Big Joule Theif

[Watson]

Thanks for the explanations. It is fascinating. I will study your circuit more closely when I got some time.

For those who are interested, I'd like to give an explanation of JT here:

Let's start with the point of time where there is ZERO collector-emitter current in the 2n3055. Now because of the battery, there will be base-emitter current through the trigger coil, limited by the resister in series. This makes the transistor close (as a switch) between collector and emitter, but keep in mind that there is a ceiling cap on the current that can flow through (roughly equals the base-emitter current multiplied by the beta parameter of the transistor). So current in the power coil starts to increase, until the current ceiling cap is reached, at which point the current stops increasing. Mind you one more crucial detail: while the current was increasing in the power coil, the trigger coil can pick up a little boost and that help contribute to the base-emitter current. Now when the current in the power coil stops increasing, such a boost ceases to exist and this immediately drops the ceiling cap quite a bit (say, 20%, depending on how much boost the trigger coil picks up from the power coil). Such a drop of ceiling cap sends the power coil into its down turn at once -- the current in it starts to decrease, and such a current decrease in the power coil further lower the cap through the pick up (inductance) at the trigger coil ( in fact, the decrease is usually fast enough to shut off the transistor completely). The decrease stops until the current in the power coil reaches ZERO, and we are about to start a new cycle. That's it, folks!

A point to remember is that there is transfer of power from the primary winding (called power coil above) to the feedback winding (called trigger coil above) *only* when there is a change of flux, which is caused by a change in current in the primary winding. Once the current reaches its maximum (called ceiling above), there is no longer a change in flux and no longer a transfer of power to the feedback winding.

Since the feedback to the feedback winding is regenerative or in phase, once the transfer stops, the current from the feedback winding to the base of the transistor drops a slight amount. And this causes the collector current to drop, which then transfers power from the primary to the feedback winding and since it's opposing the base current, it further reduces the base current.

Thus complete cutoff of the base current happens suddenly, and the stored energy in the primary winding has no place to go, so it tries to increase the voltage across the transistor, but when the voltage gets to the point where the LED conducts, it flows through the LED. Let there be LIGHT!

And then the whole process repeats itself, at about a hundred thousand times each second. Cool, huh?

Has anyone experienced this problem? I bought some LEDs on ebay and found that they have air bubbles in the epoxy, which in my opinion is something that should be caught during inspection at the factory and rejected. The seller doesn't seem to think this is a problem. What do you think? I want my money back.

[lanenal]

An explanation about JT

Thanks for the explanations. It is fascinating. I will study your circuit more closely when I got some time.

For those who are interested, I'd like to give an explanation of JT here:

Let's start with the point of time where there is ZERO collector-emitter current in the 2n3055. Now because of the battery, there will be base-emitter current through the trigger coil, limited by the resister in series. This makes the transistor close (as a switch) between collector and emitter, but keep in mind that there is a ceiling cap on the current that can flow through (roughly equals the base-emitter current multiplied by the beta parameter of the transistor). So current in the power coil starts to increase, until the current ceiling cap is reached, at which point the current stops increasing. Mind you one more crucial detail: while the current was increasing in the power coil, the trigger coil can pick up a little boost and that help contribute to the base-emitter current. Now when the current in the power coil stops increasing, such a boost ceases to exist and this immediately drops the ceiling cap quite a bit (say, 20%, depending on how much boost the trigger coil picks up from the power coil). Such a drop of ceiling cap sends the power coil into its down turn at once -- the current in it starts to decrease, and such a current decrease in the power coil further lower the cap through the pick up (inductance) at the trigger coil ( in fact, the decrease is usually fast enough to shut off the transistor completely). The decrease stops until the current in the power coil reaches ZERO, and we are about to start a new cycle. That's it, folks!

The input power is the input voltage multiplied by the iput current. The current through the LED is found by putting a 1 ohm resistor in series with the LED cathode. For every millivolt I measure across the 1 ohm resistor, there is 1 milliamp through the resistor and the LED. Assuming 3.2 volts across the typical white or blue LED, the power to the LED is the current multiplied by 3.2.

The efficiency is the power to the LED divided by the input power, and multiplied by 100 to get the percent.

Problem is that when I do the measurements and get a figure above 90%, then double check and get the same thing, I begin to think, something is not right here. But what adds further confusion is that the power supply meter says 1.5 volts and, say, 50 milliamps, and my eyes see that the LED is very bright. The input power is only 75 milliwatts, less than half of what the conventional Joule Thief takes, yet the LED looks as bright as a conventional Joule Thief. My eyes tell me that this is really putting out the light at very low power, but my head is telling me that the circuit can't be nearly 100 percent efficient.

We (meaning others and myself) believed that the measurements were not accurate because the meter was not measuring the LED current correctly. It is a high current pulse and the meter is not meauring the true current. I put a 100k resistor and 0.1u capacitor low pass filter between the 1 ohm resistor and the meter to average out the pulses. and I still get the same current.

Well, hell...

[Watson]

Nice stuff! A quite different mechanism for oscillation -- did not thought about that before. Not sure what do you mean by 90% efficiency though -- do you mean electricity -> light or something else? How did you get this figure?

The input power is the input voltage multiplied by the iput current. The current through the LED is found by putting a 1 ohm resistor in series with the LED cathode. For every millivolt I measure across the 1 ohm resistor, there is 1 milliamp through the resistor and the LED. Assuming 3.2 volts across the typical white or blue LED, the power to the LED is the current multiplied by 3.2.

The efficiency is the power to the LED divided by the input power, and multiplied by 100 to get the percent.

Problem is that when I do the measurements and get a figure above 90%, then double check and get the same thing, I begin to think, something is not right here. But what adds further confusion is that the power supply meter says 1.5 volts and, say, 50 milliamps, and my eyes see that the LED is very bright. The input power is only 75 milliwatts, less than half of what the conventional Joule Thief takes, yet the LED looks as bright as a conventional Joule Thief. My eyes tell me that this is really putting out the light at very low power, but my head is telling me that the circuit can't be nearly 100 percent efficient.

We (meaning others and myself) believed that the measurements were not accurate because the meter was not measuring the LED current correctly. It is a high current pulse and the meter is not meauring the true current. I put a 100k resistor and 0.1u capacitor low pass filter between the 1 ohm resistor and the meter to average out the pulses. and I still get the same current.

Well, hell...

[lanenal]

self-charging JT (SJT)

@all: I'd also like to suggest a self-charging JT circuit here: basically it feeds back to the same source. See attached. This makes testing OU easy: just see if the rest voltage of the source gets higher after using ...

Edit: to play save, the diodes can be 1N4007. The battery voltages can be 6V-12V, given that you used more tolerant transistors (2N3055 and MJ2955).

LEGAL NOTICE: This is released into the public domain for the overall goodness, nobody shall patent it. If anyone choose to implement this, you must take all the responsibilities for any bad consequences.

There is a circuit for it on page 6 post #154
There is also a basic Joule thief circuit on page one.

Lidmotor also has a very nice Inverted Joule Thief curcit some where in this thread.

I also sent you a PM for a link to the camera mod.

[lanenal]

Nice stuff! A quite different mechanism for oscillation -- did not thought about that before. Not sure what do you mean by 90% efficiency though -- do you mean electricity -> light or something else? How did you get this figure?

I got the schem, thanks for the link.

I read some of the thread and saw that there's a quest for efficiency.

Recently I discovered that I could double the efficiency of the JT, easily, with a diode, resistor and capacitor. I called it a Supercharged JT.

The scary part is that I sometimes got an efficiency of over 90 percent.

[Watson]

Supercharged JT

There is a circuit for it on page 6 post #154
There is also a basic Joule thief circuit on page one.

Lidmotor also has a very nice Inverted Joule Thief curcit some where in this thread.

I also sent you a PM for a link to the camera mod.

I got the schem, thanks for the link.

I read some of the thread and saw that there's a quest for efficiency.

Recently I discovered that I could double the efficiency of the JT, easily, with a diode, resistor and capacitor. I called it a Supercharged JT.

The scary part is that I sometimes got an efficiency of over 90 percent.

Quantsuff and I both experimented with the circuit and he and I confirmed the results. But we believe the measurements are causing too high readings. We monitor the LED current with a 1 ohm resistor in the cathode led of the LED; for every millivolt across it, there is a milliamp of LED current.

He modified the circuit to make it flash. Problem is that it likes to flash only with air core coils, not toroids. But the real point is that by using the circuit I gave in my link above, it will double the efficiency of the JT, or conversely it will reduce the supply current in half, while still maintaining a very bright LED.

Also, by using the circuits in the link to QS's webpage above, the drive to the transistor can be increased. He shows one circuit that can drive a 1W Luxeon to full brightness.

I'm thinking I can use this circuit to drive an old tube radio output transformer by connecting the speaker winding to the transistor, and a CFL across the winding that goes to the tube.

More on this later...

[slayer007]

OU? any chance you could post a link to it? Thanks.

I got some Fuji and Kodak flash units from disposable cameras and I'd like to modify them for driving a CFL.

There is a circuit for it on page 6 post #154
There is also a basic Joule thief circuit on page one.

Lidmotor also has a very nice Inverted Joule Thief curcit some where in this thread.

I also sent you a PM for a link to the camera mod.

[Watson]

@ Guruji

Really you don't need the capacitor in there at all.
Xee2 at OU redrew the circuit for me and he put the capacitor in there.

I didn't have it in there in my origonal circuit I posted at OU.
It's there for a filter but it's really not needed at all.

OU? any chance you could post a link to it? Thanks.

I got some Fuji and Kodak flash units from disposable cameras and I'd like to modify them for driving a CFL.

[slayer007]

@ Guruji

Running at 3v with two 1.5v batterys.

The rectified voltage across the sedondary going into a small capacitor is around 380 volts.

It should take close to 300v to run a CFL.

This was tried with the basic JT circuit with only one transistor.

One thing you could also try is to take out the 1k resistor and put a 5k pot in it's place.
That way you can go below 1k if you need to for tunning.

Also for best results the toroid should be a ferrite toroid.

P.S. I used a rectifier across the secondary because the AC comming off the secondary isn't true AC it might not be right at 60mhz.
So my meter wouldnt read it correctly.

[Guruji]

Bjt

Hi SLayer another question please.
How much ac voltage should the secondary give in 480turns cause I did not add the turns .
Now my BJT secondary is giving 14v ac. I have alot to go to light a flourescent?
Thanks

[Watson]

some ferrite torroids
Toroids | AllElectronics.com

I bought some of the "charcoal" # TOR-23 cores from them. The TOR-43 is so small it's difficult to wind for a JT, but I stacked three of them and wound 34 AWG, and managed to get it to work.

The TOR-23 cores do well with three 16 inch lengths of 30 AWG trifilar wound, which gives each winding about 150 uH. I connect two windings in parallel for the primary, leaving the third for the feedback winding.

I just bought (from Mouser) some Fair-Rite 2673002402 and 2643002402 "IM shield beads" which are really toroids. They're smaller, about 3/8" O.D. and 3/16" high, so I have to use smaller wire with the '43, but the '73 is good enough with 24 gauge telephone wire.

[vzon17]

Wow good one!!
I think that this is a "Must do" experiment replication as soon as I find a ferrite ring big enough. Any tips on where I can find one? I tried a speaker ring magnet but had very poor results. What is the amp draw on this when the CFL is on? Paul's rendition of this JT/CFL was drawing 800Ma at 10volts but he wasn't using secondary pickup winding. This 8:1 winding ratio seems to work just fine. Are the primary and secondary wires the same thickness?
Thanks,

Lidmotor

some ferrite torroids
Toroids | AllElectronics.com

[Watson]

Hi guys about the capacitor I am using 10uf 16v electrolyte cap. Is this limiting my voltage down?. I have to increase this in slayer circuit to acquire more voltage in circuit?.
Any help please?
Thanks

Is a schematic available? A link to it? Thanks.

[Guruji]

Bjt

Thanks Slayer for your help to me and to all
Thanks

[slayer007]

@ Guruji

Really you don't need the capacitor in there at all.
Xee2 at OU redrew the circuit for me and he put the capacitor in there.

I didn't have it in there in my origonal circuit I posted at OU.
It's there for a filter but it's really not needed at all.