Considering the lobing problems with line arrays, a long tweeter sounds like a good idea.
I have been considering one myself, but I have not built one.
However, I have read Ronald Wagner's book on electrostatic loudspeakers, so I would like to throw some numbers around just to see how it stacks up. This is merely a starting point.
According to Ronald Wagner's book, you need about 55 volts per mil, (thousandths of an inch). If you want to cross your tweeter over at 800 Hz, and achieve an SPL of 112 dB, it has to move 0.1 cu inch of air. This is based on the Small model where 6.2 cu inches of air must be moved to achieve 112 db @ 100 Hz, and extrapolating the answer by dividing the requirement by 4 for every octave above 100 Hz.
So 0.1 cu inches of air it is.
If we make our tweeter 5 ft long, that is 60 inches. We want good disperson from the tweeter, so we make it 0.5 inches wide. You might decide a different width, of course.
That makes our diaphragm 30 sq inches.
0.1/30 = .003333 inches, or 3.3333 mils.
3.3 mils times 55 volts/mils = 181 volts for your tweeter.
I would think that would be possible.
Making a frame for the tweeter that is 60 inches long and sturdy is another matter. But it will be only 1/2 in wide, so sturdiness might be achievable.
Crossing over at 800 Hz would eliminate the lobing region for most line arrays. Plus, I would think you would want the electrostatic to carry as much of the music as it can afford. 112 dB should satisfy most people.
Again, just throwing these things out as a point of discussion. Repeat, I have not built one.
I have been considering one myself, but I have not built one.
However, I have read Ronald Wagner's book on electrostatic loudspeakers, so I would like to throw some numbers around just to see how it stacks up. This is merely a starting point.
According to Ronald Wagner's book, you need about 55 volts per mil, (thousandths of an inch). If you want to cross your tweeter over at 800 Hz, and achieve an SPL of 112 dB, it has to move 0.1 cu inch of air. This is based on the Small model where 6.2 cu inches of air must be moved to achieve 112 db @ 100 Hz, and extrapolating the answer by dividing the requirement by 4 for every octave above 100 Hz.
So 0.1 cu inches of air it is.
If we make our tweeter 5 ft long, that is 60 inches. We want good disperson from the tweeter, so we make it 0.5 inches wide. You might decide a different width, of course.
That makes our diaphragm 30 sq inches.
0.1/30 = .003333 inches, or 3.3333 mils.
3.3 mils times 55 volts/mils = 181 volts for your tweeter.
I would think that would be possible.
Making a frame for the tweeter that is 60 inches long and sturdy is another matter. But it will be only 1/2 in wide, so sturdiness might be achievable.
Crossing over at 800 Hz would eliminate the lobing region for most line arrays. Plus, I would think you would want the electrostatic to carry as much of the music as it can afford. 112 dB should satisfy most people.
Again, just throwing these things out as a point of discussion. Repeat, I have not built one.
I have also thought about this idea. I haven't calculated the excursion (thanks for the info), but I have given some thought to implementation. One method that would be quite simple would be to use PCB material (FR4 or similar) as the spacer and frame. One can think of several ways to use copper-clad material to allow one to make contact with both the diaphragm and the stators. One may ever be able to get away with 1/32" material since the excursion requirements appear to be so small.
If one used this approach, a few segments would probably be necessary, although one could overlap the strips to form one long tweeter. The whole thing would be pretty flexible, but then one could mount it to the baffle and heat it up to tighten the diaphragm.
Finally, the transformer requirements would be a lot less stringent, so winding your own might be a reasonable thing to do.
Just throwing some ideas out since I know I won't get to this for a while.
John
If one used this approach, a few segments would probably be necessary, although one could overlap the strips to form one long tweeter. The whole thing would be pretty flexible, but then one could mount it to the baffle and heat it up to tighten the diaphragm.
Finally, the transformer requirements would be a lot less stringent, so winding your own might be a reasonable thing to do.
Just throwing some ideas out since I know I won't get to this for a while.
John
Just a thought about the narrow width. Electostatics are able to excurse because of the stretch in the diaphragm material. If the material is held to .5 or 1 inch in width and then tensioned, where is the stretch going to come from.
I realize that you have calculated a relatively low excursion requirement, but at the proposed widths, I believe typical mylars will be too rigid. There may be a reason you don't see tall and narrow (ribbon like) electrostatics. The lack of stretch across the shortest dimension may be why.
Again, just a thought to investigate before you go too far with the design,
Mark
I realize that you have calculated a relatively low excursion requirement, but at the proposed widths, I believe typical mylars will be too rigid. There may be a reason you don't see tall and narrow (ribbon like) electrostatics. The lack of stretch across the shortest dimension may be why.
Again, just a thought to investigate before you go too far with the design,
Mark
Bill, these links come to mind.
http://www.amasci.com/esloud/eslhwto.html
http://www.audiocircuit.com/index.php
http://hitechnetworks.net/bwaldron/main_old.htm
http://home1.gte.net/res0f2t3/
Prosit
http://www.amasci.com/esloud/eslhwto.html
http://www.audiocircuit.com/index.php
http://hitechnetworks.net/bwaldron/main_old.htm
http://home1.gte.net/res0f2t3/
Prosit
MarkMcK said:
A good thought. If it only excurses half of what I predicted, then the top SPL is still 106 dB, and that ain't bad.
Ronald Wagner's book contains an article where one of the early pioneers of electrostatic loudspeakers, back in the fifties, built a large diaphragm with 0.25" excursion.
Assuming his diaphragm was two feet wide, the peak would be in the middle, or 12" from the side.
That works out to an angle of 1.2º. The same deflection, when applied to a leg of 0.25"-one half of the width of the diaphragm-equals .0052 inches, or 5.2 mils.
So it looks like we have deflection to spare.
Some other things to consider:
The Mylar is stretched and put under tension. If we run into trouble on this, perhaps you just put it under somewhat less tension than you would for a larger electrostatic. Experimentation would help on this, of course. And then there is the option of making the tweeter 1" or 1.5" instead of half an inch, though I would hate to see that happen.
If we run into problems still, there is also the option of moving the crossover point upward. Moving it up one half octave equals 6 dB greater SPL for the same excursion. In this case, moving half an octave upward would mean going from 800 Hz to 1120 Hz. That is still pretty low for a tweeter crossover, and depending on the diameter of his woofer/mids, is still low enough to avoid lobing problems at the crossover point, I would think.
People don't realize this, but a 1" tweeter does not have quite the dispersion you would hope. If you are going to go through the effort of building an electrostatic, I would think you would very much want the airy sound that good dispersion affords.
Again, just throwing some things I have read and calculated as a first step.
Just two thoughts: Why not make the diaphragm a little wider than the aperture in the stator in order to have enough Vd and a narrow radiating area at the same time. One has to pay attention to different path lenghts etc though (same problem as is also encountered with compression drivers). The other one would be a diaphragm that is some inches wide but that is driven by different frequency ranges across it's width (i.e. the same as is done with th equad ESL but only in one dimension).
A third idea that just crossed my mind: Use some form of "acousic filter" (i.e. some foam or the like) that covers the left and right side of the stator partially, leaving only a narrow slot for very high frequencies while being "transparent" for mid frequencies.
Regards
Charles
I'm not so sure that would not result in a rolloff of SPL at the top end.phase_accurate said:
If we have a diaphragm 2" wide, and conver all but the 1/2" in the middle in absorbent foam, letting only the high frequency output from one quarter of the diaphragm go through, then you are getting the high frequency output of one quarter of the diaphragm, which will be 12 dB down from the output of the whole diaphragm.
Raising the power from the amp won't help, since it would increase the excursion 4 times, which would obviate the advantage of the larger diaphragm.
At least, that is the way I would envision it working.
phase_accurate said:
A fellow named Beveridge right here in Connecticut made some highly rated electrostatics using that principle. Never did get a chance to hear them.
The wavelength of a 20,000 Hz tone is 2/3 in. A slot arrangement that has a difference in path lengths from the edge to the center will be 6 dB down, (I think!), at 20,000 Hz. Which wold not be so bad. If the difference is less than 2/3 of an inch, then the output will be correspondingly less.
I think it possibly could work.
Of course, even if you do have cancellation effects at very high frequencies, those frequencies do not require much excursion to put out high SPL, so you can equalize. So you may not have to be that concerned with differering path lengths.
What Beveridge does is a little more refined than my suggestion. He is not just covering a part of the panel he is also using some sort of acoustic lens. I once heard a pair of them at an exhibition a quarter of a century ago. I can remember that they offered very transparent sound.
I also assume that every one of my suggestions would give deviations of the frequency response curve. But as always: You can't have everything.
Regards
Charles
I also assume that every one of my suggestions would give deviations of the frequency response curve. But as always: You can't have everything.
Regards
Charles
Seems to me that with the absorptive slot set up you'd be ok. The high freqs tend to radiate from the center of the diaphragm anyhow, as this is inevitably the point of greatest excursion in practice... also the effectiveness of the thin, excursion limited, edge captive diaphragm at mid freqs isn't going to be quite as good as you might imagine - meaning it will work better as the freq goes up anyhow...
having said that you probably won't need the slot - or you won't need a partially absorptive slot...
Pay some attention to the cavity resonances though...
it's an idea that I wanted to build back in 1976... (there's a pic of a big line source dynamic driver system from back then in the Archives section of my websites - submitted for your amusement)
Another advantage is that you can get a relatively high sensitivity out of it because ur driving it as a bandwidth limited driver...
As far as the rigidity of the assembly, that shouldn't pose much of a problem, one could follow any one of the "popular" construction techniques and come up with a very solid and stiff set up.
I would really try to make a continuous strip, not break it into sections... once you break it into sections one seems to get the same effect as having individual tweeters separated by the distance between the *centers* of each cell... dispite the idea that ESLs radiate uniformly at all freqs...
The membrane ought to be stable enough if the width is cleverly selected and tensioned properly...
_-_-bear
having said that you probably won't need the slot - or you won't need a partially absorptive slot...
Pay some attention to the cavity resonances though...
it's an idea that I wanted to build back in 1976... (there's a pic of a big line source dynamic driver system from back then in the Archives section of my websites - submitted for your amusement)
Another advantage is that you can get a relatively high sensitivity out of it because ur driving it as a bandwidth limited driver...
As far as the rigidity of the assembly, that shouldn't pose much of a problem, one could follow any one of the "popular" construction techniques and come up with a very solid and stiff set up.
I would really try to make a continuous strip, not break it into sections... once you break it into sections one seems to get the same effect as having individual tweeters separated by the distance between the *centers* of each cell... dispite the idea that ESLs radiate uniformly at all freqs...
The membrane ought to be stable enough if the width is cleverly selected and tensioned properly...
_-_-bear
My ESL experiences (a bit long...)
I’ve been thinking for a couple of years, off and on, about the idea of using an ESL tweeter or ESL tweeter/midrange along with a linear array of woofers or mid/woofers. I think the idea has merit but sure am having trouble finding the time to put something together! Nonetheless, I’m very interested in everyone’s ideas on the subject and am glad to see the question raised.
Here are a few thoughts and observations that might be pertinent to this thread. They’re based on my experiences building a few successful and a few unsuccessful ESL prototypes of various sizes, and a pair of large panels that I’ve been listening to for a few years now. I realize little of it's new, but the previous messages left me with the impression that some of it might be of use to somebody (my apologies to those who've seen it all before). Here goes...
1) If the smallest dimension of the panel is small (as in a narrow tweeter) the increased diaphragm stiffness shows up as an increase in the resonance frequency of the panel. We care about the resonances because if you have a major resonance in the working range of the driver, it’s typically of very high Q and so you get a lot of displacement for very little electrical input. This not only sounds bad, it also severely limits the dynamic range of the panel because the diaphragm runs into the stator---this either sounds bad, or sounds bad and arcs destructively. So, the resonance ends up defining the low frequency limit of the panel. Unless you use very steep high-pass filters, you have to start rolling off the panel well above the resonance to be safe. Acoustically damping the panel can also help.
By the way, I’ve found the resonances of a prototype can be easily heard even without driving the ESL panel electrically (and without any bias voltage). You just tap the stator with a finger and you can hear the drumhead resonances of the diaphragm. You have to use the stator construction that you’d use in the real speaker because the air load in front of the diaphragm affects the resonances, and the air load is affected by the porosity of the stators. This tapping trick makes it easy to test the viability of a panel without having to go through all the diaphragm coating and electrical connections and so forth.
2) As was briefly mentioned already, you have to be careful about Helmholtz (cavity) resonances in the supporting structure if you use a narrow ESL panel and a thick support structure. For example, if you cut a 5’ tall but 1” wide slot in ¾” MDF and mount a tweeter panel in front of or behind the slot, you’ll get major high frequency resonance effects from the air enclosed within the 1” wide, 3/4” deep slot. You can round or bevel the slot to improve things, but you end up with a badly designed horn (unless you’re a great “beveler”, in which case you end up with a wonderfully designed horn ).
3) I haven’t noticed any problems where two separate panels are mounted very close together to approximate a single tall driver. When I move my head above and below the seam between my two ESL panels, I can’t discern the seam by ear unless my ear is right up against the panels (always exciting while the voltages are applied).
4) I agree with those who have suggested that making a sufficiently rigid structure isn’t too hard, especially if you’re aiming for a relatively narrow panel. Even if it’s a foot wide it’s not too bad.
5) By using wire stators, or at least electrically isolated stator sections, you could drive the center of the panel with high frequencies (or semi-full-range) but drive the left and right edges of the panel with a low-pass filtered signal. This gives you more surface area, and therefore requires less displacement, for the lower frequencies, and you’d achieve the narrow width that provides decent high frequency distribution around the room. It also moves the frames supporting edges away from the high-frequency source (tweeter) so that cavity resonances are less problematic. This is obviously not a new idea (Wagner's book, Quad...).
6) The original Quads used a narrow tweeter panel flanked by wider panels for lower frequencies, so the idea of a tall(ish) narrow tweeter has a pretty well established precedent. If a project comes out only sounding as good as the original Quads, I could live with that.
7) When using Lincaine (sp?) stators I found 1/32" spacing from stator to diaphragm to be a pain to maintain because everything flexed (I was using double-stick foam spacers). 1/16" was much easier to work with. With heavier perforated steel stators, everything is easier--except painting them before they rust.
I’ve been thinking for a couple of years, off and on, about the idea of using an ESL tweeter or ESL tweeter/midrange along with a linear array of woofers or mid/woofers. I think the idea has merit but sure am having trouble finding the time to put something together! Nonetheless, I’m very interested in everyone’s ideas on the subject and am glad to see the question raised.
Here are a few thoughts and observations that might be pertinent to this thread. They’re based on my experiences building a few successful and a few unsuccessful ESL prototypes of various sizes, and a pair of large panels that I’ve been listening to for a few years now. I realize little of it's new, but the previous messages left me with the impression that some of it might be of use to somebody (my apologies to those who've seen it all before). Here goes...
1) If the smallest dimension of the panel is small (as in a narrow tweeter) the increased diaphragm stiffness shows up as an increase in the resonance frequency of the panel. We care about the resonances because if you have a major resonance in the working range of the driver, it’s typically of very high Q and so you get a lot of displacement for very little electrical input. This not only sounds bad, it also severely limits the dynamic range of the panel because the diaphragm runs into the stator---this either sounds bad, or sounds bad and arcs destructively. So, the resonance ends up defining the low frequency limit of the panel. Unless you use very steep high-pass filters, you have to start rolling off the panel well above the resonance to be safe. Acoustically damping the panel can also help.
By the way, I’ve found the resonances of a prototype can be easily heard even without driving the ESL panel electrically (and without any bias voltage). You just tap the stator with a finger and you can hear the drumhead resonances of the diaphragm. You have to use the stator construction that you’d use in the real speaker because the air load in front of the diaphragm affects the resonances, and the air load is affected by the porosity of the stators. This tapping trick makes it easy to test the viability of a panel without having to go through all the diaphragm coating and electrical connections and so forth.
2) As was briefly mentioned already, you have to be careful about Helmholtz (cavity) resonances in the supporting structure if you use a narrow ESL panel and a thick support structure. For example, if you cut a 5’ tall but 1” wide slot in ¾” MDF and mount a tweeter panel in front of or behind the slot, you’ll get major high frequency resonance effects from the air enclosed within the 1” wide, 3/4” deep slot. You can round or bevel the slot to improve things, but you end up with a badly designed horn (unless you’re a great “beveler”, in which case you end up with a wonderfully designed horn ).
3) I haven’t noticed any problems where two separate panels are mounted very close together to approximate a single tall driver. When I move my head above and below the seam between my two ESL panels, I can’t discern the seam by ear unless my ear is right up against the panels (always exciting while the voltages are applied).
4) I agree with those who have suggested that making a sufficiently rigid structure isn’t too hard, especially if you’re aiming for a relatively narrow panel. Even if it’s a foot wide it’s not too bad.
5) By using wire stators, or at least electrically isolated stator sections, you could drive the center of the panel with high frequencies (or semi-full-range) but drive the left and right edges of the panel with a low-pass filtered signal. This gives you more surface area, and therefore requires less displacement, for the lower frequencies, and you’d achieve the narrow width that provides decent high frequency distribution around the room. It also moves the frames supporting edges away from the high-frequency source (tweeter) so that cavity resonances are less problematic. This is obviously not a new idea (Wagner's book, Quad...).
6) The original Quads used a narrow tweeter panel flanked by wider panels for lower frequencies, so the idea of a tall(ish) narrow tweeter has a pretty well established precedent. If a project comes out only sounding as good as the original Quads, I could live with that.
7) When using Lincaine (sp?) stators I found 1/32" spacing from stator to diaphragm to be a pain to maintain because everything flexed (I was using double-stick foam spacers). 1/16" was much easier to work with. With heavier perforated steel stators, everything is easier--except painting them before they rust.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.