"Make the floor bounce / get back / give me room"*
In this thread I'm going to document some ways to make your subwoofer directional. Some of these are well known, but I think I came up with a way to do it for free which I've never seen documented.
So, first off, why do we want directional subs? If you haven't heard a set of directional subs, RUN don't walk and go find some. For a basshead like me, directional subs are probably one of the neatest audio inventions you'll ever hear. I love me some waveguides, but for my money, directional bass makes an even bigger audible difference.
The first directional subs that I heard were dipoles at audio shows, but if you want the full effect, go to a concert with a big cardioid setup and lots of power on tap. The feeling is basically the same one you get when you're on a rollercoaster and it drops at 60 miles per hours. It's a rush.
There's some other reasons to do directional subs:
1) If you live in an apartment or condo, a directional sub puts more energy into the seats and less into the walls. This isn't a small difference; it can be as much as 10-20dB. That's a huge difference; 120dB in an apartment will get you evicted but 100dB is a difference of night and day.
2) To my ears, cardioids are 'punchy.' They hit you in the chest. My hypothesis is that directional subs have a wavefront that's closer to ideal, whereas an array of omnipoles tend to interfere with each other a lot. Basically *all* arrays interfere with each other, the question is whether we use that interference for pattern control, or whether we ignore the problem and leave it up to chance.
Before we look at my 'free' directional sub, we have to understand what makes a sub directional.
Basically, as you array multiple drivers, you change the shape of the wavefront. In this pic of two subs separated by one quarter wavelength, you can see that the wavefront has gone from circular to oval shaped.
An oval shaped wave is actually the opposite of what we want, because it basically makes the subs directional... But 90 degrees away from us! The array is directional towards the sidewalls.
The U-Frame subs documented by John K offer cardioid radiation**, with no added cost.*** The way they work is fairly simple. The sound comes out of the sub in the front, and then the output from the sub in the back gives us a similar 'oval' shape as two subs. So the U-Frame is directional, and the output is focused towards the front of the sub. Ideally the length would be 1/4wl, so that the output from the rear reaches the front in-phase.
There's another way to do the U-Frame, by using two subs. Basically we'll replicate the behavior of the two sides of one cone with two sealed subs. So there's still two cone sides, but now those two sides are each in their own sealed sub. To simulate the effect of a u-frame, we have to flip the polarity on one. In the pic above, you can see that these two subs are going to interfere with each other. Note that the output from the bottom subwoofer is 180 degrees out of phase with the other sub, because the two subs are spaced a multiple of one wavelength apart. In this case, they're 2wl apart.
Something interesting happens if you position the subs one half wavelength apart. Now the bottom sub is IN PHASE with the top sub, due to the spacing. The bottom sub is still wired out of polarity, but physically moving the sub has positioned so that it's IN PHASE with the top sub.
If you look at the intersection between the wavefronts, you'll notice that it's actually 'egg shaped', and that's part of the directivity. We don't have spherical wavefronts any more, or plane wavefronts like you'd get with a horizontal array. We have an egg shaped wavefront that's focusing the sound towards us. If you look at the positive and negative wavefronts from the previous pic, you can see where they add constructively.
In the pic above, the wavefront is 100hz, and the subs are spaced 1.7m apart. (100hz is 3.4m long, so that's half a wavelength.)
Note that this behavior is basically identical to the concept behind a U-Frame. The U-Frame is spaced a 1/4wl because the sound has to travel *down* the enclosure, and then radiate back towards the front. The cardioid pictured above uses sealed subs, hence the distance difference.
* Bassnectar pic, and his big ol' cardioid array, courtesy of
** U-frame documented here: DIY-dipole-1
*** U-frame pic courtesy of Rudy from A comparison of OB, H-, M- and U-frame dipole
I was trying to figure out how to fit a cardioid array in my living room, and wasn't having much luck. My girlfriend will kill me if I put a subwoofer array in the living room that's two meters deep.
It occurred to me that tapped horns work the same way, but instead of using two subwoofers, one in polarity and one out of polarity, tapped horns use the two sides of a single cone, and the reversal in polarity could potentially be used to make the subwoofer behave like a cardioid array.
The frequency response of the tapped horn above illustrates this. You can see that there's a bump at 28hz. That bump is due to two things. A 1/4wl delay inside of the enclosure, combined with the fact that one side of the cone is 180 degrees out of polarity with the front.
The grey line in the second sim is the sealed response of the sub. See how the additional output gets higher and higher and higher as we get closer to 28hz? That's because the two wavelengths are maximally in phase at that point.
It's very similar to this:
Patrick,
Outdoors, or in rooms large in relationship to sub frequencies, cardioid or end fire arrays, both of which need separate amplification channels and delay to operate correctly at more than one frequency, do work.
In small rooms, they don't.
Measure your cardioid array in your house or garage, report back with the frequency response around the room, and various directions from outdoors, if we see any overall reduction exceeding 3 dB not accompanied by a similar increase elsewhere I will be quite surprised.
Art
As noted in my last post, I don't have room for a two meter deep array in my living room.
And a few things occurred to me:
1) The fundamental way that the U-Frame works is that we have two sources, out of phase with each other, that are separated by one half wavelength.
2) One could get similar results with two sources, separated by a full wavelength.
Therefore, if one does not way to pay for two subs, or build a big U-Frame, one could replicate the effect by using the walls to mirror the sound. By pulling the sub one half wavelength from the walls, the reflected wave is IN PHASE.
The pic above illustrates this. Similar to using two subwoofers in a line and pointed at the listening seat, but instead of using two subs, we're using a single sub and four reflections. (Floor, ceiling, and two adjacent walls.)
And a few things occurred to me:
1) The fundamental way that the U-Frame works is that we have two sources, out of phase with each other, that are separated by one half wavelength.
2) One could get similar results with two sources, separated by a full wavelength.
Therefore, if one does not way to pay for two subs, or build a big U-Frame, one could replicate the effect by using the walls to mirror the sound. By pulling the sub one half wavelength from the walls, the reflected wave is IN PHASE.
The pic above illustrates this. Similar to using two subwoofers in a line and pointed at the listening seat, but instead of using two subs, we're using a single sub and four reflections. (Floor, ceiling, and two adjacent walls.)
Yes, it's true that you need digital delay and separate channels of amplification to do a cardioid or end fire array that works at more than one frequency.
It's also true that a lot of home theater guys run their subs over 1-2 octaves of bandwidth. In my case, my subs cover 1.5 octaves of bandwidth.
In this narrow bandwidth, some clever spacing can substantially change the directivity.
I'll do that.
In the meantime, here's a measurement that John Kreskovsky did on a u-frame. In it we can see that there's significant rejection to the back:
An externally hosted image should be here but it was not working when we last tested it.
As far as an "overall reduction exceeding 3 dB not accompanied by a similar increase elsewhere", well that's what I'm after. A significant reduction of radiation to the sidewalls, with more energy directed to the couch.
An externally hosted image should be here but it was not working when we last tested it.
From my sims, it looks like the 'standard' config of having mains that go down to 40hz spaced a few feet apart gives us a pattern that's actually the *opposite* of what we want.
When spaced horizontally and played in phase, it creates a oval shaped wavefront. And as frequencies increase, there are nulls that are aimed right at the couch. (Basically to the left and the right of center.)
Flipping the low frequencies by 90 degrees improves things by focusing the energy towards the listener, and also pointing the high frequency nulls at the sidewalls.
Also, here's where I stole John's data from: NaO U-frame
So in your case, just like any PA application, you will need digital delay and separate channels of amplification.
Except that it won't work in a small room, where modal response swamps any directivity.
John's charts give no placement or room information, as far as I see they are simulations, not reality.
At any rate, I always look forward to your measured results, good luck on your latest ____adventure 🙂.
Happy hunting!
Art
Hi,
I read lots of hand waving and not much practical advice.
I completely fail to see your basic argument as two
omnipoles spaced any distance apart must produce
symmetrical wavefronts in any sensible analysis,
whether wired in phase or out of phase.
The only way to make it assymetrical is added delay.
Phase reversal is not delay at a given frequency.
I can't make any sense of a 90 degree or 180 degree
change in phase for one speaker in a stereo pair.
rgds, sreten.
I read lots of hand waving and not much practical advice.
I completely fail to see your basic argument as two
omnipoles spaced any distance apart must produce
symmetrical wavefronts in any sensible analysis,
whether wired in phase or out of phase.
The only way to make it assymetrical is added delay.
Phase reversal is not delay at a given frequency.
I can't make any sense of a 90 degree or 180 degree
change in phase for one speaker in a stereo pair.
rgds, sreten.
Last edited:
The reason the thread is titled "make your sub directional for free" is that you can create the delay by physically moving the sub.
If you have digital delay, use that, it gives you more granularity.
But the delay doesn't *have* to be digital, and if you skip the digital delay, you don't need extra amps.
Also, you can alter the phase by mixing subwoofer types. For instance, using a sealed sub at the top position and a tapped horn or vented box at the bottom position.
You can create an interference pattern by physically moving one sub away from another sub in relation to a listening position, but not a cardioid pattern, not outdoors, and certainly not indoors.
Back to the books, Patrick!
An interference pattern created by physically moving one sub away from another sub in relation to a listening position is not "directional".
If the SPL at the seat goes up and the SPL off axis goes down, you just made it more directional.
Waveguides aren't the only way to focus sound.
At one frequency, it goes up, another it goes down.
Making ripple is not the same as directionality, which you will find when you measure the results of your concept in various places around the room.
If you want to maximize level for the LP, and minimize level elsewhere, put the subs near (or under) the LP- simple inverse square law.
You have plenty of stuff laying around to use for your tests, just do it 🙂
As noted in my last post, I don't have room for a two meter deep array in my living room.
And a few things occurred to me:
1) The fundamental way that the U-Frame works is that we have two sources, out of phase with each other, that are separated by one half wavelength.
2) One could get similar results with two sources, separated by a full wavelength.
Therefore, if one does not way to pay for two subs, or build a big U-Frame, one could replicate the effect by using the walls to mirror the sound. By pulling the sub one half wavelength from the walls, the reflected wave is IN PHASE.
The pic above illustrates this. Similar to using two subwoofers in a line and pointed at the listening seat, but instead of using two subs, we're using a single sub and four reflections. (Floor, ceiling, and two adjacent walls.)
In the post above, my hypothesis was that you could make a sub more directional by placing it in a spot where the reflection from nearby walls would reinforce the on axis response.
The effect would be similar to what we get with a cardioid array (no Art, not identical, but similar) while requiring no additional subwoofer, no additional amp, and no digital delay.
To test this, I did the following:
I placed an 8" subwoofer approximately 4.5' feet from five of my walls. My ceilings are 9' tall, and I put the sub equidistant from the floor and ceiling. My living room is 10' deep, so the sub was 5' from each wall. And I pulled it 4.5' from the back wall.
The whole idea was to get a really symmetrical reflection off of the back wall, and have it sum in phase with the subwoofer on axis.
According to my sims, that should increase the on-axis output.
I would anticipate that the maximum gain would occur at 124hz. This is based on the fact that the sub is approximately 1.37m from five boundaries, and the maximum gain should occur at 1/2WL. (124hz is 3.74m long.)
Here's a measurement in my room.
1) the yellow line is on axis
2) the orange line is off axis by 15 degrees
3) the red line is off axis by 30 degrees
4) the purple line is off axis by 45 degrees
5) the grey line is the average of 0,15,30, and 45
Some observations:
There is an unmistakable reduction in output off axis. At 45 degrees off axis the output level at 132hz has been reduced by fourteen decibels. If anyone doubts that this works, just look at the graph; the reduction in energy off axis is unmistakable.
On the downside, the effect *is* very narrow in bandwidth; about one half of an octave.
At the moment I have one of my subwoofers 1.5 meters off of the floor, and 1.2 meters from the ceiling. I mostly did this because I read in the 'multiple small subs' thread that it's good to vary the vertical height. It looks like that might be a good solution to faciliate a directional sub like this. Basically get the sub off of the floor, and away from the back wall, but also vary the distances a bit to widen the frequency where the sub is more directional. (IE, instead of having 14dB of SPL reduction at 45 degrees off axis, perhaps one could widen the bandwidth to a full octave and get the reduction to 6dB instead of 14dB.)
The 14 dB off axis reduction is at only one frequency not in the usual "sub region" below 100 Hz.
Below 100Hz, the on axis sub response been reduced compared to on axis.
This response shares nothing with a cardioid response, or a decent on axis response.
It does demonstrate interference patterns nicely, good work!
"It does demonstrate interference patterns nicely, good work!"
Yep. Looks to me that you Patric have spent too much time with Hornresp and other simulation programs...
Acoustics is difficult and I am not at novice level yet myself. I don't know much about phase etc. but at some infantile level I understand all this summing/interference stuff. It is a weird world... My experiments and measurements with multi-way dipoles have opened my eyes to this materia/antimateria battle.
Your twin membrane system idea is a disaster. Just interference patterns. Corner loading is similar to horn loading and it is not so tied to frequency and backwave (dipole summing/nulling) is eliminated. When we hide the backwave, things get simplified a lot!
Yep. Looks to me that you Patric have spent too much time with Hornresp and other simulation programs...
Acoustics is difficult and I am not at novice level yet myself. I don't know much about phase etc. but at some infantile level I understand all this summing/interference stuff. It is a weird world... My experiments and measurements with multi-way dipoles have opened my eyes to this materia/antimateria battle.
Your twin membrane system idea is a disaster. Just interference patterns. Corner loading is similar to horn loading and it is not so tied to frequency and backwave (dipole summing/nulling) is eliminated. When we hide the backwave, things get simplified a lot!
Check this out if you want directional subs - phased array.
End fired Sub array (only in English) | Timo's World all around the globe
End fired Sub array (only in English) | Timo's World all around the globe
We often do this trick with our Meyer Sound subs. Stacked 3 or 4 deep with proper delay, they can really kick bass butt and not get all over the stage (important!). But that's in rooms big enough to easily fit 15 or 20 average size houses. Or outside. I can't imagine it working well in a small room.
Please let us know if it does!
Please let us know if it does!
I wouldn't call it a 'disaster', because I learned a few things from the experiment:
For at least 20 years, my 'default' location for subwoofers and furniture has basically been dictated by cosmetics. They're both up against a wall.
Here's the response of the sub when located approximately 4.5' feet from five walls, and when pushed into a corner. The dip at 132hz in the first measurement is unmistakable.
In my current setup, I have one of my subs located 5.5' off the floor, basically because Geddes said it's a good idea to vary the height of your subs. When doing that, I immediately noticed that the bass sounded 'tighter.'
So the jury is still out on why/how this works. If you have your sub located in a corner right now, I'd suggest getting it away from that corner and giving it a listen. Maybe do some measurements. Some possible ideas on why moving it away from the walls makes an improvement:
1) When placed 4.5' from a boundary, there's a strong dip at 132hz about half an octave wide, caused by the reflection mixing in to the primary wave out of phase. When that dip is combined with the sub crossover, it steepens the rolloff which may improve integration with the mains.
2) The dip at 132hz is the second harmonic of 62hz. The dip at 62hz may mask second harmonic distortion. Note that a crossover will not do this, but a reflection *will*. (Because a crossover can't reduce the distortion that's produced by a loudspeaker, but physically 'nulling' the sound via a reflection CAN.)
3) The data at musicanddesign.com indicates that dipoles have a cleaner decay than monopoles. Perhaps moving a monopole away from boundaries makes it act more like a dipole, because there are five strong reflections which occur at 3.5ms, whereas a monopole in a corner will have three strong reflections, but at a longer interval. (Ceiling, and two far walls, with reflections occurring at 7ms and beyond.)
- Status
- Not open for further replies.