There is a little bit of comb filtering, the easiest way to see it is in the impulse response where, if you look carefully, you are seeing the impulse hit 3-4 times very close together. There would be a benefit to tweeters spaced much closer together and that will enhance the illusion that the sound is coming from 3-4 feet behind the speaker.
Fortunately the CBT shading mitigates a lot of the comb filtering which would be very problematic in a standard line array with 3.5" drivers.
Still classed as a 3” like FF85wk and Alpair 5.3.
The closely related CHN-50 is another inexpensive driver that should make a good array. I wish they cam ein a rectangular baffle.
I have 4 (good) of the higher end version (FE87-something?). Along with at leasts 2 with dings. packed and ready to ship.
dave
I always look to Sd...especially on small drivers. The FE85 is showing 27 cm2 on Madisound....which is a bit less than most 3" like the faitals.
Or the good ole TC9 with 36 cm2.
Perry, clever you are ! ...using all the drivers down low along with a reflex port. Nice project.
I built a CBT with twenty-four TC9's a few years back. I might want to revisit it, and copy your frequency strategy.
Drivers share a common enclosure, including the base volume. Gotta figure out how to get a port in ..
Or the good ole TC9 with 36 cm2.
Perry, clever you are ! ...using all the drivers down low along with a reflex port. Nice project.
I built a CBT with twenty-four TC9's a few years back. I might want to revisit it, and copy your frequency strategy.
Drivers share a common enclosure, including the base volume. Gotta figure out how to get a port in ..
@mark100 thanks for the photo, and for your contribution to the conversation.
See an earlier post in this thread here: https://www.diyaudio.com/community/...th-35hz-25khz-high-output.412154/post-7670309 regarding the complications of using CBT at low frequencies.
That array in your picture above could generate some serious output with a port / passive radiator and the right modified shading strategy. I estimate you could tune it to 35Hz (with a hefty DSP peak centered right at that frequency, and steep rolloff below) and get useful output down to 30. A high Q shelf filter is ideal for that.
See an earlier post in this thread here: https://www.diyaudio.com/community/...th-35hz-25khz-high-output.412154/post-7670309 regarding the complications of using CBT at low frequencies.
That array in your picture above could generate some serious output with a port / passive radiator and the right modified shading strategy. I estimate you could tune it to 35Hz (with a hefty DSP peak centered right at that frequency, and steep rolloff below) and get useful output down to 30. A high Q shelf filter is ideal for that.
Last edited:
This illustrates delivering 100% signal to all drivers at low frequencies, vs shaded array at mid and high frequencies. The curved array simulates a virtual sound source behind the speaker, and a well executed CBT / Shaded Array seems as though the sounds are coming from that apparent location:
Last edited:
All the filters overlap from 100 to 1000
So it behaves as expected.
Really hot midrange and basically comb the rest of the way.
It is a significant challenge indeed for array crossovers.
But yes series trick makes the cost relatively lower
than having to use coils instead
Not at all, you should picture the source like it’s a constant directivity speaker, which means that a seated or standing listener hears the same SPL whether they are 2 feet away or 8 feet away from the speaker. It’s a really cool effect and it works in a room just like in an auditorium (except it’s hung from the ceiling in an auditorium so the whole thing is upside down). Every listener in the room gets a good stereo image if the speakers are toed in. A conventional speaker cannot do this, only a dipole, a large Constant Directivity design or CBT array does this.
It effectively takes the floor out of the equation and eliminates floor bounce.
Last edited:
If you audition a well designed system you’ll find it’s great for people too 🌞
I hope this is the right thread to share my attempts at this method on VituixCAD 2. The good, the bad, and the confusing.
First of all, the idea of a delay-perfect ported bass system is super interesting. I have done similar with my nearfield monitors with PEQ's to limit their excursion and maximize their output, but I would have never guessed "High Q Shelf". Neat. For my purposes, I have found varying both the Q and the shelf to be important in my experiments. The box and driver alignment seem to affect one another, and driving the Q as low as 2 had some positive effects on the timing, excursion, and port velocity. The part I can't figure is sub-bass excursion below Fs. No highpass filter will both keep delay perfect AND prevent overexcursion. It's one or the other, and overexcursion occurs at incredibly low voltages in this design, according to the sims. Which, for the record, is why I haven't tried it on any of my own speakers. VCAD is free, woofers are not.
Here's a picture that shows a simulation of a 5" woofer reaching ~100dB to 40hz in a moderate enclosure. 30hz is down ~10dB while 60hz has no delay whatsoever. 50hz is still 15ms, like most of my ported designs, giving a bit of the "Sloppy Drummer" kick, though it mostly makes them sound like they're trying to kick very hard in my experience.
The ones I built per pict above, had an apparent image center at about 1/3 of their height.
What was remarkable, was how the apparent image center didn't move up or down much, if any, seated vs standing.
Or even walking up close, almost looking down on them.
That's In contrast with a straight-line array using the same baffle with the same 24 drivers,
where the apparent image center totally moved up down with changes in ear height.
What was remarkable, was how the apparent image center didn't move up or down much, if any, seated vs standing.
Or even walking up close, almost looking down on them.
That's In contrast with a straight-line array using the same baffle with the same 24 drivers,
where the apparent image center totally moved up down with changes in ear height.
You're on the right track. In order to make this work, you need to line up the high Q of the shelf filter (typically 1.5-2.7 in my experience; -16dB shelf) with a "too small box, tuned too low" reflex box.
You can see this on page 42 figure 11 of my article https://audioxpress.com/files/attachment/2721 where the woofer response is on the left and the filter is on the right.
So for example your box is tuned to 40Hz, without DSP its response is -8dB and the curve is a low Q gradual taper. You match that to a low shelf filter with Q = ~2.4 and a F = ~25 and if you jiggle the values you can dial in a matching +8dB boost at 40Hz and a steep rolloff below that point.
Like this:
I did this with an online box calculator (top) and a MiniDSP program (bottom) and lined them up. But you can simulate both in VituixCad.
Sometimes you need to mix other notch filters, 6dB high pass filters, or an extra shelf with a different Q to get what you want. It's OK to add even more gain reduction below Fb if you need to. The danger zone for a 40Hz box isn't 20Hz and below, it's 25-35Hz. Minimizing excursion there is more important than group delay.
You WILL have phase shift, make no mistake, and there is group delay, but a shelf filter like this has half the group delay of a 24dB/octave high pass filter and protects the woofer just as effectively.
You'll find that even your 5" woofer, reflex Fb=40, can play surprisingly loud when completely protected from signals below 35Hz. My twelve 3.5" full range drivers in shaded array can shake a modest sized room which is pretty surprising to most people.
P.S. I suggest if you would like to discuss this in more detail, you move the DSP Reflex discussion to this thread: https://www.diyaudio.com/community/threads/dsp-assisted-reflex-system.404349/ and tag me in your post.
Last edited:
@perrymarshall
First of all, thank you for the publication, explanation and details of this project. To the extent of my financial possibilities, I am acquiring components for the manufacture of the OB Bitches Brew and suddenly this LA appears, which has always been one of my challenges and I have some interesting projects in the pipeline. I must say that I am a carpenter who is fond of music and lately I have been eager to learn in order to be able to make my own units. Therefore, my knowledge of electronics, acoustics and other ingredients is very limited.
For this reason I am asking this question, at some point you mention that the Dayton PS95 units could be a good replacement for the Fountek FE85, as I live in Europe it is easier and cheaper for me to use the Dayton units.
Before I wanted to do a simulation with both units to compare, the result confused me because just as the simulation of the Founteck is close to the measurements, the Dayton seems to move away from said frequency response (I understand that the measurements have been made with the miniDSP correction ) So the question is, is the simulation with the Daytons considered good and will it all be a matter of playing with the miniDSP settings?
Attached is a screenshot of both simulations with VituixCAD.
Thanks again for sharing and enlightening.
First of all, thank you for the publication, explanation and details of this project. To the extent of my financial possibilities, I am acquiring components for the manufacture of the OB Bitches Brew and suddenly this LA appears, which has always been one of my challenges and I have some interesting projects in the pipeline. I must say that I am a carpenter who is fond of music and lately I have been eager to learn in order to be able to make my own units. Therefore, my knowledge of electronics, acoustics and other ingredients is very limited.
For this reason I am asking this question, at some point you mention that the Dayton PS95 units could be a good replacement for the Fountek FE85, as I live in Europe it is easier and cheaper for me to use the Dayton units.
Before I wanted to do a simulation with both units to compare, the result confused me because just as the simulation of the Founteck is close to the measurements, the Dayton seems to move away from said frequency response (I understand that the measurements have been made with the miniDSP correction ) So the question is, is the simulation with the Daytons considered good and will it all be a matter of playing with the miniDSP settings?
Attached is a screenshot of both simulations with VituixCAD.
Thanks again for sharing and enlightening.
NO. In fact vocalists (and all instruments) sound very LARGE and three dimensional. I neglected to mention that the stereo image is huge. It's a much bigger sound stage than your typical high end tower design with a couple of woofers and a tweeter.
It does project an impression of the performers being on a stage that is at a lower elevation than the listening position. But it does not in any way sound like "the tweeters are on the floor."
It does project an impression of the performers being on a stage that is at a lower elevation than the listening position. But it does not in any way sound like "the tweeters are on the floor."
Last edited:
I'd never guess from your simulations that you feel inexperienced. I think you did a great job.
You're right, you can swap in PS95 for FE85 and the only thing you need to change is the details of the DSP EQ.
I think you should also be able to simulate the 12 drivers in the reflex box with VituixCad as well, but I'm not certain.
FWIW I think the Bitches Brew is a far superior speaker to the 3.5x12 Shaded Array. The CBT system is great, make no mistake, but the BB is much larger, much more expensive, and is in a different class altogether.
So for example most of the DSP Assisted Reflex systems I've made have been fairly small, and although they punch way above their weight class in terms of bass extension and output, compared to an Open Baffle with 15" woofers they still sound "boxy."
You could consider building an Open Baffle Shaded Array. But that would be a whole 'nother kettle of fish. You'd need to cross it over at 300Hz or so to a subwoofer. I don't know that anyone's ever done that. If I built that kind of system I would not try to have all drivers getting equal equal signal at low frequencies, I would skip the RC network and just use resistors the way Don Keele does on most of his designs and it would be much simpler.
I'm not trying to discourage you from making the Shaded Array as-is, it's a great system. But you mentioned the BB.
Finally, I used the PS95 in a single driver back loaded horn and loved it. I think the paper cone will give you a more "soft dome" sound as opposed to the harder more analytical sound of the aluminum cone Founteks.
All cone materials have a "signature sound." Paper cones sound like paper and metal sound like metal, no matter what you do. I used to design speakers for a living. I designed drivers for Acura, Honda, Mazda and Chrysler vehicles in the ‘90s, and there is no “best” cone material. There are only tradeoffs for what works best in your situation. So there are all grades of hardness and density of paper; a wide variety of plastics from polypropylene and Bextrene to Kevlar to carbon composites. I think the PS95 is on the "high speed" side of paper cone sound and unless you're after an analytical sound you'll be happy with it.
THX
Well, well, we move on, I understand perfectly that the BB is in another league, I have it in the peephole and components are arriving.
The simulation of the sub box in VituixCAD is a job for the next few days, maybe the weekend, even if it is simulated I like to know and understand what I am working on, that way it is easier or at least there is a reference on the to make corrections.
I have transferred the sketches that you published to CAD, I also have the cutting sheet, initially I want to use 18 mm Baltic birch, both units fit on one board!
If it seems appropriate and with your approval, I could publish the dimensioned drawings and cutting plans to encourage and facilitate construction.
Un saludo!!!
Well, well, we move on, I understand perfectly that the BB is in another league, I have it in the peephole and components are arriving.
The simulation of the sub box in VituixCAD is a job for the next few days, maybe the weekend, even if it is simulated I like to know and understand what I am working on, that way it is easier or at least there is a reference on the to make corrections.
I have transferred the sketches that you published to CAD, I also have the cutting sheet, initially I want to use 18 mm Baltic birch, both units fit on one board!
If it seems appropriate and with your approval, I could publish the dimensioned drawings and cutting plans to encourage and facilitate construction.
Un saludo!!!
Aah nice, I opened this thread to ask how this CBT array compares to your other open baffle speakers but I see you've already answered that.
If you had to rank between the BG ribbon array, the Bitches Brew, this CBT one and your other open baffle speaker that had won at the competition (unfortunately I forget its name), how would you roughly compare them? I would assume it would roughly be ribbon > OBs >>> CBT?
Part of my reason for asking is because open baffles have a reputation for "sounding really good" but ribbons have a reputation for sounding the "best". I'd imagine the directivity of the ribbons would make them image a bit different than the OBs but perhaps closer to this CBT. What are your personal opinions/comparisons between them?