Computer Controlled Cutting

03.1.0

Introduction

This week was all about using computer software to drive stepper motors in 2 axes to cut, burn or etch material to create designs or objects that can be constructed into 3D forms. It was also an amazing opportunity to practice parametric design, a natural next step from last week's learning on CAD. You can read about our group assignment at Fab Lab ESAN here.

[x] Demonstrate and describe parametric 2D modelling processes.
[x] Identify and explain processes involved in using the laser cutter.
[x] Develop, evaluate and construct a parametric construction kit.
[x] Identify and explain processes involved in using the vinyl cutter.

[ ] For my students: create visual aid for laser engraving and cutting settings
[ ] For my students: create visual aid to describe the different kinds of joinery
[ ] For my students: create a vinyl cutter "template" file for practicing (designing, weeding, etc.)

03.2.0

Machines

I had a lot of fun this week, finally getting my hands on some of the machines in the Fab Lab here at ESAN. This week, I primarily worked with three machines, specifications below. Jorge provided me with specifications for the older of the two machines, and with some tinkering and research, I was able to determine the rest.

1

2

3

SLIDESHOW ⇲

AVR-1400 - This chunky beast was actually a lot of fun to use - it just had a legacy toolchain that honestly was pretty simple to understand once you learned it. It has a "bed size" of 600mm by 400mm and boasts a 100W laser though you generally won't be pushing it to the maximum for cutting and engraving. I found the best cutting speeds to be somewhere between 20-40mm/s and for engraving you can push it up to 100-500mm/s with quicker speeds resulting in lighter etching. The kerf was mostly consistent - though using too much MDF can cloud the laser and it gets a bit out of focus. I found the kerf to be around 0.17mm +/- 0.02mm.

Bambu Labs H2D - Jorge's new toy in the FabLab - this thing does *too* much. In addition to being a full fledged 3D printer with multifilament support - it also boasts a slightly weaker 40W laser - but do not be fooled - this thing can cut! The bed size is slightly smaller at 310mm by 250mm but it can move a heck of a lot quicker to get the same cutting (50-100mm/s) and engraving (800-1000mm/s) jobs done. The kerf on this machine is noticeably smaller likely due to the higher precision components and self calibration features. I found it to cut about 0.14mm +/- 0.02mm of the material. One notable difference here is that the H2D cannot work with transparent materials due to the type of blue light laser used.

Roland CAMM-1 Servo - Another legacy tool in the shop but I'm already in favour of it over the Cricut I have in my classroom because it isn't trying to sell me on their crappy subscription services. This device moves a lot slower than the Cricuit (and the other cutting machines) and doesn't have the best precision but it has the advantage of having the largest material bay at 584mm wide and theoretically infinite length if you can spool it enough material.

03.3.0

Parametric Design

You can read all about my intro to real CAD software in last week's documentation but to summarize, Fab Academy is providing me an opportunity to learn some really powerful CADing techniques, including parametric design, which was a focus of this week's project.

1

2

3

SLIDESHOW ⇲

We're back in Fusion! Step one is to define your sketch's parameters - but this time we're doing it with variables (notice the little fx next to each number?)

We'll do the same with each feature as well - just type the name of the parameter and fusion grabs it from the parameters list

My first attempt was a good start but I think I could have defined more of it parametrically - this one doesn't hold up to the modularization test

To start simple, I wanted to see if I could create a simple laser cut, press-fit box. I began working in Fusion, and with practice and trial/error, I learned and practiced a few key shortcuts when designing things that are meant to be scaled or dynamically resized/modified without needing to rebuild the whole model.

TIP No. 1TOOLTIP

Define most (if not all) of your dimensions as parameters (i.e. like variables in a math function), that way you can make quick adjustments to a completed model without needing to recreate the whole thing.

➔ Parameters

For my snap fit box, these parameters included:
ModuleL: The length of the wall for the box
ModuleW: The width of the wall for the box (defined as length/2)
ModuleT: The thickness of the material I would be using (also the length of the finger joints)
FingerW: The width of each of the joints
Pitch: The distance between each finger
and most importantly Kerf: the diameter of the hole left by the laser of the particular machine used.

TIP No. 2TOOLTIP

When you've got a feature that will repeat across your design, opt to use a pattern (rectangular/circular) to generate it, instead of manually copying/pasting it.

This was actually super fun to apply, because I could use parameters to define the number of teeth to "generate" on each edge based on how long the width parameter was.

1

2

3

SLIDESHOW ⇲

Using the pattern tool to define procedurally generated features

Extrude to the thickness of your material just so you can get a visual cue of what you're making - the laser cutter won't work in 3 dimensions

Here's another version with equal side lengths - here you can use the circular pattern tool the entire time

03.4.0

Laser Cutting

From here, I could export a 2D vector image of the design from Fusion and bring it over to the laser cutting station, which for my first few projects, I used ESAN's AVR-1400. Before I could actually submit that print though, I would need to test and measure the machine's kerf, so I modelled and designed a few little "kerf test tokens" that I could compare fits with.

1

2

SLIDESHOW ⇲

Cutting out (and engraving) the kerf test tokens

The material was about 3.00mm in thickness so the kerf test tokens are all under that to see what fit is best (since we want a friction hold)

After selecting a suitable kerf (for the AVR-1400, it was about 0.17mm), I edited the parameter in Fusion and sent my .DXF file to the computer controlling the AVR-1400. Jorge has a special modified version of Corel Draw that generates G-code, rather plainly, and can control the stepper motors of the laser cutter. For cutting MDF, which was the material I chose to start with, we want to use a 75W power setting and have the laser move at 20mm/s. And here's my result:

IMAGE CAPTION ⇲

Not bad for the very first thing I laser cut. The fit was a tad bit loose - so I adjusted the kerf to cut a little bit less from the fingers and more from the slots.

IMG_FILE: endresult.jpg

TYPE: JPG

XYZ: 123

03.5.0

Construction Kit

While CADing my first model, Jorge showed me these amazing samples of "living hinges" cut from MDF, that allow the material to bend, rotate and fold by carving out alternating chunks along the transverse axis. I was wowed by this and determined to try generating my own pattern in CAD for my parametric construction kit.

In my mind, I saw my kit having circular connecting joints, consistent widths, and differing lengths to play with. The intention was less practical and more artistic, to see what I could do with the material and where it would lead me.

1

2

3

SLIDESHOW ⇲

Again starting by defining the parts parametrically

Many parameters that reference other parameters!

The hinge parts are all parametrically defined as well

My process for the living hinges required more parameters to be defined. I wanted to be able to create construction "blocks" with differing lengths with and without the hinge applied, so I would need to parametrically define how to generate how many "hinge" units for a given length.

1

2

SLIDESHOW ⇲

Since the first print was done with cardboard - I had to adjust the kerf and material thickness - easy thing to do now that things are all parametric!

Successful test!

It was also here that I decided to add a fillet to the slots, as per the suggestion heard in the lecture this week. Since the slots could be generated on radial curves, I defined the chamfer as equilateral triangles that always had a fixed distance (the material thickness) apart from the centerpoint and slot edge.

1

2

SLIDESHOW ⇲

Using equilateral triangles to fillet the slots since the fillet tool won't work on the patterned part of this design (or maybe there is a way?)

You can see the little circles here (dog bones) - they help with flexing in the roll direction

I created a few prototypes of the hinge and slot system using cardboard first (I really don't like overly wasting materials) and with a bit of back and forth, eventually developed a system that I was happy with. I opted to add these little "dog bone" terminals on the ends of each hinge gap, since I found the stress points on earlier models were always the 90 degree edges within.

Generating all different sizes of the kit was super easy! All I had to do was open up the parameters window and change two numbers (the length of the part, and a boolean to indicate if hinges should be generated or not).

I created 9 sizes to play with, 3 with hinges (80mm, 160mm and 320mm) and 6 without (20mm, 40mm, 80mm and 160mm, plus a basic unit and a basic unit with a flat edge that could be used as a "stand").

I spent an hour just playing with the kit, pressing different combinations of each together, and eventually created a simple sculpture to showcase the curves and bends. I think it turned out great!

IMAGE CAPTION ⇲

What do you think?

IMG_FILE: sculpture.jpg

TYPE: JPG

XYZ: 123

Note: If you're noticing any "burn" marks, Jorge told me that the machine had not been cleaned a bit, so lots of sticky resin from the MDF was making it on to the underside of the print, and that it wasn't actually getting burned.

03.6.0

Engraving

For engraving, I thought of an idea to create a press fit "Minecraft" grass block, where each colour of the original texture would correspond to a different laser wattage and speed, creating a monochromatic (and depth) effect.

IMAGE CAPTION ⇲

Testing what I want the finished product to look like in 3D - again because the toolpath is only in 3D - this only really serves as a virtual test to see if everything fits nicely

IMG_FILE: Pasted image 20260206121013.jpg

TYPE: JPG

XYZ: 123

Developing the box itself was the easy part. I employed the same techniques described above to create the box. The tricky part was now developing a way to assign the colours of an image to different intensities and speeds on the laser cutter. This was extra difficult because the software used to drive this machine could really only handle one set of parameters at a time, and I would need to manually start each pass configured with the correct beam intensity and speed. I guess this was a good practice, because there is no "set it and forget it" when it comes to laser cutting. You need to babysit the machine just in case!

TIP No. 3TOOLTIP

Always be in arms reach of the laser printer, and wear eye protection if you are watching it directly!

I found 16px by 16px images of the top, side, and bottom textures for the grass block online, and then asked Gemini to help me write a python script to process the images into separate .SVGs that I could send to the printer.

GEMINI PROMPTLLM

Help me create a python script using Pillow to convert a series of 16px by 16px .PNG files into layered .SVGs based on their colour data, optimized for laser engraving. Resize each image to 100mm x 100mm using nearest neighbour resampling, iterate through each pixel and convert it to HSV. We will use 8 shades overall, the bottom 4 (lightest values) will be for all the "brown" pixels and the top 4 (darkest values) for all of the "green" pixels. Within each colour, determine the brightness/value of each and assign each pixel the closest of our 4 values. Draw each pixel as a rectangle filled with black and no stroke. Ensure all SVGs are 100mm by 100mm no matter the resulting image.

Once that was generated and sent to the computer, it was time to buckle in. Engraving this was not a quick job. I sat at the laser printer for almost 5 hours just waiting for each pass to be completed. In hindsight... this was easily something I could have done quicker on a newer machine that supported multi-setting passes. I could have probably used the built in software of these newer machines to process the colour images too, but hey -- I never said I wanted it easy! It was a fun learning experience despite the long wait. I got more hours on the machine, and that's what is important.

Here's the full process that I elected to do (and document):

After the sticky MDF minecraft cube was assembled (and cleaned with isopropyl), I wanted to try out some more materials, and Jorge had just finished setting up the shiny new Bambu H2D, so it was time to play with vinyl! Over the weekend, I designed a little keychain of my cat, Loaf, whom I miss a lot while studying abroad!

1

2

3

SLIDESHOW ⇲

My sketch process exported from Procreate - do you notice when I decided to make her head bigger (proportionally it's cuter but wow is she ever tubby)

Using multiple strokes and expand function to get the different tool paths (cut vs. engrave)

Different tests on clear acrylic

I traced out an image of her in Procreate, went over it a few times to get the line thickness right, then brought it into Illustrator where it was simple to vectorize it using Image Trace because I used such a thick, clean brush in Procreate. I ran a test pass in the AVR-1400 using clear acrylic (the H2D can't do clear materials because of the laser type) and then I realized (or Jorge told me) that you have to remove the plastic film on the material before engraving or cutting. Haha, maybe aerosolizing more harmful chemicals, oops.

1

2

3

SLIDESHOW ⇲

The H2D's secret weapon - a camera on the toolhead that can "scan" the bed and give you an output that you can place your designs on

Still needed to add tape to see the cut lines from the previous pass - it isn't magic!

Voila! Two versions - one done with the AVR (left) and the other with the H2D (right) - you can visibly notice the resolution difference between the two of them

I wanted the keychain to be bidirectional (mirrored on both sides), so I had to come up with a clever way to do it. In the AVR-1400, I could create little "calibration" spots to line up the tool head, and then just pray. My first attempt I was off by maybe a millimeter. I tried again using black acrylic in the H2D this time, which has a handy little internal camera that lets you preview the cut using a picture of the build area. It's not super precise, and it was tough to see the cut lines so I had to add some tape to the edges of the design so I could properly line it all up, but it was enough to get the engraving done nicely on both sides.

03.7.0

Vinyl Cutting

I have a Cricut Maker 3 in my classroom back in Toronto, so I'm not new to vinyl cutting, so I opted to mix mediums and create something that utilized the laser cutter, engraver AND vinyl cutting for maximum Fab Academy points. I had an idea to create cut outs of the transit maps for both Toronto (where I'm from) and Lima (where I'm living right now).

1

2

3

4

SLIDESHOW ⇲

Tracing out the major streets and transit lines in Illustrator

Had to separate out each of the different coloured transit lines as they would be different jobs done in the vinyl cutter

The preview of the engraving work in the H2D software - I wanted to engrave the transit lines to give the vinyl stickers a bit of "recess" to stick inside of - to make it a bit more flush

And here's what that looked like

I traced out the map in Illustrator and separated the passes into different layers, separating the vinyl cut layers further into individual colours, then it was off to the machines! The H2D handled the engraving and cutting, while our new (old) friend, the Roland CAMM-1 handled the vinyl cutting. Honestly, I enjoyed using this machine much more than the Cricut, which crams their crappy software down your throat, trying to get you to purchase a subscription to their vector library. With the Roland, Jorge showed me a web based tool called Mods Project that could communicate with the machine over serial-to-USB (seriously!) and I could send the G-Code that way. To use this tool with the Roland, right click > programs > open program > Roland GX-GS vinyl cutters > cut, then modify the output by adding a usbserial module (right click > modules > add module > webserial, then hook up the file outputs from the Roland program to the inputs of this block).

I sent 5 individual files to the printer on 5 different colours, one to represent each of Toronto's TTC subway/light rail lines (in honour of Eglinton Line 5 and Finch West Line 6 opening up recently!), then carefully weeded out each sticker and applied it to the birch board that was laser cut. Apologies to the Torontonians reading this, Jorge did not have any purple vinyl for Line 4 Sheppard, so blue it is (Insult to injury for Line 3 Scarbourough).

1

2

3

SLIDESHOW ⇲

Weeding out line 4 - noticeably the wrong colour

Line 2 applied to the board

and the rest of the lines added!

Of course I made one for Lima too, but I didn't include the vinyl yet because only 1.5 of the 4 planned metro lines are actually built. Hopefully soon all Limeños and Limeñas can hop on board and enjoy their trains soon! The bus rapid transit system here is also really cool, more cities should adopt it!

All the files for this week's creations can be found here: