Lasersaber strikes again. A joule thief king ?

[conradelektro]

LTSpice screencapture

@Collapsingfield: thank you, I will try to add these caps. The aim is to bring the circuit to oscillation without the resistor between base (of transistor) and positive rail. So far I need a MPSA18 and a very special coil. I hope the caps will do the trick because I hate winding many different coils.

Greetings, Conrad

[TinselKoala]

More on burst oscillations and using the DSO to examine the waveforms:

http://www.youtube.com/watch?v=9FDmkbCbKP0


@conrad: Yes, getting to the point where you can eliminate the resistor is a noble goal. Resistive elements just waste "light" power by I²R joule heating. If you insist on starting out with high voltages, the base current of the transistor needs to be limited somehow. The resistor is one wasteful way. Another way is to "tune" the spike or oscillation frequency so that the transistor is only on for a tiny time, and using no base resistor.  But that method isn't compatible with this circuit because you just can't get the spike frequency high enough or short enough, especially using "cliplead" construction. The RC trick instead of just R for the base drive helps.
Then when you are making really fast short spikes, you encounter the high voltage problem: the transistors we are using can't handle the inductive collapse voltage amplitude.

Try the NPN transistors found in some CFLs: I have one here marked "W6L 13003" that works really well for short, highfrequency spikes and it can take the voltage. (Many CFLs use mosfets instead of the NPNs ... I think I got these NPN transistors from a lower-wattage CFL).

[MladenStijepic]

"W6L 13003" is mje13003, it is actualy whole range of mje13003 to 13009 ,they differ in volts and amps,mje13007 is common in atx power supply.

Had to say this, this is actualy my first post here, greets to all!

[d3x0r]

More on burst oscillations and using the DSO to examine the waveforms:

http://www.youtube.com/watch?v=9FDmkbCbKP0


@conrad: Yes, getting to the point where you can eliminate the resistor is a noble goal. Resistive elements just waste "light" power by I²R joule heating. If you insist on starting out with high voltages, the base current of the transistor needs to be limited somehow. The resistor is one wasteful way. Another way is to "tune" the spike or oscillation frequency so that the transistor is only on for a tiny time, and using no base resistor.  But that method isn't compatible with this circuit because you just can't get the spike frequency high enough or short enough, especially using "cliplead" construction. The RC trick instead of just R for the base drive helps.
Then when you are making really fast short spikes, you encounter the high voltage problem: the transistors we are using can't handle the inductive collapse voltage amplitude.

Try the NPN transistors found in some CFLs: I have one here marked "W6L 13003" that works really well for short, highfrequency spikes and it can take the voltage. (Many CFLs use mosfets instead of the NPNs ... I think I got these NPN transistors from a lower-wattage CFL).

The resistor I find is useful to start it sometimes ... some coils are better than others; the one with the foils starts almost every time (capacitance of foils) ...
I find that putting a diode from the base to the negative (so when the base is extra-low it gates to ground) helps it start up more reliably, and run at slightly higher output


I don't get a sharp high pulse on the gate(err base), it's always a low pulse...
one coil I have is like 2mH and 20mH, and can get out very high voltage on the collector side but the runtime isn't notable.

[conradelektro]

@Tinselkoala: thanks for the scope lesson, I had a vague idea about the "single shot" feature but your video brought it to the point of understanding.

I am not insisting on higher Voltage, the JSR Looper circuit I showed last also works down to about 1.5 Volt (scope shots are similar to what I showed at 9 Volt) but the two LEDs are then just glowing (which can be remedied by a 1M resistor instead of the 20M resistor between base of transistor and positive rail).

A reason for higher Voltages (9V or 12V) might be the step down arrangement in Lasersaber's JSR Looper V3.0 which I could not yet replicate in a consistent way.

Also a "magnet shaker" might produce spikes up to 20 Volt which could be collected in an electrolytic cap to drive the JSR Looper circuit.

But still, it will be easier to produce low Voltage (0.5V to 1.5V) with alternative means, therefore a low Voltage circuit has an advantage.

So far, by all these Joule Thief variants, I was most impressed by the 2SK170 circuit that works down to 50 mV
http://www.overunity.com/13175/25mv-joule-thief-powered-by-peltier-merely-using-our-body-heat-free-energy-247/msg382567/#msg382567 (from magpwr).

Greetings, Conrad