Floor-to-ceiling array vs CBT

[speaker dave]

Just after I wrote my comment I realized that the two line spacing's would alias differently. The log would smear the aliasing artifacts while they would be reinforced with a linear spacing. So yes, there is a difference at very high frequencies. But using an array way up to beyond its aliasing frequency seems to me to be hit or miss anyways. The individual drivers directivity would become important, even dominate and things would begin to fall apart (I would suspect) fairly rapidly.

The sim I showed earlier fully included driver directivity. Actually the array was designed to well match the inherent directivity at higher frequencies. I suppose that design is used beyond the aliasing frequency but there is absolutely no sign of aliasing, a byproduct of aiming for fairly wide dispersion which would in turn lead to lower frequency transitions to the surrounding elements.

Again, aliasing can be bypassed entirely if you have a long array with sufficiently dense driver spacing. The log array is just a more efficient way of using a given number of elements.

But I suspect that is what most of your design problems center on because below alias its a basically trivial problem.

Isn't that a bit like saying that, apart from higher order modes, horn design is basically a trivial problem? Or are we just saying that if someone else can tackle it, it must be trivial?

David

 

[keyser]

Hi Keyser,

With all ground plane measurements you want to tilt the speaker to measure on the intended listening axis. Otherwise you will be measuring response on the arrays extreme edge. It may still have reasonable response there but that probably isn't your listening window.

Don't worry the array being at an odd angle to the ground, the ground isn't really there for your measurement since the microphone is at the bisecting plane. Thought of another way you will have two arrays in the measurements (and no ground plane) and it is okay to aim both arrays at the mic (bending them in the middle) to sample the center axis response. This may seem like a pair of bent cylinders but that makes no difference for the measurement. You will get the +6dB response expected, relative to the free space curve.

David

David, I've given it some more thought and I think that what you're saying is not correct. In ordinary ground-plane measurements you want to nullify the effect of the floor. When you measure a line-array, the floor is a part of the system. If you tilt the array and you imagine the floor does nothing but add 6 dB to the response, you are measuring an array with a length equal to its real length. If you place the array perpendicular to the ground, you get a virtual doubling of the line length and the array will behave like an acoustical line-array down to half the frequency. Also, it then won't matter much if you put the mic on the floor or a bit above it - except of course in the frequency range where the array does not behave as a line-array.

 

[Patrick Bateman]

I'm convinced that the guy that designed loudspeaker arrays for McIntosh, Snell and JBL is correct in this case. I could draw a diagram to prove that he's right, but I have to get some work done. Suffice it to say, he's one of the world's experts on designing and measuring loudspeaker arrays and we're honored to get his expertise here for free.

 

[Overkill Audio]

Totally agree!

Well said Patrick...

Dave is as good as it gets with all things line array.
Not only is he technically qualified with lots of experience, he has real world commercial success...His ideas sell high volumes of speakers because they work really well!

Thanks Dave, please keep contributing and keeping the rest of us motivated!

 

[flg]

What they said

 

[Bob Richards]

Because a floor to ceiling line array is all about how it interacts with a room in a positive way, I don't know why I would ever want to measure it anywhere but in the room I planned to put it in. If you're talking about a truncated line, or one involving shading or one of those half way to the ceiling bent arrays, or the ones that are designed to hang over a stage, then I would probably want to measure it in other ways, but those are a very different creature. It seems that each variation will have a substantially different interaction with room acoustics.

Off axis vertically, you've got a transversal filter effect going on, with each driver being a term in the equation. Putting a cal'd mic on the ground with the array tilted or not, won't be at all the same as when it's in a typical listening room, will it?

 

[keyser]

Hey guys, I'm well aware who speaker dave is. He is most definitely one of diyaudio's most knowledgable contributors and I respect him a lot. I almost always agree with what he says, but I don't except anything he says on the basis of authority. I sometimes question what he says, and indeed it usually turns out I was wrong and he was right. Occasionally I maintain my position.

Like I explained above, I believe ground-plane measurements of line-arrays work somewhat different than ground-plane measurements of point sources. If I'm wrong, I'd really like to understand why .

 

[gedlee]

I believe ground-plane measurements of line-arrays work somewhat different than ground-plane measurements of point sources. If I'm wrong, I'd really like to understand why .

I was tempted to say that a line array is just a bunch of points so they have to work the same, but I think that your question is different that that.

If a speaker is intended to use a ground plane then its effect should be included, if not then it should be excluded. Usually when we measure speakers in a ground plane we are trying to get a free field response so we want to exclude the ground plane. If you want to include the ground plane then the array should not be tilted. If you want to exclude it then even tilting does not do that. You just get something that is harder to interpret.

Ground plane measurements are hard to do right because surfaces act very differently to grazing incidence sound waves over large distances. Unless the surface is very hard and very smooth (like a steel plate, which is what we had at Ford, concrete won't do, too porous and rough.) it will have a significant effect on the results.

 

[jeno]

This thread is about using floor and ceiling to make a line array virtually infinite. I also question if Speaker-Dave was talking about measuring a finite line array. If not I won't argue.

Edit: Earls post makes it clearer.

Earl, does that mean that a smooth rigid floor will sound better if the floor is part of the design?

 

[Bob Richards]

Correct me if I'm wrong about this, but why would anyone want to do a "ground plane" measurement? Are they trying to use a mic as a pressure zone mic? Does anyone listen to their stereo with their head on the floor? The reason a floor to ceiling line array is worth considering is because of how it works with typical listening room acoustics. It's a fairly complex thing to explain adequately. The room is then effectively an important part of the speaker system. Without the room any measurements are largely irrelevant. Speaker Dave is a student just like the rest of us. I'm not sure he always sees things in proper perspective. His criticisms of Roger Russels Line Array makes me wonder if he really understands what's going on acoustically with those.

 

[gedlee]

Earl, does that mean that a smooth rigid floor will sound better if the floor is part of the design?

It just means that different floors will have different effects. I am not going to comment on "sound better".

Correct me if I'm wrong about this, but why would anyone want to do a "ground plane" measurement?

In theory they can yield a free field measurement in a 1/2 space. And 1/2 space, on Earth, is a lot easier to obtain.

 

[Omholt]

Measuring the complete polar of a taller CBT isn't a small task. When I asked Don if he was planning to have Ron Sauro measuring the CBT36 like he did with a smaller CBT prototype, this is the answer he gave me:

The CBT36 is quite difficult to set up for a measurement of a full set of polars. A free-space CBT array would have to be created by placing two systems bottom-to-bottom. This would weigh about 100 lbs. The resultant array has to be rotated around its center of curvature so that accurate polars would be measured (which is 8.5 ft behind the front of the array; lever arm and all that).
I am in the midst of discussions with Ron to measure an alternate smaller 90-deg-arc free-standing passive CBT array to demonstrate the superior coverage of a circular-arc CBT array.

 

[fporegon]

frequency dependant shading

Not always the case. Frequently the goal is to make the effective line length proportionate to radiated wavelength. "Short at high frequencies and long at low frequencies" allows for arrays with units that are progressively low pass filtered in proportion to distance away from the array center. All units can be active at low frequencies, preserving LF power handling.

David

David,

Are you aware of a method for accomplishing frequency dependent shading? The phase shift introduced by the low pass filter(s) seems to undo the benefit of the shading. In attempting to set the amplitude of the driver element based on its distance from "center", doesn't the phase shift tend to make the driver look like it is further away from center? And so on....

I end up thinking that only a DSP approach can make this work.

Frank

 

[gedlee]

Don's CBT has the technology lock on performance if passive is required. But with DSP and individual source control his advantage goes out the window. Its only advantage is that it works so well with passive means. Dave is on the right track with individual control.

Back in the 80's we did individual control of torpedo heads with 54 (as I remember) transducers in a square array. What was possible was amazing. You could steer different signals to different locations with a single array. In other words, one array could do multiple channels.

 

[flg]

David,

Are you aware of a method for accomplishing frequency dependent shading? The phase shift introduced by the low pass filter(s) seems to undo the benefit of the shading. In attempting to set the amplitude of the driver element based on its distance from "center", doesn't the phase shift tend to make the driver look like it is further away from center? And so on....

I end up thinking that only a DSP approach can make this work.

Frank

I believe, of coarse "Correct me if I'm wrong about this",,, I'm "a student just like the rest of us", I'm not sure I always see things in proper perspective,,, wow,,, shading(analog/passive) with a lag to the drivers would be a small attempt to do what Keele has shown necessary for a strait array. Delay? There is a recent patent on this theory but I don't have it or a link to it handy. Of coarse it appears the only right or better way today is the magic DSP with the addition of several low cost classD amps?

 

[Patrick Bateman]

David,

Are you aware of a method for accomplishing frequency dependent shading? The phase shift introduced by the low pass filter(s) seems to undo the benefit of the shading. In attempting to set the amplitude of the driver element based on its distance from "center", doesn't the phase shift tend to make the driver look like it is further away from center? And so on....

I end up thinking that only a DSP approach can make this work.

Frank

David hasn't chimed in on the phase issue. Here's my take on it, if anyone's curious:

1) It is not possible for a CBT or a line array to have flat phase. The reason for this is simple; there are multiple sources radiating the same signal. For instance, in a CBT array there's something like thirty six(!) tweeters radiating frequencies above 1khz. Due to that, flat phase is literally impossible. A few months ago I measured the phase response of a planar speaker, and found the same thing. Even though it's only one driver, it's so large, it behaves like an array.

2) eXpanding arrays can have flat phase. This is because the size of the array varies with frequency. For instance, at 4khz the eXpanding array is just 1" tall. At 4khz, a CBT 36 is over a meter tall. That's why the CBT can't have flat phase; it's size dwarfs the wavelength reproduced. As long as that is true, flat phase is impossible.

 

[Omholt]

David hasn't chimed in on the phase issue. Here's my take on it, if anyone's curious:

1) It is not possible for a CBT or a line array to have flat phase. The reason for this is simple; there are multiple sources radiating the same signal. For instance, in a CBT array there's something like thirty six(!) tweeters radiating frequencies above 1khz. Due to that, flat phase is literally impossible. A few months ago I measured the phase response of a planar speaker, and found the same thing. Even though it's only one driver, it's so large, it behaves like an array.

2) eXpanding arrays can have flat phase. This is because the size of the array varies with frequency. For instance, at 4khz the eXpanding array is just 1" tall. At 4khz, a CBT 36 is over a meter tall. That's why the CBT can't have flat phase; it's size dwarfs the wavelength reproduced. As long as that is true, flat phase is impossible.

I assume you are right when it comes to the very highest frequencies regarding phase but Ron Sauro (who measured the CBT prototype) said that the phase response of the CBT was well behaved.

Here's another comment from Don:

Indeed, the impulse response of the CBT array is extremely-well behaved at most locations based on my simulations and limited measurements. As an example: if a free-standing CBT array is equalized flat at say 1 m on axis (both magnitude and phase). It's also essentially flat and linear phase at most other locations, up-down, side-to-side,or closer-farther away when the transport delay is accounted for (believe it or not!).

 

[flg]

David,

Are you aware of a method for accomplishing frequency dependent shading? The phase shift introduced by the low pass filter(s) seems to undo the benefit of the shading. In attempting to set the amplitude of the driver element based on its distance from "center", doesn't the phase shift tend to make the driver look like it is further away from center? And so on....

I end up thinking that only a DSP approach can make this work.

Frank

Like this you mean?

The David Smith (speakerdave here on this forum) line array papers conclude with a shading approach but it is termed cosine, 1/2 sine or something not Hann or Legandre.

I would like to correct or at least be a little more accurate regarding something I said earlier. The Shading David did was termed A raised co-sine I believe, that is Hann shading.

 

[speaker dave]

David,

Are you aware of a method for accomplishing frequency dependent shading? The phase shift introduced by the low pass filter(s) seems to undo the benefit of the shading. In attempting to set the amplitude of the driver element based on its distance from "center", doesn't the phase shift tend to make the driver look like it is further away from center? And so on....

I end up thinking that only a DSP approach can make this work.

Frank

Delays act more as a set back for the elements rather than added spacing apart.

We have tools at work that assume highly complex IIR filters if necessary. That doesn't mean that effective arrays can't be made with simpler filters.

All the Snell XA and McIntosh HT1 and HT3 (3 tweeter THX systems) used rolloff on the outer elements (frequency dependent) and were done with a simple passive implementation. You need to be careful with your initial rolloff rate. Its a bit like designing a 2 1/2 way system where you can't go beyond first order for initial rolloff of the lower woofer. To exceed that gives more phase shift and a dip in the transition. The simple expanding arrays are like that: rolloff rates need to be carefully considered as part of the design.

The CBT designs seem to require the delay of the outer units. Don and others have published both the standard curved version and a straight line version with electrical delay for the outer elements. The electrical delay (referenced in another post here via its patent) give pretty similar response, at least for a modest range of angles. Curved line arrays, of course, have been around since the beginning. The arc tends to broaden the far-field beam and makes near field and far field polars more uniform.

The log spacing array I am playing with has significant delay between the elements, but only in the stop band. I am hoping that I can find implementation both via DSP and via passive networks, as the drive voltages are being kept inherently simple.

Regards,
David

Speaker Dave is a student just like the rest of us. I'm not sure he always sees things in proper perspective. His criticisms of Roger Russels Line Array makes me wonder if he really understands what's going on acoustically with those.

I'm not aware that I have ever said anything detrimental regarding Roger's arrays. The McIntosh AES paper was written to help explain the observations of array phenomena, that Roger and Carl related to me.

David

In theory they can yield a free field measurement in a 1/2 space. And 1/2 space, on Earth, is a lot easier to obtain.

An empty parking lot is your friend!

David, I've given it some more thought and I think that what you're saying is not correct. In ordinary ground-plane measurements you want to nullify the effect of the floor. When you measure a line-array, the floor is a part of the system. If you tilt the array and you imagine the floor does nothing but add 6 dB to the response, you are measuring an array with a length equal to its real length. If you place the array perpendicular to the ground, you get a virtual doubling of the line length and the array will behave like an acoustical line-array down to half the frequency. Also, it then won't matter much if you put the mic on the floor or a bit above it - except of course in the frequency range where the array does not behave as a line-array.

I think the answer here depends on what you are looking for. If you want to know the anechoic response of the array on axis, then ground plane measurements with the line pointed down will be the closest representation of that.

If you want to know how the array looks in a room with floor reinforcement, then having the array straight up and the mic at a typical ear height will represent that. This is no different than putting a bookshelf speaker on a stand and measuring it in the parking lot. You will measure the bookshelf speaker and its later arriving reflected image, so your curve will show a strong floor bounce. (i.e. it will not be anechoic)

This is a bad thing for the bookshelf speaker measurement but not so bad for the line array. If the line array continues to near the floor then we see measurements of a nice array twice as long. Don Keele designs his CBTs as 1/4 arcs with the assumption that a solid floor will be there to complete the half array.

But why then stop with one surface? If your array is floor to ceiling then the practical result is that you have both lower and upper reflectors. By extension do we need to go out into the parking lot and also add an artificial roof over the speaker to represent the ceiling, and then make an "anechoic" measurement of that?

As I said, it all depends on what you are trying to measure. But the anechoic response of the line alone is taken with the mic on the ground and the array tilted forward.



Earl, Interesting about the surface roughness effect. We did ground plane measurements at JBL on asphalt and never noticed any distinct effects but we weren't really looking for them either.

Regards,
David

 

[gedlee]

Sound absorption for grazing incidence is much different than it is for larger angles. That is because the wave travels along the surface constantly being absorbed, which is not the case when its absorption coefficient is measured - that measurement is a one time incidence and reflection. The grazing incidence values are many time larger and depend on the distance involved.

I have seen results of the same model of speakers done on ground plane that I have also done on my test stand. They did not look the same. Different drivers? Maybe. One measurement is not correct? That's my guess. I have never trusted ground plane.