Massdrop X KOSS ESP/95X Electrostatic Headphone Review

[Jose Hidalgo]

 

[IAtaman]

Sorry your religious cult is under attack by heretics and the best you can do is to meme vaguely about it. I wish I could've spared you this pain and suffering.

Could you please ease on the tantrum you are throwing because some people on internet don't want to accept your beloved headphones are great.

If I understand you correctly, you are claiming estat headphones must be charged for a while to sound at their best. I don't know anything about that and am willing to learn. If you have anything more than a drop.com forum post about it, please do share - I would like to understand better. If it is you and your buddies' experiences only, not interested.

Please also note that there are people on this forum who has probably forgotten more about audio, audio electronics, estats etc than you will every know, so a bit of humility is highly recommended as well.

 

[donjoe0]

If I understand you correctly, you are claiming estat headphones must be charged for a while to sound at their best. I don't know anything about that and am willing to learn. If you have anything more than a drop.com forum post about it, please do share

In order to have "more" I would first have to have at least those first quotes, but now that they've been deleted I'm back to square one and the discussion will seem to come out of nowhere and be motivated by almost nothing instead of being motivated by comments from experienced people with years if not decades of usage of electrostatic headphones/speakers.

But anyway, I've found something more concrete about one of Nottagorilla's quotes on Stax designs putting megaohm-scale resistances in series with their headphone diaphragms:

electrostatic drivers have extremely high impedance ratings typically above 100 kΩ (100,000 Ω) [...] This seemingly extreme impedance is required to keep the electric charge from escaping the driver.

Complete Guide To Electrostatic Headphones (With Examples)

Master the technology of electrostatic headphones: Comprehensive insights into their design, sound clarity, and examples from leading brands.

mynewmicrophone.com

I assume this is pure resistance we're talking about, so the effect has to be symmetrical, so by however much that 100k/1Mohm resistance is slowing down the leakage of the membrane charge once accumulated, by the same amount it should slow down the initial accumulation itself. Now I just need more numeric data on how much slowing down is going on here, or what the total capacity of the membrane is, like in Coulombs or something. If there's any technical reason to say estats need to charge for tens of hours, it has to be tied to this.

Please also note that there are people on this forum who has probably forgotten more about audio, audio electronics, estats etc than you will every know, so a bit of humility is highly recommended as well.

Of course there are, I've seen them discuss technical topics in several places on here since I joined, and there is no sign of them in these last few exchanges on this topic. It's almost insulting to real engineers and scientists to compare them to this tantrum-throwing friend we had in here swinging back and forth between flamebaiting and insisting he didn't care to continue the discusison.

 

[Jose Hidalgo]

Since you are talking about me, then know that I just happen to be an engineer from one of France's top schools, which happens to be the equivalent of MIT.
Is that "real engineer" enough for you? Do I need to justify my studies and my knowledge just to please you?
And BTW, when you talk about somebody behind his back, at least have the courage to name him.

Like I said, I just don't have time to waste with the audiophile tantrum YOU are trowing from the start (as said by IAtaman here).

Could you please ease on the tantrum you are throwing because some people on internet don't want to accept your beloved headphones are great.

And yes, I still get notifications about this thread. So in the future, think twice about what you say and how you say it. Or else nobody @ ASR will give a damn about your posts.
Back to real life now.

 

[donjoe0]

Hah! Just as I suspected, capacitors don't charge linearly but asymptotically:

Remember that, at any finite t, Q is less than its asymptotic value CV
[...]
Upon integrating Equation 5.19.2, we obtain
Q=CV(1−e^−t/(RC)) (5.19.3)

Thus the charge on the capacitor asymptotically approaches its final value CV, reaching 63% (1 - e^-1) of the final value in time RC

5.19: Charging a Capacitor Through a Resistor

phys.libretexts.org

So since the estat membrane charges like a capacitor, we can likewise expect it to take some very large amount of time to reach exactly-full charge (math analysis says infinite time, but since physics is quantized we only have to wait for the "last electron" to get out in order to reach that 100% positive charge on the membrane and determine that the process is over).

So it turns out there should be exactly zero surprise to hear someone claim that it takes some huge amount of time to fully charge an estat membrane - that's just textbook capacitor physics. The only real question is what is the last charge increment that can still make an audible difference during estat operation (or how long do you need to charge until there is no more audible benefit left to gain).

 

[IAtaman]

Hah! Just as I suspected, capacitors don't charge linearly but asymptotically:

5.19: Charging a Capacitor Through a Resistor

phys.libretexts.org

So since the estat membrane charges like a capacitor, we can likewise expect it to take some very large amount of time to reach exactly-full charge (math analysis says infinite time, but since physics is quantized we only have to wait for the "last electron" to get out in order to reach that 100% positive charge on the membrane and determine that the process is over).

So it turns out there should be exactly zero surprise to hear someone claim that it takes some huge amount of time to fully charge an estat membrane - that's just textbook capacitor physics. The only real question is what is the last charge increment that can still make an audible difference during estat operation (or how long do you need to charge until there is no more audible benefit left to gain).

Time constant of an RC circuit can be calculated simply by multiplying capacitance with resistance. I think you said resistance was in the range of 100K ohm. If you know the capacitance as wel, we can easily calculate how long it might take for it to reach any arbitrary voltage level compared to max voltage.

 

[donjoe0]

If you know the capacitance as wel, we can easily calculate how long it might take for it to reach any arbitrary voltage level compared to max voltage.

Well yea, that's what it boils down to. Not sure they publish diaphragm capacitance though. I can find a published spec of 110 pF including cable for the Stax SR-L300, but if typical cable capacitance is 40-60 pF (if it's nothing crazy long) it doesn't make sense that the diaphragm with its comparatively huge surface area would only add another 50-ish to that. I think they're only counting the stators in that, since those are the ones in the actual audio signal path. I don't know that we can easily infer the diaphragm capacitance from the stator capacitance.

 

[IAtaman]

Well yea, that's what it boils down to. Not sure they publish diaphragm capacitance though. I can find a published spec of 110 pF including cable for the Stax SR-L300, but if typical cable capacitance is 40-60 pF (if it's nothing crazy long) it doesn't make sense that the diaphragm with its comparatively huge surface area would only add another 50-ish to that. I think they're only counting the stators in that, since those are the ones in the actual audio signal path. I don't know that we can easily infer the diaphragm capacitance from the stator capacitance.

Well, 100nF capacitance would have a time constant of about 10ms for 100K ohm resistance. At 100ms, the system would have reached +99% of its voltage. If you go down to pico range, it would be even shorter. If the assumptions and calculations are correct, it looks like it is very unlikely for longer charging to make any discernible difference.

 

[donjoe0]

If the assumptions and calculations are correct, it looks like it is very unlikely for longer charging to make any discernible difference.

Which assumptions? That published value is likely not the membrane capacitance since it's unrealistically low for such a large surface (it's comparable to the inter-wire capacitance in the cable itself - ridiculous).

Remember I quoted 110 pF, not nF. Assuming only 60 pF of that is the driver, and with Nottagorilla's hint about Stax typically using a series R>=1Mohm, we get RC=60μs and then 99% charge time = 5*RC = 300 μs. That looks like it could only be done 3.33k times/s and would only allow the reproduction of sine waves of up to 833 Hz (0-to-max charge is 1/4 of a sine period).

Can't possibly be the stators, right? Wrong. If they targeted charge states of up to 99% there would always be that asymptotic curvature enveloping the reproduced signal and making it sound like it's constantly soft-clipping. For music reproduction you can't just push a capacitor-like driver to its charge limit, you have to keep it below some percentage where the charging curve looks almost perfectly linear.

Can't say how low that would have to be, but I'm thinking it has to be damn low. Even if we used R=100kohm we would still have to keep the charging confined to where it doesn't take more than 0.25*RC so the process can be repeated at more than 160 kHz and allow the reproduction of sines of up to 40-ish kHz, which is a typical estat spec.

And even with all this we still don't know the membrane's capacitance or how long it takes to charge to its constant positive bias.

 

[IAtaman]

Which assumptions? That published value is likely not the membrane capacitance since it's unrealistically low for such a large surface (it's comparable to the inter-wire capacitance in the cable itself - ridiculous).

Remember I quoted 110 pF, not nF. Assuming only 60 pF of that is the driver, and with Nottagorilla's hint about Stax typically using a series R>=1Mohm, we get RC=60μs and then 99% charge time = 5*RC = 300 μs. That looks like it could only be done 3.33k times/s and would only allow the reproduction of sine waves of up to 833 Hz (0-to-max charge is 1/4 of a sine period).

Can't possibly be the stators, right? Wrong. If they targeted charge states of up to 99% there would always be that asymptotic curvature enveloping the reproduced signal and making it sound like it's constantly soft-clipping. For music reproduction you can't just push a capacitor-like driver to its charge limit, you have to keep it below some percentage where the charging curve looks almost perfectly linear.

Can't say how low that would have to be, but I'm thinking it has to be damn low. Even if we used R=100kohm we would still have to keep the charging confined to where it doesn't take more than 0.25*RC so the process can be repeated at more than 160 kHz and allow the reproduction of sines of up to 40-ish kHz, which is a typical estat spec.

And even with all this we still don't know the membrane's capacitance or how long it takes to charge to its constant positive bias.

I don't understand what are you talking about.

Below is a simplified diagram of a electrostatic speaker. You apply a constant voltage to the diaphragm / membrane. That creates an electric field between the the membrane and the stators. You apply the audio signal to the stators. Membrane moves to create changing air pressure. Are we aligned so far?

Now, what I thought you were saying was that it would take a lot of time to charge the membrane for it to produce sound properly. Neither the charge voltage of the membrane, nor the time constant of the RC circuit that dictates how long it would take to charge it has anything to do with the frequency of the audio signal. So what are you talking about when you say 140Hz or 833hz and such?

 

[donjoe0]

Now, what I thought you were saying was that it would take a lot of time to charge the membrane for it to produce sound properly.

Yes.

Neither the charge voltage of the membrane, nor the time constant of the RC circuit that dictates how long it would take to charge it has anything to do with the frequency of the audio signal.

No, it doesn't, I would say at first glance. (Orrr maybe it does - see below.)

So what are you talking about when you say 140Hz or 833hz and such?

Trying to estimate what frequencies the stators can reproduce based on the published Stax spec of "110 pF including cable", which I first thought characterized only the capacitor formed between the stators and said nothing about the membrane.

But if I look at that circuit again, it might be the combined capacitance of stators+membrane treated as 2 parallel capacitors, in which case all the estimations would have to be redone.

Are you reading it differently? Where do you see the 110 pF (minus cable) spec being realized on the diagram - between which terminals/plates?

 

[IAtaman]

Trying to estimate what frequencies the stators can reproduce based on the published Stax spec of "110 pF including cable", which I first thought characterized only the capacitor formed between the stators and said nothing about the membrane.

But if I look at that circuit again, it might be the combined capacitance of stators+membrane treated as 2 parallel capacitors, in which case all the estimations would have to be redone.

Are you reading it differently? Where do you see the 110 pF (minus cable) spec being realized on the diagram - between which terminals/plates?

With all due respect, I am getting the feeling that you have no idea what you are talking about and making theories up as you go.

There is one obvious capacitor in that circuit. Base voltage of that capacitor is constant, and sound is reproduced by modulating the electrical field by applying the audio signal to the stators. Even if the capacitance of was 10 times higher than what is specified at 1000pF (or 1nF), it would be almost fully charged after 5 seconds.

I don't know about you, but to me the theory that these needs to be charged for hours to sound right is starting to smell sour.

 

[donjoe0]

With all due respect, I am getting the feeling that you have no idea what you are talking about and making theories up as you go.

Well yes, I never claimed to be an expert at any of this. That's why I started by quoting the experiences of others claiming to be long-time users of the device, that's why I went on to trying to piece together what I could from different sources to explain said claims etc.

Even if the capacitance of was 10 times higher than what is specified at 1000pF (or 1nF), it would be almost fully charged after 5 seconds.

I don't know about you, but to me the theory that these needs to be charged for hours to sound right is starting to smell sour.

Tru dat, even in my "let's fumble with some numbers" analysis, things are already not looking great for the many-hours hypothesis, once the Stax spec is plugged into the formulas.

It could be that this was never about charging the membrane to the full Coulomb value but rather - as was my first suspicion before going into the capacitor discussion - about the uniform spread of that charge across the membrane surface. (I notice Ken Rockwell is also talking about it in these terms in his Stax reviews.) If the charge is not spread evenly you will not get equal forces acting at each point on the membrane and therefore you will not get perfectly planar movement, resulting in higher distortion (basically the reason planars were invented in the first place). How charge spread could be measured though is really completely beyond my pay grade. Short of redoing the distortion/noise/etc. measurements multiple times as the headphones stay powered for minutes and then hours, I don't know how this could be tested.

 

[IAtaman]

It could be that this was never about charging the membrane to the full Coulomb value but rather - as was my first suspicion before going into the capacitor discussion - about the uniform spread of that charge across the membrane surface. (I notice Ken Rockwell is also talking about it in these terms in his Stax reviews.) If the charge is not spread evenly you will not get equal forces acting at each point on the membrane and therefore you will not get perfectly planar movement, resulting in higher distortion (basically the reason planars were invented in the first place). How charge spread could be measured though is really completely beyond my pay grade. Short of redoing the distortion/noise/etc. measurements multiple times as the headphones stay powered for minutes and then hours, I don't know how this could be tested.

Or alternatively, and a lot more likely, charging them for a long time has no effect on the sound quality of these headphones

 

[donjoe0]

Or alternatively, and a lot more likely, charging them for a long time has no effect on the sound quality of these headphones

That's just a guess, I would prefer to see it confirmed with measurements before adopting it as some kind of final conclusion.

 

[IAtaman]

That's just a guess, I would prefer to see it confirmed with measurements before adopting it as some kind of final conclusion.

Sure, it is just a guess, in the same way that every time I turn my amp on, I guess it will work. Good luck to you.

 

[donjoe0]

Sure, it is just a guess, in the same way that every time I turn my amp on, I guess it will work. Good luck to you.

Nope, not in the same way. Apples are not oranges. These are electrostatic headphones, not electrodynamic, not magnetodynamic. Their membrane does not retain the same movement-relevant characteristics when unpowered vs. powered. How that transitions happens and what it does to the performance is worth investigating.

 

[Svperstar]

I'm thinking of getting the iCan Phantom, patiently waiting to see if anyone is every going to use the 95x on it.

 

[Soandso]

Why some prefer long electrostatic headphone (eStat) charging periods may have different reasons. Let's recap that the musical audio signal goes only to the flanking stators while voltage goes to the central membrane (originally in Stax 230 V and later designs 580 V). The modern eStat stator (singular) has up to 750 V of peak to peak charge so the 2 stators repelling and attracting the membrane can operate up to a peak to peak push-pull charge of 1,500 V.

A progression of eStat designs used different thickness of membranes and factored different gaps between the membrane and a stator. Stax increased that stator spacing gap to allow for greater push-pull membrane movement for better bass frequency dynamics and (since electro-static force decreases by the square of distance) raised their eStat membrane "bias" to 580V (calling it "Pro"). Somewhere along the way they also changed the chemical compound used in surface treatments (reputedly for safer human handling during production). Thus some (non-contemporary) Stax eStat design(s) can have different factors where the driving musical signal voltage being presented to the stator and the "gap" relative to the "bias" charged membrane treated with newer (less hazardous?) compounds is a factor.

An eStat will immediately play yet in some cases could improve it's membrane dynamism (sound better) to the push-pull of the stators' peak to peak push pull in the course of time. Why this would be so: when any so-called "parasitic" (residual one sided electrostatic charge) on the membrane accrues it takes time to be removed from where it is a localized static electric charge by getting new membrane "bias" charge. This is why some eStats that have initial channel imbalance usually resolves within a few minutes or even after a couple of days when supplied "bias" voltage; which encompasses the 3/4 to 1+ hour "warm up" range some adherents cite. [My vintage restored 1970s American Suprex PEP-74 eStat's channel imbalance persists approximately for over 3/4 of an hour so I leave it under constant A.C. "bias" charge to use on demand. While my 1990s Stax eStat has no initial channel imbalance and promptly sounds dynamic.]

As for why some eStat might get an audio dynamic limiting unilateral charge somewhere on it's membrane it seems there are 3 major explanations. One is how even a tiny point on the membrane surface can arise from an irregularity related to the chemical compound employed which can be from a production run or age which is vulnerable to holding a contrary charge. Another is introduced playing the music too loud for too long and that high peak to peak push-pull stator voltage left a "parasitic" charge on one side of the membrane which can be from factors of the membrane thickness relative to the "gap" with a stator. The less common cause is an actual outside contaminant.

From post #130 above for orientation here below is reproduced that eStat layout.

 

[isostasy]

I have had these for a few months, wanting to try the electrostatic / large diaphragm experience for cheap.

I use them with Vesper Audio pads which, although not fixing the poor frequency response, do make them much more comfortable and ensure better seal for bass. I put a strip of foam in them to angle them a little and have gently bent the metal headband connectors in a bit as they are were loose on my small head.

Here's my EQ:

Filter 1: ON LSC Fc 40 Hz Gain 6.0 dB Q 0.800
Filter 2: ON PK Fc 1000 Hz Gain -4.0 dB Q 1.000
Filter 3: OFF PK Fc 2000 Hz Gain 2.0 dB Q 0.500
Filter 4: ON PK Fc 3300 Hz Gain 2.0 dB Q 2.000
Filter 5: ON PK Fc 5600 Hz Gain -3.0 dB Q 4.000

If you have the velour pads just turn Filter 3 on as that approximates the difference. The low shelf is obviously to taste, sometimes I prefer 9dB at 0.6 Q for a warmer sound. Filter 5 is very subtle and not necessary unless that region bothers you, which I've found it often does me.

I do think they sound very good with EQ. They sound different to my HD6XX even though the EQ'd response is very similar - in particular they sound brighter in a way that belies the measurements and I suspect is due to the soundwave created by the large diaphragm interacting with my ear to create a different perception of treble than what is seen on a dummy head. I've noticed some spatial qualities, though not really "soundstage" which is a concept I struggle with. More like separation and openness though I realise these are unhelpful and overused terms. I get a bit more of a feeling of bass around my ears as well in contrast to the HD6XX and IEMs where the bass feels narrowly transmitted into your ear canal.

I suspect the electrostatic technology has nothing to do with it and that any perceived benefits are due to having large, open cups and diaphragms. This makes me think that planar magnetic headphones with similarly, or even larger, drivers would have the same benefits - and without the need for an energiser and the occasional appearance of an annoying whine. So now I'm tempted by those nice big (and much more expensive) Audezes.

I have flat-plate coupler measurements which I can share though they're of limited use; the cups are so large I have to take a lot of measurements and average as the response changes depending on position, so it's hard to know exactly how it relates to on-head response.

I'd like to be able to correct the response acoustically as there are electrostatic heapdhones with better response. I have no idea what kind of damping needs to be done and am not keen to open them up really.