The Finnish company Gradient produced a small OB "stand" for the '57s. It has four small size long throw speakers, passive crossover and some equalization (if I remember correctly). I've just seen it on photo. Anybody has more close experience with it?
I would even say an enclosure and accurate bass are mutually exclusive.
I am using a rebuild 'official' Gradient with the '63-s. It complements them well, both in form factor and in sound.
Jan
I can't help but be skeptical of new members like yourself... Anyway, you bring up a good point which is the issue of compressed or flat or greyscale sounding loud and complex music reproduced through Quads or other electrostatic speakers. Reminding me why I got rid of the Quads. I presume the limitation is the impractical amount of voltage needed to achieve proper low frequency excursion across a widened stator gap and using a fragile membrane when attempting to achieve midfield levels amplitudes below 100hz. Even a pair of servo driven Modified Acoustat 3s fail in this regard which are more dynamic though not nearly as nuanced as the Quad 57s. Perhaps efficiency can be improved with contemporary more conductive coating technology and or fully conductive membranes and such though a more practical approach for midfield listening is a 300hz+ electrostatic driver with conventional drivers handling bass and midbass. Quad 57s with a highpass crossover above ~200hz crossed to 12s or 15s would allow for better bass performance and presumably much higher spl capability.
The issue of observed compression and lack of dynamism is not inherent to electrostatic drivers by any means, contrary to what many state. The Stax headphones that I own driven by a mid tier solidstate amp are the most open, clear and dynamic listening experience that I've experienced. There is a volume limitation in this instance, which is indicated by mechanical failure noises when max excursion is exceeded. I never would listen at those levels regardless of whether or not they exceeded the capability of the driver.
No, not even perhaps, but quite definitely no! At the risk of upsetting you ... This is a naive, and maybe current assumption. In ESLs, increasing surface resistance of the coating only leads to more low frequency distortion, because of electrostatic load migration over time and related to the membrane-to-stator distance. A correct coating for best results would have to provide a surface resistance of some 50 MOhms ... 200 (500) MOhms. Varying these values does not change SPL at all, but the time it needs to fully load the membrane. The higher to value, the longer it takes for an ESL to be ready to reproduce sound. You may read more in depth by googleing the terms.
To expand a bit on that, we're talking about electrostatic speakers. Basically it's ideal if the charge on the membranes stays in place once charged. The coating is a compromise, you need some sort of conductance to spread the charge over the membrane but once charged, you would like the electrons to stay right where they are and not move about at all.
So far I don't know of any "electret" applications, I suppose if that would be feasible it would quite popular.
So far I don't know of any "electret" applications, I suppose if that would be feasible it would quite popular.
You're on the right track.
Imho, the biggest issue with ESL's is that they are essentially a mass-spring system with a very high Q resonance. The spring force being the membrane tension, which is needed to provide a centering force. The mass being the membrane mass plus the mass of the air coupled to it. The air mass swamps the membrane mass completely, so membrane mass is of no importance at low frequencies (but it is at high freqs).
In an open ESL design there is hardly any damping force, so the untreated resonance has a very high Q (10 or higher). This results in a sharp narrow peak in the low frequency response that can easliy be as high as 15-20dB. The common practice to deal with this is to introduce some sort of damping to lower and widen the peak. Plate stator design can introduce air resistance using smaller hole sizes, Quad uses in addition damping cloth on the rear stator on the 63 and damping material in the enclosure of the 57, Audiostatic uses silcone dots and in the later models an intricate rear frame design to break up the resonance and dampen their ESL's.
This damping lowers the Q making the peak wider and lower. This resulting bump in the low end response is then used in full range ESL's to partially compensate for the low end dipole roll-off, extending the low end frequency response of the speaker to acceptable levels. This requires carefull tuning of the resonance frequency, reason why many diy rebuilds give dissappointing results.
However, the damping methods have an effect on higher frequencies as well, resulting in loss of detail and clarity. Even worse, the Q can not be lowered to values that are truly needed for accurate bass without sacrificing too much output. This is because of practical limitations to drive voltages and the isolation properties of air, the achievable driving forces are simple too small to allow enough damping for a Q<1.
One of the most heavily damped ESL's is the Quad 63 (and it's successors), which reaches a Q of around 2.5. For this they are really pushing the drive voltages to the limit (stepup 1:250), so much ionizating detecting is required to protect the speaker. I do not have accurate data on the 57's damping, but I expect it to be less than in the 63. Audiostatics have much better stator insulation and can push the stepup to 1:300 for low frequencies. They use less damping than Quad, I once measured the Q in the 4-6 range on an ES200 panel but that was not a very scientific measurement tbh. Anyway, these Q values are all a long cry from the 0,5-0,7 that is needed for accurate bass reproduction. As a result all fullrange ESL's that I know of suffer from a resonant bass that is not very natural. The bass is very nice as long as the resonance is not triggered, but as soon as the notes enter the resonant frequency range you can hear the effect. The membrane flapping around also severely lowers the max output.
The fragility of the membrane you mention is not a problem, since the membrane is driven uniformly over it's entire surface. The problem is the insufficient driving force that does not allow for accurate low frequency reproduction with a Q<1. The lack of driving force is also why a servo system can't offer a solution either.
In my opinion, the only way around this inherent problem is not to use ESL's in the frequency range of their resonance. Which implies crossing over to a woofer system with a Q in the 0,5-0,6 range for the lowest frequencies. With active filtering or DSP it's easy to place a notch on the ESL resonance and use steep filter slopes, allowing a crossover frequency as low as one octave above Fres. Resulting in crossover freqs around 100Hz, which allows for seamless integration with a good quality (OB) woofer.
Last edited:
I'm using my 57s supported with 2x12" Ripoles, and I have been contemplating upping the crossover frequency, currently 70Hz (24db/oct). This sounds like you're telling me 90Hz would be a much better choice? (I don't want to get too high either, Ripoles have a nasty resonance peak which I think is about 170Hz with my dimensions).
You may need to place a notch on the ripole's resonance, less than one octave from the crossover point is usually trouble. Just try what sounds best.
My dipole woofer has a first resonance peak around 220Hz and it did benefit from a notch on that, with a 90Hz crossover point. That was using 3e order analog filtering though.
My dipole woofer has a first resonance peak around 220Hz and it did benefit from a notch on that, with a 90Hz crossover point. That was using 3e order analog filtering though.
Probably the exact route I will be going - what do you think of the Ripoles, and which drivers do you use?
I am looking at these very affordable drivers, having used many other Beyma units and being very impressed:
Beyma 12BR70
I'm using these: https://loudspeakerdatabase.com/Peerless/SLS-P830669 and used @CharlieM 's (Jazzman) simpler version instructions to build them (see http://jazzman-esl-page.blogspot.com/2011/01/ripole-subs-are-underway.html).
I got the dimensions from @Calvin with a front chamber width of 80mm and rear chamber width of 95mm, height 346mm (inner dimensions). I used 22mm black MDF which is probably huge overkill.
I glued together the rear chambers and fitted the speakers, then made the U form for the front chamber and put it together using the 4 threaded rods, without glue, so they can still be taken apart if need be. With my woodworking skill level that was the best option as I couldn't possibly make them as nice as @CharlieM 's.
As for the experience, I don't even "know they are there". Only the 57's seem to go 2-3 octaves lower.
NB: I do have an oddly shaped living room which allows me to face the rears of the speakers into relatively free space. It's quite essential for the 57s to have room behind them, even much more so when you add Ripoles. Don't shove them into a corner! 😳
I got the dimensions from @Calvin with a front chamber width of 80mm and rear chamber width of 95mm, height 346mm (inner dimensions). I used 22mm black MDF which is probably huge overkill.
I glued together the rear chambers and fitted the speakers, then made the U form for the front chamber and put it together using the 4 threaded rods, without glue, so they can still be taken apart if need be. With my woodworking skill level that was the best option as I couldn't possibly make them as nice as @CharlieM 's.
As for the experience, I don't even "know they are there". Only the 57's seem to go 2-3 octaves lower.
NB: I do have an oddly shaped living room which allows me to face the rears of the speakers into relatively free space. It's quite essential for the 57s to have room behind them, even much more so when you add Ripoles. Don't shove them into a corner! 😳
About the issues of membrane resonance and stored energy: These are valable points. Also an ESL will show stored energy. The membrane action will stimulate resonances on the supporting framework also, which will show an energy decay over time. Both "smearing" bass response.
This will be the test bed for the following examples:
It shows a single and already completely reworked element.
First a passive test, by strongly mechanically knocking at the center of the element, shows some major resonances of the mechanical structure at around 20Hz ... 30Hz. Instead, the energy decay at around 90Hz is the membrane resonance.
The membrane resonance might be much more pronounced if there was no membrane damping. Damping within ESL63s is partially performed by a mesh on the inner side or one single stator. So one stator is meshed, the other isn't. This is what the meshed stator looks like:
I unfortunately cannot remember how I generated the following graph. But it cleary shows that membane resonance might be a potential problem, indeed:
Now no more knocking, but applying a regular swept sine to all four panels, fully mounted and miked very closely: As mentionned before by some forists, the measurement clearly shows the low frequency peaking because of the membrane resonance. This peaking, also as mentionned before, will partially compensate for the dipole SPL resonse falloff towards lower frequencies at a regular listening distance.
So what about improving "sub" response for an ESL setup? All in all the idea of X-overing away the lowest octave(s) of an ESL might be feasibly approach, supplementing the ESP by a dedicated suitable bass transducer. Eventually one could even omit damping the ESL membrane altogether in such a system, if the resonance would be below the passband of the ESL (but then ... what about the passive stimulation of the resonance peak by the "outside" bass soundwaves from the sub ?). In any case, the sub/bass system then must match the quality of the ESL. I doubt that any "standard" magnetodynamic approach will do so, even a dipole/ripole one. This having been said, and naively stated, I wonder why bass horns have not been mentionned until now as candidate gutshakers? Shure, they are monopoles an not really handy, but ???
This will be the test bed for the following examples:
It shows a single and already completely reworked element.
First a passive test, by strongly mechanically knocking at the center of the element, shows some major resonances of the mechanical structure at around 20Hz ... 30Hz. Instead, the energy decay at around 90Hz is the membrane resonance.
The membrane resonance might be much more pronounced if there was no membrane damping. Damping within ESL63s is partially performed by a mesh on the inner side or one single stator. So one stator is meshed, the other isn't. This is what the meshed stator looks like:
I unfortunately cannot remember how I generated the following graph. But it cleary shows that membane resonance might be a potential problem, indeed:
Now no more knocking, but applying a regular swept sine to all four panels, fully mounted and miked very closely: As mentionned before by some forists, the measurement clearly shows the low frequency peaking because of the membrane resonance. This peaking, also as mentionned before, will partially compensate for the dipole SPL resonse falloff towards lower frequencies at a regular listening distance.
So what about improving "sub" response for an ESL setup? All in all the idea of X-overing away the lowest octave(s) of an ESL might be feasibly approach, supplementing the ESP by a dedicated suitable bass transducer. Eventually one could even omit damping the ESL membrane altogether in such a system, if the resonance would be below the passband of the ESL (but then ... what about the passive stimulation of the resonance peak by the "outside" bass soundwaves from the sub ?). In any case, the sub/bass system then must match the quality of the ESL. I doubt that any "standard" magnetodynamic approach will do so, even a dipole/ripole one. This having been said, and naively stated, I wonder why bass horns have not been mentionned until now as candidate gutshakers? Shure, they are monopoles an not really handy, but ???
Last edited:
The 63's frame is terrible flimsy, they benefit a lot from improving on that. Same can be said about the Audiostatic 50 and 100 models.
Nice measurements btw.
True. No ripole/dipole system will "match" the low frequency reproduction of an ESL. Which is a good thing, because ESL's are just terrible at it. A good dipole/ripole will therefore easily outperform any ESL by miles in the lowest octave. And as Advis already mentioned, with decent implemented crossovers you will never even know they are there. Not to mention the improvement in the performance of the ESL by releasing it from a task it cannot perform properly: less excursion is lower IM, more headroom and less load on the amplifier.
Take it from a guy who has been listening to a dipole/esl combo for 20+ years: once you hear what such a system is capable of, there is no way you will ever appreciate fullrange ESL's again.
But unfortunately there is this persistent belief among lots of people that since ESL's are better than magnetodynamics in many aspects, they must be better in the bass as well. No, they're not. Period.
Nice measurements btw.
True. No ripole/dipole system will "match" the low frequency reproduction of an ESL. Which is a good thing, because ESL's are just terrible at it. A good dipole/ripole will therefore easily outperform any ESL by miles in the lowest octave. And as Advis already mentioned, with decent implemented crossovers you will never even know they are there. Not to mention the improvement in the performance of the ESL by releasing it from a task it cannot perform properly: less excursion is lower IM, more headroom and less load on the amplifier.
Take it from a guy who has been listening to a dipole/esl combo for 20+ years: once you hear what such a system is capable of, there is no way you will ever appreciate fullrange ESL's again.
But unfortunately there is this persistent belief among lots of people that since ESL's are better than magnetodynamics in many aspects, they must be better in the bass as well. No, they're not. Period.
If we stretch 2 micron thick mylar on a 30-40 cm frame and bring a measuring microphone to it and tap the membrane with a finger, we will get a peak of the main resonance of approximately 25-30 Hz depending on the degree of membrane tension, but!!! if we take the microphone and move it to a distance of 1-2 m and tap the membrane again, then we will not only hear anything, but also will not see anything that the microphone can measure.
Hence the question - do we need this resonance, should we keep it in mind?
This is not at all the "drum" that is used in a brass or symphony orchestra.
An electrostatic has no problems with voicing the sub-bass, given that any room has its own main resonance and this room will happily pick up and double, or even triple this main resonance in the region of 25-35 Hz depending on the size.
The problem of electrostatics is in creating the necessary sound pressure in the range of playing bass instruments such as double bass, bass guitar and all kinds of drums, but they do NOT work in the sub-bass area, but work in the bass and mid-bass range, that is, in the range of 50-120 Hz approximately and it is in this range that electrostatics have huge problems, it cannot instantly move decent masses of air, it does not have the strength for this. I wrote about this in my topic.
Abolutely right! I have used ESL 63 with active woofers since 1994, first transmission line (TL6) with KEF B139, and then with dipole subs (different DIY) since 2001. And I agree 100% with maudio about the low frequency improvements, but the midrange and the dynamics also gets better. The headroom increases by approx 10dB, the magneto hysteresis in the step up transformer decreases resulting in less IM distortion. And of course that extra octave is also welcome when it comes to room perspective.
I can claim 47 yrs with large panels, some of those years with Dayton-Wright units and some with DIY panels made from DW cells.
I thought you always start by doing frequency runs with existing speakers to see their response and distortion. Nice that these days everybody knows how essential it is to have a DSP unit, even just for R&D.
Long-standing debates about one or stereo or dual-mono woofers. For sure, in terms of limited budgets (aren't all budgets limited?) best to have one great woofer.
The 40,000 rules seems important to me: upper Hz capability multiplied by lower Hz should be 40,000. So a great system can be harmed by adding only bass or only treble. This applies to modifying the stock ESL57s which are pretty balanced to start with (as has been commented previously) even if only so-so at either extreme.
For sure, no bass ever added more glory to my panels than a Klipschorn. Almost all the virtues you are looking for come with a horn (except super-low bass).
I thought you always start by doing frequency runs with existing speakers to see their response and distortion. Nice that these days everybody knows how essential it is to have a DSP unit, even just for R&D.
Long-standing debates about one or stereo or dual-mono woofers. For sure, in terms of limited budgets (aren't all budgets limited?) best to have one great woofer.
The 40,000 rules seems important to me: upper Hz capability multiplied by lower Hz should be 40,000. So a great system can be harmed by adding only bass or only treble. This applies to modifying the stock ESL57s which are pretty balanced to start with (as has been commented previously) even if only so-so at either extreme.
For sure, no bass ever added more glory to my panels than a Klipschorn. Almost all the virtues you are looking for come with a horn (except super-low bass).
Ooops, make that 400,000.
And unless going for a horn, harder to get a single woofer to have good performance in a small room, except maybe for a single seat.
And unless going for a horn, harder to get a single woofer to have good performance in a small room, except maybe for a single seat.