Exactly my thoughts.
We are not moving either sub through more displacement by removing one sub - both are on their own amps at 8 ohms - the displacement is the same for each subwoofer.....
So we are effectively seeing an increase in SpL due to an increase in VD - double the subwoofers!
I just can't see this any other way.
Which implies a concurrent increase in max SpL?
We are not moving either sub through more displacement by removing one sub - both are on their own amps at 8 ohms - the displacement is the same for each subwoofer.....
So we are effectively seeing an increase in SpL due to an increase in VD - double the subwoofers!
I just can't see this any other way.
Which implies a concurrent increase in max SpL?
I think people find this mentally vexing because of isobaric modelling and an assumption that the air between the subwoofers is fixed/confined within a sealed volume - it is not.
This leads down the rabbit hole of how can both subwoofers be producing output if they are closely spaced and opposed? But I feel this is just the visual image which confuses us, and we need to remember these are two naked drivers unbounded by a enclosure and therefore not behaving in an isobaric way.
This leads down the rabbit hole of how can both subwoofers be producing output if they are closely spaced and opposed? But I feel this is just the visual image which confuses us, and we need to remember these are two naked drivers unbounded by a enclosure and therefore not behaving in an isobaric way.
Here is a mental exercise for people - imagine the clamshell compound subwoofer configuration I have made.
The spacers are currently 15cm apart....and people feel they can't possibly be both producing useful bass.....they are in effect cancelling each other out......
Now imagine the spacers are 100cm apart.......will the two subwoofers produce useful bass now or still cancel each other out?
Now imagine the spacers are 300cm apart.......i.e. two 15" naked drivers at other sides of my listening room - mounted in a clamshell configuration with 300cm long spacers the width of the room - will they be cancelling each other out now, or are they operating completely separately and therefore sum?
Now consider does a 6M long wavelength differentiate between 15cm, 100cm or 300cm?
SL demonstrates it does not - as long as the distance is small compared to the wavelengths involved.
The spacers are currently 15cm apart....and people feel they can't possibly be both producing useful bass.....they are in effect cancelling each other out......
Now imagine the spacers are 100cm apart.......will the two subwoofers produce useful bass now or still cancel each other out?
Now imagine the spacers are 300cm apart.......i.e. two 15" naked drivers at other sides of my listening room - mounted in a clamshell configuration with 300cm long spacers the width of the room - will they be cancelling each other out now, or are they operating completely separately and therefore sum?
Now consider does a 6M long wavelength differentiate between 15cm, 100cm or 300cm?
SL demonstrates it does not - as long as the distance is small compared to the wavelengths involved.
Last edited:
Engineers like SL probably found limited use of this since it’s so inefficient and could just as well be replaced by a few pieces of wood. Makes litte economical sense. I’m still curious about the possible decreased distorsion and more pronounced 90 degree null, though.
Agreed - my experiments so far do seem to bear out the decreased distortion, but polars would need me to move a speaker outside to do it properly and frankly I can't be bothered
with 15cm distance I would say the air around both woofers is still "coupled", ie, the moved air between both speakers is the same. At 300cm there is more air being moved. Ok, still in phase if wavelenght is 6m, but at the midpoint between both woofers the air flow will be less, as airflow decreases just as SPL decreases when one gets further away from the speaker.
I think Charlie commented this already, and I would also be curious about the polars for such arrangement. I am thinking that maybe there is more beaming, and less output to the side in comparison to a single driver.
other thing is that, looking back at my own experiment, one would have to better define the acoustical center of the single and compound woofer, and then measure from that distance. even though, when measuring far away from the speaker these few cms of differencs in acoustical center should not matter.
for me it is mostly curiosity now, and I have no direct application for the clamshell construction (speakers limited to <20cm deep).
This is what I am trying - perhaps badly - to say, that there simply is no 'coupling' - this is a visual falsehood.
There would only be 'coupling' of the subwoofers if the space was bounded.
This, I believe is the crux of what is confusing things.
Visually they look like they are behaving in partnership - somehow interacting destructively more than they would if they were side by side or separated by a larger distance.
In reality they are behaving as two separate subwoofers even at this distance - as they would side by side, or with larger spacers as per my thought experiment.
There is simply no 'coupling' - there is no other way to explain the consistent 6dB increase in Spl with the same excursion of each driver.
This is also what SL demonstrates mathematically - the two do not interact in that way.
Again, SL's modelling is of this very configuration, "2d2" is the 15cm spacer distance:
"The model predicts that the SPL at very low frequencies is merely the sum of two dipoles with spacing D = d1. The separation 2d2 between them has no influence on the total output as long as it is small compared to the wavelength of radiation."
So he is saying that the spacer length has no effect on total output - whether 10cm, 20cm, or 100cm....
I see no compounding effect other than summing two dipoles, but the two baffles might as well be placed next to each other."
He is saying - 'yeah they sum together in a clamshell, but you may as well put them side by side'
He goes on to say - all this is very good and well - but why not just put one subwoofer in a baffle!!!!
"The model predicts that the SPL at very low frequencies is merely the sum of two dipoles with spacing D = d1. The separation 2d2 between them has no influence on the total output as long as it is small compared to the wavelength of radiation."
So he is saying that the spacer length has no effect on total output - whether 10cm, 20cm, or 100cm....
I see no compounding effect other than summing two dipoles, but the two baffles might as well be placed next to each other."
He is saying - 'yeah they sum together in a clamshell, but you may as well put them side by side'
He goes on to say - all this is very good and well - but why not just put one subwoofer in a baffle!!!!
The fact that there are 2 drivers isn't key here - you could imagine a single driver with single motor, but 2 cones, spaced by the same distance (it would look bizarre). (The cones would each be half the mass, and total compliance the same as before, to maintain the same resonant frequency.)
There may be small differences due to slightly different air load, but excursion would otherwise be the same, thanks to the fact that the same total cone mass was being moved.
There may be small differences due to slightly different air load, but excursion would otherwise be the same, thanks to the fact that the same total cone mass was being moved.
you know the stuff, maybe smoke?, they use to see air flow in a wind tunnel? Maybe we should get that between both woofers and see how it behaves during some bass notes
In an earlier post I wrote "Still, it appears that there is still the 6dB increase, so there is more to SPL than moving air?! " So yes, I agree, and I am happy, that there is the 6dB gain with same woofer excursion in a small footprint. It is just that I do not see how such close mounting could move twice the air mass, as intuitively the same air mass is moved around by both drivers. Still, the +6dB is there, and that is what eventually counts.
I am also looking forward to measurements by ayebee! Someday I may even try to measure it myself, after the other 33 projects are done...
I know the feeling regarding multiple projects Erik!
Now funny you should mention smoke.......I have a smoke/haze machine in the garage - long story involving one of my kids school discos!
I am not sure how we could do it - but I could put it inbetween the clamshell with a spotlight on and try to record the patterns?!?!?
Now funny you should mention smoke.......I have a smoke/haze machine in the garage - long story involving one of my kids school discos!
I am not sure how we could do it - but I could put it inbetween the clamshell with a spotlight on and try to record the patterns?!?!?
And I am focusing on SL's bolded and double underlined "2d1=D"
To me that says that virtual baffle of compound radiators is same as H baffle with same pathlength (which determines axial dipole null)
This means that compared to a single nude driver, compound gets a virtual baffle which gives more spl to low frequencies and shifts axial null lower. Max spl at axial dipole peak hardly gets higher.
Next pic comes from Tim Mellow, at Rudolf Finkes's homepage Dipolplus - Alles über offene Schallwände
Yes, we need more tests and measurements, also off-axis...
Last edited:
Perfect! So we agree Juhazi! I must have been misunderstanding you - my apologies!
The 'virtual baffle' effect increases SPL and consequently max SpL too just as a 'real baffle' would. (more SpL for a given x-max per driver)
It is simply a slightly less efficient version of a 'proper baffle' without the space constaints, as I have contended all along!
So for me - rather than build a 'real baffle' I have used two drivers to create a virtual one - so my speakers are smaller, and I can mount them in the slings.
A win, win.
It's great to see this thread, and the questions raised by the compound dipole, getting attention. It's definitely an interesting phenomenon, and not all that intuitive on the surface.
I think Juhazi is on to something when he talks about a "virtual baffle". SL's conclusion that the compound system behaves like an H-frame of D=2*d1 hints at this. But it is interesting that this has nothing to do with the spacing (d2) of the drivers (at least at low frequencies). This means as you separate them, the SPL will remain the same in terms of the LF output. That is quite different from a dipole, where as you separate the sources of front and rear the LF output increases.
Also, the question of the acoustic center(s) is a very interesting one. There may be some non-intuitive behavior there as well. For example, as you separate the two drivers in the compound dipole, are you getting something like two monopole acoustic centers that move with the drivers, or does the system behave as if it has a single central acoustic center between them? This would have some ramifications in terms of the phase of the front and rear sources.
Finally, I was wondering to myself whether one could create such a compound dipole with two drivers, face to face, mounted on a planar baffle of some size, as a way to effectively double the baffle size. That would be interesting. I'm not sure if a "gap" is required between the drivers or not.
Unfortunately answering most of these questions probably requires measurements, and these are not easy to make at low frequencies. Maybe ayebee will try with smaller drivers to see how that pans out.
I think Juhazi is on to something when he talks about a "virtual baffle". SL's conclusion that the compound system behaves like an H-frame of D=2*d1 hints at this. But it is interesting that this has nothing to do with the spacing (d2) of the drivers (at least at low frequencies). This means as you separate them, the SPL will remain the same in terms of the LF output. That is quite different from a dipole, where as you separate the sources of front and rear the LF output increases.
Also, the question of the acoustic center(s) is a very interesting one. There may be some non-intuitive behavior there as well. For example, as you separate the two drivers in the compound dipole, are you getting something like two monopole acoustic centers that move with the drivers, or does the system behave as if it has a single central acoustic center between them? This would have some ramifications in terms of the phase of the front and rear sources.
Finally, I was wondering to myself whether one could create such a compound dipole with two drivers, face to face, mounted on a planar baffle of some size, as a way to effectively double the baffle size. That would be interesting. I'm not sure if a "gap" is required between the drivers or not.
Unfortunately answering most of these questions probably requires measurements, and these are not easy to make at low frequencies. Maybe ayebee will try with smaller drivers to see how that pans out.
Also, if the effect of a larger "virtual baffle" is indeed true, that may suggest a higher max SPL as well.
My thinking goes something like this:
This should be very easy to test/measure. You will need 2 channels of amplification. The input power to a driver controls its cone excursion. Set up a single driver as a dipole, with a duplicate driver available. Power the single driver dipole woofer and measure the SPL at some distance. Add the second dipole driver to create a compound dipole. Power both drivers with the same input signal so that both receive the same amount of power. Re-measure the SPL. If it has increased by 6dB, then we know that for a given cone excursion the compound dipole has 6dB more output, and therefore if the input power was increased so that Xmax was reached there would be 6dB more output.
I think bushmeister and others may have already done these experiments for us.
My thinking goes something like this:
- If we first think of a plain dipole we know that as the distance between front and rear sources is doubled, the SPL at low frequencies will increase by 6dB. This is easily confirmed by modeling.
- The doubling in SPL is due to a decrease in dipole cancellation. In other words, for a given cone excursion, the SPL at low frequencies has increased.
- This would mean that for excursion=Xmax, the SPL will also be increased, and is therefore an increase in max SPL.
This should be very easy to test/measure. You will need 2 channels of amplification. The input power to a driver controls its cone excursion. Set up a single driver as a dipole, with a duplicate driver available. Power the single driver dipole woofer and measure the SPL at some distance. Add the second dipole driver to create a compound dipole. Power both drivers with the same input signal so that both receive the same amount of power. Re-measure the SPL. If it has increased by 6dB, then we know that for a given cone excursion the compound dipole has 6dB more output, and therefore if the input power was increased so that Xmax was reached there would be 6dB more output.
I think bushmeister and others may have already done these experiments for us.
Last edited:
Yeah, thanks for doing that!
Here's a couple additional thoughts of mine about an important difference between the compound dipole and an H-frame: resonances.
The H-frame encloses the front and rear outputs in short "tunnels". These tunnels give rise to a quarter wave resonance, the frequency of which depends on the depth of the tunnel, and its Q depends on the aspect ratio (depth to width) of the tunnel. The H-frame is able to reduce the visual size of the baffle by folding it back on itself on both sides.
A compound dipole can be formed with planar or nude drivers. But neither of these will exhibit any kind of resonance, since no air is enclosed inside a "tunnel".
Unless the H-frame has a very low aspect ratio (e.g. width >= depth) the tunnel resonance typically limits the bandwidth of the driver to below the tunnel resonance frequency. In that regard, the compound dipole would be able to (potentially) be used to a higher frequency compared to an H-frame.
The usable upper frequency of the compound dipole is likely to be a function of the separation (d1 in SL's figure) of the two drivers. When d1 grows large, there will be a point at which the propagation delay due to d1 becomes significant WRT the wavelength, and the sources are no longer "acoustically close" and the arrangement acts like two distinct dipoles instead of a compound one. Since d1 has no impact on the low frequency behavior, that would suggest that placing the two drivers close together would provide the widest possible bandwidth (e.g. highest usable upper frequency).
This is just a hypothesis on my part that should be confirmed with measurements.
Here's a couple additional thoughts of mine about an important difference between the compound dipole and an H-frame: resonances.
The H-frame encloses the front and rear outputs in short "tunnels". These tunnels give rise to a quarter wave resonance, the frequency of which depends on the depth of the tunnel, and its Q depends on the aspect ratio (depth to width) of the tunnel. The H-frame is able to reduce the visual size of the baffle by folding it back on itself on both sides.
A compound dipole can be formed with planar or nude drivers. But neither of these will exhibit any kind of resonance, since no air is enclosed inside a "tunnel".
Unless the H-frame has a very low aspect ratio (e.g. width >= depth) the tunnel resonance typically limits the bandwidth of the driver to below the tunnel resonance frequency. In that regard, the compound dipole would be able to (potentially) be used to a higher frequency compared to an H-frame.
The usable upper frequency of the compound dipole is likely to be a function of the separation (d1 in SL's figure) of the two drivers. When d1 grows large, there will be a point at which the propagation delay due to d1 becomes significant WRT the wavelength, and the sources are no longer "acoustically close" and the arrangement acts like two distinct dipoles instead of a compound one. Since d1 has no impact on the low frequency behavior, that would suggest that placing the two drivers close together would provide the widest possible bandwidth (e.g. highest usable upper frequency).
This is just a hypothesis on my part that should be confirmed with measurements.