Joule Thief

[resonanceman]

Gary,
I was hoping for an easy answer from one of the brains here, but before that comes take a look at this page:

http://www.doctronics.co.uk/resistor.htm
About half way down the page is a computation of the resistor needed to protect a led from a 9v battery.
One point I caught is that all the components in series will add up to the battery input.

I am not sure what counts and what doesn't on a joule thief, but maybe if you start with a 2k and see then 1k then you can easily get close.

On my extended circuit with all the secondaries, the 33r resistor drops the incoming voltage in half and the current way down.
The 10r drops the incoming voltage to about 2/3 of the battery.

So, you may only need a 100r to take the 12v down enough, but start with 2k for safety then 1k to see if you need that.

Anyway that page goes through what you need to calculate this. It is more than ohms law, but makes use of it.

~~~~~~~~~~~~~~~~~

@All,

I just had a half of a breakthrough.
It looks little, but, I think it puts me into a new place.

In Tesla's drawings, he shows the generator, then a transformer various spark gaps and lights etc, then looking to the right in all his circuits, you can see another transformer.
This then is followed by more spark gaps and more lights etc.

I THINK I have finally been able to get past that second transformer.
I can only light 3 and not 4 leds, and...
there is still a series component to it rather than a coil straight across the ends, but I think it counts.

It behaves funny, too.
It lights up at about the same brightness as the rest of the lights, but when I unplug the battery and let the supercap drain out, these 3 are the last to go.
When this happened before, I was able to greatly extend the lights by adding more inductors.
So, this is what I will be doing.

I am still NOT able to make it work with a straight transformer across the ends without the series group.
If I try that then all the lights go out everywhere.
And the scope is not much help because it often puts out one or both groups of lights when it measures.

So, a little more progress. And 3 more lights which is 1 more than reported the other day using a different set-up.

So 27 leds.

jeanna

Jeanna
Thanks for the link
I will  check  it out as soon as I have time .


I did know  enough   about  voltage drops  to know  that all  the  voltage drops in a circuit equal  the  source voltage .

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~`

It  is  interesting  that   your  new LEDs  stay on longer than the others .
That is  similar to what happens  when  I add   alot of  feedback and   flyback trying to  reach  a self charging state .
I like the effect ......the LEDs  come  on slow  and  go  off slow .    It may take  a several seconds to reach full  brightness .
Partly  because  of this  slow  on and  off  one of the projects I want to  do when I  get the 12 V  part of it perfected  is   a replacement  dome  light  for my car .     I am  going to add  a pot  or  a switch  so I can  light the LEDs at  diffrent intensity's .
I want to  have  a dim  setting that  is  about  equal  to the  original   light .
I  want  a brighter setting  that  is bright enough  to  actually read by .
I find the slow  on and  off to be very easy on the  eyes .
It   is  going to take  some doing to  get  enough  inductors  into   the  space inside the  light .


gary

[resonanceman]

Hi, Garry calculating the exact value is not that easy as you are not dealing with steady state DC voltages.The transistor is switching off and on at the circuits resonant frequency. As already suggested experiment with a pott and connect a series resistor. You are controlling the collector current by adjusting the base current. Start with a high value and steadily reduce the pott resistance until you achieve your target. I don't know your level of technical knowledge but if you guy's are working with 2n3055 transistors it would be helpfull to down load a PDF of this device. The spec should give examples of the required base bias for relative collector output current   (hfe/gain) once you know the base bias required for your project you can then calculate a value for your base resistor. Hope this helps. Merv

Merv

Thanks  for  replying
That is  pretty much  what I have  tried so  far .
I  have got it  to work .......but  then  I end up  burning  up the transistor .
I  had hoped that there  was a simple  way  to come up with a   good  value

I do have  a copy  of a spec  sheet .   I am not sure it is  as complete as  some .



For me  the next  step   is  about  taking  what  I have  learned  about JTs    flybaclk and feedback and make some real  working  stuff  with that knowledge .
I  like to  keep  things simple   when expermenting .......so I  tend to make up  test boards .
and use standard connections .
What  I am   trying to do now is come up  with  an  adapter  for my 3055  test board that will allow me to run 12 V


gary   

[resonanceman]

From what Ive read about the 2n3055 they are fairly old and rarely are not used without a heat sink. You may just have thermal runaway. You could try putting a to-3 heat sink on to see what happens using just the pot. They also can be made by different manufacturers with slightly different specs. As Crowclaw suggested you may want to download the manufacturers data sheet for their parameters.
Jenna gave a pretty good link on calculations. You would need to know how much current is needed at the transistor to make a resistance calculation easier.   

IotaYodi

The last 3055 I fried   was not  thermal  runaway .   
I have a very  good heatsink on  my  test board .
Part  of the reason I want to  limit  the  3055  to less than half  power  is to get away from the need for  a  heatsink .

I am  adding  a picture of  my  2055 test board .


gary

Edit

This is an old picture
I have  added a few things .... like  terminal blocks   so I can   change the  transistor  faster
It is also designed for a  3 wire connection to the JT  coil
I will have to add another  connection point  for  higher  input voltage ........ 

[crowclaw]

IotaYodi

The last 3055 I fried   was not  thermal  runaway .   
I have a very  good heatsink on  my  test board .
Part  of the reason I want to  limit  the  3055  to less than half  power  is to get away from the need for  a  heatsink .

I am  adding  a picture of  my  2055 test board .


gary

Edit

This is an old picture
I have  added a few things .... like  terminal blocks   so I can   change the  transistor  faster
It is also designed for a  3 wire connection to the JT  coil
I will have to add another  connection point  for  higher  input voltage ........

Have you tried Power Mosfets, basically easyer to work with when experimenting, they have a high input impedance (no high base bias current) and the source to drain resistance is extremely low, so much better for your experimental projects.

[jeanna]

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~`

It  is  interesting  that   your  new LEDs  stay on longer than the others .
That is  similar to what happens  when  I add   alot of  feedback and   flyback trying to  reach  a self charging state .

No kidding. That is interesting.
thanks for that.

...
It   is  going to take  some doing to  get  enough  inductors  into   the  space inside the  light .

true!! 
the tor-23  (10/$1)  are only 3/8 inch across and that is what I am using to light the first 10 leds from only one secondary.

---
I am thinking that you could make a very simple breadboard circuit with a 2k r resistor and a single led. back to your 12v battery.

This will allow you to check the voltage across that resistor. When it gets to 1.2v you should be home free. If that is still too low you can raise it slowly etc.
this isn't a transistor, but it is the 0.7v to turn on the led and 2 volts to run it.

I had forgotten the 2 volts part. It now seems so high.

You can also read the volts across the led to be sure of that.


jeanna