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.
If it used to work, it can work again
Today looks like a good day to fire up the snowblower and clear the driveway:
One of the bolts holding the muffler on the engine worked its way past its rebuilt locking plate and will require attention before getting out there.
Fortunately, it looks like a good day for shop projects …
Our house came with several single-LED night lights featuring a transparent light guide / reflector:
The plate had snapped off one of them and, being me, I wondered if I could replace it with something similar.
Years passed.
Obviously, this must be made from a transparent substance, which 3D printed things are not, but after some fiddling with parameters I thought the result might be informative.
The guide plate is a section of a spherical surface, here approximated by a BOSL2 spheroid():
The original is 3 mm thick, but 2 mm worked out better for my purposes by reducing the amount of infill:
The intricate base latches into the lamp’s plastic case:
The result is, at best, translucent, because it’s definitely not transparent:
The zigzag pattern seems to come from the icosohedral approximation to the sphere, because it follows the surface tesselation.
Getting the base shape right required several iterations, each printed with the model cut off just above the bottom of the guide plate:
The first two attempts needed attention from a flush cutting pliers before fitting into the case, but they don’t call it rapid prototyping for nothin’.
The original and replacement plugged into an outlet strip:
While you can see the vague outline of the strip behind the printed light guide, it’s definitely lacking in detail:
The striations throw more light into the room than the original:
Fiddling with the 3D printing parameters might make it more transparent, but it’s going back into the box it came from after giving me a better idea of which parameters to tweak the next time around.
The OpenSCAD source code as a GitHub Gist:
| // Nightlight light guide | |
| // Ed Nisley – KE4ZNU | |
| // 2026-01-13 | |
| include <BOSL2/std.scad> | |
| Layout = "Show"; // [Show,Build,Plate,Base,Pipe] | |
| /* [Hidden] */ | |
| HoleWindage = 0.2; | |
| Protrusion = 0.1; | |
| NumSides = 10*3*4; | |
| $fn=NumSides; | |
| ID = 0; | |
| OD = 1; | |
| LENGTH = 2; | |
| function ChordRadius(m,c) = (m^2 + (c^2)/4) / (2*m); | |
| PlateThick = 2.0; | |
| PlateOA = [60.0,50.0,PlateThick]; | |
| PlateRound = 5.0; | |
| PlateTaper = 1.0; | |
| PlateAngle = atan(-2/60); // original plate angle, far end closer to wall | |
| PlateM = 2.4; | |
| PlateRadius = ChordRadius(PlateM,PlateOA.x); // light guide plate | |
| echo(PlateRadius=PlateRadius); | |
| WallThick = 2.0; | |
| MountOA = [23.4,17.0,5.5]; | |
| MountRadius = ChordRadius(4.3,MountOA.x); // base arc in housing | |
| echo(MountRadius=MountRadius); | |
| PipeThick = 5.0; | |
| //———- | |
| // Define shapes | |
| // Oddly intricate base fitting into housing | |
| // Replete with magic numbers | |
| module Base() { | |
| difference() { | |
| union() { | |
| intersection() { | |
| cuboid([MountOA.x,MountOA.y,5.5],anchor=BOTTOM); | |
| back(6.5) | |
| tube(MountOA.z,or=MountRadius,wall=1.5,anchor=BOTTOM+BACK); | |
| } | |
| for (i=[-1,1]) | |
| right(i*18.5/2) | |
| back(11.5) | |
| cuboid([1.8,8.0,MountOA.z],anchor=BOTTOM+BACK); | |
| for (i=[-1,1]) | |
| right(i*22.0/2) | |
| cuboid([1.4,2.0,MountOA.z],anchor=BOTTOM+FRONT); | |
| fwd(5.0) | |
| cuboid([11.0,10.5,MountOA.z],anchor=BOTTOM+FRONT); | |
| } | |
| down(Protrusion) | |
| for (j=[-1,1]) | |
| fwd(j*(1.5 + 10.0)/2) | |
| cuboid([7.0,10.0,MountOA.z + 2*Protrusion],anchor=BOTTOM); | |
| up(3.1) | |
| back(7.5) | |
| cuboid([MountOA.x,25.0,MountOA.z],anchor=BOTTOM+FRONT); | |
| } | |
| } | |
| // Light guide plate | |
| module Plate() { | |
| xrot(PlateAngle) | |
| zrot(90) yrot(90) | |
| left(PlateOA.x/2) | |
| down(PlateM + PlateThick/2) | |
| intersection() { | |
| up(PlateRadius) | |
| difference() { | |
| spheroid(PlateRadius,style="icosa"); | |
| spheroid(PlateRadius – PlateThick,style="icosa"); | |
| } | |
| cuboid(PlateOA + [0,0,2*PlateThick],rounding=PlateRound,edges="Z",anchor=BOTTOM); | |
| } | |
| } | |
| // Light pipe between base & plate | |
| // Magic numbers to fit case opening | |
| module Pipe() { | |
| difference() { | |
| intersection() { | |
| fwd(3.0/2 – 0.2) | |
| cuboid([MountOA.x,MountOA.y,PipeThick],rounding=0.5,edges="Z",anchor=BOTTOM+FRONT); | |
| back(6.5) | |
| cyl(MountOA.z,r=MountRadius,anchor=BOTTOM+BACK); | |
| } | |
| down(Protrusion) | |
| back((1.5 + 10.0)/2) | |
| cuboid([7.0,10.0,1.0 + Protrusion],anchor=BOTTOM); | |
| } | |
| } | |
| module Assembly() { | |
| Base(); | |
| up(MountOA.z) | |
| Pipe(); | |
| up(MountOA.z + PipeThick) | |
| Plate(); | |
| } | |
| //———- | |
| // Build things | |
| if (Layout == "Base") | |
| Base(); | |
| if (Layout == "Plate") | |
| Plate(); | |
| if (Layout == "Pipe") | |
| Pipe(); | |
| if (Layout == "Show" || Layout == "Build") | |
| Assembly(); | |
As a reminder for the next time in this rodeo, the latches holding the temperature adjustment knob on the Delta 17 Series dual-handle bath / shower faucet look like this:
I am unable to apply enough force to the smooth edge of the knob opposite the handle to un-latch it, so I jammed a small prydriver into the gap and twisted enough to pop the latch, at the obvious risk of scarring the chrome plating.
A better approach would involve a plastic prydriver intended for consumer electronics case cracking.
For the record:
Ideally, I won’t need any of that information again.
One of the inline switches I installed to replace the failed switches for the LED lights got unpleasantly warm enough to prompt an investigation:
Yeah, that is not a nominal outcome, particularly in light of the claimed “10 A 250 V” rating.
The overheated plastic pulled back enough to expose the terminal inside:
There was a reason I’d wrapped those switches with known-good 3M electrical tape before deploying them.
That crimp connector took some heat and its screw looks even more unhappy:
It turned out the screw was an itsy too short to compress both the connector and the bent-metal conductor tab against the terminal block:
A 6 mm brass screw with a brass washer did a better job of compressing all parties into one conductive lump.
Although the switch now runs with the case at normal basement temperature, an allegedly UL listed replacement is on its way; it costs about five times more than that switch. If it behaves as it should, I’ll preemptively replace two other switches.
Unbelievably, the ankle zipper tab broke off in my hand:
It’s one of those zippers where the tab releases a lock preventing the zipper from coming unzipped. Mary noped out of removing and replacing the entire zipper.
Trimming a snippet of aluminum miniblind from the Small Box o’ Flat Stuff and two dots of JB Kwikweld epoxy seemed appropriate:
Ugly, but serviceable:
The stray epoxy scraped off under fingernail pressure over the next two days and the pants are ready for the next snowfall.
Because the focus pen worked on the bench, I was certain this had to be true:
There is a break somewhere along the blue wire carrying 24 V to the focus pen. The signal and 0 V wires are fine.
I updated the original post, because I’m going to use that picture a lot whenever the subject of laser machine wiring comes up.
This happened while focusing the laser before cutting the cardboard fixture for the chuck rotary:
The autofocus “pen” = switch did not operate when the rising platform pushed the cardboard against its tip, so the controller continued raising the platform. Seconds later, the platform rammed the cardboard against the laser head and I slapped the Big Red Button.
Those indentations match the focus pen and the nozzle:
Yeah, the platform shoved that pen straight up through its clamp until both punched through the cardboard.
The pen has a red LED (barely visible through the opening around the cable when you’re looking down into it) that did not light up when I manually triggered the switch: either the switch was dead or it wasn’t getting 24 V power.
Having spent considerable time diagnosing similar problems on the LightBurn forum, I was pretty sure the PVC-insulated wire connecting the pen to the controller had failed somewhere in the drag chain.
Update Yup, the 24 V wire was broken:
Another discussion there showed how to dismantle the pen, so I (turned off the power and) cut the cable a few inches from the top of the pen body.
The pen body has three parts screwed together with generous application of threadlock. After demonstrating I lack enough grip strength to break the bonds, I deployed a pair of lathe chucks designed for a death grip on cylindrical objects:
The tip came off readily enough:
The upper joint was more reluctant, to the extent I needed witness marks to show progress:
Dripping Kroil into the slightly loosened joint while twisting it back and forth eventually separated the parts:
I persuaded the last chunks of threadlock out with a stout pin (in a pin vise), eventually letting me screw the pen body together without a struggle.
Contrary to what I originally thought, the switch is a proximity sensor triggered by the reshaped head of an M3 socket-head screw also holding the brass-colored tip. Wiring it to a bench power supply verified proper operation, with the open-collector (actually, open-drain) output going low with any ferrous metal closer than about 3 mm to the sensor tip.
Which put the fault somewhere along the wiring from the controller through both drag chains to the pen, as expected.
Unlinking the X axis drag chain involved a pair of small screwdrivers prying the side plates off their pivots in the next link:
The slightly enlarged opening let me pull enough of the cable through to verify I needed more elbow room, so I dismounted the entire drag chain:
The Y axis drag chain was short enough to pull the cable out without drama.
I guesstimated the overall length from laser head to controller, cut a six conductor 26 AWG silicone ribbon cable generously longer than half of that, peeled it down the middle, then put a JST SM connector where the sections meet at the end of the gantry:
Obviously, those connector halves went on before snaking the other end of the cable sections through their drag chains. I paid considerable attention to keeping the ribbons flat and untwisted throughout their lengths, in hope they’d flex easily as the chain bends.
AFAICT there was no good way to use the old wire to pull the new wire through the chain, so running flexy silicone ribbon cable through a drag chain required tweezers, patience, and persistence. I had to realign the existing wires & tubes at various points so they didn’t twine around each other and block the path.
Another JST SM connector at the laser head allows removing / installing the pen as needed:
The connector pins and sensor wire colors:
GND = blue = common = marked cable conductorOUT = black = sensor output24V = brown = powerWiring the new cable to the controller’s 24 V / GND / LmtU- terminals showed it now worked perfectly.
Reducing the vertical offset between the tip of the pen and the tip of the nozzle was then straightforward …
The Smell of Molten Projects in the Morning 