The panel works well when it is powered (2800Volts), without making any noise, it is only when I cut the mains supply that it generates noise for about 2mn without there being any modulation.
But you're right, I'm on the limit when it comes to width.
I based myself on Pierre Frank Jensen who uses the same specifications.
http://steenfrankjensen.com/projects/electrostatic-loudspeaker/
I've also used an antistatic spray which isn't Licron and I think it could cause problems, so I'm going to rebuild a panel with 6 micron mylar and maybe increase the D/S to 2 mm.
Thanks
Serge B.
But you're right, I'm on the limit when it comes to width.
I based myself on Pierre Frank Jensen who uses the same specifications.
http://steenfrankjensen.com/projects/electrostatic-loudspeaker/
I've also used an antistatic spray which isn't Licron and I think it could cause problems, so I'm going to rebuild a panel with 6 micron mylar and maybe increase the D/S to 2 mm.
Thanks
Serge B.
The Pierre Frank Jensen design uses a 3 mm membrane to stator spacing according to link. That would be more in the ballpark. Although 3 mm is pushing the limits of drive voltage requirements. And for a panel that only operates down to 300Hz, there is no need for such large DS or for such a large panel.
I would build the next panel not as wide, maybe 150mm. You can increase the frame width left and right of the panel so the total width stays the same, that will maintain the same frequency where phase cancellation sets in. With a 1:100 transformer, there is enough drive voltage to increase DS as well. As a bonus, capacitance will be much lower making things a lot easier for the amplifier.
Just to give an idea, I play with a panel that's 20 x 100cm, it plays down to 100Hz where it is crossed over to a dipole sub. It will even play fullrange with enough output for a small room. DS is close to 3mm, bias 5kV.
I would build the next panel not as wide, maybe 150mm. You can increase the frame width left and right of the panel so the total width stays the same, that will maintain the same frequency where phase cancellation sets in. With a 1:100 transformer, there is enough drive voltage to increase DS as well. As a bonus, capacitance will be much lower making things a lot easier for the amplifier.
Just to give an idea, I play with a panel that's 20 x 100cm, it plays down to 100Hz where it is crossed over to a dipole sub. It will even play fullrange with enough output for a small room. DS is close to 3mm, bias 5kV.
Maudio, following your advice, I'm going to glue two 0.5 mm thick wooden edge strips around the stator, which will give a D/S of about 2.5 mm. I could keep the width (2.5 mm x 100 = 25 cm)?
Thanks for your advice.
Serge B.
Thanks for your advice.
Serge B.
Should work yes. And you can always play with the bias. Or add extra support for the membrane. Audiostatic uses silicone dots on each crossmember, in the middle of the membrane. It also helps to spread the fundamental resonance.
Before you tear down your current panel, try feeding it a frequency sweep. That will tell you where the resonance is, and it will tell you when the membrane touches the stators. That will make a very loud rattling noise. Start at 20Hz and sweep to a few hundred Hz
Before you tear down your current panel, try feeding it a frequency sweep. That will tell you where the resonance is, and it will tell you when the membrane touches the stators. That will make a very loud rattling noise. Start at 20Hz and sweep to a few hundred Hz
Your panel d/s is 1.5 mm and the width is 25 cm. Is that correct?
What caught my eye in your panel photo is that I don't see any center diaphragm supports-- at least none are visible.
Roger Sanders "Electrostatic Loudspeaker Design Cookbook" provides a guideline for diaphragm stability, which insures that the diaphragm will not be pulled into a stator by the recommended bias voltage-- or too-easily driven into to a stator during music playback.
Sanders' guideline states that the span between diaphragm supports should be 70-100X the d/s. Most builders follow this guideline.
A span less than 70X d/s would merely give up some efficiency, but a span greater than 100X d/s would not provide sufficient diaphragm stability.
With your panel d/s being 1.5 mm, the span between diaphragm supports should not exceed 15 cm.
In your panel photo, it appears that the diaphragm is spanning the full 25 cm width of the panel with no center supports.
If that's true, the span is excessive, and that might explain why you are still getting static after reducing the bias voltage below 3KV.
You noted that you patterned your panel/segmentation arrangement on Frank Steen Jensen's excellent speaker and website.
As the d/s is Mr. Jensen's panel is 3mm, the diaphragm span in that case would not exceed Sanders' 100X d/s limit.
If the diaphragm span is the problem, there may be simpler fix that redoing the panel. Many wire-stator ESLs with wider panels stabilize the diaphragm with a line of silicon dots (locally bonding the diaphragm to the wires) spaced a few inches apart along the mid-span of the panel. Think of the silicone dots as tent poles that keep the tent (diaphragm) lofted away from the ground (stator). A Google search would probably find some photos of this arrangement. I think builders who use silicone dots in lieu of spacers just get a tube of clear RTV and squirt some in thru the spaces between the wires.
What caught my eye in your panel photo is that I don't see any center diaphragm supports-- at least none are visible.
Roger Sanders "Electrostatic Loudspeaker Design Cookbook" provides a guideline for diaphragm stability, which insures that the diaphragm will not be pulled into a stator by the recommended bias voltage-- or too-easily driven into to a stator during music playback.
Sanders' guideline states that the span between diaphragm supports should be 70-100X the d/s. Most builders follow this guideline.
A span less than 70X d/s would merely give up some efficiency, but a span greater than 100X d/s would not provide sufficient diaphragm stability.
With your panel d/s being 1.5 mm, the span between diaphragm supports should not exceed 15 cm.
In your panel photo, it appears that the diaphragm is spanning the full 25 cm width of the panel with no center supports.
If that's true, the span is excessive, and that might explain why you are still getting static after reducing the bias voltage below 3KV.
You noted that you patterned your panel/segmentation arrangement on Frank Steen Jensen's excellent speaker and website.
As the d/s is Mr. Jensen's panel is 3mm, the diaphragm span in that case would not exceed Sanders' 100X d/s limit.
If the diaphragm span is the problem, there may be simpler fix that redoing the panel. Many wire-stator ESLs with wider panels stabilize the diaphragm with a line of silicon dots (locally bonding the diaphragm to the wires) spaced a few inches apart along the mid-span of the panel. Think of the silicone dots as tent poles that keep the tent (diaphragm) lofted away from the ground (stator). A Google search would probably find some photos of this arrangement. I think builders who use silicone dots in lieu of spacers just get a tube of clear RTV and squirt some in thru the spaces between the wires.
Last edited:
Hello CharlieM,
Thank you for your reply, I made a mistake reading Frank Jansen's site, you're right I'm going to reread Sanders and Wagner's book (which I haven't opened in 20 years) so as not to make any more mistakes.
Silicone dots would be a good thing, but I'd rather go back to a 2.5 mm D/S panel + 6 micron Mylar + Licron, I still have the materials to do it.
Thanks again, I'll keep you posted.
Serge B.
Thank you for your reply, I made a mistake reading Frank Jansen's site, you're right I'm going to reread Sanders and Wagner's book (which I haven't opened in 20 years) so as not to make any more mistakes.
Silicone dots would be a good thing, but I'd rather go back to a 2.5 mm D/S panel + 6 micron Mylar + Licron, I still have the materials to do it.
Thanks again, I'll keep you posted.
Serge B.
Hello SB,
Yes, I think you are right-- the silicone dots may have worked but once done, it would not be possible to disassemble the panel again without bending the wires at the dot locations.
Increasing the d/s as you suggest would be the best fix, and that would also allow going back to the higher biasing voltage to offset the lower drive force.
One more tip:
Your post didn't mention whether your stators are mechanically fastened together or bonded with adhesive tape.
Prying bonded stators apart can be tricky and risks bending the wires. An safe method to separate bonded stators is by cutting thru the tape bond lines using a string saw made from 60-80 pound braided fishing line.
This works quite well (I've done it).
BTW; building ESLs is a learning curve. I don't know anyone who got it right on the first attempt of an unfamiliar design. My first ESL played wonderfully for about 20 seconds before the high voltage corona found the weak spots along my stator edges and self-destructed-- I was fortunate that it didn't also smoke my amplifier.
Yes, I think you are right-- the silicone dots may have worked but once done, it would not be possible to disassemble the panel again without bending the wires at the dot locations.
Increasing the d/s as you suggest would be the best fix, and that would also allow going back to the higher biasing voltage to offset the lower drive force.
One more tip:
Your post didn't mention whether your stators are mechanically fastened together or bonded with adhesive tape.
Prying bonded stators apart can be tricky and risks bending the wires. An safe method to separate bonded stators is by cutting thru the tape bond lines using a string saw made from 60-80 pound braided fishing line.
This works quite well (I've done it).
BTW; building ESLs is a learning curve. I don't know anyone who got it right on the first attempt of an unfamiliar design. My first ESL played wonderfully for about 20 seconds before the high voltage corona found the weak spots along my stator edges and self-destructed-- I was fortunate that it didn't also smoke my amplifier.
Last edited:
That depends. When you place the dots behind the horizontal supports, disassembly is no problem at all, no risk of bending the wires. This is in fact how Audiostatic does it, I rebuild several pairs without ever damaging the wires. After removal of the old membrane, the dots that stick to the wires can be peeled off by hand. After placing a new membrane, I use a syringe without needle to place the new ones. The small tip allows me to inject the silicone between the wires. When you do so with the panel flat on the table, membrane facing down, it will form a nice round dot that is easy to control in size.
That's good to know. I was concerned that it would be a problem, but apparently not when implemented as you've shown. Thanks for sharing that!
Hello CharlieM,
It was a question I wanted to ask you, as my stators are glued as you indicate on your site, you've beaten me to it.
I'm going to try and find some fishing line.
Thank you
Serge B.
It was a question I wanted to ask you, as my stators are glued as you indicate on your site, you've beaten me to it.
I'm going to try and find some fishing line.
Thank you
Serge B.
The saw thru part will be easy. Removing the adhesive residue from the stator edges afterwards will be a pain. No solvent that I know of will dissolve it but acetone will soften it and allow you to roll it off in bits and pieces -- no fun!
I can also recommend the excellent article that Frank Verwaal from the Netherlands wrote on the subject of ESL design.
Note: This article has been available free to download on Frank's homepage for many years, but his ISP has terminated all homepages recently so the existing link (http://home.kpn.nl/verwa255/esl/ESL_English_2011.pdf) is dead now. Therefore I include the pdf here.
Thank you maudio, I had this pdf and it contains a lot of interesting information.
I measured the panel and the resonance is around 120 Hertz.
but no significant noise at this frequency.
Serge B.
I measured the panel and the resonance is around 120 Hertz.
but no significant noise at this frequency.
Serge B.
120Hz is quite high for a panel that's 25 cm wide. I would expect something in the 40-80 Hz range for this size. That means the membrane tension is indeed very high. Or the membrane is touching the stators in the middle and is not completely free to move. In that case you have the two outer free membrane parts moving in parallel, with the middle section stationary. The fact that it's not rattling makes me think the latter.
Are you planning on using it fullrange or with a sub? For a fullrange you want the resonance as low as possible. For a hybrid it's best to have it at least 2 octaves below the crossover frequency. You can reduce that to 1 octave if you add a notch filter in the crossover and/or some damping.
Are you planning on using it fullrange or with a sub? For a fullrange you want the resonance as low as possible. For a hybrid it's best to have it at least 2 octaves below the crossover frequency. You can reduce that to 1 octave if you add a notch filter in the crossover and/or some damping.
The resonance frequency is high, I think I stretched the Mylar to the maximum I measured it with a generator and an amp it would have been better to use Arta or Rew for a reliable result.
I'm doing a hybrid with a SEAS XM004-04 L26RO4Y closed-box speaker (40 Litres) + LR4 and maybe a Linkwitz transform.
I'm doing a hybrid with a SEAS XM004-04 L26RO4Y closed-box speaker (40 Litres) + LR4 and maybe a Linkwitz transform.
Hi,
efficiency, efficiency, efficiency is the mantra for good esl design as I see it.
This almost automatically leads to hybrid designs with small d/s values and high diaphragm tension.
It also means high capacitance, which is a good thing because it means higher amounts of charge, hence more output.
As such the transformation factor of the audio transformer required for impedance matching can be lower.
This has the advantage of overall lower voltages and lower voltage gradients, reducing stress on isolators (ageing).
This -maybe appearing counter intuitive at first glance- raises efficiency to well above 90dB/1W/4m) and increases dynamic range by 10-15dB easily.
My 25cm wide panel -bowed punched metal sheet stators- utilizes a front d/s of 1.0 mm and back d/s of 1.1 mm.
Horizontal orientated parallel spacers are distanced at varying ratios between 70 to 90.
Using 3.5u and 4.8u film stretched to ~1.5% elongation results in ~170-180Hz of Fs.
Still though crossing over at as low as possible frequencies is sonically desirable.
I chose ~250Hz. Like Maudio told before, this requires a precise notch filter, maybe even combined with mechanical damping.
I'd suggest to get rid of standard textbook filter thinking with ESLs, and hybrids in special.
The aim should be to have symmetrical acoustic filter flanks.
The bass may work quite amplitude linear over two or more octaves above the xover point, and a textbook filter may result in a almost identical acoustic outcome.
The ESL panel will be very different though.
Due to the pronounced Fs resonance and the necessity for an notch filter close to the required working frequency range even a first order HP would provide for a much higher order filter (rather look for elliptical filter responses than BW, LR or similar)
When notch center frequency and HP xover are close, the notch's response is by far dominating.
I use a second order HP with a Q>1 and a slightly too high xover point that at the same cancels the acoustic shortcut down from ~600Hz.
Combined the panel's amplitude response and the two electrical filters show a linear response down to ~250Hz, with a steep and roundabout 6th order falloff below.
The bass filter then requires at least a 4th order filter like LR4 to match the ESL's response curve.
jauu
Calvin
efficiency, efficiency, efficiency is the mantra for good esl design as I see it.
This almost automatically leads to hybrid designs with small d/s values and high diaphragm tension.
It also means high capacitance, which is a good thing because it means higher amounts of charge, hence more output.
As such the transformation factor of the audio transformer required for impedance matching can be lower.
This has the advantage of overall lower voltages and lower voltage gradients, reducing stress on isolators (ageing).
This -maybe appearing counter intuitive at first glance- raises efficiency to well above 90dB/1W/4m) and increases dynamic range by 10-15dB easily.
My 25cm wide panel -bowed punched metal sheet stators- utilizes a front d/s of 1.0 mm and back d/s of 1.1 mm.
Horizontal orientated parallel spacers are distanced at varying ratios between 70 to 90.
Using 3.5u and 4.8u film stretched to ~1.5% elongation results in ~170-180Hz of Fs.
Still though crossing over at as low as possible frequencies is sonically desirable.
I chose ~250Hz. Like Maudio told before, this requires a precise notch filter, maybe even combined with mechanical damping.
I'd suggest to get rid of standard textbook filter thinking with ESLs, and hybrids in special.
The aim should be to have symmetrical acoustic filter flanks.
The bass may work quite amplitude linear over two or more octaves above the xover point, and a textbook filter may result in a almost identical acoustic outcome.
The ESL panel will be very different though.
Due to the pronounced Fs resonance and the necessity for an notch filter close to the required working frequency range even a first order HP would provide for a much higher order filter (rather look for elliptical filter responses than BW, LR or similar)
When notch center frequency and HP xover are close, the notch's response is by far dominating.
I use a second order HP with a Q>1 and a slightly too high xover point that at the same cancels the acoustic shortcut down from ~600Hz.
Combined the panel's amplitude response and the two electrical filters show a linear response down to ~250Hz, with a steep and roundabout 6th order falloff below.
The bass filter then requires at least a 4th order filter like LR4 to match the ESL's response curve.
jauu
Calvin
Hello,
yes, that must be the problem.
I finished modifying the stators to 2.55 mm D/S with double-sided tape.
I glued two strips of 0.5 mm wood (1 mm thick) around the perimeter of the stator and used 6 micron Mylar and high tension.
The crossover will be digital (ADAU 1452 Analog Device), LR4 for the woofer and panel.
Thanks for the advice.
Serge B.
yes, that must be the problem.
I finished modifying the stators to 2.55 mm D/S with double-sided tape.
I glued two strips of 0.5 mm wood (1 mm thick) around the perimeter of the stator and used 6 micron Mylar and high tension.
The crossover will be digital (ADAU 1452 Analog Device), LR4 for the woofer and panel.
Thanks for the advice.
Serge B.
Hello,
Just to keep you up to date with the modifications you advised me to make:
The D/S was increased from 1.5mm to 2.6mm, and I was able to increase the bias voltage to 6000 volts without any noise problems.
Should I keep this voltage or lower it?
Thank you all for your help.
Serge B.
Just to keep you up to date with the modifications you advised me to make:
The D/S was increased from 1.5mm to 2.6mm, and I was able to increase the bias voltage to 6000 volts without any noise problems.
Should I keep this voltage or lower it?
Thank you all for your help.
Serge B.