band resonances, on the basis of Toole and Olive's research at Harman. Their conclusion was actually the opposite. Although their research was based on music, not pink noise.

Toole's experiments tended to leave and ignore uncontrolled variables. Perhaps this wasn't true of his detectability studies on high Q resonances, but without being familiar with it, I'd still probably bet against its validity. I'm in particular disagreement with him about the value of a wide "power response" for speakers.

IMO, the consequent dilution of recorded ambiance by local ambiance (wall, floor, and ceiling splash) makes recorded performances sound similar to one another, ambiance-wise, more like they're occurring in the listening room (a they are here experience), rather than in the original venue (you are there).

One's immediate preference for one or the other is almost entirely a matter of what one is used to, and most people are used to wide dispersion speakers in small-ish rooms, thanks to decades of prevalence of dome tweeters and small midranges. His subject cohort was thus inevitably and inextricably biased in favor of wide dispersion, so when tested, of course it turned out that wide dispersion is best. This conclusion happens to serve the interests of companies that make the usual sorts of speakers, and at least one has long engaged heavily in mutual admiration.

For me, this one example is enough to call into question the merit of his work in general. It even leads me to think it might not be a coincidence that a seemingly credible proclamation of the inaudibility of high Q resonances also serves the interests of most speaker makers.
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
I agree @ztatic when it comes to Toole's preference tests, e.g. those resulting in the power response conclusions. Everyone has their own preferences about this, and I think Toole's test really just determine what the most popular preferences are under double blind conditions.

But I haven't noticed any issues in his audibility tests, which are of a fundamentally different nature. Will look into it when I have some time though..

EDIT: at best these tests really just determine what the most popular preferences are under double blind conditions ;)
 

ztatic

Member
Technical Expert
Joined
Jul 13, 2018
Messages
23
Likes
15
I should have mentioned that I find peaks audible with musical source material, myself, although it's easier to pick out with a noise source.
 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,835
Likes
16,497
Location
Monument, CO
For me, this one example is enough to call into question the merit of his work in general. It even leads me to think it might not be a coincidence that a seemingly credible proclamation of the inaudibility of high Q resonances also serves the interests of most speaker makers.

Seems pretty harsh. He posts over on AVS so you might challenge him there (seems like he has posted here but not sure -- too many threads to remember). Most of his conclusions were drawn before he hooked up with Harman. I do not agree with all of his conclusions either, but then again my preferences don't match those of a number of other audiophiles, but I would not question his integrity like that.

For the record and in the interest of full disclosure, I used planar dynamic speakers (Magnepans) in one form or another since 1979, and have had a lot of ESLs roll through my system for short-term tests, so I do like what they do. They have their pros and cons, natch. My current main system is now Revel, however.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,395
Location
Seattle Area
I tend to disagree with you about narrow-band resonances, on the basis of Toole and Olive's research at Harman. Their conclusion was actually the opposite. Although their research was based on music, not pink noise.
The lower detection of low-Q resonances was thought to be due to the fact that they cover more frequencies and as such, have higher probability of being detected when playing music.

The study though is limited to bass frequencies (I think below 500 Hz). So this discussion here is different.
 

ztatic

Member
Technical Expert
Joined
Jul 13, 2018
Messages
23
Likes
15
I should have just said midrange and treble peaks are obviously audible, as pretty much anyone's ears will attest, and it's not sensible to rely on someone's data to disprove what's obvious. There's a reason why the term peaky came to prominence in the early days of microphones, phono cartridges, and speakers. People should certainly use their own ears, rather than rely on data that might not even be applicable, such as this now apparent reference to a study on bass peak inaudibility to refute midrange peak audibility. The Soundstage article is an example of how this stuff can be contagious. It's no help that in audio, people can hear what's not there and miss what is there based on expectation bias.

And by the way, narrow bass peaks are audible, but that audibility depends on whether the music happens to hit the particular note(s) with peaks. We've all noticed this, at least in our cars. Avoidance of those notes can occur only if music is the source material, rather than a sweep or a half-step sequence of tones. I think it's worth noting also that avoidance of certain bass tones can be arranged when music is the source.

Toole collected data that indicated which dispersion convention is prevalent, but represented it as indicating which dispersion convention is better. As far as I know, he hasn't renounced it to this day. Was this misrepresentation or incompetence? I know there's a rule of thumb that says, when trying to discern the basic impetus behind problematic behavior, one should bet on incompetence over intention, but in this case I'd say it's a coin flip.

As I mentioned, there's at least one speaker manufacturer that's long been all-in on using this data to support its design philosophy (not Harmon). If Toole were to change his mind about the meaning of his data, that would be awkward, and of course, many in the speaker industry would be affected, since Toole is so widely cited. So, is a guy who provides positive feedback to the industry in the guise of scientific inquiry a compromised agent, or is he incapable of scientific inquiry and conveniently taken advantage of by the industry? Either way, he might yet stumble upon data that proves something new and interesting, but the inaudibility of high Q peaks is never going to be one of them.

I've only gone into this because I think reality is important. Sharp peaks are audible, and something is wrong when it becomes possible to publish delinals of this fact and have this denial spread beyond the source rather than be summarily ignored. I'm at least partially presenting my POV on the state of the industry in the process. I don't wish to enter into a personal feather tossing session with Toole. As goes his integrity, I have no opinion.
 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,835
Likes
16,497
Location
Monument, CO
I should have just said midrange and treble peaks are obviously audible, as pretty much anyone's ears will attest, and it's not sensible to rely on someone's data to disprove what's obvious. There's a reason why the term peaky came to prominence in the early days of microphones, phono cartridges, and speakers. People should certainly use their own ears, rather than rely on data that might not even be applicable, such as this now apparent reference to a study on bass peak inaudibility to refute midrange peak audibility. The Soundstage article is an example of how this stuff can be contagious. It's no help that in audio, people can hear what's not there and miss what is there based on expectation bias.

And by the way, narrow bass peaks are audible, but that audibility depends on whether the music happens to hit the particular note(s) with peaks. We've all noticed this, at least in our cars. Avoidance of those notes can occur only if music is the source material, rather than a sweep or a half-step sequence of tones. I think it's worth noting also that avoidance of certain bass tones can be arranged when music is the source.

Toole collected data that indicated which dispersion convention is prevalent, but represented it as indicating which dispersion convention is better. As far as I know, he hasn't renounced it to this day. Was this misrepresentation or incompetence? I know there's a rule of thumb that says, when trying to discern the basic impetus behind problematic behavior, one should bet on incompetence over intention, but in this case I'd say it's a coin flip.

As I mentioned, there's at least one speaker manufacturer that's long been all-in on using this data to support its design philosophy (not Harmon). If Toole were to change his mind about the meaning of his data, that would be awkward, and of course, many in the speaker industry would be affected, since Toole is so widely cited. So, is a guy who provides positive feedback to the industry in the guise of scientific inquiry a compromised agent, or is he incapable of scientific inquiry and conveniently taken advantage of by the industry? Either way, he might yet stumble upon data that proves something new and interesting, but the inaudibility of high Q peaks is never going to be one of them.

I've only gone into this because I think reality is important. Sharp peaks are audible, and something is wrong when it becomes possible to publish delinals of this fact and have this denial spread beyond the source rather than be summarily ignored. I'm at least partially presenting my POV on the state of the industry in the process. I don't wish to enter into a personal feather tossing session with Toole. As goes his integrity, I have no opinion.

Agree with the first two paragraphs, not that it matters.

So maybe Dr. Toole didn't lie, he was just incompetent, 50-50 chance in your view... AFAIK his studies were about preference, and what most people in the studies preferred was by (his) definition "best". As I said, I am not in complete agreement with him, and in fact my main listening room is pretty much the antithesis of his philosophy (but I like it). I respect his work and do not have any desire nor need to question his character and methodology as you are.

Harman, not Harmon (typo, I've done the same).

Everyone has their own view of "reality" these days. I do think you are throwing stones, not tossing feathers, but that simply reflects my bias.

For the record, I have only rarely listened to JansZen ESLs, but very much liked their approach and what I heard back in the 80's. It was partly a memory of those sessions and discussions that led me to prefer and generally promote damping the back wave of dipole speakers. I have never listened to the headphones.

I shall try to shut up since this is straying way off-topic. - Don
 
Last edited:

mitchco

Addicted to Fun and Learning
Audio Company
Joined
May 24, 2016
Messages
640
Likes
2,397
I reviewed the NAD HP50's here and took measurements with in-ear binaural mics - I gave up on the dummy head. I agree that trying to measure the high frequency response is variable and I show that in the measurements. However, these headphones did measure remarkably flat from 20 Hz to 7 kHz on my head. I did measure distortion, but can't be sure how much of it are the mics versus the headphones... My article was peer reviewed by Dr. Sean Olive, as I did not want to misrepresent Harman's research on both headphones and loudspeakers. Which to my ears, I agree with.
 

FrantzM

Major Contributor
Forum Donor
Joined
Mar 12, 2016
Messages
4,337
Likes
7,733
I should have just said midrange and treble peaks are obviously audible, as pretty much anyone's ears will attest, and it's not sensible to rely on someone's data to disprove what's obvious. There's a reason why the term peaky came to prominence in the early days of microphones, phono cartridges, and speakers. People should certainly use their own ears, rather than rely on data that might not even be applicable, such as this now apparent reference to a study on bass peak inaudibility to refute midrange peak audibility. The Soundstage article is an example of how this stuff can be contagious. It's no help that in audio, people can hear what's not there and miss what is there based on expectation bias.

And by the way, narrow bass peaks are audible, but that audibility depends on whether the music happens to hit the particular note(s) with peaks. We've all noticed this, at least in our cars. Avoidance of those notes can occur only if music is the source material, rather than a sweep or a half-step sequence of tones. I think it's worth noting also that avoidance of certain bass tones can be arranged when music is the source.

Toole collected data that indicated which dispersion convention is prevalent, but represented it as indicating which dispersion convention is better. As far as I know, he hasn't renounced it to this day. Was this misrepresentation or incompetence? I know there's a rule of thumb that says, when trying to discern the basic impetus behind problematic behavior, one should bet on incompetence over intention, but in this case I'd say it's a coin flip.

As I mentioned, there's at least one speaker manufacturer that's long been all-in on using this data to support its design philosophy (not Harmon). If Toole were to change his mind about the meaning of his data, that would be awkward, and of course, many in the speaker industry would be affected, since Toole is so widely cited. So, is a guy who provides positive feedback to the industry in the guise of scientific inquiry a compromised agent, or is he incapable of scientific inquiry and conveniently taken advantage of by the industry? Either way, he might yet stumble upon data that proves something new and interesting, but the inaudibility of high Q peaks is never going to be one of them.

I've only gone into this because I think reality is important. Sharp peaks are audible, and something is wrong when it becomes possible to publish delinals of this fact and have this denial spread beyond the source rather than be summarily ignored. I'm at least partially presenting my POV on the state of the industry in the process. I don't wish to enter into a personal feather tossing session with Toole. As goes his integrity, I have no opinion.

z

lot of opinions... People are entitled to have some. The Internet allows us to hide behind an avatar and cast stones anyway ... Any studies to back up your views on Toole's incompetence? While we're at it any such studies conducted by you ? Anything that goes beyond mere opinions...?? Please ?
I will wait for your reply before placing you on my "ignore" list.
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
@amirm and @ztatic I'm sorry to nitpick, but I have to point out that Olive and Toole's study, as well as various other similar studies, were conducted across a wide frequency range (not just below 500Hz), and were published when Toole was still at the National Research Council of Canada (i.e. pre-Harman).

The results are summarised as follows:
  • Low-Q resonances...are more easily heard than high-Q resonances producing narrower peaks of similar amplitude.
  • The detectability of resonances decreases approximately 3dB for each doubling of the Q value.
  • In general, pink or white noise are the mot sensitive indicators of these resonances, with speech and music progressively less sensitive.
  • With transient sounds...at a given delay in excess of about 1ms, sounds containing the highest frequencies will be the most audible. In headphone listening the effect is substantial...
The following table is also provided, averaging the findings of Olive and Toole's study and those of previous studies:

1532187104002.png


As the graph shows, audibility thresholds drop as Q decreases and as frequency increases.

Floyd and Toole's methodology is also described in the attached link. I'd be interested to hear if you have criticisms of it.

I should have just said midrange and treble peaks are obviously audible, as pretty much anyone's ears will attest, and it's not sensible to rely on someone's data to disprove what's obvious.

I prefer to trust the results of rigorously conducted double blind studies than my own ears under uncontrolled sighted conditions in these matters.

Of course, I don't take the results of every study to be definitive, but I do acknowledge that the reliability of the outcomes of studies like these is infinitely greater than of my own apparent experience under uncontrolled conditions, or of anecdotal evidence.

Sharp peaks are audible, and something is wrong when it becomes possible to publish delinals of this fact and have this denial spread beyond the source rather than be summarily ignored. I'm at least partially presenting my POV on the state of the industry in the process. I don't wish to enter into a personal feather tossing session with Toole. As goes his integrity, I have no opinion.

I do not see any significant methodological failing in Toole and Olive's study in this case, nor can there said to be evidence of any an ulterior motive influencing their findings. Moreover, the results they obtained are consistent with those of others. But I'd welcome anyone who would like to review the study and point any such failings out.

Finally, I'd like to point out that I never made any statement regarding the audibility of the resonances measured in the NADs. It is worth noting though, that while on one hand these resonances are at high frequencies, on the other hand they are of extremely high Q value. If the table above is any indication, they will be audible only if they are higher in level than about -15dB with noise as the test signal, or louder (not specified) with music as the test signal. The Soundstage measurement shows that their level is about 30 to 40dB below the signal.
 
Last edited:
Top Bottom