Danny Griffin HTMAA 2023

Week 01 :: Computer-Controlled Cutting

How do we use machines precisely?

1. Vinyl Cutting

This week was a warm-up in machine callibration. I used the vinyl cutter for the first time, and learned to be more precise with the lasercutter.

Responsive image — Two sketchbook labels prepared by vinyl cutting.

I'm using two sketchbooks this semester, black for theory seminars, yellow for labwork. Thankfully, RPL is stocked with both black and yellow adhesive vinyl rolls.

The biggest learning curve I found was in understanding the how the .png would convert to a toolpath.Once I sent a file with the same text with different strokes and fills from illustrator, I realized that toolpath generation relies on pixel edge detection, meaning vector paths are not registered.

I also realized after my first test cuts that delicate typeface doesn't survive the peeling process - so I swapped to use the negative space of my sticker design instead.

2. Group Work: Laser Characterization

With Mateo Fernandez

Test square.

Focusing the laser, 2" clear of the top of sheet material.

The digital calipers in the shop were limited to two decimal places of measurement. For my own project, I was working with 0.25" interlocking joints, so I found it more useful to line up the test square on a flat metal ruler and estimate by visual inspection.

In my case, I found a 1/96" offset of the vector geometry would result in a square that was nearly 1/32" larger than 0.25", enough so that when pressed together, the cardboard would slightly compress and friction fit snugly together.

Modular Construction Kit-of-Parts

I'm fascinated by the discrepancy of ideal digital geometry versus constructed reality. Therefore, I'd like to use this exercise to test a workflow for constructing a precise surface using lasercut material.

I decided to start by plotting the function of Dini's Surface, a geometry whose equation can be found on Wolfram Alpha Version 1 of this process uses the following Grasshopper script and Rhino file.

I built most of the script framework without knowing the actual material I would use. When I got to the shop, I chose my stock material, measured, and updated my values accordingly.

Elevation view, framing derived from mathematically plotted surface.

The framing members are modeled as solids with parametrically-built notches.

Section cut reveals how the profiles are built outward from the original surface. This view also reveals the pending collison, as the nested surfaces are too close together.

Kit of parts.

Kit of parts friction-fit into four segments of the surface.

Side profile of two segments friction-fit together into a spiral surface.

At this point, I realized that I made an error in my process. The four segments are identical, and the surface repeats in a rotationally symmetric pattern about the origin. In theory, I should be able to fit together each of the four segments in sequence to extend the spiral. My mistake is due to an oversight on my part. I initially planned to use thinner cardboard, but ended up using double-ply cardboard, with a thickness of 0.25". I updated this value in my script, but I did not simulate the assembly to check for collisions. In the end, the segments collide with each other at the tightest point of the spiral, making the full assembly impossible.

Eventually, I'd like to refine the workflow a bit more to automate this process, specifically building features to test constructability of complex geometries.