I'm thinking more and more about how to control the directivity of my future loudspeaker projects. While there are several ways to get controlled directivity (horn, dipole/OB) I'd like to get have a primarily forward radiation (cardiod?) that extends down to the 200Hz range. So I thought I would start this thread to muse about the possibilities and get input from others.
I have made a couple of dipole systems of my own design and I like them but I would like to use a horn loaded tweeter and so want to generally match the primarily forward firing pattern in the mid-lower octaves.
One idea that came to mind was to use rather large OD pro-audio woofers. Starting atthe 12" horn's HP at around 1.5kHz I imagine could use a 12" driver, sealed, to match the horn's directivity. But the 12" driver will start to lose directionality as frequency decreases. So around about 500Hz I would transition to a cardiod comprised of a dipole 10" and a 15" or larger sealed driver. The dipole faces the usual way - forward. The sealed 15" driver sits directly below the dipole and fires up or is otherwise close-coupled. The combination of the dipole and monopole patterns results in an omni as the rear wave is cancelled and the front wave reinforced. Although I have not done a detailed analysis, this should have controlled directivity at low frequencies. That's also where it seems that the spatial arrangement of drivers is creating the least amount of problems because the wavelengths are long (500Hz is about 70cm) and as long as you keep the drivers physically close together they are also "acoustically close" below 500Hz.
I am aware of designs like Tom Danley's synergy horn, and that is one concept that I want to look into in more details. What other design approaches can be used to get controlled directivity at low(er) frequencies?
I have made a couple of dipole systems of my own design and I like them but I would like to use a horn loaded tweeter and so want to generally match the primarily forward firing pattern in the mid-lower octaves.
One idea that came to mind was to use rather large OD pro-audio woofers. Starting atthe 12" horn's HP at around 1.5kHz I imagine could use a 12" driver, sealed, to match the horn's directivity. But the 12" driver will start to lose directionality as frequency decreases. So around about 500Hz I would transition to a cardiod comprised of a dipole 10" and a 15" or larger sealed driver. The dipole faces the usual way - forward. The sealed 15" driver sits directly below the dipole and fires up or is otherwise close-coupled. The combination of the dipole and monopole patterns results in an omni as the rear wave is cancelled and the front wave reinforced. Although I have not done a detailed analysis, this should have controlled directivity at low frequencies. That's also where it seems that the spatial arrangement of drivers is creating the least amount of problems because the wavelengths are long (500Hz is about 70cm) and as long as you keep the drivers physically close together they are also "acoustically close" below 500Hz.
I am aware of designs like Tom Danley's synergy horn, and that is one concept that I want to look into in more details. What other design approaches can be used to get controlled directivity at low(er) frequencies?
Last edited:
I used a passive resistive enclosure using an AE TD-15M, it had a very small bandwidth(150-450hz), but was all I needed to reach a SEOS-24/BMS coax.
Resistive enclosure: http://i.imgur.com/tfITrDk.jpg
Speakers as a whole: http://i.imgur.com/c8SoqzK.jpg
I experimented with a few different enclosures. They all measured pretty similar.http://i.imgur.com/RLOfNg7.jpghttp://i.imgur.com/YpTQKUE.jpg http://i.imgur.com/hZlg6vG.jpg
As can be seen in one of the images, I eventually added a rear firing woofer behind the cardioid section, it was used below 150hz in an attempt to extend the cardioid waveform down as low as possible, but I had a very difficult time with measurements. It did flatten bass response though, but I couldn't tell if it still held any pattern or not(doubtful). http://i.imgur.com/zmqLgEU.jpg
Resistive enclosure: http://i.imgur.com/tfITrDk.jpg
Speakers as a whole: http://i.imgur.com/c8SoqzK.jpg
I experimented with a few different enclosures. They all measured pretty similar.http://i.imgur.com/RLOfNg7.jpghttp://i.imgur.com/YpTQKUE.jpg http://i.imgur.com/hZlg6vG.jpg
As can be seen in one of the images, I eventually added a rear firing woofer behind the cardioid section, it was used below 150hz in an attempt to extend the cardioid waveform down as low as possible, but I had a very difficult time with measurements. It did flatten bass response though, but I couldn't tell if it still held any pattern or not(doubtful). http://i.imgur.com/zmqLgEU.jpg
Charlie
The problem is easy to see if you look at the radiation modes. It takes two modes to get a cardioid and more than that to get anything with a higher DI. As the modes go down in frequency they loose efficiency - fast. So you always end up with a power hungry design.
The problem can be done in reverse to find out what driver sizes and locations and EQ are necessary to get any desired directivity. My book has an example for the cardioid.
I have gone through this exercise and I always come back to these issues:
1) the higher the desired DI the harder it is to do. A 4.8 dB DI is not too hard - a dipole, more than that gets tougher and tougher, 6 dB for a cardioid.
2) the higher the DI desired, it gets tougher and more and more sources are required. The system is also getting power hungry because the efficiency loss and more drivers to power.
3) then there is the simple fact that as one gets down into the rooms modal region any consideration of directivity become pointless as the room (and our ears) simply integrate the total sound over several periods of the wave meaning that all directions of the sound radiation are included equally because of the room reflections - the directions becomes meaningless.
So what you have to decide is how low do I want to go with whatever DI you want to achieve. For me there is not a reasonable solution below about 400 Hz. From 400 Hz. to 800 Hz I can get a few more dB of DI at an almost doubling of the cost of the LFs. Its not a tradeoff that I have ever been willing to take.
The "leaky" box cardioid look somewhat attractive, but again, none that I have seen achieve a DI much above about 6 dB. At HFs I get a DI of about 12 dB - flat down to 1 kHz, falling slowly to about 6 dB at 500 Hz and then dropping to zero below that.
You take your tradeoffs and live with them. I will not trade off LF CD for HF DI.
The problem is easy to see if you look at the radiation modes. It takes two modes to get a cardioid and more than that to get anything with a higher DI. As the modes go down in frequency they loose efficiency - fast. So you always end up with a power hungry design.
The problem can be done in reverse to find out what driver sizes and locations and EQ are necessary to get any desired directivity. My book has an example for the cardioid.
I have gone through this exercise and I always come back to these issues:
1) the higher the desired DI the harder it is to do. A 4.8 dB DI is not too hard - a dipole, more than that gets tougher and tougher, 6 dB for a cardioid.
2) the higher the DI desired, it gets tougher and more and more sources are required. The system is also getting power hungry because the efficiency loss and more drivers to power.
3) then there is the simple fact that as one gets down into the rooms modal region any consideration of directivity become pointless as the room (and our ears) simply integrate the total sound over several periods of the wave meaning that all directions of the sound radiation are included equally because of the room reflections - the directions becomes meaningless.
So what you have to decide is how low do I want to go with whatever DI you want to achieve. For me there is not a reasonable solution below about 400 Hz. From 400 Hz. to 800 Hz I can get a few more dB of DI at an almost doubling of the cost of the LFs. Its not a tradeoff that I have ever been willing to take.
The "leaky" box cardioid look somewhat attractive, but again, none that I have seen achieve a DI much above about 6 dB. At HFs I get a DI of about 12 dB - flat down to 1 kHz, falling slowly to about 6 dB at 500 Hz and then dropping to zero below that.
You take your tradeoffs and live with them. I will not trade off LF CD for HF DI.
I'm also using a "leaky" box with a 15 which I consider to be more like a dipole with a damped rear wave. It has a 120° forward pattern down to 200hz or so and the piston dia takes over around 1khz. Rear attenuation is about 10db at 300hz and increases as you go up in freq. I'd like more rear attenuation but this isn't bad. I'm using it currently as a 2.5 way with a sealed woofer bringing up the low end where the leaky box is rolling off.
I've also been playing with a second order gradient system, which has a narrower pattern than a dipole, but with that you get second order roll-off on the low end vs 1st order for the dipole. I haven't quite got my head wrapped around that one yet. It's pretty tricky to get right......at least the way I want to do it.
I played with the method you're proposing, Charlie, but couldn't get it to work. Didn't give it much of a shot though as my leaky box was pretty effective and didn't require another set of drivers/amp/dac etc.
I've also been playing with a second order gradient system, which has a narrower pattern than a dipole, but with that you get second order roll-off on the low end vs 1st order for the dipole. I haven't quite got my head wrapped around that one yet. It's pretty tricky to get right......at least the way I want to do it.
I played with the method you're proposing, Charlie, but couldn't get it to work. Didn't give it much of a shot though as my leaky box was pretty effective and didn't require another set of drivers/amp/dac etc.
Sorry for the OT.
Does it follow from the group delay audibility threshold that 3.2ms is integrated at 500Hz and the above applies within that window?
Second order cardioid. Two dipoles back to back and the front one electronically delayed by the amount of their distance.
.
Beamforming is an option.
I am starting the build/test phase of a 4-way passive design that will use a very tall, shallow and wide box mounted on the wall. At the lowest frequencies I will use multiple subwoofers to handle room modes. In the 200 hz range I will use many small woofers rather than a couple big ones.
I am exclusively interested in controlling vertical reflections (and eliminating the first front-wall reflection) as these change the perceived tonal character of a speaker. The lateral reflections off the side walls add a sense of envelopment and widen the sound stage, which most perceive as an improvement. (Walls are also more easily treated than floors, should I choose to tackle them later.)
With careful placement of multiple drivers on the baffle, and adhering to a few crossover design rules, I have been able to design a constant directivity speaker with a single main lobe, and no sidelobes higher than 20 dB (in acoustic space anyway). Of course, directivity does diminish at the lower frequencies, as restricted by the height of the speaker.
In effect, it's something like a 180 H x 40 V CD design.
Last edited:
The way I look at it is the integration time varies with frequency according to the impulse response of a set of gamma-tone filters.
Attachments
Here's the thread for my speaker that I was talking about above: http://www.diyaudio.com/forums/multi-way/259724-dsp-midrange-directivity-control-aka-kinda-cardioid.html. Good data starts on post 7 and there's some good discussion as well.
I've found that, like most things, when your driver spacing is around 1/4wl or less then you get the greatest effect. Using the 15 as the rear driver can make it a little trickier as it's still fairly directive at 500hz. In my recent second order gradient experiments I was trying to do it with two 15" dipoles and as best as I can tell the directivity of the 15 was the limiting factor even when I was at 1/4wl or less. I also tried a 15" dipole with a sealed 15" directly behind that and that worked fairly well.........though not better than my passive box in the areas I think are important.
I've found that, like most things, when your driver spacing is around 1/4wl or less then you get the greatest effect. Using the 15 as the rear driver can make it a little trickier as it's still fairly directive at 500hz. In my recent second order gradient experiments I was trying to do it with two 15" dipoles and as best as I can tell the directivity of the 15 was the limiting factor even when I was at 1/4wl or less. I also tried a 15" dipole with a sealed 15" directly behind that and that worked fairly well.........though not better than my passive box in the areas I think are important.
Well, I am kind of stuck with this 12" for various reasons, so I think I will go with the resistive-side/edge-slot box initially.
I see that some try to use "stuffing" in the vent, but that doesn't seem to be a very good resistive material. I have seen felt used. What type and thickness is advised in this case? Any other materials (tight or loose cloth in layers?) suitable to try?
I see that some try to use "stuffing" in the vent, but that doesn't seem to be a very good resistive material. I have seen felt used. What type and thickness is advised in this case? Any other materials (tight or loose cloth in layers?) suitable to try?
This is what I used, you'll have to experiment with density based on the porosity of your enclosure. http://m.homedepot.com/p/UltraTouch...lti-Purpose-Roll-6-Pack-60306-16482/202710055
I also use Bonded Logic.........what I've got is 3.5" thick iirc and I think I'm using two layers now. Another thing that can be done is to restrict the basket itself creating a low pass filter for the rear output. The reduced hf output to the rear allows the driver directivity to take over better. I'm going to experiment with a layer or two of felt over the basket in the next week or two. I've been trying some different cloth like materials lately to good effect but it takes a lot of layers with the thin material. Higher density/thicker material for the stuffing tends to widen the forward pattern since you are attenuating and lo passing the rear output of the driver so the cancellation to the sides isn't as great, or rather I think the null angles to the side are shifting rearward.
What sort of pattern are you looking for? True cardioid is 180° while dipole is 120°. In my second order gradient experiments I was getting about a 90° forward pattern but there was much more rear output than I want.
What sort of pattern are you looking for? True cardioid is 180° while dipole is 120°. In my second order gradient experiments I was getting about a 90° forward pattern but there was much more rear output than I want.
Thanks for the info on the bonded logic material. I will look into that.
Regarding what I want for pattern control - here is the story:
In one of my upcoming build(s) I want to use a couple of drivers that I have already purchased or acquired. Starting at the top, I have a small compression driven (about 5" wide) that works well above 3kHz (based on my measurements). Below that it totally loses directional control. Next up I have a 12" coax woofer that I picked up for a song. Rather than use that driver with a 1" compression driver up to 20kHz I thought would I would do is pair it with the Selenium D-250X GW compression driver to fill between the woofer and the small horn. The woofer-cone-as-a-horn probably only has directional control down to about 800-1kHz, but I am thinking of trying to extend that down a little by building the woofer into a shallow horn of its own. I don't have any experience with that, so perhaps it's folly. In any case, I hope to use the small horn on top, the woofer-cone-horn from about 900Hz-3kHz, and then the woofer below down to about 200Hz or so where I will cross over to yet another driver that will take it from there down to the bottom.
Wow, lots of drivers. Well I have DSP and lots of channels so why not.
What I would like to do is increase the directionality of the 12" woofer in its operating range. At the top of its range the directivity will be slightly narrowed but likely not as narrow as the compression driver playing out through the cone. The woofer only needs to operate over about 2-3 octaves, and since the low end of that range is not all that low the excursion requirements should not need to be super high under cardoid operation. Also I can use a shallow crossover to the real woofer in the system, and that will reduce the low end output requirements due to the gentle shape of the passband edge. In short (I could have said this from the beginning) I would like this 12" woofer to have a DI of 6-10dB over its operating range, primarily in the upper part. The DI could drop a bit at the lowest frequencies, but the system DI will ultimately and gently be eased downwards by the crossover to the real woofer. This system will be placed against a wall (actually both speakers in the pair are in a nook) so something like a dipole is out of the question. The monopole-below-dipole arrangement might work, but I am not convinced that it will do so as high as 900Hz because the wavelength is getting small compared to how close you can space a 12" driver. I could try it anyway. So, it appears that the resistive-side-vent cardoid might work the best, although it won't quite get me to 10dB of DI.
If the 12" driver was instead a 15" I might not even need to do anything because it would already have a sufficiently narrowed pattern by 900Hz. But alas I have a 12". My thinking is that, if I can use a couple of slots covered with felt to increase the directivity why not try it?
So, that's the long story of my plans for yet another DIY adventure. Hopefully I will learn a few things along the way. Rinse and Repeat.
CharlieLaub, in my opinion the configuration you are proposing is needlessly complex and will probably not work very well. A 12" woofer in an optimized semi-open enclosure can get you very good directivity from say 80 hz to a little above 1 khz. The directivity you get is a bit higher than that of a pure cardioid, because the driver and the baffle contribute as well. An optimized open-box cardioid can give you about 20 dB rear rejection. Couple that enclosure to a decent waveguide with a similar polar pattern at the crossover frequency and you're all set.
Rejection to the rear has its greatest advantage in the lower midrange, where front-wall reflections can mess up the timbre of the loudspeaker. At ever lower frequencies high directivity becomes less useful, because modes become the dominant factor.
If you want really high directivity in the midrange, there are two options. The first has already been mentioned by Gedlee and natehansen66: higher order gradient loudspeakers. The second is the use of a huge waveguide. Neither are very practical. You should check out the new Beolab 90 by B&O though!
http://www.diyaudio.com/forums/multi-way/280779-b-o-beolab-90-adjustable-directivity-dsp.html
Rejection to the rear has its greatest advantage in the lower midrange, where front-wall reflections can mess up the timbre of the loudspeaker. At ever lower frequencies high directivity becomes less useful, because modes become the dominant factor.
If you want really high directivity in the midrange, there are two options. The first has already been mentioned by Gedlee and natehansen66: higher order gradient loudspeakers. The second is the use of a huge waveguide. Neither are very practical. You should check out the new Beolab 90 by B&O though!
http://www.diyaudio.com/forums/multi-way/280779-b-o-beolab-90-adjustable-directivity-dsp.html
An externally hosted image should be here but it was not working when we last tested it.
like this? interesting
I don't know about more space, but certainly more amplifiers. Unless you figure how to phase the array passively like Keele does, but there aren't too many options passively.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.