Multiple Small Subs - Geddes Approach

gedlee said:
I still only see the single mode/node discussion as being misleading. Even at a node a source will excite other nearby modes since all modes are excited by all frequencies to some extent.


A 20Hz sine will not excite a 20k standing wave, and a 20k sine will not excite a 20Hz standing wave.

Will 20.1Hz see reinforcement form a 20Hz standing wave? Yes! Will a sub placed in the node of a 20Hz standing wave excite the second order resonance/standing wave at 40Hz when the signal to the sub is shifted from 20Hz to 40Hz..? Of course!

I don't see what this has to do with the things being discussed.


Then there is the direct sound, but careful study will show that this direct sound is in fact the contribution of all the modes (Welti got this wrong in his paper).

The sound that reaches a specific point in a room is the direct sound +/- the reflections and room resonances. Nothing new there.

This goes back to something that I said a long time ago that nobody accepted. Free space has to be thought of as a continuum where the modal density goes to infinity - not zero.

Could the reason be that what you write makes no sense at all?

I think it would be correct to say that high modal density means "tightly packed resonances in the form of standing waves". Standing waves occur between two or more surfaces that trap/reflect the propagating soundwave. Wihout opposing surfaces (or rather no surfaces at all) there will be no such resonances.

Hence, even outdoors in free space the direct field is carried by the modes (which are now infinite in density) just as it is in a small room. But in a small room the modes get sparse and hence the ability for them to carry the energy goes down.


Can you tell us what it is that makes you think this is the case?

The sound from a LF source does not - let me repeat - does not travel in all directions away from the source.

Yes it does IF the radiating source is small compared to the wavelength. If you can show the opposite I'll be glad to see you at the Nobel dinner next year. ;-)

(There is what is called an evanescent wave sent out, but this disapates in time and space exponentialy so it is a very small factor.

And this is something only you are aware of?


The sound wave can travel only in a discrete number of directions defined by the modes that it excites.

Nope! It travels in all direction but the sum of the pressure in a specific point at a given time depends on the reflections and standing waves within the room.

This means that the energy emitted by this source in a real room is not the same as the energy emmitted by this source in free space. This can be seen in the radiation impedance for a small room which is NOT the same as that for a source in free space.

Yes it is , more or less. An electrodynamic driver sub is controlled by the compliance and the mass of the driver and the box it is mounted in. Acoustic impedance has little effect.

People want to think that you take a free space source and bring it into a room and that it emites sound the same way, but that this gets amplified by the modes at certain frequencies. This is not correct. The presence of the room changes everything and not until the source sees a high model density does it begin to behave as it does in free space.

Again this makes no sense at all. Since you claim things not known to the rest of the world and not backed by known physics maybe it would be in order to back your claims up by a rational explanation, a model or some measurements??



/Peter
 
Earl,

Free space has to be thought of as a continuum where the modal density goes to infinity - not zero. Hence, even outdoors in free space the direct field is carried by the modes (which are now infinite in density) just as it is in a small room. But in a small room the modes get sparse and hence the ability for them to carry the energy goes down.

In other words, a room, whatever size, is a modal filter?

Thus, a very large room filters fewer modes than a very small small room?

This means that the energy emitted by this source in a real room is not the same as the energy emmitted by this source in free space. This can be seen in the radiation impedance for a small room which is NOT the same as that for a source in free space.

The presence of the room changes everything and not until the source sees a high model density does it begin to behave as it does in free space.

That is, the source will start to see an impedance similar to what it sees in free space?
 
Good to read all the debates / discussions. With all this enormous amount of information I still feel like I am unable start doing it.

You don't need to convert me now as I am already converted. I have opted to using Geddes multiple sub approach.

I am just right now designing my own PCB on veroboards after having just completed the sub woofer EQ, allpass filter (worked out 6-160dgree phase shift by turning a 100k trimpot) on LTSpice.

What XO and slope? This is the question asked many times and nobody has given an answer, or the answer would be like "they are selected based on the result of measurements". Well it is true, but ideally some samples can be given to serve as a guide.

I am considering these factors:

1. A typical room has the model region 35Hz-200Hz.
2. The subs should cover as much the model region as possible.
3. We can localize sound source starting from 150Hz.
4. Gentle slope (like LR2, BW2) may be better than steep slope (like LR4) because the model region has a transitional period and we want to maintain a smooth frequency response.

If we can localize sound source starting from 150Hz, that means the subs should be cut off at 150Hz. And we know, unless the SPL is at least 30dB down, it would have an influence on the main speakers.

Suppose we have a 2nd order LP at 50Hz, at 100Hz it is -12dB, at 200Hz it is -24dB, at 400Hz it is -36dB. This may hide the sub and make it not localizable. But it would not have sufficient output from 100Hz to 200Hz to work with the room modes.

Suppose we have a 2nd order LP at 100Hz, at 200Hz it is -12dB, at 400Hz it is -24dB. The sub would probably have the right amount of output to smooth the model region but obviously, it is very localizable.

This points me to BW3 and LR4. However, since Dr. Geddes recommends using commercial subs, and most of the commercial subs support only 2nd order LP, I can't see how the job can be done.

Regards,
Bill
 
FrankWW said:
Earl,

In other words, a room, whatever size, is a modal filter?

Thus, a very large room filters fewer modes than a very small small room?

That is, the source will start to see an impedance similar to what it sees in free space?


Correct as the room gets larger and larger its modal density increases until eventually it becomes free space. I'm not sure that I would use the term "filter" - I actually thought of using it before, but then I declined as its not quite the same thing.

You are precisely correct about the source impedance.
 
HiFiNutNut said:
What XO and slope? This is the question asked many times and nobody has given an answer, or the answer would be like "they are selected based on the result of measurements". Well it is true, but ideally some samples can be given to serve as a guide.

This points me to BW3 and LR4. However, since Dr. Geddes recommends using commercial subs, and most of the commercial subs support only 2nd order LP, I can't see how the job can be done.

Regards,
Bill


The Xo location can only be determined in-situ, but the slope can be decided. I would use 2 nd order even if I were designing the system myself. I have not found localization issues in any bandpass sub usage, but then these, acoustically, end up being 4th order LP. But that probably does not happen arround the cutoff since it would be unusual that the elctrical filter and the acoustic one were at the same frequencies. Thats why, to me, 4th order is too sharp if it designed such that all four poles are nearly the same frequency. Maybe thats why I like bandpass and 2nd order - eventual sharp cutoff, but more gradual at the cutoff.
 
Dr. Geddes,

I use sealed boxes and I guess I could shift the poles of the 4 order LP and make a BW2 simulating the rolloff of the bandpass box and another one standard BW2, which is the 2nd order XO you referred to.

So what is the usual bandpass rolloff frequency of your subs?

I guess that the precision of rolloff does not matter because it is virtually impossible to match the SPL and phase of the drivers with the numerous room modes. If random placement of subs works it simply suggests that none of the subs need to conform to any particular response and phase curves, but the random effects would make the final response flat.

I guess that a first 2nd order BW2 LP may be placed around 60-80Hz and a second 2nd order LR2 at around 175Hz, and that would probably work. If the XO points needs to be adjusted then it would be the first one.

What do you think?

Regards,
Bill
 
HiFiNutNut said:
Dr. Geddes,

I guess that the precision of rolloff does not matter because it is virtually impossible to match the SPL and phase of the drivers with the numerous room modes. If random placement of subs works it simply suggests that none of the subs need to conform to any particular response and phase curves, but the random effects would make the final response flat.

What do you think?

Regards,
Bill


I think that this is well said.

I wish that others were as perceptive and had this level of understanding. Its hard for "deterministic" people to understand the world of "random". It took the world nearly 50 years to accept Quantum Mechanics. Many still don't accept it. Einstein never did.
 
I had a very interesting experience. A parade when by while I was by the side of the road. lots of drums of various sizes. The sound was not particularly loud to me, but there was one type of drum that made my chest vibrate along with it. It was not the drums with the lowest frequency causing this. I have never noticed this kind of experience in a room even when I played back different individual frequencies.
 
gedlee said:

With "mains to 40" and "subs below that" you won't get any multi-sub benifits at all. There would be no point in doing it.

... which leads exactly to my question: Will "mains to 40" and subs (positioned as suggested by your method, adjustable level, LP filter and phase) below ~100Hz work?

I'm wondering about this because with full-range mains you already have two "subs" in the system, but you can't move them and there is no level, filter or phase you can adjust.

In other words: If my main speakers are full-range, should I add a HP filter (say 80Hz, 2nd order) to the mains for a multi-sub arrangement or not?

Regards,
Thomas
 
AX tech editor
Joined 2002
Paid Member
Pan said:



A a short transient signal that will excite resonances.

In responce to Janneman:



/Peter


I think we're straying from the issue. My response was to "messing around with subwoofers will destroy localisation because it will mess up lf transients". My point is that localisation depends on the mid/hf parts of the lf signal, and those signal parts are routed to the mid/hf drivers. Therefore, messing with subwoofers will not mess up localization.

It is beyond me how any knowlegable person here should suggest to low-pass a transient and use that as an argument in this discussion. Lowpassing a transient essentially gets rid of it, of course.
Anyways, unless my original argument is addressed, I don't think I can add anything to it.

I am learning a lot from the rest of this thgread, though.

Jan Didden
 
mat02ah said:


... which leads exactly to my question: Will "mains to 40" and subs (positioned as suggested by your method, adjustable level, LP filter and phase) below ~100Hz work?

I'm wondering about this because with full-range mains you already have two "subs" in the system, but you can't move them and there is no level, filter or phase you can adjust.

In other words: If my main speakers are full-range, should I add a HP filter (say 80Hz, 2nd order) to the mains for a multi-sub arrangement or not?

Regards,
Thomas


I'm a little unclear on this myself. I guess it would depend on the inherent Low freq roll off of the mains. It almost sounds like sealed mains is a given for proper integration into a Geddes system, not a bad thing IMO. No need for the HP, just a natural overlap of the subs low pass and the natural roll off of the mains ??
 
youngho said:
If you place a subwoofer at the node for a mode and have it play the frequency for that mode, you will hear that frequency, but the mode will not be activated. The relative level will be low because it's not reinforced by standing waves, but you will still hear it.

Is someone here able (and willing) to simulate a 2-dimensional wave superposition model of a source transmitting energy at a modes null?

Best, Markus
 
Markus - not an easy thing to do.

The rest - modal effects are significant well up to 150 Hz. in most rooms. As long as there are modal effects we need multiple sources to "tame" them. The mains are some of those sources and yes they do work better I have found as closed box (which is why thats all I sell anymore.) But its also correct that the mains LF sources are seldom if ever in desirable locations and never phased or leveled (as pointed out above). So while they might help at the very LFs they are not really all that effective. So the subs need to be "blended" into the existing sound field created by the mains. This is why I suggest overlaping and not HPing the mains. A smooth blend smoothes out the response and the main retain the imgae cues for propoer localization of the instruments. Done properly one never localizes on the subs only the mains as they should.

In short, no I never HP the mains.
 
Looking again at the equation for pressure we can make several observations.

An externally hosted image should be here but it was not working when we last tested it.


First, recognize that the pressure at any frequency, as determined by the value of k = w/c, is the sum
of contributions from all modes.
Second, the term Kn squared - k squared in the denominator controls,
in part, the magnitude of the contribution of each mode to the pressure at a given frequency. When Kn
is close to k the frequency is very close to a resonant mode and a large peak in the response may
occur. When Kn is much greater than k the mode contributes little to the pressure at that frequency.
The form of the denominator indicates that each mode behaves as a 2nd order low pass filter with cut
off frequency defined by the real part of Kn and Q defined by the imaginary part.

Examining the numerator we see that the contribution from each mode is dependent on the product of
the magnitude of the eigenfunction (Psi) for each mode at r and ro. This tells us that the placement of the
source is no more or less important than the placement of the listening position. If the eigenfunction
value is zero at the listening position then it makes no difference whether the mode is excited by the
source or not. It will not contribute to the sound pressure at the listening position.

Room response simulation tool
 
Markus,

the phase would be identical on right vs. left in your pic.

edit: also a two dimensional model has limited value since in reality we use min. two sources in a 3-D space for stereophony.

Plus, don't forget all the virtual sources due to the "mirror effect" of the surrounding surfaces in a room.


/Peter
 
John

No need to argue again about HP, LP or BP other than to repeat that this all depends on how one defines the source. It can be any one of the three depending on this definition.

Also notice that the contribution for any mode, no matter how high in frequency, is never zero unless it is at a node. But, I repeat, kn is in general complex as are the modes and as such in a real room there are no nodes where the contribution is exactly zero.

The situation is not so simple that broad reaching generalizations can be made. Put a source at any spot in the room and the listener at any other spot and play any fequency. It will always be heard - its never identically zero. It can be very small, but never zero. Its not uncommon to see deep notches in the response with a very low response sometimes only a fraction of 1% of the neighboring frequencies, but its never zero. When talking about these very small values of excitation the obscure aspects of the problem like the complex modes and the evanescent wave are all on this order of importance and have to be considered.

That said, its virtually meaningless to get into this kind of detail or discussion as it has no bearing on the real problem at hand.