Actually, the path length delta is what I always assumed I was dealing with. But sometimes I have found notches in unusual places. Made me consider the possibility that there might be a self-interference notch where the boundary was 1/4λ away, even if the angle of incidence wasn't in line with the direct path. That's actually a side-track we just got stuck on.
The main thrust of my arguments, actually, is that there are room modes (and other self-interference notches) between 100Hz and 200Hz in most listening rooms. The multisub arrangement is a large 3D array intended to create dense interference, useful below 100Hz or so, where localization isn't an issue. My argument is that a modified form of this approach, more like a traditional line array, is useful for the transition region between 100Hz and 200Hz. You don't want distant subs, as they might betray their location. But a closer blended source is useful in this range.
That argument has been presented before but I find it lacking. I think we can all agree that self-interference notches above a few hundred Hertz are easily dealt with using simple room treatments. So I have no argument with that part. I think most of us here would also agree that room modes below 100Hz can be mitigated with the multisub approach. Again, no argument there. But what I would stress is the importance of the transition range, the region between 100Hz and 200Hz. I don't think it should be casually dismissed. Absorbant materials aren't thick enough to do much, and distant subs are too far away. If you low-pass them high enough for blending with the mains, they begin to betray their location. Seems to me the best approach is to blend the midbass/midrange with sources placed just a few feet apart. This can be incorporated with more distant subs to smooth lower frequency modes, if needed.
Placing a driver near a boundary changes the combined response due to the following effects:
- reflection summation
- 4pi to 2pi loading
- potentially change the radiation impedance of a driver (very slightly)
None of these cause low frequency notches, if not due to reflection. It sounds like a lot of discussion due to a measurement that wasn't fully understood.
What physical property could possibly be creating these mysterious notches? KISS answer requested.
- reflection summation
- 4pi to 2pi loading
- potentially change the radiation impedance of a driver (very slightly)
None of these cause low frequency notches, if not due to reflection. It sounds like a lot of discussion due to a measurement that wasn't fully understood.
What physical property could possibly be creating these mysterious notches? KISS answer requested.
I think most of this stuff is pretty well understood. The disagreement, as I see it, is mostly on the significance of the 100Hz to 200Hz region. Maybe there's not even disagreement there, I don't know. Seems to me all would be concerned with this range, and maybe we all are. It just appears to me that some people say a thick rug will damp the notches in that region. I wouldn't agree to that.
As far as the cause of the notches, I'm sure all would agree that it is self-interference of some form. It is reflected energy that sums with the original delayed by an odd-multiple of one-half wavelength. Boundary notches are reflections with this phase relationship; Room modes are standing waves with this relationship. Standing waves are nothing more than periodic reflections between opposite boundaries. So it's all about cancellation notches from reflected energies.
As far as the cause of the notches, I'm sure all would agree that it is self-interference of some form. It is reflected energy that sums with the original delayed by an odd-multiple of one-half wavelength. Boundary notches are reflections with this phase relationship; Room modes are standing waves with this relationship. Standing waves are nothing more than periodic reflections between opposite boundaries. So it's all about cancellation notches from reflected energies.
No, what I am saying is that there is no floor bounce in that region at a normal listening position, and that a thick rug is at least moderately effective at the frequency where floor cancellation does occur. The notches you see in the 100-200 Hz region are either a result of microphone placement at other than a normal listening position or reflections/cancellations from something other than the floor.
Don't think in terms of just the floor bounce - think in terms of vertical modes. In fact, don't just think in terms of just vertical modes, consider them all. That's what is really in play in-room, that's what the multisub approach deals with - room modes. I think the lower midrange notches are probably mostly from the vertical modes, but it doesn't necessarily have to be so. For example, the wall behind the speakers probably plays a big role in many installations.
Seems like a lot of discussion regarding what frequency the floor notches are at in measurements.
Here you go guys, give this a shot. I created it to speed up my testing and look for anomalies in my test results (verify lack of resonances in the test set up etc.).
It has a few features you won't see in RRC type tools. Feel free to use for non-commercial use.
Dave Dal Farra
Here you go guys, give this a shot. I created it to speed up my testing and look for anomalies in my test results (verify lack of resonances in the test set up etc.).
It has a few features you won't see in RRC type tools. Feel free to use for non-commercial use.
Dave Dal Farra
Neat spreadsheet applet, Dave, thanks.
I think a lot of what keeps side-tracking us is the sole focus on the floor bounce notch, rather than the wider picture of all in-room reflections, sources of self-interference and room modes. I know I am responsible for this to some degree, because I brought it up. I really do think it is side-tracking us though, because it is just one boundary out of six.
Another thing that occurs to me is a possibility for the differences in the predicted (100-200Hz) transition region notches and the measurements. At this frequency, wavelengths are between five and ten feet long. This is a range where the walls, ceiling and floor aren't the only reflectors. Tables become reflectors too, potentially shifting the "floor" to the table surface. I think we all know this, and shouldn't kick ourselves for maybe overlooking it.
So above 100Hz, we might have even more than six reflective boundaries, depending on the furniture and stuff in the rooms. I guess in a way that's a no-brainer, but I know sometimes even the no-brainers get overlooked.
I think a lot of what keeps side-tracking us is the sole focus on the floor bounce notch, rather than the wider picture of all in-room reflections, sources of self-interference and room modes. I know I am responsible for this to some degree, because I brought it up. I really do think it is side-tracking us though, because it is just one boundary out of six.
Another thing that occurs to me is a possibility for the differences in the predicted (100-200Hz) transition region notches and the measurements. At this frequency, wavelengths are between five and ten feet long. This is a range where the walls, ceiling and floor aren't the only reflectors. Tables become reflectors too, potentially shifting the "floor" to the table surface. I think we all know this, and shouldn't kick ourselves for maybe overlooking it.
So above 100Hz, we might have even more than six reflective boundaries, depending on the furniture and stuff in the rooms. I guess in a way that's a no-brainer, but I know sometimes even the no-brainers get overlooked.
I find this also. Floor bounce can be quite dramatic when you have a microphone (or listener) fairly close to the speaker. As you move away the notch goes up in frequency and tends to get lost in the room standing wave issues.
Not to diminish the Allison studies but he did a lot of averaging of curves of woofers in similar positions in a variety of rooms. This would, of course, diminish the standing wave effects of rooms with different dimensions, while retaining the boundary effects. (making the boundary effects appear to dominate) In the end it is the combination of the 2 effects that determines LF in-room response.
David S.
Hey Wayne,
I don't know if this will work high enough in frequency for you but it truly removes the effect of the room on LF performance.
Double Bass Array (DBA) - The modern bass concept! - AVS Forum
2 things are involved: a woofer array on one side of the room launches an essentially flat wave front. On the opposite side is a second array which, after an appropriate time delay, emits a negative wavefront to just cancel the incoming wave. With the wave cancelled there is no return bounce, hence no standing waves.
The other essential element is that the woofer arrays are laid out in a pattern that prevents standing waves in the 2 dimensions along the orriginating plane. (This determines the effective bandwidth of the system.)
Very neat.
David
I don't know if this will work high enough in frequency for you but it truly removes the effect of the room on LF performance.
Double Bass Array (DBA) - The modern bass concept! - AVS Forum
2 things are involved: a woofer array on one side of the room launches an essentially flat wave front. On the opposite side is a second array which, after an appropriate time delay, emits a negative wavefront to just cancel the incoming wave. With the wave cancelled there is no return bounce, hence no standing waves.
The other essential element is that the woofer arrays are laid out in a pattern that prevents standing waves in the 2 dimensions along the orriginating plane. (This determines the effective bandwidth of the system.)
Very neat.
David
I'd love to implement this, not likely to happen. I do have one question, probably answered in the thread(s) related to it. What is the impact in other rooms in a typical home? Is there any reduction of bass energy that passes through walls/floors/ceilings? Or is it magnified? My initial thought is that it's reduced, since some room modes are not allowed to become established.
Dave
Last edited:
Absolutely. In-room, all the boundaries come into play. I've often times thrown the phrase "floor bounce" around but in reality what it was is a combination of path length deltas and the vertical modes, probably mostly the second one. I've just noticed that in practically every room, there's a notch around 100-150Hz. And like you, I've noticed the notches above a few hundred Hertz become very small.
Good link, Dave, thanks. Most appropriate in this thread.
I have seen that approach, and been impressed by it. I have also been influenced by Todd Welti's studies in the mid 2000s. That's what really pushed me in this direction, to be honest. Earl Geddes and I used to talk about it a lot in 2005 or so, as he was refining his random placement approach. That's when the flanking sub approach occurred to me.
There were a couple things I noticed back then.
One was that these troublesome 100-150Hz modes weren't removed by distant subs. If they were crossed over high enough to smooth the highest frequency modes, they started to betray their position. But if you didn't push them that high, the higher frequency modes stayed strong.
The second thing I noticed was while measuring some line arrays for a friend. They always smoothed the notches, didn't have a floor bounce notch outdoors and did a good job of mitigating the higher frequency modes indoors too. That's when I realized the higher frequency modes were probably vertical.
I have a cornerhorn that uses a couple techniques that prevent notches. One is tight placement in the corners, which reduces the Allison effect. Another is blending of a midhorn and woofer between 100Hz and 300Hz. When I measure it in a trihedral corner outdoors, it has no notch, like the line arrays.
So when I thought about all of those things, I realized that a flanking woofer, blended in the 100-200Hz range, might be very useful for stand-mounted two-way loudspeakers. They are raised off the floor, and usually some distance from the wall behind them. It does appear to work very well, and I've seen others use this approach successfully too.
Last edited:
A line array helps the floor bounce simply by its directivity. An alternative way to look at it is that, say you have a short array of 4 woofers, now you have a continuous group of 8 woofers when you consider the virtual mirrored group below the floor. In the orriginal case you had 1 woofer and a reflected woofer and a significant path length difference. Now you have 8 sources and each one is incrimentally farther than the top one. That seems likely to reduce the floor bounce effect.
I've mused over creating a purpose made array that has a null in energy in the downward direction specifically in the 100 to 200 Hz range. It isn't that difficult conceptually. Still, if you already have a floor standing cabinet (as would likely be needed) you might as well take the Allison approach and put the woofer at the floor boundary and the mid higher up.
Regards,
David S.
That's an interesting idea, David. I'd like to hear more about what you propose.
If you don't get it done for a while, please pop over on AudioRoundTable.com and let us know your progress. That's where I usually hang out. I come here sometimes, and over on AudioKarma, but mostly my spot is ART.
That's how I've usually done it. In the cornerhorns, I run the midrange down to low frequency and blend it with the woofer. Everything is pushed back tightly into the corner. When I can't put the speakers in corners, I run stand-mounted two-ways and augment them with subs, conceptually doing the same thing.
I like visualizing the source reflection as mirrored virtual sources. Makes it easier to picture what's happening.
But I wonder, do you really think directivity has anything to do with it? I mean, above the modal range I definitely see vertical arrays making a narrow vertical beamwidth. And outdoors too. But indoors, in the modal region, I consider directivity to be ambiguous. Maybe it's just semantics, maybe directivity is the way to say it. Heck, I think saying directivity is ambiguous is probably not exactly right either - the energy distribution isn't uncertain or inexplicable, it just isn't a beam. It's pockets. That's really what I mean by "ambiguous", the energy distribution in the modal range is "complex".
But however you word it, the way I see it is boundary interference and room modes create pockets of energy, like shown on page 27 of my "uniform-directivity" paper. There's a few pages there on room modes page 27 has graphs of energy distribution in the room.
As an aside, this complex energy distribution - the fact that directivity doesn't have the same "beamwidth" meaning in the modal region - is why I don't use baffle step filters in any of my speakers. Of course, the constant directivity cornerhorns wouldn't need it because they are pretty much flush with the boundaries. They're packed in tight. But even my two-way speakers that aren't usually put quite so close to boundaries don't incorporate baffle step filters.
My thinking is the baffle is physically large, so the transition from omnidirectional radiation to halfspace radiation never really occurs. Outdoors, it would, but indoors, this transition occurs below the Schroeder frequency. So there really is no such thing as "freespace radiation". I'm not sure I would want to add power to the mains in the modal region as a BSC filter would do, so I suggest using multisubs instead.
Looking at it another way, I think I see what you mean. In a sense, the narrower vertical directivity of an array prevents the vertical axial modes from forming, or at least makes them less strong. So vertical arrays are good for midbass in-rooms, because they control directivity. That's another way of looking at it.
I always saw it as where the reflected interference from one driver cancelled, sort of making it be "off" for a listener, another driver was in a different position, one that wasn't suffering destructive interference, so it was "on" for the listener. That's how I described it in the post linked below.
Maybe it's just two ways of saying the same thing, actually the reverse of one another, now that I think about it. The useful pattern, where the energy is, is created where path length deltas form constructive interference. What marks the outside of the pattern is where the path length deltas become destructive. This is what creates directivity from arrays and across the diameter of a speaker cone at higher frequencies.
Dual woofers, flanking subs or midrange/woofer blended in the lower midrange are all forms of truncated arrays. The two sound sources plus their floor reflections make four-node arrays (two physical, two virtual). So I think I'm with you on this, I just used different terminology.
Not surprised that I'd follow your approach, actually. I always loved the 4430 too. Take the 4435 and put one woofer on top of the other instead of laying them side by side, and you have the same sort of thing I'm thinking about.
Actually I was starting to swing over to your way of thinking about it.😛 If it was directivity then we would see a broader drop in level, I think.
In the end we are comparing two multielement line arrays, both the same length. One has two elements only (the end two) and the other has 8 elements. I think more energy in the middle, from the extra drivers, is giving broader directivity, but gets rid of the polar response null that we see as a frequency response null.
My brain hurts.
David
In the end we are comparing two multielement line arrays, both the same length. One has two elements only (the end two) and the other has 8 elements. I think more energy in the middle, from the extra drivers, is giving broader directivity, but gets rid of the polar response null that we see as a frequency response null.
My brain hurts.
David
Hey David, just a question I have wondered for a long time:
Why were the two woofers in the 4435 mounted side-by-side?
I mean, I know one was only used up to 100Hz, adding some oomph on the bottom end.
But still, why was the choice made to put them side by side? Did the marketing guys have a hand in it, perhaps?
Why were the two woofers in the 4435 mounted side-by-side?
I mean, I know one was only used up to 100Hz, adding some oomph on the bottom end.
But still, why was the choice made to put them side by side? Did the marketing guys have a hand in it, perhaps?
Hello Wayne
If I remember right part of the reason was to allow for a direct replacement cabinet size wise for the competing Urie's think 813 at the time. At least that's part of the scuttle butt over on Lansing Heritage. Curious to see if this is correct.
Rob🙂
Last edited:
The system format was laid out by marketing, or more properly the pro product planning group. This was primarily Mark Gander, with some input from Gary Margolis, John Eargle and sometimes Bruce Scrogin (for Export). We in engineering were just the minions charged with making it perform, we had no say in what would be designed. It was the most logical layout as the system became more square rather than tall. Remember that a lot of these were soffit mounted in studios, few were free standing in a HiFi configuration. The UREI 813 and 815 were upstart competition and they had the same format, so that was part of the equation too.
With a 2 1/2 way network there was no downside to this configuration.
David
Wife and kids road tripped one weekend and I grew a beard, ignored hygiene social conventions, and locked myself in for 2 days attempting to find a magic formula for floor reflection absorption, to try and lengthen my measurement FFT windows indoors.
I used a 3" thick futon, gobs of high absorption coefficient semi-rigid rockwool, huge battens of Solen Poly and combinations there of. Nothing resulted in reasonable absorption at these low frequencies. Not too surprising, you either need significant thickness, or resonant absorbers (ie pegboard style or constrained air volume). I even tried using an air space under the absorbers, minor improvement (see Ethen's site for more info). I'm not saying it does nothing, but you still get a deep notch down low.
The material has to be piled very thick (2 ft+) to get significant reduction in the floor notch. The problem this creates is that the higher frequency reflectivity off of the absorber INCREASES by piling it high. Most people don't know this but absorption coefficient is usually measured normal to the material. The coefficient decreases significantly as the angle of incidence shallows from 90 degrees.
Futons on floors will help higher in frequency but do next to nothing below 200Hz.
Dave
I was trying this a short while back with the dipole measurements of the UE response using HOLM and noticed this, even with fiberglass batting. What worked best was putting a layer of more dense acoustic foam on the floor, then layer the fiberglass on top of that. I've got some 18" x 24" x 3" thick convoluted foam used in packing crates. This helped the absorption of lower frequencies, provided improved midrange absorption all while reducing any reflection off of the top.
In the end it still didn't provide as much benefit as I expected. These show without/with floor damping.
As you can see, I was able to extend the window, but not as much as I had hoped.
Dave
- Status
- Not open for further replies.