Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.
Status
Not open for further replies.
Hello,

I have seen numerous Youtube videos on super high voltage Joule Thief projects. They are using small one inch ferrite toroids and from single 1.5 volt AA battery power a 3-5 watt 110-220 volt cree LED brightly consuming merely 200 mA of current from the battery! The following link will help you understand it better:

https://www.instructables.com/id/Make-a-SUPER-Joule-Thief-Light/

However, all such similar links do not specify the specific inductance of the Toroid being used or the Permeability. All they suggest is to use high permeability ferrite toroid! The authors also do not provide the answer.

Now, I want to replicate such a project at low cost. For that purpose, I have procured the following Toroids from Element14 (India):

  • Manufacturer: FAIR-RITE; Ferrite Toroid: 25.4mm x 15.5mm x 12.7mm; Core Material Grade: 77; AL (nH) : 2700 ±25% (I have procured)
  • Manufacturer: FERROXCUBE; TN25x15x10; Core Material: 3E25; AL: 5.62µH (I have procured)
  • Manufacturer: EPCOS; Toroid; 25.3 × 14.8 × 10.0; Core Material Grade: T37; AL (nH): 6970 ±25% (yet to procure based on your advice)
As, per the above mentioned link with the "Instructible", which suggests there should be 1 volt per turn induced in the secondary of the JT, I am wondering what should be the correct toroid selected for this purpose? The toroids I have already purchased do not give me such result.

Therefore, I will be more than happy if you can guide me accordingly on this.

Thank you,
With best regards,
Raj Mukherji
 
Hy Raj,

Welcome to ETO. I see that you are from India- care to tell us which part.

Most of those simple inverters comprise a relaxation oscillator. I can't get to the schematic from the link that you posted. If you can post the actual circuit you would like to build, perhaps we can help you with the design of your conductor.

spec
 
Last edited:
If the converter only draws 200 mA from a 1.5 volt battery the most power the lamp could consume would be 0.3 watts. In a real converter I think the efficiency would only be about 70% so that takes the output power down to 0.21 watts. Less than one tenth of the bulbs rating.

Les.
 
Hy Raj,

The FERROXCUBE; TN25x15x10; Core Material: 3E25; Should be OK. The core material is for suppression rather than inverters, but that should not matter too much for this application. This core will be much better than many cores you see on the net. It is a good size and should give a reasonable amount of power. If the transistor gets so hot that you cant keep your finger on it, you may have to mount the transistor on a heat sink

It would be advisable to put a 100nF ceramic capacitor, in parallel with a 470uF aluminum electrolytic capacitor across the 1.5V supply line as close as possible to the actual circuit.

just a word of caution: the secondary voltage is dangerous and can be lethal whatever is said on the net. The fact that only a 1.5V battery supplies the power makes no difference.

As Les said above, you simply cannot get more power out of any system than you put in. There are a lot of charlatans claiming otherwise on the net, but it is a fundamental physical impossibility and is refereed to as the 'acid test' of any design.

Just to explain, power in= Input Volts * Input Amps= Input Watts

Power out= Output Volts * Output Amps= Output Watts

But, as Les says, the circuit is only liable to be 70% efficient, so 30% of the power will be lost in heat.

Thus you have the formula, Vinput* Iinput= 0.7 * Voutput * Ioutput. Dont be surprised if the efficiency is more like 60% though.

Good luck with your project.

spec

Addendum:

Better Freoxcube core materials for inverters would be:
3C90
3C94
3C96
3C98
3C91
3C95
3C97
 
Last edited:
Hello,

I have seen numerous Youtube videos on super high voltage Joule Thief projects. They are using small one inch ferrite toroids and from single 1.5 volt AA battery power a 3-5 watt 110-220 volt cree LED brightly consuming merely 200 mA of current from the battery! The following link will help you understand it better:

https://www.instructables.com/id/Make-a-SUPER-Joule-Thief-Light/

However, all such similar links do not specify the specific inductance of the Toroid being used or the Permeability. All they suggest is to use high permeability ferrite toroid! The authors also do not provide the answer.

Now, I want to replicate such a project at low cost. For that purpose, I have procured the following Toroids from Element14 (India):

  • Manufacturer: FAIR-RITE; Ferrite Toroid: 25.4mm x 15.5mm x 12.7mm; Core Material Grade: 77; AL (nH) : 2700 ±25% (I have procured)
  • Manufacturer: FERROXCUBE; TN25x15x10; Core Material: 3E25; AL: 5.62µH (I have procured)
  • Manufacturer: EPCOS; Toroid; 25.3 × 14.8 × 10.0; Core Material Grade: T37; AL (nH): 6970 ±25% (yet to procure based on your advice)
As, per the above mentioned link with the "Instructible", which suggests there should be 1 volt per turn induced in the secondary of the JT, I am wondering what should be the correct toroid selected for this purpose? The toroids I have already purchased do not give me such result.

Therefore, I will be more than happy if you can guide me accordingly on this.

Thank you,
With best regards,
Raj Mukherji


Hello there,

Your example seems strange, with the LED, because we never get anything for nothing. We often look at the 100 percent efficiency numbers to be able to quickly see what might be going on, so doing that, we see that if we have 200ma at 1.5v going in and we have 150v on the output then the MAX current available to the output device is 2ma because the ratio of input voltage to output voltage is 100 and so the input current can only produce one one-hundredth (1/100) of the input current, which in this case is 2ma. THAT however is at 100 percent efficiency, so most likely we would see 1.5ma if we were lucky. That may or may not be enough to drive whatever it is we have on the output like a higher voltage LED.

[The following notes are about the two transistor version with single inductor not the transformer version]
The idea behind the Joule Thief is that when the core starts to saturate the transistor is forced out of saturation, and that means the collector voltage rises suddenly and that turns the device off. The core saturation happens when the drive current forces the core up high on the BH curve where the inductance falls sharply and so the current rises sharply too. The time it takes to get there is based on the inductor, but the value of the inductance has a lot to do with this because of v=L*di/dt, or rearranged, di=v*dt/L. In this last equation you see L in teh denominator, so di grows less with time as L is made larger. This helps set the oscillation frequency, and it cant be too high or the transistor doesnt have enough time to switch in and out of it's saturated base region mode. di cant be allowed to go too high or the core saturates too fast. Large value inductors will saturate faster because they have a lot of turns, but if you buy an inductor then they tell you the saturation level and current handling ability of the device. Knowing the saturation level and the inductance you can calculate the time to core saturation, and that cant be too fast. Probably 50kHz would be a good starting point, but even 20kHz would not be bad. You dont want to go too high because of the higher switching losses either. You'll have to buy an inductor that has an open frame so you can add the secondary winding. If you start with a raw core, you can try making a lower voltage converter first then go from there. See if you can get one 5mm white LED to light first.

To get a rough idea how all this works you can simulate it. Choose a transistor, choose an inductor value, see what happens.
 
Last edited:
I cannot help but think that this little project is on a par with much of the other madness which is published on the "Instructables" website.

Using a 1.5v battery to generate 130 volts, to light an LED. Is that sensible?

If you want to light an LED from a 1.5v battery you only need 4 or 5 volts, surely?

JimB
 
I cannot help but think that this little project is on a par with much of the other madness which is published on the "Instructables" website.

Using a 1.5v battery to generate 130 volts, to light an LED. Is that sensible?

If you want to light an LED from a 1.5v battery you only need 4 or 5 volts, surely?

JimB

If we weren't mad we wouldn't be here Jim.:arghh:

It is to light standard 110V LEDs, which are freely available and cheap. They would probably have a string of LEDs inside, or ironically a buck converter. :p

spec
 
Last edited:
I agree with JimB's comments. There will be two conversions. first from 1.5 volts to about 200 volts. then another conversion within the bulb from about 200 volts to about 11 volts (Assuming 3 LEDs driven in series) Both conversions will involve losses. It would be better and simpler to just step up the voltage to that required to drive the LEDs You would not need to wind a large number of turns for a secondary winding as it would not be required. The LED would be connected directly between the collector and the - Ve rail. There are many slight variations of the design if you search for "joule thief" I suspect the cores removed from a CFL lamp would work.

Les.
 
I agree with JimB's comments. There will be two conversions. first from 1.5 volts to about 200 volts. then another conversion within the bulb from about 200 volts to about 11 volts (Assuming 3 LEDs driven in series) Both conversions will involve losses. It would be better and simpler to just step up the voltage to that required to drive the LEDs You would not need to wind a large number of turns for a secondary winding as it would not be required. The LED would be connected directly between the collector and the - Ve rail. There are many slight variations of the design if you search for "joule thief" I suspect the cores removed from a CFL lamp would work.

Les.

But the OP didn't ask that question. :)

spec
 
Your example seems strange, with the LED, because we never get anything for nothing. We often look at the 100 percent efficiency numbers to be able to quickly see what might be going on, so doing that, we see that if we have 200ma at 1.5v going in and we have 150v on the output then the MAX current available to the output device is 2ma because the ratio of input voltage to output voltage is 100 and so the input current can only produce one one-hundredth (1/100) of the input current, which in this case is 2ma. THAT however is at 100 percent efficiency, so most likely we would see 1.5ma if we were lucky. That may or may not be enough to drive whatever it is we have on the output like a higher voltage LED.

The question of efficiency has already been done to death

The idea behind the Joule Thief is that when the core starts to saturate the transistor is forced out of saturation...

The idea behind the Joule Thief is that it accepts a low voltage, typically from a single solar cell or discharged primary cell, and converts it into a usable voltage. It uses a relaxation oscillator to do this. Relaxation oscillators of all types are used for many applications.

when the core starts to saturate the transistor is forced out of saturation, and that means the collector voltage rises suddenly and that turns the device off.
When the core saturates the collector current rises but what causes the transistor to turn off is the decrease in base drive because the core has saturated and there is no transformer action and thus no positive feedback.

The core saturation happens when the drive current forces the core up high on the BH curve where the inductance falls sharply and so the current rises sharply too. The time it takes to get there is based on the inductor, but the value of the inductance has a lot to do with this because of v=L*di/dt, or rearranged, di=v*dt/L. In this last equation you see L in teh denominator, so di grows less with time as L is made larger. This helps set the oscillation frequency, and it cant be too high or the transistor doesnt have enough time to switch in and out of it's saturated base region mode. di cant be allowed to go too high or the core saturates too fast. Large value inductors will saturate faster because they have a lot of turns, but if you buy an inductor then they tell you the saturation level and current handling ability of the device. Knowing the saturation level and the inductance you can calculate the time to core saturation, and that cant be too fast.

No. The higher the inductance the less current will flow and the slower the core will take to saturate. This is because the transformer has a voltage drive, not a current drive.

Probably 50kHz would be a good starting point, but even 20kHz would not be bad.

These inverters typically oscillate at 10Khz. Any higher and the losses rise exponentially.

You'll have to buy an inductor that has an open frame so you can add the secondary winding.

The OP already has a core which can be easily wound

To get a rough idea how all this works you can simulate it. Choose a transistor, choose an inductor value, see what happens.

I doubt the OP would have the skills to simulate the design- afraid inappropriate simulations have caused a lot of havoc on ETO recently.

spec
 
Last edited:
No marks for MrAl :)



The question of efficiency has already been done to death



The idea behind the Joule Thief is that it accepts a low voltage, typically from a single solar cell or discharged primary cell, and converts it into a usable voltage. It uses a relaxation oscillator to do this. Relaxation oscillators of all types are used for many applications.

When the core saturates the collector current rises but what causes the transistor to turn off is the decrease in base drive because the core has saturated and there is no transformer action and thus no positive feedback.



No. The higher the inductance the less current will flow and the slower the core will take to saturate. This is because the transformer has a voltage drive, not a current drive.



These inverters typically oscillate at 10Khz. Any higher and the losses rise exponentially.



The OP already has a core which can be easily wound



I doubt the OP would have the skills to simulate the design- inappropriate simulations have already caused enough havoc recently on ETO.

spec

Hi there,

What ?????? :)

Yeah this is a funny one. I was describing the 'other' Joule Thief, probably because i've used that one much more often. That's the two transistor version that does not require a transformer just a single inductor. Much simpler to build because we only need an inductor. Maybe i should post that schematic, if the OP is interested.
I jumped right into that one for some reason probably because i used it much more often. I have used the transformer version too though but yes that is somewhat similar to a Royer. I used it to build an islolated low current power supply (transformer was good for that).
I'll put a note in my original post:
"[The following notes are about the two transistor version with single inductor not the transformer version]"
But i should at least get an "E" for "effort" :)
Or a "B" for "too much Beer".
 
Hi there,

What ?????? :)

Yeah this is a funny one. I was describing the 'other' Joule Thief, probably because i've used that one much more often. That's the two transistor version that does not require a transformer just a single inductor. Much simpler to build because we only need an inductor. Maybe i should post that schematic, if the OP is interested.
I jumped right into that one for some reason probably because i used it much more often. I have used the transformer version too though but yes that is somewhat similar to a Royer. I used it to build an islolated low current power supply (transformer was good for that).
I'll put a note in my original post:
"[The following notes are about the two transistor version with single inductor not the transformer version]"
But i should at least get an "E" for "effort" :)
Or a "B" for "too much Beer".

Your a gentleman MrAl :)

spec
 
If the freq is below 100kc then I'd suggest N87 power ferrite, from memory initial permeability is about 3k.
3E25 is usable as a power ferrite, however I'd be using that more for gate drive transformers.
 
Hy Raj,

One thing that I forgot to mention is that you should be prepared to adjust the value of the resistor to suit your application. A range of about 100 Ohms to 1K Ohms would be good. Also, it would be better to make the primary wire that connects to the transistor collector a bit thicker- the thicker the better consistent with fitting the wire on the core that is. :)

spec
 
For clarity here is the circuit and winding instructions. The dots on the windings show the start of the winding. All windings are wound in the same direction. A wide range of transistors, including MOSFETs, will be suitable. The capacitor provides decoupling and also allows more energy to be extracted from the battery as it discharges.

spec​



2016_05_24_ETO_LOW_VOLTAGE_INVERTER_VER1.png
ERRATA
(1) Change transistor from TIP31x to TIP35x
(2) Change winding 1-4 from 24SWG minimum to two parallel windings of 24SWG minimum
 
Last edited:
Hi All,

Many thanks for your time in replying back to my query.

I have noted your comments and advice and would request MrAl to furnish the 2 transistor converter circuit for our reference here.

Let me also inform you that I have already made a JT using 3E25 core mentioned above with 400T of 30 SWG as Secondary, 15T and 60T with 26SWG and it is able to drive 3W 220V LED bulb. However, I am not very happy with the brightness of it. There is yet another JT circuit which I have made with the 77 toroid with 300+T using 30SWG, 4T using 22SWG and 14T using 24SWG. This also lits up the same LED bulb. The brightness of the light with both the two circuits seems to be same. To me, it can be useful as a night lamp but I want to increase the intensity more. I have tried using 2n2222, TIP41C, BD139...all give me fairly the same result!

The first circuit using 3E25 core gives me around 30 volt pulsed DC output without any load from a fuuly charged 1.5 volt AA 2100 mAH NiMH battery where as the second circuit gives around 50 volt output. I am not able to increase the voltage beyond this. I know that the LED bulb brightness will increase when the secondary output reaches 110 volt. I will try the above circuit suggested by SPEC and revert with my results. Please find the screenshot of the circuit for "Super JT" attached here.

However, any suggestion as how I can achieve 110 volt output from the secondary coil will be appreciated.

Thanks and regards,
Raj
 

Attachments

  • Schematic.JPG
    Schematic.JPG
    62.2 KB · Views: 420
A capacitor across the inverter supply might help. Try a large aluminum electrolytic (470uF or above) in parallel with a 100nF or so ceramic capacitor if you do not have a single large ceramic capacitor.

When the capacitors are fitted. measure the voltage across the battery when the inverter is running under full load.

Try putting three more 1.5V fully charged batteries in parallel, giving a total of 4 batteries to see what happens.

spec
 
Last edited:
I have just had a look at the TIP31 data sheet. It seems to be unsuitable for your application: Vbe, Vcsat, hFE. I will have a look for something better.

UPDATE: The TIP35x (where x is A, B,or C) would be much better choice for a common cheap transistor. Other transistors would be even better though.

spec
 
Last edited:
Hi All,

Many thanks for your time in replying back to my query.

I have noted your comments and advice and would request MrAl to furnish the 2 transistor converter circuit for our reference here.

Let me also inform you that I have already made a JT using 3E25 core mentioned above with 400T of 30 SWG as Secondary, 15T and 60T with 26SWG and it is able to drive 3W 220V LED bulb. However, I am not very happy with the brightness of it. There is yet another JT circuit which I have made with the 77 toroid with 300+T using 30SWG, 4T using 22SWG and 14T using 24SWG. This also lits up the same LED bulb. The brightness of the light with both the two circuits seems to be same. To me, it can be useful as a night lamp but I want to increase the intensity more. I have tried using 2n2222, TIP41C, BD139...all give me fairly the same result!

The first circuit using 3E25 core gives me around 30 volt pulsed DC output without any load from a fuuly charged 1.5 volt AA 2100 mAH NiMH battery where as the second circuit gives around 50 volt output. I am not able to increase the voltage beyond this. I know that the LED bulb brightness will increase when the secondary output reaches 110 volt. I will try the above circuit suggested by SPEC and revert with my results. Please find the screenshot of the circuit for "Super JT" attached here.

However, any suggestion as how I can achieve 110 volt output from the secondary coil will be appreciated.

Thanks and regards,
Raj

Hi,

You'll get better coupling from primary to secondary if the secondary is wound right on top of the primary winding, creating two layers instead of having the secondary wound on one sector of the toroid and the primary on another sector of the toroid core. Better coupling means less leakage inductance which translates to less loss of voltage so you could try improving the coupling to see if that helps. The way you have it shown in your drawing looks like the coupling now could be as low as 0.5, while with the secondary on top of the primary it could be closer to 1.0 which is ideal. It will never really be ideal, but that might help anyway.
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top