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.

Category: Machine Shop

Mechanical widgetry

Tour Easy: New Front Fork

A view from the wheel side shows the crack in my Tour Easy’s fork lug had opened a bit more to the rear, which is about what you’d expect from the forces involved:

Tour Easy – cracked fork lug

Removing the handlebar stem from the fork steerer tube requires removing the fairing, its mounting brackets, the fender, a speed sensor, then snipping cable ties to release all the cables and wires. Minus the prep work, removing the fork from the bike isn’t anything special.

The lower bearing (a YST 8311N in black) has rollers, not balls. The headset has J.I.S. 1 inch dimensions, captured in a screen grab to forestall link rot:

YST 8311N headset data

Which means cheap & readily available ISO standard headsets aren’t a drop-in replacement. The incomparable Harris Cyclery has J.I.S. ball-bearing headsets in stock and their Tange Levin CDS HD1002 needs just 1.6 mm of additional washer to match the YST’s 35 mm stack height…

The front side of the crown got rather graunched over the last 14 years, but I punted the problem by rotating the race half a turn to put the eroded spots toward the rear, where they’ll be under minimal stress:

Tour Easy crown bearing – damage

Re-seating the race brought an ancient Headsetter tool from the drawer:

Tour Easy fork with Headsetter

It’s basically galvanized pipe, chamfered on one end, with a set of nuts & washers on a length of all-thread rod just slightly too short for the occasion: this might be the second time I’ve used the thing and I had to supply my own all-thread & nuts. Ah, well, it probably predates the Tour Easy’s design by a decade.

The lower headset race looked to be in pretty good shape, so I left it alone. Normally, such bearing damage gives you indexed steering, but Tour Easy handlebars provide so much lever arm that nothing interferes with the bike’s steering.

The new fork didn’t have a notch for the keyed washer isolating the locknut from the upper bearing race. The usual advice is to file off the key and apply threadlocker, which makes adjusting the two nuts tedious, so I restored the notch in the steerer threads:

Tour Easy – filed steerer tube key slot

Yes, that’s a lethally sharp steel shaving from the not-very-well-reamed ID curling up in the middle of the notch.

The fender mount bridge on the new fork sits half an inch higher in relation to the brake bosses, putting the fender against the V-brake cable hardware. Anything touching the V-brake messes up the pad-to-rim alignment, so I conjured a snippet of aluminum to lower the fender just enough to clear the brakes:

Tour Easy – new fork – fender extender

I think that calls for a nice 3D printed bracket, too, but the snippet got me back on the bike faster. When I preemptively replace the fork on Mary’s bike, then I’ll do a proper bracket for both of us.

The garish red silicone tape replaces the previous black cable ties. It matches the tube paint surprisingly well and doesn’t look good on the fork, so I’ll replace it with cable ties in due course.

A few miles of shakedown riding settled the crown race against the fork, another 1/6 turn of the upper race / lock nut snugged up the bearings, and it’s all good again.

Wow, it’s great to be back on the bike!

(Due to the vagaries of writing this stuff up ahead of time, there’s actually two weeks of realtime between the post that appeared on Monday and this one.)

2015-05-13
Kenmore 158: Largest Spool Holder, Now With Eye Protection

A tip o’ the cycling helmet to Mike for catching this oversight:

Large spool holder – added bead and guide

I’d rounded the end of that steel rod, it stands behind the sewing machine, and blah blah blah. He was right: it needed a bead. That’s a fancy one pilfered from our Larval Engineer’s stash, held in place by a blob of fast-cure epoxy.

Selah.

The safety pin atop the bobbin on the left spool pin feeds the thread into the machine’s upper thread guide at the proper angle to make it all work; a direct line from the spool holder hook isn’t quite right.

2015-05-12
Tour Easy: Cracked Fork!

The fairing on my Tour Easy started making unusually loud booming sounds while we were out on an errand, so when we got home I poked around the front end to see what had worked itself loose. I finally managed to produce the sound, which turned out to be due to a very small motion in the fork:

Cracked Tour Easy Fork

That’s after 14 years and maybe 30,000 miles, so I’d say it did pretty well, all things considered.

On an upright bike a front fork failure kills you: the broken blade rotates forward, jams into the ground, and flips you over the handlebars. I rode about 8 miles with a broken fork and nothing exciting happened.

The Tour Easy’s design dates back to the mid-1970s, when custom bike parts weren’t readily available, and the front fork seems sized for 26 inch tires. A tubular bridge welded across just over the 20 inch (37-406) tire provides a fender mount, stiffens the blades, and, in my case, acts as a second bridge. On my bike, the fork supports the polycarbonate fairing and the Phil Wood hub provides an absolutely rigid connection between the blade dropouts.

For reference, the headset uses J.I.S 1 inch dimensions, with a 27.0 mm ID crown bearing. The stack height runs around 35 mm, but I don’t know the head tube ID.

A pair of forks are on their way; I’ll replace the one on Mary’s bike before it fails…

2015-05-11
Always Sign Your Work

Quite some years ago, I added a wire shelf to the bottom of the “pantry” closet to hold odds-and-ends. The most recent deep-cleaning cycle required removing the shelf, which required removing the mounting brackets to get the fool thing out of the closet.

The backside of one bracket shows I had a bit of trouble matching the mounting holes to the wall anchors:

Pantry wire shelf brackets – overview

The lower bracket bears some advice from my Shop Assistant:

Pantry wire shelf brackets – detail

Check thrice
Measure twice
Cut once

From what little we hear these days, she’s learned the value of always checking her work…

And she signs it, too.

2015-05-10
No Affordance for Pulling

Well, I didn’t expect this:

Unsleeved USB memory

Turns out the only thing holding that case in place was a blob of hot-melt glue on the bottom of the PCB. Hot-melt glue doesn’t bond well to anodized aluminum, the RPi had been sitting outside on a winter day taking time-lapse bird feeder pictures, and the USB connector seemed a bit more snug than usual.

So I slobbered more hot-melt glue on the end of the PCB, jammed the case back in place, and that was that.

The PCB has two snap lines to accommodate shorter cases, with corresponding activity LED locations; it seems I got the long-case version.

2015-05-09
Kenmore 158: Large(st) Spool Holder

Large quilting projects require lots of thread, beyond the capacity of the previous spool adapter, so we came up with a different solution:

Large spool holder

These are cheap & readily available from the usual sources, but recent reviews indicate that the “metal” base has become plastic and the build quality isn’t anything to rejoice over. My feeling is that if it’s going to become a shop project anyway, I should just conjure something suitable from the heap.

The base is a random plastic project box that came with a flimsy sheet-steel top, which I replaced with a rectangle of 0.1 inch = 2.5 mm aluminum plate for more heft. The box is filled with 1.5 pounds of wheel weights, so it’s not going anywhere on its own. The silicone rubber feet probably don’t add much to the project, but why not use ’em?

The feed hook started life as copper-flashed welding filler rod, smooth to the thread and pleasant to the eye, sitting in a hole drilled into a stainless steel 10-32 screw. It’s long enough to feed the thread just above the Kenmore’s top surface. A hook works better than an eyelet: just pass the thread over the hook and you’re done.

The central shaft is a wood dowel, shaped & sanded on the (metal) lathe, held in place by another 10-32 screw. Inside the spool sits a length of “3/4 inch” CPVC pipe (ID = 0.7 inch, OD = 0.875 inch, gotta love those plumbing measurements) that’s a sloppy fit in the just-over 1 inch spool ID.

The smaller spools fit directly on the dowel, perhaps atop the CPVC sleeve.

This seems to work OK, but I’m going to trim the dowel down to just over the length of the spool, so the thread will feed without touching the wood. I thought stacking the smaller spools atop the CPVC sleeve made sense, but that turned out to not be the case.

Took about an hour to conjure with found materials and without a hint of 3D printing…

2015-05-08
HP 7475A Plotter: Chiplotle Supershape
In this day and age, a pen plotter isn’t going to be doing anything useful, because we have better ways to draw schematics and make presentation graphics, but it can produce Algorithmic Art:

HP 7475A – Chiplotle Supershape plotting

Well, granted, that’s a rather small value of Art, but it does show that the plotter can draw 10 k points using serial port hardware handshaking.

That’s one of an infinite variety of Supershapes produced by the Chiplotle geometry.shapes.supershape() function:
```
from chiplotle import *
import math
plt=instantiate_plotters()[0]
plt.set_origin_center()
plt.write(hpgl.VS(5))
ss=geometry.shapes.supershape(3900,3900,5.3,0.4,1,1,point_count=10*1000,travel=10*2*math.pi)
plt.select_pen(1)
plt.write(ss)
plt.select_pen(0)
```
The plotter uses absolute plotter units that range from (0,0) to (10365,7962). Telling the plotter to put its origin in the middle of the page makes perfect sense, because that automagically centers the figure.

Dialing the speed back to 5 cm/s works much better with the Sakura pens than the default 38.1 cm/s = 15.0 inch/s; hand-drawing pens just don’t have the flow rate for prolonged vigorous scribbling. HP was obviously on the edge of converting to metric engineering units in the early 1980s, with the HP 7475A designed before the transition and shipped afterward.

The supershape parameters:
- 3900,3900 sets the maximum coordinate value along each axis. The plot may or may not exceed that value, depending on how weird the supershape turns out, but it’s generally pretty close
- 5.3,0.4,1,1 correspond to coefficients m, n1, n2, n3
- By default, a=1 and b=1, but you can change those as you like
- point_count=10*1000 sets how many total points appear in the plot
- travel=10*2*math.pi sets the number of complete cycles, in units of 2π
The function spits out a list of Cartesian XY coordinates, not the polar rΦ coordinates you might expect.

Slightly non-integer values, particularly for m, produce more interesting patterns. Other than that, there’s just no telling.

Use io.view(ss) to get an idea of what you got, it’s much faster than plotting!

Chiplotle Supershape preview

You may find the online superformula explorers better suited to rapid prototyping, though. There’s a list at the bottom of the Wikipedia article, although some links seem defunct.

Notice that the end of the plot doesn’t quite reach the beginning over on the far right, which is a consequence of how Python produces sequences. Adding one more point does the trick:
```
ss=geometry.shapes.supershape(3900,3900,5.3,0.4,1,1,point_count=1+10*1000,travel=10.001*2*math.pi)
```
Chiplotle Supershape preview – closed

I’ll try remembering that the next time around…
2015-05-07