First let me state that I do not advocate that fullrange-singlemembrane ESL is the holy grail. Not at all.
Nice! This is another point of potential refinement I was not immediately aware. Talking about distortions, rub and noise, more IM/Doppler will also come along for the membrane action itself for a fullrange system compared to a "bassless" ESL. Furthemore, as pointed out ...
... so it is a good approach not to stimulate all the frame's nasty eigen-modes by potentially shaking it too much while reproducing energy-rich bass layers.
But then, what's the real bass SPL potential, say, of an ESL63? Grossly comparing the volume displacement capacity of an ESL63 (70x50 cm x 2mm = 700cm^3) with the one of a single 12-inch Peerless XXLS (446 cm^2 x 25mm = 1115cm^3) you mathematically get an slight advantage for the 12-inch driver. This advantage vanishes if you allow for an Xmax = 3mm for the ESL, which might be as real as the 25mm for the XXLS. Having done that, you will be aware that the raw bass potential of an ESL's membrane array is not bad at all. Then, resorting in real-life to two or four of these 12-inchers will definitely outperform the ELS bass potential. Speaking for an ESL63. The membrane sandwiches of the two bass panels inside of the ESL57 might provide less bass SPL potential than the ESL63.
It's all about compromizing, and IMHO also adding a dedicated sub solution to an ESL is nothing but another compromize. If you add a sub stage, then go for the best possible compromize within the sub spectrum of options, not to spoil the ESL's qualities. As for me, I finally compromized not to do so, for very personal reasons. I knowingly compromized for less bass (which is a kind of sacrifice) and, and as a benefit, for less cluttering the living environment with heavy and optically proemintent audio gear. Otherwise and as mentionned at an other place, I probably would try to go with additional time-delayed ESL bass panels nearby the hearing location. Possibly I would be sitting between two ESL bass panels, placed in the middle of the room a bit like a huge electrostatic headphones. A giant Jecklin Float for bass, so to say. Or, as for me, as the sandwiched sausage, part of an audio hotdog session.
Nice! This is another point of potential refinement I was not immediately aware. Talking about distortions, rub and noise, more IM/Doppler will also come along for the membrane action itself for a fullrange system compared to a "bassless" ESL. Furthemore, as pointed out ...
... so it is a good approach not to stimulate all the frame's nasty eigen-modes by potentially shaking it too much while reproducing energy-rich bass layers.
But then, what's the real bass SPL potential, say, of an ESL63? Grossly comparing the volume displacement capacity of an ESL63 (70x50 cm x 2mm = 700cm^3) with the one of a single 12-inch Peerless XXLS (446 cm^2 x 25mm = 1115cm^3) you mathematically get an slight advantage for the 12-inch driver. This advantage vanishes if you allow for an Xmax = 3mm for the ESL, which might be as real as the 25mm for the XXLS. Having done that, you will be aware that the raw bass potential of an ESL's membrane array is not bad at all. Then, resorting in real-life to two or four of these 12-inchers will definitely outperform the ELS bass potential. Speaking for an ESL63. The membrane sandwiches of the two bass panels inside of the ESL57 might provide less bass SPL potential than the ESL63.
It's all about compromizing, and IMHO also adding a dedicated sub solution to an ESL is nothing but another compromize. If you add a sub stage, then go for the best possible compromize within the sub spectrum of options, not to spoil the ESL's qualities. As for me, I finally compromized not to do so, for very personal reasons. I knowingly compromized for less bass (which is a kind of sacrifice) and, and as a benefit, for less cluttering the living environment with heavy and optically proemintent audio gear. Otherwise and as mentionned at an other place, I probably would try to go with additional time-delayed ESL bass panels nearby the hearing location. Possibly I would be sitting between two ESL bass panels, placed in the middle of the room a bit like a huge electrostatic headphones. A giant Jecklin Float for bass, so to say. Or, as for me, as the sandwiched sausage, part of an audio hotdog session.
Last edited:
The X max for ESL 63 is closer to 2mm peak to peak = 1mm in one direction.
This is because its only the center that moves max 2mm before arcing. The edges moves close to zero.
This is because its only the center that moves max 2mm before arcing. The edges moves close to zero.
Good point that indeed is often overlooked.
And there is another factor that limits the Xmax in an ESL: The membrane will not be in the middle without a signal. It will always be attracted more to one stator then to the other one, resulting in a curvature. That's why we need membrane tension in the first place, as a counteraction force. This also puts a practical limit on the membrane width/DS spacing ratio of about 1:70-100.
True Xmax of a 63 will therefore be even less than 2 mm. So it's not so surprising that even a single 12 inch long stroke OB woofer will completely blow the ESL away when it comes to low frequency output.
And there is another factor that limits the Xmax in an ESL: The membrane will not be in the middle without a signal. It will always be attracted more to one stator then to the other one, resulting in a curvature. That's why we need membrane tension in the first place, as a counteraction force. This also puts a practical limit on the membrane width/DS spacing ratio of about 1:70-100.
True Xmax of a 63 will therefore be even less than 2 mm. So it's not so surprising that even a single 12 inch long stroke OB woofer will completely blow the ESL away when it comes to low frequency output.
This was actually one of the things I thought about when thinking about subs. Not as elaborate as you have approached things, but I wondered about the relatively large membranes and how they would be affected by the subwoofer.
I decided that it would be marginal because the ESL is on top of the subwoofer and with an open baffle design like a Ripole this would place it square in the area of cancellation so that would minimize the issue (if any). Gradient did more or less the same thing with their subwoofers (for the 57 and there was one for the 63 as well).
Interesting, I know for the 57 version of the Gradient it mentioned it added a dip in the 80Hz region when crossing to the ESL because the ESL had a bump there. I used REW and a miniDSP 2x4HD to crossover/equalize, and I didn't really see that bump, I have to admit. But that seems suspiciously close to your graphs.
Remember the dipole compensation that has to be applied 6dB/oct
so from 100Hz to 50Hz you have to increase output of the woofer with 6dB. From 50Hz to 25Hz another 6dB
And add another 2dB for 25-> 20 Hz = 14dB
so from 100Hz to 50Hz you have to increase output of the woofer with 6dB. From 50Hz to 25Hz another 6dB
And add another 2dB for 25-> 20 Hz = 14dB
Agreed. I might erroneously have been a wee bit too optimistic.
For an ESL63, the inner's stator side of the meshed stator to the diapragm distance is some 2.3 ... 2.4mm, which limits an approx. 4.5mm wide physical space for the membrane to behave in. What about arcing limiting Xmax? As for functional, electrostatic aspects limiting Xmax, add the thickness of the stator itself, as the conductive perforated copper layer is located on the other side of the stator. And also consider that this outside copper plane is further isolated by an isolating varnish.
During a regular, non-arcing, maximum-excursion of the membrane, I guess that the membrane's cross-sections might look like some never-plane, but rather ellipsoids shapes, of course maximizing the excursion at the center of the panel, and being 0 at the bound-to-the-stator locations, of course. So, what might be the real, overall integrated Xmax, or let's better talk of the max. (+_) linear volume displacement capacity of a Quad63 panel? It would really be interesting to have a theoretical model which allows to derive this value from all factors, such as membrane elasticity, tension, drive, stator assembly geometry etc. etc. For interest's sake only, nota bene, and not to argue on who's right and who's wrong.
... plus a bunch of extra dB's of input power in order to compensate for the magnetodynamic's driver Qts highpass characteristic. In the end, an open-baffle sub is really something to avoid in terms of efficiency. Use them best in wintertime. And by the way: Has anybody ever been lamenting about the power compression aspect of these OB/Dipole woofers? This is where a horn system certainly would shine. Compromize, compromize ...
Finally, an erratum:
OOps ... These values are too low. Sorry for this one. I peeled them out of my erroneous memory and did not certify them before posting. Values for surface resistance around 500MOhms square and for the ESL63 might be more standard instead. ER-Audio's Rob Mackinlay explicitly considers 50MOhm square to be way too low for fair bass behavior. Having used Rob's coating, I easily got values around the 500MOhm square target.
Last edited:
The membrane coating resistance should never be considered without a rigid connection with the polarizing voltage applied to the membrane, these are interrelated concepts.
They should be selected so that the return is maximum and at the same time the membrane does not stick to the stator.
That is, if you decide to change the factory settings in this regard with the membrane coating, then you will need to change the bias (polarizing voltage on the membrane) anyway.
They should be selected so that the return is maximum and at the same time the membrane does not stick to the stator.
That is, if you decide to change the factory settings in this regard with the membrane coating, then you will need to change the bias (polarizing voltage on the membrane) anyway.
Efficiency is a non-issue with Ripoles. If you look at the Iron Law, you have a woofer that easily gets down into the 20's and still only measures like 40x40x40cm (16x16x16") so you already know efficiency is non-existant. But if they can keep up with the 57s in terms of SPL (which they do with two fingers up their noses) that's not much of a concern.
I agree, and dipole subs don't need that much power. I run my dipole subs from 60W amps which is overkill. The esl's need much more, I run 200+W amps (with over 20A peak current capability) and they need it.
Still esl's are incredible efficient as there is nothing that can dissipate heat in them. But they require huge amounts of blind current from the amp so they make the amp rather inefficient.
Basically you're driving a capacitor of several uF. In theory there should be a resistive element in their impedance, otherwise there would be no sound output. But that is so tiny that it's impossible to measure.
Which is why there is no electrical damping effect in an ESL, unlike in a driver with a coil/magnet motor. Which is why the Q goes off the scale.
Still esl's are incredible efficient as there is nothing that can dissipate heat in them. But they require huge amounts of blind current from the amp so they make the amp rather inefficient.
Basically you're driving a capacitor of several uF. In theory there should be a resistive element in their impedance, otherwise there would be no sound output. But that is so tiny that it's impossible to measure.
Which is why there is no electrical damping effect in an ESL, unlike in a driver with a coil/magnet motor. Which is why the Q goes off the scale.
You mean several nF? Some of those Martin Logans go up to that range.
Quads are an exception as the delay lines not only create a nice point source, but also flatten out the impedance curve.
I've measured the impedance between stators on a 63' and it came out to about 400k up to 10kHz or so. Almost purely resistive.
Jan
Quads are an exception as the delay lines not only create a nice point source, but also flatten out the impedance curve.
I've measured the impedance between stators on a 63' and it came out to about 400k up to 10kHz or so. Almost purely resistive.
Jan
No, uF is correct. A 500pF ESL after a 1:150 stepup transformer is equivalent to a 500pf * 150^2 = 11uF capacitor as seen from the amplifier. That's equivalent to 0.7 ohm @ 20kHz... And without taking into account the internal capacitance of the transformer, which often is in the same ballpark as the ESL.
It's a good thing the energy content in music drops steadily with frequency. I measured peaks full amplitude up to around 4000Hz in music, after that it drops steeply. Most pain for the amp is therefore in the 1-4kHz range.Hence the large blind current demands. I 've seen many amps with overcurrent protection go into current clipping in this range.
Electrical segmentation is essential to keep the load reasonable. A big advantage of wire stator designs.
The 63 and later Quads perform a very nice trick in this regard yes.
Last edited:
I suppose that is measured on the panels with the delay lines, not on the bare panels?
I rely on the Linkwitz SPL spreadsheet for speaker VD and max. SPL.
Take 40x40x40 ripole. Fit it with two 12-inch Peerless XXLS P830845 with SD=493, Xmax=30mm, fs=29Hz, Qts=0.65, and SPL_1W=86.6dB. This driver with it's rather highish Q is very well suited for open baffles/ripoles-and-the-like.
Resorting to the Linkwitz data. D=40cm for shortest front-to-back acoustical shunt. At 20Hz and for the ripole on the floor (+6dB) you get a volume-displacement limited max. SPL of 90dB (which is 20dB less than for a non-dipole setup with the same drivers).
Now take into account the driver's Qts=0.7 and resonance fs=29Hz. For a HP filter of Fs=29Hz and Q=0.7 the attenuation at 20Hz is around -7dB (voltage, not power). So finally you get less than 80dB/W sensitivity at 20Hz. So, for listening at moderate levels, this Ripole will do it's job along a 60W amplifier. At 20Hz, max. SPL cannot exceed 90dB anyway. This comes at a stout +- 15mm breathing of both driver's membranes, nota bene.
As mentionned before, these numbers are for Qts=0.7 (red graph). The gren graph instead shows the result with the same driver data except for a more power hungy Qts 0.3. My XXLS drivers had a bit less volume-displacement capacity and, most annoying, an open-baffle-desastrous Qts of 0.25. And the W-housings were minimally dimensioned. These facts might have contributed to my personal mix-blessed OB-Sub experience.
Take 40x40x40 ripole. Fit it with two 12-inch Peerless XXLS P830845 with SD=493, Xmax=30mm, fs=29Hz, Qts=0.65, and SPL_1W=86.6dB. This driver with it's rather highish Q is very well suited for open baffles/ripoles-and-the-like.
Resorting to the Linkwitz data. D=40cm for shortest front-to-back acoustical shunt. At 20Hz and for the ripole on the floor (+6dB) you get a volume-displacement limited max. SPL of 90dB (which is 20dB less than for a non-dipole setup with the same drivers).
Now take into account the driver's Qts=0.7 and resonance fs=29Hz. For a HP filter of Fs=29Hz and Q=0.7 the attenuation at 20Hz is around -7dB (voltage, not power). So finally you get less than 80dB/W sensitivity at 20Hz. So, for listening at moderate levels, this Ripole will do it's job along a 60W amplifier. At 20Hz, max. SPL cannot exceed 90dB anyway. This comes at a stout +- 15mm breathing of both driver's membranes, nota bene.
As mentionned before, these numbers are for Qts=0.7 (red graph). The gren graph instead shows the result with the same driver data except for a more power hungy Qts 0.3. My XXLS drivers had a bit less volume-displacement capacity and, most annoying, an open-baffle-desastrous Qts of 0.25. And the W-housings were minimally dimensioned. These facts might have contributed to my personal mix-blessed OB-Sub experience.
@ jandidden:That's pretty impressive. Mr Walker knew what he was doing 😉
Did you measure the source impedance of the transformers as well? They are part of the equation for the total transfer curves.
With transformers I expect the level will have quite a noticable effect as well.
Did you measure the source impedance of the transformers as well? They are part of the equation for the total transfer curves.
With transformers I expect the level will have quite a noticable effect as well.
This is the rear stator to diaphragm capacitance on an ESL (57) bass panel. Front stator to diaphragm is the same, so total is double. And there are two bass panels parallel, total is quadruple = 1660 pF.
Attachments
But the transformer is not directly driving the stators. There's quite some network in between. That's what the amp sees.
jan
jan