I can't help you there because I don't know what a digital DSP does in the timing domain.
It has to delay the ESL to give the dynamic speaker the possibility to catch up?
But the dynamic speaker would be just as slow to return to it's original position so how to solve that? I guess this is always some kind of compromise.
Then there is the flexing of the cone versus the piston of the ESL with totally different radiation patterns. Distortion because of the flexing of the cone.
I'm not saying it sounds horribly bad, because that is subjective and if you like it, great, enjoy it.
It has to delay the ESL to give the dynamic speaker the possibility to catch up?
But the dynamic speaker would be just as slow to return to it's original position so how to solve that? I guess this is always some kind of compromise.
Then there is the flexing of the cone versus the piston of the ESL with totally different radiation patterns. Distortion because of the flexing of the cone.
I'm not saying it sounds horribly bad, because that is subjective and if you like it, great, enjoy it.
Thanks for your input. I think your comment about always some kind of compromises is very pertinent. I imagine digital DSP as well as combining ESL and drivers will have lots of compromises Im unaware of. But one thing they might do well is create a really good impulse. Heres the step response for a DSP controlled multi-way concentric ported driver KEF LSX:
and heres the Quad 2912
So possibly DSP could align the timing really well for well damped drivers and ESL. Then the DSP crossovers can be quickly changed to find what sounds good. Maybe very slow crossovers might work so the ear doesnt hear the ESL vs driver shift?
Anyway for the OP running his ESLs and his dynamics side by side looks like a relatively cheap and quick experiment.
and heres the Quad 2912
So possibly DSP could align the timing really well for well damped drivers and ESL. Then the DSP crossovers can be quickly changed to find what sounds good. Maybe very slow crossovers might work so the ear doesnt hear the ESL vs driver shift?
Anyway for the OP running his ESLs and his dynamics side by side looks like a relatively cheap and quick experiment.
Not sure what I'm looking at. Pulse starts @ zero not at 4 or 52. Also pulse is in % FS, isn't it? At what distance? On axis?
Pulse it set to 100%, then look at step related to that. On axis 1 meter 75 dB. At least in REW it works like that.
Maybe I'm totally wrong. Picture of ESL-63 pulse and step. As said pulse peak is @ 100%.
Pulse it set to 100%, then look at step related to that. On axis 1 meter 75 dB. At least in REW it works like that.
Maybe I'm totally wrong. Picture of ESL-63 pulse and step. As said pulse peak is @ 100%.
I dont worry about the graph x-axis timing of the commencement of the step. Its arbitrary./relative/ADC latency influenced/distance effect. Using a DSP it can be set for anywhere you like so with a measurement mic at the listening position I guess the idea is to exactly align the arrival of the impulses of the various drivers. It looks like magic when it can make a multiway cone speaker step response look like an ESL.
Thanks for your suggestion - I tried this with my ripole subwoofers but apart from the esthetical impact, I don't like the alteration of sound quality around the x-over frequency. I get more pressure in the lowest ocatave - but just above this area it seems the sound quality degrades a litte. I didn't correct in the time domain (because I have no experience to do that), so maybe I need to learn and try that.
Hi again,
I built a set of ripoles last year and connected them again. I added a 6 dB/oct dipole slope correction and response peakc correction using my MiniDSP 2x4 HD. I used Linkwitz site for directions (https://www.linkwitzlab.com/models.htm)
I used a cross-over frequency of 65 Hz. (I tried different frequencies). This gave the best results. Some recordings sounded better with the subs powered on. But a few songs which had a lot of repetive bass notes sounded too bassy (one note bass). Maybe there were still some problems in my filter, timing, or woofer placement. (I put both ripoled just next to my stats). Or maybe woofer level setting was not ok. But overall the woofers still added too much LF energy to my taste. It wasn't bad, but it was not as good as I've read from other reports of DIY-ers that built their own ripoles. Quality wise, I still prefered my stats without the ripole woofers, except a few recordings.
So after years of trying different LF extenstions to my stats, including sealed and ported woofers and ripoles, and transmission line woofers, I don't like them sound quality wise. I'm not sure if this is due to the nature of the woofer / stat combination, or due to lack of insight / errors made by me.
I was aware of the (dipole) line array woofer, but never considered it seriously as it would require many drivers and would thus be a costly experiment. The ripole drivers (4 units) were already around 500 euro's and currently I don't use them. But yesterday I came up with the idea of mounting all four of them in 1 large baffle that has about the same dimensions as my stats. I can make 1 OB line array woofer for channel: four 12 inch woofers will fit perfectly in the 150 cm tall panels. I can equalize them with my DSP and it would be a rather simple and cheap operation to construct just one channel to get an impression of it's capablitities.
I did another experiment as well yesterday, just to get an idea of my initial idea: doubling the stats height or width in order to get better LF extension. I just put 2 stats next to each other and compared output at different distances (2 and 4.3 meters) and compared it to one single stat. It seems that doubling their size boosts LF extension, but only at closer distances: at my current sofa location, 4 meters, the effect is gone. This gives me the idea that doubling the height of my stats won't improve LF extension at larger listening distances (> 2 meters)...
This motivates me to try the dipole line array experiment...
Below are frequency respones of in room measurements:
brown line: both panels next to each other, mic at 2 meters from stats
green line: both panels next to each other, mic at 4.3 meters from stats
yellow line: 1 panel at 4.3 meters from stat
I built a set of ripoles last year and connected them again. I added a 6 dB/oct dipole slope correction and response peakc correction using my MiniDSP 2x4 HD. I used Linkwitz site for directions (https://www.linkwitzlab.com/models.htm)
I used a cross-over frequency of 65 Hz. (I tried different frequencies). This gave the best results. Some recordings sounded better with the subs powered on. But a few songs which had a lot of repetive bass notes sounded too bassy (one note bass). Maybe there were still some problems in my filter, timing, or woofer placement. (I put both ripoled just next to my stats). Or maybe woofer level setting was not ok. But overall the woofers still added too much LF energy to my taste. It wasn't bad, but it was not as good as I've read from other reports of DIY-ers that built their own ripoles. Quality wise, I still prefered my stats without the ripole woofers, except a few recordings.
So after years of trying different LF extenstions to my stats, including sealed and ported woofers and ripoles, and transmission line woofers, I don't like them sound quality wise. I'm not sure if this is due to the nature of the woofer / stat combination, or due to lack of insight / errors made by me.
I was aware of the (dipole) line array woofer, but never considered it seriously as it would require many drivers and would thus be a costly experiment. The ripole drivers (4 units) were already around 500 euro's and currently I don't use them. But yesterday I came up with the idea of mounting all four of them in 1 large baffle that has about the same dimensions as my stats. I can make 1 OB line array woofer for channel: four 12 inch woofers will fit perfectly in the 150 cm tall panels. I can equalize them with my DSP and it would be a rather simple and cheap operation to construct just one channel to get an impression of it's capablitities.
I did another experiment as well yesterday, just to get an idea of my initial idea: doubling the stats height or width in order to get better LF extension. I just put 2 stats next to each other and compared output at different distances (2 and 4.3 meters) and compared it to one single stat. It seems that doubling their size boosts LF extension, but only at closer distances: at my current sofa location, 4 meters, the effect is gone. This gives me the idea that doubling the height of my stats won't improve LF extension at larger listening distances (> 2 meters)...
This motivates me to try the dipole line array experiment...
Below are frequency respones of in room measurements:
brown line: both panels next to each other, mic at 2 meters from stats
green line: both panels next to each other, mic at 4.3 meters from stats
yellow line: 1 panel at 4.3 meters from stat
This is a great plan.
Do you want me to do a hornresp OB sim? If you do just post the panel dimensions with the brand and model of the woofers.
Thanks for sharing your experiments in trying to get the bass augmented for ESLs. Its such a common issue.
Theoretically it may be a big OB with DSP is going to be best of all the different bass systems you have tried as:
- the large bass OB shares the dipole output and dimensions of the ESL - intuitively its a match
- the plain OB can do a similar impulse to the ESL
- the plain OB can reach up to 250Hz and higher to fill in the upper bass for the bass punch experience and also the low mids for warmth (your single panel at 4m might sound a bit thin with the 100Hz-250Hz lower SPL. That would bother me much more then the sub 70hz deficit.
- with DSP you can time align, do a very gentle crossover so the woofers and ESL can blend and get the frequency-SPL curve sorted
- By unloading the panels below 250Hz they might play much louder and cleaner
- combined with good lower mid fullness and bass and you will have a whole new system
Please can you try the double height stack and measure it for us if its not too hard? DO you have some clamps to set them up?
Stacking vertically should be totally different to horizontal stacking.
Thanks for sharing your adventures
Have you tried sealed subwoofers with DSP room correction? A few of the commercially available ones: JL Fathom, Martin Logan Dynamo 600-1600, Martin Logan BalancedForce, SVS 3000 Micro. Equalizing out the room modes tends to improve subwoofer integration.
https://www.stereophile.com/content/jl-audio-fathom-f212-powered-subwoofer-page-4
Just as I heard with a pair of Fathom f113s, two Fathom f212s used in stereo deepened and broadened the soundstage. The f212s blended so seamlessly with my ESL-989s that the subs seemed to disappear
https://www.stereophile.com/content/velodyne-digital-drive-plus-18-subwoofer-page-2
Hi,
with the first protos of my ESL and esl I utilized 8pcs of 22cm resp. 17cm drivers in a dipole tower arrangement.
The drivers were made by the same company and came from the same series of drivers ... hence I expected same sonic footprint but just more deep base extension from the larger drivers.
And while the larger tower indeed played lower the sound was less ´involving´ (corrected down to ~25Hz).
The smaller drivers performed easily audible more lively, with an astonishingly dynamic sounding kick bass (a real pusher, corrected down to ~35Hz).
This then lead to the development of the ESL MK2 and esl MK2, where both ESLs were partnered with basically the same dipole bass tower with the same 8pcs, resp. 6pcs. of the smaller drivers, corrected down to ~50Hz. A dedicated subwoofer working only below 50Hz finalized the system down to ~16Hz.
So after my experience it seems not trivial to mate even a large ESL panel with large sized dynamic woofers, regardles of the working principles, though dipole and CB appear the most logic.
Of course the lower You cross over the better .... if You can keep far enough from the ESL´s base resonance.
Still though an array of smaller mid-bass drivers seems to be the better match.
Building hybrid ESLs only I need to xover the panel anyways.
But I´m quite certain, that a subwoofer assisted panel profits from xovering also .... to control membrane excursion and to control the high-Q long-decay base resonance and -probabely the most important of all- to achieve symmetrical acoustic filter flanks of the xover branches.
The cost of a smaller driver dipole tower may even not exceed that of a large driver subwoofer, since there are so many suitable and low-cost drivers in the 15-20cm class around.
When I built the proof-of-concept ´blue Klaus´ I used cheap drivers with very promising parameters costing less than 20€/pcs.
The result was far better than any single 160€+ driver could possibly have performed.
The drivers I finally used in the ESL/esl MK2 costed almost double that, but improved matters greatly (~double excursion, lower THD, better build quality and looks).
And while the towers are quite high (160, resp. 120cm) their footprint is tiny, fitting on a DIN-A5 page of paper.
jauu
Calvin
with the first protos of my ESL and esl I utilized 8pcs of 22cm resp. 17cm drivers in a dipole tower arrangement.
The drivers were made by the same company and came from the same series of drivers ... hence I expected same sonic footprint but just more deep base extension from the larger drivers.
And while the larger tower indeed played lower the sound was less ´involving´ (corrected down to ~25Hz).
The smaller drivers performed easily audible more lively, with an astonishingly dynamic sounding kick bass (a real pusher, corrected down to ~35Hz).
This then lead to the development of the ESL MK2 and esl MK2, where both ESLs were partnered with basically the same dipole bass tower with the same 8pcs, resp. 6pcs. of the smaller drivers, corrected down to ~50Hz. A dedicated subwoofer working only below 50Hz finalized the system down to ~16Hz.
So after my experience it seems not trivial to mate even a large ESL panel with large sized dynamic woofers, regardles of the working principles, though dipole and CB appear the most logic.
Of course the lower You cross over the better .... if You can keep far enough from the ESL´s base resonance.
Still though an array of smaller mid-bass drivers seems to be the better match.
Building hybrid ESLs only I need to xover the panel anyways.
But I´m quite certain, that a subwoofer assisted panel profits from xovering also .... to control membrane excursion and to control the high-Q long-decay base resonance and -probabely the most important of all- to achieve symmetrical acoustic filter flanks of the xover branches.
The cost of a smaller driver dipole tower may even not exceed that of a large driver subwoofer, since there are so many suitable and low-cost drivers in the 15-20cm class around.
When I built the proof-of-concept ´blue Klaus´ I used cheap drivers with very promising parameters costing less than 20€/pcs.
The result was far better than any single 160€+ driver could possibly have performed.
The drivers I finally used in the ESL/esl MK2 costed almost double that, but improved matters greatly (~double excursion, lower THD, better build quality and looks).
And while the towers are quite high (160, resp. 120cm) their footprint is tiny, fitting on a DIN-A5 page of paper.
jauu
Calvin
This is interesting. Did you have explanation why the smaller woofers sounded "more lively" ?
Hi,
it could well have been only a matter of the working frequency range of 25-180Hz of the larger tower versus 35-220Hz of the smaller tower.
But since the ´next step´ included a dedicated subwoofer anyway, I opted for the same smaller drivers for both systems.
The sub added the authority and slam ´down under´ that a dipole sub leaves missing.
The smaller tower shape improved optics also
jauu
Calvin
it could well have been only a matter of the working frequency range of 25-180Hz of the larger tower versus 35-220Hz of the smaller tower.
But since the ´next step´ included a dedicated subwoofer anyway, I opted for the same smaller drivers for both systems.
The sub added the authority and slam ´down under´ that a dipole sub leaves missing.
The smaller tower shape improved optics also
jauu
Calvin
Thats a good bandwidth. For that what driver parameters would you rouhgly suggest:
- diameter:
- QTs:
- Xmax:
- Fs:
This is a very important point in that you built one large one using a set of small panels, if these panels were also of different sizes, it would be even better. And the point is the same basic resonance that we are discussing here. For excessively large panels, the main resonance is very low, somewhere below 30, or even 20 Hz., But the main resonance of the room (small and medium) “works” in this range, so the peak in this range multiplies and becomes unbearable. But the problem is that all acoustic musical instruments are mainly involved in the mid-bass register, and this is 60-160Hz, but the ESL has very problems with this in general, and large panels in particular. Therefore, by dialing small panels with the main resonance of the mid-bass range, you can greatly help in solving this problem. That is, to give the sound of a drum, double bass and other more elastic, like a stretched trampoline, character. Which will be more consistent with the nature of the sound of these acoustic musical instruments. This should be considered as another option besides isobaric.
And a few more words about the subwoofer and love for it. Connecting a subwoofer to an electrostatic in the lowest range greatly spoils the overall sound, in my opinion. If you make an additional crutch to electrostatics, then it should be a speaker with a light diffuser and not emphasize too low a frequency range, otherwise the entire low-frequency region will be amorphous, slow and stretched out in time, but a beater hitting a drum, a dynamic plucking of a double bass will obviously not get.
Last edited:
Hi,
I admit to have difficulties to understand Your point.
I agree with You in that the frequency range from 60-160Hz is of great importance, as here most of the base frequencies of voices and instruments are found.
As such the typical dip in amplitude response of fullrangers like Acoustats, Audiostatics et al lead to an anaemic powerless presentation with missing authenticity and authority.
A well made hybrid is vastly superior here.
I also agree on subwoofers in the way they are typically 'mated' with their satellite speakers.
Most subs I find are crossed over too high in frequency.
I disagree on Your general critique on Subs though.
Well executed a subwoofer improves the sound experience.
jauu
Calvin
I admit to have difficulties to understand Your point.
I agree with You in that the frequency range from 60-160Hz is of great importance, as here most of the base frequencies of voices and instruments are found.
As such the typical dip in amplitude response of fullrangers like Acoustats, Audiostatics et al lead to an anaemic powerless presentation with missing authenticity and authority.
A well made hybrid is vastly superior here.
I also agree on subwoofers in the way they are typically 'mated' with their satellite speakers.
Most subs I find are crossed over too high in frequency.
I disagree on Your general critique on Subs though.
Well executed a subwoofer improves the sound experience.
jauu
Calvin
Thanks for your message!This is a great plan.
Do you want me to do a hornresp OB sim? If you do just post the panel dimensions with the brand and model of the woofers.
Thanks for sharing your experiments in trying to get the bass augmented for ESLs. Its such a common issue.
Theoretically it may be a big OB with DSP is going to be best of all the different bass systems you have tried as:
- the large bass OB shares the dipole output and dimensions of the ESL - intuitively its a match
- the plain OB can do a similar impulse to the ESL
- the plain OB can reach up to 250Hz and higher to fill in the upper bass for the bass punch experience and also the low mids for warmth (your single panel at 4m might sound a bit thin with the 100Hz-250Hz lower SPL. That would bother me much more then the sub 70hz deficit.
- with DSP you can time align, do a very gentle crossover so the woofers and ESL can blend and get the frequency-SPL curve sorted
- By unloading the panels below 250Hz they might play much louder and cleaner
- combined with good lower mid fullness and bass and you will have a whole new system
Please can you try the double height stack and measure it for us if its not too hard? DO you have some clamps to set them up?
Stacking vertically should be totally different to horizontal stacking.
Thanks for sharing your adventures
I would love to get a hornresp OB sim. Panel dimensions would be: 35 cm (width), 150 cm (height). These are the actual dimensions of my stat panels and the four drivers I own are Peerless SLS-315, datasheet: see attachement 830669.pdf.
I will try to stack the panels and do new measurements similar to the previous ones. Not sure if I can do it as it will be more difficult to stack them. But I will try it.
I think my current woofers won't play well up to 250 Hz, I think it will be more like up to 100 Hz...
Thanks for your reply!
I tried only two sealed subwoofers, but didn't like the result. But it could be me making errors (placement, filters).
@Calvin and @havun: thanks for sharing your interesting experiences(!).
This warns me for a disappointing OB line array woofer when using larger drivers like the current four Peerless SLS-315 I own. Nonetheless I would like to implement at least one channel of those as I own them anyway and it would take relatively little time, money and effort to implement it, so at least I have some personal experience with them. But it is nice to be 'warned'
Thanks also for sharing ideas about alternative drivers and advised specs. It seems I have to test this (smaller OB line array mid/woofer) as well in future. This would make the current 'full range' trannies a bit overkill as they are built for lower frequencies whereas a higher x-over would opt for the trannies to have lower step up ratio and more HF range rather than LF range.
I will consider trying to stack my current stats also, just to try to test all options I have. It has been a long journey to reach acceptable LF performance...
This warns me for a disappointing OB line array woofer when using larger drivers like the current four Peerless SLS-315 I own. Nonetheless I would like to implement at least one channel of those as I own them anyway and it would take relatively little time, money and effort to implement it, so at least I have some personal experience with them. But it is nice to be 'warned'
Thanks also for sharing ideas about alternative drivers and advised specs. It seems I have to test this (smaller OB line array mid/woofer) as well in future. This would make the current 'full range' trannies a bit overkill as they are built for lower frequencies whereas a higher x-over would opt for the trannies to have lower step up ratio and more HF range rather than LF range.
I will consider trying to stack my current stats also, just to try to test all options I have. It has been a long journey to reach acceptable LF performance...
Last edited: