The Smell of Molten Projects in the Morning
Ed Nisley's Blog: Shop notes, electronics, firmware, machinery, 3D printing, laser cuttery, and curiosities. Contents: 100% human thinking, 0% AI slop.
Mechanical widgetry
Each target has two scorches made at the left and right ends of the gantry X axis, positioned along the Y axis as noted:
The target has maybe a millimeter of positioning error in the mirror entry aperture, so the beam remains pretty much where it should be.
The beam is hotter toward the rear, as expected, and the front target needed two or three pulses to get a good scorch in the right front corner.
Replacing the sun-rotted hose for Mary’s garden called for a new grip, because of course all hose fittings are different:
The ridges on the fitting looked close enough to half-cylinders and the fitting wasn’t tapered enough to worry about:
The OD came from the original grip, because it neatly fits Mary’s hand, and the nubbles are round-end cylinders.
Got it done the day after the old hose split, glued it on the hose with E6000+, installed it the next morning, whereupon the weather delivered three inches of rain. It’ll get screwed onto the faucet in a few days …
The OpenSCAD source code as a GitHub Gist:
| // Hose fitting grip – simple plastic extrusion | |
| // Ed Nisley – KE4ZNU | |
| // 2025-07-30 | |
| include <BOSL2/std.scad> | |
| /* [Hidden] */ | |
| HoleWindage = 0.2; | |
| Protrusion = 0.1; | |
| NumSides = 3*2*4; | |
| $fn=NumSides; | |
| ID = 0; | |
| OD = 1; | |
| LENGTH = 2; | |
| NumRibs = 8; | |
| RibOD = 3.0; | |
| GripOA = [32.5,66.0,16.0]; | |
| KnobBallOD = 6.0; | |
| union() { | |
| difference() { | |
| tube(GripOA[LENGTH],id=GripOA[ID],od=GripOA[OD],orounding=2.0,anchor=BOTTOM); | |
| for (a = [0:NumRibs-1]) | |
| rotate(a*360/NumRibs) | |
| right(GripOA[ID]/2) down(Protrusion) | |
| cyl(GripOA[LENGTH] + 2*Protrusion,d=RibOD,anchor=BOTTOM); | |
| } | |
| for (a = [0:NumSides-1]) | |
| rotate(a*360/NumSides) | |
| right(GripOA[OD]/2) | |
| up(GripOA[LENGTH]/2) | |
| cyl(GripOA[LENGTH]/2,d=KnobBallOD,rounding=KnobBallOD/2); | |
| } |
A month after the last desiccant change, the silica gel looks like this:
The top cup contains fresh-from-stock dry (regenerated) silica gel beads and the others, left-to-right and top-to-bottom, come from PolyDryer boxes:
| Material | %RH | Weight – g | Increase – g | Water gain – % |
| PETG White | 14 | 26.8 | 1.8 | 7.2 |
| PETG Black | 20 | 26.8 | 1.8 | 7.2 |
| PETG Orange | 13 | 26.8 | 1.8 | 7.2 |
| PETG Blue | 15 | 26.9 | 1.9 | 7.6 |
| PETG-CF Blue | 19 | 27.4 | 2.4 | 9.6 |
| PETG-CF Black | 28 | 27.3 | 2.3 | 9.2 |
| PETG-CF Gray | 27 | 27.1 | 2.1 | 8.4 |
| TPU Clear | 13 | 26.8 | 1.8 | 7.2 |
| Sum of weights | 215.9 | 8.0 | ||
| Measured weight | 216.3 | 8.1 |
I expected some correlation between the indicated humidity and the weight of adsorbed water vapor, but that’s not the case.
The bottom row suggests there’s also little-to-no correlation between bead color and humidity, at least at this low end of the scale.
The indicator cards tucked into the boxes roughly correlate with the meter reading, but they’re much easier to interpret in person.
The old chart of adsorption vs. relative humidity suggests the results are plausible, with the 27-ish %RH being higher than you’d expect from 9-ish % adsorption:
So they’re all set up with 25 g of fresh silica gel, although the boxes no long have the same humidity meters they started with. This likely makes little difference, as I have no way to calibrate them.
The startup ritual for a PolyDryer box’s humidity meter includes:
DO NOT EAT desiccant, to be cut open and …
However, the desiccant packets for the most recent pair of boxes (intended to simplify changing the desiccant in the collection feeding the MMU3 atop the Prusa MK4 3D printer) produced this:
The silica gel in the left cup looks OK-ish, maybe a little dark, but the fresh-from-the-bag beads in the right cup are crying out for regeneration after having adsorbed about all the water vapor they can.
If you were using that silica gel in its original DO NOT EAT bag, where you can’t see what it’s telling you, you might wonder why it wasn’t doing such a great job of drying the box + filament. The same could happen with a bag of non-indicating gel, along the lines of what I was using a decade ago.
So I dumped both in the Needs Rgeneration bottle and filled both meters with 25 g of fresh silica gel.
A long-lost repair finally made it to the top of the list:
The original string had long since rotted out, but everything else was in a plastic bag just waiting for this occasion.
The colorful cylinders are stacks of laser-cut 6 mm disks with a 2 mm hole, held to the wire & string with a tiny dot of high-viscosity cyanoacrylate glue at each end:
The disks came from acrylic leftovers:
The motion you can’t see makes the shiny bikes much more visible out there:
The string came from dismantled badge reels providing spiral springs for the auto-retracting spools in the PolyDryer boxes.
The weight ball had a 2 mm hole filled by a wood plug which I cleaned out piecemeal with a 1.5 mm drill bit in a pin vise; a short length of wood skewer holds the new string in place.
Because the upper arms support more weight, their disk stacks need fewer disks for the same leverage. The original mobile had (at most) four 6 mm chromed plastic balls at each level, so I started with eight 3 mm disks, adjusted the stack length as needed, glued them in place, then removed the surplus disks by crushing them with a Vise-Grip.
I should rip off the design (“© otagiri 1979”) to build another with recumbent bikes.
A small sewing notions cabinet, once my mother’s, now holds some of Mary’s supplies and, a few days ago, had one of its drawer pulls fall off. While preemptively tightening all the screws, I found one no longer held onto its pull:
They don’t make drawer pulls like that any more!
As I see things, it can be forgiven for losing its grip after nearly a century.
Thread the screw in as far as it will go and lay the pull flat on the bench vise anvil:
A few gentle whacks with a pin punch on top and bottom, plus a tap on each side, compressed the pull’s remaining threads around & into the screw:
Put it back in its drawer, snug the screw, and it’s all good.
That should suffice for at least the remainder of its first century …
The Inkscape file contains the overall 16×16 layout of random squares, with the color of each square indicating which of 16 (Inkscape) layers it belongs to:
The next step involves creating a corresponding set of LightBurn layouts to burn those holes out of colored paper sheets to produce layered paper art:
I know it’s art, because that’s what I was thinking when I made it.
Setting up the LightBurn layouts requires enough manual effort to make the whole thing impractical except as a testcase to see how it all worked out. An overview of the LightBurn workspace:
The little bitty grid in the upper left quadrant represents the 700×500 platform of my laser and each of the blue squares is 159 mm on a side. I tend to not delete the intermediate steps, because they serve as some sort of documentation the next time I wonder how I did that thing.
So, we begin.
Import the Inkscape SVG file:
The blue outer square and the blue text identifying it are on LightBurn’s T2 tool layer, with the black squares on the C00 layer. All of that happens automagically, because I used colors from the LightBurn palette in Inkscape.
The lonely square in the upper right is the template from which the other 256 squares were cloned, but it has no further purpose in life.
The 16×16 grid consists of sixteen overlaid groups, which need sorting out for ease of access, so drag each one off into a more useful arrangement:
Note that each of the 256 possible positions has a square in only one of those groups.
Each of the 16 groups corresponds to a sheet of paper, with the squares indicating holes exposing the sheet below it. The color of each square, as seen from the top of the stack, comes from the first sheet in the stack without a hole. Perforce, every sheet above the one without a hole must have a hole, which means you must merge all those sheets.
Line up (duplicates of) those 16 groups in the vertical line forming the left column in this arrangement:
The top group is the layer I named H000 in Inkscape, with the others in order down to H337 on the bottom. You can see why labeling them is pretty much required.
I should have equalized the vertical spaces between the groups in the left column, but it doesn’t really matter.
The rest of the triangle comes from duplicating each group using LightBurn’s Grid Array tool with a convenient space between each copy. Make 15 copies of the top group for a total of 16 H000 and no copies of the bottom H337.
Now the magic happens:
Align Horizontal Centers tool to stack each column into one tidy layoutDelete Duplicates to get rid of all the overlaid outer squaresDuplicate the line of blocks and move it upGroup each block individually to keep all the little squares together with the outlineThusly:
Combine each of those blocks with the sheet cutting template, tweak the binary sheet identification holes, and group the result:
The leftmost block has All The Holes, the next one is missing a few, and so on across the line:
So the leftmost block corresponds to the black mask atop all the layers. Because it doesn’t have alignment holes in the corners or a binary sheet number, you get to align it by eyeball after gluing up the rest of the stack.
The rightmost block has no cutout squares at all and goes on the bottom of the stack. It also lacks a sheet number, but it’s easy to identify.
Set the LightBurn speed / power values for the layers to cut your stock of colored art paper.
Position the Letter Page Holder template to put the center of the sheet cutout at the center of the platform:
Drop the fixture on the platform, use magnets to hold it down, then do a Print and Cut alignment on the corner targets so the template matches the fixture.
Then:
Ctrl-D to duplicate itP to slam it to the middle of the templateAlt-S to Fire The LaserDel to delete the blockI used a stack of paper in rainbow order roughly corresponding to the Inkscape layer colors, but you could stack them backwards or even use random colors and nobody would ever know:
I kinda like it, but wow that took a lot of prep work …
The Smell of Molten Projects in the Morning 
