In post #88 Floric wrote of the line source;-
In normal listening environments this (comb) effect is smeared by the room with the consequence of an inaccurancy in the time domain. But you will never hear the sound of the outermost speaker with a delay.
In post#97 Lynn Olson wrote;-
the power does not drop off according to the square law, thanks to the very narrow vertical dispersion. This means the acoustical joining to a radiator with spherical dispersion, such as a subwoofer or supertweeter, requires a listening-distance compensation.
Last year I tried line arrangements, and was able to attain enjoyable listening, but, the clarity of reproduction improved with distance away for the line, thus it was NOT best for use within a room at normal room listening distances. The idea was shelved.
As an additional note - I found that any form of straight side (rear back pressure reflective) baffling of the line source (including rear folded/tapered longitudinal 1/4 wave loading) had a deleterious effect upon reproduction, maybe as the graduated back-pressures caused non-coherently additive driver reactions.
The best reproduction arose by having the drivers narrowly mounted as a strip between two vertical soft long/fine pile carpet rolls with these pulled together behind the drivers to leave a rear gap equal to front cone area. There was no additional reflective rear pressure induced blurring tonality with this rounded damping baffle.
Whether the same clean sounding reproduction could be achieved by using a (well lined) curved perspex (or other) open baffle where the curve is rolled right back to a slot behind a single driver is more than I could ever farbricate or test.
Through my experiments I concluded that whizzer cones simply cannot remain time coherent at hf and yet, excepting low bass, drivers needed to be single point radiating for normal within-room listening.
So Lynn, I have been watching your thread with interest, though I have opted for cheap/cheerful 6x9 multi-way car drivers with expected aperiodic 12" bass augmentation, when (if) I get the chance.
Gee, can we ever be satisfied ?
In normal listening environments this (comb) effect is smeared by the room with the consequence of an inaccurancy in the time domain. But you will never hear the sound of the outermost speaker with a delay.
In post#97 Lynn Olson wrote;-
the power does not drop off according to the square law, thanks to the very narrow vertical dispersion. This means the acoustical joining to a radiator with spherical dispersion, such as a subwoofer or supertweeter, requires a listening-distance compensation.
Last year I tried line arrangements, and was able to attain enjoyable listening, but, the clarity of reproduction improved with distance away for the line, thus it was NOT best for use within a room at normal room listening distances. The idea was shelved.
As an additional note - I found that any form of straight side (rear back pressure reflective) baffling of the line source (including rear folded/tapered longitudinal 1/4 wave loading) had a deleterious effect upon reproduction, maybe as the graduated back-pressures caused non-coherently additive driver reactions.
The best reproduction arose by having the drivers narrowly mounted as a strip between two vertical soft long/fine pile carpet rolls with these pulled together behind the drivers to leave a rear gap equal to front cone area. There was no additional reflective rear pressure induced blurring tonality with this rounded damping baffle.
Whether the same clean sounding reproduction could be achieved by using a (well lined) curved perspex (or other) open baffle where the curve is rolled right back to a slot behind a single driver is more than I could ever farbricate or test.
Through my experiments I concluded that whizzer cones simply cannot remain time coherent at hf and yet, excepting low bass, drivers needed to be single point radiating for normal within-room listening.
So Lynn, I have been watching your thread with interest, though I have opted for cheap/cheerful 6x9 multi-way car drivers with expected aperiodic 12" bass augmentation, when (if) I get the chance.
Gee, can we ever be satisfied ?
even if subject of combing or non-combing isn't really issue-long time ago recognized in pro and PA fields,ther is just one example how pro boyz'n'grlz choose to solve it........
http://www.meyersound.com/products/industrialseries/sb-2/
hehe,I know problem of lousy intelligibility is......I have few church systems behind me.........
so- mentioning line arrays in this thread is really a side step- good for some another thread where cheapskating is first condition
http://www.meyersound.com/products/industrialseries/sb-2/
hehe,I know problem of lousy intelligibility is......I have few church systems behind me.........
so- mentioning line arrays in this thread is really a side step- good for some another thread where cheapskating is first condition
You are right Lynn,
I was thinking of the problem from the wrong direction. I thought of an array where the distance of the sources is small compared to the wavelength - which is an error if you think of audible frequencies. In addition I was thinking of sines and not of Impulses.
My first thought was that e.g. an optical grid produces a relatively sharp 0th maximum which is coherent - if it wasn't it would not exist.
But that is not valid for audible frequencies and real world drivers, there is always a part of the spectrum where the wavelength is in the same order of size than the distance of the speakers or even below that.
The picture Dave posted shows that effect very good: in the white areas you get coherent sound. In the "grey" areas you don't hear anything. If you choose the right distance for a given line length everything is o.k.
On the other hand, I did not hear the incorrectness in the time domain when I listened to line arrays right now. I was only surprised by the dynamics, the really low distorsion and the effect of a large sweet spot.
But for the correct reprodution of an impulse response which consits of a superposition of an infinite number of sine waves of a continuous spectrum they are definitely the wrong tool.
I wish you a nice day.
Best regards
Floric
I was thinking of the problem from the wrong direction. I thought of an array where the distance of the sources is small compared to the wavelength - which is an error if you think of audible frequencies. In addition I was thinking of sines and not of Impulses.
My first thought was that e.g. an optical grid produces a relatively sharp 0th maximum which is coherent - if it wasn't it would not exist.
But that is not valid for audible frequencies and real world drivers, there is always a part of the spectrum where the wavelength is in the same order of size than the distance of the speakers or even below that.
The picture Dave posted shows that effect very good: in the white areas you get coherent sound. In the "grey" areas you don't hear anything. If you choose the right distance for a given line length everything is o.k.
On the other hand, I did not hear the incorrectness in the time domain when I listened to line arrays right now. I was only surprised by the dynamics, the really low distorsion and the effect of a large sweet spot.
But for the correct reprodution of an impulse response which consits of a superposition of an infinite number of sine waves of a continuous spectrum they are definitely the wrong tool.
I wish you a nice day.
Best regards
Floric
back to back...
Lynn,
What are your thoughts regarding the value of mounting drivers back to back in a kind of warped open baffle isobarik loading magnet to magnet, with the rear facing driver wired out of phase to the front facing driver, so the two move as one unit?
Here are my own assumptions why this may be worth the added expense to some folks (and I am in way over my head here, so everyone go easy on me!):
1) Even though the "box" has no top, bottom, or sides.... the non-linearities of any driver will be somewhat improved by having a pair of them closely coupled in this arrangement, lowering distortion and other artifacts by having the "pull" driver creating a wave directly behind the "push" driver. The "push" driver could even be delayed ever so slightly, possibly introducing other advantages (depending on your take on Tierry Budge's work).
2) A little stuffing along the open sides and bottom should make the turbulence and/or any roaming non-cancelled sound waves that manage to survive in that sandwich a non factor from the listening chair. Open top still allows maximum heat escape to help keep drivers as close to optimum operating temp as possible.
3) Rear, out of phase signal is now closer in profile to front signal providing (I assume that due to all the mechanics behind a driver that the sound from the back is somewhat different than the sound from the front) so the dipole environment is now a bit closer to theoretical perfection.
4) if coaxial drivers, then you have the rear tweeter to use depending on your taste and you could eq just the rear facing tweeter for different preferences regarding this effect at the listening seat (following the linkwitz crowd hear on the potential value of the rear facing tweeter depending on room, speaker placement, and listening preferences).
5) Could also wire the two drivers in serial fashion for a nice 16 ohm load, OTL biamping mode with passive line level crossover..... By the way, what are your thoughts regarding passive line level crossovers? It seems to me that this would be a better way to optimize a system if you were building the amps as well, coupling them directly to the drivers and shaping the signal behind the preamp instead of behind the power amp.
Awesome thread by the way, and glad to see you exploring the dipole coax approach. Will be watching with great interest, and participating when possible (both in discussion and in potential build/experiment opportunities, though I don't have any measuring equipment yet...)
Lynn,
What are your thoughts regarding the value of mounting drivers back to back in a kind of warped open baffle isobarik loading magnet to magnet, with the rear facing driver wired out of phase to the front facing driver, so the two move as one unit?
Here are my own assumptions why this may be worth the added expense to some folks (and I am in way over my head here, so everyone go easy on me!):
1) Even though the "box" has no top, bottom, or sides.... the non-linearities of any driver will be somewhat improved by having a pair of them closely coupled in this arrangement, lowering distortion and other artifacts by having the "pull" driver creating a wave directly behind the "push" driver. The "push" driver could even be delayed ever so slightly, possibly introducing other advantages (depending on your take on Tierry Budge's work).
2) A little stuffing along the open sides and bottom should make the turbulence and/or any roaming non-cancelled sound waves that manage to survive in that sandwich a non factor from the listening chair. Open top still allows maximum heat escape to help keep drivers as close to optimum operating temp as possible.
3) Rear, out of phase signal is now closer in profile to front signal providing (I assume that due to all the mechanics behind a driver that the sound from the back is somewhat different than the sound from the front) so the dipole environment is now a bit closer to theoretical perfection.
4) if coaxial drivers, then you have the rear tweeter to use depending on your taste and you could eq just the rear facing tweeter for different preferences regarding this effect at the listening seat (following the linkwitz crowd hear on the potential value of the rear facing tweeter depending on room, speaker placement, and listening preferences).
5) Could also wire the two drivers in serial fashion for a nice 16 ohm load, OTL biamping mode with passive line level crossover..... By the way, what are your thoughts regarding passive line level crossovers? It seems to me that this would be a better way to optimize a system if you were building the amps as well, coupling them directly to the drivers and shaping the signal behind the preamp instead of behind the power amp.
Awesome thread by the way, and glad to see you exploring the dipole coax approach. Will be watching with great interest, and participating when possible (both in discussion and in potential build/experiment opportunities, though I don't have any measuring equipment yet...)
On his Orion, a very highly regarded speaker, Sigfreid Linkwitz has had dipole, open baffle, bass and mids from the beginning.
He recently added a back tweeter only and says it's the missing Link or similar:
http://www.linkwitzlab.com/orion++.htm
I believe it is a dipole also, like the rest of the drivers. So now all three are dipoles- when the front pushes, the rear pulls, so to speak..
He recently added a back tweeter only and says it's the missing Link or similar:
http://www.linkwitzlab.com/orion++.htm
I believe it is a dipole also, like the rest of the drivers. So now all three are dipoles- when the front pushes, the rear pulls, so to speak..
Iso-Whatsis
Once again it is useful to step out of the frequency domain, which conceals phase and time information, and look at impulses instead. Everything is travelling at the speed of sound (345 m/Sec), and for small sources, emission is hemispherical, in one polarity for the front wave, in the opposite polarity for the rear wave. We can choose to absorb the back wave - with only partial success in conventional enclosures - or let it travel freely and reflect off the walls of the room.
Putting one driver in front of another introduces a delayed image for both front and back waves, with the delay being rather short, about half-millisecond or so (a millisecond is about 14 inches long). Since the time window of 0 to 1 milliseconds is used for localization by the ear/brain/mind system, this delayed image will inevitably degrade image quality. All the fancy propaganda by well-financed hifi companies located in Scotland doesn't change these facts of physics, acoustics, and psychoacoustics.
The floor and wall reflections in the 2.5 to 20 millisecond region, by contrast, are benign, since they fall outside the localization window, and do not interfere at all. If anything, localization is improved, since reverberent content in the recording is assigned to these early reflections.
This is not new information, by the way: research into stereophonic sound in the Fifties and Sixties established the critical time intervals used for localization and perception of spatial impressions. Unfortunately, the hifi industry, with the exception of the BBC, KEF, and B&W, choose to ignore this until the early Eighties, so many speakers and surround systems are fundamentally mis-designed.
Designers who refuse to leave the frequency domain, or discuss measurement artifacts like comb filtering, are missing the forest for trees. The perception of sound involves both time and frequency domains, and ignoring one in favor of the other is a mistake. Reflections appear as what they are in the time domain: you see picket-fencing on the impulse response. If you only look at the frequency domain, it looks like comb-filtering, a series of narrow nulls in the frequency response. But it should never be forgotten that phase information is discarded in frequency-domain measurements, and time artifacts like an all-pass function (phase rotation over 180 or 360 degrees) do not appear at all.
It was probably working at the Tektronix Spectrum Analyzer division, with both a scope and an analyzer on the same bench, that underlined this. As a technical writer, it was my task to design sample measurement protocols, photograph the display, explain to the user what they were seeing, and explain the limitations of the measurement protocol. TV signals, for example, look quite different in the time (scope) and frequency (spectrum analyzer) domains. Twiddling a few controls on the SA can alter the appearance of that kind of spectrum quite a bit.
Returning to the Iso-whatsis, it's one way to make a push-pull woofer, which cancels most (but not all) of the 2nd harmonic distortion(at low frequencies). There's no change in efficiency from a single driver, but the required enclosure size is reduced in half, which is an advantage for building little-bitty speakers. The distance between the front and rear cones, though, is very troublesome, grossly degrading impulse response, and introducing our old friend comb-filtering. The usual crude fix seen in commercial speakers is lining the tunnel with felt, and not using the woofers into the midrange, hoping the lowpass crossover will roll off the midrange artifacts. (Hollow-sounding - gee, do you think a tunnel might sound that way? D'ya think?)
The whole concept has no benefits I can see, and serious and completely unnecessary drawbacks. If you want push-pull operation, fine, use two drivers, but side-by-side, with one facing in, and one facing out. Of course, the esthetics aren't as pretty, with an unfinished magnet and basket facing the customer, so I suspect the Iso-whatsis was invented as a way around the esthetic objection.
As we all know, looks win out over sound every time in the hifi biz, especially if you can get the Marketing Department to cook up some ridiculous story about making the cabinet disappear "by magic". If magazine reviewers buy it, so will the dealers and the customers, so everybody is happy.
Oh, it doesn't sound as good? That's because you're not using an Audiophile-Approved (tm) CD to listen to your system. The first time I saw my audio-pals using a StereoPile CD to assess their hifi, I knew the business had come full circle - people had so completely forgotten what music sounded like they had to buy a "HiFi For Dummies" CD with special sound tracks and liner notes to tell them what to listen for. Me, I'll take the "Rain, Steam, and Speed" soundtrack every time - nothing like a 80 mph fully-loaded freight train in a driving thunderstorm to really hear what your hifi can do!
OK, all snark aside, there are some good questions you've asked. Although line-level crossovers sidestep passive networks (good), they introduce the (much worse) problem of combining a top-quality linestage with an active crossover. This is a non-trivial problem, unless you're a member of the all-electronics-sound-the-same school, in which case you throw op-amps or digital EQ at it.
For better or worse, I'm one of those guys that can easily hear single parts changes, much less an entire linestage, and I consider designing a sonically good-sounding line-level crossover a huge and very difficult challenge, every bit as difficult as designing a new power amp or speaker system. There are very few good-sounding linestages, and the commercially available active crossovers are quite a bit worse, regardless of the high prices and fancy names on the box.
Greggo said:
Once again it is useful to step out of the frequency domain, which conceals phase and time information, and look at impulses instead. Everything is travelling at the speed of sound (345 m/Sec), and for small sources, emission is hemispherical, in one polarity for the front wave, in the opposite polarity for the rear wave. We can choose to absorb the back wave - with only partial success in conventional enclosures - or let it travel freely and reflect off the walls of the room.
Putting one driver in front of another introduces a delayed image for both front and back waves, with the delay being rather short, about half-millisecond or so (a millisecond is about 14 inches long). Since the time window of 0 to 1 milliseconds is used for localization by the ear/brain/mind system, this delayed image will inevitably degrade image quality. All the fancy propaganda by well-financed hifi companies located in Scotland doesn't change these facts of physics, acoustics, and psychoacoustics.
The floor and wall reflections in the 2.5 to 20 millisecond region, by contrast, are benign, since they fall outside the localization window, and do not interfere at all. If anything, localization is improved, since reverberent content in the recording is assigned to these early reflections.
This is not new information, by the way: research into stereophonic sound in the Fifties and Sixties established the critical time intervals used for localization and perception of spatial impressions. Unfortunately, the hifi industry, with the exception of the BBC, KEF, and B&W, choose to ignore this until the early Eighties, so many speakers and surround systems are fundamentally mis-designed.
Designers who refuse to leave the frequency domain, or discuss measurement artifacts like comb filtering, are missing the forest for trees. The perception of sound involves both time and frequency domains, and ignoring one in favor of the other is a mistake. Reflections appear as what they are in the time domain: you see picket-fencing on the impulse response. If you only look at the frequency domain, it looks like comb-filtering, a series of narrow nulls in the frequency response. But it should never be forgotten that phase information is discarded in frequency-domain measurements, and time artifacts like an all-pass function (phase rotation over 180 or 360 degrees) do not appear at all.
It was probably working at the Tektronix Spectrum Analyzer division, with both a scope and an analyzer on the same bench, that underlined this. As a technical writer, it was my task to design sample measurement protocols, photograph the display, explain to the user what they were seeing, and explain the limitations of the measurement protocol. TV signals, for example, look quite different in the time (scope) and frequency (spectrum analyzer) domains. Twiddling a few controls on the SA can alter the appearance of that kind of spectrum quite a bit.
Returning to the Iso-whatsis, it's one way to make a push-pull woofer, which cancels most (but not all) of the 2nd harmonic distortion(at low frequencies). There's no change in efficiency from a single driver, but the required enclosure size is reduced in half, which is an advantage for building little-bitty speakers. The distance between the front and rear cones, though, is very troublesome, grossly degrading impulse response, and introducing our old friend comb-filtering. The usual crude fix seen in commercial speakers is lining the tunnel with felt, and not using the woofers into the midrange, hoping the lowpass crossover will roll off the midrange artifacts. (Hollow-sounding - gee, do you think a tunnel might sound that way? D'ya think?)
The whole concept has no benefits I can see, and serious and completely unnecessary drawbacks. If you want push-pull operation, fine, use two drivers, but side-by-side, with one facing in, and one facing out. Of course, the esthetics aren't as pretty, with an unfinished magnet and basket facing the customer, so I suspect the Iso-whatsis was invented as a way around the esthetic objection.
As we all know, looks win out over sound every time in the hifi biz, especially if you can get the Marketing Department to cook up some ridiculous story about making the cabinet disappear "by magic". If magazine reviewers buy it, so will the dealers and the customers, so everybody is happy.
Oh, it doesn't sound as good? That's because you're not using an Audiophile-Approved (tm) CD to listen to your system. The first time I saw my audio-pals using a StereoPile CD to assess their hifi, I knew the business had come full circle - people had so completely forgotten what music sounded like they had to buy a "HiFi For Dummies" CD with special sound tracks and liner notes to tell them what to listen for. Me, I'll take the "Rain, Steam, and Speed" soundtrack every time - nothing like a 80 mph fully-loaded freight train in a driving thunderstorm to really hear what your hifi can do!
OK, all snark aside, there are some good questions you've asked. Although line-level crossovers sidestep passive networks (good), they introduce the (much worse) problem of combining a top-quality linestage with an active crossover. This is a non-trivial problem, unless you're a member of the all-electronics-sound-the-same school, in which case you throw op-amps or digital EQ at it.
For better or worse, I'm one of those guys that can easily hear single parts changes, much less an entire linestage, and I consider designing a sonically good-sounding line-level crossover a huge and very difficult challenge, every bit as difficult as designing a new power amp or speaker system. There are very few good-sounding linestages, and the commercially available active crossovers are quite a bit worse, regardless of the high prices and fancy names on the box.
Field Coils anyone?
Lynn, All,
there seems to be little discussion of field coils so far. I had the thought that the ability to tune the Q of a field coil speaker by altering the current density in the coil might allow using two very similar speakers on the OB and tuning them to fill their required roles.
I don't know if there are any field coil hemp cones available, I emailed the guys at Hemp Acoustics but got no response, which I believe is typical.
Lynn, All,
there seems to be little discussion of field coils so far. I had the thought that the ability to tune the Q of a field coil speaker by altering the current density in the coil might allow using two very similar speakers on the OB and tuning them to fill their required roles.
I don't know if there are any field coil hemp cones available, I emailed the guys at Hemp Acoustics but got no response, which I believe is typical.
agent.5 said:
Good point. Op-amp and digital crossovers are more acceptable below 120 Hz as far as I can tell - it's in the midrange and HF where the sonically noxious qualities of mid-fi electronics become audible. The dipole part of the system is where you want the pristine amplification - but for the sub, eh, I'm doubtful that tubes are the way to go. Both transformers and RC-coupled vacuum-tube circuits are at their worst in the deep-bass region.
I certainly would avoid passing the mid/high part of the signal through an ultra lowfi device like the Behringer DEQX, which uses cheap 741-type opamps and even cheaper electrolytic caps in the signal path. The digital part of the DEQX is decent, but the analog portion is truly terrible, using the same quality parts as a Wal-Mart boom-box.
Nothing wrong with separating the signals following the volume control, passing the mids and highs to the high-quality part of the system, and the low-bass to the subwoofer crossover, amp, and subwoofer drivers. Gary Pimm tried the DEQX for his subs and warned me that the electrolytic caps (in the DEQX input circuit) are such low quality they contaminated the mid-highs just by connecting the DEQX in parallel following the volume control. He had to replace the caps with good-quality polypropylenes in order to prevent the mid-highs from being contaminated whenever the DEQX was connected to the system.
As for field coil drivers, ooh la la, they sound so nice. I heard the Cogent horns at the RMAF and was deeply impressed at the subtlety and shimmer of the tone colors. While the signature of Alnico are bright and vivid tone colors, field-coil speakers have a sound all their own, with an almost luminous quality to the tonality. It's hard to describe unless you've heard it for yourself.
I don't know why it's even controversial that magnets sound different. The magnetic system is why the voice coil has significant inductance in the first place - the free-air value of the voice-coil inductance would be almost nothing. And of course, the magnetic system is NOT a perfectly linear inductor - what iron-core inductor is? Going further, chokes wound with ferrite and different magnetic materials sound different - and the (nonlinear) voice-coil inductance is ALWAYS in series with the signal going to the voice-coil, no matter what kind of amplifier you use.
Scottmoose said:
question for Scottmoose........and others:
what are easy predicted consequences (or differences) of using Eminence Alpha15 (97db/41Hz/Qts 1,26) vs. Eminence Beta15 (98,2db/35Hz/ Qts 0,58) in JamesD lookalike OB ?
which one is better if I need to place it as closer to wall I can?
TIA
More Thoughts ...
Not to tease you guys too much, but I've had some good ideas in the last few hours about the bass unit for the dipole - I'll share what I have in the absence of firing up Canvas and making proper drawings - so bear with me in these descriptions.
One of the key design points of the bass unit is the woofers are right next to floor level. This has the desirable quality of creating a mirror-image "virtual" driver just below the physical driver, thus doubling the effective cone area. You lose just a bit of efficiency if there's carpet right there on the floor in front of the speaker, but the loss is pretty small.
I've been thinking all along of only one or two bass drivers, but that might not be the most interesting approach. Let's stretch the envelope a bit and think of FOUR 12 or 15-inch drivers.
I'm visualizing a pair of drivers side-by-side in the front, another driver on the left side of bass unit, and another driver on the right side of the bass unit. Maybe a pair of 12's for the front - Alnico Tone Tubby's, maybe? - and a pair of 15's on the sides. All four drivers are in a quasi-box that is filled with recycled cotton home-insulation material, like ground-up blue jeans.
The box is open on the rear, the sides are no more than 16 inches deep, and there are partitions on the inside, extending from the front to the back, between the front pair of drivers, as well as a partition going from the magnet of each front driver and extending to the rear of the box. (Three internal partitions in all, with uneven distances between them.)
This is a variation of Gary Pimm's cardioid speakers, which are open-ended boxes fllled with recycled cotton insulation, about 12 to 16 inches deep. Think of them as very short transmission lines, if you will.
Now let's take the next step and look at the crossover for these four drivers, which are connected in series-parallel. The front drivers are in series (with themselves), and the side drivers are in series (with themselves). The front drivers handle the range from 160 to 220 Hz downward, and the side drivers handle the range from 80 to 100 Hz downward. All drivers are in-phase and work together at the lowest frequencies.
The 160 to 220 Hz lowpass filter is 2nd-order, and connected to the series-pair of front drivers. There is a second branch to the circuit that jumps off from the front drivers and goes to another inductor that forms a 80 to 100 Hz lowpass filter, which then feeds the series-pair of side drivers. (I don't claim credit for the 2.5 way crossover, I saw it elsewhere on the loudspeaker forum, and it'll certainly work well in this application.)
This sequential crossover ensures the low-bass drivers share the same 2nd-order lowpass filter as the midbass drivers, and all drivers share favorable phase relationships at the lowest frequencies, where they work together. The 2nd-order filter also keeps the front set of midbass drivers sounding clean, avoiding the usual 1 kHz bumps-and-dips, and keeps energy out of the frequency range where the boxes go into box modes (although these are minimized by the asymmetric dimensions and asymmetric-spaced interior panels).
At first blush it seems outrageous to have FOUR 12 and 15-inch drivers per channel - well, it does nicely compensate for the 1/f dipole rolloff of 6 dB per octave, the crossovers can be precisely tuned for just the right degree of driver overlap and room compensation, and there's no reason the low-bass pair of drivers have to be super-audiophile-quality. Gary Pimm is using modestly priced MCM or Parts Express 15-inchers for his high-Q subwoofers.
For that matter, there no necessity for four drivers at all - you could get by with a single 16-ohm front driver and two 8-ohm side drivers, depending on how the efficiencies work out. The principle is the same whether 3 or 4 drivers are used, the crossover is pretty much the same, and 12 and 15-inch drivers can be mixed and matched, since they are doing somewhat different things - deep bass and midbass.
Not to tease you guys too much, but I've had some good ideas in the last few hours about the bass unit for the dipole - I'll share what I have in the absence of firing up Canvas and making proper drawings - so bear with me in these descriptions.
One of the key design points of the bass unit is the woofers are right next to floor level. This has the desirable quality of creating a mirror-image "virtual" driver just below the physical driver, thus doubling the effective cone area. You lose just a bit of efficiency if there's carpet right there on the floor in front of the speaker, but the loss is pretty small.
I've been thinking all along of only one or two bass drivers, but that might not be the most interesting approach. Let's stretch the envelope a bit and think of FOUR 12 or 15-inch drivers.
I'm visualizing a pair of drivers side-by-side in the front, another driver on the left side of bass unit, and another driver on the right side of the bass unit. Maybe a pair of 12's for the front - Alnico Tone Tubby's, maybe? - and a pair of 15's on the sides. All four drivers are in a quasi-box that is filled with recycled cotton home-insulation material, like ground-up blue jeans.
The box is open on the rear, the sides are no more than 16 inches deep, and there are partitions on the inside, extending from the front to the back, between the front pair of drivers, as well as a partition going from the magnet of each front driver and extending to the rear of the box. (Three internal partitions in all, with uneven distances between them.)
This is a variation of Gary Pimm's cardioid speakers, which are open-ended boxes fllled with recycled cotton insulation, about 12 to 16 inches deep. Think of them as very short transmission lines, if you will.
Now let's take the next step and look at the crossover for these four drivers, which are connected in series-parallel. The front drivers are in series (with themselves), and the side drivers are in series (with themselves). The front drivers handle the range from 160 to 220 Hz downward, and the side drivers handle the range from 80 to 100 Hz downward. All drivers are in-phase and work together at the lowest frequencies.
The 160 to 220 Hz lowpass filter is 2nd-order, and connected to the series-pair of front drivers. There is a second branch to the circuit that jumps off from the front drivers and goes to another inductor that forms a 80 to 100 Hz lowpass filter, which then feeds the series-pair of side drivers. (I don't claim credit for the 2.5 way crossover, I saw it elsewhere on the loudspeaker forum, and it'll certainly work well in this application.)
This sequential crossover ensures the low-bass drivers share the same 2nd-order lowpass filter as the midbass drivers, and all drivers share favorable phase relationships at the lowest frequencies, where they work together. The 2nd-order filter also keeps the front set of midbass drivers sounding clean, avoiding the usual 1 kHz bumps-and-dips, and keeps energy out of the frequency range where the boxes go into box modes (although these are minimized by the asymmetric dimensions and asymmetric-spaced interior panels).
At first blush it seems outrageous to have FOUR 12 and 15-inch drivers per channel - well, it does nicely compensate for the 1/f dipole rolloff of 6 dB per octave, the crossovers can be precisely tuned for just the right degree of driver overlap and room compensation, and there's no reason the low-bass pair of drivers have to be super-audiophile-quality. Gary Pimm is using modestly priced MCM or Parts Express 15-inchers for his high-Q subwoofers.
For that matter, there no necessity for four drivers at all - you could get by with a single 16-ohm front driver and two 8-ohm side drivers, depending on how the efficiencies work out. The principle is the same whether 3 or 4 drivers are used, the crossover is pretty much the same, and 12 and 15-inch drivers can be mixed and matched, since they are doing somewhat different things - deep bass and midbass.
Won't you get a strong pipe resonance with that type of "box", on your woofers ?
I tried something like that : picture a cube, 50 cm wide, two ends open, 15" woofer on a board in the middle of the cube (H-baffle). It worked pretty well and sounded good, too.
Of course there was large pipe resonance. It's strong enough to be completely immune to passive crossover equalization...
Anyway, please consider active filtering : low watt high quality tubes for the midrange and up ; solid state muscle for those woofers, which permits using active EQ... since dipoles need EQ anyway.
I tried something like that : picture a cube, 50 cm wide, two ends open, 15" woofer on a board in the middle of the cube (H-baffle). It worked pretty well and sounded good, too.
Of course there was large pipe resonance. It's strong enough to be completely immune to passive crossover equalization...
Anyway, please consider active filtering : low watt high quality tubes for the midrange and up ; solid state muscle for those woofers, which permits using active EQ... since dipoles need EQ anyway.
To Box or Not To Box, That Is The Question
Yes, with the Pimm-style semi-box, active EQ w/solid-state amplification for the bass units might be a necessity. Once the drivers go into short tunnels, there's the box mode(s) to consider.
In Pimm's setup, the semi-boxes are pretty much filled, with about 2 to 3 inches of free space between the back of the driver and all that filling. All you see when when you look at the back of the box is a wall of gray fuzzy stuff - that's why they remind me of short transmission lines, or maybe resistance-loaded boxes. The filling he's using was way denser than anything I've used in a TL - there's so much it's mostly absorbing the backwave, which is what gives the cardioid response pattern. Another hint of the backwave is being heavily absorbed is the flatness of the impedance curves, which have a very TL-like appearance.
The alternative to the semi-box is a pair of 12 or 15-inchers at the base of a true dipole, and conventional subs coming in around 120 Hz or so. I'm guessing the semi-box would probably go down to 40 or 50 Hz, depending on the degree of equalization.
Since there are two quite different ways to go, that's another reason to separate the HF dipole (acrylic) baffle from the bass unit. Dipoles and short-box cardioids are somewhat different animals, and have different needs for care and feeding (crossovers and equalization).
peufeu said:
Yes, with the Pimm-style semi-box, active EQ w/solid-state amplification for the bass units might be a necessity. Once the drivers go into short tunnels, there's the box mode(s) to consider.
In Pimm's setup, the semi-boxes are pretty much filled, with about 2 to 3 inches of free space between the back of the driver and all that filling. All you see when when you look at the back of the box is a wall of gray fuzzy stuff - that's why they remind me of short transmission lines, or maybe resistance-loaded boxes. The filling he's using was way denser than anything I've used in a TL - there's so much it's mostly absorbing the backwave, which is what gives the cardioid response pattern. Another hint of the backwave is being heavily absorbed is the flatness of the impedance curves, which have a very TL-like appearance.
The alternative to the semi-box is a pair of 12 or 15-inchers at the base of a true dipole, and conventional subs coming in around 120 Hz or so. I'm guessing the semi-box would probably go down to 40 or 50 Hz, depending on the degree of equalization.
Since there are two quite different ways to go, that's another reason to separate the HF dipole (acrylic) baffle from the bass unit. Dipoles and short-box cardioids are somewhat different animals, and have different needs for care and feeding (crossovers and equalization).
I think peufeu is right. Even with a true dipole, when the depth is about 16" you will have to cope with a pretty strong resonance around 200 Hz (The 'pipe' is virtually longer than the depth of the dipole suggests). You could equalize that but even with active equalization I would stay away from such effects and use such a dipole only up to about 100 Hz.
I myself use 1 pair of 12" inch drivers per side (Peerless SLS-12), mounted on a baffle of 3.8" thick. This gives perfect dipole behaviour up to 400 Hz. They are equalized 6dB/oct with a shelving low pass set at 28Hz and 560 Hz corner frequencies. They provide me with uncoloured sound when I low pass filter them 2nd order at 200 Hz . Ofcourse, excursion limited SPL at the low end. For that, you already have a good solution in mind
.
I myself use 1 pair of 12" inch drivers per side (Peerless SLS-12), mounted on a baffle of 3.8" thick. This gives perfect dipole behaviour up to 400 Hz. They are equalized 6dB/oct with a shelving low pass set at 28Hz and 560 Hz corner frequencies. They provide me with uncoloured sound when I low pass filter them 2nd order at 200 Hz . Ofcourse, excursion limited SPL at the low end. For that, you already have a good solution in mind
.
There is no reason why the arrangement of 4 drivers couldn't be in the same kind of baffles envisioned earlier, the side panels would just need to be sized to accomodate them.
Designs with 2 or 4 bass drivers -- all on the sides push-push -- are already in the field. Given the low XO point, this arrangement could have significant advantages as far as newtonian energy loading/transmitting to the material of the baffle.
As far as 4 drivers seeming excessive -- an earlier sketch i posted had 8 x 12".
dave
Designs with 2 or 4 bass drivers -- all on the sides push-push -- are already in the field. Given the low XO point, this arrangement could have significant advantages as far as newtonian energy loading/transmitting to the material of the baffle.
As far as 4 drivers seeming excessive -- an earlier sketch i posted had 8 x 12".
dave
Member
Joined 2003
Meta Post - Waveguides for the Backwave
Well, once again I've led myself into a thicket with one of my bright ideas. Maybe stepping back a few steps will clarify things.
All of the typical acoustic radiators are intrinsically dipoles, with reverse polarity for the backwave. Note: Phase in only meaningful if the frequency is specified, while polarity exists independent of frequency. For example, it is trivial to invert the polarity of a signal with a bandwidth of 5 Hz to 5 MHz, while shifting the phase of such a signal through 90 degrees over the complete bandwidth requires an extremely complex all-pass filter. Do such wideband signals exist? Think of television, which is very sensitive to time distortion.
Returning to dipole radiators, what to do with the equal-intensity reverse-polarity backwave? One approach is a true infinite baffle, which in physical terms means cutting a hole in the living-room wall, and letting the backwave emit into the great outdoors. This is not a practical solution for most of us.
We can use a finite baffle of different shapes: let's look at four shapes.
1) A circular disk. This is the most resonant solution for an "open-baffle" loudspeaker.
2) A 90-degree circular horn, or megaphone. This is the most resonant solution for the horn.
3) An open-ended pipe with a circular cross section. This is the most resonant solution for the pipe.
4) A closed-ended pipe with a circular cross section. The other shapes can be closed off as well, creating a box.
All four shapes are actually transforms of each other, with performance that can be modified by selecting some degree of asymmetry, which decreases the Q of the standing wave, or some degree of absorbent filling, which decreases the Q and magnitude of the standing wave. In principle, 100% absorbent filling should absorb all of the backwave, and also completely remove the standing wave created by the structure. In practice, filling materials exhibit frequency selectivity as well as reflectivity at some frequencies.
Note again that shapes 1 through 3 are simply transforms, with an infinite range of intermediate shapes. All exhibit standing waves due to their finite size and the reflection created at the outer boundary of the shape. Asymmetry in the shape blurs the reflection across time, but does not decrease its total energy content. This smoothing looks good in the frequency domain, but the time display still shows a blurred reflection.
What about edge termination? Ah, now that's different. One solution not thoroughly investigated is replacing the hard material with an open-pore mesh. Perforated metal mesh, such as the pattern of ventilation holes on top of an amplifier, is typically 50% open. This kind of stands the JBL Acoustic Lens on its head - instead of forcing the forward wave to disperse through the mesh, they would have been better off terminating the horn itself with the mesh, thus softening the edge of the horn.
How does this apply to the short boxes of the dipole bass modules? Well, imagine the back portion of the box not made of wood, but perforated metal or pegboard. The back of the box is still open as before, but the last 1/3 of the sides are perforated metal or pegboard, which partially reflects and partially transmits sound. This gives a much softer edge to the end of the box, since energy leaks out progressively as the pulse travels back through the box. Internal pressures are low, since this is not a sealed or vented box in the usual sense, but more of a resistance-loaded box - or very short transmisssion line.
The mesh is also a good way to terminate the edge of the dipole on the top - although I'd be the first to admit a pattern of small holes at the outer edges of the acrylic panel would look a little odd. It would work, though.
As odd as it seems, dipoles, transmission lines, horns, and closed boxes are all topologically similar in the way they create standing waves. As this continuum of shapes approaches a pipelike shape, the energy storage increases, and as it approaches a flat disk, the energy storage decreases - although it does not drop to zero.
The usual solution of modest edge treatment (rolled corners) and/or filling materials is only modestly effective, as evidenced by high levels of remaining coloration and not-so-good time measurements. The extremely wide dynamic range of the ear is why we can hear even small reflections so efficiently, and why a 60 dB decay threshold is really only the beginning. Our hearing is optimized to detect sizes, shapes, and the acoustic properties of materials, and is extremely sensitive to resonances and correlated reflections.
Edge treatments in the broader sense of EnABL, Mamboni, and open-mesh techniques have not been investigated as thoroughly as should be, since the results will show in impedance curves and improvements in impulse response.
Well, once again I've led myself into a thicket with one of my bright ideas. Maybe stepping back a few steps will clarify things.
All of the typical acoustic radiators are intrinsically dipoles, with reverse polarity for the backwave. Note: Phase in only meaningful if the frequency is specified, while polarity exists independent of frequency. For example, it is trivial to invert the polarity of a signal with a bandwidth of 5 Hz to 5 MHz, while shifting the phase of such a signal through 90 degrees over the complete bandwidth requires an extremely complex all-pass filter. Do such wideband signals exist? Think of television, which is very sensitive to time distortion.
Returning to dipole radiators, what to do with the equal-intensity reverse-polarity backwave? One approach is a true infinite baffle, which in physical terms means cutting a hole in the living-room wall, and letting the backwave emit into the great outdoors. This is not a practical solution for most of us.
We can use a finite baffle of different shapes: let's look at four shapes.
1) A circular disk. This is the most resonant solution for an "open-baffle" loudspeaker.
2) A 90-degree circular horn, or megaphone. This is the most resonant solution for the horn.
3) An open-ended pipe with a circular cross section. This is the most resonant solution for the pipe.
4) A closed-ended pipe with a circular cross section. The other shapes can be closed off as well, creating a box.
All four shapes are actually transforms of each other, with performance that can be modified by selecting some degree of asymmetry, which decreases the Q of the standing wave, or some degree of absorbent filling, which decreases the Q and magnitude of the standing wave. In principle, 100% absorbent filling should absorb all of the backwave, and also completely remove the standing wave created by the structure. In practice, filling materials exhibit frequency selectivity as well as reflectivity at some frequencies.
Note again that shapes 1 through 3 are simply transforms, with an infinite range of intermediate shapes. All exhibit standing waves due to their finite size and the reflection created at the outer boundary of the shape. Asymmetry in the shape blurs the reflection across time, but does not decrease its total energy content. This smoothing looks good in the frequency domain, but the time display still shows a blurred reflection.
What about edge termination? Ah, now that's different. One solution not thoroughly investigated is replacing the hard material with an open-pore mesh. Perforated metal mesh, such as the pattern of ventilation holes on top of an amplifier, is typically 50% open. This kind of stands the JBL Acoustic Lens on its head - instead of forcing the forward wave to disperse through the mesh, they would have been better off terminating the horn itself with the mesh, thus softening the edge of the horn.
How does this apply to the short boxes of the dipole bass modules? Well, imagine the back portion of the box not made of wood, but perforated metal or pegboard. The back of the box is still open as before, but the last 1/3 of the sides are perforated metal or pegboard, which partially reflects and partially transmits sound. This gives a much softer edge to the end of the box, since energy leaks out progressively as the pulse travels back through the box. Internal pressures are low, since this is not a sealed or vented box in the usual sense, but more of a resistance-loaded box - or very short transmisssion line.
The mesh is also a good way to terminate the edge of the dipole on the top - although I'd be the first to admit a pattern of small holes at the outer edges of the acrylic panel would look a little odd. It would work, though.
As odd as it seems, dipoles, transmission lines, horns, and closed boxes are all topologically similar in the way they create standing waves. As this continuum of shapes approaches a pipelike shape, the energy storage increases, and as it approaches a flat disk, the energy storage decreases - although it does not drop to zero.
The usual solution of modest edge treatment (rolled corners) and/or filling materials is only modestly effective, as evidenced by high levels of remaining coloration and not-so-good time measurements. The extremely wide dynamic range of the ear is why we can hear even small reflections so efficiently, and why a 60 dB decay threshold is really only the beginning. Our hearing is optimized to detect sizes, shapes, and the acoustic properties of materials, and is extremely sensitive to resonances and correlated reflections.
Edge treatments in the broader sense of EnABL, Mamboni, and open-mesh techniques have not been investigated as thoroughly as should be, since the results will show in impedance curves and improvements in impulse response.
Hi Lynn.
I wonder if the enclosures you mention could be terminated with some modification of the Mamboni pattern, perhaps extended into three dimensional shapes of some size and made from a material that eats energy by vibrating small kinked bits of fiber?
There is a material of merit, the replacement carpet underlayment used by auto interior restoration folks. About an inch thick, with a tightly interlaced body of various length and diameter, kinked, moderately stiff fibers and small pieces of a rag like material. Has a shorter, more densly packed face, on one side, with the other face more open and with larger kinked fibers. Easily the most linear damping material I have found and good from infrasonic to about 2 k or so. Very linear in this range too.
Perhaps a Mamboni ring applied just before and entering into the peg board mesh area?
Bud
I wonder if the enclosures you mention could be terminated with some modification of the Mamboni pattern, perhaps extended into three dimensional shapes of some size and made from a material that eats energy by vibrating small kinked bits of fiber?
There is a material of merit, the replacement carpet underlayment used by auto interior restoration folks. About an inch thick, with a tightly interlaced body of various length and diameter, kinked, moderately stiff fibers and small pieces of a rag like material. Has a shorter, more densly packed face, on one side, with the other face more open and with larger kinked fibers. Easily the most linear damping material I have found and good from infrasonic to about 2 k or so. Very linear in this range too.
Perhaps a Mamboni ring applied just before and entering into the peg board mesh area?
Bud