Building the arraysWhatsNew

Smooth curves to avoid diffraction

Within audio replay, even the small details may matter. Or at least make a difference, a difference we can measure. If you look at the top picture, you’ll see a slice of the array. Sliced it right trough the center of a speaker driver cone. It should make the title: smooth curves clear if one follows the outer shapes. All done to get as little diffraction as possible.

Look closely at the cone shape and project that shape outward. You’ll see where I tried to achieve smooth curves from the speaker towards the enclosure. For this reason I created a rounded enclosure shape, designed to be smooth. Starting at the shape of the cone and going outward from there into the shape of the enclosure without any sudden changes like bends or corners. Keeping the diffraction profile of the speaker to a minimum. While the surround may still stick out a bit, there’s not much I could do about that.


Within this whole project my goal was to hear the direct sound first and have a so called ‘reflection free zone’ before the room sound follows. If there were any sharp corners on the enclosure, they may act as secondary sound sources due to diffraction and mess up the frequency curves between on- and off-axis. How do we hear something like that? It can make it easier to ‘hear’ the speaker position in the room. In other words, one can more easily identify the speaker as the source of the sound when your eyes are closed.


Let’s see that in an analysis. Years after I came up with this curve we’ve put this theory to the test. With a BEM simulation program called ABEC we compared the output of a more regular shaped enclosure to this ‘optimized’ shape I came up with.

First up, a regular enclosure shape with only a small round-over, driver is flush mounted:

Rectangle enclosure

The resulting polar frequency output horizontally:

Polar frequency output rectangle flush

Next up, the enclosure shape I have come up with and a flush mounted driver:

Smooth curves cause less diffraction with this flush mounted driver

The resulting polar frequency output horizontally:

And last but not least my enclosure shape including the back mounted driver:

Smooth curves with a back mounted driver ave an even lower diffraction profile

The resulting polar frequency output horizontally:


Clear to see if we look from top to bottom how the results become better with each step. The final smooth curves version with a back mounted driver are showing a frequency result that’s way more linear and gradual. It does show that the little things that might seem unimportant do play a role to get towards a good end result. It isn’t for ‘show only’ that I’ve chosen this particular design. In fact, this was a case where form follows function.

For me, it felt like a big feather in my cap to see these ABEC results. I had made these choices based on intuition and some playing with a wavefront simulator years before that simulation. It showed me that the gamble I took was worth it. With diffraction, avoiding it is better than fighting it.


I used these ABEC simulations of the wave front in the VituixCAD simulations to determine the frequency dependent filters. I could have used results from the diffraction tool from VituixCAD itself but I did both to be sure. The differences weren’t huge and the filters worked with both models. Not surprisingly, the bigger differences between ABEC’s smooth curves and the Vituix model were in the off-axis behavior of the array. In the end there was a strong correlation with reality of these ABEC simulations after comparing several results to the final measurements.