KrIVIUM2323
Well-known member
You are absolutely right, especially about what happen after the first 'roller coaster' ( 125 to 500hz). But for this first hump then a shelf will nicely counteract the effect.
Here again you are right, radius of roundover will help regarding diffraction ( for what happen after the 'first roller coaster', 150hz to 550hz on Olson's graph).
But it's all theorical.
Don't get me wrong, Olson's work is a reference about baffle diffraction since this paper was published in 1951 ('Direct Radiator Loudspeaker Enclosures' AES, october 27 1950). And like everybody who studyed it, i agree with what is presented in there.
But it's difficult to translate his work in here: the driver Olson used was choosen to have wide directivity until 4khz and as such it is a small driver, very small indeed (7/8" or 2,8cm diameter). It was a requirement to exacerbate the enclosure behavior.
The circular box size was choosen too in order to show off the deleterious effect that could happen too ( BSC but baffle edge diffraction too). In practice he choose a 2 feet ( circa 60cm) diameter circular box...
Both requirements leads to this awful profile you linked in post 5 and that can be seen in the article.
With the loudspeaker presented here situation is not equal: driver's sd is different and so is the directivity behavior of the driver by itself which will be more directive ( and at lower frequency range) so diffraction in high frequency will be de facto lowered/mitigated.
The ratio of driver's diameter to baffle diameter will be fairly different too, in Olson experiment it was in the order of 1/20, in the loudspeaker presented here it might be in the 1/~1,1 ratio. Here again this will have a very strong and different effect regarding baffle edge diffraction, amplitude will be way less... It's not perfect for sure, but results will be way less awful thant in Olson's paper even if they'll follow the trend displayed ( the profil of diffraction will track but the amplitude will be way, way less).
Now about roundover: for them to have a meaningful effect on edge diffraction the rule of thumb is radius must be equal to 1/4wavelength of lower frequency of interest.
Eg: to control diffraction circa 1khz you need a 10cm ( 4") radius roundover, at 2khz: 5cm (2"), 4khz 2,5cm (1"), etc,etc,...
As the driver used (the Alpair) have an sd of 50,27cm we know it's effective radiating diameter to be close to 10cm.
From there we can guesstimate it's directivity behavior: ka should be in the 3 to 4 range between approximately 3khz and 4,5khz (driver directivity around 90°at -6db) and it'll be beaming from 5,5khz ( ka:5, circa 50° at -6db).*
If we estimate the circular baffle of 'nautilus like' enclosure ( in post 1 with Alpair 7 driver) being approximately 130mm diameter then by rule of thumb the first 'roller coaster' center frequency will be around 900hz ( 115/bafle width in meter).
So our 4 octave shelf filter with 900hz center freq will take care of the baffle step compensation nicely up to 3600hz.
From there we will have the edge diffraction issues to take care about. Round over would need to be circa 2,4cm radius.
If the driver was mounted from behind or a profiled** ring was 3D printed to be put over the basket ring, then it could be a large step toward very minimal diffraction. Even more since the driver is already starting to beam at 90° and will therefore radiate way less energy toward the edge...
Imho, i don't know if i would bother to do a profiled ring anyway as my experience with this kind of enclosure with minimal (1/1,1)baffle area to driver diameter ratio are relatively stealth wrt edge diffraction ( i've built an mtm with 2x 6,5"(sealed with linkwitz transform)/1" using pvc tubing as enclosure in past years, nothing fancy except for the filter -i use dsp with FIR for more than 10years now- and was astounded by results especialy about diffraction, but don't take my words on it, if curious try it and make your own conclusion).
*: you can simulate using 'speaker directivity simulator' excel sheet from here: https://www.tonestack.net/software/speaker-directivity-simulators.html
**: profiled as the inner part of the ring ( toward the membrane) will need to be rounded over too or there will be too much diffraction from there! Ideally it should continue cone profile but MA drivers have very shallow cone profile... i don't know the Dayton drivers but for the one i have they are not vey steep either.
Like always with loudspeakers it's a pick your poison case. Compromise, compromise...
Ok but the felt is effective at which frequency? My own experiment with this prooven to me that for a tweeter freq range it's ok (from 3khz and upward), but for a larger direct radiator it's been ineffective and only roundover did it ( through blind test). Ymmv of course.
Last edited:
