This project consisted of three different parts (or milestones), the first being a comparison of several ideas / designs which would be researched and gauged in their competitiveness to conventional manufacturing methods. The next part of this project would be to improve on the initial design and filter the possible designs down to 2. The final part of this project was to choose a final design and produce physical prototypes.

 For this project we were limited to polymer processes such as Stereolithography (SLA) and Fuse Deposition Modeling (FDM). This allowed us to work with the machines at TechSpark which included the Stratasys F170 (FDM), Dremel 3D40-FLX (FDM), and Form 2 (SLA); Of which we primarily used the Form 2 for its ability to capture fine details of up to 25 microns.

Print 1

 The resulting print from this quick sketch produced a tiny keycap due to the eyeballed measurements of sample images that were found online. The print itself took around 30 minutes in the Form 2 SLA printer with the first version failing due to the lack of adequate thickness. The second version was prepared with greater thickness and resulted in a print the better resembled a keycap.

Print 2

 This keycap was designed with dimensions obtained from an existing keycap with the MX switch specification. We chose to deboss a turkey on the top of the keycap to gauge user engagement around holiday themed keycaps. As part of the course's design project, Thingiverse and Pinshape pages were created for this Turkey Keycap in order to gauge market demand for this product.

 The print was prepared and sliced with the Preform software from Formlabs with the support autogenerated for Gray resin and a layer thickness of 0.025mm. The finished print was then washed and soaked in isopropyl alcohol for 1 and 10 minutes respectively. Afterwards the supports were removed and the print was UV post cured for 15 minutes at 60 °C.

 The resulting print captured the fine texturing along the sides of the keycap and the debossing of the turkey imagery on the top of the keycap. However, there were many visible imperfection generated from the support touchpoints and the hasty removal of the support. The greatest issue was that the keycap did not fit with the MX switch as it was slightly smaller than the connector. We believed this to be mainly an issue that arose during the UV post curing as the dimensions of the keycap were accurate.

Print 3

 For this print we wanted to test how the Dremel 3D40-FLX FDM printer would fair with capturing the fine texturing and connector details. The Turkey Keycap was printed with white PLA filament with support placement around the connector as outlined in the slicing software.

 This print unfortunately was not able to capture the texturing detail of the keycap due to its rather large layer height of 0.1 mm. The placement and rigidity of the supports made it rather difficult to remove from the keycap and prevent us from testing whether or not the keycap would fit with the MX switch. The lackluster results of this print provided reason to forgo the Dremel 3D40-FLX in favor of other printers capable of reaching higher resolutions.

Prints 4 & 6

 Prints 4 and 6 were done with the Creality Ender 5 Plus FDM printer which is also capable of a print accuracy of 0.1 mm. However due to its less frequent use in a research lab setting, the prints from this machine captured more of the fine details. The following prints were printed without supports on the glass bed and on a raft as shown in the Creality slicer screenshots.

 The prints from the Creality Ender 5 Plus did a far better job of capturing the fine details of the Turkey Keycap. The raft resulted in a worse surface than the print which was directly printed on the glass.

 The resulting print was the first keycap to fit with the MX switch of the mechanical keyboard further confirming that the dimensions are correct and the possible shrinkage occured during the UV post curing process.

Print 5

 This print builds off the issues discovered in the previous SLA attempt (Print 2) and explores if a couple parameter changes would resolve issues. For this minimal supports were used to secure the keycap and the resin color was switched to clear. In addition the UV post curing step was skipped to confirm the hypothesis that the keycap would fit without this step.

 The resulting Turkey Keycap fit well onto the keyboard and supports the claim that warping occurs during the UV post curing stage. Another theory we had was that the gray resin has different density causing it to shrink more than that of the clear resin. This theory, however, was not tested as all the subsequent print were done with the clear resin.

 This model was sliced in the Preform software and the supports were autogenerated with the keycap tilted at a slight angle. This slight tilt was intended to remove any of the excess resin residue that would collect when printing completely vertically.

 This print however failed and this was most likely due to the fact that the resin tray for this specific printer was faulty. Another possible factor was that the supports were inadequate, however, this would seem to be less of a factor.

 For the first of these prints (Print 8), one instance of the textured Christmas Tree Caps Lock Keycap and the textured Shift Keycap were printed. The supports for each of these keycaps were autogenerated and oriented vertically to maximize the chance of a successful print. The next print (Print 9) builds on this and doubles the quantity of each keycap since the previous print worked well.

 The resulting prints were successful and captured the fine details and texturing of both keycaps. Also since most of the supports were attached to the flat surface of the print, the textured areas did not experience any defects from the support removal process. Since both of these prints did not experience any UV post curing, there was no warping of the part and they both fit well on the keyboard.

Design Project: Part 3

 For this last portion of the design project, the project guidelines had us focus on one of the two proposed products outlined in our previous design project proposal. In this final report we decided to move forward with the custom keycaps and fine tune the manufacturing process associated with these prints. Specifically we focused on the perfecting the UV post curing process and optimizing the orientations within the Preform slicer to reduce the resin that would accumulate on the top of each keycap.

 The keycaps in this print were angled to a maximum of 45° in each X and Y direction in order to gauge to what degree of rotation was necessary to reduce the accumulated resin. This was done with the bare keycap model to increase build rate and more clearly see the defects on what should be a smooth surface. Each keycap was labeled with its corresponding rotation on the raft and organized as seen below.

Print 11

 This next print investigates this same issue however with a subset of the previous rotations. In this print the Y axis was held constant at 0° and the X axis was rotated from 0° to -45° in 15° increments. This was done with the Bare Keycap and the Shift Keycap which kept its texturing.

 This print was a success and provided valuable information regarding the ideal alignment of the keycaps. Like before the Shift Keycaps skipped the UV post curing process so that printing orientation could remain a primary focus here. The Bare Keycaps were cured just to see if there were any significant changes to the keycap after the UV post curing process.

 One notable difference that was seen with the differences in orientation along the X axis as that the MX connecter started to lose accuracy when the keycap was rotated. This -45° orientation rotated the connector to print as a "X" rather than a "+" and lost much of the fine details around the points of connection. This made us realize that only a slight angle was needed along the Y axis to remove the excess resin and printing with 0° rotation along the X axis was the best solution.

Print 12

 After investigating the issue of resin build up on the tops of the keycaps, the next issue in question was the UV post curing and the appropriate scale needed to account for the shrinkage. This was done by testing out scales settings of 1.00, 1.010, 1.020, and 1.025 in the Preform slicer with 2 Shift Keycaps printed at each scale. This allowed us to have a set of UV post cured and as is set of keycaps at each scale for a corresponding control and experimental sample.

 Once printed, one set of keycaps were UV post cured with the supports still attached as it was noticed that the inclusion of the supports prevented warping during this UV post curing process. However, there still remained the issue of significant "yellowing" of the clear print during the UV post curing process that a adequate solution for had yet to be determined.

 The UV post curing process was done at 60 °C for 15 minutes with the supports of the prints still attached. It was noticed that for prints with higher length to width ratios, the bending and warping of the keycap during the UV post curing process was more pronounced. Keeping the supports on seems to have mitigated this issue since the supports also act as a mechanism to retain the build up of residual stress.

 With the as is Shift Keycaps it was found that the ductility of the MX connector allowed for a range of scales to fit. However, with the UV post curing the connector becomes more brittle and the range of fitting keycaps narrows to a scale of 1.020.

Print 13

 With the determined scale adjustment of 1.020 from the previous print producing adequate fittings, these settings were applied to the Turkey Keycaps in a 6 x 5 grid for a total of 30 keycaps per batch. The keycaps were placed a slight angle of 15° such that resin would not collect on the top surface.

 The resulting print was successful and each keycap captured the fine details and texturing. These prints were then washed and exposed to the UV post curing process however did not fit on the keyboard. This was an unexpected result as the same scale of 1.020 from the previous print was adequate enough to secure the Shift Keycap in place. One glaring mistake in the process was assuming that the two different keycaps would exhibit the same shrinkage from the UV post curing process. This issue would be further investigated in the next prints where different scales for each type of keycap would be tested.

Prints 14 & 15

 For the next couple of prints, a range of scales would be tested to determine the appropriate fit for each keycap. This was done within two separate prints where the first (Print 14) would test a wide range from 1.000 to 1.025 and the latter (Print 15) would test finer increments of 0.005 for each. The keycaps of both prints were arranged to maximize the density of the total keycaps that could be printed in each batch.

 After UV post curing, the results for both prints showed similar results were Shift and Caps Lock Keycaps withiin the range of 1.015 and 1.025 fit well on the MX connector. However the keycaps for the original Turkey Keycap would still not fit on the keyboard leading us to believe that another factor was contributing to possible uneven warping of the keycap.

Print 16

 The first of these prints uses the Pacman Ghost Keycap and prints it at the same angle as the keycaps from the previous prints to reduce resin build up. This was done in a batch of 20 with 4 x 5 layout of the keycaps.

 The print was rather successful, however the yield of good keycaps from this print was around 60% with several damaged and incomplete keycaps.

 When attached to the keyboard the clear resin allows for the light from the LEDs to scatter throughout the print. This creates a beautiful array of keycaps with the range of colors these keyboards are able to display.

Print 17

 For this print we decided to align the keycap with the switch placed parallel to the build direction. This was intended to see if the consitent printing angle of the "+" MX switch would provide consistent shrinkage during the UV post curing process. In addition to the Pacman Ghost Keycap, a print for an Anatomical Heart Keycap was tested as well. This layout saved space around the printing bed, however required more supports for each keycap with a total of 23 keycaps in this batch; 15 Pacman Ghost and 8 Anatomical Heart Keycaps.

 The print for this was successful and each keycap from this batch was usable. The orientation of the keycap did follow the expected behavior with the UV post curing treatment uniformly shrinking around the MX switch. Our analysis of this outcome makes us believe that the UV post curing around the other previous Turkey Keycaps was angled preventing it from fitting properly on the keyboard.

Print 18

 These last keycaps add figurines for a sleeping cat and sitting dog on the Turkey Keycap. These keycaps were organized similar to the previous iterations with the a total of 20 keycaps organized on the printing bed with a couple duplicates for each.

 Each of the keycaps from the print were successfully removed from the print bed and washed with an alcohol bath. These prints were then arranged by type with 6 Dog Keycaps, 6 Cat Keycaps, 5 Ghost Keycaps and 5 Heart Keycaps.

 The removal of the supports were done after the UV post curing process and prevent serious deformation of the print in any weak areas. Supports do leave a significant amount of imperfections around the touchpoints and further work would be done to minimize the total amount of supports and touch point size. Regardless, the resulting print displayed fine detail and look even nicer under specific lighting conditions.

 This design is primarily based off of Kevin Kennedy's tutorial for creating a simple hinged box with Fusion 360 with slight modifications to house the custom keycaps.

 The model was sliced and grouped into batches of 3 to maximize use of the printing bed and took just over 24 hours to completely print. The box provided adequate housing for the custom keycaps and worked great as Christmas gifts.

Future Work

 The process for creating custom keycaps has provided many insights into the complexity involved with fabricating the designed product. There are many nuances into refining the processing parameters of SLA printing so that the surfaces and structure of the print are satisfactory.

 One of the areas that could use more investigation is the optimization of the supports so that the print is supported while improving the surface finish of the print. Another would be exploring additional surface treatments such as acetone vapor polishing which would smooth appropriate keycaps such as the Ghost Keycap.