Ed Nisley's Blog: Shop notes, electronics, firmware, machinery, 3D printing, laser cuttery, and curiosities. Contents: 100% human thinking, 0% AI slop.
The Screen Filter (DD81-02011A) in our Samsung dishwasher (DW80K7050US) turned out to have a mold defect from the factory that’s been sitting there since the previous owners had it installed back in 2018:
Samsung dishwasher screen filter – gap
The mesh apparently didn’t quite make it into the molded plastic, so that little gap has been letting debris in the wash water circulate through the spray bars and clog the orifices.
That’s a bead of EVA hot melt glue that will probably withstand the 163 °F = 73 °C “sanitize” cycle we haven’t had any occasion to use and seems no more toxic than anything else around here.
Protip: if your dishwasher has a filter, it’s likely clogged with a nasty accumulation of gunk, too …
A wood desk chair that I’ve known since I was a pup finally got some much-needed attention, although not a restoration. By and large, I’m finally sorting out that corner of the basement and needed to put the chair’s parts back together so I can work on something else.
The wood seat consists of several slabs glued along keyed joints, one of which had fractured into a rough mess. Amazingly, the two sides fit perfectly together, albeit with the bottom no longer a planar surface, and glued up just like they should:
Wood desk chair – seat clamping
The chair isn’t up to contemporary office standards, but it has a seat elevation screw, a backrest with adjustable angle & elevation, and even a backrest tension setting:
Wood desk chair – ironwork
It was the cutting edge of desk chair technology:
Wood desk chair – patented
I vaguely recall it rolled on long-vanished steel-wheeled casters. Somewhat less long ago, one of the legs broke enough to lose its caster socket (about which, more later), so I set about yanking the three remaining sockets:
Wood desk chair – caster socket removal
During that struggle, another leg revealed a neat woodwork joint:
Wood desk chair – leg joint
It’s easy to remove a caster socket when you can bash it from the top!
Gluing that piece back in place required Too Many Clamps™ aligning it with the leg:
Wood desk chair – leg clamping
But the end result looks pretty good:
Wood desk chair – leg glued
They did a nice job of matching the wood grain; I hadn’t noticed that joint while attacking the socket.
Pending restoring the broken leg’s socket, the soon-to-arrive new casters will clash horribly with the chair’s woodwork. At least it’ll roll again and its new plastic wheels won’t scar the floors.
One of the folding wood chairs that Came With The House™ had a loose arm that turned out to be due to a missing chunk of wood:
Wood chair arm – as found
The obvious lay of the grain shows why it failed like that, surely hastened by the crack below the screw.
So I cut a snippet of brass tubing that, mirabile dictu, fit both the hole and the M6 screw, mixed up some wood epoxy and buttered it up:
Wood chair arm – brass tube epoxy fill
The crack extended entirely through the arm and was more extensive that seemed reasonable to expect the epoxy to handle on its own:
Wood chair arm – splits
So I slobberedsoaked saturated the cracks with wood hardener and clamped them shut:
Wood chair arm – clamping
The hardener is intended to solidify rotted wood, but it makes a reasonable adhesive and, being much more liquid than ordinary wood glues, seemed like it would penetrate further into the cracks than anything else on hand. We shall see how this works out.
Rummaging in the Drawer o’ M6 Screws produced a better match to the brass tube than the original flat head screw:
Wood chair arm – repaired
It screws into a fancy tee nut in the upright chair rail, where a dot of thread locker should hold it forevermore.
I hit the exposed end with some sandpaper to smooth off the last of those smears and, after a few years, it’ll probably look like it grew there.
The intended cuts are the dark lines, each with a poorly defined scorch 2 mm on its left. Knowing that the nozzle is about 4 mm, this suggests the beam is off-center enough to juuuust kiss the nozzle and splash the outer part of the beam away.
Having recently spot-checked the alignment and not seen any odd behavior on another platform-spanning project, this was puzzling. Given that the laser recently survived a move from one Basement Shop to another, with plenty of jostling while standing on end, I suppose I should have been more careful.
The biggest clue was seeing the shadow lines only near the front-right corner and noting they got worse farther into the corner. This seemed like the “fourth-corner” alignment problem described by St. Sadler some years ago and covered in a more succinct recent video.
AFAICT, the problem boils down to the difficulty of precisely aligning the beam at the longest distance it travels in the front-right corner. Careful adjustment of Mirror 1, after getting everything else lined up properly, seems to be solution.
The beam is slightly off-center at Mirror 1 and only a millimeter high on Mirror 2 at either end of the gantry travel along the Y axis.
The beam position at the laser head entry upstream of Mirror 3 shows the problem:
Beam Alignment – Initial M3 entry – 2024-05-31
The targets are left- and right-rear, left- and right-front, with varying pulse lengths obviously underpowering the last and most distant shot.
Looks like a classic fourth-corner problem!
Tweaking Mirror 1 by about 1/8 turn of the adjusting screw to angle the beam vertically upward eventually put the beam dead-center at Mirror 3:
Beam Alignment – M3 adjustments – 2024-05-31
The bottom two targets are double pulses at the left- & right-rear and ‑front, so the beam is now well-centered.
A quick cross-check shows the beam remains centered on Mirror 2 at the front- and rear-end of the gantry travel, Mirror 3 is still OK, and the beam comes out of the center of the nozzle aperture:
Beam Alignment – M2 M3 exit final – 2024-05-31
Subsequent cutting proceeded perfectly all over the platform, so I think the alignment is now as good as it gets or, perhaps, as good as it needs to be.
I’ve been using what’s now called a Multi balans chair since shortly after it came out in the 80s, during which time the plastic feet have worn flat:
Balans chair foot – foot wear
By now, the wood bases ride on the floor, which is a Bad Thing I should have fixed long ago:
Balans chair foot – wood wear
The newer Multi chairs have rolling endcaps, but AFAICT that’s not a retrofittable thing.
The feet have no obvious way to get them out, but after I saw how thin the plastic had become on one foot, some experimental carving solved the problem:
Balans chair foot – OEM foot removal
A large bolt threaded into the crude hole provided enough griptivity to yank the feet out:
Balans chair foot – removed feet
With measurements in hand, I picked up a quartet of furniture leveling feet with M10 stems and tee nuts that exactly fit into the recesses:
Balans chair foot – tee nut fit
I generally buy from sellers who include measurements in their descriptions, although I no longer believe any unit-measurement prices. Most of the time the sizes come out close enough to reality for my simple needs.
The stems were, of course, too long, but that’s easy to fix:
Balans chair foot – cutting stem
The saw does yank the stem down at the last moment, but cutting slow & steady thins the steel and reduces the drama to manageable proportions. Fitting a scrap of wood exactly under the screw would be a much better technique; be it so moved.
With the chair set to the mid-angle position I normally use, the feet meet the floor almost perfectly:
Balans chair foot – straight foot contact
At the steepest angle, things get skewed:
Balans chair foot – angled foot contact
Applying my nearly perfect hindsight, I got a set of swiveling feet and found an appropriate scrap of wood:
Balans chair foot – swivel foot cutting
Zero drama!
Which looks exactly like it should with the chair at the steepest angle:
Balans chair foot – swivel foot contact
The chair now sports two pairs of feet:
Straight feet on the rear
Swivel feet on the front
Now, to see how they survive on a chair, rather than motionless furniture.
If you have any idea why the WordPress AI image generator would come up with this, let me know:
Balans chair foot – WP AI image
That’s not hallucination, it’s just plain irrelevant.
While setting up the small table I conjured from scrap, I discovered one of the folding legs no longer had a latch to keep it from folding. Whether it never had one or the latch got lost along the way, there’s no time like the present:
Table leg latch – installed
The bolt I put there in place of the joint rivet precludes a smaller latch along the lines of the simple steel loop on the other leg, so I figured I may as well go large and, with that much surface area, plywood will work just as well as steel for my simple needs.
When those set, I glued & clamped them together in situ, then wrapped the whole mess with what’s basically high-strength friction tape to encourage it to not come too far apart under the inevitable stress when the leg tries to fold with a pile of stuff on the table.
We’ll see how long this survives; if past experience is any guide, it’ll be a while.
The WordPress AI image generator has a shaky grasp of both human anatomy and the blog topic:
Woodwork design by Escher. What is that interesting tool? So many arms, all with nightmare fuel anatomy!
Long ago, I got Mary a cheap “desk calculator” with a vital function: it beeps cheerfully with each keypress. Nothing lasts forever and the aluminum dress panel around the keys has been gradually working its way loose.
So, we begin …
Gingerly remove the panel, un-bend and flatten it, lay it on the scanner, and cover with black paper:
C-Power calculator keyboard cover
Blow out the contrast, threshold the image, do a little touchup, and get a binary mask:
C-Power calculator keyboard cover – mask
Import into LightBurn, trace and discard the image, do some shape optimization, add 0.2 mm to the height & width of one key, propagate those dimensions to other keys (Make same width and Make same height FTW), cut a paper prototype to verify the fit, iterate until it drops neatly into place, cut an adhesive sheet, then peel & stick:
C-Power KK-800A keyboard – adhesive placed
The dress panel was held in place by what was once a quick-setting gooey glue that had long since fossilized. Although it gave up on the aluminum, it was not going to come off the calculator body without more struggle than seemed warranted.
So I stuck the new glue atop the old glue and hoped for the best. You can see traces of the old glue bead through the sheet:
C-Power KK-800A keyboard – adhesive ready
Lay the dress panel in place, burnish between the rows & columns, and it looks about as good as it ever did:
C-Power KK-800A keyboard – restored
If the adhesive sheet also gives up on the aluminum, I’ll try some fancy 3M 300LSE adhesive.
The WordPress AI image generator heard I like keys, so it spat out some keys for my keyboards:
The Smell of Molten Projects in the Morning 
