Teaching Fellow: Dan Tish
SCI 6338 | Introduction to Computational Design | Final
Harvard Graduate School of Design | Fall 2019
This group’s work (including Sana Sharma and Katarina Richter-Lunn) was motived by the study of sound. Each member researched a different facet. In addition to working with sound, each group member found ways to manipulate the inputted sound. For this portion amplitude was used as a means to analysis un-paralleled layers in Fused Deposition Modeling. The original sound was printed by a LulzBot Mini 2 3D printer. The sounds it made while printing were then recorded. This information was processed into an additional 3D print to create the printer’s response to the inputted sound. 
Overview
This case study investigates amplitude as a means to analyze unparalleled layers in Fused Deposition Modeling. The original sound was printed by a LulzBot Mini 2 desktop 3D printer. The sounds the LulzBot Mini 2 made while printing were recorded. This information was processed into an additional 3D print to create the printer’s response to the inputted sound. The print reply created by the amplitude of the sounds of the printer was then compared to the original print of the song’s amplitude to ascertain if there was a correlation.  
Workflow
The workflow began in P5.JS (a JavaScript library for creative programming created by Lauren McCarthy) where the amplitude values of a MP3 (coding format for digital audio) file were recorded to a Tab Separated Value (TSV) file. These sampled amplitudes were used to alter the Z heights of a series of planar circles in Grasshopper3D (a visual programming language) for McNeel Rhinoceros (a 3D modeling software). These altered circles were divided into points to create toolpaths for the 3D printer.
INITIAL DESIGN TESTS
The first 3D prints were generated from Perlin and Sine graphs. These prints were not only modified to have un-paralleled layers but in addition to have a parabolic side profile. Difficulties with layer adhesion required that the curved profile be removed. The change limited the variables allowing for the focus to be on the extrusion settings as they related to the un-paralleled layers.
SINGLE EXTRUSION MULTIPLIER
In extrusion based 3D printing a strand of filament is moved using a driver attached to a stepper motor. The extrusion flow rate is by default one-hundred percent. The extrusion multiplier allows the stepper motor to increase its rotational speed allowing more material to be extruded through the heated nozzle melting the filament and depositing it on the build platform. 
The initial tests used a single extrusion multiplier ranging from 0.01 to 0.05. The initial prints used the 0.05 extrusion rate which resulted in clumping due to the excessive material deposited. This occurred as the compression on the layers increased due to the unparalleled nature of the design. The 0.01 extrusion rate did not deposit sufficient material for the print to adhere to the build plate. Ultimately, an extrusion rate of 0.03 was tested. Although this remedied some of the prior issues, it was evident that this rate would need to be modified based on the distance between the layers.
VARIABLE EXTRUSION MULTIPLIER
A function was developed in C# (a programming language) that allowed for the extrusion rate to vary depending on the distance of the next layer relative to each printed point. As the layers are more compressed the extrusion multiplier is lowered. The inverse is true for those layers that are less compressed. This is relative to the approximate height of each layer.
DESIGN ONE
The first test utilizing a song’s amplitude to generate the design the sample rate was increased from one hundred to twenty-five hundred values. At this scale and resolution, the fidelity of the printed part was not comparable to the vector drawing. The artist Lizzo’s music was chosen due to its great variability in volume which created more dynamic amplitude data. Thus, resulting in a more complex design. While the amplitude of the song “Juice” was printed, the sound of the 3D printer was recorded. As a result of the less extreme values, the fidelity of the print was improved.​​​​​​​
DESIGN TWO
To showcase a musical piece that could be easily identified by a larger audience, The Beatles song, “Let it Be,” was chosen. In addition to its ubiquity, it has a slow tempo with minimal amplitude variation. The printer’s accuracy and the small size of the print made denoting a song's entire amplitude as a single top layer not achievable with a great level of accuracy. As a result, the data was divided into five groups with approximately three-hundred twenty amplitude sample points. This divided the amplitudes into thirty-second intervals which continued throughout the final two iterations.
The LulzBot Mini 2 demonstrated a rough correlation between the printer’s sound and the lack of variability in the amplitude. The printer’s sound changes as a result of the required use of the stepper motors.​​​​​​​
VIDEO OF THE LULZBOT MINI 2 PRINTING LET IT BE BY THE BEATLES
SMELLS LIKE TEEN SPIRIT- NIRVANA
The song “Smells Like Teen Spirit” by Nirvana was added to diversify the range of musical styles. This song has a mid-range of amplitude variability starting off slow and ramping up at the end of the song.
The LulzBot Mini 2 demonstrates a correlation between the printer’s noise and the change of the amplitude.
VIDEO OF THE LULZBOT MINI 2 PRINTING SMELLS LIKE TEEN SPIRIT BY NIRVANA
ROLLING IN THE DEEP- ADELE
Adele’s “Rolling in the Deep” was chosen for its consistent variability as it progresses through the song. Her ability to crescendo and decrescendo regularly creates a dynamic pattern for the toolpath.
The variability from the 3D print was reflected in the LulzBot Mini 2’s interpretation.
VIDEO OF THE LULZBOT MINI 2 PRINTING ROLLING IN THE DEEP BY ADELE
CONCLUSIONS
This case study used a variety of songs to create unique 3D printed objects. Variable Extrusion rates were used to improve control of print fidelity. As the amplitude variability increased the resulting prints became more formally dynamic. The prints from the response of the LulzBot Mini 2 sounds correlated in their complexity to that of the original songs. The C# function could have potential benefits for unparalleled layer FDM 3D printing. The function allows variable extrusion rates resulting in cleaner surface qualities when 3D printing with variable layer heights. In future iterations, other data besides amplitude may offer a method to inform the curvature of the side profile of the vessel giving the object more aesthetic complexity. In addition this could allow for a more complex representation of the sound. 
Each 3D printed object was as unique as the song from which it was derived. The sound created by the 3D printer mirrored the amplitudes of the original song showing a correlation between the two. Other forms of sound such as nature or cityscape could offer alternative methods of understanding sound as it is represented through fabrication. The fabricated object would have more variability due to the spontaneity of the source. 
Back to Top