Subwoofer / Low Frequency Optimization
By Amir Majidimehr
[Note: This article was published in the May/June 2012 issue of Widescreen Review Magazine]
Do you remember the scene in the movie Stargate where Daniel Jackson decodes the symbols that allow them to open the Stargate to travel to distant galaxies? And how once the gate opens, it generates such realistic reverberations that the coffee cup falls off the table next to the scientist with a pencil in his mouth? That is what this article is about. How to get superb bass performance in your home theater/music listening room and the science that unlocks it.
But first, let me tell you a side story. I was having a forum discussion with someone in the acoustics industry and he lamented the fact that so few people try to address the acoustics of their listening rooms. I think the major problem there is the fact that there is no clear, integrated story on what people need to do in their rooms. We have endless series of conflicting advice or complicated mathematical explanations as to the optimal solution in our listening rooms, causing people to do nothing or do things that may not solve the real problems (or worse yet, make them worse). To wit, I recently did a survey of our forum (whatstbestforum.com) to see what percentage of our members had acoustic treatment. A surprisingly high 67% had visible acoustic products in their listening rooms. That was the good news. The bad news was that solutions were all over the map. Clearly there are not dozens of answers to the same question.
Fortunately as with the Stargate movie, this story has a good ending. There is indeed a logical approach to this field that is heavily rooted in the science of acoustics and how humans perceive audio. We have an equivalent to Daniel Jackson although it is not a single person but a handful of researchers working at one company: Harman International (parent company of JBL, Revel, Lexicon, and Mark Levinson among others). Through their research we now have a great roadmap on how to approach the acoustic problems in our rooms. Yes, there is an implicit plug for the company that funded this research but I hope by the time we finish this article, you will see that the techniques (for the most part) work independent of their commercial products and that a bit of implicit praise is well deserved.
Why Low Frequency Optimization
In this article, I am going to focus on the low frequency/bass performance of a home theater or listening room. Why the low frequencies? Because that is the region where the room modifies the sound of your speakers the most. One of the key principals of room acoustics is the fact that speakers operate in three regions of interest: low frequencies, transition frequencies, and everything above that. For the typical home environment, the transition frequency is in the 200 to 400 Hz. Below this range, namely the tones that define the bass and impact of movies and music, the room heavily influences the output of the speaker as you will see shortly. Above the transition frequency, the speaker literally sets the tone. In transition area, both the speaker and room matter.
Another important factor is the fact low frequency response is very position dependent meaning as you move from seat to seat you can experience completely different levels of bass. This becomes a significant problem in home theater applications where we have multiple people enjoying the movie at once and would ideally like them to all hear the same, high-quality sound.
Instead of continuing to talk about the theory, let’s look at the actual measurement of a speaker in a room (data courtesy of Harman International/Allan Devantier):
Both of the problems I talked about are apparent here. As frequencies get down to the transition area and below, our smooth frequency response suffers and becomes quite erratic. And depending on the seat (as indicated by colors of the graph), we get different response. Comparatively speaking, the frequency response is much smoother and hardly changes due to position as we move above the transition frequencies. As the graph indicates in that region the speaker is far more in control than the room.
We measure the loudness of our system using sound pressure level and here we have swings from 68 dB to 91 dB for a difference of 23 dB (decibels). To put this in perspective, we perceive a sound twice as loud when it is 10 dB higher. This means that our room has variations of more than 4:1 in its perceived low frequency loudness! Would you buy an amplifier that had that kind of variation in its frequency response? Yet that is the performance you get from this speaker as soon as it is placed in the room. And a different speaker would not do much better. The room is in control and is invariant in either case.
Next, let’s compare the levels at 30 Hz for seat 1 in red to seat 3 in blue (one notch past 20 Hz on the left). The level for seat 1 is about 83 dB or 15 dB higher than seat 3. Whoever is sitting in seat 1 is getting far more bass in that frequency relative to the person sitting in seat 3.
When we raise the frequency to say, 3 KHz, we see that the seat to seat variation disappears and so do the wild swings. What determines good performance there is the speaker and the quality of the sound that radiates around it. That is a topic for a future article. For now, let’s say that to get the response above the transition frequency right, you need to buy a well-designed speaker. Room treatment or electronic equalization cannot fix major issues there.
Back to the low frequencies, the reason we are getting such massive changes in our amplitude is that as the sound reflects from the walls, it proceeds in an “orderly” manner to combine with the direct sound of the speaker. This creates so called "room modes" where some frequencies are exaggerated while others are attenuated. This is what you are seeing in the measurement above. At frequencies above the transition the reflections become numerous and complex and the effect is very different.
At this point, the automatic reaction from many is that some acoustic material needs to be put on the walls to fix the problem. But that is not where we are going to go. Yes, acoustic material can help but at low frequencies, the options are rather limited. Most of these products lose effectiveness as frequencies go down below the transition area. Worse yet, some can act as acoustic filters, degrading the sound of a good speaker. I will cover the use of acoustic products in a future installment. For now, instead of trying to fix a bad situation, a much more effective technique is to change the nature of the problem itself. This is done through use of multiple subwoofers instead of the typical one (or none as in the case of 2-channel music).
In a nutshell, using multiple subwoofers lets us cancel some of the room modes and with it, improve the smoothness of our frequency response. The learning in this area came from research by Harman R&D in a paper published at Audio Engineering Society conference titled,How Many Subwoofers are Enough, written by Todd Welti. The paper outlines many permutations and configurations of subwoofers using computer simulations and in-room measurements. From that analysis a number of optimal settings surface. Among them is a four subwoofer configuration, placing one in the center of each wall (assuming a rectangular room). Another useful variation is four subwoofers in the corners of the room. And two in the center of each opposing wall.
We can demonstrate the effectiveness of this approach using real-life measurements. First, let’s look at what happens with one subwoofer in the left corner, measured across six seating positions (and their average which we can ignore here). To see the problem better, let’s focus on low frequencies ranging from 20 Hz to 80 Hz:
Amazing difference, no? As a side note, this doesn’t have to be expensive as you can use four lower cost subwoofers instead of a single larger subwoofer. No matter how good that one subwoofer, it cannot overcome the problems of room acoustics.
But we can do even better. Technology developed by Harman called Sound Field Management or SFM adjusts the delay, level and frequency response of each subwoofer until the best and smoothest sound is achieved across the seating positions. Since the optimization uses real measurements of the system in the listening room, it brings the bonus of compensating for differences in subwoofers so they need not be identical. SFM can also compensate for the fact that the room may not be perfectly symmetrical (an assumption in the multiple subwoofer placement research).
SFM is part of the “room equalization” logic of the JBL Synthesis SDEC-4500 digital audio processor. Here is how the output looks with SFM applied:
We see that the graphs hug each other closer yet, showing reduction in seat to seat variation. To see that better, let's change the vertical scale to seat-to-seat differential:
As we see the deviation is now quite small between the seats.
The system still has a hump (peak) in the region of 40 to 60 Hz. Fortunately too much amplitude is easy to fix as we just need to take the levels down at those frequencies. To perform this we need a high performance electronic equalizer (EQ). The EQ needs to have programmable parameters to set arbitrary frequency and narrow bandwidth. This is important as the deviations below transition frequency are often just a few hertz wide so our EQ needs to match the same. This requirement rules out the typical “graphic” EQ that comes in our audio equipment as they usually have bandwidth of hundreds of hertz. The capability we need comes in the form of “parametric” EQ which exists in either stand-alone professional audio processors or part of the automatic “Room EQ” or “subwoofer EQ” that comes with many audio products.
For our example, we are going to continue to use the JBL Synthesis SDEC-4500. This processor is able to accept up to 12 input channels and process and output up to 20. We need more output channels than input to accommodate multiple subwoofers, each optimized separately per SFM. Before we apply the EQ, we need to measure the response at all the seating positions to make sure that if we fix it in one place we don’t make it worse somewhere else. The (dealer) measurement kit for the SDEC-4500 utilizes 8 calibrated microphones which makes this task quite easy. Let’s put it in action and apply it to one of our showroom theaters:
The faint curves are the measurements before equalization (one is the original, and the other, after a level adjustment to make it easier to dial out the minor dip). We see that the “before” response ranges from 83 dB to 93 dB. Post EQ (the bright solid white curve), the variations now range from 90 to 93 db. So we have reduced the variation by a huge 7 dB. In parlance of speaker performance in a room, 3 dB variation is considered quite good!
One of the nice things about the SDEC-4500 and its calibration software called ARCOS is that it is fully transparent in showing you what it does. Per above, you not only see what the system wanted to do (called a “target curve” displayed as dashed line) but whether it accomplished that. Common “room EQ” systems lack any displays and others that do, only show the pre-optimization measurement and not whether they actually fixed anything. Indeed, some do not. In this study of effectiveness of Room EQ systems, the ARCOS system used in JBL Synthesis SDEC-4500 readily improved the system above doing no EQ whereas two popular systems as you see on the right (#1 and #2 are ARCOS with the former optimized for one seat and the other, for multiple).
Back to the ARCOS display, there is a flat line at the bottom with a set of points and curves around them. These are individual filters that are programmed to smooth the system response. The calibration engineer can at any time override the automatic generation of these filters and modify them at will, by adjusting their characteristics. Additional filters can be added to further manipulate the response.
Another great thing about ARCOS is that all the adjustments are “live” meaning you can instantly hear any changes you make to the equalization curve. Traditional automatic EQ systems take many seconds if not minutes to do their job, making this type of hand tuning rather time consuming if not impractical. As listening is always the final arbiter of what is right, having effective tools like this is critical in achieving the best possible quality.
It is important to note that we left electronic EQ to the last step, not first. It was critical to deploy multiple subwoofers and SFM to smooth out the frequency response across all the seats. Only then can EQ effectively do its job as any changes it makes will equally apply to all seats. Without it, changing the response for one person could very well make it much worse for the other people in the room. This is why an automatic room EQ by itself is rarely this effective.
Advanced Subwoofer Room Modeling
As good as our techniques have been so far, they rely on certain assumptions that may not be right in our listening room/home theater. For example, if the room is not a rectangle, our rules of thumb for subwoofer placement no longer hold. Same occurs if the room construction is not symmetrical (e.g. a door on one wall and not the other). Even simpler things like soffits in our ceiling could make the placement non-optimal. While SFM can compensate for a lot of these variations, it cannot physically move the subwoofers to a better and more optimal location.
The answer to this problem is to use computer modeling to analyze the room and its acoustic impact and use that data to identify the best number and possible location of subwoofers. The fancy term for this is Computational Fluid Dynamics (CFD). CFD is a method of analyzing fluids as they move using computer modeling. Since air is a type of "fluid," we can use this modeling technique to determine what happens when the subwoofers energize the room.
While CFD has been around for a long time and is a critical part of design in many industries from cars to airplanes, its application to audio for subwoofer optimization is rather unique. This work was pioneered by Keith Yates design group and represents the state of the art in room acoustics optimization today.
The nice thing about modeling the room and subwoofers in a computer is that it is effort-free. We can let the computer run for days if needed to find the most optimal configuration for us. In the case of our reference home theater over 30,000 potential combinations of subwoofer and location were simulated to identify the optimal solution we used. Imagine moving the heavy subwoofers in your room 30,000 times to find the best possibly location for them! Then consider that one of the subwoofers is actually on our ceiling and you see that what the computer is doing simply is not practical any other way.
The simulation is cool in other ways in that it not only generates the results we want, but provides highly educational interactive videos of its output. This is in the form of 3-D “volumetric visualization” that shows the frequency response at any point in the room (very useful in figuring out where to put the seats). Here is a single frame of that video showing what happens at 17.2 Hz.
On the left you see the simulation for a single corner subwoofer in the left rear corner as indicated by the red color. While on the right you see our theater with the optimal configuration of three subwoofers. The color coding shows the sound pressure level with blue indicating lower levels and red the highest. As with our measurements before, we see wild fluctuations in sound pressure in the room depending on the seating location (as much as 30 dB difference).
As pretty as this display is, it can be “too much” data in that we are only interested in the system performance where people are sitting. We can solve that by only showing the response at ear level of listeners rather than the entire room (I have changed the frequency up to 30 Hz to show another data point):
The filled horizontal circles in the room are the heads of audience members. The circles on the wall/ceiling are the subwoofers as before. Now a much clearer picture emerges. In the one subwoofer configuration to the left, the large differences in seat to seat response is visible as indicated by the strong shift from the color blue in the center to red in the others. The “sweet spot” in the center is getting much lower levels of bass at 30 Hz than just two seats away.
Now let’s look at the three subwoofer configuration on the right. We immediately see massive improvement in the way all the audience heads have roughly the same sound pressure as indicated by similar shade of yellow-green color (and light orange for the group behind the bar). Just as the physics predicted in Harman research, multiple subwoofers improved the situation immensely.
Keith Yates’ optimization service is now available decoupled from his general theater design through JBL Synthesis dealers in a program called FLO. Your dealer will gather the measurements of your room and work with you to identify all the possible places for subwoofer and the maximum number you would like to potentially deploy. A 3-D model is then created which after sign off will be the basis of the CFD analysis. Top two or three configurations are then furnished as the results of the analysis.
What does all of this sound like when you are done? I suggest going to listen to a system like ours as words do not do it justice. You will hear a type of bass you are not used to. It will be tight without any of the boominess of a typical home theater. It is highly dynamic but not “loud.” Non-intuitively, cleaning up the bass allows the rest of the response to improve perceptually. Get rid of that 10 dB peak in low frequencies and you will hear detail in vocals and higher frequencies that were being masked. The improvement is not subtle at all.
Coming full circle, notice that throughout this article I have avoided any mention of “art” or any kind of guesswork. Instead we are using precise science of what makes good audio to solve our acoustic problems (which are huge if left alone). The principals and solutions are simple yet powerful in the way they tame wild fluctuations in our system response resulting in a level of audio performance that few people have experienced.
I would like to sincerely thank the Harman researchers Allan Devantier, Todd Welti, and Dr. Floyd Tool for their kind permission to use their measurements and their teachings in this article. Special thanks also go to Keith Yates for his wonderful insight and innovations in this space.
Further Reading
Presentation on video for audiophiles.
External References
"Sound Reproduction: The Acoustics and Psychoacoustics of Loudspeakers and Rooms," Dr. Floyd Toole: Amazon link
"The Subjective and Objective Evaluation of Room Correction Products,"Sean E. Olive, John Jackson, Allan Devantier, David Hunt, Harman International, 127th Convention, Audio Eng. Soc. Preprint 7960: http://seanolive.blogspot.com/2009/11/subjective-and-objective-evaluation-of.html
“How Many Subwoofers Are Enough,” Todd Welti, 112th Convention, Audio Eng. Soc. Preprint 5602.
"In-Room Low Frequency Optimization,"Todd Welti, Allan Devantier, 115th Convention, Audio Eng. Soc. Preprint 5942
Keith Yates Design Group: http://keithyates.com
JBL Synthesis SDEC-4500 Product Page: http://www.jblsynthesis.com/Products/Details/120
By Amir Majidimehr
[Note: This article was published in the May/June 2012 issue of Widescreen Review Magazine]
Do you remember the scene in the movie Stargate where Daniel Jackson decodes the symbols that allow them to open the Stargate to travel to distant galaxies? And how once the gate opens, it generates such realistic reverberations that the coffee cup falls off the table next to the scientist with a pencil in his mouth? That is what this article is about. How to get superb bass performance in your home theater/music listening room and the science that unlocks it.
But first, let me tell you a side story. I was having a forum discussion with someone in the acoustics industry and he lamented the fact that so few people try to address the acoustics of their listening rooms. I think the major problem there is the fact that there is no clear, integrated story on what people need to do in their rooms. We have endless series of conflicting advice or complicated mathematical explanations as to the optimal solution in our listening rooms, causing people to do nothing or do things that may not solve the real problems (or worse yet, make them worse). To wit, I recently did a survey of our forum (whatstbestforum.com) to see what percentage of our members had acoustic treatment. A surprisingly high 67% had visible acoustic products in their listening rooms. That was the good news. The bad news was that solutions were all over the map. Clearly there are not dozens of answers to the same question.
Fortunately as with the Stargate movie, this story has a good ending. There is indeed a logical approach to this field that is heavily rooted in the science of acoustics and how humans perceive audio. We have an equivalent to Daniel Jackson although it is not a single person but a handful of researchers working at one company: Harman International (parent company of JBL, Revel, Lexicon, and Mark Levinson among others). Through their research we now have a great roadmap on how to approach the acoustic problems in our rooms. Yes, there is an implicit plug for the company that funded this research but I hope by the time we finish this article, you will see that the techniques (for the most part) work independent of their commercial products and that a bit of implicit praise is well deserved.
Why Low Frequency Optimization
In this article, I am going to focus on the low frequency/bass performance of a home theater or listening room. Why the low frequencies? Because that is the region where the room modifies the sound of your speakers the most. One of the key principals of room acoustics is the fact that speakers operate in three regions of interest: low frequencies, transition frequencies, and everything above that. For the typical home environment, the transition frequency is in the 200 to 400 Hz. Below this range, namely the tones that define the bass and impact of movies and music, the room heavily influences the output of the speaker as you will see shortly. Above the transition frequency, the speaker literally sets the tone. In transition area, both the speaker and room matter.
Another important factor is the fact low frequency response is very position dependent meaning as you move from seat to seat you can experience completely different levels of bass. This becomes a significant problem in home theater applications where we have multiple people enjoying the movie at once and would ideally like them to all hear the same, high-quality sound.
Instead of continuing to talk about the theory, let’s look at the actual measurement of a speaker in a room (data courtesy of Harman International/Allan Devantier):
Both of the problems I talked about are apparent here. As frequencies get down to the transition area and below, our smooth frequency response suffers and becomes quite erratic. And depending on the seat (as indicated by colors of the graph), we get different response. Comparatively speaking, the frequency response is much smoother and hardly changes due to position as we move above the transition frequencies. As the graph indicates in that region the speaker is far more in control than the room.
We measure the loudness of our system using sound pressure level and here we have swings from 68 dB to 91 dB for a difference of 23 dB (decibels). To put this in perspective, we perceive a sound twice as loud when it is 10 dB higher. This means that our room has variations of more than 4:1 in its perceived low frequency loudness! Would you buy an amplifier that had that kind of variation in its frequency response? Yet that is the performance you get from this speaker as soon as it is placed in the room. And a different speaker would not do much better. The room is in control and is invariant in either case.
Next, let’s compare the levels at 30 Hz for seat 1 in red to seat 3 in blue (one notch past 20 Hz on the left). The level for seat 1 is about 83 dB or 15 dB higher than seat 3. Whoever is sitting in seat 1 is getting far more bass in that frequency relative to the person sitting in seat 3.
When we raise the frequency to say, 3 KHz, we see that the seat to seat variation disappears and so do the wild swings. What determines good performance there is the speaker and the quality of the sound that radiates around it. That is a topic for a future article. For now, let’s say that to get the response above the transition frequency right, you need to buy a well-designed speaker. Room treatment or electronic equalization cannot fix major issues there.
Back to the low frequencies, the reason we are getting such massive changes in our amplitude is that as the sound reflects from the walls, it proceeds in an “orderly” manner to combine with the direct sound of the speaker. This creates so called "room modes" where some frequencies are exaggerated while others are attenuated. This is what you are seeing in the measurement above. At frequencies above the transition the reflections become numerous and complex and the effect is very different.
At this point, the automatic reaction from many is that some acoustic material needs to be put on the walls to fix the problem. But that is not where we are going to go. Yes, acoustic material can help but at low frequencies, the options are rather limited. Most of these products lose effectiveness as frequencies go down below the transition area. Worse yet, some can act as acoustic filters, degrading the sound of a good speaker. I will cover the use of acoustic products in a future installment. For now, instead of trying to fix a bad situation, a much more effective technique is to change the nature of the problem itself. This is done through use of multiple subwoofers instead of the typical one (or none as in the case of 2-channel music).
In a nutshell, using multiple subwoofers lets us cancel some of the room modes and with it, improve the smoothness of our frequency response. The learning in this area came from research by Harman R&D in a paper published at Audio Engineering Society conference titled,How Many Subwoofers are Enough, written by Todd Welti. The paper outlines many permutations and configurations of subwoofers using computer simulations and in-room measurements. From that analysis a number of optimal settings surface. Among them is a four subwoofer configuration, placing one in the center of each wall (assuming a rectangular room). Another useful variation is four subwoofers in the corners of the room. And two in the center of each opposing wall.
We can demonstrate the effectiveness of this approach using real-life measurements. First, let’s look at what happens with one subwoofer in the left corner, measured across six seating positions (and their average which we can ignore here). To see the problem better, let’s focus on low frequencies ranging from 20 Hz to 80 Hz:
We can see the predictable wild swings in frequency response and high seat to seat variations. Now let’s try four subwoofers in the corners:
Amazing difference, no? As a side note, this doesn’t have to be expensive as you can use four lower cost subwoofers instead of a single larger subwoofer. No matter how good that one subwoofer, it cannot overcome the problems of room acoustics.
But we can do even better. Technology developed by Harman called Sound Field Management or SFM adjusts the delay, level and frequency response of each subwoofer until the best and smoothest sound is achieved across the seating positions. Since the optimization uses real measurements of the system in the listening room, it brings the bonus of compensating for differences in subwoofers so they need not be identical. SFM can also compensate for the fact that the room may not be perfectly symmetrical (an assumption in the multiple subwoofer placement research).
SFM is part of the “room equalization” logic of the JBL Synthesis SDEC-4500 digital audio processor. Here is how the output looks with SFM applied:
We see that the graphs hug each other closer yet, showing reduction in seat to seat variation. To see that better, let's change the vertical scale to seat-to-seat differential:
As we see the deviation is now quite small between the seats.
The system still has a hump (peak) in the region of 40 to 60 Hz. Fortunately too much amplitude is easy to fix as we just need to take the levels down at those frequencies. To perform this we need a high performance electronic equalizer (EQ). The EQ needs to have programmable parameters to set arbitrary frequency and narrow bandwidth. This is important as the deviations below transition frequency are often just a few hertz wide so our EQ needs to match the same. This requirement rules out the typical “graphic” EQ that comes in our audio equipment as they usually have bandwidth of hundreds of hertz. The capability we need comes in the form of “parametric” EQ which exists in either stand-alone professional audio processors or part of the automatic “Room EQ” or “subwoofer EQ” that comes with many audio products.
For our example, we are going to continue to use the JBL Synthesis SDEC-4500. This processor is able to accept up to 12 input channels and process and output up to 20. We need more output channels than input to accommodate multiple subwoofers, each optimized separately per SFM. Before we apply the EQ, we need to measure the response at all the seating positions to make sure that if we fix it in one place we don’t make it worse somewhere else. The (dealer) measurement kit for the SDEC-4500 utilizes 8 calibrated microphones which makes this task quite easy. Let’s put it in action and apply it to one of our showroom theaters:
The faint curves are the measurements before equalization (one is the original, and the other, after a level adjustment to make it easier to dial out the minor dip). We see that the “before” response ranges from 83 dB to 93 dB. Post EQ (the bright solid white curve), the variations now range from 90 to 93 db. So we have reduced the variation by a huge 7 dB. In parlance of speaker performance in a room, 3 dB variation is considered quite good!
One of the nice things about the SDEC-4500 and its calibration software called ARCOS is that it is fully transparent in showing you what it does. Per above, you not only see what the system wanted to do (called a “target curve” displayed as dashed line) but whether it accomplished that. Common “room EQ” systems lack any displays and others that do, only show the pre-optimization measurement and not whether they actually fixed anything. Indeed, some do not. In this study of effectiveness of Room EQ systems, the ARCOS system used in JBL Synthesis SDEC-4500 readily improved the system above doing no EQ whereas two popular systems as you see on the right (#1 and #2 are ARCOS with the former optimized for one seat and the other, for multiple).
Back to the ARCOS display, there is a flat line at the bottom with a set of points and curves around them. These are individual filters that are programmed to smooth the system response. The calibration engineer can at any time override the automatic generation of these filters and modify them at will, by adjusting their characteristics. Additional filters can be added to further manipulate the response.
Another great thing about ARCOS is that all the adjustments are “live” meaning you can instantly hear any changes you make to the equalization curve. Traditional automatic EQ systems take many seconds if not minutes to do their job, making this type of hand tuning rather time consuming if not impractical. As listening is always the final arbiter of what is right, having effective tools like this is critical in achieving the best possible quality.
It is important to note that we left electronic EQ to the last step, not first. It was critical to deploy multiple subwoofers and SFM to smooth out the frequency response across all the seats. Only then can EQ effectively do its job as any changes it makes will equally apply to all seats. Without it, changing the response for one person could very well make it much worse for the other people in the room. This is why an automatic room EQ by itself is rarely this effective.
Advanced Subwoofer Room Modeling
As good as our techniques have been so far, they rely on certain assumptions that may not be right in our listening room/home theater. For example, if the room is not a rectangle, our rules of thumb for subwoofer placement no longer hold. Same occurs if the room construction is not symmetrical (e.g. a door on one wall and not the other). Even simpler things like soffits in our ceiling could make the placement non-optimal. While SFM can compensate for a lot of these variations, it cannot physically move the subwoofers to a better and more optimal location.
The answer to this problem is to use computer modeling to analyze the room and its acoustic impact and use that data to identify the best number and possible location of subwoofers. The fancy term for this is Computational Fluid Dynamics (CFD). CFD is a method of analyzing fluids as they move using computer modeling. Since air is a type of "fluid," we can use this modeling technique to determine what happens when the subwoofers energize the room.
While CFD has been around for a long time and is a critical part of design in many industries from cars to airplanes, its application to audio for subwoofer optimization is rather unique. This work was pioneered by Keith Yates design group and represents the state of the art in room acoustics optimization today.
The nice thing about modeling the room and subwoofers in a computer is that it is effort-free. We can let the computer run for days if needed to find the most optimal configuration for us. In the case of our reference home theater over 30,000 potential combinations of subwoofer and location were simulated to identify the optimal solution we used. Imagine moving the heavy subwoofers in your room 30,000 times to find the best possibly location for them! Then consider that one of the subwoofers is actually on our ceiling and you see that what the computer is doing simply is not practical any other way.
The simulation is cool in other ways in that it not only generates the results we want, but provides highly educational interactive videos of its output. This is in the form of 3-D “volumetric visualization” that shows the frequency response at any point in the room (very useful in figuring out where to put the seats). Here is a single frame of that video showing what happens at 17.2 Hz.
On the left you see the simulation for a single corner subwoofer in the left rear corner as indicated by the red color. While on the right you see our theater with the optimal configuration of three subwoofers. The color coding shows the sound pressure level with blue indicating lower levels and red the highest. As with our measurements before, we see wild fluctuations in sound pressure in the room depending on the seating location (as much as 30 dB difference).
As pretty as this display is, it can be “too much” data in that we are only interested in the system performance where people are sitting. We can solve that by only showing the response at ear level of listeners rather than the entire room (I have changed the frequency up to 30 Hz to show another data point):
The filled horizontal circles in the room are the heads of audience members. The circles on the wall/ceiling are the subwoofers as before. Now a much clearer picture emerges. In the one subwoofer configuration to the left, the large differences in seat to seat response is visible as indicated by the strong shift from the color blue in the center to red in the others. The “sweet spot” in the center is getting much lower levels of bass at 30 Hz than just two seats away.
Now let’s look at the three subwoofer configuration on the right. We immediately see massive improvement in the way all the audience heads have roughly the same sound pressure as indicated by similar shade of yellow-green color (and light orange for the group behind the bar). Just as the physics predicted in Harman research, multiple subwoofers improved the situation immensely.
Keith Yates’ optimization service is now available decoupled from his general theater design through JBL Synthesis dealers in a program called FLO. Your dealer will gather the measurements of your room and work with you to identify all the possible places for subwoofer and the maximum number you would like to potentially deploy. A 3-D model is then created which after sign off will be the basis of the CFD analysis. Top two or three configurations are then furnished as the results of the analysis.
What does all of this sound like when you are done? I suggest going to listen to a system like ours as words do not do it justice. You will hear a type of bass you are not used to. It will be tight without any of the boominess of a typical home theater. It is highly dynamic but not “loud.” Non-intuitively, cleaning up the bass allows the rest of the response to improve perceptually. Get rid of that 10 dB peak in low frequencies and you will hear detail in vocals and higher frequencies that were being masked. The improvement is not subtle at all.
Coming full circle, notice that throughout this article I have avoided any mention of “art” or any kind of guesswork. Instead we are using precise science of what makes good audio to solve our acoustic problems (which are huge if left alone). The principals and solutions are simple yet powerful in the way they tame wild fluctuations in our system response resulting in a level of audio performance that few people have experienced.
I would like to sincerely thank the Harman researchers Allan Devantier, Todd Welti, and Dr. Floyd Tool for their kind permission to use their measurements and their teachings in this article. Special thanks also go to Keith Yates for his wonderful insight and innovations in this space.
Further Reading
Presentation on video for audiophiles.
External References
"Sound Reproduction: The Acoustics and Psychoacoustics of Loudspeakers and Rooms," Dr. Floyd Toole: Amazon link
"The Subjective and Objective Evaluation of Room Correction Products,"Sean E. Olive, John Jackson, Allan Devantier, David Hunt, Harman International, 127th Convention, Audio Eng. Soc. Preprint 7960: http://seanolive.blogspot.com/2009/11/subjective-and-objective-evaluation-of.html
“How Many Subwoofers Are Enough,” Todd Welti, 112th Convention, Audio Eng. Soc. Preprint 5602.
"In-Room Low Frequency Optimization,"Todd Welti, Allan Devantier, 115th Convention, Audio Eng. Soc. Preprint 5942
Keith Yates Design Group: http://keithyates.com
JBL Synthesis SDEC-4500 Product Page: http://www.jblsynthesis.com/Products/Details/120
Last edited:
