Where have I gone wrong with this joule thief?

[Malalkin]

I'm using only electronics I pulled from an old TV.  I followed the schematic from wikipedia as well as I could, and I'm about 90% sure the coil is going in the correct direction.  But for some reason if I feed it 1.4V, I get 1.4V out (before the diode, so my capacitor only charges to 1.1V after the diode). 

 

Transistor: C945P (tried swapping the transistor to no avail); looking on the flat side, the pinout is emitter-collector-base. 

Resistor: 810 ohm

Inductor: I don't know what the inductance is, but there's a '12' written on the side and the top has the numbers 47E00-0110

Red magnet wire (+), brown magnet wire (-)

 

Any idea what I've done wrong here?

[WhitetailAni]

What exactly are you trying to do?

[Malalkin]

I mostly wanted to just build a joule thief to prove a point, but I was thinking in the future I could play around with it and try boosting the output enough to charge a 200V capacitor using a dead battery

[Hackentosher]

This looks horrifying, I love it.

 

Can't really tell much about what is going on without a schematic soooo no one is going to be able to tell you what's wrong with it. Also knowing the specs of your transformer is important, how do you know it's not a 1:1 isolation transformer?

[Malalkin]

Here's basically the schematic from wikipedia:

 

The windings on the transformer are the same resistance so it's probably 1:1, but anything I found online didn't seem to specify anything about winding ratios for joule thief devices.  Do I need a different transformer/coil then?

[mariushm]

1:1 is wrong ,, maybe 1:300 would make more sense

[Malalkin]

Now the ratio is 4:9 on a torroid, but the output is still 1:1. Although there's very little copper in play. 

[Unimportant]

@Malalkin Considerations:

• 1N400x has way too slow reverse recovery. Use 1N4148 or schottky diode.
• Regulation zener D2 is required for freewheeling. Whitout it (and no load) voltage will rise and kill transistor/cap. 
• Second image has way too low inductor turns (cant see first very well). Its not just the ratio between pri/sec, which can be 1:1. You need enough inductance. Wikipedia lists 20 turns for each.
• Make sure to use ferrite core toroid. Ferrite is typically black.
• Resistor color bands are unclear in picture. Last band looks like it might be orange in stead of brown. Make sure it's not many kilo-ohms.

Perhaps try BigClive's original joule thief first: http://www.bigclive.com/joule.htm

 

The original joule thief just drives a LED, soldered straight onto the transistor, which does all the clamping. This modified design, basically a poor man's boost converter, might not be as suited for a dead-bug method as it might be more suseptible to stray inductances causing ringing/resonance and killing the transistor.

[Malalkin]

37 minutes ago, Unimportant said:

• 1N400x has way too slow reverse recovery. Use 1N4148 or schottky diode.
• Regulation zener D2 is required for freewheeling. Whitout it (and no load) voltage will rise and kill transistor/cap. 
• Second image has way too low inductor turns (cant see first very well). Its not just the ratio between pri/sec, which can be 1:1. You need enough inductance. Wikipedia lists 20 turns for each.
• Make sure to use ferrite core toroid. Ferrite is typically black.

Perhaps try BigClive's original joule thief first: http://www.bigclive.com/joule.htm

 

The original joule thief just drives a LED, soldered straight onto the transistor, which does all the clamping. This modified design, basically a poor man's boost converter, might not be as suited for a dead-bug method as it might be more suseptible to stray inductances causing ringing/resonance and killing the transistor.

Didn't know about most of those things.  Although the second image is a ferrite core (it's just wrapped in plastic).  Thanks for the link. 

 

To be clear, I am trying to build something of a boost converter, just one that doesn't destroy itself.  My previous project was a 555 timer creating a square wave which went into a small 300:1 transformer and then a voltage multiplier network to yield about 200 volts.  It was ridiculously large and inefficient. 

[Unimportant]

@Malalkin I'd go for the standard boost-converter topology then: (image from wikipedia)

For the switch and diode, use parts that can handle the desired output voltage, and required current. The diode should be fast recovery, probably a high voltage schottky. For the inductor use a ferrite core with many windings, the more inductance the better (upto a limit of course, the core will saturate at some point anyway). Use your 555 to generate adjustable PWM to drive the switch transistor. Start with a low duty cycle and slowly move it up until the core saturates (you'll see the current draw spike up, a scope would help with this). Then back it off a fair bit, you don't want to be in the saturation region. That'll set it to it's maximum. With the right inductor you should be able to reach a pretty high voltage. Mind you it'll draw substantial amount of current, you cannot create energy out of nothing. High voltage, low current out means low voltage, high current in. Use a high voltage elco as capacitor but put some film capacitance in parallel with it to compensate for the elco's poor properties. Build this on a proper perf board with short connections to keep stray inductance low in both current loops.

[Distinctly Average]

On 10/30/2020 at 10:14 PM, Unimportant said:

@Malalkin Considerations:

• 1N400x has way too slow reverse recovery. Use 1N4148 or schottky diode.
• Regulation zener D2 is required for freewheeling. Whitout it (and no load) voltage will rise and kill transistor/cap. 
• Second image has way too low inductor turns (cant see first very well). Its not just the ratio between pri/sec, which can be 1:1. You need enough inductance. Wikipedia lists 20 turns for each.
• Make sure to use ferrite core toroid. Ferrite is typically black.
• Resistor color bands are unclear in picture. Last band looks like it might be orange in stead of brown. Make sure it's not many kilo-ohms.

Perhaps try BigClive's original joule thief first: http://www.bigclive.com/joule.htm

 

The original joule thief just drives a LED, soldered straight onto the transistor, which does all the clamping. This modified design, basically a poor man's boost converter, might not be as suited for a dead-bug method as it might be more suseptible to stray inductances causing ringing/resonance and killing the transistor.

Just to add, Big Clive also has a video on subject and it is actually one that got me back into electronics after many years off as I changed jobs from electronics to IT. Seems I had forgotten how much fun I used to have playing with things like this. -