Ed Nisley's Blog: Shop notes, electronics, firmware, machinery, 3D printing, laser cuttery, and curiosities. Contents: 100% human thinking, 0% AI slop.
Category: Software
General-purpose computers doing something specific
Since the PiHole runs all the time, it now hosts an FTP server to stash snapshots from the cameras onto a 64 GB USB stick. I installed ProFTPD, which Just Worked with a few configuration tweaks:
UseIPv6 off
ServerName "PiHole"
DefaultRoot /mnt/cameras
RequireValidShell off
ftp_snapshot=true
ftp_host="192.168.1.2"
ftp_port=21
ftp_username=$(/bin/hostname)
ftp_password="make up your own"
ftp_stills_dir=$(/bin/hostname)
The last line uses a separate directory for each camera, although they quickly ran into the FAT32 limit of 64 K files per directory; reformatting the USB stick with an ext3 filesystem solved that problem.
A squatter has taken over a defunct domain at the far end of a link buried somewhere in the 3800 posts you find here. In place of the useful page I saw, you’ll see this stylin’ popover:
Domain Squat – engineeration dot com
The “standard security check” is a nice touch, although you should keep in mind the Dilbert cartoon about unexpected side effects.
The actual URL, which I will not make clickable, includes the domain ffgetsplendidapps, which tells you just about everything you need to know about what’s going on.
Because they’re squatting, “continue directly to your destination” means being dumped into a Google search after they’ve meddled with your browser & system configuration. Clicking the inconspicuous × in the upper right closes the popover and dumps you into the search, perhaps before doing anything.
I have no good (i.e., automated) way to find broken links and, as far as I know, there is no way to automatically detect domain squatting, so you’re on your own.
A pair of Step2 rolling garden seats (they have a new version) served in Mary’s gardens long enough to give their seat panels precarious cracks:
Step2 Seat – OEM seat
The underside was giving way, too:
Step2 Seat – cracks
We agreed the new seat could be much simpler, although it must still hinge upward, so I conjured a pair of hinges from the vasty digital deep:
Rolling Cart Hinges – solid model – bottom
The woodpile disgorged a slab of 1/4 inch = 6 mm plywood (used in a defunct project) of just about the right size and we agreed a few holes wouldn’t be a problem for its projected ahem use case:
Step2 Seat – assembled
The screw holes on the hinge tops will let me run machine screws all the way through, should that be necessary. So far, a quartet of self-tapping sheet metal (!) screws are holding firm.
Rolling Cart Hinges – solid model – top
A closer look at the hinges in real life:
Step2 Seat – top view
The solid model now caps the holes; I can drill them out should the need arise.
From the bottom:
Step2 Seat – bottom view
Three coats of white exterior paint make it blindingly bright in the sun, although we expect a week or two in the garden will knock the shine right off:
Step2 Seat – painted
After the first coat, I conjured a drying rack from a bamboo skewer, a cardboard flap, and some hot-melt glue:
Step2 Seat – drying fixture
Three small scars on the seat bottom were deemed acceptable.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Then plotting the data points and eyeballing a straight-line curve fit:
MPCNC – Drag Knife Holder – spring constant
Doing it on hard mode definitely has a certain old-school charm. The graph highlights mis-measured data and similar problems, because, if you don’t see a pretty nearly straight line, something’s gone awry.
But we live in the future, so there’s an easier way:
Droid48 – Spring Rate – Linear Fit coefficients
Well, OK, it’s the future as of the early 1990s, when HP introduced its HP 48 calculators. I’m using the Droid48 emulator on my ancient Google Pixel: living in the past, right here in the future.
Start by firing up the STAT library (cyan arrow, then the 5 key), selecting Fit Data … from the dropdown list, then selecting the Linear Fit model:
Droid48 – Spring Rate – Linear Fit screen
Then tap EDIT and enter the data in a tiny spreadsheet:
Droid48 – Spring Rate – Linear Fit data
My default “engineering mode” numeric display format doesn’t show well on the tiny screen. Tapping the WID→ key helps a bit, but shorter numbers would be better.
With the data entered, set an X value and tap the PRED key to get the corresponding Y value:
Droid48 – Spring Rate – Linear Fit prediction
Tapping the OK button puts the line’s coefficients on the stack, as shown in the first picture. Write ’em on a strip of tape, stick to the top of the holder, and it’s all good:
Encouraged by the smooth running of the LM12UU drag knife mount, I chopped off another length of 12 mm shaft:
LM12UU Collet Pen Holder – sawing shaft
The MicroMark Cut-off saw was barely up to the task; I must do something about its craptastic “vise”. In any event, the wet rags kept the shaft plenty cool and the ShopVac hose directly behind the motor sucked away all of the flying grit.
The reason I used an abrasive wheel: the shaft is case-hardened and the outer millimeter or two is hard enough to repel a carbide cutter:
LM12UU Collet Pen Holder – drilling shaft
Fortunately, the middle remains soft enough to drill a hole for the collet pen holder, which I turned down to a uniform 8 mm (-ish) diameter:
LM12UU Collet Pen Holder – turning collet body
Slather JB Kwik epoxy along the threads, insert into the shaft, wipe off the excess, and it almost looks like a Real Product:
LM12UU Collet Pen Holder – finished body
The far end of the shaft recesses the collet a few millimeters to retain the spring around the pen body, which will also require a knurled ring around the outside so you (well, I) can tighten the collet around the pen tip.
Start the ring by center-drilling an absurdly long aluminum rod in the steady rest:
M12UU Collet Pen Holder – center drilling
Although it’s not obvious, I cleaned up the OD before applying the knurling tool:
LM12UU Collet Pen Holder – knurling
For some unknown reason, it seemed like a Good Idea to knurl without the steady rest, perhaps to avoid deepening the ring where the jaws slide, but Tiny Lathe™ definitely wasn’t up to the challenge. The knurling wheels aren’t quite concentric on their bores and their shafts have plenty of play, so I got to watch the big live center and tailstock wobbulate as the rod turned.
With the steady rest back in place, drill out the rod to match the shaft’s 12 mm OD:
LM12UU Collet Pen Holder – drilling shaft
All my “metric” drilling uses hard-inch drills approximating the metric dimensions, of course, because USA.
Clean up the ring face, file a chamfer on the edge, and part it off:
LM12UU Collet Pen Holder – parting ring
Turn some PVC pipe to a suitable length, slit one side so it can collapse to match the ring OD, wrap shimstock to protect those lovely knurls, and face off all the ugly:
LM12UU Collet Pen Holder – knurled ring facing
Tweak the drag knife’s solid model for a different spring from the collection and up the hole OD in the plate to clear the largest pen cartridge in the current collection:
Collet Holder – LM12UU – solid model
Convince all the parts to fly in formation, then measure the spring rate:
LM12UU Collet Pen Holder – spring rate test
Which works out to be 128 g + 54 g/mm:
LM12UU Collet Pen Holder – test plot – overview
I forgot the knurled ring must clear the screws and, ideally, the nyloc nuts. Which it does, after I carefully aligned each nut with a flat exactly tangent to the ring. Whew!
A closer look at the business end:
LM12UU Collet Pen Holder – test plot – detail
The shaft has 5 mm of travel, far more than enough for the MPCNC’s platform. Plotting at -1 mm applies 180 g of downforce; the test pattern shown above varies the depth from 0.0 mm in steps of -0.1 mm; anything beyond -0.2 mm gets plenty of ink.
Now I have a pen holder, a diamond scribe, and a drag knife with (almost) exactly the same “tool offset” from the alignment camera, thereby eliminating an opportunity to screw up.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Spotted high on the wall of the local USPS office:
Windows Runtime Error – VLC – monitor
A closer look:
Windows Runtime Error – VLC
Huh.
The USPS uses VLC. Who knew?
I darken their doorway so infrequently I have no idea what’s normally displayed up there. Surely it shows advertisements for USPS products, which begs the question: why VLC?
GCMC includes a typeset function converting a more-or-less ASCII string into the coordinate points (a “vectorlist” containing a “path”) defining its character strokes and pen motions. The coordinates are relative to an origin at the lower-left corner of the line, with the font’s capital-X height set to 1.0, so you apply a scale function to make them whatever size you want and hand them to the engrave library routine, which squirts the corresponding G-Code into the output file.
The scaled coordinates cover a distance L along a straight line, so putting them on an arc will cover the same distance. The arc is part of a circle with radius R and a circumference 2πR, so … polar coordinates to the rescue!
The total text length L corresponds to the total angle A along the arc:
A = 360° L / 2πR
It’s entirely possible to have a text line longer than the entire circumference of the circle, whereupon the right end overlaps the left. Smaller characters fit better on smaller circles:
Arc Lettering – Small radius test – NCViewer
The X coordinate of each point in the path (always positive from the X origin) in the path gives its angle (positive counterclockwise) from 0°:
a = 360° x / 2πR (say "eks")
You can add a constant angle of either sign to slew the whole text arc around the center point.
The letter baseline Y=0 sits at radius R, so the Y coordinate of each point (positive above and negative below the Y=0 baseline) gives its radius r:
r = R - y
That puts the bottom of the text outward, so it reads properly when you’re facing the center point.
Homework: Tweak the signs so it reads properly when you’re standing inside the circle reading outward.
Converting from polar back to XY:
x = r × cos(a) (say "times")
y = r × sin(a)
You can add an XY offset to the result, thereby plunking the point wherever you want.
This obviously works best for small characters relative to the arc radius, as the lines connecting the points remain resolutely straight. That’s probably what you wanted anyway, but letters like, say, “m” definitely manspread.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
The Smell of Molten Projects in the Morning 