[ekovalsky]

Thanks for clarifying John.  What you are referring to is indeed not yet available.  No doubt that some room correction products of the future will incorporate this technology.  

To some degree, the TBRC you refer to simulates the non-DSP "double bass array".  This quad sub setup works by mirroring the front two subwoofers on the rear wall, running them 180 degrees out of phase and with a delay proportionate to the time it takes from the front subs' output to reach them.  In other words, the rear subs don't fire until you've already heard the output from the front subs.  When they do eventually fire, the front + rear subwoofer outputs cancel each other behind the listening position, preventing reflections off the wall.  I've seen the room response measurements of normal dual sub versus such quad sub setups, and the improvement is dramatic.  Non-modal peaks and nulls from rear wall reflection are largely eliminated, as are most standing waves i.e. room modes.  

Room modes are definitely important, and most of the current room correction gear (TacT, DEQX, Meridian, Accuphase, Rives) can deal with them quite well.  Unfortunately in most rooms, the non-modal nulls and peaks are just as prevalent and not nearly as amenable to correction, particularly the nulls.  TBRC or DBA should really help.

The Meridian article is very good.  One comment about it though -- while true, saying that room correction above 250hz produces more problems than it solves depends on the speakers and the particular correction product used.  Why ?  Many speakers benefit greatly from equalization.  With the TacT and some other DSP room correction products, user manipulation of the target curve accomplishes both room correction and speaker EQ simultaneously.  In the case of my former speakers, the EQ was equally if not more beneficial to the sound than the actual room correction was.

For speakers that have in-room frequency response above 250hz that is either fairly uniform or already conforms to your desired target, then any correction/manipulation will introduce frequency response anomalies away from the sweet spot, and thus be detrimental to the sound in other locations in the room.  As my current main channels measure quite uniformly above 200hz, I can form the target curve exactly to the measured response, thus bypassing correction.  Since the TacT is only correcting below 200hz, the Meridian system would be equally effective in my current setup.

DEQX was clever to use a two stage correction process -- speaker correction (i.e. digital EQ) and room correction.  This way you can use one, both, or neither as necessary.  This can be done with TacT also, but only with the RCS and S2150 amps together and the basically undocumented but free TACS software.

[John Ashman]

To some degree, the TBRC you refer to simulates the non-DSP "double bass array". This quad sub setup works by mirroring the front two subwoofers on the rear wall, running them 180 degrees out of phase and with a delay proportionate to the time it takes from the front subs' output to reach them. In other words, the rear subs don't fire until you've already heard the output from the front subs. When they do eventually fire, the front + rear subwoofer outputs cancel each other behind the listening position, preventing reflections off the wall. I've seen the room response measurements of normal dual sub versus such quad sub setups, and the improvement is dramatic. Non-modal peaks and nulls from rear wall reflection are largely eliminated, as are most standing waves i.e. room modes.

Man, I thought I was the first to think of that because I'd never heard of anyone doing that.  That's exactly what I'm planning when I built my HT room.  Subwoofers front/rear/sides, all crossed over individually with a delayed, 180 degree bass signal.  I thought I was pretty clever

The Meridian article is very good. One comment about it though -- while true, saying that room correction above 250hz produces more problems than it solves depends on the speakers and the particular correction product used. Why ?

Because a few things (if I understand your question).  If you use TBRC on high frequencies, the higher you are in frequency, the more you have to put your head in a vice.  At that point, you're not doing room modes, you're doing wall reflections.  And, our ears dismiss wall reflections as separate from the original sound.  They have influence, but it isn't really heard as distortion.  With bass, the effect is more pronounced and we can't separate the mode from the signal.    So, it's best to EQ the speaker to be flat, perhaps run a small parametric EQ to compensate for the *overall* room sound (and personal preference).  If you change the signal above 300Hz or so, you actually hear it as a detriment to the sound, even if it measures perfectly.  Microphones work differently than our ears do.

[csero]

Man, I thought I was the first to think of that because I'd never heard of anyone doing that. That's exactly what I'm planning when I built my HT room. Subwoofers front/rear/sides, all crossed over individually with a delayed, 180 degree bass signal. I thought I was pretty clever  ...

An even more clever solution would be to launch not a simple delayed/phase altered signal from behind, but a specific pattern of multiple reflections - just like what is coming from behind in a real concert hall. And we could do it from the sides or from above, or from the front too...  ... and soon we arrive to ambience synthesis, which ( if the listening room is relatively problem free) can swamp all the room related  problems, and create a quite believable ambient filed.

And, our ears dismiss wall reflections as separate from the original sound. ...

No, your ear will not dismiss them, it will tell your brain, that you are sitting in a small listening room, while two speakers in front of you tries to imitate a big arena.

[ekovalsky]

John, some German audiophiles have been using DBA for a while and swear by it.  I have not seen it implemented here in the states, may be you and I will be the first!.  I think it is an absolutely great concept, hopefully in practice it will not disappoint.  I plan on trying it after building out my room, which will give me the needed space behind the listening position.

We are in agreement about correction at mid and high frequencies.  The benefits I heard with my old speakers above 250hz were mostly related to equalization.  But, besides changing the speaker voicing, it also equalizes the channel balance at ALL frequencies.  This really helps with imaging/soundstaging, particularly if the speaker position and/or room are not perfectly symmetric.

Appropriately setting delays makes a huge difference too.  It amazes me that I used to try to correct L-R channel imbalance with a analog volume control.  It takes significant changes in output levels to move the image.  Adding a very small delay to the closer speaker (so the sound arrival is the same for each) is vastly more effective in restoring proper balance than any volume control or L-pad.  At one point, despite both speakers having equal volumes and being equidistant from the listening position, all the imaging was coming from one channel.  It took me about 20 minutes to realize I had inadvertently added a small delay to the other channel's amp while fiddling with the RCS remote!

The TacT really acts as a measurement-inverting infinite parametric EQ in the bass range, and as a voicing equalizer and channel balancer in the mid and upper frequencies.  The results are pretty spectacular after you (1) master the software -- not easy since the documentation stinks (2) learn how to appropriately properly set the target curve.  If you use a flat target curve or any of the stock curve without molding it to measured response, the results will not be optimal and may be terrible.  Took me a long time to figure that out !

[John Casler]

 To some degree, the TBRC you refer to simulates the non-DSP "double bass array". This quad sub setup works by mirroring the front two subwoofers on the rear wall, running them 180 degrees out of phase and with a delay proportionate to the time it takes from the front subs' output to reach them. In other words, the rear subs don't fire until you've already heard the output from the front subs. When they do eventually fire, the front + rear subwoofer outputs cancel each other behind the listening position, preventing reflections off the wall. I've seen the room response measurements of normal dual sub versus such quad sub setups, and the improvement is dramatic. Non-modal peaks and nulls from rear wall reflection are largely eliminated, as are most standing waves i.e. room modes. .

Hey Eric,

Ah yes, I remember you mentioning this.

I'm not sure I am getting the description properly.

If the rear subs are run 180 degrees out of phase, any delay will simply change that phase, since phasing is simply "unity timing to directional wave pressure" is it not?

As you know, I use a "simplified" version of this, which is predicated "exactly" on positioning of the listener and the subs, and they are run "exactly" out of phase.

That is the subs are equal distances apart, and the listener is "exactly" equidistant from each sub.

This accomplishes 2 goals:

1) it creates a great movement of air and sound energy exactly between the quad Sub array, since both sets of subs act on "all the air" between them every cycle.

2) it send "equal and opposing" bass energy into the room to (as much as possible) cancel nodes/antinodes of "real time" music.

From your description, it would seem that you would get some very weird bass reflections being bounced off the front wall (back at you) from the rear subs (since they are delayed and then their return wave will be delayed even more.

I would also think that for "dynamic music" it would sound overly spacey/echoey???

Have you had an opportunity to hear this type of set up or just see measurments.

I'd really like to hear it, but like my earlier comment, it seems to focus on being an "active bass trap" or "active bass neutralizer".  My wonder is, how it can do so without fudging up the real sound, unless it is designed to do exactly what I set out to do, except it allows you to manipulate the "timing" on the rear subs to allow them to be placed at a distance that is not "equal" like mine have to be.

[John Ashman]

John, if I understand what you're doing, all you'd be doing is actively canceling the bass.  Active bass cancellation as Eric is suggesting works like this.  

1.  The bass impulse leaves the front subwoofers (let's assume in-wall subs for simplification.
2.  The bass impulse will hit the back wall (let's say 20' away) and much of that is reflected back towards the listening position.  It takes, coincidentally, 20ms to get there.  
3.  A digital delay duplicates and delays the bass signal to the rear subwoofers (wired out of phase) by 20ms.  
4.  As the bass impulse bounces off the back wall, the rear subwoofers exactly duplicates the original impulse and launches it at, hopefully, the exact same time the original impulse bounces off the back wall.  
5.  The inverted and delayed bass impulse collides with a virtual opposite of itself and mixes together and essentially cancelling that energy by the time it gets to the listening position.  
6.  There is no subsequent reflection from the front wall, thereby totally cancelling the mode created by the 20' dimension.  

Get that?

[John Casler]

Eric and John,

I find this quite interesting based on my expereinces, but don't want to sidetrack Josh's thread.

Eric, is it possible to cleave this section, and move it to say the "acoustics" Circle, where maybe Ethan and some others might have interesting input?

So as to keep Josh's thread uninterupted.

[ekovalsky]

John C.

Rather than saying "180 degrees out of phase" we should have said "inverted polarity".  This probably would have prevented some confusion.  The impulses will be offset in time by only the added delay, but the front subs will be positive polarity while the rear subs will be negative polarity.

I'm not exactly following how you have your quad subs setup, but it seems likely you are either getting some cancellation around the listening position or have setup at a 'push-pull' or isobarik type arrangement.  If you have photos or diagrams I would probably understand better.

John A. gives a very good explanation of how this works.  I had already typed my response so here goes:

1.  Sound launches from the front subwoofers at time=0.  Signal is also sent to rear subwoofers but with added delay based on distance from the front subs (i.e. 17.7 msec if 20' separation).

2.  Front subwoofer output reaches listening position, for instance at time=12.4 msec if 12' from front subs.

3.  At time=17.7 msec front subwoofer output reaches the rear subwoofers.  At that instant the rear subs fire the same signal but with inverted polarity.  The outputs of the front and rear subs cancel each other or 'annihilate'.  Therefore, output from the front subs never reaches the rear wall intact, thus cannot reflect off it.  Output from the rear subs, which is immediately cancelled by the sound arriving from the front subs, never reaches the listening position or the front wall.  The end result is a virtual subtraction of the rear wall.  Bass modes cannot establish, and there is a dramatic reduction of non-modal peaks and nulls.  The latter is most important as these nulls are what current room correction devices cannot really fix.

I don't know of anyone in the United States with a DBA setup.  Perhaps John A. and myself will be the first.  I plan on implementing it after my room is expanded, and it will complement my current absorptive treatments (which need to be rearranged) and installation of Soffittraps around the room perimeter.  

There is some literature on the Double Bass Array, alas all in Deutsche, with some measurement graphs and subjective impressions.  Here is one paper in .pdf format.  Even if you don't translate it the graphs and diagrams are fairly self explanatory.

I split this topic from a thread in the Digital Processing forum, which John C. rightfully suggested and I was already considering.  Hopefully Ethan will be willing to give us his expert opinion on this method of bass correction.

[Brian Cheney]

I can translate any German material on the subject.  When I was in Munich I made my living translating patent apps back and forth into German and English.

[youngho]

Interesting setup. Forgive my stupidity, but just so I understand better...

The rear subwoofers address the length-related axial modes.

The duplication of the front (and rear) subwoofers, wiring in phase, and orientation in opposite polarity essentially result in a large bipole, addressing width-related axial modes. I'm tired, so I can't think through the calculation of exactly where the drivers of the subwoofers should be relative to the width of the room, but it appears as though they are located at about the 1/4 and 3/4 mark.

Height-related axial modes are not addressed, but hopefully the seated position in typical listening rooms with 8' or 9' ceilings will relatively minimize these effects.

I was a little surprised that the listening position had such a problem with the length-related axial modes, as it seemed to be located in a pretty good position with respect to the front and rear walls.

How does one set the relative level of the rear subwoofers?

Why don't complex comb filtering effects from sidewall, ceiling, and floor reflections occur?

In looking at the graphs, it appears as though DBA resulted in boosting of the non-peak frequencies above, say, 23 Hz, rather than attenuation of peak frequencies, which is what I would have expected from the DBA array eliminating length-related standing waves. Why wasn't extremely low bass (sub-23 Hz) louder relative to...er...pretty low bass (higher than 23 Hz)?

Thanks!

[Ethan Winer]

> ... saying that room correction above 250hz produces more problems than it solves depends on the speakers and the particular correction product used. Why ? <

I'd put that frequency closer to 100 Hz or even lower. Heck, I'm not convinced that "room correction" is very useful at any frequency. One problem is the peaks and nulls are highly positional. Any correction you apply for one location is either inadequate somewhere else, or actually makes the response worse. I don't mean 10 feet away. I mean the next seat over on the couch.

Also, no matter how many subwoofers you have, or how sophisticated the DSP used, all that's affected is the raw low frequency response. But modal ringing stays the same, and that's at least as big a problem as the skewed response.

--Ethan

[JoshK]

Ethan,

I own the DEQX but have yet to play with the DSP and RC features, just haven't had the time yet as I just got it but in theory I totally agree with you.  I still want to play around a bit with the RC, but I don't have high hopes for it.  It isn't why I bought the DEQX.

Anywho, listening to csero's system (ambiosonic type system) and many of our (the NYAR) systems has made numerous things quite clear.  It has reset the objectives and the changed the measuring stick for me.  It will only start arguments if I was to expound upon this, but one thing is clear, that there is no subsitute for a good acoustic environment (well treated room).

One experiment that I want to try involves a lot of tedious work.  Instead of taking all measurements at one spot and then applying RC, take measurements all along a 3 dimensional area.  The result would be that you would have a 3-D response surface for every unidirection, 3 in total. Then using these 3 response surfaces with some analysis you can determine a lot about the room and also what more modest RC could be applied (by averaging, etc).  So the resultant correction curves are not the inverse of the FR with respect to the target curve but instead a more empirically derived inverse.  

It is quite possible that the above method is an excercise of futility but maybe it is the statistician in me that is determined to try.  I am certain I am not the first person to have thought of this, the concept has probably been droned to death in the field, academics, journals and such but I am completely new to this and not well read on the subject.

I am a slave to the limitations of what acoustic treatments can be done in my room due to the wife's aesthetic overruling.

[klh]

The harmon white papers also address this issue to some extent (the lower frequencies). They said that 4 subs, each in one of the corners smooths modal responses throughout the room, but that low end extension suffers a bit. If you have 4 speakers, each at the wall midpoints, the modal response is almost just as good, but there is much better low frequency extension. One may be able to extrapolate that that is becuase the front to back waves are attenuated (as previously described) as well as the side to side, provided the one of the side subs is time delayed to the seated position and the opposing side sub is delayed with reference to it's opposing sub and 180 out of phase.

[ekovalsky]

I'd put that frequency closer to 100 Hz or even lower. Heck, I'm not convinced that "room correction" is very useful at any frequency. One problem is the peaks and nulls are highly positional. Any correction you apply for one location is either inadequate somewhere else, or actually makes the response worse. I don't mean 10 feet away. ...

Room effects are definitely positional, particularly non-modal peaks and nulls.  When I have had several people all trying to listen semi-seriously to my system, I byass correction and just use two bands of parametric EQ (40hz and 72hz) to tame the dominant room modes.  

For me, and probably the majority of audiophiles, the vast majority of time spent listening is by myself.  This is a good thing, because I can listen to whatever I want and also enjoy the full benefits of correction, which is basically only active below 135hz with my current room/system.



This graph shows the actual correction filter implemented by the TacT, with frequency on the X-axis and attenuation on the Y-axis.  Above 135hz, there is a fairly uniform cut of about 6dB which allows some equalization in the bass without surpassing 0dB (except at a high Q band around 55hz -- corresponding to my major null -- where I'm cheating with up to +2dB).  If beyond 135hz this filter was up and down like a roller coaster, that would mean extensive correction or equalization is being applied by the TacT to compensate for problems that it really cannot fix.  Note that the +/- 1dB variations in each channel above 135hz result in perfect balance at all frequencies at the listening position -- this definitely helps focus and refine the soundstage.  The difference would be probably be much larger in most other rooms, as mine setup is perfectly symmetric in speaker positioning, dimensions, treatments, and even furniture.  It is fairly easy for anyone to match overall volume of the L & R channels with a simple device like the Rat Shack meter, or just using their ear, but it is likely that in most setups there will be significant variations in SPL between the channels at different frequencies.  The result is diffusion of the soundstage as fundamentals and harmonics at certain frequencies shift leftwards or rightwards.

[In case anyone was wondering about loss of resolution that occurs with correction, the RCS takes the 16 bit input and adds eight more bits for DSP headroom.  Thus with reasonable levels of correction no loss of resolution occurs.  The price of correction of course is a reduction in maximum SPL -- in my case about 6dB at most frequencies.  Since the speakers are capable of over 120dB output this is not a problem but it could be an issue for those with classic Apogees and other low efficiency speakers, and those who use low power tube or solid state amplifiers.]

Below 135hz, the "dips" on this graph actually correspond to the room modes, showing increasing attenuation at 40hz and 72hz.  These are what I tame with parEQ when correction is bypassed.  The "peaks" on this graph show how the TacT deals with nulls, by preserving output at those frequencies while reducing output at other frequencies.  Nulls really cannot be fixed, the best way to deal with them is to find the location for the listening position and speakers where they are minimized, and (as I am doing) try to incorporate them into a crossover range.  

The Double Bass Array configuration should be very beneficial in handling nulls.  If you look at the graphs below from the .pdf I linked to before, you can see how effective the arrangement was in eliminating the large null, and to a lesser degree attenuating to room modes.

PRE-DBA with two front subs


POST-DBA with two front subs plus two rear subs delayed and inverted polarity


Which bass response curve do you think would sound better ?  The PRE-DBA graph shows a nightmare for a room correction system with a huge suckout from 28-56hz.  With DBA the problems the response is incredibly good, with just some fairly mild residual suckout in a high Q band at about 30 hz.  Any room correction system would love to see that as its starting point !!!

[klh]

If one had two subs up front and one in the rear, but the seated position was closer to the rear, do you think it would work similarly? This makes sense in that the greater distance from the front would require the front subs to play much louder than the rears to have the same volume at the seated position. As such, if there were two up front (centered on the front wall), and one in the rear (also centered), then all three would be more or less equally stressed (since the ones up front would couple to each other).  Since all three would be more or less equally stressed, the SQ would be more or less the same, and thus overall SQ would be better. Am I missing something?

[ekovalsky]

If one had two subs up front and one in the rear, but the seated position was closer to the rear, do you think it would work similarly? This makes sense in that the greater distance from the front would require the front subs to play much louder than the rears to have the same volume at the seated position. As such, if there were two up front (centered on the front wall), and one in the rear (also centered), then all three would be more or less equally stressed (since the ones up front would couple to each  ...

klh, the problem I see with using only three subs is the wall, ceiling, and floor reflections from the front two subs would be very different than from the solitary rear sub.  This would then set up a whole new set of enforcement / cancellation phenomena.  If two subs are used up front and two more in the rear, in a mirror image arrangement, the reflected bass waves should somewhat cancel each other after the primary waves collide, at least if the room is symmetric.

The DBA will work best if all four subs are identical, or nearly so.  I would expect problems if, for instance, two front subs were vented and the rear subs were acoustic.  Same with front versus side firing woofers, front versus rear vents, two woofers versus one woofer, etc...

Expanding on your thought... Two subs could be tried, one along the center of the front wall and the other along the center of the rear wall (delayed and inverted phase) behind the listener.  But most audiophiles prefer stereo versus mono subs unless the crossover is extremely low, like 35-40hz.  Obviously the benefits of the sub array are limited if the subs are only operating to 35-40hz, as many bass problems are usually at frequencies somewhat higher than that.  In my room my biggest problem is at 56hz.

[John Casler]

John, if I understand what you're doing, all you'd be doing is actively canceling the bass.  Active bass cancellation as Eric is suggesting works like this.  

1.  The bass impulse leaves the front subwoofers (let's assume in-wall subs for simplification.
2.  The bass impulse will hit the back wall (let's say 20' away) and much of that is reflected back towards the listening position.  It takes, coincidentally, 20ms to get there.  
3. A digital delay duplicates and delays the bass signal to the rear subwoofers (wired out of phase) by 20ms.
4. As the bass impulse bounces off the back wall, the rear subwoofers exactly duplicates the original impulse and launches it at, hopefully, the exact same time the original impulse bounces off the back wall.
5. The inverted and delayed bass impulse collides with a virtual opposite of itself and mixes together and essentially cancelling that energy by the time it gets to the listening position.
6. There is no subsequent reflection from the front wall, thereby totally cancelling the mode created by the 20' dimension.

Get that?...

Hi John Yes I "get it", but have a feeling it isn't accurate.

The "neutralizing" Subs don't fire in a vacumn, and the sound wave it creates, while providing a scavenging affect to the leading wave "after it passes" the listener, will continue on to the front wall (and really all surfaces) and return to affect many other waves both direct (in both directions) and indirect.

I think your assumption that it "neutralizes" the energy completely is wrong.  I have a feeling it will only neutralize the energy at the "position" where the waves collide, where we will have a serious "dip", but that energy "will" continue on to the front wall and bounce off to be propelled to the listener with additional "direct" waves.

No free lunch on that one.  The only way to stop that energy is by turning it into heat.  A collision with an opposing wave won't create heat.

That energfy is in the room until it is absorbed, is dissipated, or finds a way out of the room.

By setting up an "opposing phase" mirror image, all modes and nodes are excited in their "exact" opposite, which nullifies each (theoretically) in any postiion of the room. (all things being perfect)

It is true however that the sweet seat position of equal distance between all 4 subs will produce the best result, and they must all be the same distance from all the specific walls.

My assumptions on the "neutralizing" system described is based on not hearing it (and maybe there is a point that is not coming out)

I would also wonder about the delay and the affect it has on the sound.

Maybe someday we'll be able to hear it.

[klh]

My main reason for suggesting two up front and one in the rear is I am in the process of installing an infinte baffle subwoofer with two Ascendant Audio Avalanche 18" drivers. They will be mounted in a manifold on the ceiling with the opening front and center and adjacent to the front wall. Since the two will  fire out of or through a 20.75" x 20.75" hole, I figured if I mounted another Avalanche 18" in the ceiling at the back of the room, centered side to side and adjacent to the rear wall (except the driver would be parallel to the ceiling), the sound characteristics would be similar enough to work. What do you guys think?

[ekovalsky]

My main reason for suggesting two up front and one in the rear is I am in the process of installing an infinte baffle subwoofer with two Ascendant Audio Avalanche 18" drivers. They will be mounted in a manifold on the ceiling with the opening front and center and adjacent to the front wall. Since the two will  fire out of or through a 20.75" x 20.75" hole, I figured if I mounted another Avalanche 18" in the ceiling at the back of the room, centered side to side and adjacent to the rear wall (except the driv ...

May be worth a try.  Do you have something to provide the necessary delay ?  Inexpensive Behringer gear would do the trick.

IB with 18" Avalanches should be awesome.  Good choice of woofers!

[JohninCR]

I'm doing something similar, but with one sub at the front and one at the seating position.  I'll deal with my room mode at 28hz with the XO because the subs will have 28hz as the upper cutoff.  My primary emphasis is to keep output in room because I watch late night movies, so I'm pretty sure that changing things in the time domain will mess me up.  If I start equal but opposite phase waves at the same time, my net in room is 0, which should drastically reduce what tries to escape.  If I set the waves off at different times, then I think I'll just end up with 2 separate outputs trying to escape the room.  Also, my door is close to the center of my planned dipole null and I don't want to mess up that benefit.