Cardioid Bass

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Rudolf said:


A fine example how "louder" is being judged as "better". To me in almost all cases the multiple sub response looks preferable to the corner sub.

And who wants to place four subwoofers all at half wall distance? To me it seems almost the worst way to place them.

To Tom louder is better - thats all that matters.

The wall mid-point is the location recomended by Tod Welte in his AES paper. I wrote in an objection because I did not find these locations to be best in my studies, but that was all discussed here earlier.
 
Multiple subs

A year or two back, some of us were having similar discusions on AudioRoundTable.com. I had read the Welti paper and Geddes reaction to it, so I decided to study the problem in a little more detail using CARA, a room modeling program. Dr Geddes and Dr Thomanek discussed FEM techniques, and talked about the ray tracing and eigenvalue functions used in the CARA program.

For me, either placement method was acceptable, one of Welti's symmetrical arrangements or Geddes random arrangement, having at least one corner sub and one above midpoint in the vertical. It was interesting for me to see which arrangements worked best, and if there was a clear winner. I even proposed a wager, really just for fun, but we couldn't come to an agreement on the way to determine metrics, i.e. least deviation throughout the room, largest pocket of uniformity, etc.

I used the CARA program to try various configurations. I modeled various room shapes and setup Welti configurations, Geddes configurations and some that were sort of in-between, like grouping four woofers in a room as symmetrical pairs, two per side but each in a different location in all three planes.

I took away a few conclusions from this:

1. If there are few bass sound sources, damping has more influence on the amplitude of room modes than placement.

2. Adding bass sound sources makes the sound field more uniform throughout the room.

3. As the number of sound sources is increased, placement and damping have less effect on uniformity. When interference between sound sources and reflections is very dense, damping and placement become irrelevant.

4. When limited to one or two woofers, some positions in some rooms were clearly better than others.

5. Different shaped rooms had different positions that were best and worst. For example, a square room with no open doors has more clearly defined modes than an L-shaped room or one with open halls and connected living spaces. In a sense, such a room provides some degree of randomness.

6. When the number of woofers is increased to four, the differences between best and worst case placements was much smaller than it was with just one or two woofers.

7. Symmetrical arrangements tend to form symmetrical energy patterns in the room, some with large lumps or pockets of energy in the center, higher frequency modes usually forming multiple evenly spaced pockets like flower petals or checkerboard shapes.

8. Asymmetrical arrangements tend to make odd-shaped lumps, pockets that are not as easily defined. Depending on the arrangement, they could be long and wide, broken into small patches or forming jagged shaped regions.

In conclusion, I did not think there was a clear winner for a given number of woofers for all room shapes. I think it best to model the room with an FEM program like CARA.

Certainly, the more woofers you have, the denser the interference becomes, so the more uniform the energy becomes. Another absolute is the more damping you have, the smoother the modes are. Framed drywall construction helps here, because the walls provide some damping. Concrete walls like basements are tough to get right, and multiple subs are probably a necessity.
 
There seems to be strong support for multiple subs from a wide range of people. I would be interested in pros and cons of multiple subs versus using devices such as DEQX to do room correction below say 150 Hz. ( music listening not HT )
Two possible pros for DEQX etc might be
Can do correction up to 150 Hz, where using subs with their lower crossover points may make sufficiently wide range correction more difficult .
May be able to equalise mainspeakers to give quality bass without the need for extra boxes in the room.
 
Michael Ando said:
There seems to be strong support for multiple subs from a wide range of people. I would be interested in pros and cons of multiple subs versus using devices such as DEQX to do room correction below say 150 Hz. ( music listening not HT )
Two possible pros for DEQX etc might be
Can do correction up to 150 Hz, where using subs with their lower crossover points may make sufficiently wide range correction more difficult .
May be able to equalise mainspeakers to give quality bass without the need for extra boxes in the room.


The long and the short of it is: one works, the other doesn't. DSP does not and cannot correct for the LF problems in rooms. Multiple subs does correct the problem.
 
Dr Geddes, Could you elaborate a little on why DSP cant correct for room problems at lower frequencies. I believe I understand why multiple subs is a good ( probably the best ) approach but dont really understand why it is the only way to go. Many ( I think serious ) manufacturers like Meridian seem to think that DSP can help considerably and it would be good to know why you say their approach is flawed. I think this would benefit lots of people as it seems that digital room correction at lower frequencies is becoming increasingly popular with quite a few manufacturers offering various approaches . Thanks Mike A
 
Michael Ando said:
Dr Geddes, Could you elaborate a little on why DSP cant correct for room problems at lower frequencies. I believe I understand why multiple subs is a good ( probably the best ) approach but dont really understand why it is the only way to go. Many ( I think serious ) manufacturers like Meridian seem to think that DSP can help considerably and it would be good to know why you say their approach is flawed. I think this would benefit lots of people as it seems that digital room correction at lower frequencies is becoming increasingly popular with quite a few manufacturers offering various approaches . Thanks Mike A

I've stated this so many times now, but here goes again. Lets say that yo have a room with a couple of modes. You place the mic at a point in the room and lets say that its at one modes peak and another combinations null. The DSP controller will lower the peak and raise the null. Now at other points in the room the frequencies where there was a peak are now lowered everywhere even where there was a null and the null frequency is raised everywhere even where ther was a peak. So in essence the total spatial frequency response variance has not changed, its gotten better at some places and worse at others.

With multiple subs the variance gets globally better at every point and at every frequency. No electronic means can do this. In my rooms I have never used EQ above about 200 Hz and EQ below about 200 Hz only about 25% of the time and then only a single band. DSP is not required for this simple of correction.

And companies use "room correctors" because people think it works and they can show nice curves. Everybody in the business knows that it doesn't work, its simply a marketing ploy. DSP could "improve" a really bad situation at a specific seating position, but its is guaranteed to be worse elsewhere. The only real soution is multiple sources.
 
Room damping

EQ is ineffective because it can only be set to correct specific pinpoint locations. Move in any direction and the frequency/energy distribution is different.

Room modes are caused by standing waves, and so the energy distribution is location sensitive. If you view the pattern of energy in the room, you see a pattern of hot and dark spots depending on location and frequency. At low frequency, there is usually one or two large hot spots and one or two dead zones in the room. As frequency goes up, it splits up into a pattern of hot and dark areas. Depending on room size and shape and sound source location(s), the pattern can look like a checkerboard, flower pedals or a nondescript random jagged shape.

When you use multiple sound sources, the patterns from one source are superimposed on the patterns of other sound sources. If you choose sound source positions well, then where there is a dark region from one sound source, another sound source creates a hot spot, so the two average together. This corrects the energy balance throughout the room.

If you do it in enough places, you'll completely average the sound field. This is called dense interference. One or two sources don't have enough overlap to completely smooth the field. Three or four come closer, particularly if the focus is on target areas, say perhaps the center of the room or where people are most likely to sit. That is why I like modeling the room with FEM, because you can place the speakers where you see the best energy distribution form in the target area. But no matter how you look at it, increasing the number of locations of bass sound sources tends to smooth the field.

If you don't have the luxury of using multiple sound sources, for whatever reason, the next best thing is a well damped room. Framed drywall construction is your friend here, because it provides some damping. Concrete walls like in basements are tough because they are rigid and provide little damping. In that case, membrane or panel absorbers can be employed, sort of making false walls that damp the modes.

 
Wayne P and Dr Geddes, Thanks for your responses both of which are very helpful . I think its fair to say that multiple sources is the best possible solution as it " evens the the energy balance throughout the room ".
Dr Geddes points out that with DSP as you improve the response in one place you make it worse in others. For a solution where one person is listening to music from one position I was hoping that optimising the response at this position would be an acceptable compromise. It would seem that the response at other places doesnt really matter as long as no one is listening there ? I have no idea how large an area with reasonably flat response might be created by this method but maybe instead of going for maximum flatness at one point it would be possible to do a sort of averaging a bit like what I understand Dr Geddes does across the on axis / off axis response of his speakers.
I looked a CARA several years ago and it was interesting and I guess I could get a god idea of what can be achieved with multiple sources this way. I dont suppose its possible to simulate the results of bass equalisation in CARA so seeing what sort of flatness across what sort of area could be obtained through equalisation.
I strongly agree that using the room construction to damp low frequency modes makes sense and would be a major help in getting better bass. I bought Dr Geddes Home Theatre book a couple of years ago and found it very useful. Luckily my house is of light timber construction and sound proofing is not a requirement so I have a start in that way. Mike A
 
Sorry for reactivating this old thread, but i am not sure whether discussion
was a bit focussed on steady state response.


www.linkwitzlab.com/Woofer accuracy.rtf

Elias Pekonen Home Page - Dipole Bass vs Monopole Bass


Surely it might be easier (less dependent from speaker-listener placement)
achieving a more dense modal excitation over frequencies using a cardiod
(and even a monopole) than using a dipole.

"Modal balancing" using multiple subs seems preferable with all three types.


Does it play a role for estimation of subjective quality, that critical bands
of hearing are pretty wide (when thinking in octaves) at low frequencies ?

Bark scale - Wikipedia, the free encyclopedia

Is there any influence of that on "quality weighting" between time and frequency domain ?

Do we know the "subjective weighting function" for perceived sound quality at LF ?

Is the weighting function independent from "the kind sounds" to be
judged as being reproduced in a "realistic" way ?

(e.g. organ vs. drum , string bass vs. plucked)

Further Aspects:

"They are here" vs. "You are there"


A drum e.g. will not cause room pressurization by exciting Mode 0,0,0 at
the recording venue, right ?

Using dipoles ( or cardiods) there will be no (respectively less) pressurization
of the listening room too.

-> More like "You are there" ?

The modal excitation may in some settings be more sparse using cardioids/dipoles
but the pattern in time more similar to that of the instrument at the recording venue ?

Most (all?) LF musical instruments are directional.
The monopole loudspeaker (seen as a musical instrument) seems
the only exception.

Are we used to sparsely excited modal patterns at least
in small to mid sized musical venues ?

If so, that does not mean having Schroeder Frequency high in the
listening room is a good thing ...


On the other hand: How could we detect room pressurization (Mode 0,0,0)
as such while listening ?

In cases feeling "pressurized" at low frequencies excess SPL
should be the most likely reason, which could be corrected by
proper equalisation.

Difficult to discuss matters without considering the listening room.

Some rooms may go rather abruptly from modal to statistical behaviour,
others have a more broad transitional range.

Some rooms may have modally sparse bands somewhere, others
(due to proportions) a more continouus rise in modal overlap
(with frequency).

If a room has a significantly longer side, predominantly one mode
in length direction making up the low bass response, i can hardly
think of a difference in perceived quality whether exiting that mode
using a monopole, a dipole or a cardioid, given the "balancing"
(with respect to other modes) is comparable in all three cases
(making use of placement and/or eq).

When reaching the transitional range there will be a difference, also
due to the rooms behaviour.

I can think of a monopole source in the range of lowest room modes,
continouusly morphing into a cardiod at the transitional range of
the listening room (say 80 ...140Hz depending on the room's size
and absorption) sounding very similar like a "fullrange cardioid",
given the power response at LF is equalized accordingly.
 
Last edited:
I like visualizing all these matters with the image model. If the walls of our room where mirrors we would see reflected rooms and reflected speakers off to infinity. They are all turned on at t = 0 but the sound from the further units obviously takes longer to get to the listener (and has to transition more wall boundaries).

Now if there is a regular pattern to the arrival of image sources, that is, their wavefront arrivals are evenly repeated in time by the geometry of the adjacent images, then we will get standing waves from that regularity.

If we have a woofer directional pattern that acts to line up a radiation null of many of the image speakers with the null in line with the listener, then that standing wave will be proportionately reduced.

So directional or dipole woofers can help, but only if we carefully aim them with regard to a particular standing wave.

Regards,
David
 
...
So directional or dipole woofers can help, but only if we carefully aim them with regard to a particular standing wave.

Regards,
David

Which implies that all subwoofers should be free to rotate (in case of
cardioid and dipole types) and move independently from the "main"
speakers.

I feel combining e.g. a dipole ESL (or any OB type radiating as a dipole) with a
monopole (or any kind of) woofer or subwoofer fixed in one frame/cabinet is
not a preferable solution.

Nevertheless often done this way ...
 
Last edited:
So directional or dipole woofers can help, but only if we carefully aim them with regard to a particular standing wave.

Regards,
David

I did this study some years back using the more complete modal model than the image model. In this case, regarding "smoothness" of the response, there was no clear winner. Some "types" where better in some locations and others in other locations. but no "type" of source was always better than another "type" of source. However, when one looked at the SPL capabilities of the different types this came out hands down for monopoles (no surprise there).

The simpler model (2-D) was published in AES while the more advanced one (3-D) was just done on my own to verify that the conclusions made for 2-D simulations would hold in 3-D - they do.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.