falexandru:
Here it is the pic:
- background: the circuit which works ok
- grey ferrite ring - which does not work (wired)
- blue ferrite ring which does not work
The length of the wires in blue and grey cases were 80 cm and then 40 cm.
The length of the wires in the white ferrite ring 30 cm.
You might have better luck with magnet wire, and with smaller toroids [so the windings are closer together, and more compact].
Also, [referring to another post] on those motors/generators you said "don't work", did you attach a meter to them to see if they are outputting anything [and whether your getting AC or DC]?
And whenever you say "it didn't work", I think, what does that mean. Why do I think that? Because, I don't know for sure if your aware of the concept of "a multitude of variables". It would be more informative if, instead of saying "doesn't work", you told us what the "failure mode" is. Even if it seems obvious or redundant. For instance, when you were trying the various cores [toroids], did you just hook it up, then decide it wasn't working if the LED didn't glow, or did you try reversing the wires on the primary [or secondary] to give it another try...then, even after that, did you check to make sure the power source was adequate/functioning/consistent -- i.e. with the goal of limiting variables as much as possible. And, in case you don't know what I mean by variables: The power source is a variable [in fact power source voltage is a variable, power source consistency is a variable, power source current capability is a variable], proper inductor hookup is a variable, LED polarity is a variable [i.e. is the LED connected properly]. If any if these variables are not consistent -- if more than one fail at a time, and if those failures are not consistent, then you can easily be misled.
One example of limiting variables is to use a Bench Power Supply when doing something like, testing Toroid Cores to see if they "work" or not. A bench power supply is far more likely to be consistent, than, say, a motor being used as a generator. Test the toroids with the bench supply, with parameters set to as closely model the real world power source [whatever that winds up being], as possible, then, when you have a set of cores that worked with the Bench Supply, you can have more confidence that if, when you try them with the real world power source, when they fail, it has more to do with the real world source.
Another example: when trying different power sources, use the same Joule Thief setup, each time.
If you try to "save time" by testing a bunch of things at once, when things don't work, it's hard to be sure what the source of the failure is.