Thoughts
One of the best things about working at Tektronix was the team approach to tackling huge design problems - such as the portable 492-Series Spectrum Analyzer. This was a project that had never been done before: a 40 lb. instrument that was fully self-contained, had an on-screen noise floor of -80 dB with no visible spurii at any frequency from 50 Hz to 1.8 GHz, and maintained this level of performance in the field, in a moving helicopter (it was used in the Falklands War by the UK), outdoors at a radar station in Alaska, and other severe environments.
The whole engineering group was about 50 people in Building 58 at Beaverton, Oregon. That group comprised the digital-storage, microprocessor control, microwave, RF amplification, signal processing, switching power supply, and mechanical-engineering subteams - all of them had to be harmonized, since the artifacts of one aspect of the instrument could spill over into another part.
One of more famous sayings of that group was the immortal phrase: "Once you open a can of worms, you always need a bigger can to put them back". This was literally true - despite the tight space constraints, the solution to this or that problem with low-level noise, unwanted sidebands, or phase noise, always seemed to just a bit bigger, no matter what you did, or what resources you threw at it.
The Tektronix/HP "egoless" style of engineering is almost unknown in commercial audio, where a "big" design team for a given project is two to three engineers, and five marketers. Thus, big egos, especially since audio is so commingled at every stage with marketing - the most famous "designers" in this biz are basically marketers, not designers at all, since the real grunt-work is done by anonymous toilers in the vineyard (who are usually fired after the project is complete).
The combination of corporate secrecy, combined with loss of design-team integrity over the years (high turnover), ensures that nearly the entire history of audio is two steps forward, and one-and-half steps back. Wonderful products are put on the market, the original designers are replaced, the follow-on products lose some of the original magic, and ten years later it's as if the first product had never been done.
At one time, the AES Journal was mostly about advancing the art (back in the Seventies), but the last twenty years of digital compression hasn't been about advancing the art, but concealing degradation and information loss from (most) listeners. The audio equivalent of making better and better synthetic non-dairy creamers and margerines. This has discouraged industry insiders from publishing their advancements - instead we get "White Papers" that are little more than marketing gobbledegook, disguised in pseudo-technical phrases that are just good enough to take in magazine reviewers.
The net result has been that audio designers, whether "real" engineers or not, are more isolated than ever. Most have never worked in a genuine team environment and fall prey to Big HiFi Ego syndrome. That's understandable considering their profound isolation from their peers, and the general self-promoting trends of the industry. When you're the only frog in the pond, it's easy to think you're the biggest one of all. And the ponds are isolated from each other, thanks to corporate Non-Disclosure Agreements (NDAs), the breakdown of the patent system combined with the huge volume of prior art, and a system of marketing/magazine reviewing that replaces real technical content with vague marketing buzzwords.
At least the reviews at Car & Driver and Road & Track tell you how many cylinders the engine has, naturally aspirated, turbo, supercharged, shape of the torque curve and presence of torque steer, front and rear suspension geometry, and the impact of the electronic vehicle-dynamics system on handling. This level of technical discussion is taken for granted in car-enthusiast circles, and forms the customer base for the aftermarket "tuner" industry.
In audio, this level of discussion was still common as late as the Seventies, but the transition of the high-end into a fashion-driven "boutique" industry, combined with the gradual removal of serious technical content from the Big Two review magazines, drove down the technical level of discussion over time. This gradually "dumbed-down" the buying public - which I imagine was the goal all along - but also had a severe negative effect on the design community, since they had lost a common forum. There is no English-language equivalent to "MJ" magazine in Japan, or the Audiocraft magazine of the Fifties. We have the Internet and that's it.
The drawback of the Internet is the signal-to-noise ratio - the well-known "flame" problem - but also a mismatch of audiences. DIY covers a huge range of expertise and interest. Since the AES is no longer seriously focussed on advancing the art (in terms of better sound in the home), and English-language audio-enthusiast magazines have fallen by the wayside, there aren't many places for industry insiders to gather and share stories. At the trade shows, the CEOs and marketers have to be brushed off, and the larger companies have "minders" that keep a close eye on the real designers.
This nice thing about diyAudio is that it isn't. By that I mean the forum isn't limited to do-it-yourself projects - but also covers areas that are the edge of commercial audio practice. This is the part I'm interested in, and always have been. Like most of us who've done this thing commercially, we have no interest at all in following in the footstep of XYZ famous-name manufacturers. Been there, done that, got a closet full of T-shirts with company logos on them.
The much more interesting question is, what isn't out there? There are actually large gaps in the market - in loudspeakers, there's a big gap between high-quality commercial prosound and the best audiophile loudspeakers. Although Tom Danley, Siegfried Linkwitz, JP Kreskovsky, and Dr. Geddes are coming from different design perspectives, they are all working well outside standard commercial practice - and this makes what they're doing interesting to any serious audio designer. I don't expect them to agree, but the disagreements themselves are quite illuminating, and reveal interesting sets of design assumptions.
I view audio as a modern form of artisanship, combining engineering, science, and subjective esthetic choices. Artisans do better when they meet and swap things around, just as the engineering team at Tektronix did better because they were able to meet in the same room and discuss the direction they were going. I don't expect the artisans of audio to combine forces, but I do think get-togethers in forums like this are good for all concerned. Any rough edges we see are little more than the result of the isolation we all feel in the hifi industry, which celebrates big egos and plays down the results of successful collaboration.
I'm going to build a high-efficiency dipole, but don't expect a how-to manual. What you're going to see are measurements and comments about what I heard, starting with this and that driver used in a full-range mode on a reasonably large dipole baffle. Considering the huge size of this thread, it would be probably be wise to start a new one once I start publishing hard data.
In terms of sawdust, really, the more active readers already have plenty to go on right now. You gotta measure, guys, and please use something better than that awful 1/3 octave junk - what's useful to me and many others is an unsmoothed FFT, a good impulse response over 5 milliseconds (please absorb the floor reflection), and a CSD plot. Putting the baffle on a lazy susan, marking off 7.5 or 10 degrees per tick, and making a series of measurements is good too.
In terms of getting started, it seems like the minimum is a 8" widerange driver, a single 12" driver at floor level, and either a ribbon or compression-driver+waveguide for the HF. Rather than getting lost in building horns, just buy a good one, and save yourself a lot of trouble. Dr. Geddes is selling his state-of-the-art WG at a very reasonable price, and you get XT1086 for not much at all.
But I would personally start with the wideband driver, and pick one that measures and sounds good - that driver will have a strong effect on the sonic character of the entire system. If it has a lot of HF breakup and harshness, you are committed to a high-slope crossover, and a very awkward and unpleasant system-integration with the HF driver. A driver that sound agreeable over most the audio range is going to a lot easier to integrate into the rest of the system, no matter what set of choices you make.
One of the best things about working at Tektronix was the team approach to tackling huge design problems - such as the portable 492-Series Spectrum Analyzer. This was a project that had never been done before: a 40 lb. instrument that was fully self-contained, had an on-screen noise floor of -80 dB with no visible spurii at any frequency from 50 Hz to 1.8 GHz, and maintained this level of performance in the field, in a moving helicopter (it was used in the Falklands War by the UK), outdoors at a radar station in Alaska, and other severe environments.
The whole engineering group was about 50 people in Building 58 at Beaverton, Oregon. That group comprised the digital-storage, microprocessor control, microwave, RF amplification, signal processing, switching power supply, and mechanical-engineering subteams - all of them had to be harmonized, since the artifacts of one aspect of the instrument could spill over into another part.
One of more famous sayings of that group was the immortal phrase: "Once you open a can of worms, you always need a bigger can to put them back". This was literally true - despite the tight space constraints, the solution to this or that problem with low-level noise, unwanted sidebands, or phase noise, always seemed to just a bit bigger, no matter what you did, or what resources you threw at it.
The Tektronix/HP "egoless" style of engineering is almost unknown in commercial audio, where a "big" design team for a given project is two to three engineers, and five marketers. Thus, big egos, especially since audio is so commingled at every stage with marketing - the most famous "designers" in this biz are basically marketers, not designers at all, since the real grunt-work is done by anonymous toilers in the vineyard (who are usually fired after the project is complete).
The combination of corporate secrecy, combined with loss of design-team integrity over the years (high turnover), ensures that nearly the entire history of audio is two steps forward, and one-and-half steps back. Wonderful products are put on the market, the original designers are replaced, the follow-on products lose some of the original magic, and ten years later it's as if the first product had never been done.
At one time, the AES Journal was mostly about advancing the art (back in the Seventies), but the last twenty years of digital compression hasn't been about advancing the art, but concealing degradation and information loss from (most) listeners. The audio equivalent of making better and better synthetic non-dairy creamers and margerines. This has discouraged industry insiders from publishing their advancements - instead we get "White Papers" that are little more than marketing gobbledegook, disguised in pseudo-technical phrases that are just good enough to take in magazine reviewers.
The net result has been that audio designers, whether "real" engineers or not, are more isolated than ever. Most have never worked in a genuine team environment and fall prey to Big HiFi Ego syndrome. That's understandable considering their profound isolation from their peers, and the general self-promoting trends of the industry. When you're the only frog in the pond, it's easy to think you're the biggest one of all. And the ponds are isolated from each other, thanks to corporate Non-Disclosure Agreements (NDAs), the breakdown of the patent system combined with the huge volume of prior art, and a system of marketing/magazine reviewing that replaces real technical content with vague marketing buzzwords.
At least the reviews at Car & Driver and Road & Track tell you how many cylinders the engine has, naturally aspirated, turbo, supercharged, shape of the torque curve and presence of torque steer, front and rear suspension geometry, and the impact of the electronic vehicle-dynamics system on handling. This level of technical discussion is taken for granted in car-enthusiast circles, and forms the customer base for the aftermarket "tuner" industry.
In audio, this level of discussion was still common as late as the Seventies, but the transition of the high-end into a fashion-driven "boutique" industry, combined with the gradual removal of serious technical content from the Big Two review magazines, drove down the technical level of discussion over time. This gradually "dumbed-down" the buying public - which I imagine was the goal all along - but also had a severe negative effect on the design community, since they had lost a common forum. There is no English-language equivalent to "MJ" magazine in Japan, or the Audiocraft magazine of the Fifties. We have the Internet and that's it.
The drawback of the Internet is the signal-to-noise ratio - the well-known "flame" problem - but also a mismatch of audiences. DIY covers a huge range of expertise and interest. Since the AES is no longer seriously focussed on advancing the art (in terms of better sound in the home), and English-language audio-enthusiast magazines have fallen by the wayside, there aren't many places for industry insiders to gather and share stories. At the trade shows, the CEOs and marketers have to be brushed off, and the larger companies have "minders" that keep a close eye on the real designers.
This nice thing about diyAudio is that it isn't. By that I mean the forum isn't limited to do-it-yourself projects - but also covers areas that are the edge of commercial audio practice. This is the part I'm interested in, and always have been. Like most of us who've done this thing commercially, we have no interest at all in following in the footstep of XYZ famous-name manufacturers. Been there, done that, got a closet full of T-shirts with company logos on them.
The much more interesting question is, what isn't out there? There are actually large gaps in the market - in loudspeakers, there's a big gap between high-quality commercial prosound and the best audiophile loudspeakers. Although Tom Danley, Siegfried Linkwitz, JP Kreskovsky, and Dr. Geddes are coming from different design perspectives, they are all working well outside standard commercial practice - and this makes what they're doing interesting to any serious audio designer. I don't expect them to agree, but the disagreements themselves are quite illuminating, and reveal interesting sets of design assumptions.
I view audio as a modern form of artisanship, combining engineering, science, and subjective esthetic choices. Artisans do better when they meet and swap things around, just as the engineering team at Tektronix did better because they were able to meet in the same room and discuss the direction they were going. I don't expect the artisans of audio to combine forces, but I do think get-togethers in forums like this are good for all concerned. Any rough edges we see are little more than the result of the isolation we all feel in the hifi industry, which celebrates big egos and plays down the results of successful collaboration.
I'm going to build a high-efficiency dipole, but don't expect a how-to manual. What you're going to see are measurements and comments about what I heard, starting with this and that driver used in a full-range mode on a reasonably large dipole baffle. Considering the huge size of this thread, it would be probably be wise to start a new one once I start publishing hard data.
In terms of sawdust, really, the more active readers already have plenty to go on right now. You gotta measure, guys, and please use something better than that awful 1/3 octave junk - what's useful to me and many others is an unsmoothed FFT, a good impulse response over 5 milliseconds (please absorb the floor reflection), and a CSD plot. Putting the baffle on a lazy susan, marking off 7.5 or 10 degrees per tick, and making a series of measurements is good too.
In terms of getting started, it seems like the minimum is a 8" widerange driver, a single 12" driver at floor level, and either a ribbon or compression-driver+waveguide for the HF. Rather than getting lost in building horns, just buy a good one, and save yourself a lot of trouble. Dr. Geddes is selling his state-of-the-art WG at a very reasonable price, and you get XT1086 for not much at all.
But I would personally start with the wideband driver, and pick one that measures and sounds good - that driver will have a strong effect on the sonic character of the entire system. If it has a lot of HF breakup and harshness, you are committed to a high-slope crossover, and a very awkward and unpleasant system-integration with the HF driver. A driver that sound agreeable over most the audio range is going to a lot easier to integrate into the rest of the system, no matter what set of choices you make.
Thanks Lynn, a good starting point and a pretty easy one.
Sort of back to the Lucasz Fikus speaker concept. Cover the midrage as well as you can, then fill in the top and bottom.
Sort of back to the Lucasz Fikus speaker concept. Cover the midrage as well as you can, then fill in the top and bottom.
FYI, there will be very soon, a Hemp wideband whizzerless 8, designed out to ~9KHz. I believe this will be worth exploring, for smaller scale systems.
Andy Lankester of www.capilanohornspeakers.com has just received the first run of a custom order 15" designed specifically for OB bass. He also has his own bass amp with LP filtering, etc. He sent me FR plot of the driver, and it looks to be the best I have seen yet; very impressive. I am excited about both these drivers coming on the scene. The pricepoints are $118 for the HempTone 8, and $200 for the Capilano 15.
Andy Lankester of www.capilanohornspeakers.com has just received the first run of a custom order 15" designed specifically for OB bass. He also has his own bass amp with LP filtering, etc. He sent me FR plot of the driver, and it looks to be the best I have seen yet; very impressive. I am excited about both these drivers coming on the scene. The pricepoints are $118 for the HempTone 8, and $200 for the Capilano 15.
Ray Collins said:
Ooops. Sorry, too many http in the address.
Here it is. http://www.lampizator.eu/NIRVANA/nirvana.html
Basically - Lucasz is saying start with the best midrange driver you can, "If midrange is wrong, nothing else matters." He likes vintage 8" Alnico wideband in OB, with a ribbon on top.
Similar to what's been discussed here.
round over close to the speaker
Hi
Earl, thanks for explanation but your patent seems to give me the answer already.
In case of a flat baffle ( BLUE contour) there is NO benefit of any round over at the speaker like there is for any other - less than 180 degree - angle ( RED contour for exampel).
This surprises me – hence my question – I would have thought that there is something better than just to mount the speaker flat to the baffle if NO wave guide is desired.
This also seems to imply that the OS curve becomes less beneficial at very wide wave guides, no ?
###############
1.) What is the distinction here between a " semi-direct radiator " and a "real" compression driver assembly ? Isn't it that both have less aperture than the diaphragm area?
2.) what do you mean with a " permissible contour " - are there any other contours ( than OS ) of minimal diffraction?
(Pic taken from your patent US 7.095.868. B2)
Greetings
Michael
Hi
Earl, thanks for explanation but your patent seems to give me the answer already.
In case of a flat baffle ( BLUE contour) there is NO benefit of any round over at the speaker like there is for any other - less than 180 degree - angle ( RED contour for exampel).
An externally hosted image should be here but it was not working when we last tested it.
This surprises me – hence my question – I would have thought that there is something better than just to mount the speaker flat to the baffle if NO wave guide is desired.
This also seems to imply that the OS curve becomes less beneficial at very wide wave guides, no ?
###############
1.) What is the distinction here between a " semi-direct radiator " and a "real" compression driver assembly ? Isn't it that both have less aperture than the diaphragm area?
2.) what do you mean with a " permissible contour " - are there any other contours ( than OS ) of minimal diffraction?
An externally hosted image should be here but it was not working when we last tested it.
(Pic taken from your patent US 7.095.868. B2)
Greetings
Michael
Re: Thoughts
You must have a lot of time on your hands Lynn, but I'm not complaining as I find your posts entertaining as well as informative.
Talking of the design group above......what is our next project after 'after the ariels' ha ha. Don't mind me, I definitely fall into the internet mismatch of audiences, meaning I don't understand much of this at all!!
Given the way design this has developed and grown, and the means by which it has done so, AND keeping with the greek theme for names, perhaps it should be called Diyana?
I'll let those with more knowledge than I explain what Diana did or didn't do, and I'll probably be corrected and find out she is Roman or something ha ha.
Lynn Olson said:
You must have a lot of time on your hands Lynn, but I'm not complaining as I find your posts entertaining as well as informative.
Talking of the design group above......what is our next project after 'after the ariels' ha ha. Don't mind me, I definitely fall into the internet mismatch of audiences, meaning I don't understand much of this at all!!
Given the way design this has developed and grown, and the means by which it has done so, AND keeping with the greek theme for names, perhaps it should be called Diyana?
I'll let those with more knowledge than I explain what Diana did or didn't do, and I'll probably be corrected and find out she is Roman or something ha ha.
Contact Andy Lankester at www.capilanohornspeakers.com He is in White Rock, British Columbia
He was having troubles posting the info on his site. He will be happy to forward to you, the information.
He was having troubles posting the info on his site. He will be happy to forward to you, the information.
I've been taking a little vacation from hifi.
You guys want pictures, huh? Well, check these out (it's OK, they're work-safe). Looks like we're in the middle of a UFO flap - these happen every few years.
As a connoisseur of the fringes of consensus reality, I present for your reading pleasure, the account from "Isaac" of his years at the Palo Alto CARET laboratory. The 4th-Quarter Report of the CARET team is certainly a lot more interesting than what I was doing at Tektronix in 1986.
Well, I can't claim to have reverse-engineered antigravity from recovered alien artifacts, but I think I've made a (rather small) contribution to the dipole discussion over all of these posts. To summarize:
1) Dipoles can make up the intrinsic efficiency loss below the 1/f rolloff frequency through equalization, an increase in emissive area, or both. To keep distortion the same as an equivalent monopole system, the cone area needs to increase 4 times for every octave below the 1/f frequency.
For example, if you compare monopole and dipole systems using 12" woofers, with the dipole having a 1/f of 160 Hz and meeting a subwoofer at 80 Hz, the dipole system needs to have the area of 4 woofers at 80 Hz to have comparable distortion to the single-driver monopole system that covers the same frequency range. In practice, by taking advantage of the floor image, this requirement can be relaxed to using 2 woofers at 80 Hz - assuming both woofers close to floor level.
It can be seen relying on the low-distortion properties of selected audiophile woofers may not be sufficient - the distortion probably isn't going to be several times lower (as it needs to be), and power-handling hits the SPL/excursion wall much sooner than the equivalent monopole system. Like cars, there's no substitute for cubic inches (of displacement).
Although 12 and 15-inch midbass drivers with smooth response to 1~2 kHz are readily available, the same is not true for 18-inch and larger drivers. Drivers in the 18-inch and larger category appear to be designed for subwoofer use only, with response that's quite rough above 200 Hz - bad enough that 24 dB/oct active crossovers are required.
Given this information, the best approach for dipoles made with existing drivers is to select several wide range, low-distortion midbass drivers in the 12 to 15-inch range. The number of drivers required is the result of the spectral distance between the 1/f frequency and the crossover frequency of the monopole subwoofer.
Although it is tempting to keep the number of drivers down by using a high subwoofer crossover (right at the 1/f frequency, for example), this re-introduces the problem with box colorations. Much of the reason to use dipoles is to remove box coloration; re-introducing it from the subwoofer defeats much of the "boxless" advantage.
This is a good reason to keep the subwoofer as physically small as possible, use the lowest crossover practical, and use a fairly high rolloff slope. The lower the subwoofer crossover, of course, the more drivers required for the OB, so it's a tradeoff. That, for me, is a primary advantage of the 12" closed-box Rythmik - the box is small and rigid, the driver distortion is low, and they can be used in groups of two or four, as needed.
2) Although there are no box standing-waves to contend with, diffraction is still with us, thanks to the sharp boundaries at the edge of the baffle. A lossy mesh softens the transition, spreading out the area (and time) of the boundary.
The simplest method of spreading out the boundary might be nothing more than a large number of narrow tangential (or is it radial?) slots cut into the baffle, radiating out from the center of the widerange driver. If the slots start at the edge of the driver, extend to the edge of the baffle, and there are enough of them (20, 30 or more), there should be a measurable improvement in diffraction.
3) Although this thread opened with much optimism about coaxial drivers, the frequency response curves for the woofer portion of these drivers have not been very promising, never mind the doubtful response of the horns. In particular, the 12 and 15-inch woofers for the coaxes I've looked at so far appear quite a bit worse in the critical 1~5 kHz region than equivalent, monitor-grade midbass drivers. Most of the advantage of a coax is defeated if the rough response of the drivers dictates a 24 dB/octave crossover.
The best drivers are typically the ones that have the mildest demands on the crossover, fashion trends notwithstanding. A 24 dB/octave crossover slope, although high, is not infinite, and high-Q resonances can still be audible after substantial attenuation. Although equalized-flat speakers can sound OK on first listen, there can be a subtle fatiguing quality that builds up over time - this is the biggest reason I avoid drivers with really severe peaks, or rough rolloff curves.
P.S. You tell me.
You guys want pictures, huh? Well, check these out (it's OK, they're work-safe). Looks like we're in the middle of a UFO flap - these happen every few years.
As a connoisseur of the fringes of consensus reality, I present for your reading pleasure, the account from "Isaac" of his years at the Palo Alto CARET laboratory. The 4th-Quarter Report of the CARET team is certainly a lot more interesting than what I was doing at Tektronix in 1986.
Well, I can't claim to have reverse-engineered antigravity from recovered alien artifacts, but I think I've made a (rather small) contribution to the dipole discussion over all of these posts. To summarize:
1) Dipoles can make up the intrinsic efficiency loss below the 1/f rolloff frequency through equalization, an increase in emissive area, or both. To keep distortion the same as an equivalent monopole system, the cone area needs to increase 4 times for every octave below the 1/f frequency.
For example, if you compare monopole and dipole systems using 12" woofers, with the dipole having a 1/f of 160 Hz and meeting a subwoofer at 80 Hz, the dipole system needs to have the area of 4 woofers at 80 Hz to have comparable distortion to the single-driver monopole system that covers the same frequency range. In practice, by taking advantage of the floor image, this requirement can be relaxed to using 2 woofers at 80 Hz - assuming both woofers close to floor level.
It can be seen relying on the low-distortion properties of selected audiophile woofers may not be sufficient - the distortion probably isn't going to be several times lower (as it needs to be), and power-handling hits the SPL/excursion wall much sooner than the equivalent monopole system. Like cars, there's no substitute for cubic inches (of displacement).
Although 12 and 15-inch midbass drivers with smooth response to 1~2 kHz are readily available, the same is not true for 18-inch and larger drivers. Drivers in the 18-inch and larger category appear to be designed for subwoofer use only, with response that's quite rough above 200 Hz - bad enough that 24 dB/oct active crossovers are required.
Given this information, the best approach for dipoles made with existing drivers is to select several wide range, low-distortion midbass drivers in the 12 to 15-inch range. The number of drivers required is the result of the spectral distance between the 1/f frequency and the crossover frequency of the monopole subwoofer.
Although it is tempting to keep the number of drivers down by using a high subwoofer crossover (right at the 1/f frequency, for example), this re-introduces the problem with box colorations. Much of the reason to use dipoles is to remove box coloration; re-introducing it from the subwoofer defeats much of the "boxless" advantage.
This is a good reason to keep the subwoofer as physically small as possible, use the lowest crossover practical, and use a fairly high rolloff slope. The lower the subwoofer crossover, of course, the more drivers required for the OB, so it's a tradeoff. That, for me, is a primary advantage of the 12" closed-box Rythmik - the box is small and rigid, the driver distortion is low, and they can be used in groups of two or four, as needed.
2) Although there are no box standing-waves to contend with, diffraction is still with us, thanks to the sharp boundaries at the edge of the baffle. A lossy mesh softens the transition, spreading out the area (and time) of the boundary.
The simplest method of spreading out the boundary might be nothing more than a large number of narrow tangential (or is it radial?) slots cut into the baffle, radiating out from the center of the widerange driver. If the slots start at the edge of the driver, extend to the edge of the baffle, and there are enough of them (20, 30 or more), there should be a measurable improvement in diffraction.
3) Although this thread opened with much optimism about coaxial drivers, the frequency response curves for the woofer portion of these drivers have not been very promising, never mind the doubtful response of the horns. In particular, the 12 and 15-inch woofers for the coaxes I've looked at so far appear quite a bit worse in the critical 1~5 kHz region than equivalent, monitor-grade midbass drivers. Most of the advantage of a coax is defeated if the rough response of the drivers dictates a 24 dB/octave crossover.
The best drivers are typically the ones that have the mildest demands on the crossover, fashion trends notwithstanding. A 24 dB/octave crossover slope, although high, is not infinite, and high-Q resonances can still be audible after substantial attenuation. Although equalized-flat speakers can sound OK on first listen, there can be a subtle fatiguing quality that builds up over time - this is the biggest reason I avoid drivers with really severe peaks, or rough rolloff curves.
P.S. You tell me.
Lynn Olson said:
I think you have to be a little careful here. One point that is often missed with dipole systems is what happens at Fequal. Fequal is the frequency at which a dipole and monopole have the same level. Fequal = C/(6D) where C = sound speed and D is the dipole separation. What this says is that a dipole composed of two omni directional sources, each radiating at an intensity of I, will have the same on axis intensity at Fequal as a single omni directional source radiating with intensity I. But a funny thing happens when we use a conventional driver on an opened baffle. Say the driver has a 2Pi sensitivity of 90 dB/W/M. Put the driver in a box and at low frequency (in free space) the sensitivity drops to 84dB. I.e. it’s an omni directional source with 84dB sensitivity. Now, place the driver on an open baffle such that at Fequal the driver would be operating in its omnidirectional frequency range and you will find that at Fequal the dipole sensitivity will be 90 dB, just as if the dipole were composed of to omni directional point sources in free space with 90 dB sensitivity. That is a little bonus. (pretty easy to verify with a measurement.)
The reason for this is what happens at the baffle edge which goes back to the argument that Earl and I were having some time ago. Just to clear that up a little, I would agree with Earl that the strength of the wave wrapping around the baffle in a dipole is twice that which would result from edge diffraction in a conventional speaker. But what is and isn't diffraction then becomes somewhat of a semantics argument depending on the point of reference. The bottom line is that Earl and I actually aren't in disagreement, but were we’re looking at different aspects of the same problem with different interpretations.
Lynn Olson said:
Here again I urge some caution. Suppose that you have a circular baffle. If there are no slots then the rear wave can be though of as a ring radiator at the baffle edge. On axis this reduces to a single source at the observation point. That is, the delay from all points around the circumference of the baffle to a point on the dipole axis will be the same. Cutting radial slots in the baffle would be something like spreading the edge ring radiator out over radius of the baffle. That is you are time smearing the rear wave. This may well have some smoothing effect above the first dipole peak (what you refer to as above the 1/f point) but is will also likely move the 1/f point higher in frequency since it will make the baffle look smaller to low frequency. I have some old circualr baffle lying around. Maybe if I have time this week I'll cut some slots in one and post the results. But the point here is that what the slots are doing is time smearing the rear wave. You get a similar effect by going from a circular baffle to a rectangular of other shaped baffle where the path lengths to the baffle edge vary around the circumference.
However, there are other issues you might want to consider with the slotted concept. First, since the slots will go all the way in to (or close to) the edge of the driver, the intensity of any "signal" that “leaks” through the slots will potentially be of greater amplitude that that which would exist at the baffle edge. (Attenuation due to propagation distance). Second, the slots will alter the structural integrity of the baffle and may introduce numerous individual resonances since the material between the slots could resonate at different frequencies.
Lastly there is this issue of using the driver above the 1/f point. Again, consider a circular baffle. There is a desire to use large baffles to improve the low frequency sensitivity by placing the 1/f point at a relatively low frequency. However, in doing that the driver tends to still be operating in an omni directional manner above the 1/f point and the peaks and null in the response above 1/f are severe. This behavior is also indicative of a multi-lobed, frequency dependent radiation pattern.
On the other side of the issue is using a smaller baffle. With a smaller baffle the 1/f frequency is raised and we suffer the loss of sensitivity at low frequency. However, by pushing the 1/f point higher we hopefully get it into the range where the driver is becoming directional. Since what controls the strength of the rear wave (and also any other diffraction effects) is the strength of the driver's front and rear response at 90 degrees off axis, as the driver becomes directional the response above the 1/f point naturally becomes smoother, even for a circular baffle. Additionally, pushing the 1/f point higher means that it approaches the crossover point to the tweeter and what is happening above it is no longer as much of a concern.
Lossy mesh
Cymat is a foamed aluminum material I've been looking at in another context - it might make a good panel boundary material. It's interesting looking stuff, as well.
Apparently when you cut it, the cell edges are sharp, so this would have to be resolved somehow to prevent the panel edge from being hazardous: epoxy fill, perhaps, or a thin wooden frame at the outer edges?
Regards.
Aengus
Cymat is a foamed aluminum material I've been looking at in another context - it might make a good panel boundary material. It's interesting looking stuff, as well.
Apparently when you cut it, the cell edges are sharp, so this would have to be resolved somehow to prevent the panel edge from being hazardous: epoxy fill, perhaps, or a thin wooden frame at the outer edges?
Regards.
Aengus
Hi all,
Wondering if I might get some roughing up here. I have attached an EDGE file that seems to be quite nice in it's response characteristics. I intend to use a Lowther 15 ohm PM6A on top (EnABL'd, of course)
The two lower drivers will be 8 ohm Eminence Beta eight drivers. The top one firing in open baffle and the lower one feeding a dual tube 9 foot by 4 inch diameter transmission line, with the driver set at 1/4 pipe length. Only looking for 40 Hz or so and the two Beta eights will be in series, to match the Lowther.
Looking like I can cross at 6 db per octave just about wherever I want to, though likely not higher than 300 Hz.
The Beta eights will also be EnABL'd. Ditto the baffle, which will have the "wings" bent back at about 75 degrees on either side at the 0 to -.5 meter area on the left and at the .3 to .35 meter area on the right. The stabilizer legs will allow the entire thing to rest with about a 10 to 15 degree angle to the front of the baffle.
I would appreciate any comments about obvious design flaws and general stupidity.
Bud
Wondering if I might get some roughing up here. I have attached an EDGE file that seems to be quite nice in it's response characteristics. I intend to use a Lowther 15 ohm PM6A on top (EnABL'd, of course)
The two lower drivers will be 8 ohm Eminence Beta eight drivers. The top one firing in open baffle and the lower one feeding a dual tube 9 foot by 4 inch diameter transmission line, with the driver set at 1/4 pipe length. Only looking for 40 Hz or so and the two Beta eights will be in series, to match the Lowther.
Looking like I can cross at 6 db per octave just about wherever I want to, though likely not higher than 300 Hz.
The Beta eights will also be EnABL'd. Ditto the baffle, which will have the "wings" bent back at about 75 degrees on either side at the 0 to -.5 meter area on the left and at the .3 to .35 meter area on the right. The stabilizer legs will allow the entire thing to rest with about a 10 to 15 degree angle to the front of the baffle.
I would appreciate any comments about obvious design flaws and general stupidity.
Bud
Hi Bud,
The Eminences will have quite different impedance characteristics imposed by their different mounting/loading, so they cannot be expected to work in series - only parallel. However in parallel they will present a much more respectable load.
Surely the combo will go down to 25Hz via the pipe though be displacement limited.
This is something similar which I am doing myself at the moment. 6x9 drivers in front and back for bipole. Folded line between drivers is 8ft long. External = 10" wide by 9" deep by 52" high.
Dowel rod now either edge of both front and rear panels, like a rounded baffle edge; this does make a difference with mid peaks if you take them away.
Lots of holes now drilled in sides behind the front driver (only) to make the front driver semi-OB whilst simultaneously venting the line.
Both the drivers and the vent have phase matched output at 32Hz, with respectable output down to 25Hz, esp with room gain.
Work continuing, but high WAF and good sound so a definite keeper.
Cheers ........ Graham.
The Eminences will have quite different impedance characteristics imposed by their different mounting/loading, so they cannot be expected to work in series - only parallel. However in parallel they will present a much more respectable load.
Surely the combo will go down to 25Hz via the pipe though be displacement limited.
This is something similar which I am doing myself at the moment. 6x9 drivers in front and back for bipole. Folded line between drivers is 8ft long. External = 10" wide by 9" deep by 52" high.
Dowel rod now either edge of both front and rear panels, like a rounded baffle edge; this does make a difference with mid peaks if you take them away.
Lots of holes now drilled in sides behind the front driver (only) to make the front driver semi-OB whilst simultaneously venting the line.
Both the drivers and the vent have phase matched output at 32Hz, with respectable output down to 25Hz, esp with room gain.
Work continuing, but high WAF and good sound so a definite keeper.
Cheers ........ Graham.
Attachments
Hi
Bud, looking at the pic's below, I think you will need either a lot of Earls recommendation for that shape – or astronomic distances for your listening position.
What I do with EDGE to get an idea about a specific shape is to put the mic into a 2m distance and make a rough plot
- at axis
- at 30 deg horizontally off axis
- at 30 deg vertically off axis
- and two positions in between
This almost ALLWAYS reveals that you have only the first peak that stays more or less in shape – everything at higher frequencies you better forget about.
To ease crossover design any simple baffle shape is advantageous as shown with the square baffle as an example
Its interresting to do the same with existing designs as well.
Greetings
Michael
Bud, looking at the pic's below, I think you will need either a lot of Earls recommendation for that shape – or astronomic distances for your listening position.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
What I do with EDGE to get an idea about a specific shape is to put the mic into a 2m distance and make a rough plot
- at axis
- at 30 deg horizontally off axis
- at 30 deg vertically off axis
- and two positions in between
This almost ALLWAYS reveals that you have only the first peak that stays more or less in shape – everything at higher frequencies you better forget about.
To ease crossover design any simple baffle shape is advantageous as shown with the square baffle as an example
Its interresting to do the same with existing designs as well.
Greetings
Michael