Re: Higher voltage joule thief...

[Fungus]

After my recent disassembly of a filament bulb I found myself with 4 high voltage "filaments" on my hands (48V?)

I found out I can make them flash by attaching an inductor to them and tapping the wires on a "dead" D-cell battery.

Now I'm trying to figure out how to build some sort of joule thief to light them up. 

[CaptDon]

If you tape the wires to the battery you get a dead short through the tiny inductor. Brushing or scraping the wires across the battery will give enough inductive kick to flash the lamp. And what is the goal of this post?

[The Doktor]

I think the goal of the post is to get some advice on building this higher voltage joule thief. And I think the answer is to build it much like the original, but with a higher turns ratio.

[jzx]

You can build the joule thief and add a secondary winding to have a higher voltage. You will need a rectifier for the leds, for this you can use a bridge with 4 1n4148, but the 1n4007 are slow.

[T3sl4co1l]

Yes, the standard circuit will do, plus a high ratio secondary.  You'll want it a bit better optimized for power, as those filaments will be on the order of one watt (or a modest fraction thereof).

Getting say 48V directly off a single BJT blocking oscillator, isn't the easiest thing, or the most friendly to the BJT.  But getting say 5 or 10V off it, then a 5-10x transformer, will light those easily.

For reference, here's a trusty one I made years ago, and still use regularly:



Circuit is the basic blocking oscillator,



with +V = 1.2V (NiMH cell), Rbias = 100R, Cbb = 1uF (there's a 22uF on the supply as well), Rb = 0, and, I think Ns/Np was a little less than 1..?!  You can almost count it in the photo, but mind the primary is 2 x 28AWG bifilar.  LED is simply connected Output to GND.  Switch is off/1k/100R just for a little dimming option.  BJT is PBSS303NX, a modern low-Vce(sat) type of excellent ratings, well suited to low voltage, high efficiency converters like this: with a rectified output (so, including diode loss), I measured ~60% efficiency I think, at around 3V 1W output, so it's not doing too bad at all.

The major mistake that the common circuit makes, is using Rb instead of Rbias, and no Cbb.  This ensures poor base drive during switching, and high switching losses.  It's a kitschy thing anyway, efficiency hardly matters, it blinks a light, so the poor circuit persists online (thus, it takes on a life of its own -- it is a meme as such).

Doing basically the same thing, but say 20t primary, 15t feedback, and 100t output, should light that filament quite nicely.

Tim