Archive for category Machine Shop
Having recently acquired a pair of photo lights and desirous of eliminating some desktop clutter, I decided this ancient incandescent (!) magnifying desk lamp had outlived its usefulness:
The styrene plastic shell isn’t quite so yellowed in real life, but it’s close.
Stripping off the frippery reveals the tilt stem on the arm:
The photo lights have a tilt-pan mount intended for a camera’s cold (or hot) shoe, so I conjured an adapter from the vasty digital deep:
Printing with a brim improved platform griptivity:
Fortunately, the photo lights aren’t very heavy and shouldn’t apply too much stress to the layers across the joint between the stem and the cold shoe. Enlarging the stem perpendicular to the shoe probably didn’t make much difference, but it was easy enough.
Of course, you (well, I) always forget a detail in the first solid model, so I had to mill recesses around the screw hole to clear the centering bosses in the metal arm plates:
Which let it fit perfectly into the arm:
The grody threads on the upper surface around the end of the slot came from poor bridging across a hexagon, so the new version has a simple and tity flat end. The slot is mostly invisible with the tilt-pan adapter in place, anyway.
There being no need for a quick-disconnect fitting, a 1/4-20 button head screw locks the adapter in place:
I stripped the line cord from inside the arm struts and zip-tied the photo lamp’s wall wart cable to the outside:
And then It Just Works™:
The lens and its retaining clips now live in the Big Box o’ Optical parts, where it may come in handy some day.
The OpenSCAD source code as a GitHub Gist:
The original dimension doodles, made before I removed the stem and discovered the recesses around the screw hole:
With the wrecked 5U4GB safely in the trash, I popped a smaller, somewhat less stately triode from the Big Box o’ Hollow-State Electronics and wired it up with a pair of SK6812 RGBW LEDs:
The tube’s markings have long since vanished, but, at this late date, all that matters is an intact glass envelope!
After two years, the ordinary white foam tape holding the knockoff Arduino Nano lost most of its sticktivity and easily popped off the 3D printed base:
Two layers of 3M outdoor-rated foam tape clear the bottom-side components and, based on current evidence, its stickiness should stick forever more:
The alert reader will notice the mis-soldered 1 kΩ SMT resistor above-and-right of the CH340 USB interface chip. I think those two resistors are the isolators between the 328P microcontroller and the CH340, letting you use the TX and RX lines as ordinary I/O without killing either chip.
Despite the mis-soldering, it evidently passed their QC and works fine. Seeing as how I didn’t notice it until just now, it’ll remain in place until I must open the lamp base for some other reason, which may never happen.
The data output is now on pin A5, to match the rest of the glowing widgetry:
Blobs of hot melt glue affix the SK6812 and wiring to the socket:
The original “plate cap” wiring ran directly through a hole in the hard drive platter, which I embiggened for a 3.5 mm panel-mount headphone jack. The knurled metal plug looms next to this smaller tube, but it looks better (in a techie sense) than the raw hole:
Octal tubes have an opaque Bakelite base, so I devoted some Quality Shop Time™ to the post:
Although I’d made a shell drill for 5U4’s base, this base was so crumbly I simply joysticked the spinning cutter around to knock off the rest of the post:
The shell drill would open the bottom to admit a bit more light. I may do that to see if it makes any visible difference.
I didn’t expect the serrations in the top mica plate to cast interesting patterns around the platter:
Memo to Self: use the shell drill to avoid nicking the evacuation tip!
Fortunately, the link fell off in the box and I recovered all the pieces for a failure analysis:
I’d glued the PLA together with IPS #4, a hellish mixture of plastic solvents including methylene chloride, one of the few chemicals able to chew into PLA, but there’s not much penetration or bonding going on.
Let’s try that again with a bit more solvent.
First, slide the bars into place:
I applied four solvent drops in two passes to give it time to work its way in, put four matching drops on the armor cap, squished the cap in place, tweaked the bar alignment, then applied pressure while contemplating the whichness of the why for half a minute while the solvent worked its magic.
Things look pretty good once more:
There’s no way to determine the repair’s goodness, other than by deliberately trying to snap off a bar, so I’ll just put it back in the box and hope for the best.
This table must sit across the threshold of a walk-in / sit-down shower, with the shower curtain draped across the table to keep the water inside.
Because the shower floor is slightly higher than the bathroom floor, I conjured a set of foot pads to raise the outside legs:
The sloping top surface on the pads compensates for the angle on the end of the table legs:
I think the leg mold produces legs for several different tables, with the end angle being Close Enough™ for most purposes. Most likely, it’d wear flat in a matter of days on an actual patio.
Using good 3M outdoor-rated foam tape should eliminate the need for fiddly screw holes and more hardware:
The feet fit reasonably well:
They may need nonskid tape on those flat bottoms, but that’s in the nature of fine tuning.
And, as with the narrow table, it may need foam blocks to raise the top surface to arm level. Perhaps a pair of Yoga Blocks will come in handy for large adjustments.
The OpenSCAD source code as a GitHub Gist:
For reasons not relevant here, I recently conjured a pair of tables to support an injured arm (ours are OK!) in the bathroom: one table fitting in the narrow space adjacent to a toilet and the other across the threshold of a walk-in / sit-down shower.
The raw material came from a plastic side table intended for outdoor use:
That’s the Patriotic Blue version, which seemed the least offensive of the colors on offer at the local store.
The plastic pieces unsnap easily enough:
The legs also come apart by pulling outward at the crossover points. You may need to clean the flashing from all the joints, as they’re only as finished as absolutely necessary.
A table about half the width seemed about right, so I sawed the two top plates off their struts, then angled the strut ends to match the new leg angle:
Because it’s now completely floppy, I drilled holes for 5 mm screws through the struts:
In the process, I discovered stainless steel nyloc nuts tend to gall on stainless steel screws:
I lost a pair of screws + nuts before I got a clue and began adding a drop of machine oil to each screw before tightening the nuts. Haven’t had that problem with the 3 mm SS screws, so there’s always something new to learn.
With all the screws in place, the half-table becomes a rigid contraption:
The top looks like it’s suffering from severe barrel distortion, but it really started out looking that way:
The slat sides are all curved, except the far edge that was once in the middle of the table and now fits against the wall.
It may be slightly too short, but we can stack foam slabs on the top, probably held in place with cable ties.
Memo to Self: lube all the stainless steel screws!
It’s been running more-or-less continuously since late 2016, so call it
Because I’d be crazy to replace it with another likely-to-fail WS2812, I had to remove both of them before installing SK6812 RGBW LEDs and updating the Arduino Nano.
Unfortunately, I did a really good job of bonding the side light to the tube with epoxy:
The last tube manufacturing step involved flashing the getter onto the tube envelope, so as to remove the last vestige of air. Admitting air oxidizes the getter:
It was such a pretty tube, too …
After considerable evaluation, the Customer decided the shoelaces were still too long and said the hex-crimped ferrules were entirely too rough and tended to snag on things. This time, I prepared the ferrules by chucking them in the lathe:
The steel rod inside the ferrule encourages it to remain round and not collapse while I’m filing off the flange that normally holds the plastic strain-relief doodad:
I snipped another half inch off each end of the laces and crimped on the prepared ferrules:
Which were definitely too jaggy, so they now sport an epoxy coat:
Alas, JB Kwik epoxy has a pot life measured in minutes, so the last ferrule looks a bit lumpy. They seem to work fine and the Customer is happy with the results.
Memo to Self: Next time, dunk the ferrules in a pot of slow-curing JB Weld and let them drain overnight.