Overkill Encore (2003):
Prey (2003):
Not sure on the name of this one:
Thats all. Aren't they beautiful?
An externally hosted image should be here but it was not working when we last tested it.
Prey (2003):
An externally hosted image should be here but it was not working when we last tested it.
Not sure on the name of this one:
An externally hosted image should be here but it was not working when we last tested it.
Thats all. Aren't they beautiful?
s
Derek, the posts about the Manger and the desired bass/midbass criteria are very interesting. I'm largely in agreement about the desired qualities for the 100~800 Hz region vs what's needed for the sub-100 Hz region. It's good to see there's a "secret" to unlocking this unusual driver - basically, don't try and get any deep bass out of it!
Me a master? Not really. I just collect bits of information and try and assemble them into something useful - and not always successfully, as you can see below.
Well, I'm still trying to decide if the sims are right or not. For an idealized point-source dipole emitter, adding a finite-sized baffle adds an additional non-minimum-phase delay to the direct wave - and the delay of this additional wave is greatest directly in front and directly in back. The greater the baffle size, the greater the delay of the non-minimum-phase additional wave.
I'm not too sure if I buy the argument the rear frame of the driver acts as an ideal lowpass filter for the backwave. I see it as an additional source of reflections and interference, and not an absorber at all - there's nothing back there that is soft and absorbent - it's all hard, reflective metal. So what comes out of the back is a cluttered and degraded version of what comes out of the front - lowpass filtered in the coarse sense, but not the same as an electrical lowpass filter nor the same as passage through damping material.
Overkill Audio said:
Derek, the posts about the Manger and the desired bass/midbass criteria are very interesting. I'm largely in agreement about the desired qualities for the 100~800 Hz region vs what's needed for the sub-100 Hz region. It's good to see there's a "secret" to unlocking this unusual driver - basically, don't try and get any deep bass out of it!
Me a master? Not really. I just collect bits of information and try and assemble them into something useful - and not always successfully, as you can see below.
john k... said:
Well, I'm still trying to decide if the sims are right or not. For an idealized point-source dipole emitter, adding a finite-sized baffle adds an additional non-minimum-phase delay to the direct wave - and the delay of this additional wave is greatest directly in front and directly in back. The greater the baffle size, the greater the delay of the non-minimum-phase additional wave.
I'm not too sure if I buy the argument the rear frame of the driver acts as an ideal lowpass filter for the backwave. I see it as an additional source of reflections and interference, and not an absorber at all - there's nothing back there that is soft and absorbent - it's all hard, reflective metal. So what comes out of the back is a cluttered and degraded version of what comes out of the front - lowpass filtered in the coarse sense, but not the same as an electrical lowpass filter nor the same as passage through damping material.
I apologize to John K and others if I'm taking an oversimplified view of loudspeaker systems.
I see two main sources of coloration: resonance and multipath. If the resonance is a simple system, and not multiple overlapping sources that are spatially dispersed, it is minimum-phase and can be corrected by equalization. Correction by equalization breaks down if there are multiple sources that are spatially dispersed, such as the more chaotic types of cone or diaphragm breakup.
Multipath is different. As the name implies, the sound takes multiple paths from the emission source to the listener. This is inherently non-minimum-phase - the comb filtering you see in the frequency domain is an artifact of multipath, not anything the emitter is doing. The real problem is in the time domain, where it is clearly visible if present.
Time domain problems should not be fixed in the frequency domain - if you do, the time domain problem is made worse, not better. It is possible to use tapped delay lines (in the digital or analog domains) to cancel out the multipath (this is used in digital cellphone systems), but the problem in the loudspeaker world is the correction is location sensitive - a correction good for one location in the room will make it worse in other locations. At the lowest frequencies with the longest wavelengths, it can work, but is not a good solution for mid or higher frequencies.
Antenna design for radio, television, and cellphones is simplified by the narrow bandwidth of the carrier, usually much less than a single octave. As a result, antennas and RF preamps can be selectively tuned to the desired carrier and reject the rest. It also greatly simplifies decisions that affect directivity and group-delay response (the latter critical for stereo FM, color TV, and digital transmission).
Loudspeakers have the very awkward requirement for extremely wide bandwidth (three decades!), flat response, freedom from resonance and multipath, and low harmonic and intermodulation distortion. Even for a distortionless antenna system these would be very severe requirements, and loudspeakers need to be transducers as well. The problem is nearly intractable, compared to the simplicity of the RF world. The difficulty of the problem is where the requirement for tradeoffs come in - in headroom, acceptance of multipath and resonance, and nonlinear distortion.
Direct-radiator LF drivers are reasonably well behaved in the 100 ~ 800 Hz region. Not perfect, but certainly not as bad as cones and diaphragms are at the top of their working range, where the radiator goes into chaotic breakup. The problem is in multipath. What to do with the reverse-polarity backwave?
The magnet and basket structure creates many additional unwanted reflections, and then there is the problem of where the backwave goes next. If the baffle is a simple disk, then the backwave diffracts around the edge after a transit delay that is proportional to the radius of the baffle. This happens in both directions - the front and back waves pass through each other. The transit delay in unavoidable and a simple property of acoustics.
In a box, the backwave bounces around many times, with a moderate degree of attenuation and low-passing by going through damping material with each pass. Loudspeaker cones are acoustically transparent, so the many-times-reflected backwave emerges through the cone. If the box is a vented box (bass-reflex), the vents have organ-pipe modes that selectively pass some frequencies and reject others - these modes fall in the midrange and have very high Q. The multipath within the box is a separate problem from cabinet-wall resonance, which is a structural problem, not an acoustic one. However, it is possible for acoustic standing-waves in the box to physically drive cabinet-wall resonances. It takes an accelerometer to distinguish between mechanical and acoustical problems in the box, since they can interact.
I am interested in minimizing the multipath problem in loudspeakers, which is why I've asked for Gary Pimm's assistance. He's a Tektronix guy with experience in RF systems and instrumentation shielding over a very wide frequency range. I like to give my collaborators credit, so that's why I've mentioned his name in connection with the filled-box quasi-cardioids.
I see two main sources of coloration: resonance and multipath. If the resonance is a simple system, and not multiple overlapping sources that are spatially dispersed, it is minimum-phase and can be corrected by equalization. Correction by equalization breaks down if there are multiple sources that are spatially dispersed, such as the more chaotic types of cone or diaphragm breakup.
Multipath is different. As the name implies, the sound takes multiple paths from the emission source to the listener. This is inherently non-minimum-phase - the comb filtering you see in the frequency domain is an artifact of multipath, not anything the emitter is doing. The real problem is in the time domain, where it is clearly visible if present.
Time domain problems should not be fixed in the frequency domain - if you do, the time domain problem is made worse, not better. It is possible to use tapped delay lines (in the digital or analog domains) to cancel out the multipath (this is used in digital cellphone systems), but the problem in the loudspeaker world is the correction is location sensitive - a correction good for one location in the room will make it worse in other locations. At the lowest frequencies with the longest wavelengths, it can work, but is not a good solution for mid or higher frequencies.
Antenna design for radio, television, and cellphones is simplified by the narrow bandwidth of the carrier, usually much less than a single octave. As a result, antennas and RF preamps can be selectively tuned to the desired carrier and reject the rest. It also greatly simplifies decisions that affect directivity and group-delay response (the latter critical for stereo FM, color TV, and digital transmission).
Loudspeakers have the very awkward requirement for extremely wide bandwidth (three decades!), flat response, freedom from resonance and multipath, and low harmonic and intermodulation distortion. Even for a distortionless antenna system these would be very severe requirements, and loudspeakers need to be transducers as well. The problem is nearly intractable, compared to the simplicity of the RF world. The difficulty of the problem is where the requirement for tradeoffs come in - in headroom, acceptance of multipath and resonance, and nonlinear distortion.
Direct-radiator LF drivers are reasonably well behaved in the 100 ~ 800 Hz region. Not perfect, but certainly not as bad as cones and diaphragms are at the top of their working range, where the radiator goes into chaotic breakup. The problem is in multipath. What to do with the reverse-polarity backwave?
The magnet and basket structure creates many additional unwanted reflections, and then there is the problem of where the backwave goes next. If the baffle is a simple disk, then the backwave diffracts around the edge after a transit delay that is proportional to the radius of the baffle. This happens in both directions - the front and back waves pass through each other. The transit delay in unavoidable and a simple property of acoustics.
In a box, the backwave bounces around many times, with a moderate degree of attenuation and low-passing by going through damping material with each pass. Loudspeaker cones are acoustically transparent, so the many-times-reflected backwave emerges through the cone. If the box is a vented box (bass-reflex), the vents have organ-pipe modes that selectively pass some frequencies and reject others - these modes fall in the midrange and have very high Q. The multipath within the box is a separate problem from cabinet-wall resonance, which is a structural problem, not an acoustic one. However, it is possible for acoustic standing-waves in the box to physically drive cabinet-wall resonances. It takes an accelerometer to distinguish between mechanical and acoustical problems in the box, since they can interact.
I am interested in minimizing the multipath problem in loudspeakers, which is why I've asked for Gary Pimm's assistance. He's a Tektronix guy with experience in RF systems and instrumentation shielding over a very wide frequency range. I like to give my collaborators credit, so that's why I've mentioned his name in connection with the filled-box quasi-cardioids.
"Multipath is different. As the name implies, the sound takes multiple paths from the emission source to the listener. This is inherently non-minimum-phase - the comb filtering you see in the frequency domain is an artifact of multipath, not anything the emitter is doing. The real problem is in the time domain, where it is clearly visible if present."
Lynn,
Multipath is not inherently non-minimum phase. Comb filtering typically does indicate multipath but not necessarily non-MP behavior. I presume you would consider diffraction multiplath and therefore would assume it to be non-MP. Thus, as a reference to a 3rd, uninterested party you might which to examine Kates, Loudspeaker Cabinet Reflection Effect, AES Anthology, Vol.2, which shows the conditions under which edge diffraction is MP. Similar arguments can be applied multiple sources, the back wave , etc. And, as I am sure you really know, for a linear system the frequency domain and the time domain are one in the same. And if the system is MP then time domain errors are also corrected by MP corrections in the frequency domain. Again, for a linear system. And what you have been discussing are all factors of the linear aspects of a speaker. The key is whether a simple, single eq function Eq can be applied to correct the response in a 3-D space, as Earl and I were discussing. The answer to that is yes, if the source is CD. As most of the readers here have seen, I have clearly demonstrated that.
I would agree that there can be many source of coloration in a loudspeaker, but the idea that multipath necessarily creates time domain distortion which is not correctable is in error.
Lynn,
Multipath is not inherently non-minimum phase. Comb filtering typically does indicate multipath but not necessarily non-MP behavior. I presume you would consider diffraction multiplath and therefore would assume it to be non-MP. Thus, as a reference to a 3rd, uninterested party you might which to examine Kates, Loudspeaker Cabinet Reflection Effect, AES Anthology, Vol.2, which shows the conditions under which edge diffraction is MP. Similar arguments can be applied multiple sources, the back wave , etc. And, as I am sure you really know, for a linear system the frequency domain and the time domain are one in the same. And if the system is MP then time domain errors are also corrected by MP corrections in the frequency domain. Again, for a linear system. And what you have been discussing are all factors of the linear aspects of a speaker. The key is whether a simple, single eq function Eq can be applied to correct the response in a 3-D space, as Earl and I were discussing. The answer to that is yes, if the source is CD. As most of the readers here have seen, I have clearly demonstrated that.
I would agree that there can be many source of coloration in a loudspeaker, but the idea that multipath necessarily creates time domain distortion which is not correctable is in error.
mige0 said:
Hi Michael,
I am not against damping. In fact I'm all for it. That is not the point of my posts. The point is whether or not eq can correct the time response of a system, what constitutes MP, and where can MP eq be used to advantage. Let's face it every speaker that uses any type of crossover is using EQ. We start with some raw source response and make a decision of what target acoustic response we want to meet. Call it a crossover filter if you like, but it is equalization of the raw source response to a target relative to some design point. If the source is MP and the filter (eq) is MP then the final result will be MP and will have the same time response as target response, relative to the design point, with deviations only a result of deviations form the ideal target.
Adding damping to the back side of a driver mounted as a dipole will have an effect on the raw SPL data just as mounting the driver in a sealed enclosure will change the raw SPL data. But there are several issues to look at when you consider pure open baffle, open baffle with some kind of damping over the back side of the driver, an open back enclosure of a given length and cross section, or a closed box. There are a number of sources of resonances to deal with in each case. Mechanical resonances associated with the driver such as cone breakup and basket resonances. Both of these can be controlled somewhat by driver design. The there is mechanical energy transmitted to the baffle or enclosure. Again these can be controlled by design and driver mounting. Mechanical resonance of the baffle/enclosure. These are again, mostly excited by the mechanical energy transmitted to the structure directly from the driver. And these too are controllable by design. So what we are really left with are the acoustic resonances and we must ask which are more problematic. If we stay with a pure open baffle we deal with the back wave in one manor . If we go with a sealed box or box with open back (short TL) we have other issues. Ask yourself what is the real difference between a sealed box and an open backed box and an open baffle? Lynn has suggested you talk into a sealed box to see what the colorations are. Take it a step further. Cut a 6" hole in a piece of plywood and talk into it. What is the coloration? Now, build a box say 6" x 6" x 6" with and w/o a back. Talk into it. Add damping and talk into the boxes. Depending on the amount of damping you probably won't notice much difference between the sealed and open backed boxes. In an open backed box the damping creates a low pass filter. The same effect occurs in a sealed box. The only basic difference is that, above the fundamental resonance when the wave reaches the back of the open box some of the wave is emitted to the surroundings and some is reflected back toward the driver. In the sealed box some of the wave is transmitted to the box wall and the majority is reflected back to the driver. Below the fundamental it is very difficult to get much attenuation of the back wave in an open box. As I said, the damping forms a low pass acoustic filter. Thus the low frequency radiation form the open back is pretty much identical in amplitude to the front side driver radiation. Of course, below the fundamental in a sealed box the box begins to act like a pressure vessel creating the air spring and containing the rear wave, thus preventing the "dipole" cancellation seen in an open backed box. If this is to be a midrange driver, since the damping in both open and sealed boxes is effectively a low pass filter the major differences will be based on where the pole of the filter is compared to the high pass filter for the midrange.
Below the fundamental the open backed response can be exploited to form a dipole or quasi-cardioid, depending on the damping and directional characteristics of the driver. Above the fundamental the response is nominally going to be similar to that of the sealed box with radiation pattern established by the driver characteristics and what ever anomalies arise from what ever radiation escapes form the rear, and the directional characteristics of the rear radiation.
Lynn's proposed damping approach with perforated side and internal panels aligned is specific ways will have little impact at low frequency (wave length long compared to box dimensions) other than to change the effective front to back delay. At higher frequencies there is no reason to expect that the behavior will be anything different than the low pass filter. Achieving 10 or 12 dB wide band attenuation is unlikely.
FWIW, here is a simulation of the rear radiation of 6" driver mounted in a 6" x 6" x 6" heavily damped, open backed enclosure.
An externally hosted image should be here but it was not working when we last tested it.
[edit 4.6.09]john k... said:
I should add that I am not advocating correcting comb filtering using eq. In the frequency range where comb filtering is apparent although the response may still be MP it is not CD. The peaks and nulls in the response will change in frequency as the reference position changes. However, with correct choice of driver size and baffle dimensions the dipole peak can be positioned to occur above the point where the driver becomes directional. As a result the comb filtering which would occur if the driver were still omni directional is not present. This was the case in the measurements I posted for the 8" baffle with 5" driver. The downward turn at 3K Hz would be the position of the first dipole null, but there are no other peaks or nulls above that point (other than the peak due to breakup at 5k). Add to this the correct choice of x-o frequency (2.5K in this case) and we have a CD response generated from the front and rear wave which can be eq'ed to the desired acoustic target and have the correct amplitude, phase and time response on and off axis.
I'm kind of thinking here than normally we would want to keep speakers away from walls becasue we don't want too much reflection. With open baffles firing the same amount of energy backwards, wouldn't the reflection level be still pretty high? Would this also make the multipath effect more significant?
John
Typically I think that we see things the same, but maybe use different points of view. Since MP is not a concept that is ever used in Acoustics, I find its application to 3D acoustical problems problematic. You see it as a generalized extension of the concepts, but admit that non-CD requires different EQ for different angles, which to me means that the "system" cannot be MP because there is not a single filter that can correct the time, frequency and space responses simultaneously. Now, we completely agree that CD is correctable in all these aspects and thus, to me, this can be thought of as MP. But, it seems to me that there is no generally accepted generalization of MP to space, which is why it does not appear in the acoustics literature. We will always disagree when we are free to use different definitions and choose them as such.
I also agree with your comments to Lynn about boxes. I would like to add that Lynn's comment that cones are "acoustically transparent" is completely false. They do admit to some transmission, but "transparent" is completely incorect. Even my "thin as can be" bed sheet projection screen is not completely transparent (and its intended to be). A screen of cone material would be vastly more loss. Box reflections act as a spring far more than any other effect, especially when there is absorption. The idea that sound simply "comes back through the cone" is simply untrue. There is a great deal of attenuation, and the major effect of higher frequency reflections on the cone can usually be seen in the impedance curve as a resonance, if such are significant. But there are also the far more significant and problematic spider resonances that are also visible in that curve.
Typically I think that we see things the same, but maybe use different points of view. Since MP is not a concept that is ever used in Acoustics, I find its application to 3D acoustical problems problematic. You see it as a generalized extension of the concepts, but admit that non-CD requires different EQ for different angles, which to me means that the "system" cannot be MP because there is not a single filter that can correct the time, frequency and space responses simultaneously. Now, we completely agree that CD is correctable in all these aspects and thus, to me, this can be thought of as MP. But, it seems to me that there is no generally accepted generalization of MP to space, which is why it does not appear in the acoustics literature. We will always disagree when we are free to use different definitions and choose them as such.
I also agree with your comments to Lynn about boxes. I would like to add that Lynn's comment that cones are "acoustically transparent" is completely false. They do admit to some transmission, but "transparent" is completely incorect. Even my "thin as can be" bed sheet projection screen is not completely transparent (and its intended to be). A screen of cone material would be vastly more loss. Box reflections act as a spring far more than any other effect, especially when there is absorption. The idea that sound simply "comes back through the cone" is simply untrue. There is a great deal of attenuation, and the major effect of higher frequency reflections on the cone can usually be seen in the impedance curve as a resonance, if such are significant. But there are also the far more significant and problematic spider resonances that are also visible in that curve.
john k... said:
I know.
The reason I asked – *I* actually was extensively looking after a possibility to avoid box dampening completely.
The MP concept you made absolutely clear in your last posts – *this* is what I've been looking for - if it allows for 100% substitution of dampening.
john k... said:
Would say it sounds like my OB !
But my question remains – would you say that ideal EQing could be used *instead* of dampening ?
If the MP theory holds (to even that extent) in praxis, I expect to get rid of the coloration that come with the indifferent dampening materials we have available.
Anyone else who possibly has tried?
Michael
gedlee said:
Like to back up that diaphragms of speakers are rather "transparent" than are able to provide substantial dampening / filtering of whats happening at the rear.
Made several measurements during my enclosure investigation right in front of the diaphragm and right at its back - guess what – no difference that's worth to talk about – at least not in any extent that CSD resonances would look any different.
And that's the final aim, isn't it.
Michael
That test would not show anything. The diaphragm is radiating the sound in exactly equal amounts on both sides. Hence the SPL should be identical on each side. Actually the box side should be much higher as the frequency drops. But this is simply the box compliance at work.
To do a test of cone transmission properly, you would have to put a sound source inside of the box with no speaker installed and measure at several exterior points and average and then install the speaker and take the same measurements and compare. There will be a strong peak at the drivers resonance. If the cone were actually acoustically transparent then this test would measure the same both ways. The differences are not completely transmission loss, but mostly so.
To do a test of cone transmission properly, you would have to put a sound source inside of the box with no speaker installed and measure at several exterior points and average and then install the speaker and take the same measurements and compare. There will be a strong peak at the drivers resonance. If the cone were actually acoustically transparent then this test would measure the same both ways. The differences are not completely transmission loss, but mostly so.
gedlee said:
In the end effect the same as I said - all the rest - just a more sophisticated point of view IMO
You can look at it as a combined system "speaker / enclosure" which results in radiating to the front what the enclosure dictates at the back – or put it simple (simplified if you will) saying the diaphragm behaves acoustically transparent (more or less).
Michael
gedlee said:
I know we see MP differently. I see it just as a mathematical relationship between amplitude and phase. If I stick a mic out in space and measure the amplitude and phase, if, after the propagation delay is removed, the measured phase can be replicated by applying the HBT to the measured amplitude I call it MP. I think there is more importance to this than you like to accept. For example, when a crossover is designed by defining an acoustic target and then constructing a (MP) filter to shape the raw driver response to that target we rely on the the MP characteristics of the driver, referenced to the design point, to assure that the phase of the acoustic target will be correct (other than driver offsets) so that it will correctly sum to the other drivers. Because of the MP behavior it is possible to substitute a different woofers in a two way system which may have very different raw SPL response so long as the filter, specific for each driver, shapes (equalizes) the response to the correct acoustic target (assuming offsets and sensitivity are correct). If the drivers did not exhibit MP behavior relative to the design point shaping the amplitude response to the target would not imply the correct phase and summed response.
mige0 said:
I've never looked at it. Realize that when I have talked about eq I'm really looking at eq as I posted for my 8" dipole.
mige0 said:
The comments from you and Earl about cones being acoustically transparent. Realize that for any sound that reflects off the back side of the driver to be transmitted through the cone means that an acoustic wave impacting on the back side of the cone must impart energy to the cone, creating cone vibrations which then re-radiate the energy form both the front and back sides. When a mic is placed close to the cone front or back surface what you end up measuring is more closely related to the cone motion. But that won't tell you whether the motion is due to the applied signal alone or influenced by reflections.
I've done some testing in some unique ways to look at this which I don't want to discuss just yet until I'm sure of the validity of the results. Maybe I'll bring them up to see what Earl thinks. What I will say is that cone drivers make very poor microphones.
john k... said:
John
I know all that you are saying, its not new, but to me doing a crossover design at a "design point" as you suggest is the issue. I don't do crossovers that way and so MP is of no use to me because it tells you nothing about how the crossover will work spatially - only at that single point. It's lack of spatial relevence is a considerable failing for acoustics.
Why would you want the driver to be a microphone?
A quick point on acoustic transparency of cones:
When doing the transparency test suggested by Earl, or something similar, the speaker should be connected to the amplifier of interest, with the input to the amp shorted or otherwise set to zero with the amp on. This will account for the damping effect of the amplifier on cone motion.
John G
When doing the transparency test suggested by Earl, or something similar, the speaker should be connected to the amplifier of interest, with the input to the amp shorted or otherwise set to zero with the amp on. This will account for the damping effect of the amplifier on cone motion.
John G