Even bass coverage in bassment

I just can't figure this out. I have four 12 inch subs that I'm using. The basement is 25 ft long by 20 ft wide. I can't seem to get even coverage in the marked (rectangular) area. I always seem to get a dead spot (no bass at all) just to the right of the stairs (shown by the circle).

One setup I had, shown in the second picture, the subs (green) on the right were out of phase with the subs on the left. That gave me tons of deep bass on the wall with the doors, but towards the middle of the basement, the bass would cancel. Then I had my full-range speakers (one in each corner) fill in the bass, and that sounded decent. This setup somehow took care of the dead spot in front of the stairs, but I didn't like the idea of having half the subs out of phase, and the cancellation towards the middle.

Does anyone have any ideas to get pretty even coverage? I have tried just about everything, from a sub in each corner, to all the subs in one corner, to the subs spread out along one wall and all firing one way. I got the best results with half the subs out of phase and the mids filling in. Give me something else to try, because I'm out of ideas.

[IMGDEAD]https://i207.photobucket.com/albums/bb233/xplod1236/My%20subs/01.jpg[/IMGDEAD]

[IMGDEAD]https://i207.photobucket.com/albums/bb233/xplod1236/My%20subs/02.jpg[/IMGDEAD]
 
http://www.harman.com/wp/pdf/multsubs.pdf

The short summary of lessons from that publication:
- use 2 or 4 subs
- put them symmetrically midpoints of walls for most even response

Granted, you have a non-standard room where the stairway, etc will skew the modes, but I think the general ideas behind the results still apply. I suspect all 4 of your subwoofers are coupling to the room rather well since they're either in a corner or not positioned at a low order null. The two along the long walls should cancel odd order axial modes for each other, but that's about all that's helping you.

One way you can attempt to map the modes in your room is to put a sub in the corner, play a tone, and try to find nulls with a spl meter. They should be the best locations for the subs for flat response. It's a bit error prone though.
 

nunayafb

Member
2005-08-17 11:05 pm
a sub in each corner, to all the subs in one corner, to the subs spread out along one wall and all firing one way
It sounds like you are making too large of changes, what you might want to consider is starting with a logical position and tune from there.
Based on the room having two different dimensions (not square) it is going to have different resonant modes in one direction than in the other. So if you put two subs in the middle of each wall you will get symmetry and any resonant modes will add up, same with corner loading.

I would start with this:
_.........________________________
........................................O........|
..................................................|
..................................................|
..................................................|
..................................................|
................................................O|
..................................................|
..................................................|
..................................................|
..................................................|
..................................................|
..................................................|
..................................................|
..................................................|
..................................................|
................................................O|
..................................................|
..................................................|
____________________________O__|
The "O's" representing starting woofer locations, remember you are dealing with phase relationships from multiple sources and with multiple reflections so start with moving each sub a couple of inches and one at a time.

BTW, I arrived at this arrangement by using, first an idea of asymmetry, then checking the result using a wave applet

Good luck, (be patient and you wont need luck)

-J
 

y8s

Member
2006-05-19 4:16 pm
jason_watkins said:
http://www.harman.com/wp/pdf/multsubs.pdf

The short summary of lessons from that publication:
- use 2 or 4 subs
- put them symmetrically midpoints of walls for most even response

Granted, you have a non-standard room where the stairway, etc will skew the modes, but I think the general ideas behind the results still apply. I suspect all 4 of your subwoofers are coupling to the room rather well since they're either in a corner or not positioned at a low order null. The two along the long walls should cancel odd order axial modes for each other, but that's about all that's helping you.

One way you can attempt to map the modes in your room is to put a sub in the corner, play a tone, and try to find nulls with a spl meter. They should be the best locations for the subs for flat response. It's a bit error prone though.

hey that's one of the most useful audio posts i've seen in a long time that's totally accessible and easily implemented by the commoner.

thanks!

Matt
 
Based on the room having two different dimensions (not square) it is going to have different resonant modes in one direction than in the other. So if you put two subs in the middle of each wall you will get symmetry and any resonant modes will add up, same with corner loading.

I don't believe this is correct.

There's are two symmetries you can use to put subwoofers in places that couple to the room less:

1. subwoofer along the midpoint line between two facing walls.
2. subwoofers each directly against two facing walls.

The first avoids even order axial modes because it locates the subwoofer at a null.

The second avoids odd order axial modes since the subwoofers cancel each other.

In an ideal rectangular room, the midpoints of the walls have both these properties at the same time, which is great. The room doesn't have to be square. If you check the publication/slide deck you'll see all of their analysis was in rectangular rooms.

In a more typical room with doors, windows, differing wall lengths, etc, these axial modes are still present... however, they may be significantly lower in impact, so that the room becomes dominated by tangental modes. Hard to do anything about that, but we still can at least avoid the axial modes.

Once I learn a bit more mathematica, I plan to replicate toole's analysis for multiple dipole woofers as well as more commonly found 'strange' rooms like L's.
 
How about adding different delays (or local phase shift in narrow frequency ranges with high-Q 1dB parametric EQs) to the signal going to each subwoofer?

Is there any ready-made software capable of modelling this?

There are low-cost tools to do the signal processing, like Behringer DCX2496, but the amount of possible adjustments is just too high to try to tune it by ear without a previous mathematic analysis.
 

nunayafb

Member
2005-08-17 11:05 pm
The first avoids even order axial modes because it locates the subwoofer at a null.

The second avoids odd order axial modes since the subwoofers cancel each other.

The first... at a null, this only for resonant modes, can we agree that is makes the fr worse?

The second.... how do they cancel? if they are symmetrically placed and wired in phase then any peaks and nulls due to phase addition/cancellation will be cumulative.

The general rule is don't build square rooms because the axial and lateral modes will be identical and therefore cumulative, the same thing would apply to multiple subs right, if one placed on a midpoint has peaks and nulls due to reflected energy then two at opposing midpoints will have bigger peaks and nulls.

So even with a rectangular room, both woofers placed at midpoints will share the same axial modes (assuming side wall placement) and you get two woofers operating at the same frequencies with the same peaks and nulls. I'm not talking modal resonance here, I'm talking FR, maybe this is where we are disagreeing

Look at his plots again, The flattest responses were from randomly placed woofers, all of the others had max-min of ~30 dB if you look at the graphs, or ~17dB if you read his text. It would have been nice if he did some randomly placed woofers and eliminated the against the wall restriction just for completeness. I am not knocking his good work, but with too many restrictions you limit the output and the conclusions that can be drawn from the data

I'm no expert but I would say that exciting modal resonance is only one half of the equation, the frequency response is what happens first, the resonance has to build up and that can be damped.

I may be completely wrong here, but this is what I think and I do plan on looking into this some more and hope everyone keeps discussing this and that more chime in.

-J
 
Eva: Phase shifting is an interesting idea.

I hear Harmon is working on a somewhat similar produc, for high end customers. Microphones are placed throughout the listening area. A subwoofer is placed in all practical potential locations, and measurements taken. Software optimization takes the measurements, derives the modal behavior of the room, and recommends the optimum number and placement as well as dsp correction filters.
 
The first... at a null, this only for resonant modes, can we agree that is makes the fr worse?
No, this makes frequency response flatter. Nulls are points of destructive interference. In this case, it's the subwoofer canceling itself by reflecting off the walls.

The second.... how do they cancel? if they are symmetrically placed and wired in phase then any peaks and nulls due to phase addition/cancellation will be cumulative.
Pairs of subwoofers placed symmetrically each on a reflecting wall cancel odd order axial modes between those walls. Because it's an odd order mode, the reflections are directly out of phase with the primary signal at each subwoofer.

The general rule is don't build square rooms because the axial and lateral modes will be identical and therefore cumulative, the same thing would apply to multiple subs right, if one placed on a midpoint has peaks and nulls due to reflected energy then two at opposing midpoints will have bigger peaks and nulls.
Only for even order axial modes, not for odd order. Please review the diagrams again, or an acoustics textbook.

Look at his plots again, The flattest responses were from randomly placed woofers, all of the others had max-min of ~30 dB if you look at the graphs, or ~17dB if you read his text. It would have been nice if he did some randomly placed woofers and eliminated the against the wall restriction just for completeness. I am not knocking his good work, but with too many restrictions you limit the output and the conclusions that can be drawn from the data
I don't think you read the paper very closely. While the best result simulated was for randomly placed subwoofers, it was also for 50 to 5000 subwoofers. Interesting for understanding, but obviously not practical. The best result with a practical number of subs was actually placing them into the room at the 1/4th points. This puts them at the nulls of the first even order axial mode and produces cancelation of the first odd order.

'm no expert but I would say that exciting modal resonance is only one half of the equation, the frequency response is what happens first, the resonance has to build up and that can be damped.
Evenness of response across the room is the topic of this thread and also the work I'm quoting. The primary FR of the subwoofer is largely irrelevant to the discussion. Dampening bass modes in rooms requires large traps, typically of size roughly 1/4th the wavelength of interest. They're also quite hard to match to the room. Avoiding some modes and equalizing the average of what's left is a much more practical approach.
 

nunayafb

Member
2005-08-17 11:05 pm
Ok, lets go to the master handbook of acoustics by F. Alton Everest then:
The resonance nulls have been sketched because their location is definite, but between any two nulls of a given axial mode, a peak exists. Although nulls are capable of removing a sizable chunk of spectrum, the low-frequency acoustics of the room is dominated by the wide, relatively flat peaks. pg 405,406

The modes that have nulls at a loudspeaker location cannot be energized, but those having partial or full maxima at this location will be energized proportionally. The interaction of low-frequency resonances in the listening room at the loudspeaker and listening positions is too complex and transient to grasp fully, but they can be understood if broken down into the contributions of individual modes. Loudspeakers should be located as far away from reflecting surfaces as practicable. Loudspeaker positions should be considered tentative, moving them slightly if necessary to improve sound quality. The same is true for listening position. pg 406

The sketched nulls referred to are the same as from the article, which only represent the 1st 2nd and 3rd axial modes.

Welti's article only deals with subs placed along the wall, Everest says speakers should be placed far away from reflective surfaces, does that constraint affect his results?

Let me ask you all this: if a speaker is placed somewhere in a room it will excite certain modes yes? If we say it is at a midpoint then we can all agree it will be at a null for odd order axial modes and not for other modes right? So if we put another sub on an opposite wall at the midpoint we are going to have the same nulls and will be reinforcing the same peaks right? If you stagger the woofers, you can partially excite or fully excite the 1st axial mode but not the others that sub 1 excited.

My point is that when you factor in axial, tangential and oblique modes along with room furniture, wall absorptions, stairways! and whatever else is going on in the room you cant just look at 2 or 3 axial modes and call it good.

Welti's article dealt with symmetrically placed woofers along the wall and he got rough responses, when he randomly placed the woofers (yes I read how many he used) he got nearly perfect responses, why didn't he let Matlab optimize the 4 subs anywhere in the xy plane and not just along the walls? Wouldn't that have been a more thorough representation that may have answered the question of the benefits of staggered placement?

The best result with a practical number of subs was actually placing them into the room at the 1/4th points. This puts them at the nulls of the first even order axial mode and produces cancellation of the first odd order.

Placed at 1/4 points, where did this rule of thumb come from, I didn't see any tests to validate this? Obviously if you look at only the the first odd and even mode you can see 1/4-1/4 works, but only the first even and odd order modes, what about everything else. The room response is too complex to be boiled down to a handful of analytically derived modes.


In addition room response between modes is ignored (Welti)
Why, what basis exists for saying these modes are not important.

This is good discussion lets keep it going

-J
 
Let me ask you all this: if a speaker is placed somewhere in a room it will excite certain modes yes? If we say it is at a midpoint then we can all agree it will be at a null for odd order axial modes and not for other modes right? So if we put another sub on an opposite wall at the midpoint we are going to have the same nulls and will be reinforcing the same peaks right? If you stagger the woofers, you can partially excite or fully excite the 1st axial mode but not the others that sub 1 excited.
You seem to be missing that each cancels certain axial modes for the other in the symmetric pair.

Welti's article dealt with symmetrically placed woofers along the wall and he got rough responses, when he randomly placed the woofers (yes I read how many he used) he got nearly perfect responses, why didn't he let Matlab optimize the 4 subs anywhere in the xy plane and not just along the walls? Wouldn't that have been a more thorough representation that may have answered the question of the benefits of staggered placement?
They did, in part. As I said before, the actual best result from their research is placing 4 subwoofers at the intersections of lines parallel to each wall and 1/4 the distance perpendicular.

They focused on placement along walls for two reasons: firstly because they're practical, secondly because they assume dsp correction for at least the average response will be used, so they're after consistency, not flat FR in general.

Placed at 1/4 points, where did this rule of thumb come from, I didn't see any tests to validate this? Obviously if you look at only the the first odd and even mode you can see 1/4-1/4 works, but only the first even and odd order modes, what about everything else. The room response is too complex to be boiled down to a handful of analytically derived modes.
I thought it was in the linked article, but it may be in one of their other publications.

My point is that when you factor in axial, tangential and oblique modes along with room furniture, wall absorptions, stairways! and whatever else is going on in the room you cant just look at 2 or 3 axial modes and call it good.
Yes, but there are mitigating factors. First, there's very little we can do about tangental and oblique modes without very large treatments. Second, because these modes involve more bounces (oblique in particular) they have considerably lower energy than the axial modes.

Also, it's not just the primary modes, it's also multiples from there up.

Why, what basis exists for saying these modes are not important.
I'm not sure I understand what you mean. They looked at all modes. It's true that the placements only affect improvements in axial modes, but they simulated all modes.

In any case, the proof is in the pudding, and I think the simulations agreement to measurement speaks for itself. I'm highly skeptical that randomized or asymmetric placement of a small number of subs would be anywhere near the same performance, but we won't know unless someone duplicates the first result and then simulates the second. I plan to do this, as well as look at dipoles, but as I said, I need to learn some more Mathematica. I do not that all placements they simulated with an odd number of subwoofers preformed considerably worse.

Also, I think Everest's characterization is accurate, but also entirely useless. Saying "it's hard, be willing to move stuff around a bit" is punting on trying to understand and derive principles for concrete improvement. Also, I suspect the advice about "far from reflecting surfaces" is in context of rooms larger than we're talking about. We're talking about wavelengths of 20 to 2 meters... there really isn't a way to get "far" from a wall at low frequencies in most homes. Or maybe your home has much bigger rooms than my downtown apartment :p.

I agree, this is interesting discussion. It's made me more eager to get to the point where I can crunch some numbers.
 

nunayafb

Member
2005-08-17 11:05 pm
You are making some good points, however I am not convinced based on them and this study that the four midpoint placement rule or that the 1/4-1/4 is the best, it may be it may not, like you I plan on doing some room testing in the not too distant future, until then we are all just speculating based on theory. I just have trouble believing that a complex sound field can be helped by a simple source. His tests showed that the more complex the source (increasing the number of subs) the smoother the response. What I want to know is would non symmetric arrangements converge to a smooth response faster than a symmetric arrangement.

First, there's very little we can do about tangential and oblique modes without very large treatments. Second, because these modes involve more bounces (oblique in particular) they have considerably lower energy than the axial modes.
It should be noted however that oblique and tangential waves have a lower angle of incidence and therefore transfer less energy to the wall, reducing losses, basically they decay less. (Also from Everest's book)

So we have gone back and forth, I am curious if xplod1236 is still with us and if he is willing to do some leg work for us:D Maybe start with the 1/4-1/4, then mid point and finally try something more random if you have the time. Do you have any measuring equipment?
 
His tests showed that the more complex the source (increasing the number of subs) the smoother the response.
While that's true for the simulations of large numbers of random subs, it's not true for the smaller numbers at the practical locations. There there wasn't any direct correlation between number of subs and quality: 4 was the best found, and 2 very close to it. Both of these were better than any of the simulations done for placements of 5 to 18 subwoofers.
 
sub placement

I agree with jason's analysis... and after reading the Harman paper in depth a few weeks ago, it confirms what I empirically determined with 2 sonotube 12" ers over ~ 2 years in a 25' x 30' cathedral 2 story room with open walls, etc.

Corner placement SUCKS big time... all boom... huge 30 Hz modes resonating ... not good.

The best placement for the two subs was indeed midway along the longer wall about 2 feet in (as near to as furniture, etc. allows).

SONOLOCS.jpg


Tried just about all other locations... none sounded as good as far as even bass and fairly uniform coverage (especially in the listening position) The worst was the infamous "place the subs where you sit, walk around the room totill you hear the most bass, that's the place) totally wrong...

John L.
 
Yes, I'm still with you all. School has kept me busy, but starting tomorrow, I will have all the time I want. First of all, I will have to look up and become familiar with some terms, such as axial, oblique, tangential, even order, and odd order modes. But in the meantime, I have the four 12 inch subs, so whatever you want me to do, I will do. For measuring equipment, I have a digital radio shack SPL meter, and 2 cheap computer microphones. Software wise, I can generate any tones/noise, and I have trueTRA for viewing the response.
 
measurements

I wouldn't get too excited about measuring the modal behavior, unless you're into that sort of thing. I tried using the ratshack meter, half decent mikes, software, etc. and it really didn't tell me anymore than I could hear with my own ears. Mike response is notoriously inaccurate at the low frequencies anyway; you can really only make relative measurements unless you buy expensive externally calibrated mikes.

I found the "fine tuning" of the subs, once in the best position, was quite a bit less critical than when located in the primary mode locations (corners, mainly) so there is some flexibility along the walls. With four subs, you might also try two side by side along each of the longer walls, or perhaps at the 1/3 and 2/3 positions along the walls, rather than midway along all four walls.

The idea (as far as I've experienced, is to minimize driving the main bass modes ( wall to wall, slant corner to slant corner,corner to corner, etc.) of the room. With two 12" sonosubs, I'm able to run my system flat (no equalization) in my largish space, and get even bass down below 20 hz with clean response up to maybe 108 dB or so.

hope this helps...

John L.