Hi, I just finished my fifth pair of electrostatic loudspeakers. They are smaller than the previous ones which have much lower sensitivity, because they seem to leak bias. The newer models have smaller d/s as well (about 1.6 mm compared to the about 2.2 mm of the previous stats).
One panel experienced a problem with static stability as the diaphragm sucked to one of the stators on one panel. I attached a new diaphragm with much higher tension: I use duct-tape to tension the mylar, and the first attempt resulted in a very low tension. I didn't measure it, but the mylar surface has dimensions of 23 x 130 cm and one panel had a resonance frequency of 43 Hz, the other of 65 Hz. The panel with the lowest resonance experienced the stability problems, so after a second attempt I managed to increase resonance frequency to 86 Hz!
Problem is, that low frequency extension is limited now on the 86 Hz panel. So if I'm correct there is a relationship between sensitivity and static stability: the higher the resonance frequency, the higher the mechanical tension of the diaphragm, the higher bias can be applied, but the less low frequency extension, but physical excursion of diaphragm is less with higher resonance frequency.
Right now I think the best option / compromise is to get a high as possible mechanical tension in order to get maximum static stability and sensitivity. Sensitivity of my new stats is around 86 dB @ 1 meter @ 1 kHz @ 1 VA.
They are limited in low frequency output as they fall rapidly at 65 and 86 Hz respectively. But compared to my boxed transmision-line floor standing loudspeaker it turns out this isn't as big as a problem as I would think. But in order to get more low frequency extension I would need to make the diaphragm dimensions bigger, which leads to my next question: how do other stat builders get enough low frequency extension? Like Quad: the ESL-989 seems to have a resonance peak somewehre around 55 Hz:
https://www.diyaudio.com/forums/planars-and-exotics/361665-quad-esl-resonance-damping.html#post6377852
The stators seem to consist of 6 seperate panels per channel:
Quad ESL 989 Modifikation - Counterweight Tonarmgewicht Dual 704 Zarge Rahmen Quad ESL 989 Modifikation Clockwork FineNeedle AT20SLA ADC 10E Tesla geschenkt gratis 1500 km
Dimensions of the 989 are: 52" (1335mm) H by 26" (670mm) W by 12.25" (315mm). So 1 section (out of six) would measure at most 22 x 67 cm, much smaller than my panels which measure 23 x 130 (diaphragm dimensions).
I would expect low output after resonance frequency, and the sections of the quad 989 are about half in size of my panels, but resonance frequency of the complete panels is still lower (55 Hz?) than my panels. I'm wondering how this is possible?
One panel experienced a problem with static stability as the diaphragm sucked to one of the stators on one panel. I attached a new diaphragm with much higher tension: I use duct-tape to tension the mylar, and the first attempt resulted in a very low tension. I didn't measure it, but the mylar surface has dimensions of 23 x 130 cm and one panel had a resonance frequency of 43 Hz, the other of 65 Hz. The panel with the lowest resonance experienced the stability problems, so after a second attempt I managed to increase resonance frequency to 86 Hz!
Problem is, that low frequency extension is limited now on the 86 Hz panel. So if I'm correct there is a relationship between sensitivity and static stability: the higher the resonance frequency, the higher the mechanical tension of the diaphragm, the higher bias can be applied, but the less low frequency extension, but physical excursion of diaphragm is less with higher resonance frequency.
Right now I think the best option / compromise is to get a high as possible mechanical tension in order to get maximum static stability and sensitivity. Sensitivity of my new stats is around 86 dB @ 1 meter @ 1 kHz @ 1 VA.
They are limited in low frequency output as they fall rapidly at 65 and 86 Hz respectively. But compared to my boxed transmision-line floor standing loudspeaker it turns out this isn't as big as a problem as I would think. But in order to get more low frequency extension I would need to make the diaphragm dimensions bigger, which leads to my next question: how do other stat builders get enough low frequency extension? Like Quad: the ESL-989 seems to have a resonance peak somewehre around 55 Hz:
https://www.diyaudio.com/forums/planars-and-exotics/361665-quad-esl-resonance-damping.html#post6377852
The stators seem to consist of 6 seperate panels per channel:
Quad ESL 989 Modifikation - Counterweight Tonarmgewicht Dual 704 Zarge Rahmen Quad ESL 989 Modifikation Clockwork FineNeedle AT20SLA ADC 10E Tesla geschenkt gratis 1500 km
Dimensions of the 989 are: 52" (1335mm) H by 26" (670mm) W by 12.25" (315mm). So 1 section (out of six) would measure at most 22 x 67 cm, much smaller than my panels which measure 23 x 130 (diaphragm dimensions).
I would expect low output after resonance frequency, and the sections of the quad 989 are about half in size of my panels, but resonance frequency of the complete panels is still lower (55 Hz?) than my panels. I'm wondering how this is possible?
Quad ESL-63 and newer individual panel - free diaphragm size in round numbers is 6.9" (17.5 cm) by 22.9" (58.2 cm). Diaphragm-to-stator spacing is 0.1" (2.54 mm). Bias is specified as 5-5.3 kV.
Something to keep in mind is that you may increase the panel's voltage sensitivity by increasing tension and bias, but the panel's ultimate output may not increase if you exceed the optimum voltage ratio for bias/audio drive, as discussed at link below:
https://www.diyaudio.com/forums/pla...rrent-vs-voltage-drive-esl-3.html#post2346783
Something to keep in mind is that you may increase the panel's voltage sensitivity by increasing tension and bias, but the panel's ultimate output may not increase if you exceed the optimum voltage ratio for bias/audio drive, as discussed at link below:
https://www.diyaudio.com/forums/pla...rrent-vs-voltage-drive-esl-3.html#post2346783
Last edited:
@mattstat: thanks for your reply and pointing me to this interesting topic.
Do you have any idea how the Quad esl seem to have a resonance frequency somewhere around 55 Hz while the physical dimensions of the individual panel-segments are much smaller compared to my 1 segment panels which have higher resonant frequency?
Do you have any idea how the Quad esl seem to have a resonance frequency somewhere around 55 Hz while the physical dimensions of the individual panel-segments are much smaller compared to my 1 segment panels which have higher resonant frequency?
The modern quads use a roughly 3um diaphragm, and a D/S spacing of about 1:70. If my reading comprehension hasn't failed me, you have over 1:140 d/s spacing. that will be very hard to keep from collapsing as you note. You have managed to do it by cranking the tension WAY up. but a side effect is the resonant point is now unpleasantly high. to first order, the quad panels can have roughly half the tension because they have half the d/s spacing (it's not really linear though). so that's why they have a lower Fs
Sheldon
Sheldon
I only have a single panel from an ESL-63, not a whole speaker, so can't test some of the ideas that follow. The single panel on its own in free air has about an 80 Hz resonance. It looks like an original diaphragm, but I can't be certain of that.
Going by the double hump appearance in the nearfield measurement in Stereophile, I assumed the dust cover was at least partially involved in resonance frequency reduction/broadening. I think I experimented with that idea years ago and saw a similar result, but I'd have to dig through old notebooks/files to confirm whether my recollection is correct.
Quad ESL-989 electrostatic loudspeaker Measurements | Stereophile.com
This link also talks about multiple panels together lowering the resonance frequency.
Resonance Quad Panels !
Beyond those ideas, I lean toward Sheldon's opinion: that the ratio of diaphragm free span to diaphragm-to-stator spacing is pushing your panel into the realm where more tension is required for stability at the bias you are running.
If I'm remembering correctly, your previous designs used spacer dots also, instead of a full perimeter supporting each subdivided diaphragm area. I only tend to use dots if the diaphragm area is very small and the dots would provide a significant free diaphragm area increase. For larger panels, I like the extra stability provided by full perimeter spacers.
Going by the double hump appearance in the nearfield measurement in Stereophile, I assumed the dust cover was at least partially involved in resonance frequency reduction/broadening. I think I experimented with that idea years ago and saw a similar result, but I'd have to dig through old notebooks/files to confirm whether my recollection is correct.
Quad ESL-989 electrostatic loudspeaker Measurements | Stereophile.com
This link also talks about multiple panels together lowering the resonance frequency.
Resonance Quad Panels !
Beyond those ideas, I lean toward Sheldon's opinion: that the ratio of diaphragm free span to diaphragm-to-stator spacing is pushing your panel into the realm where more tension is required for stability at the bias you are running.
If I'm remembering correctly, your previous designs used spacer dots also, instead of a full perimeter supporting each subdivided diaphragm area. I only tend to use dots if the diaphragm area is very small and the dots would provide a significant free diaphragm area increase. For larger panels, I like the extra stability provided by full perimeter spacers.
Hi, thanks again for your reply. I put much more tension on the diaphragm compared to my previous attempts - but still lower than most people seem to apply: I often see numbers of 1% or more. I managed to get about 1% stretching for the smallest dimension (width) of my 23x130 cm mylar film, but lengthwise I was unable to reach 1%, I guess I only stretched 0,2% to 0,3%.
I put rectangular dots in the horizontal middle of the diaphragm, so I got a D/S spacing of about 1:70.
The link to the discussion of combining multiple panels resulting in lower resonance frequency is interesting. If I'm correct the resonance frequency of a diaphragm is dependent on the stiffness of the suspension and the mass of the air that moves with it. Is this latter thing the same as airload?
Let's assume we have two identical diaphragms of same size and tension. When we combine them, the resonant frequency will be lower than that of the individual panels. If we had only 1 panel with the same dimensions as the two smaller panels combined, and the same tension - would the resonance frequency be the same as the two smaller separate panels combined?
I put rectangular dots in the horizontal middle of the diaphragm, so I got a D/S spacing of about 1:70.
The link to the discussion of combining multiple panels resulting in lower resonance frequency is interesting. If I'm correct the resonance frequency of a diaphragm is dependent on the stiffness of the suspension and the mass of the air that moves with it. Is this latter thing the same as airload?
Let's assume we have two identical diaphragms of same size and tension. When we combine them, the resonant frequency will be lower than that of the individual panels. If we had only 1 panel with the same dimensions as the two smaller panels combined, and the same tension - would the resonance frequency be the same as the two smaller separate panels combined?
I'd be careful looking at stereophile measurements, they are nothing like you'll measure on axis. They do some sort of on and off axis blending and close mic'ing for the bass and then smoosh them all together to try to make something meaningful. They fail.
The dust cover doesn't have a significant resonance of its own, but it does provide a first order rolloff at the very high frequencies. That is why the thickness of the dust cover matters or I should say the mass. The resonance it does have is very low, because it is one large lightly stretched diaphragm.
The modern quad panels I make have a Fs of 55hz (as a group) when first built and that slides downward a bit (~5-10hz) as they break in and the diaphragm relaxes a bit. This matches the factory quad panels I've measured.
Sheldon
The dust cover doesn't have a significant resonance of its own, but it does provide a first order rolloff at the very high frequencies. That is why the thickness of the dust cover matters or I should say the mass. The resonance it does have is very low, because it is one large lightly stretched diaphragm.
The modern quad panels I make have a Fs of 55hz (as a group) when first built and that slides downward a bit (~5-10hz) as they break in and the diaphragm relaxes a bit. This matches the factory quad panels I've measured.
Sheldon
Last edited:
The resonant frequency depends on air mass and membrane tension. The air mass increases rapidly as panels are placed near to each other. So the resonant frequency for two panels side by side is lower than either of the panels by them selves. By the time you put 4 or 6 panels together, as in the QUADs, the mass increase and lowering of the resonance can be substantial. Sorry I don't know the actual figures, only the principles.
Much has been written, also in other treads, about diaphragm tension, drive forces, aera dimensioning, air loading, stator spacing, voltages ... Despite all these very valuable informations and this precious knowledge pool, I could not find any informations about the the physical displacement of the diaphragm inside an ESL panel due to a given audio input signal. I mean by that: something like the analalogy of the "Xmax" cone/dome displacement when speaking in terms of electromagnetodynamic loudspeaker drivers.
So, does anyone of the ESL experts here know about this "Xmax" in ESL's? Is there a means to approximate a value along with an arithmetic use of known ESL parameters such as the electrostatic force, membrane resonance frequency, membrane-to-stator distance, membrane tension or so? Or is there a useful and robust, rough rule of thumb, a crude estimate for the membrane dislocation amplitude in ESL's such as "max. 1/5 of the membrane-to-stator distance"? The only thing I know from experience ist that "Xmax" has been exceeded is by the time the ESL assembly optically behaves like a lit chrismas sparkler. But even then I don't know a numeric value for "Xmax", unless "Xmax" could then be defined and calculated as some kind of arcing distance excursion limit?
By the way: I could perfectly imagine that "Xmax" has no meaning at all for ESL's because of it's completely different radiation impedance behavior compared to electromagnetodynamic drivers? E.g. that only the electrostatic force (and the membrane surface) might be relevant to approximate a max. SPL which can be produced by a given panel? I might be wrong, then.
I post this question about "Xmax" here, because I could imagine that diaphragm amplitude also depends from diaphragm tension.
Please, please enlight me now. Better not by sparklers (any kind, sparkling ESL or chrismas sparkers), but more so by solid knowledge ... Big thanks in advent!
So, does anyone of the ESL experts here know about this "Xmax" in ESL's? Is there a means to approximate a value along with an arithmetic use of known ESL parameters such as the electrostatic force, membrane resonance frequency, membrane-to-stator distance, membrane tension or so? Or is there a useful and robust, rough rule of thumb, a crude estimate for the membrane dislocation amplitude in ESL's such as "max. 1/5 of the membrane-to-stator distance"? The only thing I know from experience ist that "Xmax" has been exceeded is by the time the ESL assembly optically behaves like a lit chrismas sparkler. But even then I don't know a numeric value for "Xmax", unless "Xmax" could then be defined and calculated as some kind of arcing distance excursion limit?
By the way: I could perfectly imagine that "Xmax" has no meaning at all for ESL's because of it's completely different radiation impedance behavior compared to electromagnetodynamic drivers? E.g. that only the electrostatic force (and the membrane surface) might be relevant to approximate a max. SPL which can be produced by a given panel? I might be wrong, then.
I post this question about "Xmax" here, because I could imagine that diaphragm amplitude also depends from diaphragm tension.
Please, please enlight me now. Better not by sparklers (any kind, sparkling ESL or chrismas sparkers), but more so by solid knowledge ... Big thanks in advent!
It's a bit complicated for ESLs compared to dynamic woofers for 2 major reasons. First, the ESL diaphragm resonance is high-Q(usually on the order of 10 to 20) where as dynamic woofer resonance Q is controlled by electrical damping and is easy to measure and calculate. Second, the load the ESL motor must push against at low frequency is made up almost entirely by the airload...which is difficult to quantify. It changes with panel shape and baffle extensions...even with altitude(ie air density). By contrast, the load on a dynamic woofer motor is dominated by its physical parameters, not the airload.
A reasonable estimation of the Xmax of an ESL diaphragm at low frequency is possible, and was added to the latest version of the Segmented ESL calculator spreadsheet you can download here:
https://www.diyaudio.com/community/threads/experiences-with-esl-directivity.48120/post-5441902
For details on the calculation method and assumption made, see Item 4) on the "Notes" TAB of the spreadsheet.