Sadly, along with the diffraction, so to goes the imaging……the surface area draws attention to its own presence in the 3D space and all is lost. I’ve mixed in plenty of rooms with soffit monitors and while they’re terrific for the analytics, I’d NEVER choose this configuration for personal enjoyment.
I’ve been repeating this for decades now…….narrow baffle speakers out into the room away from walls rule unequivocally with minor room treatment. The ‘problem’ has always been that small/narrow doesn’t equate to loud ……..power compression and distortion become significant components once you turn it up and dynamics fall short of what compression drivers can do. I’m going to solve this problem someday on a 5” wide baffle…..mark my words! Lol
If you are interested in off axis behaviour of speakers - look around at Audiosciencereview. There are great measurements of good an not so good speakers and always some explanation why.
https://www.audiosciencereview.com/forum/index.php?threads/neumann-kh120-ii-monitor-review.46362/
In my speaker building jurney I detected that over all room response is more important as I thought and speakers sound more "stable" from room to room when they behave in this regard. And more and more manufacturers take that serious too.
I’ve heard a few designs (wide-baffle) that lessen the negative effect. Open-baffles in particular do so, but there is still some effect as you go higher in frequency.
Line array.
I should take some pics of my 12" box that I made (OK, after it gets varnished that is). Laboriously curved sides made with solid wood — never again! 
To go 1 better than 90°, there should just be an intermediate angle of 45° and that's it. Maybe some rustic smoothing using an angle grinder with a foam sand paper attachment (60 grit is amazingly fast on those things, but prone to burning so it needs to be on a slow gear).
Once you start feeding the OCD monster it just gets bigger and bigger, and you end up with mirror finish lacquer for no reason.
To go 1 better than 90°, there should just be an intermediate angle of 45° and that's it. Maybe some rustic smoothing using an angle grinder with a foam sand paper attachment (60 grit is amazingly fast on those things, but prone to burning so it needs to be on a slow gear).Once you start feeding the OCD monster it just gets bigger and bigger, and you end up with mirror finish lacquer for no reason.
https://www.diyaudio.com/community/...00-1-fusion-f253-project.408815/#post-7591053
same as my project, stunning match of drive units, made for each other
It shifts difraction to higher frequencies?
And high freqeuncy defraction is not defeated with a simple 15-25mm rounding?
And probably less obtrusive because the involved frequencies are lower.
But of course, to eliminate the diffraction effects of a large baffle is much harder, because it needs very large roundovers if a small diameter (relative to the baffle), wide dispersion driver is used on it at those frequencies where the diffraction occurs.
But of course, to eliminate the diffraction effects of a large baffle is much harder, because it needs very large roundovers if a small diameter (relative to the baffle), wide dispersion driver is used on it at those frequencies where the diffraction occurs.
Sound interacts with objects whose size relates to wavelength, 10cm wavelength would interact with 10cm object just the same as 100cm wavelength interacts with 100cm object.
Any/none roundover is enough as long as it starts as close to the driver as possible, in other words you have minimal baffle around your transducer.
If it doesn't, the secondary sound source that emits at the edge when sound diffracts gets wider in bandwidth. Simplified, secondary sound source at the edge is roughly bandwidth from baffle size wavelength to driver size wavelength, so the more they differ the wider bandwidth the secondary sound source is. If your baffle is 30" and driver 15", then diffraction secondary sound source is loudest between 30" - 15" wavelength, about 500-1000kHz. If your transducer was 1.5" and baffle 3", the bandwidth would be 5-10kHz. Response of both systems would look exactly the same in graphs (using ideal drivers and objects), except the other is 10x the frequency as it's 10x smaller.
More reasoning:
If there is 15" woofer, and 1" edge roundover, the edge round over has almost no effect no matter what the box is, because a 15" woofer would beam high frequencies so that very little short wavelengths get to the edge / roundover. Only lows, roughly 15" and longer wavelengths would radiate toward edge and diffract, and to mitigate secondary sound source happening on 15" wavelength, either the baffle needs to be max 15", or if it's wider the roundover must be big enough compared to wavelength to be effective. Thus such small roundover doesn't do much with such transducer, it wouldn't matter if it was there or not.
It's easy just to think that minimizing flat space around driver is the goal what roundovers need to achieve. What is big and what is small depends on wavelength. If the box needs to be wide, use as big of a woofer as fits so that it fills the whole front panel, and the condition is met. If driver is smaller than the baffle, then just make roundover that starts immediately beside the transducer. Think sphere is best. If this is not happening, then you'd have non optimal edge diffraction secondary sound source happening, and that might be just fine because it is another question what is audible and what is not. I do not know how audible each bandwidth would be as I really don't know how to listen diffraction other than feeling high diffraction (ripple) makes an uneasy sound that changes with listening position and never feels "stable", while low diffraction system enables "stable" sound that doesn't change much with location.
Any/none roundover is enough as long as it starts as close to the driver as possible, in other words you have minimal baffle around your transducer.
If it doesn't, the secondary sound source that emits at the edge when sound diffracts gets wider in bandwidth. Simplified, secondary sound source at the edge is roughly bandwidth from baffle size wavelength to driver size wavelength, so the more they differ the wider bandwidth the secondary sound source is. If your baffle is 30" and driver 15", then diffraction secondary sound source is loudest between 30" - 15" wavelength, about 500-1000kHz. If your transducer was 1.5" and baffle 3", the bandwidth would be 5-10kHz. Response of both systems would look exactly the same in graphs (using ideal drivers and objects), except the other is 10x the frequency as it's 10x smaller.
More reasoning:
If there is 15" woofer, and 1" edge roundover, the edge round over has almost no effect no matter what the box is, because a 15" woofer would beam high frequencies so that very little short wavelengths get to the edge / roundover. Only lows, roughly 15" and longer wavelengths would radiate toward edge and diffract, and to mitigate secondary sound source happening on 15" wavelength, either the baffle needs to be max 15", or if it's wider the roundover must be big enough compared to wavelength to be effective. Thus such small roundover doesn't do much with such transducer, it wouldn't matter if it was there or not.
It's easy just to think that minimizing flat space around driver is the goal what roundovers need to achieve. What is big and what is small depends on wavelength. If the box needs to be wide, use as big of a woofer as fits so that it fills the whole front panel, and the condition is met. If driver is smaller than the baffle, then just make roundover that starts immediately beside the transducer. Think sphere is best. If this is not happening, then you'd have non optimal edge diffraction secondary sound source happening, and that might be just fine because it is another question what is audible and what is not. I do not know how audible each bandwidth would be as I really don't know how to listen diffraction other than feeling high diffraction (ripple) makes an uneasy sound that changes with listening position and never feels "stable", while low diffraction system enables "stable" sound that doesn't change much with location.
Last edited:
In my experience, if big roundovers are not an option, a relative small midrange driver on a large baffle is best to used in offset arrangement, like with the larger ATCs do. It's not a perfect solution, but good enough, even for off-axis if the driver position on the baffle is clever (say golden ratio).
Of course if the baffle is so large, the involved frequencies of the diffraction could lie below the room's Schroeder frequency and the picture changes somewhat.
Of course if the baffle is so large, the involved frequencies of the diffraction could lie below the room's Schroeder frequency and the picture changes somewhat.
Last edited:
For the 3rd time, no it doesn't. You are speaking technically, not practically, at one point we did model my baffle and the difference between 90degree edges and a proper round over were miniscule. So the "must be" part is where I disagree.
Remember that as the distance between sound sources increases (delay increases), the comb filtering at higher frequencies will begin to average out (but it IS still there). The problematic destructive interference will be lower in frequency. Low enough frequencies will simply not care about changing from half-space to full-space radiation. "100 Hz doesn't care about your baffle edge diffraction."
OK above is 500mm with and without proper edges
Below 810mm ~32" with 90 degree edges.
I see now what you guys are saying, if this sim is accurate, there are "issues" in the HF. I don't know to say about that peak at ~2.6khz. Not an issue for those designing towards directivity vs omni
Yeah perhaps there is misunderstanding, everything is relative to wavelength. I should use relative dimension to avoid confusion, I'm not sure how much more simpler one could think about mitigating diffraction ill effects than minimizing flat baffle area around transducers though.
Again simplified example:
If driver diameter is d and edge roundover radius is d/10, the roundover isn't useful and could be left out.
And that is because round over is most effective of wavelengths close to it's radius, in this case d/10. As the driver is 10 times bigger and beams such short wavelengths, there is very little sound at the edge at wavelength where the roundover is effective, thus there is not much diffraction backwave either, and the roundover could be left out. Beaming is also due to the transducer being big to wavelength.
In general everything acoustics relates to wavelength. Pick any d that is relevant to your task at hand.
Last edited:
Actually, the baffle response becomes more intrusive and is easily locatable in a 3d space.....crushes imaging unless the room is large and the stereo soundfield wide.....and if the engineer summed mono up past 120hz or so?......a dead whale on the beach.