Infinite Baffle Double Bass Array

[welwynnick]

Wave Forming, Dirac ART and Double Bass Arrays (DBA) are a revolution in bass management that goes beyond equalisation. They use “subtraction subwoofers” to take bass energy away, as well as adding it. So I wondered how much energy it takes to take bass away?

To keep the box size reasonable, subwoofers force the drive unit to work below the system resonance, so they’re forced to work as static, rather than dynamic transducers, so they’re inefficient.
They don’t need lots of power to move lots of air externally, they need lots of power to compress the air internally, instead of pushing the drive unit backwards and forwards.

Question : can the subtraction subs take advantage of a positive pressure wave?
The pressure waves we hear as sound are very low pressure. The threshold of hearing, 0dB SPL, is generally taken to be 20 µPa, and 100dB is 2Pa, which is tiny. However, subwoofers have to generate much higher pressures inside the cabinet. Consider an example: a 15” driver, 50l enclosure, Sd 840cm2, 101dB at 20Hz at 1m, displacement 700cm3, 8mm travel. When the air inside is compressed quickly, the pressure and temperature rise, so it’s an adiabatic compression, so P1.V1γ = P2.V2γ . The volume decrease here is about 1.4% and the pressure increase is about 2% or 2kPa, or roughly a thousand times higher than the acoustic sound pressure level. An incoming pressure wave is miniscule compared to the internal pressure that the subtraction sub has to generate to absorb it. The answer to the question is no. So is there another way?

Transmission line speakers have always intrigued me. The drive unit doesn’t compress the air in the cabinet in a static manner, it just generates propagating sounds waves that travel around the folded transmission line. The driver doesn’t have to be heavy and inefficient to achieve LF extension, though they never seem to deliver on the theoretical advantages. If the bass wavelength happens to be double the line length, then happy days, but not otherwise. When the frequency gets low and the wavelength long, the transmission line output will start to cancel the driver output and limit the ultimate low frequency extension.

Infinite baffle loading caught my attention a long time ago. The picture above is a typical pair of IB manifolds, with opposed drivers, that vent downwards through the ceiling below. The whole attic becomes the speaker box, so the drivers behave as if they’re in free space, as long as the void is several times the combined Vas of the drivers. I’ve never heard an IB installation, but every single description of the sound has always resonated. It’s not like a box trying to make bass sounds, it’s just there (not unlike my perception of good electrostatic speakers higher up the range). Now I have a handle on that, I wondered if I could take it further.

Imagine I had an IB manifold at the front of the ceiling, and I wanted to “subtract” the bass at the rear of the room, like a double bass array. A positive wave from the front speakers will propagate backwards down the room, while the out-of-phase negative wave in the attic will propagate backwards down the attic. Could I open a hole at the back of the room and let the positive and negative waves cancel each-other out? That would be nice if it worked, and could avoid the need for expensive electronics, but I don’t think sound waves are that obliging. I think both the waves in the attic and the waves in the room would still reflect off whatever they wanted to, rather than conveniently squeezing themselves through some hole in the ceiling.

I think this would work well if I had another manifold at the back of the ceiling, instead of a hole. The speaker would have to be filtered, time delayed and phase reversed, just like a DBA, but I think it would work well.

Could I make an unfolded transmission line in the attic to channel the negative waves where I wanted them, to the back of the room? That’s not very practical because of the roof structure.
How about a false ceiling and a false wall behind the screen that the drivers can vent into? The shallow tunnel would at the same time become a manifold, a transmission line, and an IB cavity. Because the negative waves are out-of-phase with the positive waves, they should cancel each-other out when they reach the back of the room at the same time, like a DBA!

Here’s a two dimensional cross-section through the room to show the concept. I imagine the drivers would be a linear row of several units in a straight line across the front of the room. There would be positive plane waves travelling down the room, and negative plane waves travelling down the tunnel, and forced out at the back of the room. I think the cross sectional area of the tunnel would need to be at least the total area of the drive units.

It's a sort of passive infinite transmission line double bass array, it uses both the positive phase wavefronts from the front of the driver and the negative waves from the rear.

 

[staticV3]

Wouldn't these paths need be of equal length for the DSP-less cancellation to work at all?


Also yes,

I don’t think sound waves are that obliging. I think both the waves in the attic and the waves in the room would still reflect off whatever they wanted to, rather than conveniently squeezing themselves through some hole in the ceiling.

DBA works better the more surface area you can cover in woofers.

 

[olieb]

Wouldn't these paths need be of equal length for the DSP-less cancellation to work at all?

Not if you would fill the channel with helium or hydrogen, but I would not recommend that. There will be some nice resonances, too.
It is a bit like making your room a passive cardioid cabinet (instead of an active one with a "normal" DBA).

 

[welwynnick]

Wouldn't these paths need be of equal length for the DSP-less cancellation to work at all?

Yes they would, but there are lots of tuning options.
For example, the subs could be placed above the screen instead of below it.
The outlet could stop short of the rear wall, or it could go down the rear wall some way.
My original thought was to have a false floor but a false ceiling would allow more flexibility in adding or removing panels to change the position of the outlet.
I don't think this is a solution for anyone else, but I'm looking forwards to playing at some point in the future.

 

[Exprymer]

The technologies mentioned in the OP, altough interesting approaches, are in my opinion, far from being revolutionary, due to the heavily strict constraints in geometry, driver number and positioning, temperature drift and layout changes, making it essentially impossible to reproduce in home studio or home cinema applications where the user usually cannot afford to position drivers in-wall. All those constraints when added together results in skyrocketing costs that only a selected few users are willing to cover, or cinema rooms that have the budget and real need for such control. since the goal is to homogenize the pressure distribution for all the seats. Even still, huge drivers are likely used on the front to approximate a incident plane wave (STRICT REQUIREMENT). Check the Ascendo drivers made solely for cinema applications.

The idea of using the configuration proposed also is prone to fail to the lack of correlation between the incident wave and the negative wave front that supposely will meet the incident wave at the back wall. The incident wave is already not plane, will need to be cancelled by another wavefront that will very likely, look very different than the original signal, since it propagates in a different environment altogether.

When it comes to affordability and real bass management that can be used in any kind of space, the PSI Audio AVAAs are still the best value for money, which is curious given how expensive they may seem at first glance. but it's a garanteed non intrusive option that fits everywhere and it's plug and play.

 

[DVDdoug]

I think this would work well if I had another manifold at the back of the ceiling, instead of a hole. The speaker would have to be filtered, time delayed and phase reversed, just like a DBA, but I think it would work well.

The time delay affects the phase so some frequencies will be in-phase and other frequencies will be out-of-phase. But with the right length/tuning it MIGHT be possible to counteract the standing wave nodes and anti-nodes in the room. But the dips (nodes) and bumps (anti-nodes) are different in different parts of the room.

We only want to cancel the "bad bass" (anti-nodes). Otherwise, it's not hard to subtract-cancel... :} If you have a driver without a baffle, most of the bass leaks-around and cancels. Or wire a 2nd woofer out-of-phase (or wire the left & right speakers out-of-phase) and that will kill almost all of the bass. The real trick is avoiding unwanted cancelation while (partially) canceling some of the "boomy" anti-nodes.

Traditional bass traps try to absorb all of the bass that would otherwise be reflected and that smooths the dips AND the bumps caused by room modes.

A DBA (or other multiple subwoofer approaches) can have similar results. Apparently, "waveforming" works really well, with lots of cone area, lots of amplifier power, and lots of processing.

Usually, the bumps can be knocked-down with EQ but it's harder to fix the dips where the standing waves are canceled.

...In the past I played-around with transmission lines "on paper" but at 20Hz, a half-wavelength (to put the front and back radiation in-phase) is 28-feet and it just turned-out to be impractical. But with the space above your ceiling, that could be done. (My current DIY subwoofers are 15-inch drivers in rather-large ported cabinets and they go-down to the 30Hz range.)

I’ve never heard an IB installation, but every single description of the sound has always resonated.

It would depend on the driver characteristics. With speaker design software (I have WinISD which is free) you can experiment (virtually) with a particular driver and a "large box". That should get you started and then you can consider if you want to try a "port" located some distance from the driver. You can also model a large box and a (large?) port, but I'm not sure how to take the additional distance into account.

To keep the box size reasonable, subwoofers force the drive unit to work below the system resonance, so they’re forced to work as static, rather than dynamic transducers, so they’re inefficient.

Right. Most sealed "home" subwoofers are active with built-in EQ and tons of amplifier power to overcome the fall-off. That can work very well in a "small space" at home. With enough power you can compress the air in the cabinet and move the woofer! It's the most practical way to get deep bass.

Most "pro" subs used live and in dance clubs are large ported designs tuned to go down to around 40Hz. That allows for higher efficiency so you can fill a large venue with bass you can feel in your body. But the response of a ported box falls-off more steeply compared to a sealed box and if you try to extend the low-end with EQ, the woofer just flops-around without making much sound, with the wave from the port canceling the wave from the driver.

P.S.
If you do build something exotic-custom, I recommend getting a measurement mic (about $100 USD) and REW to measure the results.

 

[welwynnick]

Doing wave-forming on the cheap wasn't the sole objective of this idea. If it happens to help with spatial and tonal uniformity, so much the better.
In England we have an expression - to have my cake and eat it - which means to have everything but sacrifice nothing.
I'm a devout objectivist, and I want sound reproduction to be linear with respect to amplitude, frequency and phase, over the whole of the biggest audio envelope possible.
That affects every part of the system, and it's difficult to achieve at low frequencies.

Speakers work well within their pass-band, where they behave as dynamic rather than static transducers. Active subwoofers are compromised by being forced to operate outside their pass-band. There are two consequences. Firstly, they become inefficient, and need powerful amplifiers and motors, and strong, heavy hardware. Secondly, they have non-linear phase response, because they partly operate as a static transducer and the acoustic output is no longer proportional to the input signal, but is at least the second derivative. It's worse with a reflex port, auxiliary bass radiator or coupled cavity loading.

Transmission line speakers help to get round this by allowing low frequency waves to propagate down the transmission line reasonably unimpeded, but there's a practical limit to how long that line can be.

An infinite baffle installation allows the driver resonance to stay low, so it's operating in it's pass-band over a wider range. Instead of being limited by the length of the transmission line, and IB requires a cavity that is typically ten times the total Vas of all the drivers combined. I want a lot of low Fs drivers with a lot of displacement, so that means a big cavity.

That would give me extension, headroom and linear amplitude, frequency and phase response, but I don't have a basement. The IB / DBA gives me all those things, and keeps working at lower and lower frequencies. Although it won't pressurise the room, there's no lower frequency limit because of cabinet volume, amplifier power, transmission line length or basement volume, and I can use as many drivers as I want. In the extreme case as you tend towards DC, the air will simply flow backwards and forwards like an incompressible liquid.

I'm sure it will still need equalisation.

 

[Soniclife]

Some nice out of the box thinking here.

I don't see why it wouldn't work if the lengths are correct, and the path length is perfectly managed. But it all seems really complicated to make work in a niche way, and if you can do all this why not just use multiple IB subs thru a wall (as shown in post 1) and software. If those subs are venting into the same space they will further lower the pressure/sound in the void.

 

[welwynnick]

The time delay affects the phase so some frequencies will be in-phase and other frequencies will be out-of-phase. But with the right length/tuning it MIGHT be possible to counteract the standing wave nodes and anti-nodes in the room. But the dips (nodes) and bumps (anti-nodes) are different in different parts of the room.

Unless there are some very strange dispersion effects going on, the propagation velocity and time delay for the two paths will be the same for all frequencies. Granted it wouldn't be easy, but if the path lengths are indeed equal, then the phase difference will be preserved. If the positive wave and the negative wave are out of phase at the speaker, then they will be out of phase at the back of the room. It's not frequency sensitive.

The idea is that the cancellation will take place at the back of the room, where the reflection of the positive wave would otherwise come from. The cancellation wouldn't be taking place at the listening position. Sure, it won't take place over the whole of the wall, but it would be similar to a DBA. I expect to use a continuous row of speakers in a line across the front wall, so the advancing wave is cylindrical. Not plane waves, but not bad. On reflection it might be best to have the speakers above the screen rather than below it. That might give me the best chance to get the path lengths the same.

But it all seems really complicated to make work in a niche way, and if you can do all this why not just use multiple IB subs thru a wall (as shown in post 1) and software. If those subs are venting into the same space they will further lower the pressure/sound in the void.

That was why I said this in post 1:

I think this would work well if I had another manifold at the back of the ceiling, instead of a hole. The speaker would have to be filtered, time delayed and phase reversed, just like a DBA, but I think it would work well.

That's not a bad solution. It would need a more speakers and more electronics, but it would allow the time delays etc to be configured with software instead of carpentry.
Indeed, if the rear subs were phased properly, they would apply cancellation to both the room AND THE ROOF CAVITY. That means the roof cavity could be small - it wouldn't need to be 10 x Vas at all. The negative waves from the primary subs would be cancelled out by the rear subs by the time the waves got there.

 

[Elgrosso]

To maximize the chance for the front and back sound-waves to have a similar shape and behavior, so as planar as possible when they meet: what about adding the same tunnel in the front of the driver as well?
With the driver being hidden somewhere of the ceiling, and two similar length tunnels ending in apertures in the middle of the front and rear walls. Driver position and tunnels to be adjusted for the cancellation to happen where you want.

 

[welwynnick]

To maximize the chance for the front and back sound-waves to have a similar shape and behavior, so as planar as possible when they meet: what about adding the same tunnel in the front of the driver as well?
With the driver being hidden somewhere of the ceiling, and two similar length tunnels ending in apertures in the middle of the front and rear walls. Driver position and tunnels to be adjusted for the cancellation to happen where you want.

Something like this then?
Hopefully, that will help to equalise the path differences, and cancellation would be achieved at the back of the room.
Tuning could be done by varying the length of the vertical portion of the duct on the rear wall.
However there are some practical considerations, like how to access the drivers, and there can't be a door or window on the rear wall.

 

[welwynnick]

The other option is to have the subs opening above the screen.
One of my tuning options would be for the duct to made be made of removable panels over the WHOLE of the rear wall.
One row of panels could be left open at whatever height I wanted the duct to open.
I'm not sure where that should be, but a bit of experimentation would be in order.

The alternative would be to have a false floor, with the subs at the bottom at the front, and the duct going maybe half way up the rear wall.

 

[Flaesh]

It won't work as a DBA. In short.

 

[Elgrosso]

I was more thinking about trying to apply the same constrains to both waves, in order to get them as “symmetric" as possible.
The openings in the front and rear walls are the ones driving the shape of the soundwaves, they need to follow the real dba guideline for placement, for both waves to have a chance to get planar when they meet.
So let’s say two fake walls front and rear, few holes the right size and placement, mirrored image front and rear.
Driver in a tunnel in the ceiling, firing into each plenums.
Not talking about resonance issues, at least both waves should cancel correctly.
Then moving back and forth the driver should move the point of cancellation (with more resonance effects).

But for the amount of work it would be much simpler to just add IB subs in both fake walls.

 

[welwynnick]

It won't work as a DBA. In short.

Oh, OK.
I won't do it then, if you say so.

 

[welwynnick]

I was more thinking about trying to apply the same constrains to both waves, in order to get them as “symmetric" as possible.
The openings in the front and rear walls are the ones driving the shape of the soundwaves, they need to follow the real dba guideline for placement, for both waves to have a chance to get planar when they meet.
So let’s say two fake walls front and rear, few holes the right size and placement, mirrored image front and rear.
Driver in a tunnel in the ceiling, firing into each plenums.
Not talking about resonance issues, at least both waves should cancel correctly.
Then moving back and forth the driver should move the point of cancellation (with more resonance effects).
But for the amount of work it would be much simpler to just add IB subs in both fake walls.

So : have full-height false walls front and rear, and have a 2D array of openings in each, as if to emulate a DBA?
That's starting to seem like a lot of effort to avoid having a DBA.
I think your final comment about putting subs in both walls makes sense.
My thought is whether they can still work as IB subs - that's what I'm after just as much as travelling plane wave propagation. Emulating a DBA isn't the only objective.

 

[Flaesh]

full-height false walls front and rear, and have a 2D array of openings in each, as if to emulate a DBA?

This will work well as a DBA. A real, good DBA, not an emulation.
Have you read the threads about DBA here? Including the links. Everything has already been discussed, both theoretically and practically. For the lowest frequencies, the SSS configuration (single source to sink, as Fazenda calls it) can effectively emulate a DBA.

Regarding the invention in the start post. Consider a tube with a transducer at one end and an open end at the other. What will be the amplitude and phase characteristics?))..

whether they can still work as IB subs - that's what I'm after just as much as travelling plane wave propagation

The meaning of this is unclear to me. In any case, the DBA (and specifically the DBA*) creates a [quasi]plane traveling wave. Regardless of the volume of the boxes or the ratio of that volume to the equivalent volume.
Can you draw your room in more detail? Up to what frequency do you want to use subwoofers? Perhaps SSS or S(several))SS are sufficient.

*The SBA, too, to the extent that rear absorption is effective.

 

[NTK]

Ran a couple of simulations. The first simulated something like the proposed concept in post #1. The second has the back waves "turned off" to compare how much effects the back radiation had. The overall length of the room was 5.5 m and the overall height was 2.5 m. The excitation signal is a single period of 120 Hz signal, band-limited by ramped up and down using a Hann window. (A lower frequency will increase the wavelength and make it more difficult to visualize.)

Arguably, in these sims, the first case gave worse results. The back waves exiting at the far end of the room are delayed relative to the direct waves. Not only they don't cancel the direct waves, in this case they behaved like an extra added "reflection". (Note that these effects are frequency and room dimensions dependent.)

The animations show, for DBA to work, the importance of correct timing. Furthermore, the back waves travelled inside a narrow passage. If the walls are lossy, there will be a lot of attenuation. When sound waves travel, they don't "flow" like a fluid and flow along streamlines (like what is depicted in this picture). They travel in straight lines and bounce back and forth when they hit the duct walls (just like how light waves travel inisde optical fibres). Energy is loss in every reflection, and therefore the amplitude of the sound exiting the duct at the far end of the room can be significantly less than the direct sound. However, this loss may be partially offset by the "horn loading" of the back waves, which will depend on the geometry of the duct. So things are complicated.

 

[Flaesh]

overall height was 2.5 m. The excitation signal is a single period of 120 Hz signal, band-limited by ramped up and down using a Hann window. (A lower frequency will increase the wavelength and make it more difficult to visualize

IMO 120 Hz is too high for one sub in 2.5 m.
ps: theoretically it will work in a ideal room.

 

[welwynnick]

Have you read the threads about DBA here? Including the links. Everything has already been discussed, both theoretically and practically. For the lowest frequencies, the SSS configuration (single source to sink, as Fazenda calls it) can effectively emulate a DBA.

I did read StigEric's tremendous thread of course . I'm pretty clear about what waveforming and DBA do, but less clear about ART, just lots of arm-waving.

The meaning of this is unclear to me. In any case, the DBA (and specifically the DBA*) creates a [quasi]plane traveling wave. Regardless of the volume of the boxes or the ratio of that volume to the equivalent volume.
Can you draw your room in more detail? Up to what frequency do you want to use subwoofers? Perhaps SSS or S(several))SS are sufficient.

I want an infinite baffle solution, or something equivalent to it. That means the diver isn't loaded by a cabinet smaller than the Vas, but by a room that is much larger than Vas. The objective is to have drivers with a relatively low moving mass, high compliance, high efficiency and high displacement. The motors can be relatively modest, with a moderately small magnet, force factor and inductance, and no high power amplifiers. Those things all contrive to require a very large cavity for the subs to vent into. As well as extension and headroom, I want the sub drivers to work in their dynamic region as far as possible, to minimise phase variation within the audio band. The room is about 7 x 4 x 2.5m, and I want to work over a wide bandwidth - at least 20 to 100Hz, and preferably wider.

Ran a couple of simulations. The first simulated something like the proposed concept in post #1. The second has the back waves "turned off" to compare how much effects the back radiation had. The overall length of the room was 5.5 m and the overall height was 2.5 m. The excitation signal is a single period of 120 Hz signal, band-limited by ramped up and down using a Hann window. (A lower frequency will increase the wavelength and make it more difficult to visualize.)

Many thanks for the modelling, I appreciate that, even if the result wasn't what I hoped. Did you model it in two dimensions? My intention is to have a continuous row of subs across the whole width of the room. It will only approximate to cylindrical waves, not plane waves.
It looks like the tunnel outlet makes little difference. That's a bit of a surprise, as all that air has to go somewhere. Maybe the tunnel is attenuating the waves, which both a good thing and a bad thing. It's bad because it doesn't help cancellation, but good because it means I might be able to use this glorified transmission line as as effective IB, without incurring negative effects. It doesn't put me off, and I'm probably going to go ahead anyway. Maybe I need to play with a wider or longer tunnel, or smoother turns or an outlet like a horn.