Sound Where You Want It.

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I was studying some of Danley's posts, and I came up with a neat trick.

This isn't my idea - I stole it from him. But it has some interesting applications to a variety of loudspeakers. (home audio, car audio, portable speakers, etc...)

The idea of this speaker is to put sound where you want it. For instance, to take a large loudspeaker, and relocate the apparent source of the sound.

I got the idea from a few posts that Danley did in regards to acoustic levitation, and how the midranges work in a Unity or Synergy horn. But this 'trick' can be used in a variety of speakers, not just horns.

If I understand what he wrote correctly, we can basically 'relocate' the apparent source of a loudspeaker, as long as the duct that the sound passes through is no larger than one quarter wavelength.

For instance, if you wanted to take a speaker that played to 1500hz, the largest dimension would be 5.66cm. (speed of sound / 1500hz / 4)

Here's a "real world example."

An externally hosted image should be here but it was not working when we last tested it.
Here's an array in someone's car. (not my car.) That Dayton dome will play to 500hz. How can we relocate a midbass up there? In other words, lets design an enclosure which allows us to take our midbass, put it somewhere else entirely, then 'duct' it where we want it.

Here's how we can do this:

First, we have to decide how *high* the woofer is going to play. For instance, if we're crossing over to the Dayton dome, our crossover will be at about 700hz. (That's about as low as the Dayton can go.)

Instead of using a duct that will accomodate 700hz, let's go one octave higher. That's because the woofer will still be playing above 700hz, even with a crossover.

For this exercise, I'm going to use a duct that will accomodate 1500hz. That means our maximum dimension is 5.66cm.

Let's make the duct cylindrical, for simplicity's sake. (It doesn't have to be cylindrical; any shape will do as long as no dimension exceeds 5.7cm.)

A cylinder with a diameter of 5.66cm has an area of 25.19cm. So that's our maximum duct size, if we want to go to 1500hz.

Now that we know our duct is 25.19cm, we need to pick a woofer. The reason that we have to figure out the duct first is that we can't just cram the output of any ol' woofer into our duct. For instance, if we tried to cram the output of a 30cm woofer into a duct that's less than 6cm in diameter, it's not going to work very well.

An externally hosted image should be here but it was not working when we last tested it.
The Dayton RS100 is a good candidate. Not too big to work with our duct, but big enough to get down to 100hz.

ducted1.jpg
Here's the response of the Dayton in a 1.1 liter sealed box. That gives us an F3 of 126hz. A good match for the Dayton dome, and the addition of this midbass gives us more than two more octaves of output.
ducted2.jpg

ducted7.jpg
If anyone is playing along at home, here's my model of the sealed Dayton. I did this in hornresp, because hornresp can model ducts. There's an extremely short duct at the end of the sealed box. It's so small, it's acoustically invisible, you can delete that stub and the response won't change.

ducted3.jpg
Here's the same woofer, but now running through half a meter of duct.
ducted4.jpg
Here's a comparison between the sealed woofer running through the duct, and the sealed woofer without a duct. Sans-duct is grey.

ducted6.jpg

ducted5.jpg
If you want to try this yourself, here's the hornresp model, and what the ducted woofer looks like.

Now at this point, you're probably looking at that frequency response graph, and saying "Bateman has lost his mind, that response graph looks terrible."

First off, you're right, my forum is called "audiopsychosis" for a reason - I'm crazy about audio. But that's besides the point - there's a ton of things you can do to flatten and smooth out the response of our ducted woofer.

Here's a few ideas you might try:

  • If you offset the woofer in the duct, the response smooths out a lot. For instance, if you flip the woofer on it's side, so that the woofer is firing into the *side* of the duct, the response is much smoother. It's the same idea as what we do in a tapped horn, where moving the woofer up or down the duct smooths the response. Basically play around with the dimensions in Hornresp until you get something that looks good.
  • In my experience, Hornresp dramatically exaggerates the peaks in a horn. For instance, if the hornresp model predicts a peak of 6dB, in the real world it's often half that. If you actually built this speaker, I wouldn't be surprised to find that the 16dB peak at 520hz is more like 8dB "in the real world."
  • Tapering, expanding, or stuffing the line will reduce the amplitude of the peaks. This is all standard transmission line stuff, just play with the dimensions in hornresp until it looks good.
  • A roundover at the throat and mouth of the duct works wonders, just like it does on a subwoofer. Hornresp can model both.

In summary, you can juggle the variables to smooth this out. In the span of fifteen minutes I was able to push the peak above our crossover point of 700hz, and reduce the size by half. A bit of EQ or a steep crossover could flatten it further.

Now this probably looks like a tremendous amount of work for a 10cm woofer, but keep in mind, you can array these. For instance, it's not physically possible to 'stack' two drivers in the same point in space, but we can easily 'stack' the output from two drivers using this trick.

The key to all of this is that the sound doesn't radiate until it reaches the end of the duct. That's the trick - it allows us to put sound where we want it.

Here's where I got the idea from:

PSW Sound Reinforcement Forums: LAB: The Classic Live Audio Board => Mid cone driver slots and holes

Think of a vented box and woofer, one has several things;
First, obviously you have a source that has two phases and one has a 2nd order acoustic low pass filter attached to the rear side of the radiator.
That low pass filter is made from the air in the box, a compliance volume or acoustic spring force (which electrically appears as a series inductance) and the mass of the air in the port (which looks like a Capacitor to ground), hence a 2nd order system.
You can see that relationship if you imagine a ruler clamped to a table, you can see its resonant frequency goes down if you add mass (add a penny at the end / make a longer port or reduce its area) or make the spring weaker (make the ruler longer / or box larger).

When the radiators two opposite phases add, they cancel each other out as they are always 180 degrees apart and of equal magnitude.
The sealed box simply contains one half of the signal allowing the other half to radiate away unhindered.
The low pass filter in a vented box (the vent + box volume) is in the form of a Helmholtz resonator and “at resonance” is an inverter, in other words introduces a phase shift which makes the port radiation “in phase”, additive with the front radiation.
While one doesn’t normally think about this as a “low pass filter” it is. Progressively less comes out of the port as the frequency rises above resonance.

As one lowers the frequency from the resonance (normally the low corner in a vented box), one finds that the phase shift imposed by the L and C reduces towards zero and the sound coming out of the port reaches the same phase angle as the rear radiation.

At this point, this is the “pass” region and the rear radiation increasingly cancels out the front side as the frequency lowers and the vented box has twice the roll off compared to a sealed box.
At a point WAY below Fb, the cancellation is complete.
Above resonance, the output from the port decreases, reverting to a sealed box, which contains the anti-phase rear radiation.

When one has a normal horn, one finds that the radiators rear volume is a sealed box.
At the high frequency corner, one finds that another acoustic “low pass filter” is present although potentially not as obvious.
Here, some portion of the air in the horn throat acts like a lump of mass like a port and some portion of the air between the radiator and throat acts like a compliance or spring, forming a low pass filter.
By sizing the L and C relative to the resistances, one can often extend the hf response by having a suitable low pass filter. As with the vented box, the ultimate roll off is steeper than an unaided alignment.
Here the only output is what comes out of the port (and drives the horn) so to speak.

In the Synergy and Unity horn boxes I designed, I use that “low pass” filter effect to attenuate the distortion products that all drivers produce. The distortion products are 2,3,4,5,6, ect times the fundamental frequency and so to the degree these fall on the rolled off part of the acoustic filter, they are attenuated.
The filter here is made of the volume trapped under the cone and the mass of the air in the port and throat. I don’t use phase plugs here.

Also, when sound is introduced into the horn at some point forward of the apex (such as the cone drivers in our horns), one finds that the upper frequency limit is also set by an additional “low pass” filter effect caused by internal self cancellation.
When the wavelength is short enough (frequency high enough) the sound that went to the pointy end, bounces back and arrives out of phase with the driver pressure.
One finds (as you raise the frequency) that you eventually have a BIG cancellation notch when that “driver to dead end” distance is about 1/4 wl.
The two low pass filter effects strongly attenuate above band energy from the cone drivers and helps make the distortion especially low.
It was that notch, or pondering that notch that made me wonder about and then try what became the Tapped horn.
I thought what happens if I substitute a source of the opposite phase for that reflection? (a source which was present in the back side of the radiator), then they add instead of cancel. Some considerable fiddling in the computer eventually resulted in boxes that work better than similar sized normal bass horns using this new principal.

Anyway, up to now, the only function a phase plug has is to occupy an excess air volume that would have other wise made the acoustic “low pass filter” too low in frequency.

Once one is dealing with a radiator who’s dimensions are approaching the wavelength size, then the other function of a phase plug comes in handy.
The speed of sound governs how a pressure disturbance radiates away from its source.
If one has a radiator that is “large” acoustically and also has a single exit point, one finds that just like in the Synergy and Unity horns, one gets a deep cancellation notch when the difference in the two paths is 1 / 2 wl. The range of coherent summation is limited to the frequencies BELOW the region where cancellation begins.
This is like a pile of subwoofers, when the array is less than about 1/4 wl across, they all add together and feel “mutual radiation pressure” while a significantly larger spacing produces directivity and then lobes.

Here, a phase plug can be shaped so that the acoustic passages all have the same length or have a length appropriate to the desired exit wave front shape.
One big difference in the sound of compression drivers (IMO) after being eq’d flat is that many have a phase plug that produces a converging wavefront at the summation, while what one needs at the throat of a horn is a diverging wavefront. That “clash” can cause diffraction or interference, which produces Higher Order Modes that Earl Geddes describes.

So far as the Paraline as used in the VTC array and GH-60, this is an acoustic device which can be shaped to provide an exit wavefront that can be flat, a line source or diverge or converge, an astigmatic point source with positive or negative focal point..
It works by allowing the sound to expand radially between two plate that are too close together to support any reflected modes between them so only radial expansion takes place.

A correction slot who’s shape defines the exit wave front shape and who’s dimensions are small enough to allow the sound to bend around the corners, is placed in the radial path. The sound passes through the slot and what emerges on the other side is a wave that travels to the center from each side, bends around a corner and exist at a center slot having entered at a center hole at the rear. The VTC site had a nice graphic of the one they are using.

It probably sounds weird to suggest that you can bend sound without ill effect but you can when the acoustic dimensions are small enough. The difference as it is in the examples above is that keeping the difference in path lengths less than about 1/4 wl at the highest frequency of interest..
I used to work with 21KHz levitation sound sources and needed to place a microphone in the levitation furnace to monitor the source sound level.
Well, very very few things are “happy” at 1500 degrees C but I found that a Zirconium / Alumina tube with a 1/16 inch bore passed 20KHz sound out of the furnace to an external microphone with no problem.

Funny, the external microphone’s heatsink wasn’t large enough on the first one and the microphone melted.
Later fooling around (research) showed that a 3 foot long, 1/16 inch bore copper tube could be wound around a small coffee cup and not effect the sound passing to the microphone.

A constantly re-occurring theme in much of what I do is that many things depend on how large X is compared to the wavelength.
Anyway, I hope that makes some sense.
"
 
]I was studying some of Danley's posts, and I came up with a neat trick.

This isn't my idea - I stole it from him.
The idea of this speaker is to put sound where you want it. For instance, to take a large loudspeaker, and relocate the apparent source of the sound.

I got the idea from a few posts that Danley did in regards to acoustic levitation, and how the midranges work in a Unity or Synergy horn.

If I understand what he wrote correctly, we can basically 'relocate' the apparent source of a loudspeaker, as long as the duct that the sound passes through is no larger than one quarter wavelength.

Now at this point, you're probably looking at that frequency response graph, and saying "Bateman has lost his mind, that response graph looks terrible."

The key to all of this is that the sound doesn't radiate until it reaches the end of the duct. That's the trick - it allows us to put sound where we want it.
The short duct is fine for midrange, it creates an acoustical filter, and the ripple is above the passband.

A long duct sounds terrible in the midrange, and you won’t be able to eq the ripple in the passband to correct that.

Ever listened to your voice talking through a tube? Give it a try with a toilet paper roll or a paper towel roll pointed at a reflective surface, or have a subject talk for you.

Substitute your speaker, same effect.

Give it a try, report back from the real “real world”.

Art
 
The short duct is fine for midrange, it creates an acoustical filter, and the ripple is above the passband.

A long duct sounds terrible in the midrange, and you won’t be able to eq the ripple in the passband to correct that.

Ever listened to your voice talking through a tube? Give it a try with a toilet paper roll or a paper towel roll pointed at a reflective surface, or have a subject talk for you.

Substitute your speaker, same effect.

Give it a try, report back from the real “real world”.

Art

Yikes Art, it was your Paraline experiment that inspired this!

The same ideas in the Paraline can work for midranges, you just have to change the dimensions.

And a few minutes in hornresp illustrates that this is a juggling act. The size of the woofer matters, the length of the duct, the VAS matters.
 
Weltersys : Would the pitch improve if you talk through multiple tubes?
QUOTE]
"Pitch" is the fundamental frequency, which would remain unchanged regardless of the amount of tubes.

Multiple tubes have been used to make "shotgun" microphones, they suffer from midrange coloration (comb filtering), but it may be worth a try if no other options are available.

Just tried talking through a handful of drinking straws, less peaky than a paper towel tube, but still not "hi-fi"...

Art
 
Yikes Art, it was your Paraline experiment that inspired this!

The same ideas in the Paraline can work for midranges, you just have to change the dimensions.

And a few minutes in hornresp illustrates that this is a juggling act. The size of the woofer matters, the length of the duct, the VAS matters.
The primary purpose of the Paraline HF coupler is it sets HF vertical dispersion to a specific divergence or convergence angle independent of the horn wall.
Multiple Paraline couplers can be vertically aligned, allowing for multiple HF units on a single horn without the air non-linearity problems associated with the usual “plumbing” manifold driver designs.
The Paraline also does appear to create peaks and dips, but they are at very high frequencies where the ear is less critical, though your “bat ears” might still notice ;^).

In the tapped horns, those peaks are outside the passband, and can be equalized out with little sonic impact.

What you propose appears to put nasty peaks and dips right in the critical midrange passband.

At any rate, give your idea a try, test, listen, report back your findings compared to the “normal” Synergy coupling.

Have fun!

Art
 
If light can travel hundreds of miles via a fiber optic cable, it should be possible to make sound travel have a meter over a duct.

I concur, this will *not* be a wide bandwidth device. If I could squeeze two octaves out if it, I'd be happy.

The intriguing part of this design is the ability to relocate the apparent source.

For instance, put a midbass a few centimeters from your chair, but have it appear half a meter in front of you.

If I understand Danley's quote in the first page, this should be do-able.

It seems to hinge on a few things:

  • The apparent source of a sound is not the diaphragm of the loudspeaker, it is the point where the soundwave can expand to a fraction of it's length. For instance, I have a bandpass subwoofer that's about a meter tall. Sixty hertz is 5.67 meters. When I stand near the sub, the sound doesn't emanate from where the driver is. The apparent source of the sound is the port. I believe this is because the 60hz wave is too large to expand until it exits the port. If I am incorrect about the apparent source of the sound, someone please tell me.
  • While the duct will introduce a delay, DSP can fix that. And I'm usually the last person to recommend DSP, in this situation, it makes sense. For instance, I see a lot of people trying to use DSP to make it sound like a speaker is somewhere it is not. For instance, in car audio they'll delay the left speaker so that it's in sync with the right. But this doesn't work well, because the SOURCE of the sound is still the original location. So the DSP is a bit of an improvement, but it also creates new problems. (IE, you're improving the driver's image at the expense of the passenger's.) But what I am proposing is quite a bit different, because the actual SOURCE of the sound has been moved.

It's also a bit interesting to realize that the apparent source of sound in a horn is frequency dependent, due to the same rules. For instance, very high frequency sounds will seem to originate near the throat, while lower frequencies seem to originate from a point closer to the mouth.

 
If light can travel hundreds of miles via a fiber optic cable, it should be possible to make sound travel have a meter over a duct.

It's also a bit interesting to realize that the apparent source of sound in a horn is frequency dependent, due to the same rules. For instance, very high frequency sounds will seem to originate near the throat, while lower frequencies seem to originate from a point closer to the mouth.
There is no question sound can go through a duct for a meter, speaking tubes have been used for many centuries for communication over dozens of meters, they are still in use on warships, and still sound like speaking through a tube.

Airplanes use(d) small tubes for communication too, even though electronic communication is now primarily used, the ductwork is still in place on many commercial airliners. I recall the sound from those air tube headphones...

If you have an application that for some reason requires location of a speaker remotely from the sound exit, other than sound quality, there is nothing to stop you from piping it through a tube.

You are correct that the LF apparent source of sound is the duct or mouth exit, it is the position from which the inverse distance law applies, though the “time of flight” starts from the driver location.

Art
 
Last edited:

In the event that some of the people reading this thread would like to give this a try, here's how you go about simulating it:

  • hornresp can model this just fine. Just model the loudspeaker as a front loaded horn with a sealed back chamber. Only difference is that the front horn can be negligible or nonexistent. We're focused on the coupling chamber, which is the volume of air between the cone and the horn.

    Hornresp can figure out the length of the duct for you. Here's how you do it:
    a: Figure out the maximum diameter of your duct. That diameter is one quarter wavelength of your highest frequency. For instance, if your maximum frequency is 1000hz, you'd do it like this:
    (34000 cm per second / 1000hz / 4) =
    (34 cm / 4) =
    8.5cm or 3.34"

    Now that diameter is going to determine the area of your throat chamber, aka "ATC" in Hornresp.

    You can have a square duct, or a rectangular duct, or a circular duct. Just keep in mind that the maximum dimension will be 8.5cm in this case.

    So if your duct is square, your "ATC" will be 8.5cm^2, or 72.25cm

    The *volume* of your throat chamber is simply based on the length. If the length of your duct is 100cm, then your vtc will be 7225. (72.25 * 100)

    In the diagram, you'll notice that Hornresp is figuring out the length for you.
  • You'll likely notice an improvement in frequency response if you round over the exit from your duct. You can do that using fields S1, S2, S3, etc... Just as if there was a very short horn.
  • The weight of the air in the duct interacts with your woofer. Due to that, you'll likely get better results if the driver is smaller. I wouldn't try to get 500hz out of an 8" woofer, but you could likely do it with a 2" or a 3" driver. Keep in mind that if you need a lot of output, these are easy to array. Four 3" drivers will likely work better for you than a single 8. Low mms helps too.

 
Not to dissuade any experimentation, but when it is no problem to mount multiple drivers directly on the horn throat, what advantage do you see in remotely coupling them through a duct that severely limits the usable bandwidth, and does not attenuate out of band peaks as the throat coupling does?
 
Not to dissuade any experimentation, but when it is no problem to mount multiple drivers directly on the horn throat, what advantage do you see in remotely coupling them through a duct that severely limits the usable bandwidth, and does not attenuate out of band peaks as the throat coupling does?

The part of this puzzle that I didn't understand four months ago is that pathlength is very important with these devices.

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


For instance, let's say you take the speaker above, and you replace that 10cm pipe with a hundred 1cm pipes. As I understood it a few months ago, the pipes of narrower diameter would pass high frequency energy better, since the wavelengths can form in the large pipe, but can't form in the small pipe.

The thing that I didn't understand then - which I understand now - is that pathlength doesn't care about wavelength. Even if the duct is so small that the wave can't form, the air molecules still need to exit in-phase.

But, again, the Paraline satisfies this goal nicely, because the sound radiates radially. When I first saw it, it seemed like such an odd device, because the sound is bent 180 degrees. But now that I understand the radial expansion, it makes sense.
 
Methods to damp the resonances?

Since the original goal of the tube is to deliver sound at a distance from the actual driver, we do not need to worry about the gain (?) that a quarter wave tube would have. On the other hand, can we damp the unwanted resonances? If so, would this not give much smoother frequency response? I have only read of this done for the mouth of a horn (this is distinct from the Geddes foam thing for HOM)... here is a link

Factors Affecting Sonic Quality of Mid & HF Horns & Waveguides

Sure enough, a round tube is a waveguide, and the question is would a "towel cowl" work for Patrick's idea?
 
I am just speculating, so I don't even have a duct right now :) Probably a flare would work, or any way to smooth the impedance. Of course, Patrick's original goal (to have a small midrange port where he wanted it) might be diluted if it has to have a Le Cleac'h flare and/or a turban looking surround upon it :)
 
If light can travel hundreds of miles via a fiber optic cable, it should be possible to make sound travel have a meter over a duct.

This sounds so trivial, but it's not. Light in a fiber is very narrow-band. You get wider-band fibers (multimode), but they suffer more over long distances. Multimode also means more than one, not x to y wavelength. If you had to put a wide spectrum of light through a fiber, you would also see this peaking, and likely have absolutely nothing for many wavelengths.

Understanding Wavelengths In Fiber Optics

It might be worthwhile reading about wide-band transmission lines in electromagnetic theory. It's not a simple matter. :)
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.