Has anyone tried something like this? A thumbnail bit (without a bead on the cabinet, of course) starts at a large radius then increases curvature as it gets closer to the bearing. The initial radius seems wider than even a 1.5 inch roundover so I'd speculate the on-axis frequency response would be flatter yet, and the gently increasing curvature would allow the wavefront to diffract smoothly around the cabinet.
Besides, the edge looks more interesting than a standard roundover.
Besides, the edge looks more interesting than a standard roundover.
So this is my analysis of the data-
1) The largest round-over is the best, closely followed by the double 22.5 degree bevel, followed by the 45 degree chamfer.
2) 45 degree chamfers may be attractive, but labour intensive, and not as good as round-over.
3) Double chamfers are labour intensive, but at least closer to large round-overs.
2) Anything less than a 1" round-over is merely cosmetic, you may as well choose a square baffle.
3) the implication then, is when designing the HP crossover for the tweeter, one needs to accept some peaks and dips in the on-axis curve, because if you try to completely flatten out the on-axis (eg. DSP), you'll get peaks and drips in virtually every other angle (look at the normalised graphs)
x) Taken to the limit, what's better than a large round-over?
A large spherical baffle?
No baffle?
Thick felt?
https://vaf.com.au/shop/ols/products/signature-i93mkiii
1) The largest round-over is the best, closely followed by the double 22.5 degree bevel, followed by the 45 degree chamfer.
2) 45 degree chamfers may be attractive, but labour intensive, and not as good as round-over.
3) Double chamfers are labour intensive, but at least closer to large round-overs.
2) Anything less than a 1" round-over is merely cosmetic, you may as well choose a square baffle.
3) the implication then, is when designing the HP crossover for the tweeter, one needs to accept some peaks and dips in the on-axis curve, because if you try to completely flatten out the on-axis (eg. DSP), you'll get peaks and drips in virtually every other angle (look at the normalised graphs)
x) Taken to the limit, what's better than a large round-over?
A large spherical baffle?
No baffle?
Thick felt?
https://vaf.com.au/shop/ols/products/signature-i93mkiii
The sunken baffle then needs felt to mitigate all the reflections.
Yes the infinite baffle, which is that's why datasheets for tweeters always look so good. But it really is impractical for a speaker.
The closest approximation to infinite baffle is the in-wall speaker.
Here's the first in-wall speaker that I've seen measured with a Klippel NFS
https://www.erinsaudiocorner.com/loudspeakers/jbl_scl-6/
SF Stradivari clones are pretty close to IB regarding tweeter as well. Sadly most of us don't hae big enough room for them...
http://www.troelsgravesen.dk/PMS.htm
http://www.troelsgravesen.dk/PMS.htm
I agree. Missed that when I looked at the pics. OP, hate to make work for you but you might want to glue some sides on and rerun the tests.
I’m not sure I agree with planet10.
I mean there may be slight differences to this dataset, but the trends will be the similar.
At the frequencies/wavelengths in the region of interest, ie. emitted by a monopole tweeter, the difference between a cabinet of only 1.5-2” depth, and one which is, say 8-12” deep, should be slight, if it exists at all.
second, consider the right angle as 2 surfaces with 0 cuts. From a single cut of 45 degrees, to 2 cuts at 22.5”, to 3 cuts at 15 degrees, 4 cuts at 11.5” degrees, 5 cuts at 9 degrees… to 45 cuts at 1 degree.
similarly, let’s consider the other limit. A small round-over with a radius with us 1/2”, 1/4”, 1/8”1/16”.
The limit to which the roundover can be made smaller is… infinitesimally small. A square edge is the limit to which this can be done.
I hypothesise that the round over will still result in the best normalised off axis response.
Last edited:
Based on my take from the Olson work. It may or may not have an impact, but anecdotal listening between rectangualr and trapezoidal miniOnkens is that the one with the more teardrop shape has a lower audiable diffraction signature.
Only one way to see if this speculation holds true, and that means doing the measures again.
I will note that since i do not use traditional tweeters, the data is relative.
dave
I'm starting to see your point but I think I just confused myself....Is bafflestep an unavoidable trait?
I'm guessing that it is outside of infinite baffle and no baffle, So whether it is up high or down low, is neither here nor there? I mean that a really narrow or really wide baffle seem to be the answer with intentions of keeping it out of the middle of the passband. In particular I would say above 7000 and below 700. Edge treatment gets rid of just the issues of diffraction issues (ripples and time delay issues) ? The wider baffle provides more direct energy, which is why I personally wanted a wider baffle.
"Low enough lows (bigger than the baffle) just wrap around / diffract enough in phase with direct sound so that not much interference "ripple" is seen in axial response. Only the hump around where wavelength roughly equals baffle width which seems to make constructive interference with direct sound. Above the hump all kinds of ripple show up until the driver beams"
This wide baffle has no ripples... that I have ever seen in a sim, at least.
Milliseconds needs to be in perspective to frequency, and at a lower frequency a larger time smear of a few milliseconds doesn't mean much in a room. So I guess the question is...how many milliseconds? Or are we acknowledging that a wide enough baffle is optimal?
Infinite baffle has no edge to interact with, but then you say here "path length along the baffle just "smears" the direct sound" So infinite baffle has max direct energy smear.....This is bad? or just different.
The change in directivity is the transition of the FR above and below baffle step....
OK I think I am seeing the picture now....and its pointing back to keeping bafflestep away from some arbitrary point.
looking at post 70, my thoughts are" Oh, edge treatment can mitigate baffle step....but at a very wide baffle, this would be impractical due to the size of the edge treatment needed to be effective there?" Maybe I said that wrong....Edge treatment can remove the resonant note that occurs right before the baffle looses directivity is what I mean.
I hope I didn't miss anything.... It sounds like to me, a very wide baffle is a version of optimal...at 143hz there are no ripples to worry about and it is so deep into the room territory that would require eq anyway....also creating as much direct energy as I can.
@AllenB "What's unfortunate about an obsession is it closes your eyes. A compromise should consider everything that matters, even at the expense of what doesn't." Lol Allen this explains my approach to a lot of things...In particular "even at the expense of what doesn't"....Group Delay....hint hint lol! even though it matters lol.
If I am lucky, it may be easy enough to create my situation with @mabat 's tools so I can see something more descriptive via akabak
Last edited:
A practical example of edge diffraction mitigation...
Here is my latest project speaker's 0, 20 and 40deg horizontal measurements of tweeter and mid without eq and xo, and completed speaker with LR2@3500Hz. We can clearly see that in a finished speaker most diffraction irregularites will get mixed with midrange's directivity and xo/interference issues.
Box has pentagonal crossection, rounded edges and a trapezoidal ½" thick add-on baffle with 45¤ chamfered edges. So pretty much everything possible done to mitigate diffractions...
Here is my latest project speaker's 0, 20 and 40deg horizontal measurements of tweeter and mid without eq and xo, and completed speaker with LR2@3500Hz. We can clearly see that in a finished speaker most diffraction irregularites will get mixed with midrange's directivity and xo/interference issues.
Box has pentagonal crossection, rounded edges and a trapezoidal ½" thick add-on baffle with 45¤ chamfered edges. So pretty much everything possible done to mitigate diffractions...
Last edited:
Magico A and S series have some effort to minimize edge diifractions too...
https://audio.com.pl/testy/stereo/kolumny-glosnikowe/3487-magico-a5#laboratory
https://audio.com.pl/testy/stereo/kolumny-glosnikowe/3487-magico-a5#laboratory
Last edited:
Edge treatment doesn't mitigate the baffle step, what it can do is reduce the difference between on axis and off axis as seen in the measurements, the width and the depth work together to determine the shape of the step.
A wide baffle pushes any edge effects lower in frequency but it also changes the overall directivity in a way that does not seem optimal to me.
Good example.. akabak doesn't show all the things that matter, but you (we) will 'find' good results if that is what you are 'looking for'. Excessive reliance on the simulator exposes one to potential obsession.
Thanks Dave.
can you tell us about how to came to this conclusion?
I’ve long suspected that the room swamps any measurable differences,
OR the ear-brain interface smooths them out (or both)
Maybe a highly trained listener may be able to do it, or maybe not.
but the conditions would have to be ideal;
eg. Mono speaker/anechoic room/near field.
I’m friends with music and studio professionals (ie. working musicians as a primary job) and some of them have amazing ears- and can pick out speaker deficiencies (or in-room peaks and dips)
“Too much energy at 800Hz; and a half, maybe two dB”
but certainly I can’t critique someone else’s speaker/crossover after listening for just 30 seconds of music.
I suspect that I might not can hear the difference between a large round-over and square edges.
Some of us do roundovers because we’re perfectionist/a bit OCD and want to do everything we can to minimise doubt.