Lasersaber strikes again. A joule thief king ?

[d3x0r]

My first coil had a short between the foil windings and the case; it's also the only one that worked alright.  Wound with 24 ga widings, and aluminum sheet for foil.
I made another coil without foil, with 32ga wire, it barely works.
I made another coil, this time with thin copper foil tape for the foil windings, 24ga wire; slightly higher drive coil inductance; and it doesn't work like at all... if I use my fingers and short from positive power to the high side of the LEDs, then I can get it to oscillate, but it has very very short runtime.


I tried to insulate the edges in the first coil, but then it's too thick to fit in a pot core... so I have nothing at this point

[conradelektro]

Please see also these two postings (circuits and scope shots):

http://www.overunity.com/14591/lasersaber-strikes-again-a-joule-thief-king/msg403065/#msg403065
http://www.overunity.com/14591/lasersaber-strikes-again-a-joule-thief-king/msg403087/#msg403087

I was mistaken, the transistor switches as long as the base stays positive (during the positive half wave of the signal from the secondary winding). See the detailed scope shots. The scope shots become less and less detailed to show the small and the big picture.

I can pull the core of the coil apart a bit to make it less inductive. Less inductance makes the amplitude of the signal at the base (from the secondary winding) smaller.

If the inductance is highest (the signal on the base has a high amplitude) the primary winding is even pulled below zero (there seems to be some negative current leakage from the base to the emitter), which then causes a rather long period where the winding goes above supply Voltage and feeds back current into the cap.

If the inductance is lower (the signal on the base has a lower amplitude) the primary winding is not pulled below zero (no leakage of negative current from base to emitter), which causes a shorter  period where the winding goes above supply Voltage and feeds back current into the cap.

So, there is no mystery switching (the transistor switches as long as the base is positive), but there is a mystery leakage of negative current into the emitter (and consequently into the primary winding) if the signal on the base has a high enough negative component.

It is just pure luck that my coil was just right enough to show this phaenomen. If I reduce the impedance of the coil slightly the phaenomen goes away.


Note, the base has to go to -15V in order to cause the phaenomen. If the base only goes to -12V the phaenomen stops.

What does it mean for my coil:

The ratio of 1 : 3 (primary / secondary) should go up a bit, may be 1 : 4. And it needs more turns (on both windings, which means thinner wire) to have a bit more current specially through the secondary winding (which hopefully will leak from base to emitter while the base is very negative).

There will be a limit of "possible leakage of negative current from the base to the emitter" when the transistor will break or at least will block.

So, a bit of negative current from the secondary winding leaks into the emitter pulling the primary below negative rail, which then causes a higher back EMF of the primary which will leak from the primary (while it is above positive rail) back into the cap.

LaserSaber has found a feed back mechanism:

The negative half wave of the secondary (which normally is lost as heat in the transistor or in a safety diode) can leak back into the cap (via emitter and back EMF of the primary). This only works with certain transistors.

There is also the extremely short pulse which reduces power draw. The effect of copper strips could be to make the pulse shorter (but I have not yet seen that).

Please look a the scope shots an tell me what you see.

Greetings, Conrad

[conradelektro]

Please also see:
http://www.overunity.com/14591/lasersaber-strikes-again-a-joule-thief-king/msg403109/#msg403109 (more scope shots)
http://www.overunity.com/14591/lasersaber-strikes-again-a-joule-thief-king/msg403065/#msg403065 (circuit)
http://www.overunity.com/14591/lasersaber-strikes-again-a-joule-thief-king/msg403087/#msg403087 (circuit)

I tried to see "power draw" and "power feedback" to the 1000 µF cap. Therefore I measure with AC coupling over the cap (only the AC component is visible, the 9 V DC Voltage is not visible).

Again the left scope shot is with the high inductance core and the right scope shot is with the low inductance core (cores are pulled apart a few millimeters). The feed back with the high inductance core is higher because I think that in this case the negative current leakage through the transistor from the secondary winding on the base is happening.

I also tried to measure power consumption and as TinselKoala suspected, this is tricky. It seems to work with my Multimeter at 9 V, the result is about 20 µA. At lower Voltages (the circuit works down to 0.5 Volt) the measurements become unreliable (varies between measurements from 7 µA to 20 µA).

So, I did the calculation by help of time and capacitance.

Q = As = C * U    (Q = charge or Coulomb or Ampere-seconds, C = capacitance in Farat, U = Voltage over the cap)

1000 µF  *   9 V = 0.009 As (charge at start Voltage)
-1000 µF * 0.5 V = 0. 0.0005 As (charge at stop Voltage)
0.0085 As in 420 seconds (7 minutes) --> 0.0085 / 420 = 20 µA on average

4700 µF * 9 V = 0,0423
-4700 µF * 0.5 V =  0.00235
0.04 As in 1800 seconds (30 minutes) --> 0.04 / 1800 = 22 µA on avarage

The caps have some leakage and power draw is about 20 µA at 9 V and goes down when Voltage drops slowly to 0.5 Volt. The calculated 20 µA average power draw is not unrealistic given the 20 µA measurement at 9 Volt.


LaserSaber's circuit: 3000 µF cap

3000 µF  *   9 V = 0.027 As (charge at start Voltage)
-3000 µF * 0.5 V = 0. 0.0015 As (charge at stop Voltage)
0.0255 As in 18000 seconds (5 hours) --> 0.0255 / 18000 = 1.4 µA on average

The cause for the 14 fold difference (20 µA versus 1.4 µA) should be the core and may be the MPSA18 is even better than the 2N1304. (I will get some MPSA18 next week.) His pot core should have a higher inductance than my conventional core and therefore there should be more current on the base of the transistor (which would allow more feedback). It could also be that LaserSabers ceramic capacitor has less leakage than my old low grade electrolytic capacitors.

Greetings, Conrad

[conradelektro]

Wrong post, deleted.

[Vortex1]

You can configure a standard blocking oscillator circuit to perform in a similar manner and with nearly identical waveforms by lightly coupling the base of the transistor to the second winding with about 100 to 200 pF coupling capacitance. A high value resistor is also needed to bias up the transistor.

What you get is an extremely narrow pulse of around 400uA into the LED, but the average is closer to 4uA, the apparent brightness of the LED being being equal to one being operated at a steady current of around 10 to 100 uA.

So we are talking about a duty cycle of around 1% or less.

I have simulated this and the LS circuit in LTSpice where exact measurements can be easily had.

A fresh PP3 will put out about 10 volts with between 150 to 300 milliohms internal impedance and if touched to the 3500 for 4 milliseconds, the cap will charge to 98% of the applied voltage. Then it should have about 175 mJ available.

If dissipating 10uW with 175,000uJ, 4.86 hours are possible but this is best case not real world.

A good question to ask: Of what utility is such a device if you can hardly see it and it is unusable as a torch? I already have a clock that produces micropower ticks and runs for better than a year on a tiny battery.