Serving cervices, if you will.
mockup of our remodeled LUCIA
Let’s talk about our LUCIA redesign. Crowned “Miss charisma, uniqueness, nerve, and talent” and simplified for quick manufacture, assembly, and disassembly.
What we wanted to change from the original model:
Older LUCIA model for reference. This is not quite what we were improving upon, but close
- long assembly time — screwing in the wing nuts (6 of them!) to secure each panel took a bit of time. The L brackets are convenient (and an accessible item) for alignment, but seemed like an easy target for improvement
- awkward cervix holder — the long bolt was … questionable, and after asking members in the Richards-Kortum lab who work with the LUCIA, we confirmed our suspicions that the built in adjustability was not actually used in training applications.
- complex construction — the base was made of thick MDF and cut to sit at an angle, meaning at least three high(er)-tech tools were required to make the LUCIA kit (3D printer, laser cutter, miter saw). We thought we could simplify the design to require less manufacturing steps.
Our first thought was to remove the L brackets and instead add a finger joint for the panels to slide into place. This would address our first concern re: assembly time when building with finished pieces in the kit to a constructed LUCIA.
finger joint in our model
We received advice that these fingers might just snap off, so we kept them short and gambled that the width added from stacking frames would prevent this. In this first iteration we included L brackets, but just to hold the panels square and prevent wobble. We reasoned that this was an acceptable inclusion; since they weren’t used to attach the panels to the base we cut down on assembly time, and we needed bolts to fasten the panel components anyway (in theory, to eventually hold the fabric pieces in between). Since the holes would already be there, we could easily add a bracket to hold everything square.
We cut everything (a handful of times, since we couldn’t quite get the speed settings to fully cut through our panels), assembled, and …
it was wobbly!
Our diagnosis: the 3D printed L-brackets we used were too flimsy (there weren’t any metal brackets available). Additionally, unbeknownst to us, we hadn’t accommodated enough for the laser kerf, making the finger joint more of a suggestion than rigid joinery. To fix the floppiness, we decided the 3D printed corner joints should have upward supports that hold the first panel in place. (We were still workshopping what to do for the second panel).
for the life of us we could not get these cut settings to work (probably due to the board warping –> laser out of focus, despite using weights)
wobbly model : (
old vs revised corner joint (revised including upward supports, “this side up” arrow, + “R” side label)
While troubleshooting tolerances on these 3D printed joint holders, we also iterated designs for our cervix holder. We received confirmation that the LUCIA is typically assembled and used with cervices at one fixed distance, so we approached this problem with a fresh set of constraints. We took inspiration from shower caddies, deciding to hang the holder on the second panel. By doing this, we also eliminated the need for the third panel in the original model, which just held the long bolt in place.
post processing the final cervix holder design
After a few tweaks to get the cervix centered just right, we recognized that our plan to use Velcro to secure the cervices in the holder wasn’t needed. The LUCIA cervices are manufactured with a hole in the middle, so a peg would do just fine. We did a few more prints, adjusting infill density (increased to 65% for durability), fixing tolerances, and settled on our final design.
Heading back to the base and panels, we finally identified that the wobble was a kerf issue. We re-cut the base panel, engraving labels to simplify assembly: an arrow to indicate which side faces up, and numbers showing where panels 1 and 2 are placed (and luckily already had a “square peg round hole” situation: panel 1 is thicker, meaning it really can really only be assembled one way).
All of the clarifying labels were intended to be language-agnostic, but we slipped up when printing the corner joints — these are labelled L and R for left and right (exclusive to English), when we could have used arrows instead.
We designed and printed corner joints for panel 2 to (a) reduce wobble and (b) get rid of the L brackets completely (which were prone to snapping). Added to all of these 3D printed pieces was the arrow symbol for “this side up”. This meticulous labelling came after we attempted assembly one morning and, even as the designers, found it a little confusing.
Post processing! (wiping soot) not pictured, sanding each wood piece
More post processing! (IPA to remove glue on prints)
We re-cut the components for panel 1 (unfortunate bolt placement actually prevented the vaginal canal model from sliding into place, which needed to be fixed) and the base plate. We post processed — wiping off soot, using IPA to remove blue-tape adhesive (applied to prevent burn spread), sanding — and fastened what needed to be attached. It was time for the final assembly, save for one big problem.
The fingers for panels 1 and 2 did NOT fit into the holes we cut. Our kerf math was still wrong. Thankfully, having too much material is easy to solve. We sanded the fingers until we were satisfied with their press fit.
Now everything was sufficiently rigid, but when we did the wobble test the entire model still scooted across the table. This switched from a tolerance issue to a friction issue. We noticed, when wobbling, the base plate slipped where it contacted the table. We slapped on a rubber band for traction and BOOM! No wiggling. Our model finally came together as what we’d envisioned.
the humble rubber band
laser cut tl;dr:
- engrave: 10 power, 20 speed
- vector cut: 3 speed, 100 power, 100 frequency
- this lowkey fried the wood; with tape it was fine, but un-taped the wood will likely burn
- kerf math
- for context, instead of adding to the positive side (finger) and subtracting from the negative side (hole) to get the true measurement on both, we cut the fingers (forgot to adjust), then accounted for this when cutting holes
- ideal kerf / hole adjustment (with these speed/power/freq settings): sweet spot between 0.72 mm / 0.029″ (too little –> loose joints) and 1.1 mm / 0.043″ (too much –> tight joints)
Final model pictures:
front view (without canal insert)
side / iso view
top view, turned in on BIOE555 table
POV: you’re a physician in training
We turned it in on the class table, cleaned up, and celebrated with some carne asada at Valhalla 🕺🏽
proof of clean workspace
asada plate 😋 (shoutout latinx grads event)
Cost Analysis
the above image, but in table form (finally figured out how to get this to work):
total: $625.04 (not bad; as usual, labor is the largest share)
I had an absolute blast taking this class. It feels like this semester has been going on forever, but it’s sad to be on the other side of it. Thanks Dr. Wettergreen + the teaching team for an awesome experience in (the first!) BIOE555 : D
