Just a quick note in case someone would like to try this and not have easy access to a toroid core: Don't be discouraged if you don't have the exact components listed here; just about any reasonable setup of inductors can work just like this center tap transformer that is built in this video. I have one on a little piece of perf board that is using a pair of through hole transformers...the kind that look like regular old resistors. Despite what some results on Google would lead you to believe, just about any configuration of these basic parts will probably work out for you. Part of the fun of this circuit is modifying it to change the output and seeing just what you can and can't get away with.
The transistor oscillator doesn't stop oscillating if you remove the LED, which means that when the transistor is switched on, it will conduct between Vcc and ground freely. That's where the current is going.
Having come here from Big Clive's video (chaining from atomic14's wireless charging video) I note the last thing he does is say don't let it go open circuit. The first thing Julian does is pull the LED. No punches pulled :D
i would think you are correct over time at the 10v it jumped to.. maybe if you used a full 1.5v guessing the voltage would be just less than 40v just under its rating possibly damaging the it faster.. but i haven't seen any tests on this.
For this video my comments would be...1. You have multi channel scope and it would have been nice to see both sides of the inductor voltages at once. A serious problem when someone is not familiar with what a scope can do. 2nd point is, you need different turns ratio for two coils in order to optimize the overall current consumption. 3rd point is, your MPPT charging wont work at its best if you use that green coloured inductor. Thats a pure ferrite core and you need something with lesser permeability. This green ferrite is okay for making high value inductors but not suitable for buck/boost converters.
First of all, I wish you a happy new year. I made a few joule thief circuits myself and used iron powder cores instead because after testing I found out that the iron powder cores are better because you're using the core to store energy in a magnetic field, because of the iron powder core the magnetic field is harder to flip around and that helps with efficiency and I also don't hear any ringing in the core as you normally get when you use a ferrite core. I used the joule thief for two different reasons, the first was just to see how it works and if I could get it to light up some LEDs of an empty battery and I wanted to know how long the LED could run continuously of the dead battery, quickly I discovered that my transistors are getting way to hot when I use a 1k ohm resistor as you suggested in your video... So I replaced the resistor with a 10k ohm one and even that was not a high enough value resistor, then I used a 100k ohm one and after that a 1M ohm one and my LEDs are just as bright with the 1M ohm resistor as with all the lower values but now I don't have the thermal issues anymore. However I do have a second way that I'm using the joule thief circuit to get a string of 20 white LEDs to light up as bright as possible and I'm not using a battery power source but an old 800mA @ 5 Volt Nokia charger, that's for my helping hands with magnifying glass that I customized by adding 22 white LEDs to make my workspace much better... Again I used the iron powder core but now I wanted to see what happens if I made the windings different by adding one more winding on the collectors side so I still have the center tap with one winding going round once in one direction to the resistor and the base of the transistor and two times in the other direction making it a 2:1 coil instead of the regular 1:1 coil and I also added an extra parallel connected transistor, a diode and capacitor which is connected to the 20 LEDs string... It's not the most efficient circuit and it gets pretty warm to the touch but the 20 LEDs are lighting up very bright which is exactly what I wanted for that application, it will run on a 3.7 volt Li-Ion battery but not very long... it's completely drained in a couple hours which is an unexpected result, I thought it should keep running for days but it doesn't if you're pushing the LEDs to the limit. All in all I find the Joule Thief a lot of fun to play with and it's probably the most forgiving circuit of all, I did manage to burn a few LEDs and transistors but that was totally my own fault because of fault wiring or driving the components way out of spec just to see what happens... LOL 😆😆😆😆💥🆒
Put an oscilloscope on that transistor. I think you'll find that when you jump the resistor with a capacitor, the "resonance" (for lack of a better word) of the transistor drops. That's probably why you see a drop in power draw.
Cool, I built my first working joule thief two days ago now, following Big Clive's instruction to the letter. Works great, I have a few old ones laying in boxes somewhere that I tried to build but failed so I was quite pleased to get one working. In our motor home, which we use to travel the USA and spend the winters in areas of the states that are without snow, thus the nickname for folks such as ourselves "snow birds" we use tons of AA batteries, in nightlights, my metal detector, and even my label maker. So I always have lots of the blasted things laying around, and my main tool is a battery tester. So now I have a home built light that uses the old batteries instead of tossing them. I plan on building a small holder for my thief so it can be used as a light beside my easy chair in the rig, just to take the dark off when I actually put a tiny slide switch on mine, and used big Clive's method of casing with hot glue.
Julian: its fun to build coils (chukles) Me: remembers masters degree school, design and calculation of impedance matched coils at microwave frequencies (cries)(a lot)(in pain)
Yep at microwave frequencies even a round trace vs a square trace on a printed circuit board can have a noticeable effects on signal attenuation, leaks etc. ahhh the fun of microwaves. :D
Wondered when our "Jules" was gonna become a jewel thief.... Seriously, I was working on an MPPT-type thingy for an environmental company in Finland. Reason? They had data loggers in the wilderness, powered by 9Ah SLA batteries. Their method was to stick a 10-watt panel in series with a silicon(?) diode (Schottly?? Never heard of her, mate!) and either NO charge in winter, or just boil the hell out of the batteries in 24-hour daylight in Summer... Which is why your Arduino Muppet project was of interest... BUT, Winter...I can still get a few volts out of the panel, and use it to charge a BFC (Big..er...Friendly...Capacitor), then dump it into a Joules Thief, give the battery a bit of a kick...Then, when Summer kicked in, turn the whole mularkey to a MuPPeT device. Sadly, my work-experience ended, and the company couldn't even afford the necessary insurance for me (as unemployed) to work for free. RATS!
...Actually, once the "BFC" had got enough charge, I'd use a Unijunction Transistor* to trigger a FET switch to dump the BFC's charge into the 'theif - then to the battery. *UJT - (2N4656) - That'll get your younger readers scrabbling in Google! Us old-timers (pun intended) know what they are!
@@adamwade855 I think he refers to the fact that the oscillating voltage is not correctly reported by most multimeters in dc mode. RMS just refers to the method of getting an approximate readout for the oscillating signal.
I suspect the voltage is being clipped to 10V due to the ferite core being saturated, not any form of leakage/breakdown of the NPN junction in the transistor.
From memory, adding a capacitor raises the turn-on voltage. So it won't work on very low battery voltages unless you only add the capacitor after the battery is connected.
I made plenty of these on my cement battery experiments. I found that if you put a electrolitic capacitor between the + and - of the cell the led will flash. This happens when the battery is almost expired. keep up the great videos.
your scope is dual trace right ? Take one probe and put on the base. One on the collector. compare the timing of the pulses on base to pulses at the collector with and without the led and you will see biasing of the transistor, pulsed biasing, turning on and off the transistor. second video maybe ? if you see what I think you will see, you can explain on the video.
That's actually an interesting idea, to see the interaction of the base trigger signal vs the collector's potential. My guess is that there's a few microseconds difference around a 16khz oscillation.
While you have one channel of your scope connected to the collector, try connecting the other channel of your scope to the base of the transistor. That may help explain things while you have the LED out of circuit.
Suggest that you wind only one length of wire and after ten turns or so make a loop and then keep winding in the same direction and that will make your center tap without the mucking around. So much easier
A very mysterious circuit, love it's simplicity but complex operation! Took me ages to figure it out when I first came across it and it's hard to explain. I've seen the notation for this style of inductor drawn as a transformer but with dots marked next to the coils to indicate one winding is opposite. Then in this case the top 2 connections are commoned to positive. You're almost there, i find breaking time down slow motion like you did helps a lot. Remember that the field produced by the first coil induces current in the opposite direction in the second coil and vice versa. I wonder how measuring the voltage at collector to ground effects the operation, I'm guessing that the load of even a meter must effect things?
I built a LED dog collar with a Joule Thief to power 4 LEDs. I used two inductors (they look like dark brown resistors and also have the colored rings) instead of a coil to fit it all into a thin transparent flexible tube. It works perfectly for 60 - 100 hours with a single AA battery, even under water. It runs the battery down to about 0,6 Volts until the LEDs get too dark. I love it. It is better than all LED dog collars, i ever bought and cost me about USD 2 for parts and 1 hour of my time :)
Thank you :) I tend to overengineer everything i make and really enjoyed this one evening non-AVR project in contrast to the complex machines i usually build.
The C-E voltage rises only to about 10v. At this point, the voltage induced at the second coil (left side) is enough to trigger the transistor ON. I think the led would burn out instantly, if you'd put it back in at 40v.
Using a fixed value resistor is pointless use a potentiometer and fine tune it untill the led at its brightest use a small value cap across the resistor after fine tuning you will find two sweet spots where brightness is maximum remove the pot measure the ohms select a resistor closest to the meter reading connect the cap. Across the resistor and you have max brightness for min current
Not a bad idea, with a pot one could easily find the "sweet spot" of brightness and consumption which is a functions of the charge/discharge duty cycle. However we can't violate the laws of energy conservation no matter how hard we try it, we can only make it a bit more efficient. At 25:18 he's mentioned Colin Mitchell having been able to cut the power usage to about a third, which may be possible with a precise selection of resistor/capacitor values, including a high efficiency inductor, and most importantly a very efficient LED (expensive one). Also, putting a "snubber" cap across the LED to smooth out longer duty cycles of the oscillator may help too.
Jullian, in your search for the reason your flyback voltage without the LED limits at a value lower than you expected, check the Vebo spec limit and recall you have a coupled inductor...........
At 25:18 he's mentioned Colin Mitchell having been able to cut the power usage to about a third, which may be possible with a precise selection of resistor/capacitor values, including a high efficiency inductor, and most importantly a very efficient LED (expensive one). Also, putting a "snubber" cap across the LED to smooth out longer duty cycles of the oscillator may help too. This tiny circuit has been around since 1999 (and possible even before that in some other variation), with its remarkable simplicity and efficiency has spawned countless videos. It's a really nice little project for even beginners to try, well if it wasn't for having to make the coil by hand. :D
A few tens of microhenries per winding. I'm getting in the region of 20 to 40 uH in my experiments. The capacitor really makes a difference. It allows you to increase the resistor to 10k or more but it works best if you move the resistor from between the coil and the transistor base to between the positive supply and the secondary coil. Then put the capacitor between the junction of the resistor and coil and the negative supply. The best value seems to be somewhere between 1nF and 10nF. For such a simple circuit there's much to experiment with. The windings don't need to be symmetrical. You can use fewer turns of finer wire for the secondary (base) coil, which will allow room for more turns or thicker wire for the primary (collector) coil. You don't need such a large ring to wind the coils on. A tiny ferrite bead with 8 or 10 turns will also work. You can also experiment with the transistor as they have different voltage drops when fully turned on Vce(sat). The BC337 seems to be a particularly good choice. It's a fascinating circuit because, while almost anything will work there a huge scope for refining it and squeezing every last joule of energy out of that "dead" cell. More esoteric variations include using a germanium transistor because it will continue to oscillate when the cell voltage drops lower than 0.7 volt and substituting a MOSFET to reduce the bias current to almost zero - though in that case the gate bias voltage is a problem, requiring additional circuitry or a couple of button cells to provide it.
Have you ever checked the frequency that they are oscillating at? Mine was 170 KHZ, but the other one I built is 142 KHZ so there is a difference in the windings in each inductor and I figure that is why my two joule thief's have a different frequency.
If you're persistent enough you can fit a joule thief into regular tungsten light bulb case (without resorting for smd parts!) and make any old torch a LED torch that runs even on flat batteries. Good fun. By adjusting this resistor you can maximize light output and tune the circuit to the inductor you've made. Another neat trick is to use it for super emergency lighting because this little jewel (pun intended) can easily power 5m led strip out of dead batteries.
Isn't it simply the case that with the transistor on, it provides a lower resistance path to ground, so current stops flowing through the resistor turning the transistor off again, and that's how it oscillates?
I was able to effectively double the current lol...by not using the resistor correctly. This allowed for a much brighter light but for a much shorter period of time. I like this circuit for playing with because it is nearly indestructible and easily tweakable.
When you remove the led, the circuit is still oscillating, with the led in you are measuring the led and transistor current alternately, if you change the base coil to a switch, you can turn it on to see the transistor/coil current, turn it off to see any led/coil current, do this with the scope attached.
Having watched this I had a bit of a play session myself. I'm using 1400mah NiMh cells as a point of curiosity to see just how long they'll run before needing to be recharged. The end goal is to see if I can make my own solar light of sorts. This video helped a lot in that respect as I wouldn't have thought to put a capacitor there. Thanks for sharing this...
About 10-12v it WILL CONDUCT MUCH SOONER THAN 40V in reverse Because the LED is reverse biased and the inductor is inducing the voltage forwards into the LED and BACKWARDS into the transistor. Therefore with no LED. The transistor emitter-collector is getting a reverse voltage from normal and conducting at about 9 to 12v Thats a relaxation oscillator with a single dead AA battery Full explanation given here you go! Datasheet 2n3904 maximum collector emitter FORWARD voltage 40v when breakdown occurs reverse maximum is 10v when breakdown occurs
Well you have saved me a lot of trouble should I ever decide to make a JT! :) Maybe an exercise in greater efficiency (& further understanding of inductors) would to be to experiment with different tapping points? i.e. an asymmetrical amount of turns.
Good stuff, but I would've liked to see both the base and LED voltages on the scope, at the same time to better see what's going. And the frequency it's running at, etc. If you do make another video about this, I hope you'll scope it more!
My first functioning joule thief circuit today (no toroid, simply wound around a 1 cm diameter paper tube, if you must know), me all happy looking at the waveform on my DSO-138 when Julian let's out with "DSO-138 just won't work in this case." Nooby smile go to nooby frown.
concerning the 10v break down come to mind. 1. you are reaching the dv/dt induced break down. 2. the transistor breakdown voltage was reduced due to previous breakdowns when you removed the led before
The cores in question are most likely crappy at the speeds you want to play with to push a 2N3904 to near its limits They are good to above 200MHz with moderate collector currents. The common sorts of cores stop looking like cores well before that. In the distant past, I made an air core step up circuit. It was basically an Armstrong oscillator with a schottky rectifier. I'm sure none of my neighbors could watch TV until I moved it inside a box.
You stated that you would be revisiting this project... may I ask that you show us more with the scope... it would have been interesting to see the signals going to the base of the transistor with and without the capacitor. And what if you changed the winding ratios on the feedback portion of the circuit vs the tank coil.... what about a cap across the tank circuit... and its effects.... I find this to be a very interesting circuit, and yet so basic... thank you for making this video.
I had a few lamps but they don't have any electronics inside. Are PSU and TV's and radio's the other best option to find ferriet? I also found out that there is a color scheme, green ones are bad, yellow is good, white is best, blue is good. Black is probably iron.
After you double the length of magnet wire, you should put one end in a drill, and while holding the other end spin it into one 'double' wire--much easier to wind. (I put a piece of protective tape to prevent drill-end damage). That wire is a bit thick for burning off the enamel--too much heat---use xacto
Hi Jullian wonder what frequency the cct is oscillating at...the wonderfull strange world of RF comes to mind would also be interesting to see just the led current. ..adding the capacitor obviously changes the frequency of oscillation and as a ham i used to play around with RF to light Fluorescent tubes. ..time for another video i think....Cheers Steve Liverpool
What a weird circuit! On mine the LED is oscillating on for a few seconds and then off for a few minutes. I'm guessing it's because not enough windings around my paper tube core. Ran out of wire.
can you put 1 channel of your scope on the bottom of inductor 1 and likewise with inductor 2. test 2 can you simultaneously check current through Led and b.e current on transistor and then remove led. I have a suspicion transistor gets hotter when led is not in circuit
If only someone would invent a fantastic little sleeve that could harness all the power from a battery like this it would be AMAZIIIIIIII... Oh wait....
@Julian. A very noob question regarding joule thief...is ferrite toroid ring is necessary or any iron piece like nail or screw can be used to make the inductor. Also will there be any change in efficiency?? If possible please make a video on this too.
23:47 Your transistor on time is high, so more energy is getting wasted. Change base resistor values to increase or decrease frequency. Or give less turns on base side of the coil.
27:00 A 300 % reduction doesn't mean a reduction to 1/3 of the original current, that would be only a 66 2/3 % reduction. A 300 % reduction is impossible as reducing by 100 % would result in zero current! :-))
Maybe the "300% reduction" means it is producing more than it is using... it is a free energy device! ;-P I see people making these simple "reduction", "saving", "bonus" calculations wrong all the time. People that should know better in the sciences or consumer media. They don't seem to understand that if something cost 50% less than the other doesn't mean the other cost 50% more.
Reminds me when I use to make 4to1 balms to use on twin line for amateur radio antennas. Can you imagine if you had to make a bunch of them for a project. Would it still work if you had a few turn twisted while you made the turns.
I have a 3-LED flashlight that runs on two AA's which can light the LED at full brightness down to 1v no problem I use two rechargeables It uses a very similar joule thief circuit. except the inductor only has TWO LEADS. And the transistor shaped object gets warm. the inductor gets warm And about 400mA is drawn from the batteries when fully charged. When they're dead its still pulling enough to make the three LED's pretty bright (straw hat LED's very wide light angle. but theyre in a flashlight reflector so its more of a smooth beam)
I wonder how this would work if you tried to charge a supercapacitor off of it instead of an LED? You'd probably need to add something to limit the capacitor to lower voltages if you wanted to run something at 5v, though. I'll see if I can figure out how to use a 5.1 v Zenner diode to pass the proper voltage without going over voltage.
Julian - What about combining wireless transfer with a joule thief to allow a detachable LED to burn a long long time when connected. Maybe a coil doing double duty ?
Hi Julian, thanks for the video. I was wondering on the waste aspect of batteries when used in devices. It seems almost as though this might be able to be used to get more of the stored energy in the batteries to keep a device, such as a battery radio, going longer until the batteries drain more completely. Would this possibly work? If so there is a possibilty of adding this small amount of circuitry, with or without the LED, into the battery powered device in the battery circuitry without damaging the device and saving on wasted energy and making it truly useful other than shining an LED just to drain the batteries? Seems like a worthwhile consideration. I'm not trained in electronics, so forgive my ignorance if there''s an obvious reason why this type of circuit couldn't be used as I suggest.
+ Pyotr Leflegin It's called a bifilar winding and is as old as the hills - a standard, relatively easy way of winding two coils of the same number of turns on a core. Also the "start" of any coil is normally marked on a schematic with a dot to indicate the "polarity" of the coil so that the induced voltage polarity (direction) can be established easily. In this case, in Julian's schematic the left hand coil would have a dot at the top (common connection point) and the second (right-hand coil) would have the dot at the bottom. This is why the end connections have to be cross connected for the common point and explains why, without doing that, the coil will not work as expected since the induced voltage (from the collapsing field) will be in the wrong direction. The circuit is basically an oscillator and in order to oscillate requires positive feedback. Connecting the coil the way Julian first had it produces negative feedback thereby effectively (and actually) preventing oscillation, i.e. why it wouldn't work at first.
Thank you for your very clear explanation. I used to be an electrician but this kind of thing wasn't really in my job. Again, thanks for the advice ;) !
I built one with no resistor and used a electrolytic capacitor and a odd winding 45 turns one way 10 the other it has a slow oscillation of 64 Hz and makes a high pitch squeal , it is the most efficient one I have built to date , so far I have built around 3 dozen of all types ( I mount them into project boxes with a rocker switch and give them away when I'am done experimenting with them ) . I scope each one and do a frequency count and check the milliamps . The oddball was only pulling .188 milliamps the lowest to date . I keep everything the same as far as the transistors , torroids & LEDs ( I bpought 200 off of eBay for less then $ 15 US and the exact voltage / current . Transistor = 2N 3904 B331 / T25-52 Iron powder toroidal core .255"/6.48mm x.120"/3.05mm x.096"/2.44mm / 5mm Superbright White Round LED 20000 mcd and the same wire on most 30 AWG except the one I made with the odd number of winding's for the 45 turns I used a heavier wire of 22 AWG and the 10 turns I used 20 AWG . I keep the LEDs , transistors , resistors , wire and torroids the same for my base line , having a oscilliscope helps to see the waveform' peaks on the high side I'am getting 6.55 volts from the one with no resistor and a electrolytic capacitor 16 volts 2200 uF , some of the pones i built have a oscillation of 450 to 500 Hz . If I had a better scope I could get more accurate results , HINT HINT
Julian, could it simply be that you can't really trust your ammeter. The shape of the current is neither DC nor harmonic AC, not sure how your device can make a meaningful reading out of it. What might be happening is that when you remove the LED the shape of the current changes and even though the actual consumption drops the ammeter does not see that. You can prove it one way or the other if you put a small resistor in series with the battery and use your scope to assess voltage shape across that resistor. And with a reasonably simple calculation you can estimate the effective current draw with and without the LED.
If it goes up to 10v, you could, maby, use it to charge some lipo batteries with the remainings of dead batteries, maby adding a diode to prevent the battery from feed the circuit on it's own (a shottky one, due to the frequency obtained), and who knows, maby a zener set at 4.2V to avoid over-charging the cell.... I know that charging a lipo battery with dead AA cells could take a huge amout of them, but if, maby, you want to achieve something more than lighting up a led with it, it's maby worthed the effort....
Just a quick note in case someone would like to try this and not have easy access to a toroid core: Don't be discouraged if you don't have the exact components listed here; just about any reasonable setup of inductors can work just like this center tap transformer that is built in this video. I have one on a little piece of perf board that is using a pair of through hole transformers...the kind that look like regular old resistors. Despite what some results on Google would lead you to believe, just about any configuration of these basic parts will probably work out for you. Part of the fun of this circuit is modifying it to change the output and seeing just what you can and can't get away with.
The transistor oscillator doesn't stop oscillating if you remove the LED, which means that when the transistor is switched on, it will conduct between Vcc and ground freely. That's where the current is going.
Having come here from Big Clive's video (chaining from atomic14's wireless charging video) I note the last thing he does is say don't let it go open circuit. The first thing Julian does is pull the LED. No punches pulled :D
Removing the LED will leave the oscillator running and may ruin the transistor if the voltage exceeds the CE voltage
i would think you are correct over time at the 10v it jumped to.. maybe if you used a full 1.5v guessing the voltage would be just less than 40v just under its rating possibly damaging the it faster.. but i haven't seen any tests on this.
Those things are marked 1-2-3 because the side ones are for power rails
The smallest breadboard
For this video my comments would be...1. You have multi channel scope and it would have been nice to see both sides of the inductor voltages at once. A serious problem when someone is not familiar with what a scope can do. 2nd point is, you need different turns ratio for two coils in order to optimize the overall current consumption. 3rd point is, your MPPT charging wont work at its best if you use that green coloured inductor. Thats a pure ferrite core and you need something with lesser permeability. This green ferrite is okay for making high value inductors but not suitable for buck/boost converters.
First of all, I wish you a happy new year.
I made a few joule thief circuits myself and used iron powder cores instead because after testing I found out that the iron powder cores are better because you're using the core to store energy in a magnetic field, because of the iron powder core the magnetic field is harder to flip around and that helps with efficiency and I also don't hear any ringing in the core as you normally get when you use a ferrite core.
I used the joule thief for two different reasons, the first was just to see how it works and if I could get it to light up some LEDs of an empty battery and I wanted to know how long the LED could run continuously of the dead battery, quickly I discovered that my transistors are getting way to hot when I use a 1k ohm resistor as you suggested in your video...
So I replaced the resistor with a 10k ohm one and even that was not a high enough value resistor, then I used a 100k ohm one and after that a 1M ohm one and my LEDs are just as bright with the 1M ohm resistor as with all the lower values but now I don't have the thermal issues anymore.
However I do have a second way that I'm using the joule thief circuit to get a string of 20 white LEDs to light up as bright as possible and I'm not using a battery power source but an old 800mA @ 5 Volt Nokia charger, that's for my helping hands with magnifying glass that I customized by adding 22 white LEDs to make my workspace much better...
Again I used the iron powder core but now I wanted to see what happens if I made the windings different by adding one more winding on the collectors side so I still have the center tap with one winding going round once in one direction to the resistor and the base of the transistor and two times in the other direction making it a 2:1 coil instead of the regular 1:1 coil and I also added an extra parallel connected transistor, a diode and capacitor which is connected to the 20 LEDs string...
It's not the most efficient circuit and it gets pretty warm to the touch but the 20 LEDs are lighting up very bright which is exactly what I wanted for that application, it will run on a 3.7 volt Li-Ion battery but not very long... it's completely drained in a couple hours which is an unexpected result, I thought it should keep running for days but it doesn't if you're pushing the LEDs to the limit.
All in all I find the Joule Thief a lot of fun to play with and it's probably the most forgiving circuit of all, I did manage to burn a few LEDs and transistors but that was totally my own fault because of fault wiring or driving the components way out of spec just to see what happens... LOL 😆😆😆😆💥🆒
Put an oscilloscope on that transistor. I think you'll find that when you jump the resistor with a capacitor, the "resonance" (for lack of a better word) of the transistor drops. That's probably why you see a drop in power draw.
I wish you'd of put the scope back on when you added the caps. Great vid. I hope we all learned about inductors
Cool, I built my first working joule thief two days ago now, following Big Clive's instruction to the letter. Works great, I have a few old ones laying in boxes somewhere that I tried to build but failed so I was quite pleased to get one working. In our motor home, which we use to travel the USA and spend the winters in areas of the states that are without snow, thus the nickname for folks such as ourselves "snow birds" we use tons of AA batteries, in nightlights, my metal detector, and even my label maker. So I always have lots of the blasted things laying around, and my main tool is a battery tester. So now I have a home built light that uses the old batteries instead of tossing them. I plan on building a small holder for my thief so it can be used as a light beside my easy chair in the rig, just to take the dark off when I actually put a tiny slide switch on mine, and used big Clive's method of casing with hot glue.
I guess a lot of your things could use Eneloop NiMH's instead of primary (single use) AA's?
Cap could function as a resonance provider in C in LCR circuit, or lower the frequency by increase capacitance of the LCR circuit.
Julian: its fun to build coils (chukles)
Me: remembers masters degree school, design and calculation of impedance matched coils at microwave frequencies (cries)(a lot)(in pain)
Yep at microwave frequencies even a round trace vs a square trace on a printed circuit board can have a noticeable effects on signal attenuation, leaks etc. ahhh the fun of microwaves. :D
@@BillAnt All of us who love the electronics we are masochists XD
Wondered when our "Jules" was gonna become a jewel thief....
Seriously, I was working on an MPPT-type thingy for an environmental company in Finland.
Reason? They had data loggers in the wilderness, powered by 9Ah SLA batteries. Their method was to stick a 10-watt panel in series with a silicon(?) diode (Schottly?? Never heard of her, mate!) and either NO charge in winter, or just boil the hell out of the batteries in 24-hour daylight in Summer...
Which is why your Arduino Muppet project was of interest...
BUT, Winter...I can still get a few volts out of the panel, and use it to charge a BFC (Big..er...Friendly...Capacitor), then dump it into a Joules Thief, give the battery a bit of a kick...Then, when Summer kicked in, turn the whole mularkey to a MuPPeT device.
Sadly, my work-experience ended, and the company couldn't even afford the necessary insurance for me (as unemployed) to work for free. RATS!
Cool story - thanks. I am thinking of putting the joule thief on a supercapacitor.
...Actually, once the "BFC" had got enough charge, I'd use a Unijunction Transistor* to trigger a FET switch to dump the BFC's charge into the 'theif - then to the battery.
*UJT - (2N4656) - That'll get your younger readers scrabbling in Google! Us old-timers (pun intended) know what they are!
+Andy Crofts
UJTs. Haven't seen them around/used for quite some years. Useful little device. I wander what have taken the role now a days?
Does your diy ammeter measure (T)RMS current? If the current oscillates @25kHz you probably are measuring garbage...
What? What are you trying to say? Sounds like you are trying to be critical of someone doing more than you...
@@adamwade855 I think he refers to the fact that the oscillating voltage is not correctly reported by most multimeters in dc mode. RMS just refers to the method of getting an approximate readout for the oscillating signal.
I suspect the voltage is being clipped to 10V due to the ferite core being saturated, not any form of leakage/breakdown of the NPN junction in the transistor.
From memory, adding a capacitor raises the turn-on voltage. So it won't work on very low battery voltages unless you only add the capacitor after the battery is connected.
I made plenty of these on my cement battery experiments. I found that if you put a electrolitic capacitor between the + and - of the cell the led will flash. This happens when the battery is almost expired. keep up the great videos.
your scope is dual trace right ? Take one probe and put on the base. One on the collector. compare the timing of the pulses on base to pulses at the collector with and without the led and you will see biasing of the transistor, pulsed biasing, turning on and off the transistor. second video maybe ? if you see what I think you will see, you can explain on the video.
That's actually an interesting idea, to see the interaction of the base trigger signal vs the collector's potential. My guess is that there's a few microseconds difference around a 16khz oscillation.
*Joule thief Joule thief* ua-cam.com/play/PLh8HTLB-VWMkqsKIpgXFHg5sJDN5eGqtN.html
Love the header pin mount for the ferrite Lego style micro boards...who knew?
While you have one channel of your scope connected to the collector, try connecting the other channel of your scope to the base of the transistor. That may help explain things while you have the LED out of circuit.
Suggest that you wind only one length of wire and after ten turns or so make a loop and then keep winding in the same direction and that will make your center tap without the mucking around. So much easier
A very mysterious circuit, love it's simplicity but complex operation! Took me ages to figure it out when I first came across it and it's hard to explain.
I've seen the notation for this style of inductor drawn as a transformer but with dots marked next to the coils to indicate one winding is opposite. Then in this case the top 2 connections are commoned to positive.
You're almost there, i find breaking time down slow motion like you did helps a lot. Remember that the field produced by the first coil induces current in the opposite direction in the second coil and vice versa.
I wonder how measuring the voltage at collector to ground effects the operation, I'm guessing that the load of even a meter must effect things?
Those CFL bulbs seem to last much longer if they are mounted facing upwards rather than downwards from the ceiling, it must be the heat
I built a LED dog collar with a Joule Thief to power 4 LEDs. I used two inductors (they look like dark brown resistors and also have the colored rings) instead of a coil to fit it all into a thin transparent flexible tube. It works perfectly for 60 - 100 hours with a single AA battery, even under water. It runs the battery down to about 0,6 Volts until the LEDs get too dark.
I love it. It is better than all LED dog collars, i ever bought and cost me about USD 2 for parts and 1 hour of my time :)
You are a true maker :)
Thank you :)
I tend to overengineer everything i make and really enjoyed this one evening non-AVR project in contrast to the complex machines i usually build.
The collector winding provides negative feedback, just like in a Blocking Oscillator, to turn off the transistor. Then the whole thing repeats.
You can try changing the code in your ammeter, so it averages out the reading
The C-E voltage rises only to about 10v. At this point, the voltage induced at the second coil (left side) is enough to trigger the transistor ON.
I think the led would burn out instantly, if you'd put it back in at 40v.
This is a useful demonstration of back EMF from an inductor.
Using a fixed value resistor is pointless use a potentiometer and fine tune it untill the led at its brightest use a small value cap across the resistor after fine tuning you will find two sweet spots where brightness is maximum remove the pot measure the ohms select a resistor closest to the meter reading connect the cap. Across the resistor and you have max brightness for min current
Yes, that is really important. I did that and the LED is much brighter and the circuit is more efficient.
Not a bad idea, with a pot one could easily find the "sweet spot" of brightness and consumption which is a functions of the charge/discharge duty cycle. However we can't violate the laws of energy conservation no matter how hard we try it, we can only make it a bit more efficient.
At 25:18 he's mentioned Colin Mitchell having been able to cut the power usage to about a third, which may be possible with a precise selection of resistor/capacitor values, including a high efficiency inductor, and most importantly a very efficient LED (expensive one). Also, putting a "snubber" cap across the LED to smooth out longer duty cycles of the oscillator may help too.
Jullian, in your search for the reason your flyback voltage without the LED limits at a value lower than you expected, check the Vebo spec limit and recall you have a coupled inductor...........
At 25:18 he's mentioned Colin Mitchell having been able to cut the power usage to about a third, which may be possible with a precise selection of resistor/capacitor values, including a high efficiency inductor, and most importantly a very efficient LED (expensive one). Also, putting a "snubber" cap across the LED to smooth out longer duty cycles of the oscillator may help too.
This tiny circuit has been around since 1999 (and possible even before that in some other variation), with its remarkable simplicity and efficiency has spawned countless videos. It's a really nice little project for even beginners to try, well if it wasn't for having to make the coil by hand. :D
What's the actual inductance on those things? Does your little component tester tell?
A few tens of microhenries per winding. I'm getting in the region of 20 to 40 uH in my experiments. The capacitor really makes a difference. It allows you to increase the resistor to 10k or more but it works best if you move the resistor from between the coil and the transistor base to between the positive supply and the secondary coil. Then put the capacitor between the junction of the resistor and coil and the negative supply. The best value seems to be somewhere between 1nF and 10nF. For such a simple circuit there's much to experiment with. The windings don't need to be symmetrical. You can use fewer turns of finer wire for the secondary (base) coil, which will allow room for more turns or thicker wire for the primary (collector) coil. You don't need such a large ring to wind the coils on. A tiny ferrite bead with 8 or 10 turns will also work. You can also experiment with the transistor as they have different voltage drops when fully turned on Vce(sat). The BC337 seems to be a particularly good choice. It's a fascinating circuit because, while almost anything will work there a huge scope for refining it and squeezing every last joule of energy out of that "dead" cell. More esoteric variations include using a germanium transistor because it will continue to oscillate when the cell voltage drops lower than 0.7 volt and substituting a MOSFET to reduce the bias current to almost zero - though in that case the gate bias voltage is a problem, requiring additional circuitry or a couple of button cells to provide it.
That is awesome how you built the inductor onto that breadboard friendly header!! Nice!
Have you ever checked the frequency that they are oscillating at? Mine was 170 KHZ, but the other one I built is 142 KHZ so there is a difference in the windings in each inductor and I figure that is why my two joule thief's have a different frequency.
Julian, the coil is acting as a load and the current is so low is because of the pulse width modulation
If you're persistent enough you can fit a joule thief into regular tungsten light bulb case (without resorting for smd parts!) and make any old torch a LED torch that runs even on flat batteries. Good fun.
By adjusting this resistor you can maximize light output and tune the circuit to the inductor you've made.
Another neat trick is to use it for super emergency lighting because this little jewel (pun intended) can easily power 5m led strip out of dead batteries.
Isn't it simply the case that with the transistor on, it provides a lower resistance path to ground, so current stops flowing through the resistor turning the transistor off again, and that's how it oscillates?
I'm running a small pice of 12V LED strip. (2 × 3 LEDs) from a single AA cell. This simple circuit just keeps on giving.
I have measured up to 50V across the CE junction (with no LED connected) on my scope.
I was able to effectively double the current lol...by not using the resistor correctly. This allowed for a much brighter light but for a much shorter period of time. I like this circuit for playing with because it is nearly indestructible and easily tweakable.
When you remove the led, the circuit is still oscillating, with the led in you are measuring the led and transistor current alternately, if you change the base coil to a switch, you can turn it on to see the transistor/coil current, turn it off to see any led/coil current, do this with the scope attached.
Having watched this I had a bit of a play session myself. I'm using 1400mah NiMh cells as a point of curiosity to see just how long they'll run before needing to be recharged. The end goal is to see if I can make my own solar light of sorts. This video helped a lot in that respect as I wouldn't have thought to put a capacitor there. Thanks for sharing this...
About 10-12v it WILL CONDUCT MUCH SOONER THAN 40V in reverse
Because the LED is reverse biased and the inductor is inducing the voltage forwards into the LED and BACKWARDS into the transistor.
Therefore with no LED. The transistor emitter-collector is getting a reverse voltage from normal and conducting at about 9 to 12v Thats a relaxation oscillator with a single dead AA battery
Full explanation given here you go!
Datasheet 2n3904 maximum collector emitter FORWARD voltage 40v when breakdown occurs
reverse maximum is 10v when breakdown occurs
Well you have saved me a lot of trouble should I ever decide to make a JT! :) Maybe an exercise in greater efficiency (& further understanding of inductors) would to be to experiment with different tapping points? i.e. an asymmetrical amount of turns.
Yeah, I want to do more extensive joule thief experiments soon
Good stuff, but I would've liked to see both the base and LED voltages on the scope, at the same time to better see what's going. And the frequency it's running at, etc. If you do make another video about this, I hope you'll scope it more!
I will :)
when cfl first came out they lasted for ever
My first functioning joule thief circuit today (no toroid, simply wound around a 1 cm diameter paper tube, if you must know), me all happy looking at the waveform on my DSO-138 when Julian let's out with "DSO-138 just won't work in this case." Nooby smile go to nooby frown.
concerning the 10v break down come to mind. 1. you are reaching the dv/dt induced break down. 2. the transistor breakdown voltage was reduced due to previous breakdowns when you removed the led before
2) Transistor BREAK DOWN voltage hmmm yes to words seem to suggest just that ;)
thextremeking Thank you for your kind remark and correction.
I doubt he is running into dV/dT problems with a 2N3904. I know they live while doing:
dV/dT = 5V * 2 * pi * 100MHz = 6V/nS
Ken Smith hmm... this starts to be interesting. the voltage to consider is about 10v. I think i am going to do some experiments.
The cores in question are most likely crappy at the speeds you want to play with to push a 2N3904 to near its limits They are good to above 200MHz with moderate collector currents. The common sorts of cores stop looking like cores well before that. In the distant past, I made an air core step up circuit. It was basically an Armstrong oscillator with a schottky rectifier. I'm sure none of my neighbors could watch TV until I moved it inside a box.
You stated that you would be revisiting this project... may I ask that you show us more with the scope... it would have been interesting to see the signals going to the base of the transistor with and without the capacitor. And what if you changed the winding ratios on the feedback portion of the circuit vs the tank coil.... what about a cap across the tank circuit... and its effects.... I find this to be a very interesting circuit, and yet so basic... thank you for making this video.
Nice, time to open lamps and get out the ferriet. Good hint!
Dippo6905
I always do that 😂
What a coincidence : I saw something on the interweb yesterday which suggested that 'ferriets' make good pets. Well, I think it was 'ferriets'
Ferret is a pet. Ferret also like electronics, they tear everything apart. Really.
I had a few lamps but they don't have any electronics inside. Are PSU and TV's and radio's the other best option to find ferriet? I also found out that there is a color scheme, green ones are bad, yellow is good, white is best, blue is good. Black is probably iron.
Good idea! I muse rummage around in my junk box and see if I can ferret out a ferrite...:)
After you double the length of magnet wire, you should put one end in a drill, and while holding the other end spin it into one 'double' wire--much easier to wind. (I put a piece of protective tape to prevent drill-end damage).
That wire is a bit thick for burning off the enamel--too much heat---use xacto
lol that little ferrite ring's like a little pearl popping out of a clam.
Hi Jullian wonder what frequency the cct is oscillating at...the wonderfull strange world of RF comes to mind would also be interesting to see just the led current. ..adding the capacitor obviously changes the frequency of oscillation and as a ham i used to play around with RF to light Fluorescent tubes. ..time for another video i think....Cheers Steve Liverpool
What a weird circuit! On mine the LED is oscillating on for a few seconds and then off for a few minutes. I'm guessing it's because not enough windings around my paper tube core. Ran out of wire.
can you put 1 channel of your scope on the bottom of inductor 1 and likewise with inductor 2. test 2 can you simultaneously check current through Led and b.e current on transistor and then remove led. I have a suspicion transistor gets hotter when led is not in circuit
If only someone would invent a fantastic little sleeve that could harness all the power from a battery like this it would be AMAZIIIIIIII... Oh wait....
Hello Julian. The symbol for a ferrite core is dotted lines ======
you don't have to make inductors, you can though.
really good video Julian, understand a lot more about the joule thief now, thanks.
watching the last few videos, any bets on when Julian will sharpen that pencil?
Yes yes, that's always and exciting thing to watch for. :D
@Julian. A very noob question regarding joule thief...is ferrite toroid ring is necessary or any iron piece like nail or screw can be used to make the inductor. Also will there be any change in efficiency?? If possible please make a video on this too.
Bread boards have parasitic capacitance between the plates. That might explain the fly back not going as high as you expected.
23:47 Your transistor on time is high, so more energy is getting wasted.
Change base resistor values to increase or decrease frequency.
Or give less turns on base side of the coil.
Try putting a piezo disc as a resonator or across coil to keep constant voltage.
Thanks for the video Julian, with this video I am finally starting to wrap my brain around this.
Loud and clear, a lot off practicing is required. Thanks.
The CFL bulbs don't last here in the USA either where we have 120volts.
I am wondering if the scope's 1M ohm input impedance loads the circuit down enough so that you only saw 10 volts?
As soldering goes that was reasonably steady
Damn, never thought of using Blu-tack like that before, I'll give myself and uppercut later.
27:00 A 300 % reduction doesn't mean a reduction to 1/3 of the original current, that would be only a 66 2/3 % reduction. A 300 % reduction is impossible as reducing by 100 % would result in zero current! :-))
LOL
Maybe the "300% reduction" means it is producing more than it is using... it is a free energy device! ;-P I see people making these simple "reduction", "saving", "bonus" calculations wrong all the time. People that should know better in the sciences or consumer media. They don't seem to understand that if something cost 50% less than the other doesn't mean the other cost 50% more.
This is what happens when you learn maths from the batteriser datasheet. :P
Silver Legend on Talking Electronics it says improvement of 300%. See www.talkingelectronics.com/projects/LEDTorchCircuits/images/LEDTorchCctB.gif
That's as illogical - an improvement of 100% would be a reduction to zero!
i have a light bulb shaped camping light that uses one AA and I think it must use a joule thief to make it so bright.
Reminds me when I use to make 4to1 balms to use on twin line for amateur radio antennas. Can you imagine if you had to make a bunch of them for a project. Would it still work if you had a few turn twisted while you made the turns.
If it's kink free then it can't be Big Clive style.
I have a 3-LED flashlight that runs on two AA's which can light the LED at full brightness down to 1v no problem
I use two rechargeables
It uses a very similar joule thief circuit. except the inductor only has TWO LEADS. And the transistor shaped object gets warm. the inductor gets warm
And about 400mA is drawn from the batteries when fully charged.
When they're dead its still pulling enough to make the three LED's pretty bright (straw hat LED's very wide light angle. but theyre in a flashlight reflector so its more of a smooth beam)
thanks for the video. now I can muck about with the pile of inductors in my drawer.
I wonder how this would work if you tried to charge a supercapacitor off of it instead of an LED? You'd probably need to add something to limit the capacitor to lower voltages if you wanted to run something at 5v, though.
I'll see if I can figure out how to use a 5.1 v Zenner diode to pass the proper voltage without going over voltage.
finally made my own joule thief, i have been wanting to build one. it is easier that i thought.
Brill! I was hoping to see a scope in the mix with this one and didn't come away disappointed : )
I figured-out how to hook-up a PNP transistor for the joule thief circuit.
I'll explain it to anyone that asks.
loved the video. really helped me understand inductors as well. look forward to seeing more..
Julian -
What about combining wireless transfer with a joule thief to allow a detachable LED to burn a long long time when connected. Maybe a coil doing double duty ?
Hi Julian, thanks for the video. I was wondering on the waste aspect of batteries when used in devices. It seems almost as though this might be able to be used to get more of the stored energy in the batteries to keep a device, such as a battery radio, going longer until the batteries drain more completely. Would this possibly work? If so there is a possibilty of adding this small amount of circuitry, with or without the LED, into the battery powered device in the battery circuitry without damaging the device and saving on wasted energy and making it truly useful other than shining an LED just to drain the batteries? Seems like a worthwhile consideration. I'm not trained in electronics, so forgive my ignorance if there''s an obvious reason why this type of circuit couldn't be used as I suggest.
im not sure what that all meant but im learning little by little tks from an old guy from canada :)
What happens to brightness and current consumption when a fresh new battery is used?
Very good! One of the best JT posts I've ever seen -- I really like the way you wound the inductor!
+ Pyotr Leflegin It's called a bifilar winding and is as old as the hills - a standard, relatively easy way of winding two coils of the same number of turns on a core. Also the "start" of any coil is normally marked on a schematic with a dot to indicate the "polarity" of the coil so that the induced voltage polarity (direction) can be established easily. In this case, in Julian's schematic the left hand coil would have a dot at the top (common connection point) and the second (right-hand coil) would have the dot at the bottom. This is why the end connections have to be cross connected for the common point and explains why, without doing that, the coil will not work as expected since the induced voltage (from the collapsing field) will be in the wrong direction. The circuit is basically an oscillator and in order to oscillate requires positive feedback. Connecting the coil the way Julian first had it produces negative feedback thereby effectively (and actually) preventing oscillation, i.e. why it wouldn't work at first.
Thank you for your very clear explanation. I used to be an electrician but this kind of thing wasn't really in my job. Again, thanks for the advice ;) !
great tutorial on wiring of the inductor
Julian, have you tried to run a series of LEDs? It would be interesting to see it light 2 LEDs, etc. and see how the current changed.
Big Clive used a string of LED's and it seemed to work fine. I am not sure if they were parallel or series. I would think parallel.
cool I built my first one in 1993 ! crammed it into a space of 25mm by 10mm and it used a BFY50 !
I built one with no resistor and used a electrolytic capacitor and a odd winding 45 turns one way 10 the other it has a slow oscillation of 64 Hz and makes a high pitch squeal , it is the most efficient one I have built to date , so far I have built around 3 dozen of all types ( I mount them into project boxes with a rocker switch and give them away when I'am done experimenting with them ) . I scope each one and do a frequency count and check the milliamps . The oddball was only pulling .188 milliamps the lowest to date . I keep everything the same as far as the transistors , torroids & LEDs ( I bpought 200 off of eBay for less then $ 15 US and the exact voltage / current . Transistor = 2N 3904 B331 / T25-52 Iron powder toroidal core .255"/6.48mm x.120"/3.05mm x.096"/2.44mm / 5mm Superbright White Round LED 20000 mcd and the same wire on most 30 AWG except the one I made with the odd number of winding's for the 45 turns I used a heavier wire of 22 AWG and the 10 turns I used 20 AWG . I keep the LEDs , transistors , resistors , wire and torroids the same for my base line , having a oscilliscope helps to see the waveform' peaks on the high side I'am getting 6.55 volts from the one with no resistor and a electrolytic capacitor 16 volts 2200 uF , some of the pones i built have a oscillation of 450 to 500 Hz . If I had a better scope I could get more accurate results , HINT HINT
Wow, you're a joule thief guru!
No just really bored
Julian, could it simply be that you can't really trust your ammeter. The shape of the current is neither DC nor harmonic AC, not sure how your device can make a meaningful reading out of it. What might be happening is that when you remove the LED the shape of the current changes and even though the actual consumption drops the ammeter does not see that. You can prove it one way or the other if you put a small resistor in series with the battery and use your scope to assess voltage shape across that resistor. And with a reasonably simple calculation you can estimate the effective current draw with and without the LED.
there goes the world's supply of 32 swg tinned copper wire
So how much does the mA drop when you remove the LED with the capacitor with the resistor?
If it goes up to 10v, you could, maby, use it to charge some lipo batteries with the remainings of dead batteries, maby adding a diode to prevent the battery from feed the circuit on it's own (a shottky one, due to the frequency obtained), and who knows, maby a zener set at 4.2V to avoid over-charging the cell.... I know that charging a lipo battery with dead AA cells could take a huge amout of them, but if, maby, you want to achieve something more than lighting up a led with it, it's maby worthed the effort....
What if you put a capacitor instead of an led?
Remove the LED and you are still ramping up the current in the inductor.
you can use a joule thief to charge batteries as well.
Nathan Dean Hmm, so if it were running off of a nearly empty NiCd or NiMH, it could charge it back up again? That'd come in useful.
Where did you get the ammeter?
Such high votiage the current might be pushes back to the battery?
So it works by just pulsing the LED really fast at a usable voltage so it looks constant to your eye?