Joule Thief

[Pirate88179]

Yes, series.
Like resistors: inductors in series simply add. Inductors in parallel do the inverse thing: L_final = 1/(1/L₁ + 1/L₂ + ....)

Thanks TK. 

[Farmhand]

Yeah less switching less switching losses, diodes also rob us at low voltages. I've ordered some some small Schottky diodes 1N5819's.

I'll look into booster IC's as well.

I'm thinking of trying to set up this oscillator as a tiny boost converter and use a 1 Meg pot to form a voltage divider with the 100k resistor, by applying a positive voltage from the output cap charged by the diode output through the (adjusted) 1 Meg pot to the junction of the diode anode and the 100 k resistor, then when the cap reaches the desired voltage it should stop the oscillator, I think.

This would allow a voltage of about 4 to 5 volts for logic circuitry to run from, preventing over voltage possibility. We could charge 0.22 Farads to 4.3 volts and have 2 joules to dump into a battery or load every so often.

Or a higher voltage for other things even.

If we use the circuit to switch a transformer to charge a cap and reference the cap 0v to circuit ground we can also use that method maybe.

..

P.S. It works but the divided voltage from the output needs to be applied to the PNP base directly.

Blue trace is output capacitor voltage, Yellow is the collector of the NPN.

[Vortex1]

Switching losses from the two transistor regenerative switch can be minimized by :

a) adding some resistance between the collector and base of the pair (drive path)

or

b) configure the pnp as a current source, limiting turn on current that is pure loss

and with high gain transistors, (super beta) this loss can be further minimized.

I was wondering what function you think the two 1N4148 diodes on the npn base to ground may perform.

A diode on the pnp base to V+ would definitely be a good idea.

Using a 10 meg pot on the base to ground of the pnp I get very wide range control from pulses in the order of a minimum of 1 sec PRR to as high as would give good brightness, but for safety, limiting to a few tens of kHz.

Nice thing about this circuit is it automatically adjusts duty cycle as voltage falls. The elementary circuit I posted worked down to about 0.75 volt.

Perhaps you could explain the expected function of your components in the feedback path.

I think overall this is a very worthy circuit that needs further development in a separate thread.

I am not at all averse to using an extra transistor or resistor here or there if it leads to improved overall performance. Transistors and resistors are inexpensive compared to coupled inductors.

I like the simplicity of the single off the shelf inductor.

Also I have played with a bit of magnetic bias of the inductor (with a small permanent magnet) Properly placed it allows tuning of the inductor properties to get a very linear current ramp both on inductor charge and discharge. This could also be done with a bias winding, but in either case, the bias eliminates the need for a gap at higher duty cycles.

Kind Regards
Vortex1 (ION)

[Dave45]

any advice on the gates

[Farmhand]

Yes it's a good circuit, Wanna start the thread ?

The diodes between the npn base and ground are intended so that the npn base can go negative, although it seems to make no difference in this circuit. I don't use bipolar transistors much, mosfets usually work best for me for on/off switches, I've never experimented much with bipolar transistor circuits, except mainly for triggering a circuit or for driving or turning off mosfets.

The feedback circuit did work to stop the oscillations when the cap reached a given value but it also affected the oscillator frequency in odd ways, so that's out. I have another plan, but it would be a separate secondary circuit, lets say an (output control circuit)

It's good to keep part count low, but not if using another part means better efficiency, especially with very low voltages.

The circuit I posted works without the 1 nF capacitor across the diode at the npn collector and without the diodes from ground to npn base, and it works well, in that configuration it's the basic "Stingo" circuit that Sucahyo used at EF, I think he discovered it while experimenting. For low voltage use a better design I think is needed. Time to try something slightly different.

A simple darlington pair (npn) works to oscillate a coil too, in a way that connecting a wire from the Earth (stake) to the base of the pair starts it off, depending on input voltage. Getting the coil switch to turn off quicker will help me.

I want an efficient little oscillator to work from 1 volt or less and low current draw so as to harvest energy at low potential levels and boost it into high capacity at higher voltage, then store it for use in common batteries or just use it for night time lights.

I've got a removable core coil I can adjust from 800 uH air cored to over 4.5 mH big steel core and t measures 3.2 mH with a neat fitting iron powder core. So I can use it as a test coil.

I like the idea of being able to oscillate the primary of a regular transformer (scavenged) and use the secondary as output, but with the flyback clamped in some way so the load can be disconnected intermittently.

I was also wondering about making use of the low voltage output of a homo-polar machine or whatever they be called, but that would require some high current handling, an initial boosted voltage could provide the power for a micro controller and mosfet switching in that case.

Cheers