2 component negative feedback oscillator and voltage "multiplier"

Maybe it's not a transistor.

--

John Larkin Highland Technology, Inc

The cork popped merrily, and Lord Peter rose to his feet. "Bunter", he said, "I give you a toast. The triumph of Instinct over Reason"

Standard NPN, i think. Things seem to not oscillate in Spice.

That circuit would never oscillate. If it did, it would pump energy

Might be something like this:

but not that exact circuit.

The original circuit oscillates if you put 100k from supply to base. But only puts 20uA RMS through the LED used

The LED is back-biased.

With lossless L this is about 95% efficient:

Version 4 SHEET 1 5945676 13421556 WIRE 2160 -592 1680 -592 WIRE 2160 -512 2160 -592 WIRE 1680 -480 1680 -592 WIRE 1904 -384 1904 -496 WIRE 1680 -368 1680 -400 WIRE 2160 -368 2160 -432 WIRE 2336 -368 2160 -368 WIRE 2160 -320 2160 -368 WIRE 1904 -224 1904 -304 WIRE 1968 -224 1904 -224 WIRE 2080 -224 2032 -224 WIRE 2336 -208 2336 -368 WIRE 2160 -160 2160 -240 WIRE 1904 -112 1904 -224 WIRE 1952 -112 1904 -112 WIRE 2080 -112 2080 -224 WIRE 2080 -112 2032 -112 WIRE 2096 -112 2080 -112 WIRE 2160 0 2160 -64 WIRE 2336 0 2336 -144 FLAG 1680 -368 0 FLAG 2160 0 0 FLAG 2336 0 0 FLAG 1904 -496 0 SYMBOL voltage 1680 -496 R0 WINDOW 3 -394 13 Left 2 WINDOW 123 48 57 Left 2 WINDOW 39 0 0 Left 2 WINDOW 0 72 26 Left 2 SYMATTR Value PULSE(0 1.5 10n 1n 1n 1 2 1) SYMATTR Value2 AC 1.5 SYMATTR InstName V1 SYMBOL LED 2320 -208 R0 SYMATTR InstName D1 SYMATTR Value NSCW100 SYMATTR Description Diode SYMATTR Type diode SYMBOL ind2 2144 -528 R0 SYMATTR InstName L2

SYMBOL ind2 1920 -288 R180 WINDOW 0 36 80 Left 2 WINDOW 3 36 40 Left 2 SYMATTR InstName L1

SYMBOL npn 2096 -160 R0 SYMATTR InstName Q1 SYMATTR Value 2N3904 SYMBOL ind2 2144 -336 R0 SYMATTR InstName L3

SYMBOL res 2048 -128 R90 WINDOW 0 0 56 VBottom 2 WINDOW 3 37 56 VTop 2 SYMATTR InstName R1 SYMATTR Value 100 SYMBOL cap 2032 -240 R90 WINDOW 0 0 32 VBottom 2 WINDOW 3 32 32 VTop 2 SYMATTR InstName C1 SYMATTR Value 470p TEXT 1288 -448 Left 2 !.tran 0 5 0 1u startup uic TEXT 1936 -336 Left 2 !K L1 L2 L3 0.99

It must be sucking the light _in_!

That needs a 3-winding transformer and doesn't start reliably. It has a do-nothing state.

All one-transistor positive feedback oscillators have a "do nothing' state my dude...

I don't know any 1 winding or 2 winding one-transistor oscillators that have very good efficiency driving an LED, the current ramp needs to be well-controlled and the transistor driven hard on and off. A sine oscillator won't do that, a naive blocking oscillator doesn't do it either and taking the LED output from the transistor collector s**ts it up. Rectifying and filtering would improve it but then you burn power in the rectifier.

Not so. Any decently-designed oscillator will always oscillate when power is present, and has no hang state. Quench it and it starts back up.

A good blocking oscillator puts most of the cycle energy into the LED. Neither ideal inductors nor switches dissipate power, and it's not hard to get close to ideal. And sensible blocking oscillators always oscillate.

But there are cheap chips that power an LED from one cell and only need one inductor.

Probably not. The solar lights & cheesy flashlights I've dissected use little 3 or 4-pin boost converters, one inductor, and not much else.

QX5252F is one such beast, and the S-8351 series is another.

Cheers, James Arthur

I don't seem unijunction transistors much these days. I would think they'd do pretty well for this sort of circuit. Sort of a single transistor 555 timer.

I have to say no one has been able to explain to me why they disappeared.

The list that comes up when you search "high efficiency Joule thief" is long, but I've found that actual numbers whether simulated or experimental are pretty thin on the ground. In a real circuit with small-ish value of inductance to get the oscillation frequency up the core is always going to be going into saturation on every cycle I dunno how that could be very efficient.

This one works great it can do a string of 7 LEDs with PWM dimming:

It can even be pressed into service as a CM flyback controller for well-behaved secondary loads

The ZXSC380 circuit has only three parts, a sot-23 "transistor", an inductor and LED. The same three parts in the .asc circuit are arraigned differently, but as you pointed out, the LED is back biased, and cannot work as drawn. So, the reverse engineering on the Dollar Tree LED Flashlight must be in error. The ZXSC380 likely costs pennies, in high volumes.

I took some measurements and note that indeed you are correct, except the circuit on page 5 of the ZXSC380 is the one i cited. Thanks for the part type and data sheet.

What about the joule thief? Seems to run until the cell is damn near dead.

• ...very much like the one you cited from Zetex.

I messed on the trace-out to schematic. Sorry.

Ya, it can boost a very low voltage that doesn't automatically imply its very efficient at transferring energy for any given combination of components and supply voltage. The very simple ones with one coupled inductor and a resistor I don't know how they would achieve very high efficiency (>90%) AFAIK they operate by allowing the transformer core to saturate that's not efficient.

Could probably cut-and-try component values to get it in the 80s but performance would depend heavily on the properties of the inductor and LED.

There are lots of variations like:

and:

Some may be inherently better than others.