Joule Thief

[SkyWatcher123]

Hi nick, nice to hear from you again, thanks for sharing the information.
Happened to have a solar garden light here and it has a nice light sensor on top, so placed it at same spot on outside of glass bulb covering for each bulb test.
With meter set at 20kohm, the 6 watt cree ungutted placed in socket shows .4 on meter.
With light sensor placed on gutted cree bulb, meter showed 1.37 on meter.
If we divide 1.37 by .4 we get, 3.425 ratio.
And if we multiply 3.425 times the gutted bulb input of 1.1 watts, we get 3.77 watts.
If this is any indicator, it may mean that if the efficiency stayed the same, we would only need 3.77 watts to achieve the 1.37 light sensor meter reading, maybe, hehe.
If a 1.5 kohm base resistor is used on JT circuit and using the secondary capacitor, it cuts exactly in half the current draw compared to without secondary capacitor.
So my guess is, is that the ratio would be almost exactly 50 percent or ratio of 3 on light meter reading, instead of the 3.425 that it shows using the 1kohm resistor for transistor base.
I was laying down earlier trying to figure out what the 100 nanofarad capacitor in series with led bulb is actually doing to lower input by almost half and raise light output by large margin as well.
My only thought at the moment, is that the capacitor may be somehow charging quickly on one phase, then when secondary coil collapses, the capacitor is acting like a battery and adding its voltage/current to the collapsing pulse through leds.
Making them much brighter and in turn, reflecting less load back to primary input coil, which may explain the almost cut in half, current draw.
If anyone has any other ideas on how the capacitor may be interacting with this circuit, much appreciated.
Keep us updated on your garden solar panel tests, sounds good.
peace love light
tyson

[NickZ]

  "My only thought at the moment, is that the capacitor may be somehow charging quickly on one phase, then when secondary coil collapses, the capacitor is acting like a battery and adding its voltage/current to the collapsing pulse through leds."

  That very well may be.  It may also be similar to the reason that Slayer's Exciter type circuits are so economical on battery draws. As he uses a 1meg resistor, and 4148 diode to achieve his results. Hard to beat 30 mA to brightly light the 2.5 watt gutted Led bulb.

  All this would not matter much if using only the grid as the input source, but that all changes when using batteries.

  What is the cap value that you are using?  I'm always burning up those little green caps on both the AC led bulb circuitry, as well as on the CFLs, when connecting the bulbs to my Ringer 3.0 circuit. As it gives over 1000- 1500v output from a 12v battery.

Maybe a trim pot would also help to further adjust the voltage to find the sweet spot.

  Here's a couple pics of some of my circuits. As you know I always liked the Cfls, and I have so many of them. The big tall bulbs are 65 watts Cfl, along with 60 watt incandescent bulb both running on the same Ringer circuit, using only a single TIP 3055 transistor, or also the 2n2222. I love simple circuits.
  Lately I've been working on the Tariel Kapanadze or Akula 0083 self runner type circuits.
Although I've had no luck getting them to self run, YET.
Hopefully that will all change, soon.

[crowclaw]

The capacitor is acting as a DC blocking element thus isolating the adc current path directly to the LED. Remember the JT wavefoms tend to be a little unstable for want of a better term, displaying a rather unusual complex pulse type waveform with AC properties. Some degree of experimentation will indeed reveal a sweet spot as mentioned and the capacitor value will further aid the best output coupling to the LED. The LED is also it's own DC retifier of course. The JT waveforms can be very complex in nature and also rich in harmonic content, the very nature makes repeat circuit design results vary considerably. tunning each circuit will attain the best possible results so you are well on your way here to produce an effecient  and economical circuit design. Good work...Crow

[SkyWatcher123]

Hi nick, awesome lights you have there, thanks for tips and good luck on the self runner.
Hi crow, thanks for the explanation, maybe another way of saying a similar thing, i tend to need a simpler visualization method to comprehend how something works, words are not sufficient for me, so i try and visualize it in mind.
So, if we base the efficiency off the light sensor comparative numbers, this light is around 120 lumens per watt.
Tried 12 volts, though this coil/core seems to like around 4 volts input for best efficiency so far.
Lowered the base resistor to 750 ohms and draws around 360 milliamps at 3.8 volts input and numbers work out to similar, 120 lumens per watt.
At this input of 1.36 watts, based on light sensor numbers, led bulb is probably around 164 lumens.
Nice useful light for any purpose.
peace love light
tyson

[crowclaw]

The higher 12v could be driving the circuit into saturation, do you have a scope to view the waveforms.