I have a turn table I use to rotate the speaker with a fixed mic position. I will have to modify for an on-wall speaker. I have only measured two speakers, commercial ones that I wanted to understand the sound of, but the measurement were reasonable. I think I get it. I will certainly post any measurements of this project, but I've got a way to go.
The cancelation would either be at 570 Hz (for 150 mm deep) or 745 Hz (for 115mm deep). The 90-180 degree radiation above 570 Hz on typical speakers looks to generally be -10db or less relative to the on-axis response and not much different between 570 Hz and 745 Hz. So, the cancelation would be a fraction of a db at these frequencies. Should I expect the rear radiation to be higher for an on-wall speaker or the SS Illuminator? Thanks!
Whoops, I see that it can be louder than -10 db, more like -6db relative to the on-axis. How do I calculate how much the cancelation would be between two sources with different SPL?
Right, thanks. So...
Cancelation=db_onaxis-10*log10(((10^(db_onaxis/10)*10^-12)-(10^(db_180/10)*10^-12))/10^-12)
If at a cancelation frequency A, the on axis is 90db, and the 180 off axis is 84db, the cancelation will be 1.26db. If the cancelation moved up to a higher frequency (B) where the 180 off axis is 82db, the cancelation will be 0.75db.
For my comparison of a 150mm vs 115mm deep on-wall enclosure, this represents something of a worse-case scenario. The differences are small, but worth considering.
Cancelation=db_onaxis-10*log10(((10^(db_onaxis/10)*10^-12)-(10^(db_180/10)*10^-12))/10^-12)
If at a cancelation frequency A, the on axis is 90db, and the 180 off axis is 84db, the cancelation will be 1.26db. If the cancelation moved up to a higher frequency (B) where the 180 off axis is 82db, the cancelation will be 0.75db.
For my comparison of a 150mm vs 115mm deep on-wall enclosure, this represents something of a worse-case scenario. The differences are small, but worth considering.
To get an idea of the size of the cancellation and reinforcement one can look at measurements such as those in the first post here for a conventionally shaped 2 way. Wide and shallow will have the same reinforcement at lower frequencies but the cancellation dip should be reduced and raised in frequency. To quantify how much is probably best tackled with a detailed 3d simulation using, for example, a BEM solver. If you're not keen then to get a feel there ought to be some measurements for shallow on-wall designs around but I am not aware of any. Anyone?
Here is a likely fairly well designed shallow 2 way on-wall speaker with an overly smoothed plot which will be hiding some of the details. It suggests dips and power response deviations of around 5 dB although without the values of the diffraction off the various edges, radiation pattern of the woofer, applied corrections, etc... it is hard to see quite what is being juggled with what.
Attached are a couple of plots from an initial set of BEM simulations from years ago using python, gmsh, acousto, paraview and matplotlib which are all freely available. The test case was a rectangular box with a circular driver and the dimensions of the speaker used in the measurements in the previous post. The frequency of the dip indicated the speaker was actually mounted a fair distance from the wall presumably to accommodate cables and/or mounting brackets (or the speaker dimensions were not what I found and used!). I spent about a day on it with an initial coarse grid before concluding I would likely need to improve acousto to use for my purposes. The authors didn't respond when approached so I didn't make any modifications to the software and dropped the project. Nonetheless acousto is likely to be fine for a basic analysis of on wall cabinet shapes (but without testing it would be unwise to state with certainty). Akabak seems to be a more popular choice on these forum but it is fairly expensive commercial windows software with a free restricted version. A chap in the software forum is currently considering using bempp as an alternative.
I can't recall the details of what I did (or even what the plots relate to) because it was just an initial quick look and I decided not to proceed. The spike at 900 Hz will be a numerical artefact possibly a resonance of the box or exceeding what the coarse grid can resolve. The point being that if you are comfortable with software it is not a major exercise to quantify the acoustic processes involved in placing a speaker on/near a wall.
Attached are a couple of plots from an initial set of BEM simulations from years ago using python, gmsh, acousto, paraview and matplotlib which are all freely available. The test case was a rectangular box with a circular driver and the dimensions of the speaker used in the measurements in the previous post. The frequency of the dip indicated the speaker was actually mounted a fair distance from the wall presumably to accommodate cables and/or mounting brackets (or the speaker dimensions were not what I found and used!). I spent about a day on it with an initial coarse grid before concluding I would likely need to improve acousto to use for my purposes. The authors didn't respond when approached so I didn't make any modifications to the software and dropped the project. Nonetheless acousto is likely to be fine for a basic analysis of on wall cabinet shapes (but without testing it would be unwise to state with certainty). Akabak seems to be a more popular choice on these forum but it is fairly expensive commercial windows software with a free restricted version. A chap in the software forum is currently considering using bempp as an alternative.
I can't recall the details of what I did (or even what the plots relate to) because it was just an initial quick look and I decided not to proceed. The spike at 900 Hz will be a numerical artefact possibly a resonance of the box or exceeding what the coarse grid can resolve. The point being that if you are comfortable with software it is not a major exercise to quantify the acoustic processes involved in placing a speaker on/near a wall.
Attachments
Well, Heißmann certainly used a way of gating on his measurements. That smooths things of course. But I think with such a shallow (and shaped) enclosure sound power might be quite a tad better than the classical deep monitor speaker shoved at the front wall. Even better than the monitor some feet away from it, actually.
Here is an in room measurements of my Dali Ikon OW speakers with no gate or smoothing. The center of the woofer is about 135 mm from the wall and there should be and is a cancelation at ~640 Hz. It is much more dramatic than my calculations full of assumptions suggests, but it is also very narrow. I suppose part of this is the effect of the parallel walls of the room. I don't think I've ever looked at these measurements without smoothing and I have always assumed that these narrow cancelations are not audible. I will sit down with some sweeps and test tones and try to test that assumption.
Quite possibly their presence alone, regardless of them being parallel. Despite the way it appears, you can't tell much unless you don't include later reflections.
http://audiocompaniet.no.24nb8.srv....ere/stativhoyttalere/bremen-production-3d8-v2
An idea worth trying, it doesn't have to be egg-shaped of course, it's just a fancy design, but the idea is sound (pun intended)
An idea worth trying, it doesn't have to be egg-shaped of course, it's just a fancy design, but the idea is sound (pun intended)
I've gated the measurement to limit it to the front wall's first reflection. There still appears to be the expected baffle distance cancelation ~640 Hz, but now the magnitude is closer to what I expected. I suppose it is important to remember that the first reflections are the most important as they are perceived as part of the original signal in a normal sized room.
Hi, why you have this as goal? This is limitation, which is fine, it just limits the performance and thus important to be sure why it's there.
On wall speaker has multiple issues, one is how it interacts with the wall, and another is stereo image which is literally nailed down.
On stereo image, I find toe-in and positioning in general be the most important thing. When ability to adjust the system this way is taken away the performance is now limited. This might be just fine if one doesn't care about it, or if the system is further optimized. Basically, one would need as smooth off-axis response as possible because it's highly likely that reference angle doesn't hit the ear. Thus, custom waveguide could help optimize wiggle, flatten out DI to any angle. For same reason edge diffraction should be avoided as much as possible, to keep sound the same to listening window. For these reasons I'd do three way speaker, tweeter and mid with as small "enclosure" as possible, to get any edge as close to the transducers as possible, everything as flat against wall as possible.
Separate bass unit somewhere. I'm afraid that in worst case the whole wall becomes a transducer if the speaker is not decoupled sufficiently from the wall. I'd imagine any wall resonance is quite a low in frequency and it is another reason to not put bass on the wall. If the whole wall resonates, I'd think it's sure way to muddy up sound, perhaps worse thing than some SBIR, what's what you try to avoid here.
Interaction with the wall cannot be avoided, because wavelength goes from very long to very short, and the enclosure size is somewhere in between, there is by necessity some frequency bandwidth that has some interference show up. For above reasons I think hanging speaker on wall isn't optimal, perhaps it's fine enough. I suggest building quick prototypes and checking it out how well it works, before committing full surface finnish and so on. Have fun
Keep in mind that gating has a smoothing effect, so the cancellation may be stronger than shown in the gated response.
Happy to use a wave guide. The Seas DXT has a small one. But I don't want to design my own. Both these constraints are about keeping the project manageable for me.
The Seas DXT tweeter your looking to use is a very good driver, dare I say better than most inexpensive AMTs in regard to HD in the lower treble. Mind you, I'm coming from my own preference here and don't like the typical rough sound of most AMTs on the market. Exceptions to this rule are some higher end Mundorfs, which have a special coating on the membrane surface to reduce the surface resonances which cause the elevated H3 down low. A decent 25mm metal dome tweeter won't have this issue by design, being pistonic over most of its range up to past 10k.
The benefits of a WG are reduced by the wall boundary given the cabinet is shaped to reduce edge diffraction. I highly recommend using surface dampening media (thin foam) on the sloped sides to eliminate HF edge reflection modes and diffraction ripple.
What works even better IMO is to design a sloped baffle which insets the tweeter as close to the wall as possible. This aligns the acoustic centers of both drivers and produces a more cohesive, better integrated sound at the xover overlapping response region, simplifies xover design and can improve transient response.
The 2 pi radiating field produced by wall mounting is effective down to 1/2 WL of the distance from the front baffle to the wall, as is the distructive cancelation frequency mode. The ceiling and floor will produce its own modes here and its important to account for the collective destructive and reinforcing modes of all boundary effects interacting with each other. This again can be minimized with wall and ceiling treatment, so at a minimum your reducing the first direct reflected HF modes. The diffuse field modes aren't as damaging here given the listening distance doesn't allow for the convergence of secondary wall / ceiling reflections.
From my experience installing soffit mounted studio monitor setups, the main influence of the hemispherical radiating field over a floor standing speaker is mainly across the lower midrange, discounting cabinet edge diffraction components and the resulting ripple. The typical range of boundary reinforcement of wall mounting mainly boosts everything under 1kHz by up to 6 dB with a gradual declining 2-3 dB/Oct slope. This isn't linear and will have its own ripple to contend with. You'll also want to avoid horizontal driver orientation on the baffle for the most uniform polar response (sweet spot) at the listening location.
The benefits of a WG are reduced by the wall boundary given the cabinet is shaped to reduce edge diffraction. I highly recommend using surface dampening media (thin foam) on the sloped sides to eliminate HF edge reflection modes and diffraction ripple.
What works even better IMO is to design a sloped baffle which insets the tweeter as close to the wall as possible. This aligns the acoustic centers of both drivers and produces a more cohesive, better integrated sound at the xover overlapping response region, simplifies xover design and can improve transient response.
The 2 pi radiating field produced by wall mounting is effective down to 1/2 WL of the distance from the front baffle to the wall, as is the distructive cancelation frequency mode. The ceiling and floor will produce its own modes here and its important to account for the collective destructive and reinforcing modes of all boundary effects interacting with each other. This again can be minimized with wall and ceiling treatment, so at a minimum your reducing the first direct reflected HF modes. The diffuse field modes aren't as damaging here given the listening distance doesn't allow for the convergence of secondary wall / ceiling reflections.
From my experience installing soffit mounted studio monitor setups, the main influence of the hemispherical radiating field over a floor standing speaker is mainly across the lower midrange, discounting cabinet edge diffraction components and the resulting ripple. The typical range of boundary reinforcement of wall mounting mainly boosts everything under 1kHz by up to 6 dB with a gradual declining 2-3 dB/Oct slope. This isn't linear and will have its own ripple to contend with. You'll also want to avoid horizontal driver orientation on the baffle for the most uniform polar response (sweet spot) at the listening location.