[youngho]

Howdy,

Hope all are well. For fun, I'd been idly considering the problem of square rooms. It seemed to me this morning that the Harman paper may present several solutions:

1. Square or cubic rooms with fairly rigid walls should present a set of highly predictable modes.
2. The height mode of typical rooms should be obviated by typical seating position heights (say, 36" in a 8' room)
3. The mid-wall placement of two subwoofers on opposing walls, say front and rear, should prevent excitation of all odd-order modes corresponding to the width modes
4. The opposing mid-wall placement should result in mode cancellation for the length modes (all or only odd-order? I forget)
5. A few bands of parametric equalization should correct the even-order width modes, which should theoretically be minimum phase phenomena, and should prevent problems with ringing. Any remaining bands could deal with relevant height modes.

This solution should theoretically give fairly even bass response over a reasonable listening position, except perhaps at even-order mode frequencies as one moves laterally from a central listening position

How about if one set up the room in a diagonal home theater arrangement? This should theoretically help in terms of reflections, but subwoofer setup would benefit from a four-subwoofer arrangement. This would give the maximum benefits in terms of avoidance of odd-order mode excitation, as well as mode cancellation. Theoretically, this should be a great setup with easily arrangeable channel speakers that could be wall-mountable with appropriate phase or group delay, right?

Clearly, some broadband and reflection point acoustic treatment would be needed, but the highly symmetric nature of the room might make it easy to place, no?

Young-Ho

[Ethan Winer]

Young-Ho,

> 2. The height mode of typical rooms should be obviated by typical seating position heights (say, 36" in a 8' room) <

Not really. For any given height you still have numerous peaks and severe nulls. The ideal height may be flatter than a less-ideal height, but the response is still badly skewed. And of course you'll still have modal ringing to deal with.

> 5. A few bands of parametric equalization should correct the even-order width modes, which should theoretically be minimum phase phenomena, and should prevent problems with ringing. <

EQ might help the raw response a little below 100 Hz, but there is no way it can reduce ringing. The decay times will be exactly the same, just at a lower level.

> How about if one set up the room in a diagonal <

The advantage of facing a corner is the side walls splay enough that you don't need to treat the first reflections on those walls. But this creates an even bigger problem in most rectangular rooms: The corner that's now behind you focuses all sound directly back at you. You could reduce that by treating everything back there with absorption, but that probably requires even more absorption than the traditional orientation.

--Ethan

[youngho]

Ethan, thanks for your response. Again, this was just an idle thought as I was shampooing my hair, and I appreciate your practical experience.

In terms of the height modes, the seated listening position would probably avoid the nulls and peaks for the first, second, and third modes, depending on the actual height of the room. The square or cubic room would benefit from ceiling treatment, but many rooms would.

In Floyd Toole's second white paper on loudspeakers and rooms, he states that room resonances at low frequencies exhibit minimum phase behavior and that fixing the frequency response via parametric equalization should also have a corrective effect in the time domain. This explains the dramatic improvement in both frequency and impulse response from one of the real world examples in the third white paper where  parametric equalization was applied. For the fixable low frequency resonances, this should fix problems with ringing, then, right? It wouldn't help higher frequency resonances, though, as you point out, so some absorption would still be needed.

My thought in the diagonal setup was that the rear corner would have a significant amount of fiberglass absorption, which would help with the corner reflections and resonances. In fact, by treating all three front channel speakers and acting as a bass trap to boot, wouldn't a sufficiently wide and deep single absorber be more efficient than three separate absorption panels plus bass trap?

Let me put it a different way--wouldn't the Harman paper's suggested subwoofer placement be a relatively elegant solution to the square or cubic room problem? I'm not saying that everyone should go and build square rooms. Instead, I'm wondering if one has a square or cubic room, what should one do? The other answers would seem to be parametric equalization (which wouldn't help the nulls), TONS of absorption, and/or Helmholtz resonators. The additional subwoofer(s) and appropriate placement would seem to reduce a lot of the acoustic problems, unless I'm more clueless than usual.

Thanks for your thoughts

[Carlman]

My thought in the diagonal setup was that the rear corner would have a significant amount of fiberglass absorption, which would help with the corner reflections and resonances. In fact, by treating all three front channel speakers and acting as a bass trap to boot, wouldn't a sufficiently wide and deep single absorber be more efficient than three separate absorption panels plus bass trap?
...

I know nothing of the papers/research you speak but I found that doing a rectangular room at a diagonal is a nightmare to deal with... twice.  It's impossible to tame to the jarbled information reflecting back into the listening area... it gets harsh and fatiguing fast.  At first you think, hey wow.. this sounds cool and dreamy... but then you crank it up a little and the bass is sucked out and the mids get blaring.... Treatments did nothing for me.  I needed too many to be acceptable stylistically.

I like wider over deeper rooms in my experience.  But, I reserve the right to change my mind if I hear a deeper room that sounds great.  Which I know it can...

I have almost a square room.  9x11x13 (HxWxD)  It sucks.  I'd like to make that room bigger.  Who knows... if I live there long enough, maybe I will.  It's cheaper than a system upgrade...

[Ethan Winer]

Young-Ho,

> In terms of the height modes, the seated listening position would probably avoid the nulls and peaks for the first, second, and third modes <

But what about all the other, non-modal frequencies? No matter the room dimensions or where you are in the room, there are still numerous peaks and nulls all over the place caused by simple boundary interference - comb filtering.

> Floyd Toole ... states that room resonances at low frequencies exhibit minimum phase behavior and that fixing the frequency response via parametric equalization should also have a corrective effect in the time domain. <

Perhaps in theory, but in practical terms he is wrong. Maybe if you could exactly match the required boost or cut, and the bandwidth, and the phase shift, you could reduce ringing for a location 1 inch cubed. But as soon as you move your head even one inch all the ringing will be right back again.

> This explains the dramatic improvement in both frequency and impulse response from one of the real world examples in the third white paper where parametric equalization was applied. <

Again, only for an area one inch cubed. And I bet it took him an hour to tweak the EQ to achieve even that.

> It wouldn't help higher frequency resonances <

Actually, it could, assuming the same caveats. There are some very sophisticated DSP algorithms available, but they all suffer the same fatal flaw: The more correction you apply, the smaller the corrected area becomes.

> wouldn't a sufficiently wide and deep single absorber be more efficient than three separate absorption panels plus bass trap? <

It depends on how you define "sufficiently wide." You'd probably need it to be six feet wide or even wider, floor to ceiling.

> wouldn't the Harman paper's suggested subwoofer placement be a relatively elegant solution to the square or cubic room problem? <

Sorry, I can't see how.

> if one has a square or cubic room, what should one do? <

Get as much bass trapping as you can possibly manage.

> TONS of absorption <

Bingo!

--Ethan

[youngho]

Ethan, once again, I appreciate your responses. I'm slow, but I do learn.

But what about all the other, non-modal frequencies? No matter the room dimensions or where you are in the room, there are still numerous peaks and nulls all over the place caused by simple boundary interference - comb filtering.

Would SBIR be worse in square rooms than rectangular rooms? I certainly acknowledge the need for reflection treatment in any shaped room. I thought that the primary disadvantage of square rooms were the modal resonances. Also, the Dolby-recommended loudspeaker setup would seem to make most of the primary reflections points relatively easy to find without even needing to use a mirror.

Incidentally, my original statement was that "the height mode of typical rooms should be obviated by typical seating position heights." I should have said that "the problems associated with the heigh mode of typical rooms should be relatively minimized" instead, but I certainly didn't mean to imply that the seated position would experience a perfectly flat frequency response.

Perhaps in theory, but in practical terms he is wrong. Maybe if you could exactly match the required boost or cut, and the bandwidth, and the phase shift, you could reduce ringing for a location 1 inch cubed. But as soon as you move your head even one inch all the ringing will be right back again.

I don't understand why this is the case. Let me walk through my (limited) understanding:

1. Minimum phase behavior means that a bump in the frequency response will mean ringing in the time domain.
2. The peaks due to modes is greatest when one sits in the antinode of the mode, for example the listener sitting in the middle of the square room listening to the second-order length and width mode (boom!)
3. Parametric equalization (where one simply sets central frequency, width of curve, and amplitude) can correct the mode-related peak. Because the mode-related resonance is a minimum phase phenomena, the phase shift that automatically occurs also automatically corrects the ringing. There seems to be a new program to make it much simpler to adjust these three parameters of parametric equalization.
4. Moving away from the antinode will actually result in a relative null, rather than a peak. Nulls are not associated with ringing. However, if one were instead approaching an antinode, then you're absolutely right.

Again, only for an area one inch cubed. And I bet it took him an hour to tweak the EQ to achieve even that.

I happen have the Infinity speakers with the RABOS system, which was easy to use to make improvements that were subjectively and measurably one or two orders of magnitude better than previously. I agree that Toole's measurement and correction tools are considerably (probably also one or two orders of magnitude) more sophisticated.

It depends on how you define "sufficiently wide." You'd probably need it to be six feet wide or even wider, floor to ceiling.

Yes, but you'd need more than this absorption, anyway, to ameliorate the modal-related problems of the square room, anyway, no? I thought a lot of depth was needed to treat low frequenceis. In the diagonal setup, wouldn't this absorption consequently serve quadruple duty?

Sorry, I can't see how.

In a square room home theater, for a relatively centrally positioned listener, which is where I assume he or she would end up to be fairly equidistant from all speakers, the simple addition of a second subwoofer with opposite mid-wall positioning for both subwoofers would result in elimination of all odd-order modes related to the width and length of the room, regardless of where one sits in the room. No odd-order length and width modes, period. The addition of parametric equalization would substantially improve observed bass response, at least at the listening position--moving away from the listening position would result in relative nulls at the even-order modal frequencies, not peaks, so there should not be ringing. Absorption would still be necessary, but much less (particularly in terms of depth, since the second-order mode is necessarily higher in frequency than the first-order mode) would be needed than in the square room with only one subwoofer. At least in theory. Isn't this an improvement?

Young-Ho

[youngho]

I know nothing of the papers/research you speak but I found that doing a rectangular room at a diagonal is a nightmare to deal with... twice.  It's impossible to tame to the jarbled information reflecting back into the listening area... it gets harsh and fatiguing fast.  At first you think, hey wow.. this sounds cool and dreamy... but then you crank it up a little and the bass is sucked out and the mids get blaring.... Treatments did nothing for me.  I needed too many to be acceptable stylistically.

Thanks for posting your experiences. I thought I had seen a number of posts in the past from users with square rooms who had reported improvements with diagonal setups.

Young-Ho

[ctviggen]

Can anyone point me to a paper that says that "the simple addition of a second subwoofer with opposite mid-wall positioning for both subwoofers would result in elimination of all odd-order modes related to the width and length of the room"?  I thought these were room modes, which occurred regardless of where subs were put.

[Ethan Winer]

Young-Ho,

> Would SBIR be worse in square rooms than rectangular rooms? <

SBIR is affected entirely by the distance between the speakers (S) and the boundaries (B). So I see no relation to room shape.

> I thought that the primary disadvantage of square rooms were the modal resonances. <

Yes, that is correct.

> the Dolby-recommended loudspeaker setup would seem to make most of the primary reflections points relatively easy to find without even needing to use a mirror. <

You don't need a mirror regardless. The article How to set up a room on my company's site shows an easy way to calculate the reflection points. Though that simple formula assumes the walls and ceiling are not angled.

> Minimum phase behavior means that <

Stop right there. All gear has all kinds of phase shift, and the room only adds more. And it's not simple phase shift (one or two poles), but rather the sum of all the shifts from every piece of gear and acoustic sound path. There's no direct way to counter something that complex with an equalizer. And unless you counter it completely and precisely there's no chance you could reduce ringing.

> Parametric equalization ... can correct the mode-related peak. <

How could it? The peak changes amplitude around the room, so if you cut 12 dB to fix the prime seat, you've cut too much for a location nearby.

> you'd need more than this absorption, anyway, to ameliorate the modal-related problems of the square room, anyway, no? <

Not if you have enough absorption! It also depends on how you define "ameliorate." You'll never get a perfectly flat response in any small room. The best you can hope for is to make it much better than it was without treatment.

> the simple addition of a second subwoofer with opposite mid-wall positioning for both subwoofers would result in elimination of all odd-order modes <

But that too does nothing to reduce ringing or broaden the modal peak bandwidths. Those are at least as damaging as the skewed low end response.

> At least in theory. Isn't this an improvement? <

I'm not enough of a math guy to understand the theory at the level needed to comment intelligently. I can say that I've been doing this stuff for many years, and I've played with EQ and speaker placement and all things related. In practice, I have never seen EQ or even speaker placement be nearly as effective as broadband low frequency absorption.

--Ethan

[youngho]

Bob,

I looked at the Harman white papers at
http://www.harman.com/wp/index.jsp?articleId=122
http://www.harman.com/wp/index.jsp?articleId=1003

I can't remember if they explicitly state so, but mid-wall positioning puts a subwoofer into the node of odd-order modes corresponding to the axis represented by the wall. A subwoofer (or speaker) fails to excite standing waves for which that position represents a node, which would be those odd-order modes.

Mode cancellation puts a second subwoofer on the other side of the node, which results in cancellation of the mode. In the scenario below (centrally positioned listener in a square room), I believe that this would represent all modes for which the listening position represents a node, or all odd-order modes corresponding to the axis represented by a line drawn between the two subwoofers.

Am I incorrect in my reasoning?

Young-Ho

[Carlman]

I read opinion on the diagnol thing.. which is partly why I tried it... and at first it does sound good... but something bothered me within a few minutes... and I couldn't get it to sound accurate.  

Please experiment and let us know what you find... I'm just 1 guy that had bad results...

[youngho]

Thanks again for your response. As you point out, theory can be one thing and practice another. Again, I'm not proposing that square rooms are ideal. I'm just wondering if the Harman solution might be particularly relevant and helpful to home theater owners with square rooms, particularly if the necessary amount of low frequency absorption is not possible or acceptable.

Stop right there. All gear has all kinds of phase shift, and the room only adds more. And it's not simple phase shift (one or two poles), but rather the sum of all the shifts from every piece of gear and acoustic sound path. There's no direct way to counter something that complex with an equalizer. And unless you counter it completely and precisely there's no chance you could reduce ringing.

I'm sorry, I wasn't suggesting that parametric equalization will fix all ringing at all frequencies. I simply meant that by decreasing the amplitude of the frequencies related to the primary room modes it should also reduce ringing related to the primary room modes.

How could it? The peak changes amplitude around the room, so if you cut 12 dB to fix the prime seat, you've cut too much for a location nearby.

Absolutely. This is what I meant when I said, "Moving away from the antinode will actually result in a relative null." I agree 100%. For my part, I think I would rather have a null than a peak of equal frequency, width, and amplitude, though.

Not if you have enough absorption! It also depends on how you define "ameliorate." You'll never get a perfectly flat response in any small room. The best you can hope for is to make it much better than it was without treatment.

That is *exactly* what I would hope for. I absolutely agree that the square room would still need absorption, but it still seems to me that the amount of absorption needed would be less with the opposite mid-wall subwoofers.

I'm not enough of a math guy to understand the theory at the level needed to comment intelligently. I can say that I've been doing this stuff for many years, and I've played with EQ and speaker placement and all things related. In practice, I have never seen EQ or even speaker placement be nearly as effective as broadband low frequency absorption.

Again, I appreciate your practical experience, having none myself. Still, it seems reasonable to me to provide the following answer to the question, "I have a square room, and I'm getting really boomy bass with home theater system, especially when I sit in or near the middle of the room. What should I do?"

1. Use as much fiberglass in the corners as possible and tolerable. Depth is critical to treat those low combined length/width modes.
2. Treat primary reflection points of all speakers.
3. Purchase a second subwoofer (same model as the first) and place both subwoofers on opposite mid-walls. This could help significantly, at least in theory, particularly in rooms where absorption depth is limited by cosmetics or other reasons, by eliminating the odd-order room modes related to length/width axes.
4. Use parametric equalization for remaining mode-related peaks (particularly even-order length/width-related ones).
5. Setting up the home theater diagonally may make an additional improvement.

Step #3 is what I mean by "the Harman solution," and I suggest, may reduce, but certainly not eliminate, the need for step #1.

Young-Ho

[Ethan Winer]

Young-Ho,

> I simply meant that by decreasing the amplitude of the frequencies related to the primary room modes it should also reduce ringing related to the primary room modes. <

I realize, but I just don't see how. Even considering one mode you have the sum of several phase shifts. At minimum there's the speaker itself and the crossover. If you have an EQ in the circuit there's more sources of phase shift. With a turntable there's yet more shift for the RIAA compensation.

> I think I would rather have a null than a peak of equal frequency, width, and amplitude, though. <

Not if it loses all of the impact of every bass instrument. Many rooms have the deepest null somewhere between 80 and 120 Hz, and that's right in the middle of the fullness range.

> it still seems to me that the amount of absorption needed would be less with the opposite mid-wall subwoofers. <

Again, I can't see how adding another sub will affect anything other than the raw LF response.

> "I have a square room, and I'm getting really boomy bass with home theater system, especially when I sit in or near the middle of the room. What should I do?" <

Definitely #1, "Use as much fiberglass in the corners as possible and tolerable."

> 2. Treat primary reflection points of all speakers. <

Sure, that too. All rooms need that.

> Step #3 is what I mean by "the Harman solution," and I suggest, may reduce, but certainly not eliminate, the need for step #1. <

I guess neither of us will know for sure until you get a second sub, and measure it with ETF. If you do that, please report back.

--Ethan