Hello All,
First off, I am new to the board. My goal is to build an ESL with subwoofer.
Before I do so though I would like to understand as much as possible. I have read the books by Sanders, Fikier and Wagner. I have browsed the web for days on end and read a lot on diyaudio.
But still, questions remain.
The big one right now is the falloff in ELS SPL towards the low frequencies. And if this is different for wire stators and plate stators.
For example Sanders claims in his book (page 16) that the rolloff starts at the frequency for which the minimum width is 1/4 of the wavelength. He further claims that this is due to phase cancellation (from the backside).
But in Fikier's boek, he shows a measurement of the ESL 175 in fig 9.16 where the rolloff is at about 60Hz. The width of the ESL 175 is only 260mm, which according to Sanders should have the rolloff starting at around 4.5Khz.
Unfortunately Sanders does not provide measurements but shows only calculated frequency charts.
Wagner does not mention this effect at all, though he does mention directivity on page 76/77.
At first I thought that the discrepancy between Fikier and Sanders could be due to the use of a wire stator by Fikier. As the wires act as an acoustic lens, they would disperse the higher frequencies more effectively pushing the rolloff frequency lower.
However that cannot be the case as in fig 9.4 he shows the frequency curve of a metal plate ESL that also has the rolloff below 100Hz (but which according to Sanders should have been at about 2.5kHz)
Thus I am left puzzled. Does the effect mentioned by Sanders exist? And if so, how can it be that Fikier did not see it?
I would be grateful for any answers.
First off, I am new to the board. My goal is to build an ESL with subwoofer.
Before I do so though I would like to understand as much as possible. I have read the books by Sanders, Fikier and Wagner. I have browsed the web for days on end and read a lot on diyaudio.
But still, questions remain.
The big one right now is the falloff in ELS SPL towards the low frequencies. And if this is different for wire stators and plate stators.
For example Sanders claims in his book (page 16) that the rolloff starts at the frequency for which the minimum width is 1/4 of the wavelength. He further claims that this is due to phase cancellation (from the backside).
But in Fikier's boek, he shows a measurement of the ESL 175 in fig 9.16 where the rolloff is at about 60Hz. The width of the ESL 175 is only 260mm, which according to Sanders should have the rolloff starting at around 4.5Khz.
Unfortunately Sanders does not provide measurements but shows only calculated frequency charts.
Wagner does not mention this effect at all, though he does mention directivity on page 76/77.
At first I thought that the discrepancy between Fikier and Sanders could be due to the use of a wire stator by Fikier. As the wires act as an acoustic lens, they would disperse the higher frequencies more effectively pushing the rolloff frequency lower.
However that cannot be the case as in fig 9.4 he shows the frequency curve of a metal plate ESL that also has the rolloff below 100Hz (but which according to Sanders should have been at about 2.5kHz)
Thus I am left puzzled. Does the effect mentioned by Sanders exist? And if so, how can it be that Fikier did not see it?
I would be grateful for any answers.
hI
The wires won't act as a lense, as the wavelength of the highest frequency is larger than the dimensions of the wires. It would act as a lense in case of a single slit as shown in different books of physics. Such single slit can be found in the beveridge esls. So there should be no difference between perforated sheets and wire stators regarding dispersion.
There have been some measurements performed by amateurs of the esl175 which shows a cancellation dip indeed. It is centered between 100 and 250 Hz roughly. So this is at a lower frequency as predicted by Sanders.
The wires won't act as a lense, as the wavelength of the highest frequency is larger than the dimensions of the wires. It would act as a lense in case of a single slit as shown in different books of physics. Such single slit can be found in the beveridge esls. So there should be no difference between perforated sheets and wire stators regarding dispersion.
There have been some measurements performed by amateurs of the esl175 which shows a cancellation dip indeed. It is centered between 100 and 250 Hz roughly. So this is at a lower frequency as predicted by Sanders.
Hy Skyseeker,
The difference is caused by the setup off the measurement.
If you do a close distance measurement (say 10 cm), the dipole cancelation does not (or hardly) occure.
If Fikier had made the measurement at say 2 to 3 meters, the LF rolloff
would show up.
You should also check out the web site of Linkwitz (I don't remember if he mentions the effect of the distance on dipole cancelation).
See http://www.linkwitzlab.com
Edwin
The difference is caused by the setup off the measurement.
If you do a close distance measurement (say 10 cm), the dipole cancelation does not (or hardly) occure.
If Fikier had made the measurement at say 2 to 3 meters, the LF rolloff
would show up.
You should also check out the web site of Linkwitz (I don't remember if he mentions the effect of the distance on dipole cancelation).
See http://www.linkwitzlab.com
Edwin
Sy: "The issues are effective width (not the same as actual width, you have to account for the length, too)"
Thanks, I would like to find out more about this. Any suggestions?
Dijkstra: "The wires won't act as a lense, as the wavelength of the highest frequency is larger than the dimensions of the wires"
Yes, I wondered about that too. Still it seems as if the dispersion characteristic of the ESL175 is much better than the "laser focus" Sanders talks about. Are there any other explanations?
Dijkstra: "It is centered between 100 and 250 Hz roughly"
That is very useful information. Thus a crossover to a dynamic woofer would need to be at say, 200 or 300 Hz.
Edwin: " ... measurement at say 2 to 3 meters, the LF rolloff would show up."
Unfortunately Fikier does not say how he did perform the measurements. But as he also included 30 degrees of-axis measurements, so I think it is safe to assume that it must have been more than 1 meter. Otherwise even 30 degrees of axis would be in front of the speaker. Distance is a factor though, as also Sanders says that having the speakers really close makes it a much more immersive experience (than at a distance).
Oh, and yes, I spend a few days over Linkwitz's website. He almost convinced me to stick with dynamic dipoles instead!
(I will use an open dipole dynamic woofer though)
--------------
Does anyone know of direct comparisons between a metal stator ESL and a (identical size, transformer etc) wire stator ESL?
And for a wacko idea: How about an ESL with one metal plate stator and one wire stator?
Thanks, I would like to find out more about this. Any suggestions?
Dijkstra: "The wires won't act as a lense, as the wavelength of the highest frequency is larger than the dimensions of the wires"
Yes, I wondered about that too. Still it seems as if the dispersion characteristic of the ESL175 is much better than the "laser focus" Sanders talks about. Are there any other explanations?
Dijkstra: "It is centered between 100 and 250 Hz roughly"
That is very useful information. Thus a crossover to a dynamic woofer would need to be at say, 200 or 300 Hz.
Edwin: " ... measurement at say 2 to 3 meters, the LF rolloff would show up."
Unfortunately Fikier does not say how he did perform the measurements. But as he also included 30 degrees of-axis measurements, so I think it is safe to assume that it must have been more than 1 meter. Otherwise even 30 degrees of axis would be in front of the speaker. Distance is a factor though, as also Sanders says that having the speakers really close makes it a much more immersive experience (than at a distance).
Oh, and yes, I spend a few days over Linkwitz's website. He almost convinced me to stick with dynamic dipoles instead!
(I will use an open dipole dynamic woofer though)
--------------
Does anyone know of direct comparisons between a metal stator ESL and a (identical size, transformer etc) wire stator ESL?
And for a wacko idea: How about an ESL with one metal plate stator and one wire stator?
My experience with balancing the low end of ESL's is that a 315Hz is indeed a sweet spot for crossover to some dynamic drivers.
The best solution is to build a companion line-array of mid-bass drivers for the 40 or 60Hz to 315Hz range.
This maintains the same depth of field characteristics of the ESL line source in the critical mid-bass.
For more details, measurements and theory on how I applied this to create a very capable, large Center Channel to match my MartinLogan Monoliths, read this thread:
http://www.martinloganowners.com/~tdacquis/forum/showthread.php?t=2018
This is the result:
The best solution is to build a companion line-array of mid-bass drivers for the 40 or 60Hz to 315Hz range.
This maintains the same depth of field characteristics of the ESL line source in the critical mid-bass.
For more details, measurements and theory on how I applied this to create a very capable, large Center Channel to match my MartinLogan Monoliths, read this thread:
http://www.martinloganowners.com/~tdacquis/forum/showthread.php?t=2018
This is the result:
Hi,
the broader distribution character of a wire stator against a flat metal stator is solely due to electrical segmentation and not a matter of stator building material. You can achieve similar results to a electrically segmented stator by curving the stator -as ML and other do.
Acoustic phase cancellation starts above 1kHz with usual panal widths.
So You have to equalize the freq-response in any case or You get the -sadly too often heared anemic, thin sound. The lower You chose the crossover-freq the more You have to equalize. Experience shows that a equing does no sonical harm as long as You eq less than ~6dB. To stay below this value most panels should be crossed over around 300Hz, with wider panels allowing for lower crossover-points and thinner panels for higher ones.
I agree with R. Sanders when he says that a too low crossover point is a trap in which most ESL DIYers fall. It kills output and distortion figures and doesn´t gain on sound quality on the other hand!
The imo optimum way how the problem is solved is shown in the ML Statement2, JonFo´s setup and -as I may humbly add- my own setup.
jauu
Calvin
the broader distribution character of a wire stator against a flat metal stator is solely due to electrical segmentation and not a matter of stator building material. You can achieve similar results to a electrically segmented stator by curving the stator -as ML and other do.
Acoustic phase cancellation starts above 1kHz with usual panal widths.
So You have to equalize the freq-response in any case or You get the -sadly too often heared anemic, thin sound. The lower You chose the crossover-freq the more You have to equalize. Experience shows that a equing does no sonical harm as long as You eq less than ~6dB. To stay below this value most panels should be crossed over around 300Hz, with wider panels allowing for lower crossover-points and thinner panels for higher ones.
I agree with R. Sanders when he says that a too low crossover point is a trap in which most ESL DIYers fall. It kills output and distortion figures and doesn´t gain on sound quality on the other hand!
The imo optimum way how the problem is solved is shown in the ML Statement2, JonFo´s setup and -as I may humbly add- my own setup.
jauu
Calvin
Floor dip/bounce.
That will create a hole or "suck out" in the response of any speaker that is some distance above the floor (or other surface). This can not be overcome with simple means.
One way to tell if this is happening is to get the speaker by some means onto the floor (lay it down?) and remeasure, and/or do a FFT type measurement with the mic very close to the diaphragm so as to avoid the measurement of the acoustically created suckout.
Briefly stated, this suckout effect sucks.
Don't overlook its contribution to these issues.
_-_-bear
That will create a hole or "suck out" in the response of any speaker that is some distance above the floor (or other surface). This can not be overcome with simple means.
One way to tell if this is happening is to get the speaker by some means onto the floor (lay it down?) and remeasure, and/or do a FFT type measurement with the mic very close to the diaphragm so as to avoid the measurement of the acoustically created suckout.
Briefly stated, this suckout effect sucks.
Don't overlook its contribution to these issues.
_-_-bear
"To stay below this value most panels should be crossed over around 300Hz, with wider panels allowing for lower crossover-points and thinner panels for higher ones."
My current plans call for a 400mm wide (wire) stator, 1.5mm DS, 1600mm height. Slightly curved (20 dergees opening). Crossover at 500 Hz to a (one per channel) peerless XXLS 12" in an open dipole.
http://www.tymphany.com/datasheet/printview.php?id=321
(3KV bias, 1:100 audio transformator, 100W amplifier, but the amplifier might change)
Any suggestions?
My current plans call for a 400mm wide (wire) stator, 1.5mm DS, 1600mm height. Slightly curved (20 dergees opening). Crossover at 500 Hz to a (one per channel) peerless XXLS 12" in an open dipole.
http://www.tymphany.com/datasheet/printview.php?id=321
(3KV bias, 1:100 audio transformator, 100W amplifier, but the amplifier might change)
Any suggestions?
Hi,
as MJ pointed out 1:100 is too much if You don´t use segmentation
A metal-sheet stator will have a capacitance of app. 1.7nF. This translates to ~17µF with a 1:100 tranny and calls for a extremely stable amplifier and loss of highs because of very low impedance values. If You´re then using firstclass trannies with low losses You´ll hardly find a suitable amp to drive this beast. 1:50 is a good choice to start with (the suggested pair of power trannies with 230/6V
will rather give 1:68
)
With a panel this large (a ML Statement uses similar size) You can easily cross over at 300Hz. I wouldn´t go for a higher crossover point in this case. The dynamic headroom will still be higher than a single 12" driver could deliver. When designed and built carefully the panel will have an efficiency above 90dB/2.83V/1m and an even higher value @4m distance. A single 12" driver in OB will be much lower in value (my wild guess: 10 to 15dB). The XXLS is a true subwoofer driver with high inductance voice coil, high mass and a rather weak motor. Those are not intended to work good in the midrange. JonFoo and I have shown different solutions with multiple 6"-8"-drivers that fit better to such a large panel and outperform a single XXLS in probabely every point -even price!
3kV polarization is very optimistic ;-) You´ll probabely end up with 2kV, at best 2.5kV for noise freedom and practicability.
jauu
Calvin
as MJ pointed out 1:100 is too much if You don´t use segmentation
A metal-sheet stator will have a capacitance of app. 1.7nF. This translates to ~17µF with a 1:100 tranny and calls for a extremely stable amplifier and loss of highs because of very low impedance values. If You´re then using firstclass trannies with low losses You´ll hardly find a suitable amp to drive this beast. 1:50 is a good choice to start with (the suggested pair of power trannies with 230/6V
will rather give 1:68
)With a panel this large (a ML Statement uses similar size) You can easily cross over at 300Hz. I wouldn´t go for a higher crossover point in this case. The dynamic headroom will still be higher than a single 12" driver could deliver. When designed and built carefully the panel will have an efficiency above 90dB/2.83V/1m and an even higher value @4m distance. A single 12" driver in OB will be much lower in value (my wild guess: 10 to 15dB). The XXLS is a true subwoofer driver with high inductance voice coil, high mass and a rather weak motor. Those are not intended to work good in the midrange. JonFoo and I have shown different solutions with multiple 6"-8"-drivers that fit better to such a large panel and outperform a single XXLS in probabely every point -even price!
3kV polarization is very optimistic ;-) You´ll probabely end up with 2kV, at best 2.5kV for noise freedom and practicability.
jauu
Calvin
the suggested pair of power trannies with 230/6V will rather give 1:68
Hmm, I see that 230:6 gives 38, not 48 or 68
2.5kV?, ok. I plan to use an adjustable bias supply thus I can experiment with the bias.
300Hz x-over. OK, so be it
I have been wondering about the x-over and my choice was heavy influenced by R. Sanders's book.
The choice for the XXLS was made because I like o.a. organ music. So I really want to get to that 20Hz. Are there better driver choices for that? (PS: I would use two drivers if needed)
10-15db less SPL than the ESL is quite unacceptable of course.
I am quite taken in with your (Calvin) concept of having a line source for the 100-300 Hz region. What driver would you use if you would build that again?
Btw: I am still reading the back posts on this forum, you guys have done a great job of providing extremely useful information. It just takes ages to wade through all of it
Hmm, I see that 230:6 gives 38, not 48 or 68
2.5kV?, ok. I plan to use an adjustable bias supply thus I can experiment with the bias.
300Hz x-over. OK, so be it
I have been wondering about the x-over and my choice was heavy influenced by R. Sanders's book.
The choice for the XXLS was made because I like o.a. organ music. So I really want to get to that 20Hz. Are there better driver choices for that? (PS: I would use two drivers if needed)
10-15db less SPL than the ESL is quite unacceptable of course.
I am quite taken in with your (Calvin) concept of having a line source for the 100-300 Hz region. What driver would you use if you would build that again?
Btw: I am still reading the back posts on this forum, you guys have done a great job of providing extremely useful information. It just takes ages to wade through all of it
Hi,
just in short.
You need two voltages, one each for each stator, referenced to the polarizing voltage. This means one tranny with 1x6V/2x230V, or two 6V/230V trannies with their 6V-coils connected in parallel and their 230V-coils connected in series. The connection is the reference point for the polarization.
Driver:
Well I would use between 5" and 8" size, 8 pieces per side.
The Westra I used is no longer in production, but I assume that the dayton 7" drivers (from PartsExpress) like the 295-335 or the RS180-8 will be ok. With these You need a sub-bass below +50-60Hz. with an array of 8"-drivers not. So if choosing an 8"driver it should have a fs of ~30-35Hz, while the smaller drivers allow for fs of up to 60Hz.
jauu
Calvin
just in short.
You need two voltages, one each for each stator, referenced to the polarizing voltage. This means one tranny with 1x6V/2x230V, or two 6V/230V trannies with their 6V-coils connected in parallel and their 230V-coils connected in series. The connection is the reference point for the polarization.
Driver:
Well I would use between 5" and 8" size, 8 pieces per side.
The Westra I used is no longer in production, but I assume that the dayton 7" drivers (from PartsExpress) like the 295-335 or the RS180-8 will be ok. With these You need a sub-bass below +50-60Hz. with an array of 8"-drivers not. So if choosing an 8"driver it should have a fs of ~30-35Hz, while the smaller drivers allow for fs of up to 60Hz.
jauu
Calvin
"two 6V/230V trannies"
Yes, of course <slap to the forehead>.
Somehow I had this picture of using a 230/2x6V, but of course that won't work. Silly me
"So if choosing an 8"driver it should have a fs of ~30-35Hz"
Since I seem to be drawn to alu cones (somehow any graph I like seems to come from an alu cone), I was thinking the Dayton RS225S-8 an 8" with an Fs of 27Hz.
I do not have any experience with drivers, so if these would be inappropriate I would very much like to know. Anyone?
For the record: X-over at 300Hz, 8x RS255S in a line source. No extra subwoofer. I would reduce the width of the ESL to 350mm and increase its length to 1800mm. D/S remains at 1.5, wire stator, 20 degrees curved stator.
Yes, of course <slap to the forehead>.
Somehow I had this picture of using a 230/2x6V, but of course that won't work. Silly me
"So if choosing an 8"driver it should have a fs of ~30-35Hz"
Since I seem to be drawn to alu cones (somehow any graph I like seems to come from an alu cone), I was thinking the Dayton RS225S-8 an 8" with an Fs of 27Hz.
I do not have any experience with drivers, so if these would be inappropriate I would very much like to know. Anyone?
For the record: X-over at 300Hz, 8x RS255S in a line source. No extra subwoofer. I would reduce the width of the ESL to 350mm and increase its length to 1800mm. D/S remains at 1.5, wire stator, 20 degrees curved stator.
Hi,
Ok I would further reduce the panel width to 250-300mm.
Stay with the 1800mm length
Make a (good isolated) wire stator, and dont attempt to make a curved one.
a curved design makes it difficult, if not ,almost impossible to get an exact well build stator. You will not get a better panel
It is most important to get an efficient panel, and the only way to get there is to built with tight tollerances. You normally won't achieve that by building a complex curved stator. ( We are talking in 1/10 s of mm's)
One's you have that efficient panel, you can easily choose a lower step up and get better hights and better impedance.
About the line source with 8 woofers, I don't have experience, but you should follow Calvins advice here.
Geert
Ok I would further reduce the panel width to 250-300mm.
Stay with the 1800mm length
Make a (good isolated) wire stator, and dont attempt to make a curved one.
a curved design makes it difficult, if not ,almost impossible to get an exact well build stator. You will not get a better panel
It is most important to get an efficient panel, and the only way to get there is to built with tight tollerances. You normally won't achieve that by building a complex curved stator. ( We are talking in 1/10 s of mm's)
One's you have that efficient panel, you can easily choose a lower step up and get better hights and better impedance.
About the line source with 8 woofers, I don't have experience, but you should follow Calvins advice here.
Geert
Hi,
the Dayton looks good The magnetic shielding is a nice feature if You plan Homecinema usage too.
when You mount them in a dipole-array keep that thing small!!
A width just slightly more than the OD and a depth of app. 7-8", not more! You can use a U or an A dipole. With the A the Fs is lowered a bit (3-5Hz, when the sidewalls are angled at 10°). You can safely do this, because the large number of drivers allows for such a low Fb.
You have to keep it small to reach a upper fs of 300Hz and still have enough distance to the upper resonance. You´ll very probabely have to kill the upper resonance with a notch, a steep crossover filter won´t be enough. Additionally You will have to eq the low base with a High-Qt-subsonic, that pushes the bass around 30-40Hz and cuts it below the fs. This can only be done with an active crossover.
So this concept is a fully active one, don´t waste one thought about a passive solution here!
I wouldn´t change the panel to a taller and thinner one!
Getting less wide means you have to eq the panel more (300-600Hz) or You add wings on every side. With the wider panel You can ´spare´ the wings and use a thin but strong frame to hold the panel. Looks more transparent and pleasing. You dont need the last 10" to 15" down to the floor acoustical wise. So a length of 1600mm is really enough for sitting or standing listening positions. Adding more length doesn´t help in any listening position but increases the capacitance considerably and makes the thing only harder to drive for the amp.
jauu
Calvin
the Dayton looks good
The magnetic shielding is a nice feature if You plan Homecinema usage too.when You mount them in a dipole-array keep that thing small!!
A width just slightly more than the OD and a depth of app. 7-8", not more! You can use a U or an A dipole. With the A the Fs is lowered a bit (3-5Hz, when the sidewalls are angled at 10°). You can safely do this, because the large number of drivers allows for such a low Fb.
You have to keep it small to reach a upper fs of 300Hz and still have enough distance to the upper resonance. You´ll very probabely have to kill the upper resonance with a notch, a steep crossover filter won´t be enough. Additionally You will have to eq the low base with a High-Qt-subsonic, that pushes the bass around 30-40Hz and cuts it below the fs. This can only be done with an active crossover.
So this concept is a fully active one, don´t waste one thought about a passive solution here!
I wouldn´t change the panel to a taller and thinner one!
Getting less wide means you have to eq the panel more (300-600Hz) or You add wings on every side. With the wider panel You can ´spare´ the wings and use a thin but strong frame to hold the panel. Looks more transparent and pleasing. You dont need the last 10" to 15" down to the floor acoustical wise. So a length of 1600mm is really enough for sitting or standing listening positions. Adding more length doesn´t help in any listening position but increases the capacitance considerably and makes the thing only harder to drive for the amp.
jauu
Calvin
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.