Computer-Controlled Cutting

At the Waag, there is a large 130 Watt CO₂ BRM laser cutter with a 1300mm by 900mm bed.

We conducted some tests for the group assignments.

I wanted to create a simple fruit bowl: a basic cross raster with a sphere cut out.

The laser cuts straight through so no angles are possible, making the model somewhat inaccurate. The next step is to create actual plates with a circle cut out straight.

Calculating the radius of the circle to be subtracted required some extra effort but I got to use the Pythagorean theorem again.

This is better but now we need slots for press-fit joints. The following code adds slots at the top or bottom. The slots are chamfered to make inserting the plate for a joint easier.

This is quite good! Figure 17 shows a preview of the result is visible, along with two plates for the X and Y axes. Note that kerf is not the actual kerf we measured in the group assignment but a value used to get to the 2.3mm sweet spot found in the press fit test (see figure 16).

Creating the chamfer on the slots on the top side was a bit challenging due to the curve they are placed on. This curve sometimes resulted in a small, ugly hook on the slot (see figure 18). To mitigate this, I added an extra square cut-out (see figure 19).

I created a layout for cutting using the following function:

columns = 4;
distance = 5;

module layout_translate(i) {
  x = (i % columns) * (bowl_w + distance);
  y = floor(i / columns) * (bowl_h + distance);

  translate([x, y, 0]) { children(); }
}

for (pos = [step_size:step_size:bowl_w - step_size]) {
  i = (floor(pos / step_size) - 1) * 2;

  layout_translate(i) plate(pos, false);
  layout_translate(i + 1) plate(pos, true);
}

I was very impatient and created a smaller version to create on a 3D printer for testing. I borrowed this idea from the week 3 documentation of Bas Pijls.

Meh, it's acceptable.

After some trial and error on with the layout (to make it fit on a piece of cardboard), I copied my SVG file to LightBurn and ran the cutter at 100mm/s with power 30%/20%.

It took a while to finish, but the cutting was perfect. I forgot to add numbers to the design for assembly, so I used a pencil to number them.

Putting it together was difficult because the slots were too tight. As clearly visible in figure 24, the "fingers" between the slots began to bend. I needed a flat nose pliers to assemble it and gave up after a couple of rows.

Henk advised me to make two by skipping slots. The first one turned out fine, but the second did not, as I used too many plates in the first. In the second bowl the plates became misaligned. It was too much work to fix it, so I left it as is.

This was a failure, so I cut a second, smaller bowl with wider slots: 2.5mm instead of 2.3mm.

I made the mistake of changing the laser power because I disliked the smoke stains on the back of the cardboard after cutting. I tried 25%/15% but this was unsuccessful. The cut did not go straight through, as visible on the back in figure 26.

I ran it again with the preview settings: 100mm/s and 30%/20%.

Again, I agot a perfect (though stained at the back) cut. Assembly was better but still very tight.

Cardboard is not a precise material. It is uneven and gets compressed when measured for thickness. The joint slots I created were all too narrow because the measurement I took was 2.5mm, but I should have measured a stack of pieces cut from the cardboard, then divided by the number of pieces to obtain a more reliable value.

The precision of a laser cut is very impressive. Removing cut cardboard from the machine feels like magic, as cutting it by hand would leave all kinds of dents and other imperfections. It's also loud and smells like fire.

I like the Atari ST busy bee and want to turn it into a sticker by replacing pixels with black and white dots.

Converting the pixels of such a small image into a text editor is quite easy by converting it to XPM (using convert busy-bee.png busy-bee.xpm) because the XPM-files are just C-code, and the image is stored in a clear array of strings.

/* XPM */
static const char *busy_bee[] = {
/* columns rows colors chars-per-pixel */
"17 17 3 1 ",
"  c black",
". c white",
"X c None",
/* pixels */
"XXXXX.XXXXXXXXXXX",
"XXXX. .XXXX....XX",
"XXXX. .XXX.    .X",
"XXXXX...X.  ... .",
"X..XX.  .  .... .",
".  ...  . .... .X",
"X...  ..  ... . .",
"XXX.  .  . . . .X",
"XXXX..  .   ...XX",
"XXX.   . . .  .XX",
"XX.  ..        .X",
"X.  .... .  ....X",
"X. .... .  .    .",
"X. ... ..  .  ..X",
"X. .. . .. . .  .",
"XX.  . .XX.. . .X",
"XXX..X.XXXXX.X.XX"
};

In the above example, the Xs represent transparency, the spaces black, and dots white. I turned this into an OpenSCAD program:

data = ["XXXXX.XXXXXXXXXXX",
        "XXXX. .XXXX....XX",
        "XXXX. .XXX.    .X",
        "XXXXX...X.  ... .",
        "X..XX.  .  .... .",
        ".  ...  . .... .X",
        "X...  ..  ... . .",
        "XXX.  .  . . . .X",
        "XXXX..  .   ...XX",
        "XXX.   . . .  .XX",
        "XX.  ..        .X",
        "X.  .... .  ....X",
        "X. .... .  .    .",
        "X. ... ..  .  ..X",
        "X. .. . .. . .  .",
        "XX.  . .XX.. . .X",
        "XXX..X.XXXXX.X.XX"];

// select color here
c = ".";

// calculate step size
booklet_w = 130;
booklet_h = 195;
s = min(booklet_w / len(data[0]),
        booklet_h / len(data));

$fn = 200;
for (i = [0:len(data) - 1]) {
  for (j = [0:len(data[i]) - 1]) {
    if (ord(data[i][j]) == ord(c)) {
      translate([j * s, i * s, 0]) circle(s * .45);
    }
  }
}

// alignment markers
m = 5;
if (ord(c) == ord(".")) {
  translate([0, -m, 0]) square([m, m]);
} else if (ord(c) == ord(" ")) {
  translate([-m, 0, 0]) square([m, m]);
}

In this program, c can be set to select the color to render. I placed a square in the bottom left corner for alignment when applying the cuts as stickers. These should be removed after application.

On the day of cutting, I decided against the previous design because I didn't have enough time to do two colors (and I'm unsure if both black and white vinyl were available). So, I opted for a chicken I discovered a couple of weeks ago while experimenting with fonts.

I created the above image using Gimp (Inkscape on my laptop seems broken as it does not produce text) and received instructions to transform it into an outline using modsproject.org. These are the steps to turn my image into something the Roland GCX-24 at the Waag can process:

• programs -> open program -> roland GX-24, cut
• select png file
• Roland GX/GS-24 Relative -> set force 80g
• on/off -> save file
• cut raster -> calculate
• save file

This yields a CAMM-file that I can cat to /dev/usb/lp1 on my (GNU/Linux) computer.

I had all kinds of trouble trying to get the above to work on Google Chrome (using that because I was instructed to use WebUSB to send it to the cutter). The image would= not popup and send file would not do anything. So I opened mods in Firefox (Librewolf to be precise) and went for the save file path.

Next, on the cutter:

• Check if the knife is correctly sticking out of the head.
• Check if the head is tightly fastened.
• Position the sticker roll using the unlock switch on the back.
• Lock the sticker roll in place with the switch on the back.
• Make last position adjustments using the arrow buttons.
• Press Origin to make sure it starts at the current position.

Finally: cat chicken.png.camm > /dev/usb/lp1

I cut out the part with the chicken on it, peeled off the extra material (weeding), put transfer tape on the rest and removed the release liner.

#+name transfer

Then I stick it onto my phone are rubbed the parts with the sticker firmly. The last step: remove the transfer tape.