For those interested in optimizing multiple subs in a room. This is a description of a commercial subwoofer that uses 4 sub modules. The first 2/3rd of the page describes the system and the reason for multiple subwoofers.
Towards the bottom of the page is a description of how to set the subwoofers up in a room. Most processes I have read about for setting up multiple subs seem to involve a lot of trial and error which can make them hit or miss. The process described here seems very straight forward, repeatable & simple.
DEBRA Distributed Subwoofer, 5+ pieces- James Romeyn
The system is ~ 3k USD, looks like a good value. But the process should work for anyone building or who have multiple subs.
BTW I have no involvement with the company that makes these subs.
Towards the bottom of the page is a description of how to set the subwoofers up in a room. Most processes I have read about for setting up multiple subs seem to involve a lot of trial and error which can make them hit or miss. The process described here seems very straight forward, repeatable & simple.
DEBRA Distributed Subwoofer, 5+ pieces- James Romeyn
The system is ~ 3k USD, looks like a good value. But the process should work for anyone building or who have multiple subs.
BTW I have no involvement with the company that makes these subs.
My personal experience leads me to believe that the concept is perhaps not a bad approach to solving problems of bass response in a room. It "might" solve the problem for most listening environments, however, there is no reason to believe that this is the only "good" approach to dealing with lobing and loading related issues in a listening environment.
Well based on the papers by Geddes and Toole it appears that 3-4 subs are needed to solve the issues with bass in a small (ie residental) room.
Could that be related to close field coupling. Bass is hard to control, but if the sub is very near to where one is sitting it may subdue the boom and reduce that -echo bass- sound.
PeterC.
PeterC.
Low volumes and a bit of eq to tame the boom gives one a bit of control in a small room.
Where my listening area, is in a basement, low ceiling, a focused listening zone of very nice sound. Outside of that there is a large dip and peak in bass response.
PeterC.
Where my listening area, is in a basement, low ceiling, a focused listening zone of very nice sound. Outside of that there is a large dip and peak in bass response.
PeterC.
There are a few good things about the writeup. Multiple sub woofers, obviously, and attacking the (IMO idiotic) conventional wisdom about low crossover points. $3k for what look like 4 Dayton RSS265's in nice cabinets with an SA1000 strikes me as a fair price, too.
Certainly, his setup will be better than 1 $3k sub. I'm not convinced that inverting the polarity of one sub will provide the benefit of having finer control of phase (delay) and separate level controls by using external processing and separate amps, though.
Certainly, his setup will be better than 1 $3k sub. I'm not convinced that inverting the polarity of one sub will provide the benefit of having finer control of phase (delay) and separate level controls by using external processing and separate amps, though.
This setup process can be applied to any subwoofer array. As to the part of reversing the polarity on 1 of the subs that is something you can determine experimentally.
I need to get another sub to try this out.
Do not forget about flanking subs or helper woofers, as they smooth the important range above 80Hz, up to the Schroeder frequency:
This region is a difficult range to deal with, because the sound field begins to transition between modal and statistical behavior. It is also the most noticeable range. Modes in this range tend to really cry out at you.
I've studied the modal problem for a long time, since first reading Todd Welti's papers about ten years ago. I began to setup woofers in four corners or along two wall midpoints, per his recommendations. Then, in 2005, I talked with Earl Geddes at the Great Plains Audiofest, where he debuted his Summa loudspeakers with Duke LeJeune. We talked at length about multiple subs during that show and in the months that followed. This is when Earl began suggesting his random arrangement.
There were some disagreements about where the best places were to position multiple subs, but everyone seems to agree that for deep bass response smoothness and uniformity, once you get to four subs, placement almost doesn't matter as long as they are not stacked. Whether Welti (symmetrical) or Geddes (asymmetrical), whether ordered (Welti and Romeyn) or random (Geddes), smoothness is assured below ~80Hz when four subs are used.
What is not so assured is smoothness above 80Hz. This range becomes very placement sensitive. Subs placed too far away from the mains cannot be blended at higher frequencies because they become localizable. And if a sub is placed within a few feet - where it can be blended higher - it is probably best to be tied to the main full range speaker it is flanking.
This brought me to a conclusion that a pair of flanking subs or helper woofers could be employed for blending in the midbass and lower midrange region. This strategy can be accomplished with a variety of methods, a separate helper woofer or "flanking sub", a built-in bipolar woofer or 2.5-way speaker with helper woofer placed behind and below the main woofer in the cabinet, or a three-way speaker with a midrange that has sufficient low-frequency capacity to be blended with the woofer in the 80-200Hz range.
Any of these flanking sub strategies can be used with more distant subs used to smooth lower frequency modes, positioned as described by Welti, Geddes or Romeyn. In fact, some Welti configurations can be employed with flanking subs as a matter of natural consequence. But no matter how the (distant) deep bass multisubs are configured, it is very important to employ some form of flanking sub / helper woofer arrangement to smooth the midbass. I think smoothing the higher frequency modes is a greater priority, because they are almost always the most noticeable.
This region is a difficult range to deal with, because the sound field begins to transition between modal and statistical behavior. It is also the most noticeable range. Modes in this range tend to really cry out at you.
I've studied the modal problem for a long time, since first reading Todd Welti's papers about ten years ago. I began to setup woofers in four corners or along two wall midpoints, per his recommendations. Then, in 2005, I talked with Earl Geddes at the Great Plains Audiofest, where he debuted his Summa loudspeakers with Duke LeJeune. We talked at length about multiple subs during that show and in the months that followed. This is when Earl began suggesting his random arrangement.
There were some disagreements about where the best places were to position multiple subs, but everyone seems to agree that for deep bass response smoothness and uniformity, once you get to four subs, placement almost doesn't matter as long as they are not stacked. Whether Welti (symmetrical) or Geddes (asymmetrical), whether ordered (Welti and Romeyn) or random (Geddes), smoothness is assured below ~80Hz when four subs are used.
What is not so assured is smoothness above 80Hz. This range becomes very placement sensitive. Subs placed too far away from the mains cannot be blended at higher frequencies because they become localizable. And if a sub is placed within a few feet - where it can be blended higher - it is probably best to be tied to the main full range speaker it is flanking.
This brought me to a conclusion that a pair of flanking subs or helper woofers could be employed for blending in the midbass and lower midrange region. This strategy can be accomplished with a variety of methods, a separate helper woofer or "flanking sub", a built-in bipolar woofer or 2.5-way speaker with helper woofer placed behind and below the main woofer in the cabinet, or a three-way speaker with a midrange that has sufficient low-frequency capacity to be blended with the woofer in the 80-200Hz range.
Any of these flanking sub strategies can be used with more distant subs used to smooth lower frequency modes, positioned as described by Welti, Geddes or Romeyn. In fact, some Welti configurations can be employed with flanking subs as a matter of natural consequence. But no matter how the (distant) deep bass multisubs are configured, it is very important to employ some form of flanking sub / helper woofer arrangement to smooth the midbass. I think smoothing the higher frequency modes is a greater priority, because they are almost always the most noticeable.
Last edited:
Wayne, thanks for the post. I see you approach as a variation of Welti and Geddes, if I am reading the link correctly if you are running say a 4 sub configuration with your 3 pi speakers for example (similar to the Geddes Harper) a 2 way with a 12 in pro woofer a horn. You would have 2 'flanking' subs close to the main speakers and then the other (2 or more) subs positioned in the room. Is this correct?
What method do you use for placing the other subs?
What I think Romeyn is doing that is a bit different for the other approaches in 2 areas. The speakers are placed with the drivers placed close to the wall ~1.5 in (<4cm), providing some loading I presume, and you reverse the polarity on 1 of the subs until you determine the best sound.
Have you tried any of these things when you have set up subs?
What method do you use for placing the other subs?
What I think Romeyn is doing that is a bit different for the other approaches in 2 areas. The speakers are placed with the drivers placed close to the wall ~1.5 in (<4cm), providing some loading I presume, and you reverse the polarity on 1 of the subs until you determine the best sound.
Have you tried any of these things when you have set up subs?
That's correct. I run flanking subs near the mains when I have matched-directivity two-ways as my main speakers. But I do not use flanking subs with my constant directivity cornerhorns, because they are sort of "built-in" on that design. Constant directivity cornerhorns don't need flanking subs because they are within 1/4λ of the corner apex and so do not suffer from self-interference from the adjacent walls. Also, the mid and bass are blended between 100Hz and 250Hz, which smoothes vertical modes.
As for the more distant subs, as I said earlier, placement doesn't matter. You can use a Welti method or the Geddes method. This new approach described by Romeyn would be just fine as well. The good news is, it doesn't really matter where you put the other two subs, as long as they aren't grouped together. That can help with WAF - Put them where they fit with the decor. When four subs are used, modal smoothing is assured.
Back in 2005, I did a lot of simulations as well as in-place measurements in a couple homes. I wouldn't consider two homes to be conclusive, except that I also did a LOT of simulations in different shaped rooms using different source locations. This started out as an excercise to compare the original Welti placement suggestions to Geddes pseudo-random placements.
What I found was that some placements were marginally better than others, depending on the room layout. But as long as four woofers were used, the differences were really small. If you used only three subs, the placements were a little more important, and if just two, placement became fairly important. Having just two really subs requires mains that have good extension, which then means you really have four bass sound sources, but I digress.
The thing that struck out at me most significantly was the differences above ~80Hz. Uniformity above 80Hz was usually poor unless (what I now call) flanking subs were employed. There are some Welti placements that have a couple subs flanking the mains. Those sound great. But this was missing on most other configurations. The conclusion I drew is simple - Include flanking subs or something equivalent, no matter what method you use to choose the more distant subwoofer positions.
This is really key here. It is the most important thing, in my opinion. When flanking subs are used, the response is good through the midbass and midrange. Without flanking subs, it most likely won't be, whether or not multisubs are used to smooth the lower bass range. So to me, that's the first priority. Run the first pair as flanking subs. Run them in stereo, one sub per main. Place it slightly below, behind and beside the main it is flanking, and run it up to somewhere in the 80-120Hz range, with a gentle low-pass, like second-order. That way there will be sufficient energy in the low 100s to provide modal smoothing, blended with the mains up through the lower midrange, up to approximately the Schroeder frequency. Being physically close to the mains, there are no localization problems and being between the mains and the nearest boundaries, self-interference from the reflections are mitigated.
Last edited:
I like to support that strategy, as it meets my experience from real
setups and simulations too.
Although >120Hz might be above the Schroeder Frequency in many
rooms. But even any modally sparse region around and above
Schroeder is hard to get right.
Those upper bass/lower midrange modes are likely to introduce
serious coloration:
Multisub configurations having wide spatial distribution in the room
cannot practically be used to smooth that frequency range, but are
restricted to lower frequencies <80 ..100Hz if you want the subwoofers
to be "unnoticeable" and not disturbing localizaton.
Having sound sources up to lower midrange distributed in height at
least is also a beneficial stragegy for usual floorstanding boxes.
Making that sources also distributed in depth and to the sides is
even more beneficial and yields more robust results
("robust" in a statistical sense: Reproducible in a variety of rooms
and settings).
Getting upper bass and lower mids right by "modal smoothing"
(spatially distributed excitatation of the room), might indeed be
priorized over smoothing the lowest two octaves, if you have to
decide for one of them (e.g. due to budget and/or space restrictions):
You can do even a lot using moderate seat and speaker's position
variations in the lowest octaves unless you choose more widely
separated subwoofers.
If you want some spatial uniformity/consistency in lowest octaves
for a couple of seats however, maybe you won't get around widely
distributed subwoofers a la Geddes or a la Welti/Devantier.
Btw. Dipole and Cardiod Woofers might also be seen as "distributed
sources" typically having separations of <1m. For the lowest octaves
also special concepts having larger separation are possible.
Those source separations <1m - which you propose if i do understand right -
are typically not sufficient for modal smoothing of the lowest octaves in
a living room (and for smoothing the spatial variation of the inroom
frequency response in that range), but are truly beneficial to reduce
coloration in the upper bass and lower midrange, especially if the room
is not heavily treated due to that range (which usually is not the case).
Kind Regards
Last edited:
Example for "modal smoothing" upper bass / lower midrange
The following pictures may not be a "proof" as this evaluates only
one specific setting, but it may serve as a demonstration of
"modal smoothing":
1st picture: 2 sources exciting the room from the very same position
...(or "single source with double volume displaced")
2nd picture: 2 spatially separated sources
...0.6m along X-axis (depth of room)
...0.1m along Y-axis (width of room)
...0.6m along Z-axis (height of room)
The "listening position" in the opposite corner under the ceiling may not be
"typical" ... but instructive
The frequency range from 80Hz ... 600Hz was chosen, because there
is only little change below 80Hz: The separation of the sources is too small
to show significant effect in the lowest octaves.
In the frequency range shown a multiple subwoofer setting can do close
to nothing while a "multiple woofer" with moderate separation of sources
<1m setting can do a lot.
>> Which range should have priority in smoothing the loudspeaker room
interaction:
a) 80Hz.....600Hz or
b) 20Hz ......80Hz
Kind Regards
The following pictures may not be a "proof" as this evaluates only
one specific setting, but it may serve as a demonstration of
"modal smoothing":
1st picture: 2 sources exciting the room from the very same position
...(or "single source with double volume displaced")
2nd picture: 2 spatially separated sources
...0.6m along X-axis (depth of room)
...0.1m along Y-axis (width of room)
...0.6m along Z-axis (height of room)
The "listening position" in the opposite corner under the ceiling may not be
"typical" ... but instructive
The frequency range from 80Hz ... 600Hz was chosen, because there
is only little change below 80Hz: The separation of the sources is too small
to show significant effect in the lowest octaves.
In the frequency range shown a multiple subwoofer setting can do close
to nothing while a "multiple woofer" with moderate separation of sources
<1m setting can do a lot.
>> Which range should have priority in smoothing the loudspeaker room
interaction:
a) 80Hz.....600Hz or
b) 20Hz ......80Hz
Kind Regards
Attachments
Last edited:
Some more examples:
- there are 3 pairs of graphs, pairs have filename prefixes "02", "04", "06".
- within each pair there is a single source and a double source member.
- the single source member in each pair is placed at the mean coordinates
of its double source twin.
Absolute position and separation is varied slightly between the 3 pairs.
One cannot say "use separation and you are fine". But using separation
there is a tendency towards more smoothness of the inroom response,
which would have to be evaluated in a more systematic experiment of course.
- there are 3 pairs of graphs, pairs have filename prefixes "02", "04", "06".
- within each pair there is a single source and a double source member.
- the single source member in each pair is placed at the mean coordinates
of its double source twin.
Absolute position and separation is varied slightly between the 3 pairs.
One cannot say "use separation and you are fine". But using separation
there is a tendency towards more smoothness of the inroom response,
which would have to be evaluated in a more systematic experiment of course.
Attachments
Last edited:
- Status
- Not open for further replies.