It certainly is nicer to have "as-linear-as-possible" sound with controlled uniform directivity .
And it is possible over a wide range, from around 300- 400 Hz up using a Synergy or Co-entrant style horn, which puts all the separate drivers output within 1/4 wavelength at their crossover points, eliminating the comb filter effects that will always be present in the vertical plane with standard front mounted cone drivers, or a cone and waveguide/horn.
The only lobing issues with that type of horn would be between the horn and the sub (if required, larger versions can be made full range), and room issues usually dominate that frequency region anyway.
Art
And this is ambiguous at 150-500Hz, as this region is troublesome to treat and notoriously sensitive to specular reflections, and opined as possibly the most fundamental of regions.
You have repeatedly stated that we are not very sensitive to frequencies below 500Hz, which I take to mean with regards to where they are coming from, and when they originated.
If I were to go and study the group delay sensitivity charts, and take the cynical view that's had some of us wondering why somebody doesn't just build a 21" waveguide....then I wouldn't be posting here....I'd probably be crunching, drinking coffee and pulling my hair out.
I spent several hours today studying the relationship between my particular floor and ceiling, and basically, if I want to attenuate the ceiling reflection at all I may as well go all the way because if it is there at all, it is better at full strength. I don't care about anything above 500Hz...except that they need to go and a nice 2" absorber can be put to good use there.
My point is that I coordinated a 150-500Hz total response with both vertical and horizontal walls with little absorption. There is reasonable cooperation between walls of the same plane and I did use a helper woofer (sub) placed front and centre "barely" helping up to around 500Hz. I then needed only two more subs, and one was doing nothing more than supplementing the low 'almost non-modal' region.
So with no horizontal absorption besides the rear wall, and only >500Hz elsewhere I find the results pleasing. It is not easy, nor have I even seen fit to try to isolate time delay effects or horizontal versus vertical directivity down at these lowish frequencies.
This is important to me. If male vocalists cannot sing through their range without sounding like they're changing rooms.... if female vocalists cannot let go and wail, unhampered enough to induce emotions within me that only my wife is capable of on that rare occasion....well, I'm already moving on and planning to reassess the power response of my crossover.
Yes, this frequency region is a problem to treat, but I see nothing in your post that would sway my opinion, which remains that this region is not very important to imaging and the like.
I also see your view on the three broad ranges. One sub suggested for VLF duty, one in the central bass and one filling in holes which is lower in level and may go higher.
Of course going through it all the hard way (like I've been told I sometimes do
), may have given me the knowledge to reduce this down to a design process that avoids extraneous problems, rather than starting with a random room and throwing subs at it with limited confidence.
Of course going through it all the hard way (like I've been told I sometimes do
), may have given me the knowledge to reduce this down to a design process that avoids extraneous problems, rather than starting with a random room and throwing subs at it with limited confidence.
I have changed my position a bit, mostly in regard to the VLF sub. I have not found them to be all that effective and now recommend just using a "good" sub (like mine) for the entire range and relying on EQ to get what you want. In several studies, I found that tuning a sub lower just lowers the efficiency at higher frequencies, it doesn't really achieve any more output at the lower freqs. So there is no point to changing the tuning if you have EQ, just let the electronics change the tuning for you.
Member
Joined 2009
Yes, while completely ignoring the horizontal ones. For me the horizontal is the more important.
Member
Joined 2009
The horizontal directivity is indeed much more important!
I was very interested in discovering the effects of the vertical response so I built a point source design by placing the drivers <1/4 wavelength. The speaker sounds great indeed but I was very surprised to discover that when I move the crossover to a point where it creates ugly lobes on the vertical axis, the effect on the sound is negligible. Horizontal off-axis response on the other hand is audible!
---
Speaking of Keele's CBT I don't think that he ignores the horizontal response, I can't recall seeing graphs but I would guess it illuminates almost equally in 180deg. In other words the horizontal directivity is constant but wider than more typical CD designs.
There are plenty of speakers out there that deal with directivity in one way or another. Most line arrays will give high vertical directivity, the CBT is just more well behaved than equal weighted mid length lines. At McIntosh I designed raised cosine weighted arrays that had a nice vertical directivity with good smoothness.
Your first problem is to decide how you want to control directivity. Then over what frequency range it is important for you. Then the available technology will be decided.
In essence you need to look at the reflections in the typical room/listener relationship and decide which ones you would like to reduce. Those choices will be based on strength, arrival delay and arrival direction.
Most of what I read suggests that wide horizontal dispersion with fairly strong lateral reflections is fine if the reflections don't arrive too early. Similar studies suggest that vertical reflections from either the floor or ceiling are problematic in that the ear can't seperate tham from the direct sound and they are perceived as frequency response errors. I frequently hear treble bounces off of carpeting and the ceiling as comb filtering at HF. Kates and Salmi both showed that midrange reflections from the floor are audible. The bigger design challenge is controlling MF directivity to the point of reducing them.
For floor bounce problems there have been plenty of speakers that deal with the floor (but not necessarily the ceiling) these include the Allison designs, the AR9, the NHT, anything with a woofer at the floor boundary and a low crossover point. Some large arrays such as the Snell XA reference had usefull low-midrange directivity. The Ken Kantor Magic speaker used short arrays and a foam lined recess to increase directivity and reduce the floor bounce.
CD horns are good for contributing directivity but they tend not to go low enough. If they are axisymmetrical they will give up dispersion in the horizontal plane, which doesn't seem to be a good thing.
Regards,
David S.
Your first problem is to decide how you want to control directivity. Then over what frequency range it is important for you. Then the available technology will be decided.
In essence you need to look at the reflections in the typical room/listener relationship and decide which ones you would like to reduce. Those choices will be based on strength, arrival delay and arrival direction.
Most of what I read suggests that wide horizontal dispersion with fairly strong lateral reflections is fine if the reflections don't arrive too early. Similar studies suggest that vertical reflections from either the floor or ceiling are problematic in that the ear can't seperate tham from the direct sound and they are perceived as frequency response errors. I frequently hear treble bounces off of carpeting and the ceiling as comb filtering at HF. Kates and Salmi both showed that midrange reflections from the floor are audible. The bigger design challenge is controlling MF directivity to the point of reducing them.
For floor bounce problems there have been plenty of speakers that deal with the floor (but not necessarily the ceiling) these include the Allison designs, the AR9, the NHT, anything with a woofer at the floor boundary and a low crossover point. Some large arrays such as the Snell XA reference had usefull low-midrange directivity. The Ken Kantor Magic speaker used short arrays and a foam lined recess to increase directivity and reduce the floor bounce.
CD horns are good for contributing directivity but they tend not to go low enough. If they are axisymmetrical they will give up dispersion in the horizontal plane, which doesn't seem to be a good thing.
Regards,
David S.
I'm not sure I understood what you said in this sentence. You hear treble bouncing off of carpeting? I understood that treble is absorbed by carpeting -- maybe you could elaborate what you define as "treble".
As well, I'm not sure about the math so don't now how to work this out but my assumption is that the tweeter would be sufficiently far from the floor (and perpendicular to it) that any reflection off of it would not create comb filtering.
Last edited:
Actually, more treble bounces off of carpet than you would guess. By treble I suppose I mean the 5k to 20kHz range.
Typically you hear a distinctive treble pitch due to the comb filtering. As you walk away from the unit it goes up in pitch due to the path lengths getting closer together. You can hear a similar effect if you listen to small waves breaking on the beach at the shore. Doing deep knee bends will also raise the pitch as you get closer to the floor. (Try it.)
David S.
Typically you hear a distinctive treble pitch due to the comb filtering. As you walk away from the unit it goes up in pitch due to the path lengths getting closer together. You can hear a similar effect if you listen to small waves breaking on the beach at the shore. Doing deep knee bends will also raise the pitch as you get closer to the floor. (Try it.)
David S.
They do if they are big enough.
If by "give up dispersion" you mean that they have a narrow horizontal coverage ("dispersion" implying a wide coverage), how is that not "a good thing"? It elliminates the very early reflections that yield poor imaging.
Otherwise I don't understand what you are saying.
Member
Joined 2009
The Kates and Salmi papers show issues with floor bounces down to at least 200Hz. Most horns aren't that big. I'm pretty sure yours aren't.
There is good agreement that later lateral reflections are fairly benign and that earlier vertical reflections are troublesome. The only gray area is whether we want any reflections to get a sense sound sources in a live space and arrivals from outside our speaker triangle, or not. I would refer you to Toole, but I know you don't agree with him. Perhaps you can point us to a study that shows that any and all lateral reflections degrade imaging?
David S.
Here's my take on the items discussed in this thread, e.g. directivity, forward lobe and vertical nulls, floor/ceiling bounce, Allison effect, etc. It kind of sums up the things I've learned over the last 35 years or so, bowing to guys like David Smith (speaker dave) and Earl Geddes (gedlee) among many others.
There are other, and probably more effective, ways to deal with florr bounce, perhaps not down to 200 Hz (I can't imagine anything working down that low).
Agreed
I am completely baffled by what this means.
Of course there are no direct studies, who is going to do them? But I would not claim that "all" lateral reflections degrade imaging. Only the very early ones (< 10 ms, approx.), after that they are all good. And to say that "no" early refelections degrade the image is simply not reasonable either. Its the Impulse response "gap" that is important. (I recently learned that Beranek claims importance for this gap as well, so its not without its support.)
Floyd studied what he wanted to prove and used situations that are quite unlike what I do. So his results are not applicable to my situation.
My hypothesis is based on the psychoacoustical work of Blauert and Moore. It is an unproven hypothesis I grant you, but the studies that you keep pointing to simply are not close enough in basic assumptions to apply to the situation that I am talking about.
I will rest my "proof" on my commercial success (not that this is a proof of anything, of course, but it works for me
Do the studies give an explanation why this might be ? I'm assuming its something to do with the fact that the HRTF is very different between 30 degrees off axis and 60 degrees off axis in the horizontal plane (direct path from the speaker and typical reflected path from a side wall) in a way that helps the brain separate the two signals using both spectral balance as well as time delay...
With a vertical reflection you still have a time delay but the HRTF difference between 30 degrees off the horizontal axis and 0 degrees elevation and 30 degrees off axis and a low or high elevation isn't that much, except around 8Khz...
I guess it depends a bit on what type of carpet. Deep shag pile carpet might absorb more treble but fairly flat dense loop pile carpet probably would reflect quite a bit of treble.
I've told this story before but I used to have a very long listening room that was 4 metres wide by 8 metres long but with only a standard 2.4 metre ceiling, my normal listening position was about 3 metres from the speakers but I could easily sit 5-6 metres back, and sometimes did.
In comparing different tweeters I noticed a large difference in the imaging at the extended listening distance of 5-6 metres between a dome tweeter (wide symmetrical dispersion) and a ribbon. (semi-cd horizontal, fairly narrow and progressive with frequency vertically)
As I got further back from the dome tweeter the image location started to become indistinct and "stretched" in the vertical plane, an effect I would describe as a "wall of sound", which I attributed to ceiling (and perhaps floor) reflections of the treble.
With the ribbon tweeter this stretching and defocusing of the vertical image was far less noticeable if at all, even at 5-6 metres - the vertical location of the image was still fairly sharply defined and roughly at the height of the tweeter. Again, I put this down to reduced reflection of high treble from the ceiling and to a lesser extent the floor.
Whats interesting is that the bulk of our elevation perception comes from a fairly narrow frequency range centred around 8Khz, and with the ribbon tweeter in question the vertical directivity at 8Khz is 56 degrees, (+/- 28 degrees for -6dB) which is narrow enough to cut the floor and ceiling reflections in this frequency range dramatically even at an extended listening distance let alone a normal 2.5 - 3 metre distance.
So it seems to me that at least for imaging, floor and ceiling reflections of treble are a bad thing, and it may also contribute to excessive power response in the treble.
I think there's definitely something to be said for asymmetrical directivity at high frequencies - wider (but still controlled) in the horizontal plane and narrower in the vertical plane.
Narrower vertical directivity is frequently cited as a disadvantage with ribbon tweeters but I really think that a modest narrowing of directivity in the treble in only the vertical plane is a good thing, and it certainly seems to benefit imaging.
You're obsessed with knee bends!
Last edited:
I think it is all similar to the cocktail party effect where you can use your directional hearing to perceptually seperate a lateral reflection from a frontal source. Our lateral perception of location is very good, our front to back (say 60 degrees vs. 120 degrees) is fairly poor and the ability to split a ceiling bounce from above the speaker (same azimuth) from the source is nearly nonexistent. Pinna effects might let you weakly guess elevation, but when both sounds come together we just seem to lump them as one frequency abberated arrival.
Its not just HRTF, but the phase or level difference between the two ears that tells us where around the horizontal circle sources come from. For reflections from directly above or below the primary source there is no phase or level difference between our two ears, between the two sources.
Carpet is not as absorptive as you think. Even a small amount of reflection is enough to give audible comb filtering. I went out on a acoustics site survey where a good sized court room had a large hemispherical dome. You could walk to the middle of the room and be at the center of the hemisphere. Because the acoustics had been attrocious the whole dome had been covered with NRC 0.9 material (90% absorptive over the voice range). This was essentially 1" thick best-grade fiberglass. In spite of the treatment you could still hear a pretty dramatic slap back. 90% absorption is still 10% reflection.
Knee bends are good for you! Plus changing a stationary event to a time varying event (frequencies of comb filtering) makes it more obvious to all.
David
Its not just HRTF, but the phase or level difference between the two ears that tells us where around the horizontal circle sources come from. For reflections from directly above or below the primary source there is no phase or level difference between our two ears, between the two sources.
Carpet is not as absorptive as you think. Even a small amount of reflection is enough to give audible comb filtering. I went out on a acoustics site survey where a good sized court room had a large hemispherical dome. You could walk to the middle of the room and be at the center of the hemisphere. Because the acoustics had been attrocious the whole dome had been covered with NRC 0.9 material (90% absorptive over the voice range). This was essentially 1" thick best-grade fiberglass. In spite of the treatment you could still hear a pretty dramatic slap back. 90% absorption is still 10% reflection.
Knee bends are good for you! Plus changing a stationary event to a time varying event (frequencies of comb filtering) makes it more obvious to all.
David
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.