Re: hyper-vivid tonality
Hi
I guess it's rather 2nd harmonics involved as there is no symmetry around anything and IM would be an even better start for description.
In addition to that there also should be found kind of delay or phase shift involved due to the difference in the speed of sound.
In this respect it is also different to something created by simple electronics as the "signature" is created along the way when passing compressed or expanded air along the horn and within the compression chamber. This takes some time depending on the distance the waves have to travel through.
The result should have quite a unique sound pattern, actually nothing else than a specific kind of distortion we are all familiar with from real world but is not specified in technical terms yet.
Just my private not fully thought through theory!
Greetings
Michael
Hi
I guess it's rather 2nd harmonics involved as there is no symmetry around anything and IM would be an even better start for description.
In addition to that there also should be found kind of delay or phase shift involved due to the difference in the speed of sound.
In this respect it is also different to something created by simple electronics as the "signature" is created along the way when passing compressed or expanded air along the horn and within the compression chamber. This takes some time depending on the distance the waves have to travel through.
The result should have quite a unique sound pattern, actually nothing else than a specific kind of distortion we are all familiar with from real world but is not specified in technical terms yet.
Just my private not fully thought through theory!
Greetings
Michael
Been thinking about a 4-driver version, with two 18Sound 12NDA520's and two 18Sound 15NMB420's. I figured out how to do a series-parallel crossover for all 4 drivers, so the impedance stays between 4 and 8 ohms - although multi-amping is still preferable, and certainly more flexible in terms of room EQ and control of driver overlap. I should add the 4-driver version would work with almost any sets of 12 and 15-inch drivers, most certainly including Tone Tubby's.
In terms of the low Qts of the 18Sound drivers, my friend John Atwood is working on a Class D subwoofer amplifier with variable damping, yes, re-visiting the old Fisher "Z-Matic" circuit. Actually, it's not that hard to do with any feedback amplfier - the amp is re-configured to sense current as well as voltage.
There's a reasonable chance this is the one I'll build - the more I've thought about the "issues" of OB speakers, over-equalization has to be at the top of the list. This is where the horn guys have it all over us. Compare the surface areas of midbass horns to dipoles - well - most dipoles lose out, particularly when you think of just how much EQ the commercial systems are using (a lot). This system has a surface area equal to a 27-inch driver; when you add the floor image for the closely-coupled 15-inch drivers, it equals a 34-inch driver - Hartley territory. The Variable-Geometry concept means the surface area decreases with increasing frequency, so we're back to the single 12" driver by 400~500 Hz.
In terms of the low Qts of the 18Sound drivers, my friend John Atwood is working on a Class D subwoofer amplifier with variable damping, yes, re-visiting the old Fisher "Z-Matic" circuit. Actually, it's not that hard to do with any feedback amplfier - the amp is re-configured to sense current as well as voltage.
There's a reasonable chance this is the one I'll build - the more I've thought about the "issues" of OB speakers, over-equalization has to be at the top of the list. This is where the horn guys have it all over us. Compare the surface areas of midbass horns to dipoles - well - most dipoles lose out, particularly when you think of just how much EQ the commercial systems are using (a lot). This system has a surface area equal to a 27-inch driver; when you add the floor image for the closely-coupled 15-inch drivers, it equals a 34-inch driver - Hartley territory. The Variable-Geometry concept means the surface area decreases with increasing frequency, so we're back to the single 12" driver by 400~500 Hz.
Attachments
That fits most of the things you've mentioned. And the driver layout is about the only one that works.
One of my thoughts is that these dipoles start to get a comparable area ,and advantages of, a flat panel speaker but a whole lot more excursion if needed. At normal volumes they'll barely move most of the time.. While not cheap, they won't be as expensive as many high end projects, and you sure get an impressive speaker!! Very little doubt that it has the basics for good sound. At this point it makes sense to go for the best drivers possible. I suspect that having identical 12" drivers will be best- think of single malt vs. blended scotch. The blends seem logical, but the single malts have the character. Soon you will find out though, I could be wrong.
A much cheaper version might be doable later. As you mentioned this basic approach will work with many other drivers of the same size.
Nelson Pass had a current amp design published here, but possibly less power than you have in mind..
Congratulations- I really like it!
One of my thoughts is that these dipoles start to get a comparable area ,and advantages of, a flat panel speaker but a whole lot more excursion if needed. At normal volumes they'll barely move most of the time.. While not cheap, they won't be as expensive as many high end projects, and you sure get an impressive speaker!! Very little doubt that it has the basics for good sound. At this point it makes sense to go for the best drivers possible. I suspect that having identical 12" drivers will be best- think of single malt vs. blended scotch. The blends seem logical, but the single malts have the character. Soon you will find out though, I could be wrong.
A much cheaper version might be doable later. As you mentioned this basic approach will work with many other drivers of the same size.
Nelson Pass had a current amp design published here, but possibly less power than you have in mind..
Congratulations- I really like it!
Hi
Lynn, this concept looks impressive! Nothing for small rooms!
Double 15" bass seems to be the minimum needed for what you are looking for in SPL (assuming around 5-10mm linear X-max).
What exactly do you expect to be the benefits from the second 12"?
Do you expect the max excursion of the WR to drop enough to balance the additional cost and size or is it for a possibly less complex crossover?
Greetings
Michael
Lynn, this concept looks impressive! Nothing for small rooms!
Double 15" bass seems to be the minimum needed for what you are looking for in SPL (assuming around 5-10mm linear X-max).
What exactly do you expect to be the benefits from the second 12"?
Do you expect the max excursion of the WR to drop enough to balance the additional cost and size or is it for a possibly less complex crossover?
Greetings
Michael
mige0 said:
We're doing pretty well for linear Xmax. The 12NDA520 has 8mm total linear Xmax, and 22mm total peak-to-peak excursion. The 15NMB420 has 13mm total linear Xmax, and 36mm (!) total peak-to-peak excursion.
The Eminence 15" mentioned earlier has an Xmax one-third the 15NMB420 - not a small difference in terms of SPL at midbass frequencies. Then again, guitar speakers like the Tone Tubby frequently have Xmax so small it can't be measured ... 1~3 mm in the real world.
The second 12-incher (which can be replaced with a third 15-incher, possibly with some benefit) gave noticeably smoother transitions in the midbass region when I looked at the far-field in Edge. The vertical array of three drivers seems to be the minimum for smooth transitions from midrange to close-to-floor bass drivers - and if I want to have TWO bass drivers close to the floor, well, that's four drivers overall, like it or not.
In terms of size, the tradeoff between four 12-inch, two 12 & two 15's, and one 12 and three 15's, probably comes down to a tradeoff of LF extension vs baffle width. All of the variations use a singe Widerange, one Midbass, and two Bass drivers, with the lowpass filters for the MB and B drivers adjusted to the baffle width and room.
What I'm aiming for is a fairly steep area-doubling with frequency, instead of a sudden "step" at one frequency. I think this, combined with over-equalization, is one of the potential faults of OB design. I'm particularly wary of EQ response boosting.
I like EQ notch-filtering, since it withdraws energy from a frequency region where the driver is departing from constant-acceleration operation and entering an ultrahigh gain near-oscillation - mechanical resonance, which is a special-case operating region, with different rules. Thanks to resonance, you can have psuedo-efficiencies that exceed Theile-Small theoretical limits - that's what's happening with 10~15 dB peaks in metal-cone and the really wild guitar speakers. I feel it is quite reasonable practice to (greatly) reduce the power going into these frequency regions.
But boosting power - I don't feel so good about that. Drivers in the ideal case are already constant-acceleration, which means excursion going up at a 12 dB/octave rate, and LF boosting increase this slope to an astonishing 18 dB/octave!
This seems to be just asking for trouble from IM distortion at LF, cross-modulating with higher frequencies. This crossmodulation is especially malign since the maximum audibility of distortion follows the Fletcher-Munson curve, peaking at 1~5 kHz. So excursion that's been electronically boosted in the 100~200 Hz region is going to crossmodulate with 1 kHz and higher frequencies, where distortion products are much more audible.
That's the main reason Linkwitz uses multiway crossovers with high slopes - to control IM distortion from good, but low-efficiency, drivers that are being pushed really hard. This is a legitimate approach for both dipole and horn speakers - partition the spectrum to protect from out-of-band IM products, which also gives an additional degree-of-freedom to select drivers optimized for fairly narrow bands.
Speaking only for myself, it has been my experience that true multiway speakers are very difficult to integrate sonically, especially at low and high sound levels. If a complex system been optimized for 70~90 dB SPL, it probably won't sound good at less than 50 dB, or more than 100 dB. The real attraction of simple 2-ways is that system integration is much simpler, and you don't get that terrible "falls apart and lies on the floor" sensation with overly complex systems. I never really liked my 3 and 4-way systems - they just sounded too "busy" and never as natural and relaxed as the simpler systems.
This system is a 2.5 ... N design, with large, intentional overlap regions, and only one true highpass filter (for the tweeter). If there's a key design principle, that's it. I think 2.5 ... N is an interesting alternative to 3 and 4-way systems with high-slope crossovers, and I expect the new system will sound quite different than more traditional multiway designs.
The thing to be watched are the polar patterns from all those drivers. What makes it even possible are the low frequencies involved, frequencies that are well into the 1/f rolloff region. It also makes it hard to simulate and measure, since I'm bringing in addition drivers in the same region where the velocity component is getting large, and room interaction is becoming significant in terms of wavelengths. I'm pretty sure the "simulated" models will dictate excessive bass compensation - as it is, the Edge sim is showing a response down to 75 Hz for the system, a surprising result.
Well, as far as I remember, the ideea was that a freestanding horn allows you to move it along the baffle edge, for the best audible configuration.
One of the ideas was to strap the horn in a set of hinges. This way the vibrations from the massive midbas and bass units will not affect the compresion driver.
Some thoughts.
I have found it quite a challenge to achieve EQ goals with purely acoustic means (driver Fs and Q, baffle dimensions, driver placement). In my ca. 6, 7 year OB building experience I always tried to get rid of some of the filters and EQ by choosing a compensating configuration. This will work to some extent, but if you want a really exact solution, a blank slate and suitable EQ is far easier. In the 50-200 Hz region, measurement alone will be a huge challenge, because the data will contain a mixture of several effects: dipole rolloff, resonance at the knee of the rolloff (Linkwitz describes this in detail and I have consistently found the same effect), beginning driver rolloff if low Q, and "baffle step" (floor transition), plus floor reflections in an essentially un-windowable frequency range if not measured at "ground plane". To correct all these lumped effects with any precision when they in fact measure in compound, will require long trials. I found it much easier to do what SL does: address each issue in isolation, measure it in isolation (!! important), fix it independently, and then cascade the elements (in fact the filters).
An example: for my current OB implementation, I chose a 24x40" baffle, 7" deep. Width and depth were chosen to get the dipole rolloff to start just below 200 Hz, just when the gradual floor transition adds 6 dB, with the aim to begin dipole EQ at now 100 Hz and not 200 Hz. (In addition to that I wanted diffraction from baffle edges to begin at >700 us to avoid timing smearing, so a wide baffle was indicated). The depth was chosen to completely hide the back of the woofers yet without creating a cavity resonance. Driver placement and baffle height were chosen to smooth out nulls and peaks of the dipole effects over a wide range.
The result works, sort of: raw response is fairly flat down to 100 Hz, and I *could* now get rid of SL's recommended 200-100 Hz 6 dB floor transition shelving highpass, start dipole EQ at 100 Hz, and with another shelving lowpass EQ the low Q (0.22) bass driver, all down to 20 Hz. By starting dipole EQ a bit lower than calculated, the dipole resonant peak around 200 Hz would not be too apparent on measurements.
But in reality things work only approximately that way. The 200 Hz resonance is still apparent that way ( on certain notes a muddying is noticeable, especially on piano), and starting dipole EQ lower to mask it only resulted in missing EQ in the lower bass. So the peak EQ went back in and the start of the dipole EQ back to where the calculations by theory had it put. Then I noticed that the intended dipole EQ down to 20 Hz just created too much excursion for optimal sound quality. By calculation it was all sufficient, a 7.5mm Xmax 15" woofer should be enough no? It is, but not at optimal performance. But if I stopped EQ say at 40 Hz, the intended EQ starting frequency at 100 Hz meant that then the corner frequencies weren't far enough apart, so I wouldn't get the full gain and slope. So the 100-200 Hz shelving highpass went back in, and the dipole shelving lowpass went back to 200-40 Hz even though it is partially cancelled by the shelving highpass ... Basically in the end I did what SL does: address issues separately and fix them separately, usually by EQ.
In addition to that, measurements of what is going on are not really interpretable if you can't take the various filter elements in or out while tailoring the measurement setup precisely to the effect you want to pin down. For instance a ground plane measurement will not show the floor transition, so you can't use the 100-200 Hz filter - which is OK if you planned on using one. But if you plan on building the baffle such that you won't need the 100-200 Hz filter, you have to tailor the response such that a ground plane measurement will in fact show the sub-100 Hz bass 6dB down, starting at 200 Hz. (I found non- ground plane measurements completely unreliable from 100-400 Hz, floor bounce is just too erratic as an influence)
Another thing to consider: pro woofers such as my 15ND930, or the ones considered here, are spec'd at insane power levels: 500W pink noise for 2 h before taking T/S parameters, in my woofer's case. Actual T/S in-baffle at home power levels are *dramatically* different, I suppose because the suspension never softens up enough to reach the spec'd low Fs and Q. So planning should allow for deviation from spec'd T/S data, and EQ will very likely be the only option below 200 Hz.
I have found it quite a challenge to achieve EQ goals with purely acoustic means (driver Fs and Q, baffle dimensions, driver placement). In my ca. 6, 7 year OB building experience I always tried to get rid of some of the filters and EQ by choosing a compensating configuration. This will work to some extent, but if you want a really exact solution, a blank slate and suitable EQ is far easier. In the 50-200 Hz region, measurement alone will be a huge challenge, because the data will contain a mixture of several effects: dipole rolloff, resonance at the knee of the rolloff (Linkwitz describes this in detail and I have consistently found the same effect), beginning driver rolloff if low Q, and "baffle step" (floor transition), plus floor reflections in an essentially un-windowable frequency range if not measured at "ground plane". To correct all these lumped effects with any precision when they in fact measure in compound, will require long trials. I found it much easier to do what SL does: address each issue in isolation, measure it in isolation (!! important), fix it independently, and then cascade the elements (in fact the filters).
An example: for my current OB implementation, I chose a 24x40" baffle, 7" deep. Width and depth were chosen to get the dipole rolloff to start just below 200 Hz, just when the gradual floor transition adds 6 dB, with the aim to begin dipole EQ at now 100 Hz and not 200 Hz. (In addition to that I wanted diffraction from baffle edges to begin at >700 us to avoid timing smearing, so a wide baffle was indicated). The depth was chosen to completely hide the back of the woofers yet without creating a cavity resonance. Driver placement and baffle height were chosen to smooth out nulls and peaks of the dipole effects over a wide range.
The result works, sort of: raw response is fairly flat down to 100 Hz, and I *could* now get rid of SL's recommended 200-100 Hz 6 dB floor transition shelving highpass, start dipole EQ at 100 Hz, and with another shelving lowpass EQ the low Q (0.22) bass driver, all down to 20 Hz. By starting dipole EQ a bit lower than calculated, the dipole resonant peak around 200 Hz would not be too apparent on measurements.
But in reality things work only approximately that way. The 200 Hz resonance is still apparent that way ( on certain notes a muddying is noticeable, especially on piano), and starting dipole EQ lower to mask it only resulted in missing EQ in the lower bass. So the peak EQ went back in and the start of the dipole EQ back to where the calculations by theory had it put. Then I noticed that the intended dipole EQ down to 20 Hz just created too much excursion for optimal sound quality. By calculation it was all sufficient, a 7.5mm Xmax 15" woofer should be enough no? It is, but not at optimal performance. But if I stopped EQ say at 40 Hz, the intended EQ starting frequency at 100 Hz meant that then the corner frequencies weren't far enough apart, so I wouldn't get the full gain and slope. So the 100-200 Hz shelving highpass went back in, and the dipole shelving lowpass went back to 200-40 Hz even though it is partially cancelled by the shelving highpass ... Basically in the end I did what SL does: address issues separately and fix them separately, usually by EQ.
In addition to that, measurements of what is going on are not really interpretable if you can't take the various filter elements in or out while tailoring the measurement setup precisely to the effect you want to pin down. For instance a ground plane measurement will not show the floor transition, so you can't use the 100-200 Hz filter - which is OK if you planned on using one. But if you plan on building the baffle such that you won't need the 100-200 Hz filter, you have to tailor the response such that a ground plane measurement will in fact show the sub-100 Hz bass 6dB down, starting at 200 Hz. (I found non- ground plane measurements completely unreliable from 100-400 Hz, floor bounce is just too erratic as an influence)
Another thing to consider: pro woofers such as my 15ND930, or the ones considered here, are spec'd at insane power levels: 500W pink noise for 2 h before taking T/S parameters, in my woofer's case. Actual T/S in-baffle at home power levels are *dramatically* different, I suppose because the suspension never softens up enough to reach the spec'd low Fs and Q. So planning should allow for deviation from spec'd T/S data, and EQ will very likely be the only option below 200 Hz.
BASTA! versus SoundEasy
Hi
Lynn
MBK
Taking a rough guess from the intermodulation I measured in post 824
http://www.diyaudio.com/forums/showthread.php?postid=1215943#post1215943
"http://www.diyaudio.com/forums/showthread.php?postid=1215943#post1215943"
I doubt, that a XO 2,5...N system will be sufficient for let's say 110db / 1m down to 40 Hz (or as in MBK's case the chosen single 15" for more moderate listening levels)
Does SoundEasy software allow to simulate max excursion / SPL with different multiple drivers on open baffles in 2.5 way arrangement?
If not, BASTA! software from
www.tolvan.com
"www.tolvan.com "
would be a cheap alternative I would consider to buy for myself.
If I understood right with a 2.5....N approach the 12" wide range speaker would run down to its natural resonant frequency splitting the load by half every time an additional driver comes in. First with the 12" low-mid and second with the two 15" woofers.
Greetings
Michael
Hi
Lynn
MBK
Taking a rough guess from the intermodulation I measured in post 824
http://www.diyaudio.com/forums/showthread.php?postid=1215943#post1215943
"http://www.diyaudio.com/forums/showthread.php?postid=1215943#post1215943"
An externally hosted image should be here but it was not working when we last tested it.
I doubt, that a XO 2,5...N system will be sufficient for let's say 110db / 1m down to 40 Hz (or as in MBK's case the chosen single 15" for more moderate listening levels)
Does SoundEasy software allow to simulate max excursion / SPL with different multiple drivers on open baffles in 2.5 way arrangement?
If not, BASTA! software from
www.tolvan.com
"www.tolvan.com "
would be a cheap alternative I would consider to buy for myself.
If I understood right with a 2.5....N approach the 12" wide range speaker would run down to its natural resonant frequency splitting the load by half every time an additional driver comes in. First with the 12" low-mid and second with the two 15" woofers.
Greetings
Michael
I'm in agreement with both MBK and mige0. The plan all along was to transition to stereo subwoofers between 70 and 100 Hz. All the 2.5 ... N does is merely take the speaker from the 1/f transition of the widerange driver down to the point where the subwoofers take over - basically, from 200 to 70 Hz, a one-and-a-half octave range.
Within this range I'll be using separate amplification and equalizing as necessary, measuring on a flat boundary surface (the rear wall of the room, which I used before to confirm room energy spectra). There will be three sets of amplifiers: my Class A1/A2 Karna triode amplifiers, powering the WR and HF drivers; a Class AB or D amplifier for the Midbass and Bass drivers; and another Class AB or D amplifier for the stereo subwoofers. I'll use a Rane or DBX parametric equalizer for the MB, B, and SW amplifiers, and use passive crossover EQ (if necessary) for the mids and highs.
Within the 70 Hz to 20 kHz range, I want to keep upward EQ (response boosting) to the mininum, or failing that, have the boosted range covered by multiple drivers, instead of one highly stressed driver. I'm sure you can see what I'm aiming for - dipole sound with near-Klipschorn dynamics, which isn't going to happen with heavy EQ and jamming 200 watts of Class AB power into 88 dB/metre audiophile drivers - which were never designed for that kind of abuse.
So the whole business of 2, 3, or 4 midbass and bass drivers (and the way they work together) is wide open, and to be determined by measurement and audition. I don't know if I want the slant the OB backward by 5 or 10 degrees; I'll have to build it so it can adjusted by set-screw (probably on the base of the vertical strut) and audition that as well.
Similarly, the compression driver + horn will need some kind of cat's-cradle that mechanically isolates, supports the 10-kilo weight of driver+horn, and provides adjustment for front-to-back alignment and left-to-right rotation. The mounting system should also accomodate ribbon tweeters, which will need the same location in space as the CD for the horn system.
This project is not a ready-to-build speaker. People who want that should visit the websites that offer a range of plans built around popular audiophile drivers; you're getting a known quantity that follows the template of nearly every commercial audiophile speaker.
The winding path this project is taking is similar to the way I wrote about the Ariel speaker in 1992, followed by the Amity, Aurora, and Karna amplifiers, sharing my thoughts as I go along. The Ariel was a modest departure from a minimonitor, using well-known drivers of the day, a MTM, and a twin-path transmission-line cabinet. Nothing too astonishing there.
The fully-balanced triode amps, though, were a jump into the wild blue yonder, using circuits that hadn't seen the light of day since the mid-Thirties. These amplifiers were very different animals than the Williamson variants that dominated tube-amps since 1947 - and as I found out, measured and sounded quite different as well. When I started in 1997, all of the SET builders ridiculed the idea of transformer coupling and all-balanced circuits - many still do, ten years later. And of course transistor amp guys think any kind of tube amp is just goofy, a waste of time, since transistor amps solved all the problems worth solving decades ago.
In terms of builder risk, this project is about halfway between the Ariel and Amity. There's no risk of electrocution, and you don't need to know the finer points of ground-system layout and magnetic-field induction. But owning instrumentation, a decent measurement mike, and software similar to SoundEasy is a must, just as a scope and a DVM are essential for building an amplifier.
That's why the feedback and experience of ScottG, MBK, mige0, and so many other is so important - they've built systems outside the commercial sphere, and I'm very curious what they've come up with, what works, and what doesn't.
Rather than present perfect, 100% debugged, ready-to-build systems to the public, I'm willing to make my mistakes in public - and anyone that reads my stuff know I make plenty of mistakes!
I also know audio designers, like pro photographers, make lots of mistakes all the time. They just don't show them, hiding behind the image of the audio-expert. (And I do that too.) Well ... unfortunately, we don't have many geniuses in audio. The days of Major Armstrong, Blumlein, Bell Labs, and D.E.L. Shorter of the BBC Labs are long gone. We're all just working on the margins, assisted by computer modelling and better instrumentation.
Within this range I'll be using separate amplification and equalizing as necessary, measuring on a flat boundary surface (the rear wall of the room, which I used before to confirm room energy spectra). There will be three sets of amplifiers: my Class A1/A2 Karna triode amplifiers, powering the WR and HF drivers; a Class AB or D amplifier for the Midbass and Bass drivers; and another Class AB or D amplifier for the stereo subwoofers. I'll use a Rane or DBX parametric equalizer for the MB, B, and SW amplifiers, and use passive crossover EQ (if necessary) for the mids and highs.
Within the 70 Hz to 20 kHz range, I want to keep upward EQ (response boosting) to the mininum, or failing that, have the boosted range covered by multiple drivers, instead of one highly stressed driver. I'm sure you can see what I'm aiming for - dipole sound with near-Klipschorn dynamics, which isn't going to happen with heavy EQ and jamming 200 watts of Class AB power into 88 dB/metre audiophile drivers - which were never designed for that kind of abuse.
So the whole business of 2, 3, or 4 midbass and bass drivers (and the way they work together) is wide open, and to be determined by measurement and audition. I don't know if I want the slant the OB backward by 5 or 10 degrees; I'll have to build it so it can adjusted by set-screw (probably on the base of the vertical strut) and audition that as well.
Similarly, the compression driver + horn will need some kind of cat's-cradle that mechanically isolates, supports the 10-kilo weight of driver+horn, and provides adjustment for front-to-back alignment and left-to-right rotation. The mounting system should also accomodate ribbon tweeters, which will need the same location in space as the CD for the horn system.
This project is not a ready-to-build speaker. People who want that should visit the websites that offer a range of plans built around popular audiophile drivers; you're getting a known quantity that follows the template of nearly every commercial audiophile speaker.
The winding path this project is taking is similar to the way I wrote about the Ariel speaker in 1992, followed by the Amity, Aurora, and Karna amplifiers, sharing my thoughts as I go along. The Ariel was a modest departure from a minimonitor, using well-known drivers of the day, a MTM, and a twin-path transmission-line cabinet. Nothing too astonishing there.
The fully-balanced triode amps, though, were a jump into the wild blue yonder, using circuits that hadn't seen the light of day since the mid-Thirties. These amplifiers were very different animals than the Williamson variants that dominated tube-amps since 1947 - and as I found out, measured and sounded quite different as well. When I started in 1997, all of the SET builders ridiculed the idea of transformer coupling and all-balanced circuits - many still do, ten years later. And of course transistor amp guys think any kind of tube amp is just goofy, a waste of time, since transistor amps solved all the problems worth solving decades ago.
In terms of builder risk, this project is about halfway between the Ariel and Amity. There's no risk of electrocution, and you don't need to know the finer points of ground-system layout and magnetic-field induction. But owning instrumentation, a decent measurement mike, and software similar to SoundEasy is a must, just as a scope and a DVM are essential for building an amplifier.
That's why the feedback and experience of ScottG, MBK, mige0, and so many other is so important - they've built systems outside the commercial sphere, and I'm very curious what they've come up with, what works, and what doesn't.
Rather than present perfect, 100% debugged, ready-to-build systems to the public, I'm willing to make my mistakes in public - and anyone that reads my stuff know I make plenty of mistakes!
I also know audio designers, like pro photographers, make lots of mistakes all the time. They just don't show them, hiding behind the image of the audio-expert. (And I do that too.) Well ... unfortunately, we don't have many geniuses in audio. The days of Major Armstrong, Blumlein, Bell Labs, and D.E.L. Shorter of the BBC Labs are long gone. We're all just working on the margins, assisted by computer modelling and better instrumentation.
Re: Re: BASTA! versus SoundEasy
Right now, I have one REL Strata II, which complements the sound of the Ariels. But REL has moved on, and the Strata III is a different design with a different LF rolloff slope from the Strata II, so buying another REL to match isn't an option. (Huge phase shifts between L and R channels, very undesirable.)
I'll just have to sell what I have now and build a pair of subwoofers. The easiest thing is probably copy Gary Pimm's W-baffle subwoofers, which if I recall right, are built around pairs of Parts Express 15-inch drivers.
agent.5 said:
Right now, I have one REL Strata II, which complements the sound of the Ariels. But REL has moved on, and the Strata III is a different design with a different LF rolloff slope from the Strata II, so buying another REL to match isn't an option. (Huge phase shifts between L and R channels, very undesirable.)
I'll just have to sell what I have now and build a pair of subwoofers. The easiest thing is probably copy Gary Pimm's W-baffle subwoofers, which if I recall right, are built around pairs of Parts Express 15-inch drivers.
Lynn Olson said:
Thats the only commercial DIY sub that will work. Not only does it lower distortion above driver/port resonance, it so well controls an otherwise "droning" driver architecture, that it can actually be used as a bass driver (i.e. freq.s *above* 31 Hz) with more transient capable loudspeakers. Note though that its still cr@p near driver/port resonance, AND its overall output is still compromised at lower freq.s. My suggestion would be to vent it around 12-13 Hz - keeping the lossy resonance (and distortion) out of the way while providing higher power handling and a more modest size cabinet. (..of course such is suggestion is counter intuitive - most will gravitate to a sealed design without fully understanding the variables). Note: as a result I would choose the 2 12's over the single 15.
Using the adaptive EQ is interesting - since the driver is sensing its own output, I guess the vented version would have to have the vent frequency (and Q) programmed into the little feedback-control card in the power amp.
I was thinking of a sealed box, actually, since the feedback servo takes care of most of the distortion. The primary benefit of vented is substantially reduced excursion (and distortion) close to the vent frequency. With servo, that benefit mostly disappears, leaving increased SPL as the remaining asset - but with the drawbacks of vent coloration and a 4th-order highpass characteristic.
In a servo setup, what are the sealed-box DISadvantages? Curious to know.
-----
It is interesting that Rythmik is addressing the same concerns that I'm looking at, just in a different frequency band. It does make you think about custom drivers for the Bass array in the OB, although that's a project for another day.
I was thinking of a sealed box, actually, since the feedback servo takes care of most of the distortion. The primary benefit of vented is substantially reduced excursion (and distortion) close to the vent frequency. With servo, that benefit mostly disappears, leaving increased SPL as the remaining asset - but with the drawbacks of vent coloration and a 4th-order highpass characteristic.
In a servo setup, what are the sealed-box DISadvantages? Curious to know.
-----
It is interesting that Rythmik is addressing the same concerns that I'm looking at, just in a different frequency band. It does make you think about custom drivers for the Bass array in the OB, although that's a project for another day.