Hi MarkusG,
May I suggest that you go ahead and make the ESLs first? You might be surprised just how low a pair of ESLs can go. If you are going to biamp your hybrid panels, then you can really worry about the crossing frequency later. I'm sure that once you've listened to them, you will know right away from where you should put the crossover.
There are so many factors affecting the frequencies responses of each panel and each pair of ESLs might just need different cross over frequency, IMHO though.
Wachara C.
May I suggest that you go ahead and make the ESLs first? You might be surprised just how low a pair of ESLs can go. If you are going to biamp your hybrid panels, then you can really worry about the crossing frequency later. I'm sure that once you've listened to them, you will know right away from where you should put the crossover.
There are so many factors affecting the frequencies responses of each panel and each pair of ESLs might just need different cross over frequency, IMHO though.
Wachara C.
Markus,
I haven’t tried heat shrinking but I’ve had good results using 1/16” thick x .75 wide foam tape holding 6 micron polyester mechanically stretched to 1.5% elongation (which I think is more tension than you would typically get with heat shrinking) and the panels have been playing fine since July 2008. The shear load on the tape will result in a slight amount of creep and slight tension loss but it hasn’t been a problem on my panels to date. The shear load would likely be less with heat shrinked film so creep shouldn't be a problem. And, if you wrap the diaphragm film over the edge of the stator and secure it with tape on the opposite face also, as Calvin suggested in one of his threads, that should eliminate any creep and keep full tension on the film indefinitely. If you do use tape, use only 3M brand polyurethane foam. The 1/16” x .3/4” wide or 1/16 x 1” wide is perfect for hybrid panels or 1/8” thick for full range.
Good luck with your project!
Charlie
I've been experimenting for fun in hifi and professionally as a researcher for 50 years. I take for granted nothing works right the first time. It almost always is a throw-away trial. People with an engineering frame of mind (I generally use the term to illustrate "Brand X" or naive boobies) suppose that you can look up stuff in a handbook and make it right; might work for real simple things like designing paper clips.
My over-riding concern is doing it. It is relatively easy to make a four little panels crossing over at 1000 Hz using some plastic frames you find at the bargain store in the cookingware section along with 5-cents worth of SaranWrap and odds and ends from your transformer storage box. (And conductive paint, graphite for locks, high voltage high resistance resistors, power supply ladder pieces, high voltage test-lead wire, hair dryer to shrink the SaranWrap... just a few things more.)
Had you said, "Right. Now I will begin" two weeks ago... by today, you'd be into building Model 2, crossing over at 300 Hz and maybe using dipole woofers.
And I'd give the same advice for any lurkers too timid to get going making ESL speakers.
And Chinsettawong says roughly the same thing.
Last edited:
Now it makes more sense, thanks. 🙂
I think I'm ready to start planning and hunting for materials now.
I think I'm ready to start planning and hunting for materials now.
.... or you'd know enough by that point about the construction challenges, like the sound enough, and have the knowledge to buy some old beat-up commercial ESLs and modify them, like a skilled DIYer, to your liking.
I'm not sure what the 3M number is. I'm even holding a roll of it in my hand right now and I don't see a part number on it, just a 3M logo. My guess is that it's either 4016 or 4026. I buy it online from McMaster-Carr, as McMaster-Carr part #7626A114 and it's described as multi-purpose double sided polyurethane foam tape, .063" x 3/4" x 36yd roll. (I went with 3/4" rather than 1" width on my latest panels in order to minimize the stray capacitance).
Just so you know... foam tape significantly simplifies panel construction but it almost certainly would not give the same durability as solid plastic spacers because it deteriorates over time due to ozone and UV attack. I use 3M tape exclusively because it's formulated to resist attack and the other/cheaper brands may not be as durable. Even if I have to rebuild my panels some years down the road, the ease of construction makes foam tape worth it to me.
Hi,
I suggest You visit 3M´s Website. Theres alot of info about their tapes, because there are tapes and then there are tapes. As with every glue or adhesive, the features of the to-fix partners define the type of adhesive/glue. There are tapes specialized for low surface energy partners and tapes for high surface energy partners.
The sometimes bad reputation of tapes and glues may have its reason simply in the all to often usage of the wrong tape or glue for a certain application.
I´ve made good experiences with the 3M tapes though they are indeed quite costly. This relativates because of the ease, speed and cleanliness of handling.
There are two drawbacks to consider.
1) You have only one attempt to mate diaphragm and stator - no room for correction or failures.
2) I have not been able to get the stuff off of a stator for repair reasons. 3M themselves have no procedure to get the stuff off of a surface apart from a lot of heat and serious rubbing (You won´t believe how big of a bubble Your thumb may develop). If anyone has a quick, cheap and easy way to get the tape off of a stator please tell us.
On the pro-side these tapes really stick to the right partner like mud to the proverbial boot. The available thickness range is just perfect for hybrid panels with rather small d/s values. I wouldn´t recommend d/s values of more than 1.5mm, because of the considerable shear forces that stress the tapes. Thicker tapes of course give in more than thinner tapes, which may release diaphragm tension too much.
The constancy of thickness is very good. Spacers from hard plastics are no better, some, like acrylics are typically worse. The softness of the tapes allows contact strips to be a bit ´embedded´ into to the tape. With hard plastic spacers the thickness of the contact strip adds to the d/s.
The softness of the tapes insures that the membrane won´t rip along the glue edges. Hard spacer´s edges may cut the membrane.
The tapes can take quite alot of shear force if sufficient glue area is provided.
The glueing force is quite high from the beginning (app. 90% of the final value). It allows quick building. The diaphragm can be released from the stretching frame almost immediately after the mating.
3M offers highly transparent clear tapes which are perfect as ´visible´ spacers or dampers, but don´t glue as well as the black, grey or white foam tapes and they are a bit more flexible.
The soft character of the tapes introduces alot of damping. When I knock on my panels they give a rather low-pitched well dampened sound whicle the same done on the bigM´s panels gives a more high-pitched rattling sound.
Get Your tape rather from a 3M dealer or distributor, since most electronic distributors just sell the standard types in typical widths and thickness ranges. Very probabely You need different.
jauu
Calvin
I suggest You visit 3M´s Website. Theres alot of info about their tapes, because there are tapes and then there are tapes. As with every glue or adhesive, the features of the to-fix partners define the type of adhesive/glue. There are tapes specialized for low surface energy partners and tapes for high surface energy partners.
The sometimes bad reputation of tapes and glues may have its reason simply in the all to often usage of the wrong tape or glue for a certain application.
I´ve made good experiences with the 3M tapes though they are indeed quite costly. This relativates because of the ease, speed and cleanliness of handling.
There are two drawbacks to consider.
1) You have only one attempt to mate diaphragm and stator - no room for correction or failures.
2) I have not been able to get the stuff off of a stator for repair reasons. 3M themselves have no procedure to get the stuff off of a surface apart from a lot of heat and serious rubbing (You won´t believe how big of a bubble Your thumb may develop). If anyone has a quick, cheap and easy way to get the tape off of a stator please tell us.
On the pro-side these tapes really stick to the right partner like mud to the proverbial boot. The available thickness range is just perfect for hybrid panels with rather small d/s values. I wouldn´t recommend d/s values of more than 1.5mm, because of the considerable shear forces that stress the tapes. Thicker tapes of course give in more than thinner tapes, which may release diaphragm tension too much.
The constancy of thickness is very good. Spacers from hard plastics are no better, some, like acrylics are typically worse. The softness of the tapes allows contact strips to be a bit ´embedded´ into to the tape. With hard plastic spacers the thickness of the contact strip adds to the d/s.
The softness of the tapes insures that the membrane won´t rip along the glue edges. Hard spacer´s edges may cut the membrane.
The tapes can take quite alot of shear force if sufficient glue area is provided.
The glueing force is quite high from the beginning (app. 90% of the final value). It allows quick building. The diaphragm can be released from the stretching frame almost immediately after the mating.
3M offers highly transparent clear tapes which are perfect as ´visible´ spacers or dampers, but don´t glue as well as the black, grey or white foam tapes and they are a bit more flexible.
The soft character of the tapes introduces alot of damping. When I knock on my panels they give a rather low-pitched well dampened sound whicle the same done on the bigM´s panels gives a more high-pitched rattling sound.
Get Your tape rather from a 3M dealer or distributor, since most electronic distributors just sell the standard types in typical widths and thickness ranges. Very probabely You need different.
jauu
Calvin
Absolutely right! Trying to remove 3M foam tape from stators once the adhesive has set, is VERY frustrating. I did have some luck using the thick textured rubber thumb covers used by postal workers. The best procedure I found was to use a stiff plastic bristle brush to remove all the foam layer of the tape, leaving only the thin adhesive layer on the stator. Then, I would apply toluene to the adhesive layer and let it sit overnight. The next day, the adhesive layer was much easier to peel off.
Do you think the difference in resonance sounds has mostly to do with your choice of tapes? or perhaps you use a thinner diaphragm material than "bigM" ?
Hi Calvin,
3M offers several 1/16" urethane tapes... is there a particular 3M part number that you would recommend?
thanks,
Charlie
Hi,
as I wrote, the choice of the optimal tape depends on the glue partners.
This is mostly the PET film on one side and a resin or plastic or laquer on the other the stator side. Besides 42 😉 there´s no onefitsall number recommendation.
The number system of 3M also depends on the thickness measurement. Same tape in inches has a different number than in mm.
jauu
Calvin
as I wrote, the choice of the optimal tape depends on the glue partners.
This is mostly the PET film on one side and a resin or plastic or laquer on the other the stator side. Besides 42 😉 there´s no onefitsall number recommendation.
The number system of 3M also depends on the thickness measurement. Same tape in inches has a different number than in mm.
jauu
Calvin
Hello Calvin,
Very impressive "small" system. 🙂
I had a couple questions which you may choose to answer if you don't consider the information proprietary to your design.
ESL resonance
1) You state the crossover is ~250Hz. What frequency is the fundamental diaphragm resonance? and how do you deal with it? Do you use it to partially equalize the dipole roll off? based on your past posts, I assume you probably use a notch filter.
2) With the woofer tower placed so close the the ESL, do you have much problem with the acoustic output from the woofer exciting the ESL fundamental resonance? Does the U-frame or A-frame help with this issue?
Woofer radiation pattern match to ESL radiation
1)As you have mentioned, ideally you want the radiation pattern of the woofer array to match that of the ESL. The ESL radiation pattern is dipole in nature at low frequencies. Have you compared your "small" system using an H-frame woofer array vs the A-frame woofer array? I understand the H-frame is less visually appealing, but its radiation pattern is dipole in nature at all frequencies up to the crossover with the ESL, while the A-frame is not. This may not be nearly as important as using a line source woofer array to match the SPL vs distance behavior of the ESL.
With the U-frame or A-frame baffle shape, the front and rear loading of the woofers is not the same so asymmetries in the radiation pattern result . In particular, dipole behavior is only seen at very low frequencies. As frequency increases the nulls off to the side disappear. At the 250Hz crossover point, the radiation pattern of a U-frame does not match that of the ESL very well, although your woofer array still behaves as a line source which may be the more important factor.
2)Have you experimented with adding damping inside the cavity of your A-frame woofer array?
With proper damping, the radiation pattern of a U-frame or A-frame woofer can approach a cardioid. This may be advantageous when dealing with placement close to a rear wall.
For those unfamiliar with H-frame and U-frame radiation patterns, you might check out John K's informative website.
DIY-dipole-1
Last edited:
Hi,
The panel resonance Fs is around 150Hz. So it´s not part of the working frequency range. Its dealt with by electronic means in the active crossover, i.e a notch filter.
Placing a woofer in close proximity to a panel always bears the potential that the diaphragm may be blown into the stators when the woofers excurse. Dipolar woofers by the acoustic phase cancellation fan a lot of air around the cabinet. The easiest measurement is to have a ´slot´ and a bit of distance between panel and woofer cabinet. This allows for the air to circle around the casing without influencing the panel overdue. Higher Fs of the panel helps a bit too in reducing the effect.
I haven´t tried an H-frame because I find this style of casing rightout too big and way too ugly. If I can get similar or even better results with something that pleases my eyes far more, why should I bother? 😉
Yes, the A-frame introduces a asymmetry, but to my experience this is in praxis not an as serious effect as JK´s simulations describe. You have to keep in mind too, that our systems is much smaller in dimensions, which shifts the effects up in frequency. When we had our system measured in an big anechoic room there was unfortunately not enough time to do extensive measurements on many different angles, but just on a few different angles.
The nulling effect of the dipole perpendicular to its main axis was clearly audible though, so for us the proof of the effect was already in the listening. If You consider the sensitivity of the human hearing in that low-frequency range, such a distinct recognizable effect requires several dBs of SPL-difference. Side-note: The amplifier-rig was positioned right between the ESLs and I was close to killing the system, when I propped up the volume because it sounded too soft. Well, taking just one step back and I was smacked in the face with Discotheka-level plus. 🙄
Under normal listening conditions with reflecting walls the effect is much less audible. One should keep in mind that either dipole or cardioid cease working as true such in a real listening room situation. What remains are the small compact size and the low group delay figures.
I do use a bit of stuffing, but mainly to reduce the Q of the cavity resonance a bit. Theres hardly any damping in this frequency range because of the low effectiveness of typical wadding. Considerable damping effects come into effect rather only above 200Hz.
jauu
Calvin
The panel resonance Fs is around 150Hz. So it´s not part of the working frequency range. Its dealt with by electronic means in the active crossover, i.e a notch filter.
Placing a woofer in close proximity to a panel always bears the potential that the diaphragm may be blown into the stators when the woofers excurse. Dipolar woofers by the acoustic phase cancellation fan a lot of air around the cabinet. The easiest measurement is to have a ´slot´ and a bit of distance between panel and woofer cabinet. This allows for the air to circle around the casing without influencing the panel overdue. Higher Fs of the panel helps a bit too in reducing the effect.
I haven´t tried an H-frame because I find this style of casing rightout too big and way too ugly. If I can get similar or even better results with something that pleases my eyes far more, why should I bother? 😉
Yes, the A-frame introduces a asymmetry, but to my experience this is in praxis not an as serious effect as JK´s simulations describe. You have to keep in mind too, that our systems is much smaller in dimensions, which shifts the effects up in frequency. When we had our system measured in an big anechoic room there was unfortunately not enough time to do extensive measurements on many different angles, but just on a few different angles.
The nulling effect of the dipole perpendicular to its main axis was clearly audible though, so for us the proof of the effect was already in the listening. If You consider the sensitivity of the human hearing in that low-frequency range, such a distinct recognizable effect requires several dBs of SPL-difference. Side-note: The amplifier-rig was positioned right between the ESLs and I was close to killing the system, when I propped up the volume because it sounded too soft. Well, taking just one step back and I was smacked in the face with Discotheka-level plus. 🙄
Under normal listening conditions with reflecting walls the effect is much less audible. One should keep in mind that either dipole or cardioid cease working as true such in a real listening room situation. What remains are the small compact size and the low group delay figures.
I do use a bit of stuffing, but mainly to reduce the Q of the cavity resonance a bit. Theres hardly any damping in this frequency range because of the low effectiveness of typical wadding. Considerable damping effects come into effect rather only above 200Hz.
jauu
Calvin
I had experimented with a dipole woofer array placed next to my ESL panel and even with a 6 inch gap between them there was considerable excitation of the diaphragm resonance. But, it just struck me that the resonance of my panels were much lower than yours...about 70Hz. I guess it makes sense that the dipole woofer array would be "fanning" more air past the ESL the lower the frequency is. I will have to try the experiment again sometime with a higher resonance panel.
Measurements on my H-frame and U-frame woofers matched JohnK's simulations and measurements very well, including the loss of dipole radiation pattern at higher frequencies with the U-frame. But, as you just brought to my attention, the length of your A-frame baffle is considerably less than what I used for my experiments. Mine were 12 inches long, and it looks like your are about half that. So, the dipole radiation pattern of your woofer array would be nearly intact all the way to the crossover frequency. Very nice 🙂
I was using the longer sides to improve the output capability at low (<50Hz) frequencies. For this application, ugly or not, the H-frame has the advantage that the cavity resonances are higher than a U-frame for the same path length front to back. I see for your "big" system you add a sealed subwoofer to help out with these lowest frequencies.
Is there any particular reason you use stuffing to reduce the Q of the cavity resonance rather than use a notch filter the way you do for the ESL diaphragm resonance?
Last edited:
Hi,
In fact I use a notch in the crossover for two reasons.
1) to get rid of the left over peaking response
2) to shape the acoustic flank of the crossover response to symmetrically match the ESL´s.
In this design certain parameters coincided so well that I could use one notch twicefold.
jauu
Calvin
Who told You I didn´t use electrical notching too? 😉
In fact I use a notch in the crossover for two reasons.
1) to get rid of the left over peaking response
2) to shape the acoustic flank of the crossover response to symmetrically match the ESL´s.
In this design certain parameters coincided so well that I could use one notch twicefold.
jauu
Calvin
Yeah, I guess I didn't ask my question very well. What I was getting at, was why you bothered to use some damping material to reduce the resonance peak in the A-frame cavity when you obviously have no problems using notch filters to do this.
After thinking about it a bit more, I realized that using a notch filter would affect both the response of the front and rear equally. This is fine when you are dealing with an H-frame which has a cavity induced response peak in the front and rear. But, with your A-frame it would be advantageous to use damping material in the cavity as it would reduce the response peak in the rear without substantially affecting the front response. This would minimize the required "notching".
Funny, I thought dipole enthusiasts didn't care about flapping cones or diaphragms, Doppler distortion, or irregularities of many sorts arising from big excursions found in dipoles?
A notch filter controls the input signal but has no influence on the flopping and intermodulation arising from acoustic/mechanical excitation of the natural resonance of the cone or diaphragm or the severe non-linearity of off-center diaphrams.
A notch filter controls the input signal but has no influence on the flopping and intermodulation arising from acoustic/mechanical excitation of the natural resonance of the cone or diaphragm or the severe non-linearity of off-center diaphrams.
Hi,
well, if You want You can dig for a hair in every soup. 😛
Doppler distortion and irregularities of many sorts can be found in any electromechanic transducer using membranes. Those inherent to dipoles are not necessarily worse than that of others.
First measurement is to keep the bandwidth small, the second is to keep the lower bandwidth limit high to reduce excursion demands, third is to supply for as much membrane area to keep excursion demands low.
A subwoofer or a rather low crossed over hybrid features low bandwidth, thereby fulfilling measurement No. 1.
A dipole doesn´t pressurize a room. Especially in larger rooms a dipole leaves something to be desired and is less suited to it. A dedicated Subwoofer of a different style is advantageous here.
Using a subwoofer frees the dipolar midbass from high excursions, since with rising frequency the SPL of a dipole not only rises, but eventually even surpasses the SPL of a monopolar source. So if a reasonable low lower bandwidth limit is chosen, the dipole is just a little or not at all inferior to a monopole. Measurement No.2 fulfilled.
Stacking dipoles increases membrane area, thereby fulfilling measurement No. 3 and allows for shaping the distribution character similar to a strip-shaped panel.
In short, respecting a dipole´s limitations and designing accordingly, leads to similar or even better results as with other cabinets. Distortion values need not be higher than with other solutions (the audibility of low distortions is very questionable anyway).
Excursion demands depend on frequency and level and can be higher as well as lower with an dipole.
Since the dipole builds very compact, one can put a lot more membrane area into the same volume.
My small system features 6 drivers with ~125cm² membrane area in a cabinet of ~27L volume. The resulting 750cm² are close to that of a 15" driver. I put the lower bandwidth limit at ~35Hz. The lower bandwidth limit of the big system with its dedicated subwoofer is at ~60Hz. The 8 drivers feature ~1000cm² of membrane area, which is somewhere between that of a 15" and 18" driver. The cabinet volume measures ~33L.
But while a driver in a comparatively small closed Box (BR would simply be impossible) features a very high Fb (in built resonance), a high Qtb and needs alot of excursion generating equalization to reach a low F-3, the Fb of the dipole even drops slightly against the free-air Fs and Qtb rises only slightly against the free-air Qtc. Of course does the dipole also need electrical equing, but while the overall Q of the system might be the same in both cases, its my experience that the mechanical generated Q is more audible than the electrically generated Q.
As explanation: If we want the system to behave like a highpass with a Q of say 0.7 and the Qtb (mechanical Q) of the built-in driver differs, we can trim the Q with electronical means. The resulting Q will be the product of mechanical Qtb and the electrical Qel of the filter.
Now the Qtb of a dipole is roughly the driver´s free-air Qtc, typically below 0.5. The Qtb of a smalll closed box may be 1.0. In the dipole case we need a equing filter with a Q of 1.4 in the CB-case we need a Q of 0.7 (0.5*1.4=1.0*0.7=0.7).
The impulse response of both systems will be same as both systems feature a resulting Q of 0.7. But sonically the lower Qtb of the dipole combined with a higher electrical Qel sounds better to my ears than the higher Qtb of the CB combined with a lower electrical Qel.
jauu
Calvin
well, if You want You can dig for a hair in every soup. 😛
Doppler distortion and irregularities of many sorts can be found in any electromechanic transducer using membranes. Those inherent to dipoles are not necessarily worse than that of others.
First measurement is to keep the bandwidth small, the second is to keep the lower bandwidth limit high to reduce excursion demands, third is to supply for as much membrane area to keep excursion demands low.
A subwoofer or a rather low crossed over hybrid features low bandwidth, thereby fulfilling measurement No. 1.
A dipole doesn´t pressurize a room. Especially in larger rooms a dipole leaves something to be desired and is less suited to it. A dedicated Subwoofer of a different style is advantageous here.
Using a subwoofer frees the dipolar midbass from high excursions, since with rising frequency the SPL of a dipole not only rises, but eventually even surpasses the SPL of a monopolar source. So if a reasonable low lower bandwidth limit is chosen, the dipole is just a little or not at all inferior to a monopole. Measurement No.2 fulfilled.
Stacking dipoles increases membrane area, thereby fulfilling measurement No. 3 and allows for shaping the distribution character similar to a strip-shaped panel.
In short, respecting a dipole´s limitations and designing accordingly, leads to similar or even better results as with other cabinets. Distortion values need not be higher than with other solutions (the audibility of low distortions is very questionable anyway).
Excursion demands depend on frequency and level and can be higher as well as lower with an dipole.
Since the dipole builds very compact, one can put a lot more membrane area into the same volume.
My small system features 6 drivers with ~125cm² membrane area in a cabinet of ~27L volume. The resulting 750cm² are close to that of a 15" driver. I put the lower bandwidth limit at ~35Hz. The lower bandwidth limit of the big system with its dedicated subwoofer is at ~60Hz. The 8 drivers feature ~1000cm² of membrane area, which is somewhere between that of a 15" and 18" driver. The cabinet volume measures ~33L.
But while a driver in a comparatively small closed Box (BR would simply be impossible) features a very high Fb (in built resonance), a high Qtb and needs alot of excursion generating equalization to reach a low F-3, the Fb of the dipole even drops slightly against the free-air Fs and Qtb rises only slightly against the free-air Qtc. Of course does the dipole also need electrical equing, but while the overall Q of the system might be the same in both cases, its my experience that the mechanical generated Q is more audible than the electrically generated Q.
As explanation: If we want the system to behave like a highpass with a Q of say 0.7 and the Qtb (mechanical Q) of the built-in driver differs, we can trim the Q with electronical means. The resulting Q will be the product of mechanical Qtb and the electrical Qel of the filter.
Now the Qtb of a dipole is roughly the driver´s free-air Qtc, typically below 0.5. The Qtb of a smalll closed box may be 1.0. In the dipole case we need a equing filter with a Q of 1.4 in the CB-case we need a Q of 0.7 (0.5*1.4=1.0*0.7=0.7).
The impulse response of both systems will be same as both systems feature a resulting Q of 0.7. But sonically the lower Qtb of the dipole combined with a higher electrical Qel sounds better to my ears than the higher Qtb of the CB combined with a lower electrical Qel.
jauu
Calvin
- Status
- Not open for further replies.