Hi wesayso,
I'm glad you went for it! You definitely won't regret it, and you definitely made the right choice on the drivers.
What are your plans in terms of EQ? Do you have a DSP based EQ like the Behringer DEQ2496? If not, then I really would suggest that you invest in one. I was inspired by Roger's articles in Speakerbuilder, but i never really did like the analog EQ implementation he used. To his credit, there wasn't really any other options out there at the time, but with the advent of really cheap DSP based EQ, you can truly fine tune the response on a pair of arrays to be perfect over as much of the audio band as you want. I have four different EQ presets I use depending on what I'm listening to, and how I'm listening to it. I have one that is perfectly flat at the listening position for critical listening, one that is intended to sound best when I'm moving around the room, one for movies/TV and one with a leaner bottom end if I really want to listen excessively loud (who doesn't from time to time). Having that kind of flexibility is a huge advantage!
As for comparisons to the arrays 18hurts built, it's going to be hard to do. I don't mean this as a knock against the guys who have built arrays here, but proper EQ is an extremely critical aspect of making these work. My arrays sound like absolute junk without any eq, and I knew full well that they would going into it. I've attached some pictures of the EQ I have implemented for my "Sounds best everywhere" EQ setting which is basically aiming for a flat anechoic response. If you look closely you'll see that they needed almost 10dB of cut through the lower midbass (around 160-800Hz) and about 15dB of boost from about 80Hz down. They're relatively flat through the midband until you get into destructive interference territory above about 8kHz where you need between 6dB and 8dB of boost depending on how you like the top end to sound.
For the arrays 18hurts built, he's probably fine without boost in the top end thanks to the tweeter array, and if he was careful with the crossover between the woofer and tweeter arrays then they might actually be pretty flat from 1k upwards. Anything below that though is probably not too great, and will definitely require good EQ and a measurement mic to get it right.
When properly implemented, I would honestly rate these as a true world-class loudspeaker. They have all the magic of a really good fullrange loudspeaker with none of the drawbacks. They image extremely well, and as I mentioned before, the bottom end integration is like nothing I've ever heard before. It's the best bass I've ever heard. The top end has all the sparkle and air of a good tweeter, and doesn't sound like most single fullrange drivers which always remind me of a 1950's radio, with no bottom or top end to speak of. The dynamics they're capable of is incredible, and if you relax the bottom end EQ you can play them at absurd levels with minimal distortion. I've never owned a pair of loudspeakers that were able to to all of the above as well as the line arrays can. My Ariels (Lynn Olsen design) have a slightly better midrange, but almost no bottom end. The Usher TL's I've used in the past have great bottom end but they can't go as deep and they don't sound as well integrated, and the only thing I've heard that is notably better in the top end is a NeoPro5i with a well implemented active crossover. Basically, you can do better in terms of individual aspects, but as a complete package, nothing even comes close to the line arrays.
As for box construction, I'd really suggest that the large box be broken up into smaller sections to prevent standing waves in the enclosure. You should also never let drivers wired in series share the same sealed volume. If you have the patience, it might actually be beneficial to give each driver its own sealed enclosure using 1/2" or 3/8" dividers between each driver. This would make for an extremely stiff cabinet, and give each driver its own volume to work in.
You also need to be very cautions with mounting drivers that are so small. Use a thinner front baffle and make sure to chamfer the rear of each cutout. Letting the drivers breath freely is very important.
I can share my box plans with you if you'd like. I'll see if I can dig up some of the pictures I took during the construction phase.
Cheers,
Owen
I'm glad you went for it! You definitely won't regret it, and you definitely made the right choice on the drivers.
What are your plans in terms of EQ? Do you have a DSP based EQ like the Behringer DEQ2496? If not, then I really would suggest that you invest in one. I was inspired by Roger's articles in Speakerbuilder, but i never really did like the analog EQ implementation he used. To his credit, there wasn't really any other options out there at the time, but with the advent of really cheap DSP based EQ, you can truly fine tune the response on a pair of arrays to be perfect over as much of the audio band as you want. I have four different EQ presets I use depending on what I'm listening to, and how I'm listening to it. I have one that is perfectly flat at the listening position for critical listening, one that is intended to sound best when I'm moving around the room, one for movies/TV and one with a leaner bottom end if I really want to listen excessively loud (who doesn't from time to time). Having that kind of flexibility is a huge advantage!
As for comparisons to the arrays 18hurts built, it's going to be hard to do. I don't mean this as a knock against the guys who have built arrays here, but proper EQ is an extremely critical aspect of making these work. My arrays sound like absolute junk without any eq, and I knew full well that they would going into it. I've attached some pictures of the EQ I have implemented for my "Sounds best everywhere" EQ setting which is basically aiming for a flat anechoic response. If you look closely you'll see that they needed almost 10dB of cut through the lower midbass (around 160-800Hz) and about 15dB of boost from about 80Hz down. They're relatively flat through the midband until you get into destructive interference territory above about 8kHz where you need between 6dB and 8dB of boost depending on how you like the top end to sound.
For the arrays 18hurts built, he's probably fine without boost in the top end thanks to the tweeter array, and if he was careful with the crossover between the woofer and tweeter arrays then they might actually be pretty flat from 1k upwards. Anything below that though is probably not too great, and will definitely require good EQ and a measurement mic to get it right.
When properly implemented, I would honestly rate these as a true world-class loudspeaker. They have all the magic of a really good fullrange loudspeaker with none of the drawbacks. They image extremely well, and as I mentioned before, the bottom end integration is like nothing I've ever heard before. It's the best bass I've ever heard. The top end has all the sparkle and air of a good tweeter, and doesn't sound like most single fullrange drivers which always remind me of a 1950's radio, with no bottom or top end to speak of. The dynamics they're capable of is incredible, and if you relax the bottom end EQ you can play them at absurd levels with minimal distortion. I've never owned a pair of loudspeakers that were able to to all of the above as well as the line arrays can. My Ariels (Lynn Olsen design) have a slightly better midrange, but almost no bottom end. The Usher TL's I've used in the past have great bottom end but they can't go as deep and they don't sound as well integrated, and the only thing I've heard that is notably better in the top end is a NeoPro5i with a well implemented active crossover. Basically, you can do better in terms of individual aspects, but as a complete package, nothing even comes close to the line arrays.
As for box construction, I'd really suggest that the large box be broken up into smaller sections to prevent standing waves in the enclosure. You should also never let drivers wired in series share the same sealed volume. If you have the patience, it might actually be beneficial to give each driver its own sealed enclosure using 1/2" or 3/8" dividers between each driver. This would make for an extremely stiff cabinet, and give each driver its own volume to work in.
You also need to be very cautions with mounting drivers that are so small. Use a thinner front baffle and make sure to chamfer the rear of each cutout. Letting the drivers breath freely is very important.
I can share my box plans with you if you'd like. I'll see if I can dig up some of the pictures I took during the construction phase.
Cheers,
Owen
Attachments
Thanks Owen,
After seeing the IDS-25 on the net I figured I could build something like that and get away with it towards the girlfriend.
I have been running some old big speakers with 15" woofers that take up quite a bit of space. My girlfriend kept nagging me about the floorspace they use up and beeing this big they where never allowed proper placement to do them justice. The arrays could be placed close to the wall and would take up less space. I immediatly thought of the Behringer DEQ 2496 beeing the ticket to making this work. In my research on line arrays I stumbled upon this thread. It was entertaining and re-assuring to find your replies on this thread with the same ideas right down to the DEQ and the drivers.
My cabinet plans looked like this but are open to change as I have not cut anything yet:
The plan was to buit this from 18mm and 12mm MDF in a (very) big stack.
The 12mm would be used for bracing and the 18mm for the actual chambers.
Stacking gives me the opertunity to make a round shape on the outside, helping with the WAF (GAF in my case
) and limiting difraction while the inner cavity could be made up with a wavy shape to limit standing waves.
The four holes around the circumfence are used for dowels while the bottom stack used threaded rods to connect to the footplate.
The inner parts would get a round over but because I want to water cut these parts I haven't drawn that.
As a backup plan if water cutting is too expencive I have a friend who is willing to CNC these for me.
The outer shape would get a layer of fibreglass bonded to it to strengthen and smooth it out. The plan is to mount the drivers with the frame touching and fill in the sides to the outer box to have them close to flush.
So what do you think, crazy huh?
After seeing the IDS-25 on the net I figured I could build something like that and get away with it towards the girlfriend.
I have been running some old big speakers with 15" woofers that take up quite a bit of space. My girlfriend kept nagging me about the floorspace they use up and beeing this big they where never allowed proper placement to do them justice. The arrays could be placed close to the wall and would take up less space. I immediatly thought of the Behringer DEQ 2496 beeing the ticket to making this work. In my research on line arrays I stumbled upon this thread. It was entertaining and re-assuring to find your replies on this thread with the same ideas right down to the DEQ and the drivers.
My cabinet plans looked like this but are open to change as I have not cut anything yet:
The plan was to buit this from 18mm and 12mm MDF in a (very) big stack.
The 12mm would be used for bracing and the 18mm for the actual chambers.
Stacking gives me the opertunity to make a round shape on the outside, helping with the WAF (GAF in my case
) and limiting difraction while the inner cavity could be made up with a wavy shape to limit standing waves.
The four holes around the circumfence are used for dowels while the bottom stack used threaded rods to connect to the footplate.
The inner parts would get a round over but because I want to water cut these parts I haven't drawn that.
As a backup plan if water cutting is too expencive I have a friend who is willing to CNC these for me.
The outer shape would get a layer of fibreglass bonded to it to strengthen and smooth it out. The plan is to mount the drivers with the frame touching and fill in the sides to the outer box to have them close to flush.
So what do you think, crazy huh?
Last edited:
Thanks for the information, Owen
You are 100% correct, those things NEED EQ to get the proper sound. I went with the Apex Jr. tweeters since they having a 3dB rising response from 4 to 8 KHz which gets the response fairly flat past 10 KHz. The 2-way design was chosen to limit the amount of EQ required over full range designs. Where the upper mids mesh with the tweeter line is an EQ issue and I'm going to construct a tapped horn for 16 to 60Hz bass response (I like pipe organs)
Since I will never again build a two-way array, I find the full range versions very interesting. Tangband almost builds something that might be better for full range arrays with a low 60Hz Fs, 86dB 1w/1m, 5mm Xmax and a rising response at 8 KHz that jumps 4 to 8dB from 8 to 15 KHz. The TB W4-1720 returns to "normal" at 20KHz so might work well with arrays?
http://www.parts-express.com/pdf/264-872s.pdf
All they need to do is change the frame so the 4" frame can be trimmed from 125mm (4.95") to around 110mm (4.33") and I'll start saving my money for 32 of them running 16 per side.
You are 100% correct, those things NEED EQ to get the proper sound. I went with the Apex Jr. tweeters since they having a 3dB rising response from 4 to 8 KHz which gets the response fairly flat past 10 KHz. The 2-way design was chosen to limit the amount of EQ required over full range designs. Where the upper mids mesh with the tweeter line is an EQ issue and I'm going to construct a tapped horn for 16 to 60Hz bass response (I like pipe organs)
Since I will never again build a two-way array, I find the full range versions very interesting. Tangband almost builds something that might be better for full range arrays with a low 60Hz Fs, 86dB 1w/1m, 5mm Xmax and a rising response at 8 KHz that jumps 4 to 8dB from 8 to 15 KHz. The TB W4-1720 returns to "normal" at 20KHz so might work well with arrays?
http://www.parts-express.com/pdf/264-872s.pdf
All they need to do is change the frame so the 4" frame can be trimmed from 125mm (4.95") to around 110mm (4.33") and I'll start saving my money for 32 of them running 16 per side.
wesayso:
That looks like quite the undertaking, but it's a good plan! You'll have your work cut out for you getting all those pieces stacked and glued.
I did something like that, but with a meager 16 layers. I'll never do it again since it was an obscene amount of work, but the result was better than I had expected. Just glue and flush trim each layer, one at a time!
If you do decide to go that way, you should really consider a thin layer of wood between each driver since it would be really easy to implement. That would allow you to do away with the braces entirely, and would have a minimal impact on internal volume. Giving each drive its own volume would definitely be beneficial.
If you're good with fiberglass and auto body filler, I think the finished result could be very impressive looking.
18Hurts:
Those drivers do look very impressive, and I've used the larger 6.5" version (W6-1721) and I can attest to the build quality being superb. They are, however, painfully overpriced at $50 a driver. To build an array like mine would be a whopping $2500!
With the larger Sd and higher Xmax though, they would have some pretty incredible bottom end. You could get away with much less EQ.
As for the FR graph published by TB, I would really take that with a grain of salt. They tend to be a little on the optimistic side, and the actual drivers seldom measure as well as those graphs imply. Zaph has some measurements for it on his site that are probably more representative.
Cheers,
Owen
That looks like quite the undertaking, but it's a good plan! You'll have your work cut out for you getting all those pieces stacked and glued.
I did something like that, but with a meager 16 layers. I'll never do it again since it was an obscene amount of work, but the result was better than I had expected. Just glue and flush trim each layer, one at a time!
If you do decide to go that way, you should really consider a thin layer of wood between each driver since it would be really easy to implement. That would allow you to do away with the braces entirely, and would have a minimal impact on internal volume. Giving each drive its own volume would definitely be beneficial.
If you're good with fiberglass and auto body filler, I think the finished result could be very impressive looking.
18Hurts:
Those drivers do look very impressive, and I've used the larger 6.5" version (W6-1721) and I can attest to the build quality being superb. They are, however, painfully overpriced at $50 a driver. To build an array like mine would be a whopping $2500!
With the larger Sd and higher Xmax though, they would have some pretty incredible bottom end. You could get away with much less EQ.
As for the FR graph published by TB, I would really take that with a grain of salt. They tend to be a little on the optimistic side, and the actual drivers seldom measure as well as those graphs imply. Zaph has some measurements for it on his site that are probably more representative.
Cheers,
Owen
Attachments
Last edited:
Thank you very much for the heads up, Owen
At this point, I'm just happy that my hands quit vibrating from all the machining required for 96 tweeters!
My wife is thrilled with the arrays, it gets me out of the house so she can listen to her music--and I can listen to mine. At this point, it is tweak time as I attempt to get them to sound correct for classical music and they are getting close. The feedback destroyer might work well for me as it has 12 parametric bands per channel... more stuff to mess with! Since a "digital EQ" will eventually be thrown in the mix, I'll ponder that solution after I go with a proper tapped horn subwoofer and mixer (guitar connection).
I have been asked at least a dozen times to "build a pair of those for me!" since the populace at large either likes the sound or the "look" of the things. For house use, I'll leave it up to those that have built and operated arrays in their house for wisdom. A nice solution for acoustic hell in a metal walled garage with cement floors and plenty of objects to reflect from but I'll leave it to you for proper music/HT design philosophy. My wife absolutely refuses to allow those monsters back in the house.
At this point, I can highly recommend arrays for garage/party speaker use since they sound better than point source in a highly reflective environment. For those pondering building them, I give two thumbs up for full range construction since 2-way fixes EQ problems but also creates a few issues at the same time. Basically, you'll need EQ anyway so go with full range and get the proper EQ and run.
Passive solutions need not apply for DIY--JBL can hook you up if passive is required.
So tow thin could that layer be without possibly going to resonate/vibrate? A 12 mm layer would take up most space there is between the drivers if they are mounted flange to flange.
Mr. Russell used one large space only devided by about 12 braces with two big (1-14" or 63mm) holes. I sort of copied that concept but made sure the other brace shape blocked the holes when looked at vertically. They have the opening near the baffle to be able to wire the speakers there.
So in short, I know what you're saying but it is kinda hard to realise.
You'd need to use a 6-8mm sheet between each driver if you were going to go that way. That's about as thick as you can make it while still mounting the drivers basket to basket.
That would be more than enough thickness to prevent any sort of resonance. Keep in mind that they have strength in numbers (there would be 24 internal dividers and 2 thicker external tops and bottoms), and there is technically no actual net pressure drop between the internal dividers (since the adjacent drivers will all be in phase) so they will not be prone to resonate. For added effect, you could always add a 2mm layer of bitumen sheet to both sides of each divider once you're an inch or two back from the driver.
Wooden airplane wings are built like that, with multiple thin wooden braces which provide incredible strength simply because there were so many of them. This would be the exact same principle.
Cheers,
Owen
That would be more than enough thickness to prevent any sort of resonance. Keep in mind that they have strength in numbers (there would be 24 internal dividers and 2 thicker external tops and bottoms), and there is technically no actual net pressure drop between the internal dividers (since the adjacent drivers will all be in phase) so they will not be prone to resonate. For added effect, you could always add a 2mm layer of bitumen sheet to both sides of each divider once you're an inch or two back from the driver.
Wooden airplane wings are built like that, with multiple thin wooden braces which provide incredible strength simply because there were so many of them. This would be the exact same principle.
Cheers,
Owen
Yep. Nice speakers, though I've grown to dislike the Infinity hissy treble sincy my first Primus system. Anyway, those drivers seem really far apart compared to some people people who are overlapping driver frames to minimize c-c.
But I guess I misremembered the tweeter line. The ribbon line looks like a good idea, but those dome mids look far apart. I was figuring that 2" mids (yes, I'm looking into 3-way designs using the Aura 2x3" drivers) lined up edge to edge comb at around 6500 Hz. Those tweeters probably go down to 2500,and those domes probably reach to 6kHz, but that doesn't explain 4-6" C-C. I'd love to know the design theory behind the QLS
hmmm... I'm more than a little perplexed by that design. I don't get the c-c spacing on the "dome midranges" which apparently go all the way down to 600Hz, but more importantly, I don't get the single tiny midrange which apparently covers from 200Hz to 600Hz. That's such a tiny range to implement using a passive crossover, and I would think that having the one small driver would significantly decrease the maximum output without really adding much benefit. Seems like a bit of an afterthought to me, but then again I fundamentally don't understand the design.
Cheers,
Owen
Cheers,
Owen
Yeah, I'm having troubles implimenting a 1000-5500 Hz crossover on the midranges without strange peaks, even 4th order. The mix of line and point sources is probably fine for home use. A midrange at that size and frequency doesn't seem like it could be efficient, although the size,may be required for directionality since those midranges are about on the floor. Overall, I wasn't impressed by those speakers when I heard them, but I did just hear the Martin Logan CLS's and Beverage series 2's just behind me and with much better demo program... definitely good, but I can't get over that hissy treble.
OPC: For some reason, I thought you were using the Aura NS3's instead of the Vifas. On paper, the Auras should be better (except for the frame), especially for the low end. The FR is FLAT through to the tweeter range (yeah, no tweeters). Krutke didn't seem to like them, but I'd imagine that it has something to do with being twice as much as the HiVi 3"ers. X-max is huge for that size. Distortion figures aren't as good as others, but I'd imagine that this isn't as crucial for LA use at home. Were they $9 when you were looking? Is there a reason why you stayed away from them?
I've never listened to the Infinity arrays exactly,
Liked the Kappa 9's and IRS Betas from 20+ years back--very dynamic and clean sounding at the rather high levels 20 somethings enjoy. The Kappas belonged to a friend and I heard the Betas in a sound room--liked them although the price was as breathtaking as the speakers.
Love the mid-bass response (drums are effortless) and the 2-way allows amazing highs a 6 foot stack of 48 dome tweeters can give. Coupled with a very high output subwoofer and some EQ--I think I can live with the 5" woofers for now. I did it on a lark and am very happy I did, very educational and it shows the direction I need to go for eventual improvement.
The Aurasound Whisper 2" inverted dome full range I'd like to hear. They quote the frequency response at 250 to 15KHz, 3mm Xmax and 15 watts of power handling. EQ the highs in and run bass bins at 250Hz and down?
Infinity seemed to get away with it so maybe that would work?
I'll just keep watching as 2 to 4" drivers develop with high Xmax for their size, decent efficiency, smooth response and decent power handling. I'll revisit actually physically changing drivers in the coming years, for now it's back to tapped horns and tweaking to improve the sound. My wife won't let the monster arrays or giant tapped horns in the house so now when I think of arrays, I think of
Liked the Kappa 9's and IRS Betas from 20+ years back--very dynamic and clean sounding at the rather high levels 20 somethings enjoy. The Kappas belonged to a friend and I heard the Betas in a sound room--liked them although the price was as breathtaking as the speakers.
Love the mid-bass response (drums are effortless) and the 2-way allows amazing highs a 6 foot stack of 48 dome tweeters can give. Coupled with a very high output subwoofer and some EQ--I think I can live with the 5" woofers for now. I did it on a lark and am very happy I did, very educational and it shows the direction I need to go for eventual improvement.
The Aurasound Whisper 2" inverted dome full range I'd like to hear. They quote the frequency response at 250 to 15KHz, 3mm Xmax and 15 watts of power handling. EQ the highs in and run bass bins at 250Hz and down?
Infinity seemed to get away with it so maybe that would work? I'll just keep watching as 2 to 4" drivers develop with high Xmax for their size, decent efficiency, smooth response and decent power handling. I'll revisit actually physically changing drivers in the coming years, for now it's back to tapped horns and tweaking to improve the sound. My wife won't let the monster arrays or giant tapped horns in the house so now when I think of arrays, I think of
gmcalabria:
The Aura drivers look suspiciously like the Dayton drivers I tried, but the specs seems a little different. A few things stand out just looking at the datasheet, but since I've never actually measured or tried these, take this with a grain of salt...
Pros:
- Massive x-max would allow for lots of bottom end
- Lower Fs would require a little less EQ
- Underhung motor - which is cool
Cons:
- Extremely inefficient
- Aluminum cone break-up starting at 8k
- Very high Le (look at impedance curve above 1k) points to mediocre motor design
- stamped steel basket
Overall I feel like you're paying a high price for all that excursion. The sharply rising impedance seems to be at least a partial cause of the diminishing output above 8k, which is actually working against you if you want to run these full range. What you really want is a driver with a gradually increasing output at higher frequencies. Having such a low efficiency will also mean that you'll need a huge amount of power to drive these. I'd say upwards of 1000W for a 25 driver array.
I was so put off by the Dayton aluminum cone drivers that I think I just wrote these off because they looked so similar. Probably not fair overall, but I am weary of anything with an aluminum cone when you plan to run full range. The exception being the Fountek driver which was stunning.
Just my 2 cents...
Cheers,
Owen
The Aura drivers look suspiciously like the Dayton drivers I tried, but the specs seems a little different. A few things stand out just looking at the datasheet, but since I've never actually measured or tried these, take this with a grain of salt...
Pros:
- Massive x-max would allow for lots of bottom end
- Lower Fs would require a little less EQ
- Underhung motor - which is cool
Cons:
- Extremely inefficient
- Aluminum cone break-up starting at 8k
- Very high Le (look at impedance curve above 1k) points to mediocre motor design
- stamped steel basket
Overall I feel like you're paying a high price for all that excursion. The sharply rising impedance seems to be at least a partial cause of the diminishing output above 8k, which is actually working against you if you want to run these full range. What you really want is a driver with a gradually increasing output at higher frequencies. Having such a low efficiency will also mean that you'll need a huge amount of power to drive these. I'd say upwards of 1000W for a 25 driver array.
I was so put off by the Dayton aluminum cone drivers that I think I just wrote these off because they looked so similar. Probably not fair overall, but I am weary of anything with an aluminum cone when you plan to run full range. The exception being the Fountek driver which was stunning.
Just my 2 cents...
Cheers,
Owen
I keep forgetting you were looking for a full-range speaker... They seem like a better choice for 2-way until you weigh LF extension with efficiency and come out about the same either way. Better off going with the proven Vifa.
Dayton calls their 3.5" driver a direct replacement to this Aurasound, but the Aurasound claims that theirs is a 3". I wonder what they're measuring as the nominal is still a rough measurement.
I finally took a look at Jim Griffin's white papers and a better look at Zobsky's build. I've come up with a few fundamental questions.
Zobsky's big complaints seemed to be a lacking upper-midrange and top end, both of which he attributed to comb filtering. If it's really just the drivers, then that's one thing that's not easy to work around on a tight budget, but what if it's really the comb filtering that's causing this. The midrange can be helped by using smaller woofers (3" isn't bad going to tweeters, 2" would be better), but smaller woofers with good excursion/bass response/efficiency are expensive. Alternatively, the array can be built 3-way with 2" extended range drivers for mids, and with an active crossover that doesn't have a bum 6th channel (one that I didn't pick up on eBay for cheap...), this is no problem whatsoever. These mid drivers can be had cheaper than better quality 3" drivers, and allow for larger/more efficient/lower ranged woofers, which are better for PA use.
But what about the top end? No dome tweeter is small enough to reach 15-20kHz without combing. Ribbons and line array-style tweeter waveguide lenses are way too expensive for a budget build. If that's the case, than Zobsky's super tweeter setup becomes necessary. If the problem with mixing line and point sources is relative drop-off slopes over distance, then there's nothing wrong with a properly designed single tweeter/super tweeter system so long as it's tuned to a single listening distance (fine for HT use, and with active crossovers or an L-pad, adjusting this distance is easy). If that's the case, I'm returning back to my single waveguide compression tweeter design since the compression driver will have the dynamics of a row of tweeters without all of the sawdust.
Now, one might try to build these arrays for dual use PA/HT drivers, but if they are, then it seems like a can of worms trying to make the nearfield/farfield cutoff point the same for all three rows of drivers (or just 2), especially considering differing reflection and power tapering issues. The line of tweeters would definitely be better than a point source for this, but again, is the dual use part really reasonable? In-room reflections make the single tweeter not have a true 6 dB/double distance drop off anyway.
So if the point source tweeter/line source woofer works, why not do the same thing for the midrange? Seems like a D'Appolito array for the middle is a good idea. Wait, a sec, we're back at the design that SpeakerDave has as his avatar....
I see good reason to keep the mid array instead of MTM or horn, but not nearly so much for the tweeter, not on a cheap build. The single tweeter seemed to work well on Lou's Thin Blue line array project, maybe a mildly loaded dome tweeter is a reasonable alternative to a compression driver.
If I had tons of money, I'd go with high sensitivity ribbons or a line array-style horn. Actually, if I had oodles of money, I'd be REALLY interested to see how an open baffle woofer and a huge stack of ESS Airmotion tweeters would do. I have no idea what the implications of a dipolar line array tweeter would be, but it sounds line a great idea...
But then again I don't have tons of money. But back up a few here; if ther 10+ kHz trouble is the driver, then EQ won't help, but if it's part of the combing, then why not just compensate with EQ? I am using a digital x-over. If so, what would the loss profile be like? Of course, there might be some issues with the 10+kHz due to the dipolar nature of the woofers and the monopole tweeters not bouncing off of the walls when beaming. Any thoughts there?
18Hurts: you've built a similar system to Zobsky but didn't mention the lack of air/sparkle. Is it an issue? Is this something that's hidden by the acoustical nightmare garage?
Greg
Zobsky's big complaints seemed to be a lacking upper-midrange and top end, both of which he attributed to comb filtering. If it's really just the drivers, then that's one thing that's not easy to work around on a tight budget, but what if it's really the comb filtering that's causing this. The midrange can be helped by using smaller woofers (3" isn't bad going to tweeters, 2" would be better), but smaller woofers with good excursion/bass response/efficiency are expensive. Alternatively, the array can be built 3-way with 2" extended range drivers for mids, and with an active crossover that doesn't have a bum 6th channel (one that I didn't pick up on eBay for cheap...), this is no problem whatsoever. These mid drivers can be had cheaper than better quality 3" drivers, and allow for larger/more efficient/lower ranged woofers, which are better for PA use.
But what about the top end? No dome tweeter is small enough to reach 15-20kHz without combing. Ribbons and line array-style tweeter waveguide lenses are way too expensive for a budget build. If that's the case, than Zobsky's super tweeter setup becomes necessary. If the problem with mixing line and point sources is relative drop-off slopes over distance, then there's nothing wrong with a properly designed single tweeter/super tweeter system so long as it's tuned to a single listening distance (fine for HT use, and with active crossovers or an L-pad, adjusting this distance is easy). If that's the case, I'm returning back to my single waveguide compression tweeter design since the compression driver will have the dynamics of a row of tweeters without all of the sawdust.
Now, one might try to build these arrays for dual use PA/HT drivers, but if they are, then it seems like a can of worms trying to make the nearfield/farfield cutoff point the same for all three rows of drivers (or just 2), especially considering differing reflection and power tapering issues. The line of tweeters would definitely be better than a point source for this, but again, is the dual use part really reasonable? In-room reflections make the single tweeter not have a true 6 dB/double distance drop off anyway.
So if the point source tweeter/line source woofer works, why not do the same thing for the midrange? Seems like a D'Appolito array for the middle is a good idea. Wait, a sec, we're back at the design that SpeakerDave has as his avatar....
I see good reason to keep the mid array instead of MTM or horn, but not nearly so much for the tweeter, not on a cheap build. The single tweeter seemed to work well on Lou's Thin Blue line array project, maybe a mildly loaded dome tweeter is a reasonable alternative to a compression driver.
If I had tons of money, I'd go with high sensitivity ribbons or a line array-style horn. Actually, if I had oodles of money, I'd be REALLY interested to see how an open baffle woofer and a huge stack of ESS Airmotion tweeters would do. I have no idea what the implications of a dipolar line array tweeter would be, but it sounds line a great idea...
But then again I don't have tons of money. But back up a few here; if ther 10+ kHz trouble is the driver, then EQ won't help, but if it's part of the combing, then why not just compensate with EQ? I am using a digital x-over. If so, what would the loss profile be like? Of course, there might be some issues with the 10+kHz due to the dipolar nature of the woofers and the monopole tweeters not bouncing off of the walls when beaming. Any thoughts there?
18Hurts: you've built a similar system to Zobsky but didn't mention the lack of air/sparkle. Is it an issue? Is this something that's hidden by the acoustical nightmare garage?
Greg
You are very perceptive Greg--good questions
My row of tweeters are 10mm domes with a C to C spacing of around 36mm between them. This will lead to beaming over 10KHz and a reduction in output in the top octave. The Audax tweets have a natural rise of 3dB from 90dB to 93dB peaking at 8 KHz then tapering back down to 90dB at 10+KHz. This helps with the downward "tilt" that is natural with arrayed tweeters in the 5 to 10KHz band.
Since arrays will cut the output as they go up, EQ would work to "get it back". As you say, the ultimate solution would be 2 to 3" mid-woofers as a midrange for a three-way array. Before I started the project, I figured I'd go with the 5" Sony woofers for the neo magnets, 3mm Xmax and poly coated cones/included grills and rubber surrounds for durability in a garage. They put out really good mid-bass but the upper mids fall apart as they cross to the tweeter line. I knew it was not perfect but at $60 for two dozen of them, figured I was not out a lot of cash if I hated line arrays. The sound is distinctive so my process continues to make them better over time. With a tapped horn for the bottom, it would actually be 4 way so I'm about as far away from full range arrays as I can get!
The woofers are 6" (152mm) the tweeters are 2.3" (59mm) wide and my removable bezel is 11.25" (286mm) wide. To prevent serious brain damage with such tight tolerances, the 2" mid would be the solution. The 48 tweeter line would be cut out as a strip so I won't have to do that again--I won't do that again! The Vifa 2" at 55mm would fit 32 of them in the bezel and their output would match the twelve 5" woofers. The 1mm of Xmax and Fs of 178Hz dictate at least a 400Hz crossover frequency but I'd feel better at 500Hz for protection.
Although my ears are pushing for a 3-way array, my hands don't want to cut 24 woofer and 64 2" mid holes in plywood right around now and my wallet refuses to hand that kind of cash out with Christmas next month.
The Tangband 2" looks really good but blowing $1,400 on 64 of them, add croosover parts and wood for $1,500 upgrade in a garage. It is not critical at this point--eventually something good will show up since 400 to 4.3KHz is not too hard to reproduce for a 2" speaker.
Anyone want to send me around 64 Tangband 2" mids? I'll brave the cold and post pictures to thank you.
I'm still not convinced center to center spacing is the only factor involved. If you look at the length of an array the outer ones are going to be much further away than the ones who are at or around your ear level. So maybe it would be enough to make the tweeter array shorter to prevent combing? See the distances from my IDS example:
Closer center to center spacing would add more sources inbetween but the time difference (difference in length) between the top tweeter and center tweeter would still exist. So maybe you could get away by just using a shorter tweeter array to keep the difference to the listeners ear minimal. I would guess it would create a difference in sound between sitting down and standing up though.
Closer center to center spacing would add more sources inbetween but the time difference (difference in length) between the top tweeter and center tweeter would still exist. So maybe you could get away by just using a shorter tweeter array to keep the difference to the listeners ear minimal. I would guess it would create a difference in sound between sitting down and standing up though.
Last edited:
gmcalabria:
Take a look at the EQ profiles I posted a few posts back. I have a PEQ at 20kHz with a gain of 8dB and BW of 2. This is specifically to counteract the HF rolloff caused by comb filtering. With that EQ in place, the array both measures and sounds perfectly balanced in the top end.
Again, it's a little more complicated than just taking the C-C spacing and saying "It's going to comb filter above this". In actual fact, you start getting lobes in the far field at the frequency with a wavelength that is equal to the C-C spacing, but the first actual comb "cancellation" doesn't happen until twice that frequency. So in the case of the Vifa drivers, the C-C spacing is 83.75mm which is a frequency of 4100Hz. The first actual cancellation doesn't happen until 8200Hz. That's where the output is going to start to diminish, and where you'll need EQ.
The problem is further compounded by the fact that the driver dispersion is no longer spherical at these high frequencies, and to get comb filtering, you need spherical dispersion. The little drivers start to be become more and more directional at higher frequencies, and this counteracts the effects of combing, much like an array of ribbons would. The reason an array of ribbons doesn't comb filter is because the dispersion patterns of the ribbons don't overlap, meaning there is no cancellation and no combing. All of this is covered in Jim's paper.
If you want a really great and still affordable solution, take a look at an array built with these:
Bohlender Graebener Neo8-PDR Planar Transducer 264-713
12 of those in an 8 ft line would:
- Have excellent horizontal dispersion
- Have absurd efficiency
- Eliminate all comb filtering and minimize EQ requirements
- Go down to 350Hz
- Cost $1200 if you shop around
- Could be run dipole or in a box.
All you would need is a large sub to fill in the bottom end, or use an array of 6-7" drivers for frequencies below 350 Hz.
wesayso:
The issue you mentioned genuinely doesn't exist thanks to temporal masking:
Temporal masking - Wikipedia, the free encyclopedia
The difference in arrival times between the top driver and the centre driver in your diagram would be 0.56us which is so short that your brain would never be able to distinguish the two arrivals as separate or distinct. The generally accepted minimum value for when your brain can tell the difference is 20ms, so it's not even in the ballpark.
Some people have built curved arrays to account for this, but all they end up doing is making comb filtering worse, and then the array can only be focused on a single point in space, which is pretty restrictive. It's also very difficult to build a perfect curved baffle.
Regards,
Owen
Take a look at the EQ profiles I posted a few posts back. I have a PEQ at 20kHz with a gain of 8dB and BW of 2. This is specifically to counteract the HF rolloff caused by comb filtering. With that EQ in place, the array both measures and sounds perfectly balanced in the top end.
Again, it's a little more complicated than just taking the C-C spacing and saying "It's going to comb filter above this". In actual fact, you start getting lobes in the far field at the frequency with a wavelength that is equal to the C-C spacing, but the first actual comb "cancellation" doesn't happen until twice that frequency. So in the case of the Vifa drivers, the C-C spacing is 83.75mm which is a frequency of 4100Hz. The first actual cancellation doesn't happen until 8200Hz. That's where the output is going to start to diminish, and where you'll need EQ.
The problem is further compounded by the fact that the driver dispersion is no longer spherical at these high frequencies, and to get comb filtering, you need spherical dispersion. The little drivers start to be become more and more directional at higher frequencies, and this counteracts the effects of combing, much like an array of ribbons would. The reason an array of ribbons doesn't comb filter is because the dispersion patterns of the ribbons don't overlap, meaning there is no cancellation and no combing. All of this is covered in Jim's paper.
If you want a really great and still affordable solution, take a look at an array built with these:
Bohlender Graebener Neo8-PDR Planar Transducer 264-713
12 of those in an 8 ft line would:
- Have excellent horizontal dispersion
- Have absurd efficiency
- Eliminate all comb filtering and minimize EQ requirements
- Go down to 350Hz
- Cost $1200 if you shop around
- Could be run dipole or in a box.
All you would need is a large sub to fill in the bottom end, or use an array of 6-7" drivers for frequencies below 350 Hz.
wesayso:
The issue you mentioned genuinely doesn't exist thanks to temporal masking:
Temporal masking - Wikipedia, the free encyclopedia
The difference in arrival times between the top driver and the centre driver in your diagram would be 0.56us which is so short that your brain would never be able to distinguish the two arrivals as separate or distinct. The generally accepted minimum value for when your brain can tell the difference is 20ms, so it's not even in the ballpark.
Some people have built curved arrays to account for this, but all they end up doing is making comb filtering worse, and then the array can only be focused on a single point in space, which is pretty restrictive. It's also very difficult to build a perfect curved baffle.
Regards,
Owen