Joule Thief

[SkyWatcher123]

Hi crow, do not have a scope and would guess at least the copper is saturated, not so sure about the ferrite core.
It has very little copper wire on it, maybe a ferrite half or full core from TV will work better with 12 volts.
Then again, this particular ferrite flyback salvaged from TV, was wound for lighting up gutted cfl's previously, whereas leds do not need such high voltage spikes.
Think i will take apart another ferrite TV flyback i have sitting here and rewind that one for 12 volts, which will probably mean, more turns for primary or whatever ratio is best.
Ideally, the perfect coil wind ratios to achieve resonance would be good.
peace love light
tyson

[crowclaw]

Youre doing a fine job experimenting, just keep your eye on the base drive current at higher voltage though... better to use a preset pott for adjustments. Unless you already do so of course. Kind Regards Crow

[SkyWatcher123]

Hi crowclaw, thanks for the kind words and the helpful tips, are you building anything yourself, if so, would love to see anything you have to share.
Well, decided to test a TV ferrite yoke core, which comes with two separate halves.
This ferrite core is working more efficiently than the TV flyback ferrite core.
This yoke also has the same winding turn ratio, 20 to 1, as the flyback core and same turns, secondary has 200 turns per layer, 2 layers for 400 turns 30 awg.
Primary has 40 bifilar turns, 24 awg.
The light sensor ratio is better.
Gutted cree 6 watt led bulb, meter shows 1.3 on meter compared to .4 for AC wall powered bulb.
Which is 3.25 ratio.
Input of .8 watts using same, 1kohm base resistor.
3.25 times .8 equals 2.6 watts.
So if efficiency stays the same, it should only require 2.6 watts to achieve around 450 lumens.
So at this .8 watt input, 4 volt - 200 milliamp, it is outputting around 139 lumens.
So based on the sensor numbers, light output is around 173 lumens per watt, keep in mind, the bulb powered from the line AC, is only 75 lumens per watt.
Not sure if fully winding this ferrite TV yoke will add any efficiency,
all thoughts appreciated.
Think I'll experiment with 4 volt input for awhile, until no more efficiency can be squeezed out.
peace love light
tyson

Edited the lumens  to 173 lumens per watt based on light sensor numbers, let me know if the number calcs seem off.

Also, the same 100 nanofarad non-polarized capacitor is in series with the led bulb at secondary coil, cap is from cfl circuitry, not sure of its voltage rating.

[SkyWatcher123]

http://www.youtube.com/watch?v=a38_EN3U_pc&feature=youtu.be
Hi folks, here is a video of the latest ferrite TV yoke core version.
As said, based on the light sensor numbers, relative to a non-modified cree 6 watt led bulb, the calculations show about 173 lumens per watt.
Gutted led bulb in video is around 140 lumens light output at 800 milliwatts input.
Let me know folks what you think about these results and what you folks think about these lumen numbers.
I know lynxsteam with his lynx joule ringer, with jouleseekers help in testing, achieved 114 lumens per watt.
peace love light
tyson

[crowclaw]

Very encouraging resuslts and good progress. Ideally we need to know the specifications for the LED's used in the Cree Lamps i.e. their maximum permissible drive current for rated luminance output! taking into account that they are working from an AC line voltage and hence using a converter circuit to hold a stable operating current. These lamps have to work at varried temperatures so its important to hold a constant current through the LED's to prevent thermal runaway. The converter circuit also takes a portion of the total current and hence wattage which also applies to the JT drive circuit of course. It is the actual safe region LED current that is important in case they are being over or under driven! I'm not keen on using digital meters around pulse circuits as results can be missleading, an analogue meter would be prefered (see Bill_ Ha Ha) Don't be tempted to measure the LED current through an unmodified Cree bulb due to hazardous voltages!!!
So to sum up... you're making excellent progress with your experiments LED's will give more light output the harder you drive them, but to maintain their life expenctancy have a safe working limit avoiding thermal runaway problems. You could also experiment with very simple buck boost circuits as an alternative, (do a Google search) 
My attentions recently have centred around inductive power transfer and solar energy projects, many of my JT experiments still remain within this thread going way way back.

Kind Regards Crow