It’s basically ten identical identical spacers cut from 3 mm plywood, with a side benefit of dramatically reducing my scrap plywood stash, then skewered by a pair of absurdly long 4 mm self-tapping metal screws into holes drilled half an inch into the ¾ inch solid wood cabinet floor.
It clears some clutter atop the microwave and, at least to my deflicted ears, sounds much better. At some point I must screw the Raspberry Pi under the cabinet, too, but that awaits further rearrangement.
Confirming the diagnosis, the cool white LEDs worked fine with the light turned on:
With nine spare SI2306 transistors in hand from the last time in this rodeo and minus the sticky adhesive foam covering the PCB, replacing the other driver transistor was no big deal, whereupon the lamp once again worked the way it should:
While I was in there, I spotted a dent in the input filter cap:
Most likely I squished a wire between the cap and the U-shaped steel strut joining the two halves of the pole. I relocated the replacement cap off the circuit board into an open space with a bit more room:
The fragile wires running to the lamp head got their own sheet of silicone tape (not shown here) to isolate them from the U-strut:
Tuck all the wires back inside, snap the housing together, and it should be good for another uhh half year or two.
It’s hard to be sure about such things, but I now have eight spare transistors …
The four control “buttons” on the SmartHeart kitchen scale are copper-foil tabs that sense the presence of your finger though about 5 mm of white plastic and glass:
The main failure mode seemed to come from the microcontroller locking up and refusing to recognize any of the buttons, most annoyingly the On/Tare button, while continuing to measure whatever weight was on the scale with whatever zero point it chose. Recovery involved waiting until the thing timed out and shut itself off.
The two buttons on the left select Kilocalories for any of the various foods arrayed around the display. Depending on how it jammed during startup, it might display the Kilocalorie value for, say, sugar, while ignoring all button presses. As the manual does not mention any way to return to weights after activating the Kilocalorie function, other than turning it off, it’s not clear recognizing the other buttons would be much help.
Because we have no use for those functions, I unsoldered the wires to those sensor pads and it no longer jams in that mode:
The alert reader will note the PCB legend says I have unsoldered the ON/OFF and UNIT wires. If one believes the silkscreen, the PCB dates back to 2015, so it now carries a reprogrammed microcontroller with functions that no longer match the silkscreen.
The overall soldering quality resembles mine on a bad day.
With those out of the way, the scale still jammed and refused to recognize the remaining two buttons. I wondered if it was somehow sensing ghost fingers over both sensors and waiting for one to vanish, so I added a shield ring around the power tab:
That reduced the sensitivity of both sensors to the point where they pretty much didn’t work, without reducing the number of jams.
So I tried increasing the sensitivity of the power tab by replacing it with a larger copper foil sheet:
That definitely got its attention, as it will now respond to a finger hovering half an inch over the glass, as well as a finger on the bottom of the case: it can now turn on and jam while I pick it up.
More tinkering is in order, but it’s at least less awful in its current state than it was originally, so I can fix a few other things of higher priority.
The health plan I use pays $100 toward the year’s over-the-counter healthcare stuff, although with a caveat: you can only buy the stuff from a specific website. As you might expect, what’s available consists of no-name generic products with absurdly high sticker prices and, just to rub it in, the hundred bucks gets paid in quarterly use-it-or-lose it installments.
Seeing as how it was free, I got a kitchen scale:
It has two catastrophically bad design features:
Terrible battery life
Overly sensitive controls
It runs from a pair of series-connected CR2032 non-rechargeable lithium coin cells. Which would be fine, except that the blue LED backlight stays on for 30 seconds after each button touch and draws about 10 mA.
The battery lifetime is best measured in days.
The four control “buttons” on either side of the backlit LCD are touchless sensors using copper foil stickers:
The alert reader will spot those the empty CR2032 coin cell contacts over on the left and a pair of NP-BX1 batteries in the middle.
I figured there was no need to keep feeding it coin cells while I played with it, so I conjured a holder from the vasty digital deep. Normally, that would be an OpenSCAD solid model suited for 3D printing, but in this case the lithium cells exactly filled the space between the PCB and the bottom of the case, so it became a 2D design neatly suited for laser cuttery.
I planned to stick the orange cutout (in 1.5 mm acrylic) as a stabilizer around the pogo pins making contact with the cell terminals from the red cutout (in 3 mm acrylic), but just melting the pins into the acrylic seemed sufficient for the purpose. Strips of adhesive sheet saved from the margins of previous projects affix the holder (not the cells!) to the scale’s upper glass layer.
As far as I can tell, the scale is perfectly happy running on 7.4 V, rather than 6.0 V. The PCB has two terminals marked +3V and +6V, so it probably depends on which LEDs they use for backlights:
The alert reader will notice a peculiarity concerning the sensor pad connections along the top edge.
Reasonable people disagree as to the cause of the failure, but a replacement controller for the (new) Bafang motor I’m installing on my bike just arrived in the mail.
Disassembling the motor is straightforward, except for the part where you must excavate an internal plug from the silicone snot gluing it into place, eventually revealing its socket:
Regrettably, there seems no way to do that without destroying the dense closed-cell gasket around the connector:
Equally regrettable: a replacement gasket wasn’t included with the replacement controller. Although I don’t have any of the specific foam, some marginally less dense foam from the Big Box o’ Padding seemed suitable for laser cuttery:
The upper left prototype suggested a slightly larger rear bar that produced the gasket in front, which fit snugly:
It lacks the latch cutout, but the foam is squishy and I expect to never touch it again.
A generous glob of hot melt glue holds everything in place:
Although the usual Youtube videos show folks slathering RTV silicone caulk on these connectors, that’s a Very Bad Idea™, because RTV caulk releases acetic acid as it cures. That’s not a problem in the open-air siding-and-lumber environment the caulk was intended for, but sealing a glob of the stuff inside an enclosure will eventually corrode all of the electronics therein.
Cutting intricate doodads has become trivially easy: if you can draw it, you can pretty much cut it, just like that:
A new ResMed ClimateLine headed CPAP hose arrived on schedule and let me measure the old hose:
The center two they-are-not-USB contacts on the input end of the hose (on the right) are for to the heating element spiraling around the tube and measure 10.0 Ω.
The outer two contacts report back from what must be a 10 kΩ thermistor embedded in the dingus sticking into the hose lumen at the output end (on the left). It reads 12 kΩ in a 68 °F room and responds to warmth with a lower resistance, which is what you’d expect.
Plugging the ClimateLine hose into the AirSense 11 unit enables temperature sensing at the end of the hose, with closed loop control from 60 °F to 86 °F. Mary set it to 80 °F in manual mode, which apparently produces different results from the same temperature in Auto mode, and declared victory.
A humidity setting of 4, in the middle of the 1 – 8 range, works for her.
Both the AirSense 11 and its power brick claim 24 VDC at a suspiciously exact 2.71 A. The hose heater could soak up 2.4 A of that, but the AirSense 11 also heats the humidifier’s water tank (“tub”), so it’s unlikely they’re both seeing the full 24 V.
I am prohibited from further investigation. [grin]