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.

Author: Ed

Amazon Basics Alkaline AA Cell Failures

A few weeks ago, the house seemed unusually warm when I crawled out of bed. Checking the heat pump thermostat woke me right up:

Heat pump – battery critical

This, as they say, is not a nominal outcome.

A pair of AA alkaline cells powers the thermostat and, due to its wireless communication link to the heat pump’s air handler in the attic, it chews through two pairs a year. As you’d expect, it displays a “Battery Low” message for at least few days at the end of their lifetime, which was not the case for this failure.

After replacing the cells, the thermostat reported that, yes indeed, the house was much warmer than usual:

Heat pump – high temperature

A temperature monitor showed the heat had jammed on in the deep of the night:

Heat pump – runaway temperature

The heat pump exhaust temperature showed a similar event:

Heat pump – exhaust temperature

One of the AA cells showed about 1.3 V, but the other was around 0.25 V, suggesting an abrupt failure, rather than the normal gradual voltage decrease with plenty of time to replace the cells.

It’s reasonable to jam the heat on when the thermostat isn’t communicating, rather than let the house gradually freeze, but it did come as a surprise. I don’t know how the heat pump reacts to a battery failure during the cooling season; not refrigerating the house would be perfectly fine in most circumstances.

The Amazon Basics AA cells I’ve been using have worked as well as the Name Brand ones, so I was willing to write one off as happenstance.

However, during the recent Daylight Saving Time dance, I discovered the clock in Mary’s Long Arm Sewing Room had stopped, with an Amazon Basics AA alkaline cell from the same lot inside:

Failed clock AA cell

The date shows I’d replaced it in March, with the previous cell lasting an amazing 3-½ years. This one was completely dead, reading barely 0.1 V, after seven months. Mary hasn’t had a quilting project at the long-arm stage in recent months, so the clock may have been stopped for quite a while.

Perhaps something has gone badly wrong with Amazon’s battery supplier QC.

As the saying goes: Once is happenstance. Twice is coincidence. Three times is enemy action.

2025-11-04
Prusa MK4+MMU3 vs. Spool Join: Whoopsie
Because nobody will ever see the Radiator Sleds, I started a batch with the tail end of the white PETG spool and set up the Spool Join function to switch to the retina-burn orange PETG when the white filament ran out.

The two colors combined nicely on that layer:

Prusa MK4 MMU filament joining

Unfortunately, the Spool Join didn’t work out quite right and I had to extricate the white filament from the MMU3, then coerce the orange filament into position.

The key section of the MMU3 looks like this:

Prusa MK4 MMU filament joiningPrusa Mk4 MMU3 selector

The Selector assembly rides on the smooth rods, driven by the stepper motor on the far end of the leadscrew. It stops at one of the five filament tubes (visible to the left of the upper smooth rod, with filament tips showing), whereupon a drive gear pushes the filament into the Selector, under the FINDA sensor (the threaded fitting sticking out of the top), into the PTFE tube, down to the Nextruder, through the idler to trip the Filament Sensor, then into the extruder’s planetary drive gear.

I think this happened:
- The rear end of the white filament passed through the FINDA sensor
- The MK4 reversed the Nextruder to drive the filament back into the MMU3
- The rear end of the filament didn’t reenter its filament tube and escaped out to the side
- The MMU3 drive gear couldn’t pull the filament backward, because the back end was misplaced
- The Extruder planetary drive gear couldn’t pull the filament forward, because the front end was now above the gear
- Both the FINDA and the Filament Sensor showed the filament was present, so the MK4 knew something was wrong
Fortunately, I was watching the whole operation and could intervene.

The MMU3 works well when the filament behaves properly, but it’s very sensitive to bends in the filament and misshapen ends. In this case, the white filament had the usual tight curve due to being would around the spool hub, which was enough to mis-align its end with the MMU3 tube while backing out.

Trust, but verify.
2025-11-03
Handlebar Grip Sleeve
Mary’s zero-mph crash loosened the starboard handlebar plug enough to let it eventually decamp for parts unknown. Its replacement, a somewhat fancier aluminum plug with an expanding cone retainer using an actual M3 nut, worked fine for the last year, but Mary recently noticed the socket head screw had worked loose.

In the interim, I’d moved the Bafang thumb control from its original position on the crossbar to just above the rear shifter:

Tour Easy – right handlebar control stack

Which moved the clamp on the shortened grip off the end of the handlebar tube, so I flipped the grip around, tightened the clamp, and installed the plug.

Unfortunately, the grip ID is 4 mm larger than the tube ID, which meant the plug’s cone retainer was struggling to hold on in there. Perhaps the plastic cone has relaxed bit, but I figured giving it more traction would be a Good Idea™ before I declared victory:

Handlebar Grip Sleeve – PrusaSlicer

It’s a little plastic sleeve with slots to let it expand against the inside of the grip:

Handlebar grip sleeve – installed

Yes, it’s sticking out slightly; you can see the corresponding gap up inside next to the tube.

A wrap of double-sided sticky tape glues it in place as the retainer presses it against the grip ID and a dot of low-strength Loctite should keep the screw from loosening again.

The OpenSCAD source code:
```
// Handlebar grip sleeve
// Ed Nisley - KE4ZNU
// 2025-10-25

include <BOSL2/std.scad>

/* [Hidden] */

ID = 0;
OD = 1;
LENGTH = 2;

HoleWindage = 0.2;
Protrusion = 0.1;
NumSides = 3*2*4;

$fn=NumSides;

Sleeve = [18.5,22.0,14.0];
Kerf = 1.0;

  difference() {
    tube(Sleeve[LENGTH],id=Sleeve[ID],od=Sleeve[OD],anchor=BOTTOM);
    for (a=[0,90])
      zrot(a)
        up(Sleeve[LENGTH]/4)
          cuboid([2*Sleeve[OD],Kerf,Sleeve[LENGTH]],anchor=BOTTOM);
  }
```
That was easy …
2025-10-31
Monster Restocking

This being the season of monsters, I rebuilt and deployed a few from last year:

Alligator – installed

Real spiders have better camouflage, but cardboard works pretty well:

Spider 1 – camouflaged

I must move them to the driveway on Friday:

Spider 2 – camouflaged

This one stands out in any situation:

Green spider

Tiny T-Rexes are just cute:

Tiny T-Rex with pillar lamp

They need support, because their heavy head bends the spine just above the hips, so they’re snuggled up against pillar lamps or each other:

Mantis and Tiny T-Rexes

The cardboard Mantis turned out entirely too fragile:

Mantis – dismantled

Reassembling the poor thing with a blob of hot-melt glue on each joint held it together.

If I used something other than corrugated cardboard, they’d likely survive longer out there.

A handful of RGBW tea lights provide just enough illumination to make them visible.

2025-10-30
Mantis Durability

A Praying Mantis appeared on the house wall:

Mantis – alert

The next morning found it huddled against the cold:

Mantis – chilled

It had reached operating temperature and gone about its business a few hours later.

I deployed a cardboard Mantis in its honor as a seasonally appropriate yard decoration, but mine didn’t survive the night nearly as well as the real one:

Mantis – dismantled

I doubt a predator was involved …

A site search will reveal previous encounters with their kind.

2025-10-29
No Kings
I discovered this commentary, in several variations in different contexts, after attending the Poughkeepsie No Kings protest last weekend:

You are allowed to say, at any point, “I can’t support this”.

Even if you did.

Even if you were unsure.

You can say, at any point, “This has gone too far.”

And, while the best time to say that was earlier, the second best time is now.

That is relevant, because the Executive branch of the United States government has internalized two facts:
- There are no rules
- There are no consequences
The President and the Executive branch now act with the knowledge that the separation of powers, the checks and balances, and the restrictions written into the US Constitution no longer apply.

Justifications based on Constitutional hairsplitting are irrelevant. The Founding Fathers did not intend the Executive branch to operate as it does now.

Justifications based on “But what about …?” are irrelevant. The scale of current malfeasance dwarfs all precedent; there are no valid comparisons.

Justifications based on “But Congress is dysfunctional!” are irrelevant. It takes only one to dysfunction and the parties have been swapping irresponsibility for decades.

I commend to your attention the Army Talk Orientation Fact Sheet 64 from March 1945. It is straight-up US WWII propaganda with a rosy view of the Soviet Union, but you should fact-check all items in the section “Can We Spot It?” on page 4 against current events.

Should you think your particular identity, institution, tradition, behavior, property, possessions, protection, legality, or beliefs will remain untouched because you’re in a particular group, you are incorrect.

I changed my voter registration to “No Party” several decades ago, when it became evident the Republican Party had lost interest in whatever small-government / low-deficit / personal-responsibility principles it may have once had; thinking it had those principles was likely a misunderstanding on my part.

I cannot support many planks of the Democratic Party’s platform, either, but they remain based in rule-of-law and have some appreciation of what functions a government should perform.

I still vote in every election and intend to continue doing so.

WordPress likes images and this seems appropriate:

No Kings Logo
2025-10-28
Dirac Delta: Hacker News

Every day, about 200 readers peruse 300 pages, but sometimes this happens:

Softsolder vs Hacker News – 2025-10-24

Unlike AI or robot scrapers, this spike came from actual humans reading a note on Hacker News.

I was surprised at how much traffic came from the various Referrers based on Hacker News.

Here today, gone tomorrow …

2025-10-27