You are on page 1of 10

The "FLEET" ("Forever Lead-out Existing Energy Transformer") device is a self-powered

electrical generator which has no moving parts and which can be constructed cheaply. It has
been developed by a Hong Kong based team of people: Mr Lawrence Tseung, Dr. Raymond
Ting, Miss Forever Yuen, Mr Miller Tong and Mr Chung Yi Ching. It is the result of some years
of thought, research and testing and it has now reached an advanced stage of testing and
demonstration and is nearly ready for commercial production.
Mt Tseung has applied his "Lead-out" theory to the category of low-power circuits known as
the "Joule Thief" circuits. These circuits originated with an article by Mr Z. Kaparnik, in the
"Ingenuity Unlimited" section of the November 1999 edition of the "Everyday Practical
Electronics" magazine.
The initial circuit allowed the very last energy to be drawn from any ordinary dry-cell battery,
and used to light a white Light-Emitting Diode ("LED") for use as a small torch. It allows a
battery which is considered to be fully discharged, to drive the circuit until the battery voltage
drops right down to 0.35 volts. The initial circuit uses a bi-filar coil wound on a ferrite ring or
"toroid". Bi-filar means that the coil is wound with two separate strands of wire side by side, so
that each adjacent turn is part of the other coil. A coil of that type has unusual magnetic
properties. The Joule Thief circuit is like this:

It is important to notice how the coil is wound and how it is connected. It is called a "toroid"
because it is wound on a ring. The ring is made of ferrite because that material can operate at
high frequencies and the circuit switches On and Off about 50,000 times per second ("50
kHz"). Notice that while the wires are wound side by side, the start of the red wire is
connected to the end of the green wire. It is that connection which makes it a "bi-filar" coil
instead of just a two-strand coil.
This "Joule Thief" circuit was then adapted by Bill Sherman and used to charge a second
battery as well as lighting the Light-Emitting Diode. This was achieved by adding just one
more component - a diode. The diode used was a 1N4005 type because that was to hand at
the time, but Bill suggests that the circuit would work better with a very fast-acting Schottkytype diode, perhaps a 1N5819G type.
The circuit produced by Bill is:
When driven by a 1.5 single cell
battery, this circuit produces about 50
volts with no load and can supply 9.3
milliamps of current when the output is
short-circuited. This means that you
could charge a 6-volt battery using a

1.5 volt battery.

Gadgetmall of the Joule Thief forum has taken the circuit further and
found a very interesting situation. He has modified the circuit and used a batt-cap which is a
very high capacity, very low-loss capacitor. This is his circuit:

He has added an additional winding to his one-inch (25 mm) diameter ferrite toroid, and he
uses that to power a 1 watt LED. Why he has done this is not immediately clear to me, except
possibly, that it shows when the circuit is operating. He runs the circuit driven by a small
rechargeable battery, which feeds 13 milliamps into the circuit, for a period of fourteen hours.
At the end of that time, the batt-cap has gathered enough energy to fully recharge the driving
battery in a minute or two, and then power a heater winding of nichrome wire (as used in
mains-powered radiant heaters) for four and a half minutes. Alternatively, that amount of extra
power could boil a kettle of water. The really interesting thing about this is that the driving
battery gets recharged every time and so the circuit is self-sustaining although it is not a
powerful circuit.
However, Mr Tseung has taken the Joule Thief circuit and modified it to become a circuit with
a very serious output, moving it into a completely different category.
As a first step towards what the team calls their "Fleet" device,
the toroid has been enlarged to a much greater diameter. The
coil is now wound on a section of plastic pipe, 170 mm (6.5
inches) in diameter and 45 mm (1.75 inch) deep:
This section of pipe is "bi-filar" wound with two wires side by side
as already described for the Joule Thief construction. As before,
the start of one wire is connected to the end of the other wire.
Then, the winding is given a layer of electrical tape to hold it in
place and to provide an easy working surface for a second
The wire used for the winding is the widely available red and black pair of wires, sometimes
called "figure of eight" because the cut end of the wires looks like the numeral 8. The wire
should be able to carry 2.5 amps. It must be side-by-side wire and not one of the twisted
varieties. It
looks like this:
The second winding is made in the same way
but the connections are slightly different. As
before, the end of the first wire is connected
to the start of the second wire, but that
connection is then insulated and not used in
the following circuitry. This just connects the
two windings one after the other, known

technically as being connected "in series" and is the equivalent of making the winding with
just a single strand of wire. The completed coil may look like this:
This particular design is still in it's early stages and so many different coils sizes and
constructions are being tested:
The arrangement is for the inner winding of
the toroid to be oscillated by the Joule Thief
circuit already described. This causes a
pulsating magnetic field to envelope the outer
winding of the toroid, producing an electrical
output which is capable of doing useful work.
The really important thing about this
arrangement, is the fact that the amount of
power coming out of the circuit is very much
greater than the amount of power needed to
make the circuit operate. The additional
power is led out of the local environment and
drawn into the circuit, becoming available to
do useful work.
The overall circuit then looks like this:

While the outer winding is shown here with thicker wire of a different colour, this is only to
make the arrangement easier to understand. In reality, the outer winding is with exactly the
same wire as the inner winding, and it will normally go all the way around the toroid. The total
amount of wire needed to make the windings is about 70 metres and so it is normal to buy a
full 100 metre reel of the twin-core wire, which allows both windings to be made and leaves
spare wire for other things.
For those of you who are very technically minded, the
output waveform looks like this:
and the voltage pulses in this output are occurring about
290,000 times per second.
What has worked better for me is using a bridge of four
diodes rather than a single diode:

A Suggestion
(Placed in the Public Domain on 26th October 2012 and so cannot be
This is an arrangement of readily available, cheap parts for operating a useful light. Probably
the most efficient circuit for this task is the Joule Thief circuit and the light bulb usually
chosen for this circuit is the compact fluorescent lightbulb (CFL) which is widely available
around the world and although there are many variations, looks something like this:
The arrow is pointing to a join in the construction where the bulb section joins the lower
section which houses circuitry and the mains connector which can be one of many different
types. However, that type of bulb has mains circuitry built into it which means that in order to
get the higher efficiency which we would like, each bulb needs to be physically modified which
is far from ideal. The circuitry inside a bulb of this type, converts the alternating current of the
mains into Direct Current and then uses that direct current to supply an oscillator circuit which
generates high-frequency pulses which power the bulb. Unfortunately, that circuitry gets in the
way of low-power operation and so needs to be removed. People who have done this, say
that the bottom can be popped off just by inserting a screwdriver into the crack and twisting
the screwdriver to force the two pieces apart. If you try that, then let me wish you luck as it
has never worked for me, even when applying enough force to permanently damage the
material on both sides of the crack. Instead, I do it by cutting the plastic base 8 mm below the
arrow line resulting in this:
The 8 mm clearance is because the glass tubes project down below the join and we need the
remaining section of the plastic housing to support the bulb when we mount it on our circuit
box. The cut should be made when holding the base section as the glass tubes are very
fragile and easily damaged. If available, a Dremel or similar tool which has a small cutting
disc very well suited to making this cut. The cut should be just deep enough to go through the
plastic wall but not further than that. There is a tiny circuit board contained in the lower part,
generally, with some very nice components which can be used for other circuits. The cutdown light bulb can be mounted on the circuit box using a strong adhesive, or a hole can be
drilled very carefully in the centre, between the tubes (holding the base and not the tubes
when doing this) and the base can then be bolted to the component box.
The glass tubes are U-shaped and there are two inter-tube horizontal connecting tubes in
order to make all of the U-tubes function as one long zig-zag discharge path and get all of the
tubes lit up at the same time. The two tubes which do not have this horizontal interconnecting
tube, have two wires coming out of the bottom of them, and used to connect the circuit to the
tubes. These four wires need to be cut, leaving them as long as possible, then each pair has
the insulating enamel scraped off them and then soldered to a length of wire which will be
used to connect to the new circuit, or alternatively, direct to the board if very small fiddly work
is no problem.

This adaption makes this low-cost method unsuitable for applications where an off-the-shelf
bulb is needed so that replacements can be made without the need for any technically skilled
person to be available. What we need therefore, is a bulb which does not have the mains
circuitry (called ballast) built into it, and the earlier PLD bulbs are in that category.
Unfortunately, they are more expensive and not so widely available. They look like this:

These have the enormous advantage of not requiring any work on the bulb in order to operate
immediately with our circuit. We can build a suitable circuit from scratch, but it is very popular
with home constructors making a one-off experiment to use the very cheap circuit found in
Fujis Quick Snap disposable camera which is very widely available. It looks like this:

There are various ways of adapting the circuit board found inside the camera, and thanks is
due to Gadgetmall of this forum for sharing his method of adaption and expertise with these
circuits, which has allowed him to run a fluorescent light for 38 hours powered by just one AA
size 1.5 volt battery
Disclaimer: This document must not be considered to be a recommendation for you to
actually attempt to undertake any of the following modifications, and should you decide to do
so, then any loss, damage or injury are wholly your responsibility and not that of anyone else.
The camera needs to be taken apart in order to get at the circuitry inside it. A word of warning
here, there is a high-voltage capacitor inside the camera and if it happens to be charged, then
it is quite capable of giving you a really nasty shock, so as soon as the circuit board is
exposed, I strongly recommend that you take great care to avoid a shock, even though it is
not likely to be a fatal shock. As soon as the capacitor is exposed, then short across its wires
using a metal tool which has a plastic handle, such as a screwdriver or pair of pliers with a
well-insulated grip. If the capacitor happens to be charged, then that may produce a bright
spark which makes a loud crack.
The camera is taken apart like this:
1. The green covering which is a piece of very sticky, strong plastic is peeled off. The join
is underneath, where the black colouring ends.
2. In the middle of the bottom, there is a flap which you lever up, uncovering the battery.
There are various varieties of Fuji Quick Snap camera with different circuitry, the one
shown here was supplied by Asda (Wal-Mart in the UK) in 2012 and the circuit board is
marked A07 or A60 while some earlier versions have a different layout for some
components and even have the battery inserted the other way round. Before you
remove the battery which in the UK is a 1.5V AAA alkaline battery, make a careful note
of which way round the battery is inserted. In this case, the plus of the battery connects
with the long copper arm. Remove the battery.
3. Pull off the black plastic covers on the underside of the camera, located at each end of
the battery compartment, and then using a screwdriver, force the two halves of the
black camera case apart, which leaves the front of the camera looking like this:
4. Make sure that the flash is not charged, first, by using a non-conducting item to press

together the switch contacts marked A in the following picture, and then using some
metal object with a plastic handle, bridge across the gap between the soldered points
ringed around and marked B as they are the ends of the high-voltage capacitor. If the
capacitor happens to be charged, then there will be a spark and a loud sound, but this
is unlikely with a new camera unless you have been pressing the buttons since
unwrapping it.
5. Press the black plastic latch marked C in the picture above, over towards the left and
that releases the circuit board which can be lifted out and looks like this: And seen from
the top:
6. The board is quite small, being about 40 mm x 25 mm when the capacitor and flash
unit are removed which is the next thing to do, probably by cutting the very tough
capacitor leads and then cutting off the plastic pins holding the flash unit to the board,
levering it upwards and cutting away its metal contacts which link it to the board.
7. The two switches which form the cameras flash-charging switch and its shutter
release switch need to be wired permanently closed. These are marked Switch 1 and
Switch 2 in step 5. above. I suggest that you cut the arms of Switch 1 to about half
length, clamp them together with a pair of long-nosed pliers and bend them over to
crimp them flat, and then solder them together. Switch 2 needs to be bridged across to
make it permanently closed. A clip can be used to clamp the upper and lower contacts
together so that they can be bridged with solder.

8. The rest of the modification is seen from the top of the board: This is to achieve this
Gadgetmall circuit:


Gadgetmall comments that increasing the voltage supplied to this circuit, puts the 2SD1960
transistor at risk as it will overheat. That transistor is tiny, has no heat sink or space to fit one
and was only intended to be on for a few seconds while the capacitor charges up in
preparation for operating the flash bulb. This circuit modification runs the transistor
continuously for very long periods and so we are already stepping outside the Fuji circuit
designers operating conditions. Also, we would like to run the circuit with somewhat higher
voltage in order to get improved operation of the fluorescent tube or bulb. Consequently, we
might consider using a more powerful transistor. The 2SD1960 transistor is rated at 30-volts,
5-amps, 170 MHz and 0.75 watts, so we might consider swapping it for, say, a BD245C
transistor rated at 100-volts, 10-amps, 3 MHz and 80 watts as our circuit runs at under 0.1
MHz and the BD245C transistor can be mounted on a heat sink, although with its much
greater handling capacity, it should stay cool at these tiny powers. We can boost the gain of
the BD245C by a factor of 200 or so, by using a BC109C or a 2N2222 transistor to form a
Darlington pair, making the circuit:

At this point we notice that the only component from the Fuji camera circuit is the ridiculously
tiny transformer. However, as that transformer is very cheap and since it appears to work well
driving all sorts of fluorescent loads, it does appear to be a worthwhile component in spite of
its minute size. The wire sizes used in the transformer are very small, with the #26 AWG wire
having a diameter of just 0.4038 mm, the #32 AWG wire a diameter of 0.2032 mm and the
#45 AWG having a ridiculously small 0.0447 mm diameter, which means that twenty turns of
that wire laid side by side cover less than one millimetre! There is, of course, a strong
temptation to wind a ferrite-cored version of this transformer, using larger diameter wires for
greater reliability and current handling capacity. That would not be difficult to do as under
1800 turns are involved and the voltages are well within the wire insulation capabilities.
However, for the moment we
might just use two of the cheap
Fuji transformers arranged like
While the circuit shows a long
fluorescent tube being driven by
each of the output transformers,
any type of fluorescent bulb can
be driven, and for the long-term
practical considerations, one of
the PLD plug-in push-fit types
would be the optimum choice for a
This Joule Thief style of circuit is
very efficient and draws relatively
little current, so if we can replace
that small current as it is being
drawn, then we will have a self-

powered light. There are relatively cheap solar panels measuring just 68 mm square and
supposedly providing five volts with a theoretical maximum output of 70 milliamps. It would be
excessively optimistic to expect one of these panels illuminated by a fluorescent bulb to
produce anything like that amount of current, but with several panels, it should be possible to
achieve the necessary current output for continuous operation, especially if the panels are
very close to the bulbs. They are 2.5 mm thick and they look like this front and back:

We want to
these panels
so that they
collect as
much light as
possible and yet do not obscure the users light
unduly. One obvious position is to use two panels
back to back between the two bulbs, where each
bulb would obscure the light coming from the other
bulb anyway:
Here, each bulb illuminates the area cast in shadow by the other bulb, giving full 360 degree
all-round light as well as giving maximum illumination to two panels. Also, as the bulbs are
taller than the panels, the top of each bulb also illuminates the whole of the area.
Seen from the side, it looks like this:

However, whether or not two panels provide sufficient current to meet our needs, (and it
needs to be remembered that these bulbs become
dimmer as they get old), we would be tempted to
add additional panels anyway, in order to provide
additional current for other things. Very few light
bulbs are arranged to give all-round light as most
are mounted on or near walls, ceilings, etc. So we
might well be tempted to double the number of
panels like this:

This arrangement is also very good in that is obscures very little of the light which the user
would consider to be useful. With four panels, it is unlikely that the current needed for
continuous operation of the light bulbs would not be generated by the panels. However, if we
provide a USB output socket, the unit could act as a solar panel charger during the day if
placed in the sun. (Actually, if we provide enough panels, it might well be able to do that
during the night as well). Such an output socket could allow devices such as eBook readers,
mobile phones, LED torches, etc. to be charged and that would be a useful additional feature
for people who live without mains electricity.
The circuit arrangement might be like this:

The solar panels are connected directly to the USB output socket and that allows them to act
as a straight solar charger when placed in sunlight. The USB connections are like this:

These views are what the user sees when looking at the outside of the socket. Pin 1 is the
+5V power line and Pin 4 is the 0V or Ground line. These sockets are readily available and
are not expensive. The B type is often used when connecting a printer to a computer. If both
an A type and a B type of socket is provided, then a standard A to B USB cable could be
used to connect two units together for greater charging capability when placed in sunlight.
In this very simple circuit, a diode D passes the output from the panels directly to a large
capacitor C which charges up quickly to about 0.7 volts less than the panel voltage. There is
no need to add in over-voltage protection as the panels are not able to provide more than five
volts (unless you were to connect them in series). The diode prevents the USB socket load
from discharging the capacitor. Late at night, when the light is turned off and the user goes
asleep, the capacitor remains charged as there is no load on it at all. If the capacitor is of
good quality, then it will maintain its charge for a long time, ready to kick off the circuit
operation the next time that the light is switched on. There are alternatives to this. For

With this arrangement,

three AAA dry cells are
provided along with a press-button press-to-make switch. The user switches on and then
taps the press-button switch to get the circuit running. The batteries should have their normal
shelf life as the usage is so trivial. An alternative would be to use three rechargeable batteries
(fully charged when given to the user) as the circuit will keep the rechargeable batteries
topped up at all times, in daylight if at no other time, although there should be sufficient output
from the panels to provide the absolutely minimal battery power used. If the battery holder is
easily accessible, it might well be possible to use the device as a charger for small

rechargeable batteries, again, in sunlight if at no other time.

Bulb voltages of 110V or 220V are only related to the circuitry built into the base of the more
common bulbs, the actual tubes which light up are all the same. However, it should be noted
that all fluorescent tubes generate radiation other than light and so users should stay more
than one foot (300 mm) away from a compact fluorescent tube and double that distance when
powering the long fluorescent tubes (which can also be driven by this circuit and which have
large areas above the tube where panels can be mounted without obscuring any part of the
tube or tubes. Please remember that compact fluorescent light bulbs contain highly poisonous
mercury and so must not be just thrown away as they need to be disposed of carefully to
avoid contaminating the local water supply. If a compact fluorescent bulb breaks, then have
people and pets leave the room. Do not walk over the affected area. Ventilate the room before
you start the clean-up. Mercury vaporises readily at room temperature. Open all windows and
leave the room for at least 15 minutes. The broken pieces can be disposed of as normal
waste as the mercury has already vaporised.
For people in the UK, there are these suppliers:
The 5-volt, 70 mA 68 mm x 68 mm solar panels: from eBay trader Discount Devices Ltd. At
3.68 delivered.
The black ABS box 118 x 98 x 45 mm (Code 400-560) from for 2.11, the
BD245C transistor 1.18, heat-sink 0.78, ceramic disc capacitor 0.08, preset resistor 0.15
and the 2N2222A transistor 0.23, all plus VAT plus delivery.
The 18 watt 4-pin PLD fluorescent light bulbs from local hardware store Wilkinsons for 2.95
each. A production version of this light would have bulb sockets
( but this prototype just had
the top of the box cut appropriately and the bulbs inserted and glued in place.
The mesh used for the ventilation holes was cut from a 1.49 Splatter Guard of 32 mm
diameter, intended to make sure that nothing escapes from a frying pan when it is in use. The
adhesive used was Evostick Impact.
Other panel arrangements might be as shown here:

Long tube fluorescent tube fittings usually have plenty of spare space above the tube for
mounting the solar panels and the fixing box room for the (insulated) circuit board:
Users should not come closer than 0.5 metres to a fluorescent light fitting of this type, for
more than a few minutes at a time. Consequently, the fitting is usually positioned high up,
which may make a cord-pull On/Off switch the appropriate type for this style of fitting.