So I'm dreaming of building the ultimate speakers and I want to see if I've got some of the right priniciples, so far.
1. Active crossovers - the benefits of active vs. passive is a given, so no explanation needed, right?
2. Use large drivers for the bass because of doppler distortion or whatever you want to call it. Try to have these drivers operate cleanly to about 300Hz because of midrange drivers (see 3). Given this, you'll probably want a 4 way system if you're going to use it for home theater, because finding drivers that play cleanly and loudly from 20 Hz to 300Hz is very unlikely or very very expensive.
3. Use approximately 5" metal drivers. OK, here's where I get a little contronversial. The more I read about the mid/woofers, the more I see the value in smaller drivers. Those 7 and 8" drivers are great for getting down to close to 40Hz in a 2 way but they are often difficult to crossover at 2-3kHz because of the the resonance peaks they have near 4-6khz. The smaller 5" drivers seem to give you a much smoother transition to the tweeter, as long as your bass drivers can help it out on the other end.
The other thing that I've been looking at is the use of metal cone drivers. If you check out the harmonic distortion figures for the metal Seas L12 http://home.new.rr.com/zaph/audio/4.5test/ compared to the damped paper Vifa TC14 http://home.new.rr.com/zaph/audio/surface5test/ you'll notice that the metal less distorted - I think I'm reading that right? With the right active crossover (metals all have such high resonances that I wouldn't want to use all the passive components necessary because of the phase shifts, etc) they might sound pretty good. Linkwitz chose the metal Seas W18 for his midrange (with an active crossover) so maybe there is something there.
The other benefit of using a smaller midrange driver and crossing it over around 200 hz is that you wouldn't have the floor cancellation that you see by mounting a midwoofer close to ear level, right? Maybe use a MTM arrangement to give greater sensitivity, power handling and even less vertical dispersion acoustic issues.
4. Buy a really great tweeter - but since you've used a smaller midrange you don't have to worry about one that crosses low (like 1.5kHz) and get's close to it's Fs.
Am I on the right track, or have I missed something?
1. Active crossovers - the benefits of active vs. passive is a given, so no explanation needed, right?
2. Use large drivers for the bass because of doppler distortion or whatever you want to call it. Try to have these drivers operate cleanly to about 300Hz because of midrange drivers (see 3). Given this, you'll probably want a 4 way system if you're going to use it for home theater, because finding drivers that play cleanly and loudly from 20 Hz to 300Hz is very unlikely or very very expensive.
3. Use approximately 5" metal drivers. OK, here's where I get a little contronversial. The more I read about the mid/woofers, the more I see the value in smaller drivers. Those 7 and 8" drivers are great for getting down to close to 40Hz in a 2 way but they are often difficult to crossover at 2-3kHz because of the the resonance peaks they have near 4-6khz. The smaller 5" drivers seem to give you a much smoother transition to the tweeter, as long as your bass drivers can help it out on the other end.
The other thing that I've been looking at is the use of metal cone drivers. If you check out the harmonic distortion figures for the metal Seas L12 http://home.new.rr.com/zaph/audio/4.5test/ compared to the damped paper Vifa TC14 http://home.new.rr.com/zaph/audio/surface5test/ you'll notice that the metal less distorted - I think I'm reading that right? With the right active crossover (metals all have such high resonances that I wouldn't want to use all the passive components necessary because of the phase shifts, etc) they might sound pretty good. Linkwitz chose the metal Seas W18 for his midrange (with an active crossover) so maybe there is something there.
The other benefit of using a smaller midrange driver and crossing it over around 200 hz is that you wouldn't have the floor cancellation that you see by mounting a midwoofer close to ear level, right? Maybe use a MTM arrangement to give greater sensitivity, power handling and even less vertical dispersion acoustic issues.
4. Buy a really great tweeter - but since you've used a smaller midrange you don't have to worry about one that crosses low (like 1.5kHz) and get's close to it's Fs.
Am I on the right track, or have I missed something?
Pretty much similar to my implementation of a front running speaker. One high quality treble, a good strong true midrange and finally 8" drivers from 380hz and down. Only 3 way but it has a trick up its sleeve:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=52876
Only thing I've deviated on is the large bass driver and instead chose 2 or 3 x 8" drivers along with a LT circuit to boost low end. Response should be in the order of 15hz-40Khz in room.
This is only one way and there are others, though this is definitely the easiest way for me to get top performance from good drivers.
http://www.diyaudio.com/forums/showthread.php?s=&threadid=52876
Only thing I've deviated on is the large bass driver and instead chose 2 or 3 x 8" drivers along with a LT circuit to boost low end. Response should be in the order of 15hz-40Khz in room.
This is only one way and there are others, though this is definitely the easiest way for me to get top performance from good drivers.
First thing - you shouldn't be concerned about budget.
Seems to me that the best way method of picking drivers is to pick the largest ones that won't beam at the frequency you want to cross at.
Larger drivers tend to have lower distortion at the same given frequency and output level, because they have to move less.
Note that the crossover frequencies aren't static - you can change them to suit, not the other way around. Linkwitz uses a 1.4kHz crossover between tweeter and midrange - you can use one that's lower or higher depending on what drivers you use.
Now, the driver(s) covering the modal region (generally starting around 200Hz and going down to the frequency whose wavelength is 1/2 the length of the room), should be operated as a dipole, not monopole. This is because dipole bass is almost always less peaky/nully (not a word, I know) than monopole bass in this region. With dipole bass, you'll need either EQ or lots of space in order to achieve a flat response, and I'd prefer the latter, as you need extra power (i.e. more distortion) to get the same output level below the point where dipole cancellation (6dB/octave rollof) occurs.
Since this has all been covered in numerous threads in this forum, just do a search.
Seems to me that the best way method of picking drivers is to pick the largest ones that won't beam at the frequency you want to cross at.
Larger drivers tend to have lower distortion at the same given frequency and output level, because they have to move less.
Note that the crossover frequencies aren't static - you can change them to suit, not the other way around. Linkwitz uses a 1.4kHz crossover between tweeter and midrange - you can use one that's lower or higher depending on what drivers you use.
Now, the driver(s) covering the modal region (generally starting around 200Hz and going down to the frequency whose wavelength is 1/2 the length of the room), should be operated as a dipole, not monopole. This is because dipole bass is almost always less peaky/nully (not a word, I know) than monopole bass in this region. With dipole bass, you'll need either EQ or lots of space in order to achieve a flat response, and I'd prefer the latter, as you need extra power (i.e. more distortion) to get the same output level below the point where dipole cancellation (6dB/octave rollof) occurs.
Since this has all been covered in numerous threads in this forum, just do a search.
Anyway, my pick for a system with minimal need for EQ and no separate subwoofer, listening distance around 4m or less:
Bass: Precision Devices 24" PD2450 - dipole operated on a flat baffle up to 200Hz-250Hz, 4th-order linear phase crossover or steeper. Although all the drivers are vertically aligned, the drivers aren't placed symmetrically on the baffle for smoother response.
Midrange: JBL 12" 2206, probably cardiod operation (theoretically, more seamless merging between dipole and monopole), crossed to tweeter at about 1kHz-1.2kHz.
Treble: Probably a 1"-throat beryllium-diaphragm compression driver + plug-filled OS waveguide designed by Dr. Geddes (phase plug and waveguide matched). The treble response will have to be EQed as the on-axis response drops as frequency increases.
Quick and easy, although that can be improved upon tremendously. A good start for a highly dynamic, relatively flat (in any room) system.
Bass: Precision Devices 24" PD2450 - dipole operated on a flat baffle up to 200Hz-250Hz, 4th-order linear phase crossover or steeper. Although all the drivers are vertically aligned, the drivers aren't placed symmetrically on the baffle for smoother response.
Midrange: JBL 12" 2206, probably cardiod operation (theoretically, more seamless merging between dipole and monopole), crossed to tweeter at about 1kHz-1.2kHz.
Treble: Probably a 1"-throat beryllium-diaphragm compression driver + plug-filled OS waveguide designed by Dr. Geddes (phase plug and waveguide matched). The treble response will have to be EQed as the on-axis response drops as frequency increases.
Quick and easy, although that can be improved upon tremendously. A good start for a highly dynamic, relatively flat (in any room) system.
Is there a problem? If you take a look at the response chart, at 1kHz the off-axis response (the line indicated is 45 degrees off-axis, I believe) is down only about 2.5dB. Going up to 1.2kHz, it's down 4dB. It's fine.
EDIT: I would like to change "listening distance 4m or less" to "about 4m", as you can definitely choose drivers that are smaller in size for closer distances like 2m. That's Post #4.
EDIT2: Also in Post #4 is a typo. "cardiod" should be "cardioid".
EDIT3: Another reason why I chose that woofer is because the response is very very flat. Other high-sensitivity woofers with the required power handling (over 100W or so, for a headroom of over 20dB) either have unlisted distortion - which tends to mean relatively high distortion - or have a crappy response.
EDIT4: Forgot the requisite link:
http://www.jblpro.com/pub/components/2206.pdf
EDIT: I would like to change "listening distance 4m or less" to "about 4m", as you can definitely choose drivers that are smaller in size for closer distances like 2m. That's Post #4.
EDIT2: Also in Post #4 is a typo. "cardiod" should be "cardioid".
EDIT3: Another reason why I chose that woofer is because the response is very very flat. Other high-sensitivity woofers with the required power handling (over 100W or so, for a headroom of over 20dB) either have unlisted distortion - which tends to mean relatively high distortion - or have a crappy response.
EDIT4: Forgot the requisite link:
http://www.jblpro.com/pub/components/2206.pdf
David,
Where in the great "Republic of Vancouver Island" are you
You'll probably find as many different opinions as there are people... as much to taste as to differing needs.
Metal drivers -- IMHO these that are worth using are few & far between. 1st most of them ring like crazy... any ringing needs to be down at least 40 dB, 50 in a world class loudspeaker. I don't like to see any XOs in the critical part of the midrange, so XO points outside of 250Hz to 5 kHz, a little more would be better. My favorites are the Joran JX92 and the Fostex FE108E Sigma -- my 1st choice would probably be the latter. Crossing these over this high, open baffle is doable & good... otherwise some sort of aperiodic TL
Your woofers should reach up quite high... 1 kHz isn't unreasonable... you won't find many better than the Llambda 15" Apollos (you'll also have trouble finding them)... they are probably usable up to 2.5-3k. Since these are pretty much unobtainum one has to start looking at what is available, but it sets a goal to strive for.
Tweeters... i'm not a big dome tweeter fan. Ribbons are probably your best bet of what is available... the smaller the better. A dipole would be best, but you'd have to make them yourself (not all that hard actually if you are mechanically adept -- or know someone who is). And if you are going to make your own an ESL tweeter also becomes a consideration,
dave
Where in the great "Republic of Vancouver Island" are you
You'll probably find as many different opinions as there are people... as much to taste as to differing needs.
Metal drivers -- IMHO these that are worth using are few & far between. 1st most of them ring like crazy... any ringing needs to be down at least 40 dB, 50 in a world class loudspeaker. I don't like to see any XOs in the critical part of the midrange, so XO points outside of 250Hz to 5 kHz, a little more would be better. My favorites are the Joran JX92 and the Fostex FE108E Sigma -- my 1st choice would probably be the latter. Crossing these over this high, open baffle is doable & good... otherwise some sort of aperiodic TL
Your woofers should reach up quite high... 1 kHz isn't unreasonable... you won't find many better than the Llambda 15" Apollos (you'll also have trouble finding them
)... they are probably usable up to 2.5-3k. Since these are pretty much unobtainum one has to start looking at what is available, but it sets a goal to strive for.Tweeters... i'm not a big dome tweeter fan. Ribbons are probably your best bet of what is available... the smaller the better. A dipole would be best, but you'd have to make them yourself (not all that hard actually if you are mechanically adept -- or know someone who is). And if you are going to make your own an ESL tweeter also becomes a consideration,
dave
True. There are many ways to achieve low distortion, low power compression, uniform power response, minimal lobing, minimal beaming, etc.
However, many of the drivers you listed have a relatively low maximum SPL. Ideally, one should be able to re-create the levels of live listening.
EDIT: A 15" will beam like crazy at 2.5kHz. At that frequency, most 15" drivers with a flat on-axis response tend to be more than 10dB down at 45 degrees off axis. How do you suggest using it up that high?
However, many of the drivers you listed have a relatively low maximum SPL. Ideally, one should be able to re-create the levels of live listening.
EDIT: A 15" will beam like crazy at 2.5kHz. At that frequency, most 15" drivers with a flat on-axis response tend to be more than 10dB down at 45 degrees off axis. How do you suggest using it up that high?
Not so much as the size. You can engineer the diaphragm so that while breakup the radiating area decreases and effectively radiates like a smaller diaphragm, but not much else.BassAwdyO said:
A rule of thumb is that a rigid piston starts beaming when the frequency being reproduced is shorter than the circumference of the piston.
454Casull said:
I had trouble digging up that much... Nick was unfortunatly no where near as good a business man as he was a driver designer/builder.
Stryke (now Acoustic Elegance -- i guess someone else had rights to the name), biught all the leftover inventory and perhaps more from Nick, so he is John is the one to contact. His website is a bit of a mess rigt now so not much info there.
If you want an Apollo you would need to find used ones... after Nick shipped the last one, he said there would be something like 10k of tooling to do to do any more (ie the 1st par will cost >$12k)
dave
Although I think we're starting to see some real convergence in understanding the 'important principles' in speaker design, there is still room for a fair range of opinions. IMHO you have to define the design criteria first, and then let the rest of the parameters fall where they will. If you follow SL's design of the Orion, this is what he did - he didn't say 'I want an 8" mid', but rather defined the operating parameters and the driver that fell out was an 8.
IMHO the work of Earl Geddes and Tom Danley is blazing the 'state of the art' trail, although SL and his mainstreaming of dipole bass are there too. If I were to put these together, I'd say the important criteria are:
- controlled directivity over as large a frequency range as possible
- (following directly from above) close match between on-axis response and power response
- directivity pattern chosen so as to reduce first-reflection points as far as possible (90x90 for larger rooms, 60x60 for smaller)
- single acoustic source in vocal range (say ~300 up). ie lack of either vertical or horizontal lobing
If you put this together, you get essentially what John Hancock was working on. A Unity concept horn from 300 on up, mated to a dipole midbass system from 300 down to wherever it pooped out, and a monopole sub to pinch-hit down low as needed. There *might* be a slight discontinuity in power response going from the dipole to the Unity, but I think this can be managed with selection of the horn flare and xover freq. I think modifying the 'normal' Unity to bring in some of the Geddes OS waveguide ideas would be a good thing, although I'm not convinced I understand the implications.
My current prototypes kinda satisfy this, although I'm not using dipole midbass as my tiny room requires the speakers to be *in* the corners. My mid/tweet is currently an experimental coaxial horn rather than a Unity, but it's similar in that both sections have 60-degree patterns and are are coaxial; I digitally time-align them to bring the effect pretty close to the Unity, although in the nearfield there is a mismatch in their coverage patterns. Very promising initial results, although much to still do.
Two aspects that I specifically don't think should be *constraints* on the design are
- selection of xover frequency. Unlike Planet 10, I feel the selection of xover points is a moot point *if the system is designed properly*. The Orion, the Unity and Geddes' waveguide speaker all use xovers between ~1k and ~1k4, which according to much 'conventional wisdom' is a no-no.
- distortion of drivers. IMHO there are enough 'good' drivers that this is a moot point. Not to say distortion is irrelevent, simply that sacrificing other aspects of the system simply to be able to use a metal cone driver doesn't make sense to me. Select the lowest distortion driver that satisfies the other system constraints.
IMHO the work of Earl Geddes and Tom Danley is blazing the 'state of the art' trail, although SL and his mainstreaming of dipole bass are there too. If I were to put these together, I'd say the important criteria are:
- controlled directivity over as large a frequency range as possible
- (following directly from above) close match between on-axis response and power response
- directivity pattern chosen so as to reduce first-reflection points as far as possible (90x90 for larger rooms, 60x60 for smaller)
- single acoustic source in vocal range (say ~300 up). ie lack of either vertical or horizontal lobing
If you put this together, you get essentially what John Hancock was working on. A Unity concept horn from 300 on up, mated to a dipole midbass system from 300 down to wherever it pooped out, and a monopole sub to pinch-hit down low as needed. There *might* be a slight discontinuity in power response going from the dipole to the Unity, but I think this can be managed with selection of the horn flare and xover freq. I think modifying the 'normal' Unity to bring in some of the Geddes OS waveguide ideas would be a good thing, although I'm not convinced I understand the implications.
My current prototypes kinda satisfy this, although I'm not using dipole midbass as my tiny room requires the speakers to be *in* the corners. My mid/tweet is currently an experimental coaxial horn rather than a Unity, but it's similar in that both sections have 60-degree patterns and are are coaxial; I digitally time-align them to bring the effect pretty close to the Unity, although in the nearfield there is a mismatch in their coverage patterns. Very promising initial results, although much to still do.
Two aspects that I specifically don't think should be *constraints* on the design are
- selection of xover frequency. Unlike Planet 10, I feel the selection of xover points is a moot point *if the system is designed properly*. The Orion, the Unity and Geddes' waveguide speaker all use xovers between ~1k and ~1k4, which according to much 'conventional wisdom' is a no-no.
- distortion of drivers. IMHO there are enough 'good' drivers that this is a moot point. Not to say distortion is irrelevent, simply that sacrificing other aspects of the system simply to be able to use a metal cone driver doesn't make sense to me. Select the lowest distortion driver that satisfies the other system constraints.
dwk123 said:
With a crossover that's steep enough, you can restrict the lobing region to a very small range. I'd like to try the Lake Contour digital processor - that linear phase 180dB/octave crossover looks delicious in theory. I wonder how it sounds?
http://www.lake.com.au/proaudio/documentation/Lake Contour Linear Phase Crossovers.pdf
http://www.lake.com.au/proaudio/documentation/Linear Phase Brick Wall Crossover.pdf
Forgot to mention that with vertically-mounted time-aligned drivers, horizontal lobing doesn't exist (at least to my understanding) because the difference in path length stays constant regardless of the horizontal angle. Also, vertical lobing shouldn't be much of a problem if everybody is the same height.
454Casull said:
Sure - this is what NHT is doing with the DEQX system as used in their Xd system. Small baffle with tightly spaced drivers, so it exhibits pretty close to point-source behavior once your out of the immediate nearfield. In terms of conventional 'driver in a box' speakers, this can have some clear advantages. As a free side-effect, you get very good stopband rejection to avoid exciting break-up modes of hard-cone drivers. Aside from the lack of constant-directivity behavior (dispersion starts too wide and will still narrow with increasing frequency), there isn't much to complain about in this approach.
Well, I got turned off by the pre-ringing that linear phase filters exhibit. Oh well, can always fall back on old reliable - the 4th-order L-R filter.454Casull said:
With a crossover that's steep enough, you can restrict the lobing region to a very small range. I'd like to try the Lake Contour digital processor - that linear phase 180dB/octave crossover looks delicious in theory. I wonder how it sounds?
http://www.lake.com.au/proaudio/documentation/Lake Contour Linear Phase Crossovers.pdf
http://www.lake.com.au/proaudio/documentation/Linear Phase Brick Wall Crossover.pdf
Forgot to mention that with vertically-mounted time-aligned drivers, horizontal lobing doesn't exist (at least to my understanding) because the difference in path length stays constant regardless of the horizontal angle. Also, vertical lobing shouldn't be much of a problem if everybody is the same height.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.