Professional Documents
Culture Documents
As shown in the image above, the polymeric disks that result from the process
can vary in thickness across the disk. This stems from a number of possible issues
(warping of the bottom plate, gunk buildup along the bottom surface, etc). At the time, I
skirted the issue by discarding faulty molds. For this week’s assignment, I wanted to see
if I could leverage elastic averaging to improve disk thickness consistency without
adversely affecting my usual workflow. An elastically averaged mold, combined with a
level, flat wedge of aluminum in place of the typical oven rack, could save me hours per
week.
Functional Requirements:
I started by first qualitatively identifying my overall goals:
• The resulting coupling needed to serve as a repeated-use PDMS mold – the
material had to be sufficiently heat resistant, non-reactive to PDMS, and allow for
easy disk removal.
• The process of making these molds had to be on the same order of convenience
of manufacturability as my current mold making procedure. I would be making
dozens of these molds, and would need to make more over time. Spending hours
fabricating the One Perfect Mold, or designing around using a Sink EDM (or
some other tool I did not have consistent access to), would not help.
• The coupling had to have sufficient stiffness in the linear z-direction and angular
X and Y directions. It needed to be able to handle the weight of one polymeric
disk (30g), and to allow for uneven distribution of liquid polymer weight during the
pouring process.
• I identified a maximum variation in thickness of 2.5% across a produced
polymeric disk as my desired goal.
This rough list of requirements was developed and given quantitative values,
resulting in the FRDPARRC seen in the associated spreadsheet (“2 - FRDPARRC,
Analysis, and Data.xls”). Most notably, I determined an angular repeatability of within
0.12° as my most critical requirement (analysis documented in spreadsheet).
Initial design:
There is little variance across both stiffness and angular repeatability when
switching between surfaces, indicating that friction might not be a significant factor in
this system. Additionally, the experimentally found and predicted stiffness agree closely
(within 5%), suggesting that my beam bending analysis was sufficient. On the other
hand, there is a wider gap between the experimentally found and predicted values for
angular repeatability – the setup was found to be twice as repeatable as predicted. My
best guess as to the reason behind this disparity would be a mistake in my error
analysis.