Hello,
I am fairly novice in my understanding of ELS design. But it seems like the surface area determines how low a frequency the panel can do and the larger the excursion also. ie. high voltages in anti phase on the stators.
Most ELS's seem to have low SPL compared with other speaker types. Anyway, I was wondering what is requirement to produce high SPL electrostats?
Kevin
I am fairly novice in my understanding of ELS design. But it seems like the surface area determines how low a frequency the panel can do and the larger the excursion also. ie. high voltages in anti phase on the stators.
Most ELS's seem to have low SPL compared with other speaker types. Anyway, I was wondering what is requirement to produce high SPL electrostats?
Kevin
I like electrostatic speakers. I have Martin Logan SL3 myself and am quite pleased with them.Simple to get loud ESLs.
First, you feed the bass elsewhere and then you feed the treble elsewhere. Lots of compelling reasons to do so in addition to getting more Bels. This helps both the amp/speaker output in an obvious way but also does statistically remarkably beneficial things to the electric power from the amps when you have two amps instead of one.
Below maybe 150 Hz, you can send the bass to a single speaker almost anywhere in the room (not so close to your ear as to hear the speaker rustling).
First, you feed the bass elsewhere and then you feed the treble elsewhere. Lots of compelling reasons to do so in addition to getting more Bels. This helps both the amp/speaker output in an obvious way but also does statistically remarkably beneficial things to the electric power from the amps when you have two amps instead of one.
Below maybe 150 Hz, you can send the bass to a single speaker almost anywhere in the room (not so close to your ear as to hear the speaker rustling).
Fanuc said:
I think the key is the third word in your second sentence--"seems". Full range ESLs produce quite large SPLs but your perception may be that they do not, especially when compared to point source speakers. I have measured 104dB at 3m with my full range ESLs. Of course you can walk right up to them and place your ear near the panel when they are doing this without cringing in pain. Don't try that with a point source speaker when it is producing 104dB at 3m! Have you ever heard a large, full range ESL? They are loud.
...j
Hi,
the ´secret´ of building high-SPL-ESLs lies in sufficient membrane area and smallest stator-stator-distances and high mechanical diaphragm tension tension. A ESL can´t provide for as much power per unit diaphragm area as a dynamic speaker, so it has to have a larger area.
Since efficiency of the panel is a function of distance (inversely quadratical with rising distances), You should see, that You get higher SPLs with smaller d/s values. It´s a typical beginner´s mistake to design the d/s too large. The best compromises lie between 1.0 and 1.5mm d/s. It´s another very common failure to use low diaphragm tension. Besides placing the resonance at a too low value, it kills output by reducing the d/s considerably.
So what I´m talking here of is a panel for a hybrid ESL. When You measure the distortion of a panel You´ll soon realize that distortion values quickly rise as soon as the membrane needs to execute some excursion. This is the case for frequencies below 200Hz. This is the region where dynamic divers can shine...so leave this freq-range to them. A well executed panel can deliver astonishingly high levels of SPL above this freq-range.
I had my panels measured in an anechoic room last week at Aachen University and got really satisfying results ;-)
(Panel: 125x25cm, 1.1mm d/s, 1.4nF, Fs:120Hz, Tranny: 1:68, pol-voltage: 1.5kV)
Measurements: 93dB@2.83V@4m with distortion values below 0.1% above 200Hz, below 0.15% at 10W@8Ohms and below 0.3% at 50W@8Ohms. The last power value translating to ~110dB@4m and beginning of clipping (since the measurement was fullrange the maximum value of SPL above 200Hz is even higher than that!)
The SPL in a living room is considerably higher that in a anechoic chamber and for correlating signals You can add 6 more dBs of SPL.
These values are high enough to satisfy most demanding needs.
It´ll be no easy task to find a similarly capable bass
The described panel can play in rooms from 20m² size and up.
jauu
Calvin
the ´secret´ of building high-SPL-ESLs lies in sufficient membrane area and smallest stator-stator-distances and high mechanical diaphragm tension tension. A ESL can´t provide for as much power per unit diaphragm area as a dynamic speaker, so it has to have a larger area.
Since efficiency of the panel is a function of distance (inversely quadratical with rising distances), You should see, that You get higher SPLs with smaller d/s values. It´s a typical beginner´s mistake to design the d/s too large. The best compromises lie between 1.0 and 1.5mm d/s. It´s another very common failure to use low diaphragm tension. Besides placing the resonance at a too low value, it kills output by reducing the d/s considerably.
So what I´m talking here of is a panel for a hybrid ESL. When You measure the distortion of a panel You´ll soon realize that distortion values quickly rise as soon as the membrane needs to execute some excursion. This is the case for frequencies below 200Hz. This is the region where dynamic divers can shine...so leave this freq-range to them. A well executed panel can deliver astonishingly high levels of SPL above this freq-range.
I had my panels measured in an anechoic room last week at Aachen University and got really satisfying results ;-)
(Panel: 125x25cm, 1.1mm d/s, 1.4nF, Fs:120Hz, Tranny: 1:68, pol-voltage: 1.5kV)
Measurements: 93dB@2.83V@4m with distortion values below 0.1% above 200Hz, below 0.15% at 10W@8Ohms and below 0.3% at 50W@8Ohms. The last power value translating to ~110dB@4m and beginning of clipping (since the measurement was fullrange the maximum value of SPL above 200Hz is even higher than that!)
The SPL in a living room is considerably higher that in a anechoic chamber and for correlating signals You can add 6 more dBs of SPL.
These values are high enough to satisfy most demanding needs.
It´ll be no easy task to find a similarly capable bass
The described panel can play in rooms from 20m² size and up.
jauu
Calvin
bentoronto said:
Hello,
Yeah this is the arrangement I have used for years. Started with Quad esl57 & 63's and I bought a top of the range B&W active subwoofer. The sub was too damn heavy though!, plus I was moving so I sold it. First time I had heard high SPL uncoloured sub bass before. I thought the drive unit was blown but could shake the windows. Even phoned B&W about it! (think I was naively expecting a bass speaker foolishly not a sub bass speaker, or was listening to to many cheapy car boom boxes subs!)
Anyway, The reason I asked was when I went to a hi fi show in bristol, england I got to listen to Wilson Audio Benesch speakers. (excellent all round speakers though they probably cost the same as a car!). Can't remember the exact model but what struck me was the punchiness to the midrange and high end. The bass is a part of that obviously but there's something more. Have also heard some horn speakers also and they too have a punchiness to cymbals, percussion.
Having said that I still think the ELS's are excellent generally and can do something in midrange and highs in terms of timbre that no other speaker can do. ie moving coil. including the best MC headphones.
Thanks for the replys.
Calvin said:Hi,
the ´secret´ of building high-SPL-ESLs lies in sufficient membrane area and smallest stator-stator-distances and high mechanical diaphragm tension tension. A ESL can´t provide for as much power per unit diaphragm area as a dynamic speaker, so it has to have a larger area.
Since efficiency of the panel is a function of distance (inversely quadratical with rising distances), You should see, that You get higher SPLs with smaller d/s values. It´s a typical beginner´s mistake to design the d/s too large. The best compromises lie between 1.0 and 1.5mm d/s. It´s another very common failure to use low diaphragm tension. Besides placing the resonance at a too low value, it kills output by reducing the d/s considerably.
So what I´m talking here of is a panel for a hybrid ESL. When You measure the distortion of a panel You´ll soon realize that distortion values quickly rise as soon as the membrane needs to execute some excursion. This is the case for frequencies below 200Hz. This is the region where dynamic divers can shine...so leave this freq-range to them. A well executed panel can deliver astonishingly high levels of SPL above this freq-range.
I had my panels measured in an anechoic room last week at Aachen University and got really satisfying results ;-)
(Panel: 125x25cm, 1.1mm d/s, 1.4nF, Fs:120Hz, Tranny: 1:68, pol-voltage: 1.5kV)
Measurements: 93dB@2.83V@4m with distortion values below 0.1% above 200Hz, below 0.15% at 10W@8Ohms and below 0.3% at 50W@8Ohms. The last power value translating to ~110dB@4m and beginning of clipping (since the measurement was fullrange the maximum value of SPL above 200Hz is even higher than that!)
The SPL in a living room is considerably higher that in a anechoic chamber and for correlating signals You can add 6 more dBs of SPL.
These values are high enough to satisfy most demanding needs.
It´ll be no easy task to find a similarly capable bass
The described panel can play in rooms from 20m² size and up.
jauu
Calvin
Excellent reply and excellent work too!. Have always had an urge to look at making an els panel (totally agree on the hybrid approach) but there is no point if the panels turns out to be average. ie. See my post above for my reasoning on this.
Have you found it very hard work to make your panel ? and do you have prior experience making them ?
Kevin
Hi,
actually it was relatively easy to build. But that's a toe loop for a professional skater too
Since I'm interested in ESL-design for roundabout 25years by now and having build different types of panels, I think I roughly know what to do ;-)
The quoted numbers of SPL are very probabely not the end...I´ve got some ideas to raise the SPL even further, to reduce distortion even more and to raise reliability and longevity...but I´ve to test that before I like to go official with it
jauu
Calvin
actually it was relatively easy to build. But that's a toe loop for a professional skater too
Since I'm interested in ESL-design for roundabout 25years by now and having build different types of panels, I think I roughly know what to do ;-)
The quoted numbers of SPL are very probabely not the end...I´ve got some ideas to raise the SPL even further, to reduce distortion even more and to raise reliability and longevity...but I´ve to test that before I like to go official with it
jauu
Calvin
Another way: high voltage bias.
Excuse me if I am repeating unreliable information, but the SPL goes up with some power of the bias voltage. But the bias voltage is limited by the air gap. At some point, and here I reveal my Dayton-Wright "bias," you need the welding gas, SF6 (dunno what, but the "F" is probably something bad for the ozone layer) to prevent arcing. But that is all do-able in your shop, like Dayton-Wright did it, just wrap your panels in dry-cleaners bags.
OK, OK, I admit it... may not be the way to get crystal clear sound. Not sure.
In as much as my study of the subject is very aged, glad to become better informed and freed from error by other posters.
Excuse me if I am repeating unreliable information, but the SPL goes up with some power of the bias voltage. But the bias voltage is limited by the air gap. At some point, and here I reveal my Dayton-Wright "bias," you need the welding gas, SF6 (dunno what, but the "F" is probably something bad for the ozone layer) to prevent arcing. But that is all do-able in your shop, like Dayton-Wright did it, just wrap your panels in dry-cleaners bags.
OK, OK, I admit it... may not be the way to get crystal clear sound. Not sure.
In as much as my study of the subject is very aged, glad to become better informed and freed from error by other posters.
Calvin,
And what about the amplifier driving your panels ? a trannie with a ratio of 1/68 and panel capacitance of 1,4nF give a final capacitance seen by the amplifier about 68*68*1,4e-9 and that is about 6,5uF, that give about 1,22ohm of impedance at 20kHz, that is a significant capacitance load on a normal amplifier, I can design such amplifier, but for it to be reliable, it would be an "overkill monster".
How you deal with it ? Or you simply rely on the fact, that these fequencies have very rarely a significant energy to blow an amp ?
And what about the amplifier driving your panels ? a trannie with a ratio of 1/68 and panel capacitance of 1,4nF give a final capacitance seen by the amplifier about 68*68*1,4e-9 and that is about 6,5uF, that give about 1,22ohm of impedance at 20kHz, that is a significant capacitance load on a normal amplifier, I can design such amplifier, but for it to be reliable, it would be an "overkill monster".
How you deal with it ? Or you simply rely on the fact, that these fequencies have very rarely a significant energy to blow an amp ?
Hi,
indeed, since my panel is of the sheet-metal type with no electrical segmentation, the panel+tranny put some serious stress on most amps, especially those typical feedbacked class-AB-designs with undersized power supplies.
I feed the trannies via a serial resistor of ~0.8Ohms. The reason is that the stray-inductance of the tranny and the capacitance of the panel form a resonant circuit (the resonance frequency lying at >20kHz with a well executed tranny) with a rather high Q. The resistor damps this resonance, thereby linearising the frequency response and easing the stress on the amplifier considerably.
An old Rotel 980BX works quite well in this setup (on of the few affordable amps where the power supply and the associated cabling(!) is as it should be). Even better work class-D designs, since those handle this complex load with more ease and they can deliver larger high freq high current peaks. I´m using a pre-filter feedback design which has just one point to be watched. The output inductance adds to the trannies stray-inductance and as such is reducing the resonance-frequency, hence the bandwidth. It migt be hard to reach 20kHz with such a setup, but since the highs still sound far better with an Electrostat of 15kHz bandwidth than with any dome tweeter, let alone a dynamic fullranger, the reduced bandwidth is just a ´cosmetic flaw´ ;-)
I want to try out some other D-designs with smaller inductances and maybe some post-filter-designs to have the bandwidth limit of >20kHz again ;-)
jauu
Calvin
indeed, since my panel is of the sheet-metal type with no electrical segmentation, the panel+tranny put some serious stress on most amps, especially those typical feedbacked class-AB-designs with undersized power supplies.
I feed the trannies via a serial resistor of ~0.8Ohms. The reason is that the stray-inductance of the tranny and the capacitance of the panel form a resonant circuit (the resonance frequency lying at >20kHz with a well executed tranny) with a rather high Q. The resistor damps this resonance, thereby linearising the frequency response and easing the stress on the amplifier considerably.
An old Rotel 980BX works quite well in this setup (on of the few affordable amps where the power supply and the associated cabling(!) is as it should be). Even better work class-D designs, since those handle this complex load with more ease and they can deliver larger high freq high current peaks. I´m using a pre-filter feedback design which has just one point to be watched. The output inductance adds to the trannies stray-inductance and as such is reducing the resonance-frequency, hence the bandwidth. It migt be hard to reach 20kHz with such a setup, but since the highs still sound far better with an Electrostat of 15kHz bandwidth than with any dome tweeter, let alone a dynamic fullranger, the reduced bandwidth is just a ´cosmetic flaw´ ;-)
I want to try out some other D-designs with smaller inductances and maybe some post-filter-designs to have the bandwidth limit of >20kHz again ;-)
jauu
Calvin
Hi,
the amplimos are really fine stuff. But in my application they achieve just a slightly higher upper bandwidth limit than my toroids. They have clear advantages over the toroids at the low-freq-end, which I don´t need with the hybrid design. Since the toroids proved to be excellent with regard to distortion figures too, I prefer to save money and invest it at other points of interest ;-)
The bandwidth limitation is no sonical problem but just a measurable ´flaw´ and only arises with bad trannies or in conjunction with the output inductance of a prefilter-feedback-D-amp. Usng the toroids with a typical solid state amp with series resistor or a tube amp with elevated output impedance the bandwidth lies well above 20kHz.
The freq-response can be easily tailored by the serial resistance and the achieved linearity up to 15kHz does not leave anything to wish for. Actually the -3dB-frequency lies above 15kHz, but since the -up to this point- very flat response exhibits a clear cut off instead of the usual smooth transition, the 12dB/oct-drop looks worse than it really is. Sonic qualities of the panels are still outstanding.
@ben: I agree that trying to cover fullrange is a very difficult task for dynamic as well as electrostatic speakers. Theoretically it is easy to achieve the goal with ESLs. But the problems lie in size, efficiency and distortion-figures. As long as You keep the needed excursion small -which is the case for a freq-range >150Hz, the mentioned problems reduce vastly and the advantages of a wide bandwidth, box-less driver come through. And there is no dynamic driver system that achieves the levels of naturalism, coherency, resolution, low distortion, dynamics and speed in the mid-highs as a well executed ESL.
jauu
Calvin
the amplimos are really fine stuff. But in my application they achieve just a slightly higher upper bandwidth limit than my toroids. They have clear advantages over the toroids at the low-freq-end, which I don´t need with the hybrid design. Since the toroids proved to be excellent with regard to distortion figures too, I prefer to save money and invest it at other points of interest ;-)
The bandwidth limitation is no sonical problem but just a measurable ´flaw´ and only arises with bad trannies or in conjunction with the output inductance of a prefilter-feedback-D-amp. Usng the toroids with a typical solid state amp with series resistor or a tube amp with elevated output impedance the bandwidth lies well above 20kHz.
The freq-response can be easily tailored by the serial resistance and the achieved linearity up to 15kHz does not leave anything to wish for. Actually the -3dB-frequency lies above 15kHz, but since the -up to this point- very flat response exhibits a clear cut off instead of the usual smooth transition, the 12dB/oct-drop looks worse than it really is. Sonic qualities of the panels are still outstanding.
@ben: I agree that trying to cover fullrange is a very difficult task for dynamic as well as electrostatic speakers. Theoretically it is easy to achieve the goal with ESLs. But the problems lie in size, efficiency and distortion-figures. As long as You keep the needed excursion small -which is the case for a freq-range >150Hz, the mentioned problems reduce vastly and the advantages of a wide bandwidth, box-less driver come through. And there is no dynamic driver system that achieves the levels of naturalism, coherency, resolution, low distortion, dynamics and speed in the mid-highs as a well executed ESL.
jauu
Calvin
Your comments about going wide-range sound sensible and well put.
Some design parameters can be fudged to get satisfactory results. For example, cutting off the bottom at 150, as you suggest or limiting power handling. Most design parameters end up being compromises albeit OK acceptable ones... and, as you say, give me a so-so ESL over a good dynamic any day.
But some design issues such as dispersion or transformer choices (or adding resistors in series) remain recalcitrant.
Ultimately, a matter of personal choices and going for a single panel is a wonderful ideal. I used 6 to a side early-generation Dayton-Wright panels that way on a curved window-frame open box.
BTW, the whole issue of transformers or amps for that matter, can be sidestepped by using a high voltage "Sanders" amp. That was my pride and joy for many years. Currently I am using stock Dayton-Wright transformers... pretty fine units but not nearly as good. I can't urge that people make these amps enough.
Some design parameters can be fudged to get satisfactory results. For example, cutting off the bottom at 150, as you suggest or limiting power handling. Most design parameters end up being compromises albeit OK acceptable ones... and, as you say, give me a so-so ESL over a good dynamic any day.
But some design issues such as dispersion or transformer choices (or adding resistors in series) remain recalcitrant.
Ultimately, a matter of personal choices and going for a single panel is a wonderful ideal. I used 6 to a side early-generation Dayton-Wright panels that way on a curved window-frame open box.
BTW, the whole issue of transformers or amps for that matter, can be sidestepped by using a high voltage "Sanders" amp. That was my pride and joy for many years. Currently I am using stock Dayton-Wright transformers... pretty fine units but not nearly as good. I can't urge that people make these amps enough.
Hi,
the effort of creating a good and somehow efficient HV-amp is still ahead of me ;-)
First I liked to get a top-class-panel up and running with the rather conventional means of a stepup-tranny and classical amp.
And the results are at the moment that good, that it´ll be hard to design an equally well sounding HV-amp (the acoustat-like class-B designs, heavyly relying on strong feedbacking sound worse)
Since my panel is efficient enough to be driven from a rather low HV-supply I´ll probabely try a solidstate-design like Neill McKein´s or similar ones like the Kimmel µ-follower. Those are ccs-loaded class A designs. In case I need more than 1kV, I could get my fingers on a new triode which would allow for 4kV.
jauu
Calvin
the effort of creating a good and somehow efficient HV-amp is still ahead of me ;-)
First I liked to get a top-class-panel up and running with the rather conventional means of a stepup-tranny and classical amp.
And the results are at the moment that good, that it´ll be hard to design an equally well sounding HV-amp (the acoustat-like class-B designs, heavyly relying on strong feedbacking sound worse)
Since my panel is efficient enough to be driven from a rather low HV-supply I´ll probabely try a solidstate-design like Neill McKein´s or similar ones like the Kimmel µ-follower. Those are ccs-loaded class A designs. In case I need more than 1kV, I could get my fingers on a new triode which would allow for 4kV.
jauu
Calvin
Yeah, whole new world of construction thrills when you work with high voltages. Not for the feint hearted. But easy enough once you collect the parts. Took me a while to sort out gremlins in my Sanders-like amp but couldn't be happier when I was using it. Hint: not safe when kids and small animals are in the room.
I used an anode voltage of 2400 volts and the drive swing was like 700 volts rms. This drove my Dayton-Wright panels medium loud and then have, I would guess, fairly big air gaps, although fairly large surface area for 6 combined per side.
I use quite elderly amps and tend not to think there's much difference among electronics (and don't have any special liking for tubes). But now-a-days, not hard to check for basic noise and distortion even with only a test CD and a computer oscilloscope with frequency analysis (others on this list are surely more current on testing that I am). In short, it is feasible to build and test a HV amp.
I used an anode voltage of 2400 volts and the drive swing was like 700 volts rms. This drove my Dayton-Wright panels medium loud and then have, I would guess, fairly big air gaps, although fairly large surface area for 6 combined per side.
I use quite elderly amps and tend not to think there's much difference among electronics (and don't have any special liking for tubes). But now-a-days, not hard to check for basic noise and distortion even with only a test CD and a computer oscilloscope with frequency analysis (others on this list are surely more current on testing that I am). In short, it is feasible to build and test a HV amp.
Hello... must be 23 years ago!
Yes, I had 9 small woofers in series parallel and front-and-back mounted in what looked like a closet. But it was really a door opening in what had been the outside wall of the house before expansion - brick and concrete walls, very large and solid. After all, Klipschorns don't go really low.
That was a previous house and previous wife. Can't say my present house compares sonically speaking to that wonderful beamed-ceiling LR, but the new wife is a vast improvement.
Perhaps you were present for a comparison of driving the Dayton-Wright speakers with a HV amp versus the D-W transformers?
Remind me who you are with a PM, please.
Yes, I had 9 small woofers in series parallel and front-and-back mounted in what looked like a closet. But it was really a door opening in what had been the outside wall of the house before expansion - brick and concrete walls, very large and solid. After all, Klipschorns don't go really low.
That was a previous house and previous wife. Can't say my present house compares sonically speaking to that wonderful beamed-ceiling LR, but the new wife is a vast improvement.
Perhaps you were present for a comparison of driving the Dayton-Wright speakers with a HV amp versus the D-W transformers?
Remind me who you are with a PM, please.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.