My underwater kinetics 2AAA flashlight died. I had it on my
keychain, and it took a pretty hard knock I guess when I dropped
the keys the other day.
The SMT inductor broke off its solder pads, and a tiny widget
busted off the top of one of the transistors.
I soldered the inductor back on, and after a bit of head
scratching I decided that the tall, dark colored, grainy
cylidrical thing that was epoxied to the SMT transitor was
probably a heatsink. So I simply used 5 minute epoxy
to glue it back together at the break.
I also gussied the other SMT devices up with a few dabs
of epoxy as well, because the original dabs were a bit
on the spare side.
Likely it's a dc to dc converter of some sort, much more efficient than
wasting power in a dropping resistor. I'm amazed at the kind of
efficiencies I see claimed in chip manufacturer's ads these days.
Jim just confirmed what I've been saying for the last ten years or so,
repair of consumer electronics has become more of a mechanical job than
an electronic troubleshooting one. The parts themselves hardly ever fail
these days, it's mainly "loose disconnections" that keep the parts from
doing their intended tasks.
Resistors waste power. The best way to get a lot of light out of LEDs is to
pulse them at a relatively high rate putting lots of current through them.
I've designed a flashing red LED beacon for a client (it's used in an alarm
system) that puts 250 mA through two strings of LEDs (0.5 A total) . They
are pulsed at 35 kHz using a small, very cheap microcontroller (25c in
quantity) with the power stored in an inductor being switched across them
with a MOSFET. Current taken from the supply is < 10 mA. The whole thing is
on a tiny 30 mm by 45 mm PCB using mostly surface mount parts.
The electronics are two transistors, a couple of resistors, and a
flyback inductor, to boost the voltage so that a white LED that takes
four volts to turn on, can be run off of two AAA batteries that only
make three volts.
Now if I only could have found the loose connection that made my
daughter's cell phone stop working - not that I didn't inspect
every bit of the board under a microscope for an hour or so!
Deja vu all over again Jim. My youngest (17 year old) son sat on or
whacked the stuubby little antenna on his two month old Motorola camera
phone (which I'd just bought him to replace the previous phone which he
lost somewhere.) and it twisted over in the plastic threads of the
housing so that it's inside end scrubbed several tiny surface mount
parts right off the board.
Because our cellphone carrier (Cingular) is no longer doing any service
work on phones at their stores here in Taxachusetts, sick phones have to
be mailed in for repair, so I figured there was no chance I could FTF
sweet talk a local repair guy into doing me a favor and fixing it on the
sly. The mail-in service operation wanted $150 to exchange a "damaged"
phone. (Yeah, I know, I should have sprung for the "all risk" insurance
Cingular wanted to sell me, but I must have been projecting my cautious
experience and not thinking about the kid's, so I declined it.)
I couldn't Google up anything in the way of a print for the phone, and I
figured I'd go nuts trying to reverse engineer it and then obtain the
components it needed, so I just pulled the SIM card out of it, chucked
the phone in the trash, and bought the kid a rugged, non-flip, internal
antenna ex-Cingular Nokia complete with wall and car chargers for $30 on
Stuck the SIM card into it, made a quick phone call to Conversent so
they could plug the phone's ID numbers into their computer, and he was
on the air again. He can bloody well live with a plebian phone until
he's on his own - Ha!
(It's times like this when I think children are G-d's punishment for
Yep! Like Jeff said, some sort of a DC to DC thingie. Comments regarding
mechanical repair to modern electronics reminds me: RCA Colortrack TV
receiver - mid '80s model. Modularized; a problem tracked down to the tuner
control module - opened up tuner control module and found a note under the
shield that the "new" set had already been serviced before it was sold to
me. The pecker tracks therein must have been the first fix for an
"intermittent" in the module because it came clear (again) when I
Some time later, when the problem reappeared I was able to fix it with a
popsicle stick propped under the rear of the tuner control module. Now this
was "really cool" - complex electronic problem fixed with a popsicle stick!
Over the years, the module would exhibit the same problem which was handily
repaired via the popsicle stick fix. Ultimately, a stack of popsicle sticks
failed to clear the problem - like it was still intermittent, over a cycle
of months, or sometimes years.
Optovisor (finally) to the rescue revealed a tiny PCB crack around an edge
card connector on the front of the module. This was 4 or 5 years ago.
After soldering over the crack, the old RCA has given trouble-free service
as a bedroom set ever since.
You didn't say you got it working again, but I gather you did. Nice work,
Jim. When it comes to electronics, I'm lost. Only good fortune in my shop
yields a repair once the magic smoke gets out. I don't have a clue how to
trouble shoot the vast majority of devices these days.
I seem to remember reading about pulsing LEDs to achieve higher
percieved brightness. But the LEDs would have a much shortened life
compared to running them at their rated current even though the
average current was the within spec. Wish that article was handy now.
You had some doubts??????
"To be civilized is to restrain the ability to commit mayhem.
To be incapable of committing mayhem is not the mark of the civilized,
merely the domesticated." - Trefor Thomas
Yep, it was working again when I was done. It was actually a pretty
easy diagnosis, as there were *two* parts floating around free inside
the housing, and one of them (the heatsink do-hicky) was a brittle
fracture off its base, so it could only go back on one way!
I felt kinda bad because I didn't take the time to reverse engineer
the schematic for it, but my free morning time was up at that point.
The apparent brightness is the same if the average current is the
same. Pulsing them does not make them look brighter.
This assertion is based on an experiment I did today.
I ran two identical red LED's (automotive brake-turn parts if it
matters, LiteOn LTL 912-VRKSA), one with DC current and the other
getting pulses of 10X that current with a 10% dutycycle so the
average current was the same.
I ran this at a variety of currents and frequencies. In every case,
the LED's looked identical and indistinguishable unless the frequency
was low enough so there was noticable flicker. At 40 Hz, where there
was noticable flicker, the pulsed LED didn't look any brighter but it
was certainly more irritating.
Try varying the duty cycle, at a fixed current/frequency (say 1000 hz,
something up out of noticable by a good ways). IIRC, running one this
way the apparent brightness hit a plateau (visually indistinguishable to
an LED running the same current 100%) somewhere between 30-50% duty
Take that duty cycle and ramp the current up, and you might get a more
noticable difference in output. In specific, if the rated current of the
LED is 20mA, and you find no noticable brightness increase between 50%
and 100% on time, you should see a difference when you run 40mA at 50%.
It has been 16 years or so since I ran that experiment, but I recall it
Particulary if you're an epileptic....Blinking lights, even LEDs, at the
"wrong" rates can trigger epileptic seizures. That's why you sometimes
see warnings in theatre programs when they are using flashing strobe
lights in the show. See for example:
Quite a while ago I heard reports of helicopters crashing in Viet Nam
because a few pilots pilots had previously unrevealed predispositions to
epilepsy which caused them to go into siezures when the rotor blades
chopped sunlight reflecting off something in the cockpit. But, that may
just be an urban legend, it's tough for me to believe that the folks in
charge of such things wouldn't have checked prospective student pilots
out well enough to spot that.
More than likely it was a FET in a constant Current mode driver.
So the 'resistance' of the FET constantly changes as the battery depletes
giving the same light output until there is not enough control or battery.
One concern with the 5 minute - it might be to hot and hurt semiconductor.
know the direction of hit - then look at the massive object - how the gravity
does stuff. Might be a break in the pcb due to a pull...
Mechanical fit that isn't mated directly...
jim rozen wrote:
Ah, it's a moot point now.
Last night I mentioned to my wife that I had fixed the flashlight.
"Oh, what went wrong with it?"
"Some of the parts had come off the circuit board, and I
was eminently clever and managed to fix it. Damn I'm smart."
Or some other kind of preening to that effect.
I proceeded to unscrew the lens to show her what a great
microelectronics handyman I really am.
The lens flipped out of my fingers as it screwed off the
barrel of the light, and flipped across the room. Because
I tracked it closely as it spun across the floor, I found
it right away under the kitchen table.
The circuit board, with LED attached, had come loose and
flew off in some other direction.
Correction: some other *dimension*. It's gone now, I can
say that for sure after seaching for an hour. After looking
high and low, I finally decided that it must have dropped down
into the hole in the kitchen floor where there had been an
old heating pipe at one time.
So I went down the basement and peered up onto the top of the
beam that was right under the hole. Yep, there was something
there. With the aid of an inspection mirror and a (different)
flashlight I extracted the object.
Dumbfounded, I held an adapter that I had made years ago, to
fit PR-style flashlight bulbs into antique navy battle lanterns,
which take DC bayonette lamps. I'm sure I must have looked
for *that* thing for a few hours, ten years ago.
Maybe in ten years I'll be looking for something else and
find that LED on its board. In the meantime I think
I'll plug that damn hole up.