The offset like this can create the flat response but only for on axis. The off-axis frequency response is only as good as narrow baffle at the short side (near driver), but worse on the far side of baffle. Considering that what we hear is both on- amd off-axis sound, I am happy to buy one more smaller midrange and use no baffle rather than try to do this offset trick.
Yes you can do that. The edge radius is difficult though, as the bigger the baffle the bigger the radius must be as it needs to work on lower frequency (longer wavelength), which means the baffle must gain thickness with the size in order to accommodate the needed roundover radius. Same goes with closed box speakers. On boxed speakers there likely is enough thickness to the structure, but open baffle there usually isn't.
Basically, fully effective edge radius must start rounding over immediately beside the transducer, otherwise it is not as big as it is needed for full effect. Any round over is better than nothing, but let's stick to idealized thought experiement. 9cm radius roundover means baffle must be 18cm thick to accommodate it, which makes the dipole pattern break in reality, unless it's donut like baffle so that both sides of the woofer turn into waveguides of similar properties, which they do not as the structure is not symmetric with basket and cone shape and all. Very hard to manufacture. Continuing thought experiment, If we start with big ideal donut shape with big roundovers and perfect response, and start to shrink it also the needed roundovers shrink keeping the effectiveness same (keeping frequency response without direction varying interference), and you can shrink it down to transducer size, which makes the baffle and roundover just disappear, which makes no baffle the cheapest and easiest and cheapest baffle ever. But, it increased the cancellation at some particular low frequency as you've been pointing out, which makes the good bandwidth go up in frequency, so no you'd need another driver to cover the lost bass. But on a multiway system we have multiple drivers, one for bass and one for highs, and perhaps some between so it's matter of juggling things around to make a complete system. Also tthe main diffraction hump would be there, but since DI is flat it can be EQ:d flat.
Example from previous post, say I'm thinking making multiway OB system but don't know what it looks like. I could have monopole bass and dipole from 200Hz up. For example. Here few tests I might do:
10" woofer without baffle. Cool cool, nice polar pattern down from about 1kHz, high DI. Good bandwidth can be read from the normalized polar data on bottom right image, or from DI, which is red line on left center image. It would be below the driver beaming / edge diffraction issues, basically where the polar pattern is smooth and response wiggle free (enough) to any angle.
10" driver without baffle, seems it would do it, dipole pattern from ~1kHz down.
6" driver offset on 10" inch baffle seems to do it as well, but it's useful bandwidth is about the same. Interestingly DI is lower in general, with small hump. Simple thought is there is less attenuation of lows but also less attenuation to sides, which might be good or not good thing depending on what's the plan.
15" without baffle, about the same useful bandwidth.
Axial responses overlaid
All the three have more or less same axial output at 200Hz, difference within 2db so all of them have about same usable bandwidth in a system. Big baffle/small driver has bit wider pattern than big driver small baffle. I think it comes down to matter of which one to choose for any reason, cost, aesthetics, manufacturing and that stuff. 15" and 10" driver might have more volume displacement and thus more SPL potential than a 6", if that is relevant for the plan.
Basically, fully effective edge radius must start rounding over immediately beside the transducer, otherwise it is not as big as it is needed for full effect. Any round over is better than nothing, but let's stick to idealized thought experiement. 9cm radius roundover means baffle must be 18cm thick to accommodate it, which makes the dipole pattern break in reality, unless it's donut like baffle so that both sides of the woofer turn into waveguides of similar properties, which they do not as the structure is not symmetric with basket and cone shape and all. Very hard to manufacture. Continuing thought experiment, If we start with big ideal donut shape with big roundovers and perfect response, and start to shrink it also the needed roundovers shrink keeping the effectiveness same (keeping frequency response without direction varying interference), and you can shrink it down to transducer size, which makes the baffle and roundover just disappear, which makes no baffle the cheapest and easiest and cheapest baffle ever. But, it increased the cancellation at some particular low frequency as you've been pointing out, which makes the good bandwidth go up in frequency, so no you'd need another driver to cover the lost bass. But on a multiway system we have multiple drivers, one for bass and one for highs, and perhaps some between so it's matter of juggling things around to make a complete system. Also tthe main diffraction hump would be there, but since DI is flat it can be EQ:d flat.
Example from previous post, say I'm thinking making multiway OB system but don't know what it looks like. I could have monopole bass and dipole from 200Hz up. For example. Here few tests I might do:
10" woofer without baffle. Cool cool, nice polar pattern down from about 1kHz, high DI. Good bandwidth can be read from the normalized polar data on bottom right image, or from DI, which is red line on left center image. It would be below the driver beaming / edge diffraction issues, basically where the polar pattern is smooth and response wiggle free (enough) to any angle.
10" driver without baffle, seems it would do it, dipole pattern from ~1kHz down.
6" driver offset on 10" inch baffle seems to do it as well, but it's useful bandwidth is about the same. Interestingly DI is lower in general, with small hump. Simple thought is there is less attenuation of lows but also less attenuation to sides, which might be good or not good thing depending on what's the plan.
15" without baffle, about the same useful bandwidth.
Axial responses overlaid
All the three have more or less same axial output at 200Hz, difference within 2db so all of them have about same usable bandwidth in a system. Big baffle/small driver has bit wider pattern than big driver small baffle. I think it comes down to matter of which one to choose for any reason, cost, aesthetics, manufacturing and that stuff. 15" and 10" driver might have more volume displacement and thus more SPL potential than a 6", if that is relevant for the plan.
Last edited:
Pure imagination
Since post #34
has same off axis response.
Above directivity.
So yes you just crossover to a smaller driver.
better to hear with ears , not eyes
why even worry about directivity though
since you guys seem to think directivity is what you listen too.
And non directivity needs to be cancelled for some reason.
and again it will magical fall out the same every time.15" without baffle, about the same useful bandwidth.
15" becomes directive around 500 / 600 Hz
So again it is a huge surprise everything is gone below that
Im not going to fix it again for you guys.
Hi,
in context of comparing the sims between each other it's perfectly clear from the data what the directivity is and what the relative performance is. In reality there is more differences due to cone shapes, breakups and so on.
Making an open baffle speaker is about directivity and it's worth it try to perfect it for as wide bandwidth as possible. There is lot of stuff that goes to listening and what matters and what doesn't, perhaps small variations don't and what is small depends on room, prefereces, listening distance and listening skill for example. In my experience small bandwidth deviations are one thing what makes sound of a system identifiable, sound of the system overlaid on what is it you want to listen the music fighting for attention. If you have "perfect" dipole for most of the bandwidth it shouldn't make sound of it's own stick out, thus directivity matters. Small bandwidth phenomena, like directivity mismatch on crossover, resonances, all are small bandwidth phenomena and can take attention from music. Perhaps same goes for diffraction, perhaps wider bandwidth is better.
Trung is right that response is better on one side on offset driver baffle than the other, except got the sides reversed. Edge diffraction makes wave emit backwards where the original sound came from. Thus better response is on the opposite side to close edge, because from driver to edge and back is shorter distance than on the other side the delay is less and bandwidth of interference is higher up in frequency, perhaps out of band if very close to edge. Can be imagined, but is not just imagination, you could check it with simulator and take what you wanna.
edit. There is actually a listening test for edge diffraction audibility hidden: Take a speaker with offset drivers, and listen from say 45 deg angle. Which side sounds better, when the offset driver is closer to you or when its farther? Repeat the test with great listening distance, say at other side of the room, and close listening distance, perhaps at 1-2m. I bet that on the far side of the room there is about no difference, while closer up other could sound better. I'm not sure if I notice any difference with edge diffraction other than when I move around and sound changes, which makes it apparent. Staying still for a while and sound is perhaps fine as long as there is no significant peaks in frequency response, which are readily audible and take attention.
in context of comparing the sims between each other it's perfectly clear from the data what the directivity is and what the relative performance is. In reality there is more differences due to cone shapes, breakups and so on.
Making an open baffle speaker is about directivity and it's worth it try to perfect it for as wide bandwidth as possible. There is lot of stuff that goes to listening and what matters and what doesn't, perhaps small variations don't and what is small depends on room, prefereces, listening distance and listening skill for example. In my experience small bandwidth deviations are one thing what makes sound of a system identifiable, sound of the system overlaid on what is it you want to listen the music fighting for attention. If you have "perfect" dipole for most of the bandwidth it shouldn't make sound of it's own stick out, thus directivity matters. Small bandwidth phenomena, like directivity mismatch on crossover, resonances, all are small bandwidth phenomena and can take attention from music. Perhaps same goes for diffraction, perhaps wider bandwidth is better.
Trung is right that response is better on one side on offset driver baffle than the other, except got the sides reversed. Edge diffraction makes wave emit backwards where the original sound came from. Thus better response is on the opposite side to close edge, because from driver to edge and back is shorter distance than on the other side the delay is less and bandwidth of interference is higher up in frequency, perhaps out of band if very close to edge. Can be imagined, but is not just imagination, you could check it with simulator and take what you wanna.
edit. There is actually a listening test for edge diffraction audibility hidden: Take a speaker with offset drivers, and listen from say 45 deg angle. Which side sounds better, when the offset driver is closer to you or when its farther? Repeat the test with great listening distance, say at other side of the room, and close listening distance, perhaps at 1-2m. I bet that on the far side of the room there is about no difference, while closer up other could sound better. I'm not sure if I notice any difference with edge diffraction other than when I move around and sound changes, which makes it apparent. Staying still for a while and sound is perhaps fine as long as there is no significant peaks in frequency response, which are readily audible and take attention.
Last edited:
Rudolf Finke helped me ten years ago to understand the issue here at diyaudio.com. His homepage is in German, but here is a paper in English too They cover all the issues in this thread very well and in more compact form than at linkwitzlab
http://www.dipolplus.de/index.html
Whitegragon, I really don't understand what you are trying to say in post #43. English is only my second language (for 55 years now), and mostly as written. Please try to understand that you are on an international forum - it is really hard to get the idea/message of your posts
http://www.dipolplus.de/index.html
Whitegragon, I really don't understand what you are trying to say in post #43. English is only my second language (for 55 years now), and mostly as written. Please try to understand that you are on an international forum - it is really hard to get the idea/message of your posts
Last edited:
Directivity is definitely what we are listening in a room, especially in small room (smaller than the auditorium). If you listens outdoor then I agree directivity is not a problem.
frequency response is down right decreased below dipole peak, -6dB/octave, but we can make it flat with a shelving boost filter. Of course we need to sacrifice xmax for it, but that is why we use one 10-12" driver for lower mid (to 80-100Hz) and 2x 12" or one 18" driver for bass. That is a well-known trade off to get an open sound and 3D soundstage in dipole speaker.
Last edited:
I've been doing some ABEC simulations of OB with coaxials and showing results here:
ABEC OB sim thread
There is a steep learning curve to ABEC but ABEC combined with Vituix removes most of the (gu)estimation about how baffle size and shape and driver placement choices work together as a system.
I was concerned with getting dipole characteristic to extend up to the XO to the coax tweeter. Tapering the baffle upwards was all I could do given the coax constraint. One wants a wide baffle at the low end and a narrow or no baffle at the high end.
One thing I realized recently is that directivity at the high end is determined by cone shape. To get the deepest dipole side nulls one has to match the directivity from the inside of the cone with that from the outside of a cone. I simulated several cone depths comparing different drivers and eventually realized that the shallowest cone works best. This stands to reason. If not using a coax, I would definitely use planar mid and tweeters, as others have shown in this thread. However, I want the wide, lobe-free, vertical listening window and simpler construction provided by a coax top end.
ABEC OB sim thread
There is a steep learning curve to ABEC but ABEC combined with Vituix removes most of the (gu)estimation about how baffle size and shape and driver placement choices work together as a system.
I was concerned with getting dipole characteristic to extend up to the XO to the coax tweeter. Tapering the baffle upwards was all I could do given the coax constraint. One wants a wide baffle at the low end and a narrow or no baffle at the high end.
One thing I realized recently is that directivity at the high end is determined by cone shape. To get the deepest dipole side nulls one has to match the directivity from the inside of the cone with that from the outside of a cone. I simulated several cone depths comparing different drivers and eventually realized that the shallowest cone works best. This stands to reason. If not using a coax, I would definitely use planar mid and tweeters, as others have shown in this thread. However, I want the wide, lobe-free, vertical listening window and simpler construction provided by a coax top end.
Haa yes, perhaps there is culture difference that makes some misunderstanding. WhiteDragon are you mostly focusing on axial response? I'm focusing on directivity, because when directivity is fine the rest is adjustable in crossover, but failed to explicitly state that as this is the norm for many years now, since VituixCAD made it possible to hobbyists get full data to good use.
Diffraction for example is something that cannot be fixed other than building better structure, nothing in crossover can touch it. Thus, planning the system so that enables building the structure so that makes good measurements (nice DI) is the way to go, that's what one must do if trying to get best performance.
Instead of shelf boost one can add big series inductor to make one pole of low pass. This would reduce amplifier his and driver motor distortion entering acoustic domain. It necessarily attenuates the signal some and one would just boost it back, just like with shelving filter. The smaller the baffle the lower one can low pass for wider bandpass response. Of course all of it is malleable with DSP. Pperhaps a nothing-burger, thought to mention as it's another dimension how baffle size affects things, passive or hybrid xo.
Example, optimizer is set to show LR2 target filters, adding a coil makes response roughly that way. Bigger baffle has much more sensitivity and less bandwidth just with series inductor. Smaller baffle would need more power from amplifier, perhaps offsetting the distortion reduction from the inductor.
Last edited:
Truly appreciate so detailed and clear an explanation and roadmap -- maybe a career some day for me after retirement ;-) Due to a recent NP outcome (not possible; near perfect) I'm scratching my head to explain 1+1>10 while lining up further candidates for experimentation. About three years ago I had done an "LXmidi" 4/8-inch baffleless-OB/omni-TL fiberglass honeycomb, crossover-less. This time I paired 5/5 "micor55" LX (not OB but possibly could be) and got NP flat response XO-less. So it can't be that hard to diy this thing!?
(edit) To align acoustic centers effectively just up-fire toward the ceiling and adjust until a palpable image floats midair above and beyond the one speaker. I don't know why.
Last edited:
wchang, it is very difficult to localize a sound source below 300Hz in-room, because of so called Schröder effect (late energy and room modes dominate). Midrange and tweeter range can usually be located well, but at listening distance 2-3m they mix together
More about psychoacoustics http://www.davidgriesinger.com/
In history of loudspeakers there are many multiway speakers with down- back- or upfiring bass drivers, but they have never been popular universally. And check car audio installations! In Scandinavia Sonab had many models with upfire or tilted midbass I have a pair of OA-14 from 1980s!
https://www.diyaudio.com/community/...home-claims-stig-carlsson-why-so-rare.388437/
https://carlssonplanet.com/en/
Combining different dispersion sources, they mix together to some degree based on wavelength, distance and xo. Eg. dipole+mono=cardioid
https://musicanddesign.speakerdesign.net/craw_cross.html
More about psychoacoustics http://www.davidgriesinger.com/
In history of loudspeakers there are many multiway speakers with down- back- or upfiring bass drivers, but they have never been popular universally. And check car audio installations! In Scandinavia Sonab had many models with upfire or tilted midbass I have a pair of OA-14 from 1980s!
https://www.diyaudio.com/community/...home-claims-stig-carlsson-why-so-rare.388437/
https://carlssonplanet.com/en/
Combining different dispersion sources, they mix together to some degree based on wavelength, distance and xo. Eg. dipole+mono=cardioid
https://musicanddesign.speakerdesign.net/craw_cross.html
Last edited:
Thank you so much tmuikku for taking the time for such detailed and exhaustive responses.
It lays out, clarify and solidify a lot of my thoughts, that were to this point, mostly assumptions based on empirical testings and semi-understanding of related papers.
No problem, I try to share what I've found out so far. If you find out more, please reply, as I'm here to learn like many others 🙂
Sure, but I don’t think I can add to your knowledge guys.
I can only share my path and it went kind of the opposite way, I mean it started with no baffle mainly for aesthetics reasons.
Then the ease of the build, the quickness to test allowed to iterate a lot, like 20+ variations when I stopped counting.
And on the path to constant directivity this relation driver/baffle size and usable bandwidth became very obvious.
From a simple 2 way w/ horn I ended with 2x18”+12”+4”+amt, and H-frame subs.
I was ready for severe compromises with nude drivers but realized that it was in fact a good starting point.
Well, some extra terrestrial in his home should have a stereo system; maybe kill two birds by making your speakers be a part of the movie set and when they're done filming...
Please share how you supported drivers baffle-less and iterated quickly. (Followed your U-frame thread.) Thanks.
I think opposite, that my tweeter size is smaller. measured with baffle ant it is 62mm wide. According to datasheet NEO 3 is 89mm height.plus some wood. But now i am thinking here two things come into play. Baffle width and radiating area width. Did't thought about that. Need to check on edge simulator just lazy for now 🙂 but if this thinking correct then this is good way for improving horizontal directivity just by rotating tweeter by 90 degrees. i was measuring at 2m distance outside.. Frequency dependant windowing with 1/6.9 octave resolution
no sorry i don't have build thread. Some info is on my site www.baffless.com , but is more nice than DIY friendly 🙂