So i was doing a run-time test of my well-used Black Shadow modded PadMe. This Padme came with a dead original driver, so i modded it back in the fall with the 20-mode / 1.5 volt 7880 driver and a Nichia 219B.
I used this light quite often after the mods, and it worked great with great tint & modes on three AA NiMH cells in Parallel. I t had a great run time in all modes, and good throw & output on Turbo mode, so i decided to test the run time on Turbo with three freshly charged Ladda 2000mah NiMH cells.
Fortunately i had the light apart and in my view while it was running, ( with the main driver plate held to the cells on the flashlight body with a rubber band) to monitor heat build up on the custom LED aluminum pill plate, and after running for roughly 20 minutes i heard a “pop” sound and when i looked at the light there was a visible candle-sized flame flickering off the driver, with the torrid coil glowing red hot. After i blew the flame out and turned off the light, i checked the cells and LED and the volts were still at 1.31 V, and LED survived, but the driver is literally cooked with the smell of burned wiring and thermoplastic in the room.
This is the first time i had a light “burst into flame” , and its not even a “xxxfire” light.
*+
Edit Update: - i discovered the solder joint and resistor near the Emitter leads has de-soldered and shorted together, causing the driver failure. (You can barely see the resistor near the red wire lifted off the board on an angle in the second photo, and the solder on the red lead is now contacting the black wire shorting the connection.+*
It might be a test worth doing, perhaps it would extinguish the available oxygen and put itself out, or overheat and blow a li ion (or NiMH), its something i think worth testing (outdoors)
Is there a way to determine which component on the board failed?
I think the most likely conclusion to this fire is that it would put itself out quickly once a component “failed open”, ending the shenanigans. I could be wrong about that though!
Thanks for posting this, it’s interesting stuff.
I think that I have some of those 7880 drivers. In fact, now that I think about it… I think I even tested one some. It may be in an SK68 right now. As I recall the efficiency was low. Maybe very low, I don’t recall. I do recall that input current would increase with voltage while output current remained stable - that means that efficiency was dropping like a stone above a certain voltage.
FWIW, I’ve just pulled a 7880 driver out and taken a look at it. I think that it may be a flyback driver. I’m basing this on the fact that the inductor has 4 leads rather than 2. Flyback converter - Wikipedia
The torrid coil definitely has two windings on it, (four leads) i have used this light quite often with no problems. And have other lights with the same driver modded in it. I like the modes settings it has, regardless if its inefficient of not. Though it did have really good run times on the lower modes with the three NiMH cells in parallel.
its probably what happened here, with the three NiMH cells in parallel the current supplied might have been to high. I also use this same driver in single-AA lights with no problems at all.
I’m not familiar with the term “Forced-Boost” driver either, as i known this 7880 as a Regulated standard boost type driver. ( steps the voltage up from 1.5 to the voltage needed to drive LEDs.
I modded an old plastic incan billy light last year. The type that took a 6V lantern battery.
It's now running an XM-L and one of these drivers - powered by 8AA Eneloops in parallel for runtime.
The emitter is mounted on a PC chipset heatsink but the driver is simply sitting in place held by hot-glue.
I don't often run it at max, but perhaps I should do some spot checks on the driver temperature before leaving my latern unattended. Thanks for the warning DBSAR!
Of course not! Fortunately only one needs to fail open. Or desolder itself, or whatever. Like I said, maybe I’m wrong… but that’s where my money would be: I would bet that the driver (as a unit) would fail open rather than build up enough pressure to pop the tailcap.
How much current is running to the LED on turbo? (and how much current draw from Battery) With rough estimate of driver (in)efficiency may be enough to glow a toroid.