It is a really nice move by Lasersaber to publish his experiments in all detail.

I will rewind my transformer to the 1:2 ratio tomorrow or on Tuesday (enameled copper wire 0.4 mm diameter, may find some Litz wire).

The "magnet shaker" power supply is neat. And it works because the circuit needs very little power.

I wonder why the MPSA18 does not burn through at 70 Volt (Lasersaber mentions a 70 Volt charge of the electrolytic capacitor in the video). The VCEO of the MPSA18 transistor is 45 Volt absolute maximum? But there might be a huge Voltage drop over the DC resistance of the 200 turn winding?

Notice that the transformer steps down to the LEDs, so, the circuit needs higher Voltage in the electrolytic cap (may be at least 12 Volt)?

70 Volt at the primary (driven by the collector/emitter of the MPSA18) will lead only 35 Volt to the base of the transistor (secondary step down 2:1). And the Voltage drop over the primary (200 turns, may be 20 to 30 Ohm) will help the transistor to survive. Will have to measure that.

Greetings, Conrad

it would be a stepdown; but it's still got a spike because of the sharp turnoff... it's not just 'volts in = volts out' but 'how fast current changes in  = how fast current changes out'
that is an interesting note though; I should reverse my coil.
the LEDs are also driven low.... it's a low voltage spike that hits the base ...

http://www.instructables.com/id/Electricity-Generating-Footwear/step12/Installing-The-Insole/
Another way to charge the caps using piezo electric disc .. can integrate   this as well
totoalas
 

Hi folks, Hi d3x0r, thanks for the information, i kinda figured that.
Though i wired one with 200 turn as first coil 30 awg. and outer coil 100 turns 30 awg.
Got lasers v3 to work, charged a 10 volt 3300uf electrolytic cap and it lights led for about 10 seconds and sometimes sputters back lighted again. 2n3906 pnp transistor was used.
Though i may have used to small of an e-core.
I get the feeling one needs the core he is using to see similar results.

So after that, today i decided i wanted another non-modified led bulb lamp that was a bit brighter than the child's led bulb lamp i built.
Well i got it brighter with another ferrite flyback core, using joule thief style wiring with secondary coil.
Though it was a little less efficient than the child's lamp, so i decided to wire it joule ringer style and it was much brighter and higher frequency.
Thing is, the efficiency dropped to around 105 lumens per watt for a 75 lumens per watt rated bulb.
Then i put in the older version of this 6 watt ecosmart led bulb and it was less bright, but the efficiency went very high, to around 167 lumens per watt at 1.22 watt input.
Again same bulb, just a previous version, that must have different circuitry inside and it is built differently, the newer one seems much cheaper built, though solidly built.
This joule ringer is giving 2.2 times more lumens per watt from this bulb, compared to wall powered.
Granted it's only giving 207 lumens at that efficiency, though again, i can imagine what that g7 filament led bulb would do at its stock 138 lumens per watt and not to mention that bulbs power factor is around .5 or so, so even more room for better efficiency.
I watched that video by OUbrads and it just proves how we can get very usable light for little input.
Let me crunch some numbers here.
If we used that g7 vintage led bulb rated at 138 lumens per watt and used an efficient ferrite flyback or equivalent core, maybe even like laser is using now and since i'm getting 2.2 times more lumens per watt than stock rating, we may be able to get 303 lumens per watt with this type of setup.
I do observe that these high lumen per watt results, only seem to occur at lower brightness levels.
Though that is not a problem really, because if using the one bulb is not providing the needed light output, then we just use 2 or 3 to do the job, while maintaining the same high lumen per watt setup.
I have a gutted 6 watt cree led bulb here and i will test to see how well it runs with this joule ringer setup and i will probably gutt one of these newer 6 watt ecosmart bulbs to see if it can run more efficiently, though it doesn't look like it was meant to be taken apart. (el cheapo)
peace love light
 

2n3906 pnp transistor was used.

Though i may have used to small of an e-core.
I get the feeling one needs the core he is using to see similar results.

My experiments with LaserSabers very low power circuit show without any doubt, that the transistor is the most crucial element.

My best result was with a MPS18 transistor (which I bought from MOUSER http://at.mouser.com/Search/Refine.aspx?Keyword=MPSA18, but you also find them on ebay). All other transistors I tested cause a power draw which is at least 10 times higher.

Look at my post http://www.overunity.com/14591/lasersaber-strikes-again-a-joule-thief-king/msg405416/#msg405416

The MPSA18 transistor (in this circuit) has a very short "on time" of about 600 nano seconds, much shorter than with any other transistor I tried.

The type of core, the wire and the exact number of turns of the two windings (ratio primary/secondary) have much less influence on the result. The right coil can still improve the result, but not by more than by a factor of two or three.

The brightness of the LED depends much on the LED type.

So, get some MPSA18 transistors and you are in the game (in case you are interested). There might be other transistors with a similar behaviour, but one has yet to find them.

I also suspect that the MPSA18 transistor allows some current feedback to the power source (electrolytic cap). Negative current from the secondary leaks from the base into the emitter and draws the primary below ground which then causes a higher "back EMF" which feeds current back into the cap. But this is just a wild theory to explain the 10 fold power requirement drop with a MPSA18. My expertise is not good enough to clarify that.

Greetings, Conrad

Well, i don't have any small transistors here with the high gain of the mpsa18.
Though i wonder if using two 2n3906 in a darlington pair might work, though that will slow down the rise and fall time.
Will test it out later today.
peace love light
 

I just had a total failure with the new windings on my transformer core, see the attached photo for the specs.

I put this coil into LaserSabers latest circuit http://laserhacker.com/?p=420 and it did not work without a resistor between base and positive rail. And with the resistor (1M to 20M) it did not do the very short ringing bursts but the normal JouleThief mode at 14 KHz. The LED was fairly bright but the power consumption was 200 µA to 600 µA depending on supply Voltage.

It seems to be more difficult than I thought to come up with the right transformer.

Greetings, Conrad