[mgalusha]

John,

For DIY, the Pi speakers, six Pi, seven Pi and eight Pi are all corner horns and Wayne Parham will provide the plans and parts info upon request if you desire to build them yourself. He is extremely giving with his knowledge; the easiest way to contact him and the folks who have built or bought his designs is on his forum at audioroundtable.com. The main website is pispeakers.com but it doesn't provide all the info.

mike

[JohnR]

Thanks Mike I'd been meaning to look those up, but it's not "corner horns" I'm interested in really but corner speakers. I need to draw up a diagram to explain what I'm thinking about more. (The allison above is the closest)

[Edit: don't get me wrong, there's definitely something to be learned from the Pi speakers... more later)

[Duke]

I'm not looking to buy anything, more interested in experiences and knowledge for a DIY one. I'm sure Duke's speakers are very good but they have rear ports, so I don't see how they can be used in a corner. (I mean right in a corner, per the picture of the Allisons.)

The baseboard gives enough clearance for the ports; they're flared which eases the transition to the outside world, and any "interference" from proximity to the wall would lower the effective tuning frequency, which is the direction we want to go anyway.

In a corner, I'd have my speakers tuned quite low, like in the lower 20's, so that the pre-boundary-reinforcement rolloff starts pretty high up.  We'd want to be -3 dB around 80-100 Hz or so,  and then -10 dB down in the 30's ballpark, before boundary reinforcement.   Depending on how well the rest of the room reinforces the low end, we might find it best to plug the ports and have a low-Q sealed box.   Or maybe one box sealed, one box low-tuned ported, whatever it takes.

The inherent directivity of a large-diameter midwoofer and then the waveguide-style horn reduce coloration in the midrange and treble region that could otherwise arise from significant early interaction with the nearby walls.   Klipsch and PiSpeakers do the same basic thing, and their larger horns (including for the woofer) maintain good pattern control down lower than my approach does. 

If you're going to use small-diameter direct-radiator cones 'n' domes, do your crossover design taking the effects of corner loading into account, in particular near the lower end of each driver's range, where it inherently tends towards omnidirectional radiation. 

[Duke]

sorry, double post; meant to "modify" my post above and accidentally hit "quote".

[JCS]

The BIB thread:
http://www.diyaudio.com/forums/full-range/66173-terry-cains-bib-why-does-work-does-anyone-have-those-fostex-craft-handbooks.html

You can't see from the front, but the horn mouth is (usually) at the top such that the room corner acts as an extension of the horn mouth.

Cheers, Jim

[AJinFLA]

Has anyone here used or built/designed speakers that are specifically designed to be used or built into room corners?

Hi John,
If I were to do one, it might look a bit like the http://www.gradient.fi/Gradient/Helsinki_1.5.html. I'd make the HF more directional and possibly double up the corner dipole - http://www.linkwitzlab.com/faq.htm#Q30



Luckily, I don't have to place my speakers in corners, but I'm sure some do.

cheers,

AJ

[JohnR]

I had forgotten about that diagram... will it actually work?

Anyway, some diagrams to illustrate where I'm at so far in thinking about this. First, let's assume that walls nearby are unavoidable (small room, say). Here's an illustration of the first reflections with conventional placement:



The delay to the first reflection is too short. If the speakers have limited directivity down to a certain frequency then the reflectino will be reduced. Moving the speaker into the corner (ignoring effects on bass for now) will give something like this:



That's the scenario with say backing a Frugal Horn into a corner. With a corner horn like the Pi speakers, the controlled directivity would minimize the reflection at higher frequencies and at lower frequencies it would not matter as the corner is part of the horn. I believe that's how it works anyway.

But what if you put the speaker right into the corner, like this:



Assuming that the walls don't have obstructions, how well will this work?

[WC]

I guess the speaker would need to front ported or sealed, right?

[AJinFLA]

I had forgotten about that diagram... will it actually work?

Not sure, never tried it .
Keep in mind also that you diagrams are quite frequency dependent, i.e. you won't have those sort of angles at increasingly HF with a piston source/directional setup. and/or at least reduced intensity.
Btw, have you ever seen this http://www.trueaudio.com/array/index.htm? Might be useful in your research.

cheers,

AJ

[Letitroll98]

Ref drawing # 3 my thoughts are this.  Ignoring bass, if you use wide dispersion mid/tweeter drivers, they are not going to beam as in the drawing, they'll reflect all along the wall giving a diffuse wall of sound very much like you actually hear in concert halls.  This is what you got with Allison 3's, extraordinarily clean and distortion free as the room interaction could be predicted and controlled, but not much of what we call pinpoint imaging today.

On the other hand, if you use waveguides/horns with a high degree of directivity, wadda ya need the corner for?  Unless you plan on adding the bass component and your using the corner to control in room bass response as Duke describes (or Roy Allison put into effect) and you employ CD to eliminate side and front wall reflections in mids and treble.  Is this what you're getting at? 

[JohnR]

On the other hand, if you use waveguides/horns with a high degree of directivity, wadda ya need the corner for?  Unless you plan on adding the bass component and your using the corner to control in room bass response as Duke describes (or Roy Allison put into effect) and you employ CD to eliminate side and front wall reflections in mids and treble.  Is this what you're getting at?

Well, turn it around the other way - it's not that I want to design a speaker and deciding to use a corner as part of it, it's that I have a corner (two) and so what is the best way to utilize that space for listening? Imagine the driver is very very close to the wall, you basically have a 90 degree waveguide at all frequencies (you're sitting in it). Because of real dimensions, this would be the case only up to a certain frequency, above which controlling the directivity of the driver independently of the wall seems desirable. Depending on the frequencies and dimensions, an actual CD may not be needed at all.

The lines in drawing #3 are not indicating dispersion pattern, but the first reflection path. I'll try to make a clear drawing

[JohnR]

Keep in mind also that you diagrams are quite frequency dependent, i.e. you won't have those sort of angles at increasingly HF with a piston source/directional setup. and/or at least reduced intensity.

I'm thinking this can be employed to advantage. With a full range driver or perhaps a planar, reduced directivity will mean that the "reflection" shown is not an issue at higher frequencies. At lower frequencies, the wavelength will be such that the effective multiple sources don't interfere destructively. Off the top of my head though, that seems optimistic, will investigate.

Btw, have you ever seen this http://www.trueaudio.com/array/index.htm? Might be useful in your research.

Awesome, I had looked briefly a while ago but at the time skimmed over it. Very helpful 

[JohnR]

I guess the speaker would need to front ported or sealed, right?

Not necessarily, if you made it tall it could be some sort of transmission line.

[JohnR]

Here is an updated diagram, showing the "phantom" drivers as per Murphy:



I believe Murphy's diagram is incorrect, however, and that the phantom driver at the top right should not be there, so I have shaded it out. (More on that later.)

In green is my crude illustration of dispersion pattern at low frequencies. At low frequencies, the driver is omnidirectional. In red is the dispersion pattern at high frequencies. Depending on the width of the driver, the output "beams" and thus does not (in this simplified ideal scenario) interact with the wall or create phantom drivers.

In between, there will be a frequency range where the real and phantom drivers interfere to create "comb filtering" effects. If "X" were a quarter-wavelength (yes?) at the frequency where the driver itself narrows to a 90 degree pattern, then that would be minimal. Suppose that's 15 cm, then that frequency would be 550 Hz - too low to actually realize that "ideal" situation.

If the phantom driver at the top right doesn't exist, then that frequency would be doubled to 1100 hz - still too high to do with anything other than a proper horn I think.

As to why that driver doesn't exist, consider Murphy's diagram:



Image F exists because sound source A is radiating in all directions. This will be case with a speaker out in the room. However, if the rear of the driver is enclosed and the "box" is in the room corner, there is no radiation between the driver and the corner to create driver F.

[JohnR]

The thought experiment continues...... referring to the second diagram in the post immediately above, the "phantom" drivers are only present because of the reflection off the wall. For example, the reflection at "C" creates the phantom driver image at "D". It's a useful tool to model the behavior. With more than one driver interacting, then you get comb filtering etc. But what if you could make it so that there is no reflection off the wall at all? Then there would be no phantom drivers to cause any comb-filtering at all......

So, suppose the geometry of the baffle is changed as shown below. Now there are no reflections from the front and side wall reaching the listening position. I've drawn the paths of direct reflections to illustrate that none can reach the listener. We've created a "reflection free zone" around the listener - with respect to the front and side wall anyway.



That still leaves the rear wall, the floor and ceiling, and the opposite wall. With regard to the rear wall, it is further away than originally shown, so I've redrawn the diagram with a more realistic room dimension, with the paths of first reflections shown for the opposite and rear walls shown. For matters of scale, the room is 2.7m wide (I think).



The floor and ceiling can be addressed by using an array as per Murphy, or just living with it as almost everybody does anyway.

Now, this all seems a little "too good to be true"...  Duke and AJ, any comments? 

[puppet]

As an amateur, only thing I see wrong with your illustration is the "laser beam" rendition of the sound waves. Seems to me that as soon as the actual wave left the confines of the enclosure wing(s), you'd see reflections off the adjoining side walls.   

[AJinFLA]

John,

With a system such as this, I think it comes down to minimizing, rather than eliminating the early reflections, making the angle such that the path lengths are extended or not as direct toward the LP (which is..?).
The spatial rendition should be different from your avg, away from the walls speaker, but all speakers are compromises...and I'm sure you're either doing this out of curiosity or because the room calls for such a compromise.

cheers,

AJ

[Quiet Earth]

As an amateur, only thing I see wrong with your illustration is the "laser beam" rendition of the sound waves. Seems to me that as soon as the actual wave left the confines of the enclosure wing(s), you'd see reflections off the adjoining side walls.

I was thinking the same thing when i looked at the pictures. But I think the general idea is that the reflections are imediate or very "early" when the baffle's proximity is so close to the side wall(s). The time smear is supposedly minimized this way.

The other benefit is efficiency, which is always a good thing.

[JLM]

Why mess around?  Just fill the entire front wall with 15 inch woofers in an infinite baffle design (2 or 3 dozen should surfice) and cross over to sealed quint-pole (5 drivers pointing up, left, right, forward, and back) mid/tweeters siting out into the (large) room.  The bass would be a single wave into 1 Pi space, while the mid/tweeters provide point source imaging.  Note that the mid/tweeter design would have to address/minimize backwave reflections like some sort of transmission line/pipe/horn.

By large room I mean one that allows the mid/tweeters first reflection distances to be at least 11 feet (0.01 second delay) greater than the direct sound.

[JohnR]

As an amateur, only thing I see wrong with your illustration is the "laser beam" rendition of the sound waves. Seems to me that as soon as the actual wave left the confines of the enclosure wing(s), you'd see reflections off the adjoining side walls.   

Ah, the arrows are rays, not beams. That is, they are perpendicular to the wavefront. e.g.



(http://www.physics.uiowa.edu/~umallik/adventure/phys-optics/lightwave.html)

Thinking about it in terms of circular wavefronts, perhaps there would be diffraction from the change in the direction of the baffle/wall. I really don't know. Perhaps the baffle could be made curved... might even be easier to build...

With a system such as this, I think it comes down to minimizing, rather than eliminating the early reflections, making the angle such that the path lengths are extended or not as direct toward the LP (which is..?).

The LP is the blue circle.

you're either doing this out of curiosity or because the room calls for such a compromise.

Well, both, but it's a small room.