An Improved Joule Thief--An Unruly Beast Tamed?




About: I am a retired analytical chemist living with my wife Cynthia in Cornwall, south west England. I have held the UK radio amateur call sign G3PPT since 1961. I have been interested in computing since the day...

Few simple electronic circuits have given as much fun to so many as the Joule Thief and vast amounts of internet bandwidth have been expended on its construction. It is with some trepidation that the author is proposing to add to this!

Teacher Notes

Teachers! Did you use this instructable in your classroom?
Add a Teacher Note to share how you incorporated it into your lesson.

Step 1: The Basic Circuit Before Improvement

Virtually every Joule Thief circuit that you can find will be based around a toroidal inductor that has two parallel windings with the start of one winding being connected to the finish of the other. There is a tendency to make the toroidal transformer as small as possible. Some may have noted that the way the inductor is connected into the circuit with the common connection to the two windings being taken to the supply line is similar to the centre tapped secondary winding of a mains transformer and then wonder if such a mains transformer could be used in a Joule Thief circuit.

The question is asked here for instance:

This is one of those instances when it is easier to 'try it' rather than prevaricate and accordingly the circuit shown in the first picture was assembled on a breadboard and shown in the second picture.

The parts are non-critical. Specifically, the mains transformer is rated at 15-0-15 Volts 250mA A/C with a 240 Volt A/C primary which is not used at this stage. This one was bought new from the Maplin chain in the UK but almost any small transformer with a centre tapped secondary winding will do and a suitable item could be found in the junk box or salvaged from old equipment. In some countries like the USA your transformer is likely to have a 110VAC primary but I do not think that this will greatly alter the performance of this circuit. The transistor is a BC549 but again, not critical. The 1000uF electrolytic capacitor simply decouples the supply line.

The result here is a noticeably well tempered Joule Thief that powers the LED with supply voltages down to 0.4Volts. Old Joule Thief hands will be horrified by the size of the inductor even though the one used here is a small example.

Immediately apparent is an inherent disadvantage of the basic Joule Thief circuit which is the total lack of control. At low supply voltages you get efficient generation of low light output that may be useable for some applications. As you increase the supply voltage with better batteries the light output escalates but with it the supply current increases disproportionately which means that better batteries will have their life unnecessarily shortened. a new battery might even burn out the LED. I have tabulated some readings in the third picture above.

This circuit is crying out for some form of control!

Step 2: Adding Simple Control--Stage 1

I will add the control components in stages explaining as I go.

So far I have have used the centre tapped low voltage winding on the transformer and now the other winding is brought into play. This winding serves as the mains input when the transformer is used in normal applications.

I have connected a full wave bridge rectifier across this winding to rectify the alternating voltage generated by the Joule Thief action. This alternating voltage is asymmetrical and will be a quirky waveform and a bridge rectifier ensures that we get the best positive voltage yield regardless of which way round we connect to the winding.

The bridge rectifier is a frequently encountered electronic configuration and I have made it here from individual IN4004 diodes as I had some to hand and the voltages generated by the circuit are fairly low.

The circuit diagram in the first picture shows our progress so far and the second picture shows a close up of the bridge. The white ring on the diodes indicates the cathode where the positive voltage appears.

The blue wire connects the negative side of the bridge to earth out of the picture.

Step 3: Adding Simple Control--Stage 2

We can now connect a 100k potentiometer from the positive output of the bridge rectifier to earth and then a 1uF 35V electrolytic capacitor in parallel with this as shown in the circuit diagram in the first picture. Leave the potentiometer wiper contact floating for the present. Note also that the anti-clockwise potentiometer connection is connected to the bridge rectifier output--this is counterintuitive but otherwise you will get maximum light output when the potentiometer is fully turned to the left.

The second picture shows an enlarged view of the additions.

The circuit thus far can be tested by connecting a battery and observing that a positive voltage is built up across the potentiometer and electrolytic capacitor. In this case a 0.52 Volt battery yielded around 5 Volts and a 1.3 Volt battery yielded 19 Volts. These measurements answer a query that some may have in that surely high voltages should be expected across what is the high voltage input of the transformer. In practice this is not an issue. The transient voltage peaks could charge a small high voltage capacitor up to 50 or 60 Volts but here the tiny 100kOhm load of the potentiometer can drop the voltage to safe low levels. Further we cannot extract any usable power from this part of the circuit--it is for control only.

Do not be tempted to run the circuit without the potentiometer because as stated above, the voltage can rise sufficiently to destroy the electrolytic capacitor.

Step 4: Adding Simple Control--Stage 3--Closing the Loop

Here we see the final circuit. The circuit diagram is in the first picture and the breadboard layout is shown in the second.

R1 is the resistor connected to the base of oscillator transistor Q1 and note that it has been increased to 10K.

A second general purpose NPN transistor Q2 is added with the collector connected to the base of Q1, the emitter to earth and the base to the potentiometer wiper via 10K resistor R2. The circuit functions by feeding positive voltage back from the bridge rectifier to Q2 which is turned on thus reducing oscillator amplitude. The base/emitter junction has a threshold of about half a Volt and Q2 does not conduct until this threshold is exceeded giving us a very crude form of control which is adequate for our purpose. A Zener diode would give a sharper function but the circuit has to generate sufficient voltage to make it conduct, that said, Zener diodes are available down to 1.8 Volts or so and might be worth experimenting with.

The LED output can be controlled by the potentiometer setting but only up to a level dictated by what the battery will give and do not expect a 0.5 Volt battery to give you a searchlight! On the other hand a new lithium AA cell can give a very bright light indeed.

Step 5: Performance of the Improved Circuit

The table in picture 1 shown above gives the current through the improved circuit for various battery voltages and is a drastically improved picture over the results for the basic circuit in Step1. The control is not brilliant but shows that the beast has indeed been tamed. One difference is that the control now available has allowed batteries up to around 2.5 Volts to be used which is unthinkable with conventional Joule Thief circuitry. You can now consider placing a couple of tired batteries in series to get up to 2.5 Volts and then use this combination all the way down to 0.5 Volt but please take precautions against battery leakage when discharged to such a low level. Note that you cannot go above 2.5 Volt as there is DC path through the oscillator transformer to the LED and the LED will light up directly from the battery. (Perhaps putting two LED's in series would be an interesting experiment.)

The current values are drastically lower making the circuit very efficient and the 0.1 mA current consumption at minimum setting means that batteries can actually recover whilst operating giving the prospect of the circuit operating almost indefinitely or at least until there is battery failure due to catastrophic leakage.

There is the possibility of putting similar dead batteries in parallel.

Step 6: Some Different Power Sources

The circuit is now very versatile in in what it can accept as a power source.

The first composite picture shows the circuit operating with button cells. On the left it is an AG4 and on the right an AG12. In the UK ,and no doubt elsewhere, selections of these cells are sold on cards cheaply. In my case after the AG1 cells have been used in my watch I am left with a number of others for which I have no obvious use. The AG4 will run this circuit at minimum setting for some three days.

In the second composite picture on the left we see a couple of 'tired' AA cells which have been taped together and soldered in a series combination giving us 2 Volts in this case. The ability of the circuit to run at up to 2.5 Volt makes it potentially very useful and could enhance the way we use dead cells. You could try two lots of two cells in parallel and then put these in series. You may find it advantageous to put any combination of cells in a seal-again plastic bag with just the wires poking out to guard against leakage.

On the right of the second picture we see the circuit running from a 3 Farad Super Capacitor charged to 2 Volts. This will run at minimum setting for around 5 hours.

Step 7: To Finish

The use of a mains transformer will surprise if not horrify many Joule Thief aficionados but it does seem to have merit. There is much more iron in which to build up the magnetic field and the laminated construction with soft iron must help. If you are able to replicate what is shown here then you have the basis for a superb bespoke night-light for a child and it could be designed so as to contain the transformer in the base of the construction. A lighthouse perhaps? At minimum setting a single new AA alkaline cell or even two in parallel might have sufficient longevity to outlast a child's need for a night-light.

At times the performance of a Joule Thief seems so good as to defy basic laws but it may be possible to partially explain this. The above picture shows the oscilloscope waveform taken from a 10 Ohm resistor in series with the LED in our circuit. The current through the LED is occurring in needle sharp pulses with a sloping trailing edge and a width of 10 to 20 microseconds and these are spaced 600 microseconds apart. Persistence of vision means that the eye/brain combination sees the peak and not the bits in between. This is similar to how we see patterns when a sparkler is waved about. This is outside my field of knowledge so I merely offer it up as a suggestion.

Thank you for reading this contribution.

Be the First to Share


    • Made with Math Contest

      Made with Math Contest
    • Multi-Discipline Contest

      Multi-Discipline Contest
    • Robotics Contest

      Robotics Contest

    11 Discussions


    Question 1 year ago on Introduction

    Lionel Seer - I must say you have managed to present a very successful model with lots of options to the JT original version. My one question which overhangs in my mind all the time, is that, if this modification or the original JT circuit -

    1. Can it produce more power on the Output than the Input from Batteries.

    2. What if we want to SCALE up this version to use 24v (2 x 12v Marine Batteries) then drive an Inverter to produce AC 240v / 50Hz and power some serious Home Appliances.

    I understand if both my questions are irrelevant to the topic and/or if you don't have answers to such out of the box questions. In anycase, thank you very much for enlightenment.

    1 answer
    Lionel Searsamtani2013

    Answer 1 year ago

    Hello samtani2013 - Apologies for my late reply but I have had problems recently logging on to Instructables.

    1. You can never produce more output power than that supplied by the batteries--we must live our lives by basic physical laws :-)

    I think that the spiky waveform produced by this circuit makes current measurement extremely difficult and this is where false hopes are generated. Proper measurement of the efficiency of a Joule Thief circuit is therefore vital for serious experimentation and this is the way that I do it.

    Make sure that the power input is well decoupled using a high value electrolytic capacitor e.g. 1000 microfarad connected between the supply rail and ground. This will ensure that the supply current and voltage can be properly measured and then the power into the circuit is given by:

    Supply voltage multiplied by supply current (V * I)

    To calculate the output power I place a 10 microfarad electrolytic capacitor in parallel with a 10 Ohm resistor between the LED and ground and then measure the voltage developed across the resistor/capacitor combination. You will need a quality digital voltmeter to do this as a 1 milliamp LED current will generate only 10 millivolts across 10 Ohms. Hence you will see that the output LED current is given by :

    Voltage across the 10 Ohm resistor multipled by 10 (the value of the resistor)

    To calculate the output power in the case of this 1 milliamp: we know that the voltage across a working white LED is 2.5 Volts, (you may like to measure it accurately and separately with a power supply and resistor in series,) and then the LED power is 2.5 X 0.001 = 2.5 milliWatt.

    The circuit efficiency is given by (LED power / power in) times 100.

    In practice I am finding that a well tempered Joule Thief can have an efficiency of 50 to 70% with a supply battery voltage of over 1 Volt but this efficiency drops away as the supply voltage drops down even though the circuit is still producing light.

    2. Mains transformers can be used to make transistor inverters but they are not designed for this duty and the efficiency will not be very high especially with a small transformer.


    3 years ago

    The mortal coil reels from such heresy, you have chosen to blaspheme accepted science by divulging forbidden knowledge, would that such a dangerous act rip a hole through the time/ space continuum and cause all mankind to perish.

    prevaricate; now that's an interesting word, thanks for sharing ti. ☺

    2 replies
    Lionel SearBeachsideHank

    Reply 3 years ago

    Is divulging forbidden knowledge going to destroy widely held illusions? You can take away all of a person's worldly possessions but never take away their illusions.

    Sorry for being dilatory in replying and thanks for prompting me to consult dictionary and thesaurus. Mea culpa, I should have said procrastinate and not prevaricate!

    In fact I always wanted to be a procrastinator but never quite got round to it.

    BeachsideHankLionel Sear

    Reply 3 years ago

    Ahh but prevaricate so aptly describes our current aspiring politicians' stance on matters, it boils down to "vote for me 'cause I suck less than my opponent". ☺


    3 years ago

    if you conect about 50 to 100 micro henr ferrite coil on + your batery, you wil be able to boost your jt and reduce usage of batery (sorry about my english) this is from Hartlay oscilator and works.


    3 years ago

    the idea of a joule thief is to get every last joule out of a battery so the idea is to make then run on the lowest amount of current possible but give lots of light so the centre tap side of the transformer would be used on the joule thief but you would not add the led the led would go on the high voltage side of the transformer thus make the led brighter when at the lowest input voltage

    1 reply
    Lionel Seardasimpson1981

    Reply 3 years ago

    A good point but you have to take into account the resistance of the high voltage winding in question. For this small transformer it is a massive 990 Ohm so you are not going to be able to draw much power out of that, however it does provide the small amount we need to control the oscillator.

    Even the secondary windings of this transformer measure 12.5-0-12.5 Ohm so there are some associated losses.

    If you need to extract power from a Joule Thief circuit then I feel that you need a separate low resistance winding and if you can find a mains transformer with two such windings e.g. mains in and 0-12V plus 0-12V out then it could be a useful tool with which to experiment.

    There is not much power to be had but I was able to drive a clock with a Joule Thief based on a toroid, see :


    3 years ago

    Thank you for looking beyond the proverbial fence.


    3 years ago

    so the idea is the old batterys in that tv remote that no longer power it can be used in the light and when they drop to a power they cant run the light you put them in series and so on untill they really is no joules left in them at all


    3 years ago

    Ty for posting.