[31 Aug 2024]
My notes on this Mega Thread on Speaker Directivity
Jay Mitchell is an experienced speaker designer and guitarist who came up with the novel and apparently effective solution to make guitar speakers less pathologically directional. He posted this online in 2008 and I just came across it. Dumping some quick notes here for later reference.
Recipe:
Get open-cell polyurethane foam, either 1/2" or 3/4" thick, large enough to cover the front of the speaker. Cut an outer circle to fit the speaker baffle hole (the piece of wood that the speaker is mounted on) and a smaller hole in the center. For a 12" speaker, use a 3" hole. Mount this "donut" in the speaker baffle hole - you can use spray-adhesive to glue it to the grill cloth if the speaker is rear-mounted on the baffle. Obvs remove the grill or remove the speaker to get access to the hole.
Theory of operation: guitar speakers are so directional at higher frequencies (above 1KHz) because the sound is emitted by the whole speaker cone, and at wavelengths smaller than the speaker (1KHz wavelength is approximately 12"), the waves don't disperse very much. It's science! Hi-fi speakers usually have tweeters to make the higher-frequency sounds, and the tweeters are much smaller, so they disperse better. That's why this problem is particular to guitar speakers which don't normally use tweeters. The foam donut preferentially absorbs higher frequencies due to its thickness and density, and so at higher frequencies, the 3" hole constitutes a virtual emitter. Due to the smaller diameter, shorter wavelength sounds disperse better.
Jay's descriptions in that thread are more detailed, so check them out for background. There are lots of stupid/entertaining flamewars as well, people insisting it can't possibly work (without having tried it), people insisting it doesn't work but it turns out they used couch cushion foam instead of open-cell polyurethane foam, etc.
Mounting technique, from Jay: "Attach it to the rear side of your cab's grille using spray contact adhesive (e.g., 3M Super 77, available at Lowe's). Spray a light coating of adhesive on the foam only, and press it against the grille cloth within about 30 seconds of spraying it. You can easily remove the foam with no ill effect on the grille material, if you decide you don't like the effect."
You don't have to mount it exactly in this way. The important things are to get the center hole centered, get the foam close enough to the speaker, don't touch the speaker's moving parts, and prevent "flanking paths" for sound around the outside of the foam. So, for example, you can cut a big rectangle and tape it to the outside of the grill to try things out, or if there's the right amount of space between the grill and the baffle, put the foam in that gap, etc.
Foam material - use McMaster-Carr part 85735K72 or similar (there are many online vendors for open-cell polyurethane foam). People on the thread report different results for different foams.
This looks like appropriate foam The Foam Factory
If you lose too much high end, try thinner foam. Less dense foam also a possibility.
10" speakers: directivity occurs at a higher frequency, so try 1/2" foam and cut a 2" center hole.
15" speakers (like for bass): use 3/4" foam? Or even thicker? Not sure on center circle diameter, maybe start at 3" and enlarge based on results?
From the thread, Jay said: 'A 15" speaker will become "beamy" at a lower frequency than a 12. The 3/4" foam will still probably do a good job with one, but you want to keep the center hole about the same size. I haven't done any testing with 15s. I know the idea will still work, but I would expect the treated speaker to still be somewhat beamier than a 12, just as is the case with the untreated speakers.'
Cabs with multiple speakers: they will be directional due to the extra drivers, but applying the donut to each speaker should still help tame beaminess. Sinkhole's 2x2x12 Marshall cabs were venomous in this regard.
There was some discussion of how to close-mic a speaker that has a Mitchell Donut installed. There seemed to be consensus that micing in the center still worked about the same as with an untreated speaker, but micing at the edge was darker. There was a suggestion that micing at an angle near the edge of the 3" hole could give a result with moderate brightness. There was also a poster who said micing off-center as usual worked fine, even with the foam betwen the mic and the speaker.
Jay has a nice description of how sound emanates hemispherically from each point on the speaker cone as it vibrates, and the cone moves essentially like a rigid piston (not exactly, but close enough), and the directivity etc can be modeled as a summation of time-dependent interference/reinforcement of all these point sources. It puts me in mind of writing an audio ray-tracer to numerically compute this stuff; it actually sounds kind of easy/cheap, compared to graphics. You could produce an IR for different speaker+listener positions and analyze (and hear) the results without building anything physically.
Jay later designed the Atomic CLR "full range flat response" monitor, intended as an accurate PA-type speaker system for accurate reproduction, including guitar modelling amp outputs (with ~ideal dispersion).