Archive for July, 2018
A long time ago, a pair of white LED + red laser flashlights powered by an AA cell diverged: one flashlight worked fine, the other always had a dead battery. The latter ended up on my “one of these days” pile, from which it recently emerged and accompanied me to a Squidwrench Tuesday session:
The black wire trailing from the innards goes to the battery negative terminal, with the aluminum body providing the positive terminal connection to the wavy-washer spring contact visible atop the rear PCB inside the front shell.
The switch connects each red wire to the battery negative terminal, so there’s a color code issue in full effect. The two red wires burrow through holes in the rear PCB (shown above) and connect to the negative terminal of the laser module (the brass cylinder near the top) and the negative terminal ring on the front PCB holding the seven white LEDs:
Continuing the color code issue, the black wire from the laser is its positive terminal. The out-of-focus wire (an LED pin) sticking up near the top of the picture carries the positive connection to the LED ring. The red wires from the switch are the negative connections for the LEDs and laser.
Voltages applied to the LED ring and the currents flowing therein:
Seven LEDs at 20 mA each = 140 mA, so the voltage booster must crank out slightly more than 3.2 V. They’re not the brightest white LEDs I’ve ever seen, but suffice for a small flashlight.
A crude sketch of the PCB layout, with a completely incorrect schematic based on the mistaken assumption the SOT23-3 package was an NPN transistor:
Obviously, that’s just not ever going to oscillate, even if the
2603 topmark meant a 2SC2603 transistor, which it doesn’t.
A bit more searching suggests it’s a stripped-down Semtech SC2603A boost converter, normally presented in a SOT23-6 package. If you order a few million of ’em, you can strip off three unused pins, do some internal rebonding, and (presumably) come out with an SOT23-3:
That topology makes more sense!
Before going further, I had to rationalize the colors:
Soldering longer leads to the PCB allows current & voltage measurements:
With the LEDs and laser disconnected, the converter seems to be struggling to keep the capacitor charged:
Those purple spikes come from the current probe at 200 mA/div: maybe half an amp in 5 μs pulses at 6 kHz works out to a 15 mA average current, which is pretty close to the 11 mA I measured; it’s not obvious the Siglent SDM3045 meter was intended to handle such a tiny duty cycle.
Obviously, the output capacitor is junk and, after removing it, the AADE L/C meter says
NOT A CAPACITOR. Perhaps it never was one?
Measuring the cap in the good (well, the other flashlight) suggests something around 100 nF, so I installed a random 110 nF cap from the stash. The current peaks are about the same size:
The cap voltage (not shown) is now nearly constant and the 50 Hz PWM rate reduces the average battery current to 100-ish μA:
Not great, but tolerable; a 1000 mA·h battery will go flat in a few months.
The LED current runs a bit hotter than I expected:
The bottom is about 200 mA and the average might be 350 to 400 mA.
Compared with the other flashlight:
So the cap is maybe a bit too small, but it likely doesn’t matter.
This parts collection appeared atop the driveway wall, arranged just as shown:
It seems something snagged a large bee (not a honeybee!), ate the contents, and left the wrapper behind. We’ll never know the rest of the story.
Puts one in mind of Turner in Count Zero, though.
The glaring white ring around the drain comes from Magic Porcelain Chip Fix epoxy:
What looks like a blob on the left side covers the missing chip, with the rest of the ring filled in to make it look like I knew what I was doing. The drain dried out while we were on vacation, having been scrubbed clean before we left, making for the best surface preparation I could provide.
As it turns out, our resident iron bacteria took about a week to set up shop along the bottom of the ring, producing a pair of small rust-colored dots that will inevitably spread to encompass the whole thing. They’re endemic in the plumbing, impossible to kill off, and nothing more than an unsightly nuisance.
The running lights on our Tour Easy fairing sit just about perfectly level, despite how they appear in relation to the fairing edge:
And, because they’re firmly attached to the fairing mount, there’s no way to tilt them to extract the 18650 cell.
This took entirely too long to figure out:
The LC40 end caps have a recess exactly where it’ll do the most good: capturing the tab inside the cap means it can’t interfere with the rear contact spring:
Swapping cells no longer requires muttering!
Being that sort of bear, I (sometimes) note the date on cells when I change them, as with this notation on the AA alkaline cells in the Logitech trackball:
These Amazon Basics AA cells lasted almost exactly two years, compared with 15 and 20 months from the previous two pairs of Duracell AAs. A few months one way or the other probably don’t mean much, but the Amazon cells aren’t complete duds.
The new Amazon Basics cells have a gray paint job, so they’ve either changed suppliers or branding.
A soaker hose leaped under a descending garden fork and accumulated a nasty gash:
Mary deployed a spare and continued the mission, while I pondered how to fix such an odd shape.
For lack of anything smarter, I decided to put a form-fitting clamp around the hose, with silicone caulk buttered around the gash to (ideally) slow down any leakage:
As usual, some doodling got the solid model started:
A hose formed from chopped rubber doesn’t really have consistent dimensions, so I set up the model to spit out small test pieces:
Lots and lots of test pieces:
Each iteration produced a better fit, although the dimensions never really converged:
The overall model looks about like you’d expect:
The clamp must hold its shape around a hose carrying 100 psi (for real!) water, so I put 100 mil aluminum backing plates on either side. Were you doing this for real, you’d shape the plates with a CNC mill, but I just bandsawed them to about the right size and transfer-punched the hole positions:
Some drill press action with a slightly oversize drill compensated for any misalignment and Mr Disk Sander rounded the corners to match the plastic block:
A handful of stainless steel 8-32 screws holds the whole mess together:
These hoses spend their lives at rest under a layer of mulch, so I’m ignoring the entire problem of stress relief at those sharp block edges. We’ll see how this plays out in real life, probably next year.
I haven’t tested it under pressure, but it sure looks capable!
The OpenSCAD source code as a GitHub Gist:
We negotiated the Belmar Bridge
connection stairway from the Allegheny River Trail to the Sandy Creek trail:
We’re maneuvering Mary’s bike, but you get the general idea. Our bikes aren’t built for stairways, particularly ones with low overheads:
For some inexplicable reason, I didn’t have a roll of duct tape in my packs, so the temporary repair required a strip of tape from a battery pack, two snippets of hook-and-loop tape, and considerable muttering:
It was good for two dozen more miles to the end of our vacation, so I’d say that was Good Enough.
The new version has holes in the ferrules ten stay diameters deep, instead of six, which might eliminate the need for heatstink tubing. I added a small hole at the joint between the curved hooks and the ferrules to force more plastic into those spots:
I also bent the hanger extension to put the fender’s neutral position closer to the wheel.
We’ll see how long this one lasts. By now, I now have black double-sticky foam tape!
The OpenSCAD source code as a GitHub Gist:
As a bonus for paging all the way to the end, here’s the descent on the same stairway:
No, I wasn’t even tempted …