gedlee said:
Dr. Geddes that one confuses me. Would that be from a front view or a view from above?
Because, if it's from above (a half donut) , that has been my ideal outcome as this has come together. Especially when combined with the "vitamin pill" Lynn designed years ago. Basically, cutting the back half of the "vitamin pill" off? It would not be difficult to router the shape and edge glue a baffle to this kind of round shape. I would use Carbon Fiber and a very stern verticle support for the drivers at the rear. All to be held in place by a thick layer of 3M's Marine Polyurethane sealant/adhesive (5200). This stuff is soft (forever) and almost impossible to separate. I don't care what size driver you have it can hold it. They use it on fiberglass boats and it virtually delaminates the fiberglass hulls when they try to take it apart.
If it is a front view there is at least one speaker that does this. I don't recall who it is. It is only the LF driver so I'm thinking it doesn't match up to what you're proposing.
Ok, with that in mind what are you suggesting? If you're speaking of a frontal view, then, a spearate "chamber", if you will, would be designed for each respective driver, is this what you mean? If so just how close together would they need to be? What about baffle step correction? This would become really easy, but would it solve your "time delay" issue?
Teh
I remember the PowerPoint presentation where Dr. Geddes is presenting HOMs attenuation visible on impulse response tail. The effect is delicate.
It is interesting for me that in Manger's house we used bipolar arrangement (do you remember donut Diskus?) and room interaction was negligible. There was no fuzzy imaging normally associated with omni radiation. We used normal and binaural signals. In the newest AES convention paper Christian Langen from Schoeps wrote that Manger is able to mimic best condenser mic response (near Dirac) and in this case the others can be deconvoluted.
Of course dynamic range is fine thing to have if you are alone
It is interesting for me that in Manger's house we used bipolar arrangement (do you remember donut Diskus?) and room interaction was negligible. There was no fuzzy imaging normally associated with omni radiation. We used normal and binaural signals. In the newest AES convention paper Christian Langen from Schoeps wrote that Manger is able to mimic best condenser mic response (near Dirac) and in this case the others can be deconvoluted.
Of course dynamic range is fine thing to have if you are alone
gedlee said:
Yep, tried it but didn't like it. With open alignments you have to obtain different driver to edge differences around the baffle. With a fixed single distance delay for the rear wave, you maximize the classic dipole ripples above the baffle cutoff.
Some say that an infinitely thin baffle is the ideal, so edge diffraction of the front and rear waves totally cancel. I've explored that along with big radius roundovers, but haven't been satisfied with anything yet. The problem is when you do a big roundover for the front side, you have to figure out what to do with regard to the rear wave.
My front and rear 90° conical waveguides with 2" radius roundovers aren't too far from the donut shape and the square shape along with some rear wave attenuation is sufficient for me wrt dipole ripples, but I'm getting significant vibration in my solid wood roundovers that I've yet to do anything about. The foam inside the waveguide did wonders, so thanks for sharing that about your Summa's.
gedlee said:
I want to draw everyone's attention here to what Earl is saying. This bears very close reading and analysis.
If I understand right, we're dealing with a 3-dimensional multipath problem with a level-sensitive detector. Not that dissimilar to distortions experienced in long-distance transmission paths with NTSC and stereo-FM systems, where the subcarriers are badly degraded by multipath. We don't have the option of developing PAL and SECAM
- we're stuck with the perceptual systems we were born with.The distinction between the minimum-phase behaviour of reflections and non-minimum phase behaviour of diffraction is extremely significant, especially regarding the different ways they are perceived.
I should bring to the attention of the reader that distortions of this nature may be impossible to correct with DSP, since moving only a few inches at the listening position may have a completely different set of time dispersion distortions. This is not a minor point - a mis-aligned DSP "corrector" could greatly aggravate the existing time dispersion of a horn system.
Combine the severe HOM's of standard theater CD horns with almost univerally-used DSP, and it could be a real mess - in the time domain. The freq resp is more-or-less flat, but the time response in the first 2 mSec ain't gonna be pretty.
Attachments
What's not shown is how the time response changes with microphone position. Any changes in time signature with spatial angle renders DSP correction counterproductive.
Why? When the DSP is misaligned, you get entirely new classes of time dispersion that don't exist in the original uncorrected speaker system. A misaligned DSP is a worst-of-all-worlds scenario - and if the DSP correction is only good for a extremely small listening area, it is worse than useless, it'll actually degrade the sound.
If HOM's are highly directional, it makes all but the most minimal high-frequency DSP correction undesirable. What's appalling is that CD-horn arrays + automatic DSP is the way pretty much all prosound is done these days, movie theaters included. No wonder I preferred the sound of old-school multicells with only the mildest of FR correction.
In his own quiet way, I think Earl has just pulled the rug out from the prosound world.
Why? When the DSP is misaligned, you get entirely new classes of time dispersion that don't exist in the original uncorrected speaker system. A misaligned DSP is a worst-of-all-worlds scenario - and if the DSP correction is only good for a extremely small listening area, it is worse than useless, it'll actually degrade the sound.
If HOM's are highly directional, it makes all but the most minimal high-frequency DSP correction undesirable. What's appalling is that CD-horn arrays + automatic DSP is the way pretty much all prosound is done these days, movie theaters included. No wonder I preferred the sound of old-school multicells with only the mildest of FR correction.
In his own quiet way, I think Earl has just pulled the rug out from the prosound world.
Lynn Olson said:
Lynn
Quite perceptive and quite correct. I have been trying to make this point over and over again, how DSPactually makes things worse in a real sound field NOT better, but no one, until now, seems to have gotten it.
There simply is no way to correct these aberations with anything but a diffraction free design. And when you hear such a design you will know what has been missing.
I don't usually make subjective claims and so I'll hedge this one with some real data, albeit NOT psychoacoustic.
My partner took my class in loudspeakers and he was struck by my knowledge of psychoacoustics NOT loudspeakers - he knew that I knew that stuff. He lives in Thailand (a NY'er) and flew here one week to talk to me about loudspeakers and to hear my system. Kenny is an ex-NY-recording-engineer and producer (he did Jaco Pastorius stuff), and he was a recording engineer and producer in Thailand for a number of years (anyone who has ever been to Bangkok knows why he stayed there. I wish I had known about Bangkok when I was young and single!!) He now designs and builds recording studios, nightclubs and concert halls in Asia. He has built some of the best recording studios on that side of the world. Kenny knows good sound.
After he heard my system he offered me a partnership in a new company in which he has dumped multiple 100,000's into. He has bet this kind of money on what he heard at my home. Say what you will he obviously thought that the sound was "pretty good". We are exstartic about the stuff that we are building as it is exceding all expectations.
You see, we use NO electronic processing at all, we do virtually everything acoustically. We designed out all diffraction and built several sub designs since the ESP (Enhanced Sound Perception)line with multiple staggered subs (another innovation) has everyone who hears it simply blown away.
It may be some time before we can import into the US (because its so expensive - like $1,000,000 just to get the first shipment in the door.) But all are welcome to Bangkok to hear the systems at our factory.
Once again Lynn superb catch on realizing the implications of our research.
gedlee said:
I've also tried it - and dido. IMO The problem with such a design is the cavity for the driver, not the baffle edge.
Now it could be, just like a good horn design, that a good dipole cavity/waveguide could be designed that overcomes these limitations. Perhaps a VERY low expansion rate? ..and it might well be accomplished with such a large baffle as Lynn has proposed. Unfortunately I think its a dead-end for DIY'ers because of the complexity/cost.
An alternative that Paul W uses to exceptional effect - dampening the rear wave with a lossy material and having no "cavity" for the front of the baffle and driver. That will however change the subjective spl for the listener - but that can be accommodated. Moreover (depending on the driver), lossy material should be moderately clear of the driver (the higher in freq. you go, where very short excursions are, the more important this becomes - especially with a more compliant driver).
Perhaps another solution that Lynn has already suggested is changing the pressure differential at the edge by perforating it (i.e. little holes near the baffle edges). An "aperiodic edge"?
I've also mentioned the possibility of a "disappearing edge" - i.e. virtually no edge, (similar to a knife's cutting edge), that can act as a boundary for pressure build up/release. Still, a *very* narrow dominate artifact might appear (..high amplitude, near discreet freq.). A key to this design is to have equal spl, for each phase, at that edge diffraction freq. for a more complete "null". Of course you could also add-in Lynn's solution.. and/or a low storage thermal loss paint near the edge.
On another note:
One other thing that I've mentioned before (and that I think would be an EXCELLENT way to extend your dominant research), are HOM's and typical DRIVERS - which necessarily must diffract, (because of the angle change), at the driver's surround-to-frame. I find this particularly pertinent with this thread because here we are discussing large diameter ("deep" profile) drivers operated at relatively high freq.s. The funny thing is, as Lynn has mentioned, some large diameter drivers sound good at higher freq.s and some don't. Certainly resonant "break-up", IMD, and THD, can and often are serious problems.. but I keep thinking there is a LOT more going-on, and I wouldn't be at all surprised if HOM's are at least in part a culprit to that "equation".
BTW, nice discussions!
Most hifi hornys don't listen to CD horns. I know none in my area that uses DSP. It also seems most good pro manuf. have moved away from CD. There are many alternatives that sound great.
So Lynn. what's keeping you from building a near full range horn system? Last time I looked here you were craving real dynamics. Using a direct radiator is a major compromise in dynamics and low level mid and bass detail - Low compromise low mid and bass is vital to serious playback IMO. I've used four 21" woofers in one dipole system and twelve tens in another and they can be pretty good but not as good as a horn when it comes to realism and detail. Plus it will give a wider range of choice for amplification and IME will mate better with a compression driver. With my 21" dipole I used a compression driver above 1.6K with twin 10" pro mids in an WMTMW. The Heil AMT in it's place was better being a dipole itself. All the best compression driver systems I've heard use bass horns.
Have you had bass horns in your room? If you haven't you should try some rather than relying on theory. I tried to bring Lee into this discussion but he ignored me.
So Lynn. what's keeping you from building a near full range horn system? Last time I looked here you were craving real dynamics. Using a direct radiator is a major compromise in dynamics and low level mid and bass detail - Low compromise low mid and bass is vital to serious playback IMO. I've used four 21" woofers in one dipole system and twelve tens in another and they can be pretty good but not as good as a horn when it comes to realism and detail. Plus it will give a wider range of choice for amplification and IME will mate better with a compression driver. With my 21" dipole I used a compression driver above 1.6K with twin 10" pro mids in an WMTMW. The Heil AMT in it's place was better being a dipole itself. All the best compression driver systems I've heard use bass horns.
Have you had bass horns in your room? If you haven't you should try some rather than relying on theory. I tried to bring Lee into this discussion but he ignored me.
ScottG said:
An alternative that Paul W uses to exceptional effect - dampening the rear wave with a lossy material and having no "cavity" for the front of the baffle and driver. That will however change the subjective spl for the listener - but that can be accommodated. Moreover (depending on the driver), lossy material should be moderately clear of the driver (the higher in freq. you go, where very short excursions are, the more important this becomes - especially with a more compliant driver).
Perhaps another solution that Lynn has already suggested is changing the pressure differential at the edge by perforating it (i.e. little holes near the baffle edges). An "aperiodic edge"?
There is a well known way to remove diffraction effects in optics. It is called apodization. Basically the aperture/entrance pupil is modified to taper the transmissivity following a known mathematical function (involves Bessel functions). I don't know what the equivalent of apodization could be in acoustical domain, but waves are waves, isn't it
Some sort of progressive attenuation towads the edges perhaps.
Bratislav said:
Clever! Apodization is used in some camera lenses to achieve a nice "Bokeh", the Japanese term for the out of focus parts of a photograph. A quality that the West is beginning to appreciate, at at least put a name on.
Maybe Bokeh has some acoustical equivalent. Reading about all these edge diffraction artifacts, one might think so.
I've got two more days here in the US, then I move to Thailand for the summer. Once there I will not have as much time as I do now. I am interested in these discussions and I think it important to clear up points on waveguides, etc. But this tread is about dipoles and I feel like I am steering away from the main topic when I comment on waveguides. I tied to get this discussion going on some other threads that made more sense, but no one seems to be going there.
I'd like to take these last couple of days to talk about waveguides to an audience that will listen - you guys seem to be doing that, but I think that I'm going to try and steer the waveguide audience to one of the other threads.
I just don't have much experience with dipoles - not much to add.
I'd like to take these last couple of days to talk about waveguides to an audience that will listen - you guys seem to be doing that, but I think that I'm going to try and steer the waveguide audience to one of the other threads.
I just don't have much experience with dipoles - not much to add.
SY said:
Hey, I'm a FFT user and we never use term "windowing"
Apodization is a widely used term in optics. Nothing to do with bokeh really (maybe someone coined that term to impress potential lens buyers). To properly apodize you'd need to progressively attenuate first optical element toward the edges. I'm only aware of it being used in astronomy to increase resolution in cases of objects having widely different brightness levels (also in radio astronomy).
Lots of amateurs use quasi-apodising masks using wire mesh, and it actually works quite well. That is why I am optimistic it can be done in acoustics 'on cheap'. One would only need to progressively slow down sound waves as sound moves towards the edge of the box/baffle/horn mouth/whatever. And that is done relatively easy, no?
This ribbon driver claims to use foam to slow down sound waves at the edge.
http://www.raalribbon.com/products_flatfoil_140-15.htm
http://www.raalribbon.com/products_flatfoil_140-15.htm
Earl,
thank you for answering my questions so thoroughly, quite convincing evidence, especially on frequency and SPL dependency.
Diffraction in OB's:
- Siegfried Linkwitz has a discussion on his diffraction page ,
and in FAQ 38.
Bottom line, he thinks it is not too important in dipoles because dipole cancellation will partly remove its effects, leading in the net effect to a virtual non-illumination of the baffle edges. Of course the effects on FR should be minimized.
In my own OB's that have evolved over 6, 7 years now, I chose a very wide baffle (24") to *maximize* the delay of the first edge diffraction. My goal was to get it out of the 0.6 ms window (from your quoted research, more towards 0.2 ms actually) where delays are still conflated with the original sound, into the delay range where the diffraction sounds more like an early room reflection. Ideally this should be 1.5 ms or more but that would make the speaker too large, 24" seemed like a good compromise.
As a side note my own OB speaker has quite the dimensions of the Summa, of which I was not aware at the time when I designed it... I just followed my own set of constraints and got to similar dimensions. The way it is going it might end up looking not too different from your Summas, but shallower and in OB or cardioid in the mid/low end (preferred, if I can pull it off): now I got interested in your waveguide approach to improve top end power handling, which is also a principal concern of Lynn in his thread here.
Finally, there is the Large midrange for OB thread, relevant to a low end OB / waveguided HF approach.
thank you for answering my questions so thoroughly, quite convincing evidence, especially on frequency and SPL dependency.
Diffraction in OB's:
- Siegfried Linkwitz has a discussion on his diffraction page ,
and in FAQ 38.
Bottom line, he thinks it is not too important in dipoles because dipole cancellation will partly remove its effects, leading in the net effect to a virtual non-illumination of the baffle edges. Of course the effects on FR should be minimized.
In my own OB's that have evolved over 6, 7 years now, I chose a very wide baffle (24") to *maximize* the delay of the first edge diffraction. My goal was to get it out of the 0.6 ms window (from your quoted research, more towards 0.2 ms actually) where delays are still conflated with the original sound, into the delay range where the diffraction sounds more like an early room reflection. Ideally this should be 1.5 ms or more but that would make the speaker too large, 24" seemed like a good compromise.
As a side note my own OB speaker has quite the dimensions of the Summa, of which I was not aware at the time when I designed it... I just followed my own set of constraints and got to similar dimensions. The way it is going it might end up looking not too different from your Summas, but shallower and in OB or cardioid in the mid/low end (preferred, if I can pull it off): now I got interested in your waveguide approach to improve top end power handling, which is also a principal concern of Lynn in his thread here.
Finally, there is the Large midrange for OB thread, relevant to a low end OB / waveguided HF approach.