KaiserSoze
Addicted to Fun and Learning
- Joined
- Jun 8, 2020
- Messages
- 699
- Likes
- 592
With DIY kits, the focus is on making the speaker cabinet as simple as possible to reproduce.
Rounded side edges would certainly improve the radiation pattern. However, the radius must have a certain size for this.
This is nothing new, however, as the magazine Hobby-Hifi, for example, already showed in detail 20 years ago.
Here is a small excerpt from the article (Have removed the German comments from the picture, the results speak for themselves, the frequency responses 0° to 60° are shown):
Source: http://hobby-hifi.de/Archiv/01/05_01/05_01.html
View attachment 75493
The test object had a baffle width of 0.17m. This means that the distance of the tweeter to the lateral edge of the baffle is 0.85m, which at a wavelength corresponds to the frequency of 4kHz - exactly the problematic frequency range.
UPDATE: The percentages in the image refer to the baffle width of the loudspeaker. In order to achieve a similar effect as in the bottom diagram with a 0.3m wide loudspeaker (problematic frequency range would be around 2.3kHz), the radius of the rounding must be 45mm.
Two important remarks on this:
- A radius that is too small further exacerbates the problem of edge diffraction - the frequency response becomes even more unsteady.
- Rounding the side edges reduces the "effective" baffle width, the effect of edge diffraction shifts to higher frequencies and "smears" the edge diffraction over a wider frequency range, which means that the crossover will most likely have to be adjusted.
Nowadays, even for hobby loudspeaker developers, it is no longer a problem at all to optimize the radiation of a loudspeaker through simulation (before the first prototype must be produced)- it is almost child's play![]()
I had several thoughts after reading this.
First and foremost, there is little reason to hypothesize that the irregularity in the off-axis response of the X-LS Encore is related to edge diffraction. The overall shape of the off-axis response is strongly indicative of an off-axis response that you get when, at the crossover point, there is a directivity mismatch between the woofer and the tweeter. At face value there is the obvious question of how it would make sense to think that this could be corrected through a technique that deals with diffraction-related problems.
In the MLSSA graphs you show from the article, the curve at the front appears to be the on-axis response; the curve at the back is the response 60 degrees off-axis. We are concerned here with correcting an anomaly in the off-axis response, whereas in the article the response anomaly that is being dealt with is in the on-axis response (primarily at least it is the on-axis response). Thus, the diffraction-related technique as illuminated in the article is not concerned in any useful or particular way with how the off-axis response per se is affected by diffraction ripple or is likely to be affected by a diffraction-mitigating technique that applies expressly to the on-axis response.
It might turn out that in the off-axis response the diffraction ripple for this speaker will produce a dip at about 1.6 kHz and a peak about an octave higher in frequency, such that there will be reasonably good frequency alignment between the problem and the proposed solution. If so, there is at least a possibility (albeit remote) that diffraction reduction will have the desired effect on the off-axis response. But there are two points I'll make. First, nowhere in the article that you included or in your comments is there any hint that in order for this approach to have even a slim possibility of working it is first necessary to investigate whether the off-axis response is affected by the diffraction ripple in the particular way by which reduction of the diffraction ripple would have the desired effect as opposed to making it worse (i.e., the good frequency alignment that may in fact exist was in no way a given). Second, if it happens that the frequency alignment is such that reducing diffraction would have the desired effect on the off-axis response, then it will also be true that the diffraction ripple is beneficial to the on-axis response and that if you reduce the diffraction ripple by rounding the edges (enough for it to matter), the on-axis response will suffer correspondingly. This implies the need to make adjustments to the crossover, superficially analogous to the need you encounter if you add a waveguide to the speaker. Except that it is relatively straightforward to make the indicated adjustment to compensate for the effect of a waveguide, whereas if the on-axis response owes its flatness to the presence of diffraction ripple and you take away the diffraction ripple, the correction that would need to be applied to restore the on-axis response to flatness would be essentially equivalent to correcting for the effect of diffraction ripple in an on-axis response that would be flat were it not for the effect of the diffraction ripple. You might be able to do this using DSP and using multiple parametrically-defined corrections, one for each peak and one for each dip in the diffraction ripple, but it is obviously not something that is realistic in the context of a passive crossover.