Exponential Horns are amazing acoustical machines with no moving parts. They also look gorgeous. Here I'll show you how I made several thin-walled aluminum and steel exponential horns. They work great as loudspeakers, and (my original intent) as smoke-ring cannons aka vortex-ring cannons. Whether you've got a 1920's Victrola, or a 2014 High fidelity sound system, exponential horns like this are unbeaten as the most efficient way to spread sound from a limited source.

Acoustical Horns are Like Gears on a Bicycle

Horns are like gears for sound pressure waves, coupling forceful motions of a small quantity of air into small motions of a large quantity of air, or vis versa. They can make a loudspeaker broadcast sound more efficiently, and more directionally, and they can also make a receiving sensor like your ear or a microphone more sensitive and directional. I experimented with using an exponential horn to couple the motions of air from an 18" subwoofer to an aperture, from which swirling toroidal (or 'poloidal' if you're a nit picker) vortexes could propagate long distances as an action-at-a-distance effector for various art and educational projects I'm working on. Nothing like blasting someone's hair back at 20 feet!! You'll also see some photos of that project, including a stainless steel iris, and heavy duty servo-motor controlled pan-tilt gimbal capable of aiming all the above.

You can read more about acoustical horns here: http://en.wikipedia.org/wiki/Horn_(acoustic)

What you'll need

  • Sheetmetal to make your horn out of. I've used 0.10" thick aluminum for the 4' long horn shown, and 0.065" (16 gauge) mild steel for the smaller horns. (Steel is definitely easier to TIG weld, if you're starting out.)
  • TIG welder and accessories (filler metal rod, tungstens, etc).
  • MATLAB software (only if you want to define arbitrary horns with the code I've written, instead of using the files for the particular horns I've included here). You can probably make this code work in the open source free alternative programs to matlab, Scilab and Octave, too.
  • A CNC waterjet cutter (or a laser cutter powerful enough to cut your thickness of metal, >400W) reaaaaally helps to make the precisely shaped parts that fit together in precisely the right 3D curve. I'm lucky enough to have access to one in Techshop Pittsburgh. You could also make the necessary segments with a bandsaw and a file, if you are up for detail work, by printing the diagrams on paper, scribing over the printed lines with an X-acto knife, then cutting down to the line with a bandsaw followed by grinding. Or farm out the water jet cutting to a shop. But the panels need to fit together SUPER accurately, otherwise your welding job will be much harder.
  • A Slip Roller to smoothly bend the sheet metal after cutting it to shape.

Step 1: Size the Horn and Generate (or Borrow) the CAD File

Generate_Exponential_Horn.m is a matlab script I wrote to generate the flattened patterns of the facets of a faceted exponential horn.

The following variables should be modified:
numfacets %default 5, the number of facets of the horn. the number of sides of the polygonal cross section
dia1 %default 43 inches, the outer diameter of the horn at the widest end

dia0 %default 3 inches

If you don't want to run or modify the code to generate your own .DXF files, you can use the ones I've attached here. (Also Vcarve .CRV and flow jet .ORD files). These files should be able to be scaled in either axis and still join together nicely, too. If it's not overwhelming, you could also message me and I might be able to run off a series of files for you.

NOTE: some post-processing of the .DXF file was helpful to make sure that all the lines are in a closed, connected, curve. If I make another horn in the near future I'll update the details here as I refresh my own recollection.

An additional method of specifying your geometry would be to use Pepakura with a CAD model of the horn. But in my innocence of CAD at the point I attempted this the first time, I found it easier to write my own geometry code in Matlab than to figure out how to get Autodesk Inventor to specify a faceted exponential horn and export the flattened facets as 2D cut patterns.

Instructables / Autodesk folks, I asked the local pittsburgh autodesk employees about generating a faceted exponential horn in Inventor then exporting the flattened sheet metal facets (as well as every Techshop Pittsburgh DC and knowledgeable patron and Inventor Instructor here) and still drew a complete blank. I'd be happy to write an instructable about that aspect if you could get me started. Message me?

<p>Great stuff, very high quality work.</p> <br><p>Is there any specific advantage to using these high tech, expensive technologies over using something like doorskin, tape and glue construction, and epoxy coatings? A lot of classic speakers were made of paper.</p> <br> <br><p>Thanks again, beautiful work.</p>
<p>Great Question, David. <br><br>The simple answer is I used what I had. With the tools that are available to me (at techshop), I could design, cut, bend, and weld a horn like the smaller steel ones shown here in an afternoon, nowadays as little as 4 hours, since I've done it a few times. The water jet time for the smaller steel horns was less than 10 minutes for each horn's parts ($30) and the steel maybe another $10, so this approach is actually not very expensive. I'm not familiar with the paper horn building process (a description would be much appreciated!) but I'd wager that the time to make one out of paper, with multiple epoxy curings and laminations, might be just as time consuming, if not moreso, than doing it this way out of metal. There's also the durability of the result, and the simplificity of not requiring any formwork -- I understand that paper-mache horn building is often made over a meshwork which is first shaped by cutting plywood to the correct contours). <br><br>There's a rationale I often give credence to, that if you're going to spend amount of your time on something, it doesn't make sense to skimp on the materials. So $40 in materials / machine costs was acceptable to me, especially as a learning project.<br><br>Then there's the added benefit that I really like TIG welding, and making the horns this way afforded me an opportunity to practice that. Most people at TSpgh know that if they need help, I'll usually enjoy any excuse to get under the helmet. This project was no different. I strive towards ultimately building thin walled tubular steel aircraft frames and bicycles, so practice on thin walled materials is something I take any chance to do.<br><br>Then, there are the additional benefits of steel construction: high temperature resistance. I'd been talking with Jim Newton about the possibility of launching vortex rings of flammable gasses, which could be ignited, forming legit, aimable, fireballs. As I made these, I had that potential adaptation in mind, with a smile.</p><p>While enumerating the benefits of the horn, It seems likely that in some esoteric use, the electrical conductivity of the horn could be helpful. I'm not sure I'm going to try to articulate that one here, though:-)</p>
<p>AWESOME! Your welds are gorgeous, too. Is that stainless??! I like the speaker array and mounting fixtures. Did you enjoy using it as a megaphone before mounting the speakers? I recently downsized and gave the giant 4' long aluminum version of this I made to the FIRST Robotics team I coached, at the Sarah Heinz House in Pittsburgh PA. They use it now to learn about impedance matching, and of course magnify their voices...</p>
yes; 15/10 stainless steel.<br>After acoustic mesurement , It's very linear, with visaton sc8n speaker (http://www.visaton.com/en/industrie/breitband/sc8n_8.html )<br>usable from 150hz to 5000hz :)
<p>I'd love to see the plots if you're inclined to share them!</p>
<p>I presume you have spoken/sung through the horn; I am always impressed with how the improved coupling of vocal chords to the air results in a tangible sensation of singing in a more viscous medium; Do you feel your vocal chords differently while singing/speaking through it?</p>
Thanks for the great 'ible! I've been trying to sort out the math for a faceted exponential horn myself, and finding this instructable was a godsend!<br><br>I just had one question; In order to convert the function of my facet from terms of x into terms of the exponential curve, (in other words, 'flatten' it), I am having trouble rewriting the arc length formula ( integrated from 0 to x) to solve for x, so that I can input a known length along a curve from x=0 and receive the x coordinate of the point that given distance along the curve.<br><br>It seems you've it figured out with the MATLAB program, so do you know the rewritten arc length formula so that x is isolated?<br><br>Thanks :D<br><br>Dan
<p>Hi Dan, flattening the sheetmetal is exactly the challenge in doing this. I struggled for a while to get Autodesk Inventor to do it for me, but eventually (after a couple of hours, and also asking the staff at Techshop, and even some Autodesk Engieers), I realized that I could write the code myself faster than I could find whatever buried option in the CAD software, if it even existed. It took me only a couple hours to write the code. <br><br>The approach is to define the curve of the panel in 3D, then flatten it by re-parameterizing the panel. You start with panel width as a function of height (or position along horn's axis) and end with panel width as a function of the length of the curved centerline of each panel. I included my Matlab code which you can follow along if you're curious!</p>
<p>Gordon,</p><p>Thanks for getting back to me! I eventually gave up on the tricky calculus involved in &quot;flattening&quot; the horn facets... It seemed to be more trouble than it was worth. If Autodesk engineers couldn't figure it out, then I certainly wouldn't be able to!</p><p>I used your excellent MATLAB program to create the facet model for my horn, and I recently cut out the 6 facets of my horn using printed out PDF's of the MATLAB figure and my scroll saw. I am using thin, transparent PETG for my segments instead of metal like you used... Now I just have to figure out how to bond the facets together!</p><p>Thanks again for the excellent instructable and useful MATLAB code.</p>
<p>I am going to attempt to make this cool contraption. I was wondering what was the size of the sheet metal you used and how many I should buy.</p>
<p>I made this of 16 gauge steel, which is quite coincidentally about 1/16th inch, or about 0.06&quot; thick. You could make it out of anything of course, but your ability to bend it into shape will be significantly more difficult with thicker materials. </p>
<p>Darn...where's the video? Cool project though.</p>
<p>added video!</p>
<p>Do you have any video of this piece?</p><p>It's super rad!</p>
<p>I posted some video above:-)</p>
<p>we demand a video !</p><p>also, great work!</p>
<p>your wish is my command! I posted it above. </p>
<p>oh darn! its a private video :[</p>
<p>It is public now.</p>
<p>Whoa. The green &quot;jelly fish&quot; shots are fantastic. thanks for sharing. </p>
<p>What a great project...I want to scatter my Moms ashes with music over the ocean.</p>
<p>Thanks. That is perhaps the most noble application I've heard conceived.</p>
<p>Guess you longed after the pleasure of making smoke rings with your mouth after you quit smoking ! &hellip; ;))</p><p>Nice post. Thanks !</p>
<p>&quot;Sound&quot; advice.</p>
So cool
<p>It looks so awesome! I love the shooting of smoke rings, looks alien in the best way possible. Thanks for sharing!</p>
<p>My pleasure! </p>

About This Instructable




Bio: I'm a guy with diverse interests in San Francisco, CA. I enjoy solving problems, inventing and making things, exploring the unknown, making music, and ... More »
More by GordonKirkwood:CNC it however it lays: adjusting CAM toolpaths for skewed or hard-to-align stock.  The CNC Bubble Iris:  A Computer Controlled Giant Bubble Machine Laser Magnetic Field Mapping 
Add instructable to: