Joule Thief

[MileHigh]

Bill:

The output from the JT's will not be 350 volts.  The JT voltage output will directly follow the supercap voltage.  So as the supercap voltage slowly increases so will the JT output voltage.  That means there is a risk for over-voltaging the supercap only when the supercap is fully charged.

Note I am talking abut JT circuits, not your flash circuits.  I can't comment on the flash circuit without seeing the circuit.  There is a decent chance that your flash circuit is similar in operation to a JT circuit but that's all that I can say.

What I am trying to figure out with this current project is how/why the light gets brighter when I have been told by several folks that you can't put ac circuits in parallel and get any increased output?

In a standard JT circuit the LED flashes when the coil discharges, and we know the LED is a diode.  If you put two JT circuits in parallel driving an LED light bulb you can imagine something like this:  You swap out the original LEDs in each standard JT circuit for a diode.  Then the outputs of those two diodes are bridged together and they drive the LED light bulb.  In a scenario like this when one JT is discharging, the diode on the other JT becomes reverse-biased and that protects the other JT from being affected by the discharge of the first JT.  So you end up with a circuit that funnels the discharges from both JT's through the LED light bulb.  It looks like something like that is happening in your clip but I can't be 100% certain.

The same principle of funneling the discharges together could apply to charging the supercapacitor.

MileHigh

[MileHigh]

Shock and awe, somebody has done something big with Jule Thief technology!

http://www.youtube.com/watch?v=kiI77x6VBMc

[Lakes]

Nice and geeky, but $120 is way over priced for a flash light!

[MileHigh]

Just looking at the video I can take a decent shot at reverse engineering the flashlight in my head.  The micro inside the flashlight has programmable timers that can output a pulse train with a programmable frequency and duty cycle.  So the software just has to poke the registers to change the nature of the pulse train.  The precision for changing the low and high pulse time is very fine, most likely the crystal clock for the micro.

The output pulse switches on a transistor that starts to energize an inductor.  When the transistor switches off the inductor discharges through the LED array.  Since inductors discharge pulses of current, you can precisely control the average current that flows through the LED array.  The average current determines the brightness.  You don't really need to measure the voltage, that takes care of itself.

The "secret" here is that you don't use any resistors to control the LED intensity.  No resistors means no wasted energy burning off in the resistor.  However, even the coil has a resistance.  So the challenge is to energise the inductor so that it transfers power into the LED load with the minimum of lost energy burnt off in the resistance of the coil wire.  So you assume that the designers of the LED flashlight tried to optimize the pulse timing for maximum power transfer efficiency.

You can probably find an LED flashlight in a Dollar store that has two intensities.  So you can imagine that something similar is going on in the cheap LED flashlight that might only cost $3.  So it would be an interesting project to reverse engineer one of those and it's cheapo shmeapo.

Then if you figured out how it works you could try hacking into it and using a larger coil and battery, as an example.  Or, make the same circuit but using your own discrete components.

[Farmhand]

The flashlights are already in use and for quite a reasonable price, I got a surefire LED flashlight with 1000 Lumens High-Low and 10 Hz strobe settings (the strobe kinda makes me ill to look at). It uses a 7.4 volt LED and works from a 3 volt battery, they use the battery way down because they use an oscillator and and inductor. Calling it a "joule thief" without seeing the schematic is a bit rich. Any kind of oscillator could produce the same coil collapse to get higher than supply voltages to run the LED on, it is Boost converter topology in my opinion, then a bit more, Micro processors may be used to produce the signal. Hardly a JT. And in use for quite some time now.

Mine is a WF501b which mounts well and is bright as (blinding) and I intend to get a Cree T6 2000 lumen like listed below.

All the listing don't brag about the booster circuit but they do use it. An entire set with battery and charger for 27 dollars.
http://www.ebay.com.au/itm/2000Lm-CREE-XM-L-T6-LED-Zoomable-Flashlight-Torch-Lamp-18650-Charger-/261215609941?pt=AU_Sport_Camping_Hiking_Lamps_Torches&hash=item3cd1a9e455&_uhb=1

http://www.ebay.com.au/itm/CREE-XML-T6-White-LED-Zoomable-Torch-Flashlight-Lamp-Light-1600LM-Camping-/190923317086?pt=AU_Sport_Camping_Hiking_Lamps_Torches&hash=item2c73ea9f5e&_uhb=1

CREE XML T6 LED Torch Flashlight
Brand new and high quality
Materials: aluminum alloy
Applied high efficient booster circuit, can utilize the batteries in the largest extent
The beam can be focused.
Dimension: about 14*4*4CM
Housing Color: black
Flux luminous: about 1600LM
Lighting color: white
1 x CREE XM-L T6 LED
Lighting modes: 100% On, 75% On, 50%On, Strobe, Quick flash
Powered by 1 x 18650 battery
Zoomable
++ Package included ++
1 x Torch
1 x Battery holder
Packed in paper box

Is there anything new in the world, not much. 
To think that no one was using the boost converter topology to make better use of things is just ignorant.

Strings of LED lights also use a boost circuit to get the voltage required to run a string of LEDs from a 1.5 volt battery.

It's no big deal, nor is it new or fantastic or over 100% efficient.

For some reason a lot of posters think the rest of the world has their head in the sand. Obviously not true.
If it is useful it will get used.

Here is a picaxe controlled boost converter capable of over 50 Watts output fairly efficiently at up to and over 24 volts I made, it can do all kinds of jobs and is programmable to control power and maximum voltage. No Joule Thief required. The trick to use the battery right down is to use the circuit to boost it's own supply voltage, which I did not need to do. But there is even a chip sold to do it, can't recall the part no. I can try to find it, it was shown in a thread by SeaMonkey, it's a very ingenious chip.
https://www.youtube.com/watch?v=eLnNgfaha10

I've shared the schematic and code for my booster before a couple of times.

Cheers