We came up with the initial idea for this speaker as a response to analyze our sonic relationship to space. The process of creatively fabricating the idea is closely linked to ideas surrounding spacial philosophy and design philosophy. Initially we had even debated gathering impulse responses from a collaborative improvisational sound piece, however, the significance of mapping the acoustics within a space posed to have a greater impact to our perspectives on space, sound design, acoustic ecology, acoustic mapping, etc.
This is the final rendering of the 3d model which we will construct into a speaker that is 45 cm in height and cut into 25 layers of plywood, split into 3 pieces per circular section. Working with Matthew’s brother Chris exposed us to the processes that makeup 3d modeling, and we were collectively exposed to new issues in creative processes related to this type of engineering. We have learned how to expose ourselves to the options that make up processes 3d models live in, and for me at least it was very inspiring to realize the possibilities that exist and are possible in design, both in access to software and fabricating imagined and processed shapes and objects.
In order to represent the shape and the fabrication of the shape we attempted to construct a scale model of the final shape. This was in order to experiment with fabricating the full size model when we have access to the CNC machine. After organizing all of the layers onto an Adobe Illustrator file we setup a sheet of 3mm birch plywood on the machine (The final design of the scale was rendered into a 6:1 ratio to what would be the final model).
I then proceeded to label each individual piece (1-50 separated into two sides A and B) in order to track the construction of the shape.
The pieces were then removed from the laser cutter and organized onto a cutting sheet in order to begin assembly. Matt and I began gluing the pieces from the bottom up and quickly realized we made a mathematical and a design error in the design of the 2d layout. We had not accounted for the bevels we had designed for the cnc router to cut into a full 18mm thick sheet of plywood, and in that we had layed out 50 layers onto the 2d plane instead of the originally mapped 25 layers (which were designed and exported in a 3d context). We continued assembly as this was a test to work towards the final model and Matt noticed that our reference point on the shapes did not align to the ratio that was exported as well. We were successful in the assembly plan, however, we did not allign the two halfs because of the issue with the ratio and distortion of the reference point.
Final test shape and fabrication, the gaps were distorted as a result of the ratio issue we faced, however, they are designed to house the top and bottom speaker drivers.
We began the process of constructing a scale model by separating the top and bottom portions of the 3d model of the shape. Each layer is separated into two portions because the top and the bottom of each layer are space out in order to be beveled by the CNC router. In order to format the design for the laser cutout we needed to export organized 2d files of the layers separated for the final 3d shape into Adobe Illustrator.
In experimenting with fabricating the shape we initially struggled with configuring software in order to splice the final shape into multiple layers, and then for the scale model, 2d layers. This export identifies 50 total layers of the shape exported from the curve layout in a software program called Rhino. Constructing the shape in itself uses spatialization and impulse response proved to be difficult, and the organization of these layers was as well. The format was reconfigured to be placed on a 2d plane and we ended up using jpeg exports into adobe illustrator for the scale model. The sides were widened and the shapes needed to be grouped in order to be analzed by the laser cutter.
This is how we layed out the sections onto an Adobe Illustrator file in order to maximize space used on the 3mm birch plywood sheet.
In preparation for the construction of the final speaker units we have been researching various approaches to designing a crossover. A crossover is essentially a high/lowpass filter circuit that limits one speaeker driver to high frequencies and another to lower frequencies. They are categorized and arranged in various sequences and order as are frequently used filters. The progression is parallel to greater attenuation at the cutoff per octave. Each step in the order increases the attenuation by 6db. “A 1st order crossover filter 6 db/octave, a 2nd order 12 db/octave, a 3rd order 18 db/octave, and so on’.
Components involved include capacitors and inductors, which are similar but operate in different ways. A capacitor stores voltage in a circuit and limits the total output, while an inductor also limits the output but stores the current in a magnetic field. We have the means to wind our own copper coil inductors if we wish to save money.
Phase changes introduced by a crossover:
“Each order of crossover introduces a 90 degree phase shift. A 180 degree shift is an inverse of the wave. If 2 speakers are 180 degrees out of phase then they will cancel each other wherever they produce the same frequencies. Even with crossovers, both speakers will produce sound for several octaves beyond the crossover point. If this problem occurs, there will be a noticeable dip in the frequency response at the crossover point. To solve this problem, wire one, but not both, of the speakers backwards (+ to -). Usually, phase shift problems only occur with 2nd order (or 6th order) crossovers, but can also occur when using multiple 2-way crossovers in a 3-way (or more) speaker system. The only way to really find and fix a phase shift problem is trying all possibilities in reversing the speaker leads.”
Phase changes introduced by Inductors:
“When using more than one inductor in a crossover, the electro-magnetic fields of the inductors can interfere with each other causing an unpleasant result. That is why it is best to keep the inductors as far apart as possible. Also, keep the fields out of phase with each other by rotating the inductors 90 degrees. It is possible to have 3 inductors out of phase, as shown below.”
It is important to note that when building a speaker it is important to use components that have the least resistance possible so that resistence in the overall speaekr system doesn’t build up too much for the amplifier. Saia has been experimenting with how to arrange the amplifier for the passive speaker we are building with the other components and we have landed on using an external amplifier.
There are three different types of capacitors, Electrolytic, Mylar, and Polypropylene. They are categorized by performence, Electrolytic to Polypropylene. Metalized polypropylene being the best, but most expensive.
Back to the crossover, it is important to focus the crossover onto the middle of the overlapping frequency response of the two drivers used. This is in order to have the maximum amount of flat octave response on both speakers.
“mid/woofer crossover there are 4 octaves between 200-3.2k Hz, 200-400-800-1600-3200. 800 Hz is the middle frequency, with 2 octaves flat in either direction. For the tweeter/mid crossover, there are only 1 octaves, 2000-4000. 3k Hz is the crossover point with 1/2 octave stable in either direction. These two drivers have little overlap, and normally would not be used together.”
The final shape has been altered to house a 5.5cm tweeter and a 13cm driver. The shape has also been spliced into 25 separate layers, and each layer split into two pieces. We are currently attempted to graph all of the pieces on a 2d plane in order to cut and assemble a scale model of the cabinet.
As we’ve finalized the plans for the final cabinet, we have positioned sections designated for the tweeter and the subwoofer speaker drivers. The final model has been divided into 3 sections and a total of 25 layers. We need to adjust the size of the tweeter because we had to order new ones for the final speaker build, however, this will not affect the design principles we have established. The model will need to be laid out of a full sheet of plywood and cut in two parts, the top layer beveled and flipped so that the bottom layer can be beveled.
The process for using the CNC machine has panned out to be more strenuous than initially thought so we are planning on making a scale 1/3 model in anticipation of assembling the full-size cabinet. Currently, I am separating the final model into 50 seaparate pieces on a 2d plane that can then be lazer cut into a smaller model. The assembly is difficult as we need to label all pieces, however, this is in preparation for the full scale one. The model will not have bevels in it, however, the 2d layout requires percise organization, as it is harder to decipher the different pieces.
Construction for the final cabinet will begin once the CNC machine is ready to be blocked off for the day, because there will be 100 total cuts made into the sheet of plywood we will use.
For our submission we will have completed the 3d renderings of the final cabinet as well as a scale model and the electronics for the final speaker. We have started arranging the electrical components necessary for the speaeker drivers. Saia has begun calculations for the inductors we will need, and the crossover for the drivers as well.
In finalizing the final shapes for the speaker cabinet design we ran into issues with formatting and exporting the file types required for Matthew’s brother Chris to render the final layout in Solidworks. We exported the files from TouchDesigner as FBX files on a PC, however, the files were not usable in the software Chris was going to splice our design in.
The logistics were solved after a couple hours of attempting various formats, and it took installing another 3d rendering software to format the files correctly.
These are the final shapes exported from TouchDesigner – We used the top middle two in our final design of the speaker.
The final shapes we chose were stills from the sine wave sweep altering a torus shape, with a hollowed-out center. We chose two to make two separate cabinets in our realisation of this concept. The thickness of the model will equate to roughly 1″ to be able to support the driver that will be mounted to the exterior wall. There will be a flattened area digitally designed to hold the speaker driver, for aesthetic and practical purposes.
Here are the components purchased and organzied for the final speaker build. We have researched crossovers and inductors for what will be the final stereo pair.
Two drivers, two tweeters, two inductors, capacitors, and resistors.
In the initial process of developing a shape to use in our speaker design, we used TouchDesigner to see how different shapes react to the sine tone sweep we recorded at Westminster station. In TouchDesigner we have been experimenting with distorting cuboids, ellipsoids, and cylinders. The audio file that was imported into TouchDesigner was the 6 second sine tone sweep from the station. The audio was then visualized in a frequency spectrum and analyzed by the program taking the RMS of the sound and averaging out the maximum and minimum amplitudes. Then a math input is used to control the size-to-output ratio that will manipulate the shape. The path that follows is a shape into noise, which distorts the shape, and then a RMS controller to direct the amplitude of the noise from the audio file. This outputs a final shape, which is distorted in a 6-second loop following the frequency response that occurs from the audio file initially imported.
The issue we are currently facing is rendering a 3-dimensional still image that can be used to design the final cabinet. Right now the file can be exported as a video file, however, we need to decide on a still frame to position the speaker drivers. The final shape will then be divided into 25 separate layers and each layer will be divided into three pieces with a center cavity subtracted from the shape.
For this project we have collaborated with Christopher McConway in order to impliment functionality in both the fabrication and structure of our shape from the impluse response recorded in Westminster Station. In discussion and with his perspective in Mechanical Engineering he was able to divide the shape for us into 25 layers divided into two sections each. This is an excerpt from his notes on doing the porject with us and images demonstrating the procedure.
” Import base FBX file into Fusion 360, as an unstitched surface.”
“Use the “patch” Feature, within the “surface” tab of the fusion360 modelling environment, to mend/close the open ends of the “Unstitched Surface”. You can see, in the screenshot (Fig1) attached, that before using this the patch feature the body is essentially hollow. Using the patch feature is fundamentally adding another surface that will represent the top/bottom of the body. Visually this will make the body seem complete but it is still hollow (Fig2).”
“At this stage the model consists of THREE separate surfaces”
“To combine these separate surfaces use the “Stitch” feature (Fig3), this will collate the surfaces into one SURFACE BODY, model remains hollow.”
“The next objective is to “Create” a solid body from these three surfaces that will essentially fill the empty space within the boundaries of the surfaces. The tool used is call “Boundary Fill”, seen in Fig4.”
“The model can then be exported from Autodesk Fusion as a STEP file. For simplicity down the line, I choose to go with a step file as it can be imported straight into SolidWorks as a solid body, avoiding working between meshes and solids. In an early test export, using meshes and solids, only the solid features I had added to the model were preserved when exported as IGES, which wasn’t ideal.”
“Once the export is complete, import this file into SolidWorks. The benefit of the step file here being that I can start modifying the model immediately.”
“The first step in SolidWorks is to scale the model up, to meet the desired end dimensions. You can see in Fig5 that I scaled the body up by a factor of 9, increasing the height of the body from 50mm to 450mm.”
“I then chose a direction to which I would model the flat surfaces to house the electrical components. This happened to be the right plane in the case of *MODEL NAME*, as I felt it demonstrated the complex geometry of the body more favorably than the other planes. I then created another plane parallel to the front plane that was coincident with the boundary of the body (Fig6).”
“From here I made a series of extruded bosses and cuts to form the 170mm outer diameter flat surface for the speaker to mount to. The extruded cut at this stage was made as a circle into the original imported body. This gave me what I needed to now shell the body without sacrificing the top or bottom of the original body. You can see this in Fig7. Shell parameters were set to create an 18mm thick “wall”.”
– I then created the features that would house the tweeter, with the centers of the two holes aligned vertically and spaced 160mm.
– After adding the main geometry to hold the speakers, I made some adjustments to minimize the protrusions from the original surface. You can see the final geometry in Fig8.
– The body was then separated into 18mm thick layers to suit the selected material. I did this by adding split lines every 18mm, the finished product seen in Fig9.
– This model was then exported as an IGES file. I imported this IGES file back into fusion360 to check the geometry.
