Double Bass Array (DBA) - The modern bass concept! - Page 4

Very interesting, I will have to use this design in my next HT build, if the housing market ever turns around

I would like to create a DBA in my nicely rectangular 7'11-1/2" x 14'9" x 21'5-1/2" (4.50 m x 2.43m x 6.54m) WxHxL room.
Nils's flat subwoofers would work, but sufficient spaces behind one end wall and above the entire ceiling practically beg for Infinite Baffle subwoofers.
So, in striving to combine IB's very low bass extension with DBA's suppression of modal effects:
(1a) Can I use ceiling-mounted IBs at the non-IB-able end to cancel any of the SBA's wavefront?
(1b) What if I put a quad array of IB subwoofers at the end where I can, and use planar ported subs for the opposite wall?
(1c) What about IBs at the ceiling of one end with only floor-level IBs at the opposite end?
(1d) How is the distance from IB drivers to manifold opening related to manifold area?
(For my case, could I make 6-ft tubes to bring the manifold down from the ceiling?)
(1e) Ceiling IBs both ends? (Definite FAF improvement, but don't properly delayed ceiling-only IBs do at least some mode-quenching magic?)

For extra credit, I'd like to increase the upper cutoff frequency above the room's quad-array limit of 76.6 hz.
So, I need to decrease the minimum driver separation.
One driver in the center of each wall would help, but the room won't permit.
The IB wall can handle any configuration of drivers, but the other end won't allow drivers in the middle 45% of the bottom 55%.
So,
(2a) add one ceiling IB center one end, floor IB other?
(2b) add one each ceiling IBs both ends?
(2c) Nils's flat subwoofer a little above center at non-IB end, and another at opposite end above/at/below center
(2d) other ideas?

Lastly, how should I go about beating down higher-frequency reflections?
DBA's infinite array of mirror-image subwoofers relies on having little absorption in the listening room walls, floor and celing.
But the room will still resonate at all the higher harmonics of 1/2 the room dimensions,
so these frequencies still require treatment.
I guess the usual approach is cloth-covered "rigid fiberglass" (gotta research this still),
but (3a) how might one best approach the tradeoffs between absorption and subwoofer imaging?
(3b) Is there a better approach than slapping a rug/pillow over each of the mid+tweeter images?

Thanks in advance,
Yoga Man

Some general answers, yogaman:

For a DBA to work properly, the rear wall sub positions must exactly match the front sub positions. Front and back subs should be the same type so that their frequency response is the same. I supposed this could be equalized for with different sub types, but then you have to worry about time delay through the equalization filter.

The sub positions on the walls cannot be chosen arbitrarily. The inter-sub spacings must be 2x the spacings from sub to nearest wall, floor or ceiling. The idea of this is that if you created the mirror images of the subs reflected in side walls, floor, and ceiling, the original sub positions plus the mirrored sub positions will all lie on a reguler rectangular grid. If you research the "image source method" of acoustic modelling, you can learn why this has to be so.

Regards,
Terry

i described the same exact concept on Partsexpress forum when i was discussing various possibilities for active room correction ...

if i thought of it - it must be good

Vasyachkin,
Kudos to you for prior art, but which scenario did you favor? Did you implement? Results, pointers, etc, appreciated.

Terry,
Thanks for a first jab at my problem.

I can see that my presentation has not conveyed my design questions very well.
Please allow some review and more detail.

As we all know, the simplest DBA is one speaker each in the center of the front and back walls.
The walls create acoustic images of these drivers every room-width to each side;
the ceiling and floor image the drivers every room-height above and below; and
this repeats, including all the diagonals, even out to infinity, for perfectly reflective walls, floors and ceilings.

A two-speaker DBA array centered vertically doubles the horizontal density of this simplest DBA,
but leaves the vertical density the same.
The two-speaker DBA array also shifts the driver horizontally by 1/4 of the room width, but the vertical offset is unchanged.
Nils's beautiful quad DBA then doubles the vertical density, too, also shifting the vertical placement by 1/4 room-height.

In all these configurations, the assumption is that the rear wall array is configured identically to the front wall array.
But it's also clear that the number of speakers and their horizontal and vertical "phasing" can be varied.
So, my fundamental inquiry is, "What would happen if the front and rear arrays were not identical?"

For 1a, the proposed arrays are Nils-style quad at one end, and dual ceiling at the other end.
Note that the horizontal alignment is the same at both ends, with two vertical lines of driver images half a room-width apart, 1/4 room-width on either side of each wall image.
The dual ceiling drivers are vertically offset by 1/4 of the room-height relative to the quad array, and every other driver (image) is missing.
Still, can't the sparser array cancel at least some of the back wall reflections? (How much?)

Regarding 1b: I understand that mismatches of the original and reflected wavefronts will result in non-cancellation,
but wouldn't "different but equalized" speakers get you at least most of the way there?
(I'm gonna have furniture, too; is this speaker mismatch more distorting than furniture?)

The images in 1c similarly are aligned horizontally and offset vertically between the end walls, this time in an attempt to minimize the number of drivers.
How well will this work?
(I observe that one way to think about this configuration is to treat the wavefront as moving along the diagonal
rather than parallel to the walls. Is that useful? Btw, I bet this configuration is most susceptible to furniture.)

1d reflects ignorance about IB manifold design.
I realize this may not be the best thread to pose this question.
Sorry, but if someone can point me to...

1e is similar to 1c, but the vertically sparse emitters (and images) are at the same heights
on front and back walls (and images), so does this mean they'll cancel each other better or worse than floor vs. ceiling?
(My guess is 1e is better than 1c.)

The added drivers and their images in my part 2 above are attempting to decrease the maximum separations
between emitters (real and virtual) so that the cutoff frequency could be raised.
Thus, they would be placed in the horizontal center.
2a and 2b enumerate the IB options.
Again, sure, the match won't be perfect, and/or the array will not be fully filled in, but won't there be *some* bandwidth gain?
Which should be better?
(How much?)

In 2c, regarding vertical alignment between the two ends, one suggestion is to place them symmetrically, ie, where one end is slightly above and the other below the horizontal midline;
another suggestion is maintaining the same height from the ceiling/floor.
Or misalign them, and place one end in the exact center where possible.
My guess is symmetry is best, especially for small offsets. But can I get increased bandwidth?

Part 3 seeks advice on how to simultaneously achieve low-frequency reflection and high-frequency absorption.
I suspect the key is that the LF reflections utilize the whole walls, ceiling, and floor, while the HF reflections
can be suppressed by treating their reflections as narrower beams,
but I hope a DBA enthusiast might have already resolved this dilemma.

Thanks to everyone who has taken the time for considering my "out-of-the-box imaginings".
(Seriously evil pun intended.)

Yogaman

here is the original thread:

http://techtalk.parts-express.com/sh....php?p=1532641

having looked over it just now i don't see myself talking about this particular design. probably because i didn't think it was practical for a residential setting. i was thinking for a dance club to have the entire floor as one array and entire ceiling as another ( delayed and inverted ). that way the entire dance floor, no matter the size, would have perfect bass.

read the whole thread, well, not the whole just up to the point where i stop posting ( because i am banned ) ...

Yogaman,

I don't think most of your questions can be answered without running a simulation of the specific layout. It would just be speculation for us to try.

I will say, I don't think ceiling speakers by themselves would work very well. The driver-driver spacing should be under 1/2 wavelength and the d-d spacing of a ceiling speaker is twice the room height. For an 8' ceiling, that means your max 'recommended' frequency would be about 35Hz. I have a room similar to yours where I could do an IB array at one end but not at the other. I have considered doing shallow floor-ceiling boxes that are open to the attic at the top and lightly stuffed to prevent pipe resonances. Pretty close to IB. Or I could just build sealed boxes and use a little EQ.

Terry,

I don't think using digital EQ is a problem since it's one of the requirements of the DBA that the back speakers be delayed by the time of flight. The Behringer DEQ2496 or DCX2496 would be very cost effective and give you powerful EQ as well as the required digital delay.

I've wondered if you couldn't do some EQ trickery to fight nulls like the one Nils has up around 80Hz. You can't boost a null but what you can do is put a notch filter on the inverted rear signal. Would it work? I dunno. The same concept could be used to take advantage of room gain below the first longitudinal mode. The DBA, in theory, cancels all room gain. However, you could roll off the rear array at very low frequencies so the front array gets its room gain.

Bottom line, I think you could deviate quite a bit from the 'ideal' DBA and still take advantage of the DBA concept with some clever EQ, adjusting the front and rear arrays separately, not as a single unit. My living room has openings to other parts of the house and I'm not inclined to install doors. I suspect, with a 6-channel DCX2496, I could still dial in different parts of the array so the sum is pretty good.

PS -- I think we can safely ignore Vas's prior art as this thread is two years older than his thread at PE.

while i still haven't read this thread ( nor do i plan to ) i do believe the subject matter is somewhat different from the one we had at PE ...

this whole bass control is such a big subject - it makes me really reluctant to write an article about it - i am too lazy to even read about it in fact lol

Thanks, Dennis, for the helpful suggestions, especially your insight that the upper cutoff of ceiling-only drivers is 35 hz.
Happily, this reduces the number of choices for my DBA configuration.

What tool (free, I hope?) would you recommend for simulating non-identical arrays?

Regarding your question about improving the null:
The planar wave from the front array should not reflect at all from the back wall, thanks to the rear array.
(And not from the walls, floor or ceiling because it's planar, of course.)
So, if the null you discuss is due to reflection(s) from the walls floor or ceiling, then notching the cancellation wave from the rear drivers should make the null worse.
So whence does the null arise? Furniture? Near-field effects?
If so, then, yes, maybe you will see more amplitude at that null location,
but you'll also be reintroducing all the room modes at that frequency.
I suppose that could turn out to be what the room needs, but it seems unlikely.

I also thought about making cabinets that could open through the ceiling as you describe.
I asked it as a manifold distance question.
I think the IB design doesn't care where drivers are placed, as long as you avoid constricting the flow.
Your idea to stuff openings for tuning seems like a good one.
Were you thinking to put both drivers in the same enclosure?
If so, does that mean the enclosure volume should double to 20x Vas?
Should the ceiling aperture area then be 2x the area of a single driver?

Besides those IB questions, my new list of DBA questions is:

4a) Should I put IB drivers where I can, and use Nils-style flat drivers such as Nils's, EQing out any response mismatches?
4b) Or should I use only Nils-like drivers everywhere?

5a) Does the distributed (ie, non-lumped) model of Nils-style subwoofers mean that he ignored the locations of four of his sources, ie, the ports?
5b) If so, could these extra radiators be considered as increasing the effective density of the driver arrays, and thereby the cutoff frequency for the DBA?
For example, leave speakers and ports at 1/4 and 3/4 room-height, but move speakers to 3/8 and 5/8 room-width, and recognize the port radiators at 1/8 and 7/8.

Uh, in ascii, that's something like:

0 1 2 3 4 5 6 7 0 <- eighths of a room-width
| | | | | | | | |
v v v v v v v v v

------------------------- Ceiling
| +-------+ +-------+ |
| |p D| |D p| | <-- 3/4 room-height
| +-------+ +-------+ |
| | <-- 1/2 room-height
| +-------+ +-------+ |
| |p D| |D p| | <-- 1/4 room-height
| +-------+ +-------+ |
------------------------- Floor

+-------+
|D p| = a subwoofer with a single driver (D) and a single port (p) located as shown
+-------+

(Display the above in a fixed width font such as Courier to make it look right. -ym)

6) And, renaming (3), how can I integrate DBA's reflection-based, nearly anechoic bass with the need to minimize higher frequency reflections?

7) Hey! How about designing a subwoofer with multiple ports? Flat drivers seem to be the perfect shape for this. Not bunched together like SVS PB12 tuning ports, but strategically placed according to the wall coordinates. Wouldn't this yield an array with even greater radiator density?

Thanks again to all,
Yogaman

PS. After hearing vasyachin describe his invention and his approach to knowledge, I'm disinclined to recognize his claim to the concept of mismatched arrays. Mismatched arrays seems like a fairly obvious concept, though. Equally obvious, I think, is the observation that the port and its images can be treated as additional array elements, although if it's so obvious why didn't I think to ask about it until this morning? I bet RF engineers have already thought a whole lot about this stuff...

Sorry, I'm not aware of any reasonably priced software that could do it. Terry Montlick (above) uses his own custom room software but I doubt he'd be willing to run custom sims for you unless you wanted to pay his hourly rate.

About nulls, they're surprisingly easy to create and the way to fix them isn't always obvious. JohnK over at the PE board modified his room software to simulate a 2-front, 2-rear DBA. In a perfectly reflective room (concrete cave) it was almost perfect up to 70Hz or so. However adding some wall absorption introduced some deep nulls at (I think) 22 and 66 Hz. I'm 99% sure it's because he didn't decrease the gain on the rear array, to account for the sound of the front array being absorbed by the walls, but he wasn't interested in modifying his software (again) to check it out. There was also some question about the mic position (too near the front or rear wall, I forget).

As for the rest, we're all just guessing until we build something.

Thanks again, Dennis. I'll hunt down the PE forum thread you suggest.

Also, thank you very much for helping me better understand your suggestion about altering the reflection-cancellation wavefront image to compensate for distorting objects in the room.
I believe you are 100% correct.

Couldn't a modern pc adjust delay and equalization for the drivers in real time using something like the freeware program, Convolver? (Rather than an separate equalizer boxes for each (pair of) channels.)

Maybe a future version could even compensate for objects/people moving around in the room... That'd be some nice software, now wouldn't it?

And, as long as we're making ultra-skinny speaker boxes, why don't we mount a bunch of mid-ranges and tweeters all over all the walls and ceiling, and actively cancel all reflections everywhere?
That'd sure solve my number 6 or 3, wouldn't it?

May I borrow your checkbook, please?

Best wishes,
-yogaman

Acknowledgement: I see that Vas did discuss active equalizers in his other thread. But I still bet AE is old news. -ym

Who has a pointer to a multi-port subwoofer design, please?

How about using rear subs with local microphones in a feedback loop? They would cancel any signal within the loop's bandwidth, so there is no need for digital delay, and they could probably also deal with non-rectangular rooms.

Or to have the best of two worlds, have the delayed, attenuated and inverted signal on the positive input and the mike signal on the negative.

Has anyone had a chance to exercise this approach? I have 4 Ficar IB18's that I was planning on using for dual IB manifolds. I am in the planning stages of building a dedicated HT and am toying with the idea of a DBA or SBA. Now I am thinking I can use either 2 manifolds (2 drivers each) midwall in height or 4 front facing drivers (would need something like floor to ceiling lam beams to prevent vibration) at the aforementioned locations. My concern is with real world conditions i.e., chairs, soffits, acoustic treatments for higher frequencies etc. Will these negate the benefit?

Mike

Have you checked out Paul W's HT? He went with an array on the front wall and bass traps on the back wall.

http://advanceddefinition.com/The%20Octagon.htm

Yes, that's what really piqued my interest. I asked him about his approach using a limp mass absorber (hanging several layers of tar paper). Apparently it works reasonably well.

This thread need a bump!
I have finally bought 16 15" woofers for my DBA project. I have thought about this for over 2 years now and is eager to get started. I have a square room 5,3 meters in width, 2,35 meters high and 5,9 meters long. The woofers I bought were on offer this week and seemed suited for the project. With 8 woofers on the front wall the longest distance between woofers will be 1,38 meters and I should get usable frequency response up to 122Hz. I have read this thread i couple of times but there is still one question that is not answered.

* How to calculate the total SPL from a DBA?

The way I see it: In my case I will get 8 "tunnels" where the sound wave from each woofer travels. The sound have to stay in its "tunnel" because on each side of the tunnel there is either a wall/floor/ceiling OR another tunnel with the same pressure from a neighboring woofer. Let’s assume this is true and take a closer look at one tunnel.
A woofer is omnidirectional in its frequency range, but in this case where its situated in the "end" of a tunnel the pressure can only travel forward. We will therefore get a 6dB increase from each of the directions it cannot travel and 3dB because it cannot go backwards either. In total 15dB. If each tunnels cross section is about 1 meter we can take the woofers SPL @ 1 meter and just add 15dB to it. Does anyone have any thoughts about this?
Another thing is: How much will a bigger "tunnel" decrease sound pressure?

My woofers can generate maximum 90dB SPL @ 20Hz @ 1 meter distance (closed box and maximum excursion 6,5mm p-p.) My "tunnels" are 1,2x1,4 meters in cross section. According to my theory I will get an added 15dB minus maybe 2dB from my tunnels being larger than 1 meter. And will end up on a maximum of about 103dB @ 20Hz in the entire room.
If I on the other hand mounted all 8 woofers in one big cabinet it would give a total of 90+6+6+6=108dB SPL plus added 6-9dB if placed in corner for a total of about 115dB instead of 103dB... Is it really true that DBA is this inefficient?

Any help on the subject would be appreciated.

the tradeoff will be that you will need more power for the same spl.

however, in a world where power is cheap, i like this approach.

not the first time that i have seen this approach...

very good solution though.

"I have finally bought 16 15" woofers for my DBA project."

holy crap. i can't wait to see what you come up with aaa. i must admit that i don't understand your concept of 'tunnels', but you may be on to something.

Don't hold your breath when it comes to my projects. They tend to take some time. But 142 kilos of woofers arrived today and I will post my results in this thread when I have some progress. The woofers I bought are not exactly top of the line with only 6,5mm excursion p-p but they have pretty good price/performance ratio at 110$ each. Time will show if put an accelerometer in then too.

Digging up an old thread! ...

One question - how far are your rear subs behind your listening position? Do you have a problem with localisation?

^^ I'd be interested to read more details of the design and implementation of your new mains.

I don't see precedence as having the slightest effect in this instance. Precedence was tested in everything I have read anyway, with clicks, a predominantly high frequency function where effects like HRTF manifest, not at 100Hz where your 2m distance is barely more than half a wavelength.

I am extremely interested in doing a DBA setup. If all goes to plan, I will have brand new basement at my disposal.

My question has to do with the 1/4 spacing.

Lets assume for a minute a front IB setup is done with the correct spacing (4 sets of dual opposed 15's with manifolds exiting the 1/4 spacing. Assume 17' x 8' wall.

Now on the the rear.

What if I have seating on risers?
Do I need to make sure the risers end before the rear wall?
As in, do I have to build in a gap from the rear of the risers/seating that steps back down to the full 8' height of the wall?

Does that mean that if the risers "level" continued all the way to the back wall, I would be not be able to place the rear IB correctly? If the manifolds were placed on the 1/4 distance based on the new wall height, would it still work, or do I need the original 8' height?

Thanks
Steve

Steve, I would let the risers go all the way back to the wall, then adjust the positions according to the rule. It should not be a problem. Think of it this way. Imagine the boundaries are all tinted glass mirrors so that you see an array that extends to infinity, but becomes progressively darker due to losses at the boundary. Alignment of front and rear wall is not the critical issue.

Thanks Paul,

I will start planning accordingly. I will start a thread when the new house becomes a reality.

follgott, very nice response from the mains!

seaton made a comment back on the first page of the thread that makes a lot of sense. the idea was to "roll" the filter on the back wall of subs below the point where you transition from modal behavior to pressure vessel behavior so that the subs on the back wall can contribute spl to ultra low frequencies. have you experimented with that approach yet?

An all pass filter would be perfect for that since the delay is frequency dependent. It's something I've been meaning to do, but just haven't had the right combination of resources to make happen.

That said, I would argue that there are some psycho-acoustic benefits to not having pressure gain effects. Is that more important than the corresponding distortion/output limits? The majority here could probably care less, but in the world of striving for accuracy it poses some interesting differences. The cool thing is you could set up your processing to choose your flavor with the flip of a switch...