Hmmm. My first post on the What’s Best forum – what to start off with? Well, I would like to discuss two methods of optimizing subwoofers in a room which have some similarities. One is Sound Field Management, which Allan Devantier and I worked on starting in 2003. The other one is a “Double Bass Array”, also known as “Controlled Acoustic Bass System (CABS)”. The two basic approaches are certainly not the only ways to optimize subwoofers in rooms, so this is not a comprehensive overview of such methods. The two methods discussed here have at least been thoroughly documented. Though there are some interesting similarities between the two, certain situations may favor the use of one over the other. That is the ultimate focus of this discussion.
In order to avoid a too-long post, I’m going to describe DBA and SFM first, and follow up with the meaty part – the comparison.
In the following discussion, we are generally assuming rectangular rooms and low frequencies.
Double Bass Array
The term Double Bass Array (DBA) was coined in a white paper by Anselm Goertz et. al for Klein + Hummel [1]. The original paper is in German, but I found an English translation at:
http://forums.klipsch.com/forums/storage/8/1485263/tmt2002_eng.pdf
The paper does not give much detail however. There is also some discussion and pictures of this technique at
http://www.avsforum.com/avs-vb/showthread.php?t=837744
The original Goertz et al paper outlined the basic idea: try to produce a plane wave going from the front to the back of the room, thus eliminating standing waves throughout the enclosed space. The speakers in the front of the room are placed on the front wall in a way that minimizes sidewall and floor/ceiling modes, leaving only the front\back modes. The speakers in the rear of the room are delayed and their polarity reversed, so that they cancel out the sound from the front. The idea is essentially to create a plane wave traveling from the front to the back of the room, then cancel the reflected wave when it reaches the back. Some earlier researchers had employed methods to create such a plane wave, but using much more complicated speaker placements, FIR filters, and many microphones. Such advanced methods work very well, but are cumbersome to say the least.
Starting in around 2005, Adrian Celestinos and Sofus Bierkedal Nielsen applied a more rigorous approach to DBA by developing the Controlled Acoustic Bass System (CABS)[2]. They systematically investigated and formalized this concept, using sophisticated models and in real rooms. They made some very practical investigations into simplified speaker layouts, error due to non-optimal layouts, and other areas. Coincidentally, Sofus was a professor of mine at Aalborg University. Small world!
The basic DBA setup normally uses 4 subs on the front wall, located ¼ of the room width in from the walls and ¼ of the room height in from the floor/ceiling. The same configuration is used on the rear wall, and of course we are assuming a rectangular room. The signal processing is simply a delay unit, set to the time-of-flight for sound from the front of the room to the rear. Obviously, inverting polarity requires no special processing.
The ideal result would be flat magnitude and linear phase anywhere within the room volume. This assumes that a perfect plane wave is produced at from both the front wall and the rear wall (i.e. the cancellation wave). The subs need to be spaced closer than about ½ wavelength at the frequency of interest to approximate a plane wave (perhaps less if the room has significant low frequency absorption). Based on the simulations and actual room measurements done by Celestinos and Nielsen, the method seems to work very well.
A few miscellaneous notes and questions on DBA… I have heard it said that “Genelec has a patent on the DBA”. Unless there is another patent that I don’t know about, the one I do know about does not look like a patent on DBA [3]. It does have a source in a room and another out of polarity and inverted source at the other, but it also uses microphones to sample the space, and I don’t see the optimized speaker layout to cancel the sidewall and floor/ceiling modes. There are also a slew of other patents that use some sort of cancellation scheme. So perhaps not so clear. I don’t currently know of any product by Genelec using this. In fact it is hard to find mention of it on their web site.
Sound Field Management. (SFM)
Sound Field Management (SFM) was developed by Allan Devantier and myself circa 2003 [4]. In SFM, individual complex transfer function measurements (i.e. impulse responses) are made from each subwoofer to each seat in the listening area. For example if there are 4 potential subwoofers locations and 4 seats, there would be 16 measurements made. Notice I said “potential subwoofer locations”. The idea is to measure more subwoofer locations than you have subwoofers. So, the SFM algorithm would predict the total acoustical response at each seat, for each of the 6 possible ways you can put 2 subs into 4 possible locations (for example). The best configuration is then chosen according to whatever your criteria is. Generally the criteria used is to maximize the seat to seat consistency, though other criteria such as efficiency and flatness of magnitude response are also calculated and can be used.
But wait, there’s more! SFM also can include a number of different possible signal delays and simple one-band filters for each subwoofer, and predict the resulting response for each combination. This is known as a brute force optimization, can blow up the number of possible configurations to be simulated to many millions, so the algorithm has to be efficient. You can tailor the optimization to use whatever you have available: extra possible sub locations and/or signal delay and/or simple filters. Obviously the result is best if you have all three. Once you have run the simulations, you have a database, from which you can select any of the top rated solutions you like. In the case of SFM, we are optimizing at a few individual seats, not the whole room, like in DBA\CABS. This is an important difference, as we shall see later.
SFM has been thoroughly validated in real rooms and works very well. At least 2 patents have been granted and the system has been implemented the Harman Audio Test System (HATS) for use in the JBL Synthesis calibration toolbox.
Comments and questions welcome, and next post will get to the meaty part: point by point comparison of the two methods.
Bye for now,
Todd Welti
Research Acoustician
Harman International Inc.
[1] Anselm Goertz, Markus Wolff, and Lutz Naumann, “Optimization of Sound Reproduction in Listening Rooms for Surround Sound Loudspeaker Setups”.
White paper for Klein+Hummel, 2001.
[2] Adrian Celestinos and Sofus Nielsen, “Controlled Acoustic Bass System (CABS) a Method to Achieve Uniform Sound Field Distribution at Low Frequencies in Rectangular Rooms”. J. Audio Eng. Soc., Vol 56, No 11, 2008 Nov.
[3] European Patent Specification EP 1 088 298 B1 (US Patent 6,795,557 B1), Mäkivirta et el., 2004 Sept.
[4] Todd Welti and Allan Devantier, “Low Frequency Optimization Using Multiple Subwoofers”. J. Audio Eng. Soc., Vol 54, No 5, 2006 May.
In order to avoid a too-long post, I’m going to describe DBA and SFM first, and follow up with the meaty part – the comparison.
In the following discussion, we are generally assuming rectangular rooms and low frequencies.
Double Bass Array
The term Double Bass Array (DBA) was coined in a white paper by Anselm Goertz et. al for Klein + Hummel [1]. The original paper is in German, but I found an English translation at:
http://forums.klipsch.com/forums/storage/8/1485263/tmt2002_eng.pdf
The paper does not give much detail however. There is also some discussion and pictures of this technique at
http://www.avsforum.com/avs-vb/showthread.php?t=837744
The original Goertz et al paper outlined the basic idea: try to produce a plane wave going from the front to the back of the room, thus eliminating standing waves throughout the enclosed space. The speakers in the front of the room are placed on the front wall in a way that minimizes sidewall and floor/ceiling modes, leaving only the front\back modes. The speakers in the rear of the room are delayed and their polarity reversed, so that they cancel out the sound from the front. The idea is essentially to create a plane wave traveling from the front to the back of the room, then cancel the reflected wave when it reaches the back. Some earlier researchers had employed methods to create such a plane wave, but using much more complicated speaker placements, FIR filters, and many microphones. Such advanced methods work very well, but are cumbersome to say the least.
Starting in around 2005, Adrian Celestinos and Sofus Bierkedal Nielsen applied a more rigorous approach to DBA by developing the Controlled Acoustic Bass System (CABS)[2]. They systematically investigated and formalized this concept, using sophisticated models and in real rooms. They made some very practical investigations into simplified speaker layouts, error due to non-optimal layouts, and other areas. Coincidentally, Sofus was a professor of mine at Aalborg University. Small world!
The basic DBA setup normally uses 4 subs on the front wall, located ¼ of the room width in from the walls and ¼ of the room height in from the floor/ceiling. The same configuration is used on the rear wall, and of course we are assuming a rectangular room. The signal processing is simply a delay unit, set to the time-of-flight for sound from the front of the room to the rear. Obviously, inverting polarity requires no special processing.
The ideal result would be flat magnitude and linear phase anywhere within the room volume. This assumes that a perfect plane wave is produced at from both the front wall and the rear wall (i.e. the cancellation wave). The subs need to be spaced closer than about ½ wavelength at the frequency of interest to approximate a plane wave (perhaps less if the room has significant low frequency absorption). Based on the simulations and actual room measurements done by Celestinos and Nielsen, the method seems to work very well.
A few miscellaneous notes and questions on DBA… I have heard it said that “Genelec has a patent on the DBA”. Unless there is another patent that I don’t know about, the one I do know about does not look like a patent on DBA [3]. It does have a source in a room and another out of polarity and inverted source at the other, but it also uses microphones to sample the space, and I don’t see the optimized speaker layout to cancel the sidewall and floor/ceiling modes. There are also a slew of other patents that use some sort of cancellation scheme. So perhaps not so clear. I don’t currently know of any product by Genelec using this. In fact it is hard to find mention of it on their web site.
Sound Field Management. (SFM)
Sound Field Management (SFM) was developed by Allan Devantier and myself circa 2003 [4]. In SFM, individual complex transfer function measurements (i.e. impulse responses) are made from each subwoofer to each seat in the listening area. For example if there are 4 potential subwoofers locations and 4 seats, there would be 16 measurements made. Notice I said “potential subwoofer locations”. The idea is to measure more subwoofer locations than you have subwoofers. So, the SFM algorithm would predict the total acoustical response at each seat, for each of the 6 possible ways you can put 2 subs into 4 possible locations (for example). The best configuration is then chosen according to whatever your criteria is. Generally the criteria used is to maximize the seat to seat consistency, though other criteria such as efficiency and flatness of magnitude response are also calculated and can be used.
But wait, there’s more! SFM also can include a number of different possible signal delays and simple one-band filters for each subwoofer, and predict the resulting response for each combination. This is known as a brute force optimization, can blow up the number of possible configurations to be simulated to many millions, so the algorithm has to be efficient. You can tailor the optimization to use whatever you have available: extra possible sub locations and/or signal delay and/or simple filters. Obviously the result is best if you have all three. Once you have run the simulations, you have a database, from which you can select any of the top rated solutions you like. In the case of SFM, we are optimizing at a few individual seats, not the whole room, like in DBA\CABS. This is an important difference, as we shall see later.
SFM has been thoroughly validated in real rooms and works very well. At least 2 patents have been granted and the system has been implemented the Harman Audio Test System (HATS) for use in the JBL Synthesis calibration toolbox.
Comments and questions welcome, and next post will get to the meaty part: point by point comparison of the two methods.
Bye for now,
Todd Welti
Research Acoustician
Harman International Inc.
[1] Anselm Goertz, Markus Wolff, and Lutz Naumann, “Optimization of Sound Reproduction in Listening Rooms for Surround Sound Loudspeaker Setups”.
White paper for Klein+Hummel, 2001.
[2] Adrian Celestinos and Sofus Nielsen, “Controlled Acoustic Bass System (CABS) a Method to Achieve Uniform Sound Field Distribution at Low Frequencies in Rectangular Rooms”. J. Audio Eng. Soc., Vol 56, No 11, 2008 Nov.
[3] European Patent Specification EP 1 088 298 B1 (US Patent 6,795,557 B1), Mäkivirta et el., 2004 Sept.
[4] Todd Welti and Allan Devantier, “Low Frequency Optimization Using Multiple Subwoofers”. J. Audio Eng. Soc., Vol 54, No 5, 2006 May.