Joule Thief

[gyulasun]

...

---------     What also puzzles me is the frequency.   -------------

---------     15 KHz seems to be rather low. Is this the impedance of the Toroid? How could I get a higher frequency?    -------------

...

Hi Conrad,

My understanding is that the oscillating frequency is determined mainly by the coil inductance and the base resistance (I mean the 1 kOhm).
You have a pretty high permeability toroidal core with its A_L value of 6440nH/N², your coils inductance is L=N²*A_L=9*9*6.44=522uH (microhenry) for each one coil. This high inductance for the 9 turns means a high permeability core.
(Also, if you clip your scope probe directly to the base of the transistor (ground is on the emitter) and watch also the pulse frequency on the collector on the other scope channel, you will find the oscillation frequency slightly shifts for the probe 16-20pF capacitance too.) 

So if you use a core with less A_L value (lower permeability) or you reduce the 9 turns to say 4 or 5, your frequency will go up.
Also, as Bill wrote try to change the 1 kOhm base resistance too.

rgds,  Gyula

[gyulasun]

Hi Jeanna and Bill,

I understand you work by practice and seeing the color of a toroidal core you can pretty much estimate its possible behavior in the joule thief.
While normally I agree with such practice I still wrote to Conrad he has a high permeability core...
He included the induction factor for his core, the A_L value, and it is a pretty high number. This means also high permeability because normally for a core with OD=16mm ID=9.6 and h=6.3 the nine turns usually does not give 522uH inductance.

Would like to show you a very handy program that includes helps on getting permeability and/or A_L from a know number of turns and the measured inductance for any unknown toroidal core. (you need the mechanical dimensions for the core too).

Here is the link:  http://www.dl5swb.de/html/mini_ring_core_calculator.htm

Using this program, the permeability comes out as 10006 (ten thousand). In your practice with cores so far does this count as a low perm core?  I think this to be a high value.

rgds,  Gyula

[gyulasun]

Hi Conrad and All,

I apologize if the following link has already been referred to here but I found it very practical and informative (and scientific) on joule thiefs:

http://watsonseblog.blogspot.com/2009/02/joule-thief-high-efficiency-01.html 

Gyula

[Pirate88179]

Gyula:

I had not seen that before.  Thanks for posting it.  The only thing I do not understand is that he said he wanted low permeability for his toroids and yet, I believe, most of us have had our best results with high perm.

I do not pretend to understand everything about the JT circuit, far from it.  I just can't figure out why he wants low perm?  Can you or anyone explain this?

Thanks,

Bill

[jeanna]

Would like to show you a very handy program ...
Here is the link:  http://www.dl5swb.de/html/mini_ring_core_calculator.htm

Using this program, the permeability comes out as 10006 (ten thousand). In your practice with cores so far does this count as a low perm core?  I think this to be a high value.

rgds,  Gyula

@Gyula,
Thank you.

I use linux and I have never used wine, but I do have an inductance meter which helps me.
I am sure this calculator will help many others.

About the permeability, Yes, 10,000 is high perm.
I have a 2 inch with 11,000u and 14,000u and the 14,000 is needed to get the same results as the 3 1/3 inch core with 11,000u.
I am sure Conrad's core is fine, too, but the blue covering sometimes makes it a choke and very low u.

@Conrad,
I actually think there is nothing wrong.
It is likely you need to 'tune' the turns.

I do not have the kind of scope you have, so what I have to say may not apply.
If I move to higher resolution I sometimes see  a more detailed waveform and see that where I thought the frequency was 10Khz, it turns out to be 60Khz or even higher.
Maybe, something like that is happening there??

Please make a 10 turn pick-up winding and tell me what the volts and frequency are on the 10 turns?

thank you,

jeanna