i think if you just need an arm computer for software projects, sure, but the pi still has a lot of use where you also need a lot of GPIO and don't want to deal with dongles
... with all the light i subjected it to? i don't know, but now with a broken light and a broken charger, it will be a while until i revisit this, but i'll post about it when i do :)
it was good to get a couple of these lights working!
resistor carbonara for all!
yeah! when it didn't work, i wasn't expecting that to be the failure at all, i was hoping for a blown fuse or a shorted capacitor or something.
I found another failure with these lights, one of them, the one in the teardown photo, doesn't respond to movement, probably a damaged sensor...
a zoomed in view of the cremated circuit showing the resistor's caps completely rusted, and the body almost entirely turned to ash and molten metal
a different view of the same circuit in another orientation to show more of the damage
surprisingly, the fuse did blow! I checked before posting because i had a silly joke ready about how the circuit had protected it just in case it survived.
that is the cremated remains of the resistor, but there was a tiny solder blob rattling around under the PCB
a photo of the insides of one charger with its magic smoke still safely inside
I took another charger apart to see what the circuit should look like, and it's simple enough that i think i'll just repair the burnt one. I'll probably replace everything because i no longer trust components that were cooked so hot for who knows how long, even if they read ok on the meter.
the inside of the charger unit showing a carbon layer on the plastic cover as well as covering most of the components of the charger
When i plugged in the charger to place the lights around the house, I noticed that one of them didn't start charging (as indicated by a brief blue flash) so I took the charger apart and was met with the distinct smell of magic smoke and the view of a finely cooked resistor carbonara
the light turned on in the dark, it's brighter near where the LEDs are, and becomes gradually darker along its length
the light turned on with a ceiling light also on
they produce a dim warm light which is bright enough to see my way around at night and are not painful like the ceiling lights are after a few hours in the dark
I noticed there was no battery protection on board, so I took the little boards from the dead batteries and put them onto the replacement batteries which didn't have them and I reassembled the lights
now i'm going to look for batteries that fit from my salvaged battery collection, and they should be good!
a close-up of the main board with 7 LEDs at the top, the motion sensor in the middle, with "S427E" written on it, a PIC16F684 at the bottom right, and the charging circuit on the left of the board. two screws and two plastic bits hold it in place
taking a closer look at the main board, i'm surprised to find a PIC16F684 microcontroller. I thought these lights worked just with plain old switches and jellybean logic ICs for the motion sensing, but i guess a microcontroller is cheap enough these days. I can't find any info on that motion sensor.
the light taken apart. 2 PCBs inside, one for control and the main night light LEDs, another for a pair of LEDs that you can use as a flashlight. one button which turns on the flashlight mode, a receiving charging coil for wireless charging and the motion sensor all visible. next to it, the inflated battery, so inflated that the pouch is starting to unroll at the edges.
once inside, i found the culprit, a spicy pillow! the batteries in all of them had become inflated and measured between 0.8 and 1V. that metal block is the magnet, and the round thing is the wireless charging receiving coil
on the left of the image, the mains plug and cradle for the night light. the light goes on it magnetically. on the right, the light, a rounded rectangular shape, with a motion sensor at the bottom.
with my night vision struggling a bit more these days, i went to find some old motion sensing night lights that were in the house a few years ago that died of unknown causes. i found them and took them apart
the FBSOA curves for the IXTP80N075L2 MOSFET, the curve @Tc=25ยฐC shows a lower Id for the same Vds compared to the curve @Tc=75ยฐC when operating at DC, so it looks like continuous power dissipation is higher when the transistor is hotter. the shorter pulses do have lower power under FBSOA when Tc=75ยฐC. I may just not be interpreting this correctly, or there might be a mechanism i'm not aware of that would make higher DC power dissipation when hotter a thing.
am i not interpreting the SOA correctly? it looks like when the transistor is hotter, it can handle more power than when it's at ambient temperature? is that right? it doesn't sound right, what am i missing? hmmm
they're a bit evil tho... "DC barrel jack is tip negative"
almost symmetry!
the joke was that the box looked kind of like a jewelry box, the more i read it the worse it gets ๐ฌi'll see myself out!
if anyone ever wants to marry me, they better propose in style.
I decided to upgrade my quickly cobbled together dual supply from last year. I hope this turns out well.
bsky.app/profile/anth...
curious indeed!
finally unlocked the power of multiplication!
about the hobbyist bit, yeah! i just picked the first result in the package i wanted, and this will still do what i need it to, just a curious thing that two versions exist
looking further down in the datasheet i found that the thicker tab one is called "dual gauge" and has a tab that's ~1.23mm thick. I just chose the first LM317 result i found for the TO-220 package without realising there were two versions. this is the only information on thermal specs
yeah! i got this after my order. no mentions of the thinning of the linear regulators, so that probably is a much older change
so linear regulators have become thinner! I thought only the fake ones had tabs this thin! for a moment i thought mouser had started shipping fake parts, but then i looked in the datasheet and, well, they did get thinner indeed! ~0.5mm specified in the datasheet. i wonder when this happened, hmm!
I forgot to consider the MOV's capacitance, which measured 2.5nF. at 60 Hz that's ~1.05Mฮฉ and that accounts for the 108ยตA I measured! ._.
the extremely old varistor with one of the parallel capacitor's terminals de-soldered and standing on the newly acquired third leg
i then connected the extremely old varistor after removing the parallel capacitor, and the current through it was about 50ยตA. 108ยตA leakage for the new varistors doesn't feel right. Any thoughts on this?
B72220Q0151K101 connected to mains 120VAC in series with a multimeter in microamps mode, another multimeter is measuring the voltage across the varistor/multimeter series. 108.3ยตA and 120.4V
i/v curve of the varistors in the "EnergetiQ" series, specificallt the SIOV-Q20 ones, which is the series my varistors are from.
almost 40 year old varistor! in parallel with a capacitor.
So, I found an old power strip my dad had in storage. It had MOVs from the 80s in it. I wanted to replace them, so I got new ones from farnell. I got these: B72220Q0151K101. the datasheet shows this curve for ileak. i'm getting 108ยตA at 120V. have i gotten duds? i've tested 3, am i testing it wrong?
at least it had personality, not made by the lifeless corporate cookie cutter template that spat out the new one ๐
