Simple Free-Energy Devices

There is nothing magic about free-energy and by “free-energy” I mean something which produces output energy without the need for using a fuel which you have to buy.

Free energy devices have been around for a very long time now. I have stood beside a water mill and it’s power is scary as it could crush you in moments and never even notice. That mill is on a quietly flowing small river and it can operate at any time day or night without having to pay anything for the power which it uses. It will have cost quite a bit to build the mill in the first place, but after that, it produces major power year after year. Most free-energy devices are just like that as it costs to build them in the first place, but after that they run for free.

This presentation is mainly for people who have never come across free-energy and know nothing about it. So, each chapter deals with just one device and tries to explain it clearly.

Chapter 6: The South African Generator

This self-powered generator started out as a device with a rotating disc (called a “rotor”) which required considerable skill to make accurately, but it passed through several versions and eventually ended up as a motionless design which is easy to make.

The first version put out 40 watts of mains voltage, mains frequency AC and so could power ordinary household equipment such as lights and fans. The prototype was run continuously for three weeks (500 hours) before being modified to give greater output. This is the circuit for the first version:

The “Load” was a 150-watt inverter which converts 12-volts DC to mains voltage and frequency AC, which in South Africa is 230 volts at 50 cycles per second.

The circuit is very simple. The rotor “C” is given a spin. It has five magnets inserted in it and as one of the rotor magnets passes the triple-wound coil “B” it creates a current in winding “1”. This switches the transistor on, powering coil winding “2” using current supplied by battery “A”. That transistor pulse reinforces the current in coil winding “1” and creates a current in coil winding “3” as well as pushing the rotor magnet away, maintaining the rotor spin.

The current in coil winding “3” is rectified by the diodes shown in blue and the resulting DC is fed back to the driving battery “A”, maintaining it’s charge.

When the rotor magnet has moved away the transistor switches off and that generates a high voltage at the Collector “c” of the transistor. That high voltage is siphoned off through the Ultra Fast UF5408 diode and used to charge all five batteries in the circuit. So, this circuit not only sustains itself, outputs 40 watts of mains power but it also charges three additional 12-volt batteries. That is very impressive!

However, in addition to charging three spare batteries like this one:

The developer wanted a greater level of continuous output power and so he modified the circuit to this:

This is a major change, with four of the five batteries removed. The one remaining battery is not being used to power the inverter but instead it is a passive component which holds the circuit’s feedback to about 12-volts in order to protect the inverter. The circuit produces an output of 60 watts of mains power continuously. A mains transformer steps the voltage down to charge a 100 microfarad capacitor to 42 volts and that is used by the transistor to pulse the coils. Four additional coils are used for extra output which is fed back to the capacitor which powers the circuit. This arrangement works very well, sustaining itself while outputting continuous power. You will notice that now the transistor powers six separate coils.

The triggering method used with the transistor in this circuit is really not the best and it needs careful adjustment to get it’s very best performance, so the developer changed the arrangement and switched to using two solid-state Hall-effect magnetic sensors and ten small coils. This gave a major improvement, more than doubling the output to 150 watts of continuous power:

The rotor has five magnets as before but with the two sensors pulses are triggered ten times per revolution instead of the original five pulses per rotation. The developer also switched to using a Field Effect Transistor (“FET”) type IRF840 instead of the previously used powerful MJE13009 bi-polar transistor. Not shown here is the fact that the developer now uses two powering batteries, one to drive the circuit while the other is recharging. Those batteries get swapped over very five minutes or so. The ten coils are connected in two chains of five (because there are five magnets in the rotor) and so a pulse is generated every 36 degress of rotation of the rotor.

The circuit used is this:

A C5353 transistor is used to produce a strong pulse when either of the two A3144E Hall-effect sensors is triggered by a passing rotor magnet, and that switches the IRF840 FET on.

This is an excellent circuit and used to charge a battery bank it could provide adequate electrical power for most of the needs of a household. However, the rotor needs to be made very accurately, either by lathe or perhaps by using a 3D printer. One replicator shows his 3D printed rotor which is made in two halves which are bolted together. It looks like this:

However, the developer used his existing two sets of five small coils and instead of driving them with a rotor he drove them with a simple 555 timer circuit. He found that he got the same level of output without his rotor if the coils were pulsed at 40% on and 60% off. This is a major step forward as it now becomes a motionless, solid-state circuit which is easy to build and need no special equipment. There is also now no restriction on the number of coils which can be used as they no longer have to fit around a rotor. Extra coils produce extra output power.

If we want to build one of these generators for our own use, then just initially, we will use low frequency (due to assumed iron core coil limitations) and run the testing using a circuit of this type:

The resistor “R” and the capacitor “C” control the frequency of the pulsing and the result is very good. However, as the developer has powered both coil chains of his rotor circuit from a single transistor (although they generate at least 600V feedback pulses), he used just one IRF840 transistor driving his two coil sets for his tests. He also likes to use his circuit which swaps over two drive batteries, one to provide current while the other one is recharging, but that is a minor matter.

So, let’s say for argument sake, that the above circuit is running at about 500 cycles per second (“Hz”). For this, C and R might be 100nF and 1.5K, in order to keep the coil frequency down, then there will be some 500 pulses per second returned to the drive battery. But, if we were to connect the circuit like this:

Then when the first transistor switches on, the second transistor switches off and vice versa. Doing that returns twice as many pulses per second to the drive battery without increasing the rate of pulsing of either of the coil chains. Remember also, that the transistors are powerful enough to drive several coil chains simultaneously, and each extra coil can be expected to increase the excess output power available.

However, testing shows that the output from the first transistor is not very good for switching the second transistor and so a better result is produced with the addition of a monostable circuit as that allows you to specify exactly what length of voltage pulse you want for the second transistor:

This technique of keeping the coils pulsed slowly while increasing the rate of pulses passed back to the output, can be extended further. It is perfectly possible to cascade ten or more coil chains during each of the 500 Hz pulses. That raises the output pulse rate without raising the coil pulse rate. This can be done by using a Divide-By-Ten chip, such as the CD4017B which can be wired to act as divide-by-9, divide-by-8, etc. down to divide-by-2. This is achieved by connecting the Reset pin (pin 15) to the next output. In the following circuit diagram, a divide-by-3 arrangement is shown and the divide-by-4 output is connected to the reset as that bounces the output back to output 1 again. The 555 clock is speeded up by a factor of three as it will take three times as long before the high voltage output of the 4017 chip returns to output 1 (on pin 3). The chip connections are like this:

For a divide-by-4 output, pin 10 would be connected to Reset pin 15 and the fourth output would be from pin 7 and the 555 clock pulse rate increased to four times the original rate by lowering the value of “C” or decreasing the value of “R”.

Please remember that the transistor needs to be able to handle high voltages if you decide to use a different type, also, you will need a more powerful DC/AC inverter to handle higher output power. There is essentially no limit to the output power you can achieve with solid state as you just add more coils and possibly more transistors. Please use a heat sink with each transistor.

If you decide to use a 24-volt input, please remember that both the 555 chip and the 4017 chip need to be kept down to 12-volts as they are not able to handle 24-volts. Also, you need a 24-volt inverter if you decide to do that.

If experimentation shows that your particular construction of the circuit works better at higher and higher frequency of clock pulses, and that results in each coil driving transistor needing a longer drive voltage period than the length of one divide-by-N clock period, then that can be dealt with by using a monostable on each output as shown by the shaded portions of this diagram:

Now that there is no need to construct a precision rotor with magnets, the only significant task is to wind the coils which generate the excess power. It is perfectly possible to wind perfect coils without any equipment at all. First, you need to choose the wire diameter and buy in the wire needed. Wire of 0.71 mm diameter is popular (swg 22 or AWG 21) and is easy to work with. Then you need to choose the core material – iron (not steel) or ferrite and create a spool with that core by attaching stiff flange discs of about 30 mm diameter at the ends of the core for iron. The coils shown here are wound on 8 mm iron bolts with windings 75 mm long, eight layers of wire and 40 mm diameter flanges (which could be much smaller):

Three of these coils can be wound from a single 500 gram reel of 0.71 mm wire and the iron cores can certainly operate at more than 6000 Hz. Each of these coils has about 315 turns and a DC resistance of 1.6 ohms. However, ferrite is generally considered to be a better core for high frequency operation and these can be wound quite easily Using the same 0.71 mm diameter wire (swg 22 or AWG #21), a 140 mm long ferrite rod of 10 mm diameter can be wound quite easily without any equipment, and six coils with three layers each can be wound from a single 500 gram reel of wire, and each coil has about 590 turns and a DC resistance of one ohm.

The basic ferrite rod has a 20 mm diameter disc of stiff cardboard glued to each end. It looks like this:

Cut a 140 mm wide piece of paper 32 mm long. This width matches the gap between the spool flanges. Attach a strip of Selotape to the paper so that it overlaps by half its width all along the paper strip and set it aside until the first layer of wire has been wound.

You can hang the full spool of wire on a rod hung from the edge of a table or desk. Push the first few inches of wire through a hole through the flange near the core and start winding by turning the spool in your hand. The winding needs to be done carefully so that the turns lie cleanly side by side with no gaps between them and no turns overlapping any other turn:

When the far end of the spool is reached, stick the piece of paper to the layer of turns using the Selotape already on the paper, bend the paper round the layer of winds and pull it tight using other strips of Selotape to hold it in place as you move progressively along the length of the spool. The paper will not be long enough to go all the way around the layer as the core now has the wire thickness making the core larger, but that is quite intentional as you don’t want more than a single layer of paper. You will need the paper layer to allow you to see the next layer of wire clearly as you wind it. If you don’t have that paper layer it is enormously difficult to see the next layer well enough to detect winding errors as the wire is exactly the same colour as the first layer.

You now have a perfectly wound first layer. Before starting the second layer, cut the next strip of paper, measuring 40 mm wide. Stick a strip of Selotape along the length of the paper, again, with half of the width of the Selotape overlapping the paper and set it aside. Wind the next layer in exactly the same way, finishing by sticking and securing the paper around the core with its two layers of wire.

That process is repeated until all of the desired layers have been wound. Finally, the wire is cut with a few inches left for connecting the coil in the circuit, and the wire is passed through a second hole in one of the flanges:

This generator can be built in thousands of variations, the main difference being the coils being used – the core material, the core length, the wire diameter, and the number of layers wound. You can, of course, start with one coil and see how your circuit performs, and later on, add one or more coils to boost the performance.

The way that coils perform is not at all obvious. It is generally agreed that the larger the number of turns, the greater the voltage produced when the coil is pulsed. BUT, other factors are also important. The impedance of the coil (it’s AC resistance) makes a very big difference when the coil is being pulsed. That is affected by the core material, the wire diameter, the wire material, the number of turns, the quality of the winding, how spread out the turns are, the number of layers, etc. Generally speaking, it is probably best to wind a series of coils and test them to see which works best for you, and then wind the remaining coils to match your best result.

If you wish to use two separate drive batteries, one to power the circuit while the other is recharging, then that is perfectly possible. Batteries which are providing power to a load don’t charge nearly as well as disconnected batteries which are being charged. However, the mechanism which switches between the two sets of batteries needs to have extremely low current draw in order not to waste current. One possibility for that would be to use a latching relay like this:

This is the electronic version of a mechanical two-pole switch. A brief pulse of current between pins 1 and 16 locks the switch in one position and later, a pulse of current between pins 2 and 15 locks it in the other position. The current drain on the circuit would be almost zero.

While standard NE555 integrated circuits can operate with a supply voltage down to 4.5 volts (and in practice, most will operate well at much lower supply voltages), there are several more expensive 555 ICs which are designed to work at much lower supply voltages. One of these is the TLC555 which has a supply voltage range from just 2 volts right up to 15 volts, which is a very impressive range. Another version is ILC555N with a voltage range of 2 to 18 volts. Combining one of those chips with a latching relay produces a very simple circuit as the 555 timer circuit is exceptionally simple:

The capacitor used has to be high quality with very low leakage in order to get this waveform which is On for exactly the same length of time as it is Off. This is important if we want the two batteries to receive the same length of time powering the load as the time they receive being recharged.

A weakness of the 555 chip timer from our point of view is that it has only one output while we need two outputs, one falling when the other rises. That can be arranged by adding a transistor and a couple of resistors like this:

With this circuit, when pin 3 of the 555 chip goes low, the capacitor connecting it to pin 2 of the relay pulls that pin 2 voltage low and causes the relay to change state as the relay pin 15 is connected to +12V, causing a current surge through the coil as the capacitor charges. A few moments later, when the capacitor has charged up, the current drops away to zero. Five minutes later pin 3 goes high again and that switches the transistor on causing its collector voltage to drop rapidly to near zero. That pulls pin 1 of the relay down low causing it to change state before the capacitor has a chance to charge up.

This is fine if the capacitors shown in blue are poor quality and their charge bleeds away in a period of five minutes. Nowadays, even cheap capacitors are generally much too good quality to allow that to happen and so we need to connect a resistor across the capacitor to create that drop in charge. But that additional resistor is connected continuously and so it needs to be of a high enough value not to waste any significant current – perhaps 18K would be a reasonable choice. An 18K resistor with twelve volts across it draws only 0.667 of a milliamp of current. The capacitors are about 100 microfarads.

So, if we prefer, we could use this circuit, perhaps laid out like this:

The TIP3055 transistors are there only to raise the current carrying capacity of the tiny latching relay. Let’s decide to build a very simple version of the circuit but allowing for later expansion for greater output power. Let’s try this circuit arrangement:

This arrangement allows for considerable alteration of the operating frequency by merely turning a knob. Experienced constructors will have their own preferred methods of construction, but we might choose to use a layout on an open board in order to make it easy to see what is happening and to give good cooling during the development stage, perhaps something like this:

This arrangement keeps soldering to a minimum and allows for easy alterations as the circuit is extended for higher output power. The timer board can be swapped out later on if you decide to use a Divide-by-N style of operation. It is suggested that a single, long thin coil shape is the most effective for this application but that has not yet been proved.

Two types of screw connectors are used. One type has all of the connectors connected so that many wires can be connected to a single point. They look like this:

Unfortunately, these connectors cost about £5 each which is several times more expensive than the standard connector which has each connector insulated from all of the other connectors in the block:

If cost is a major factor, then a standard connector strip can be converted to a single multiple output strip by wiring one side with a thick piece of wire like this:

We have a problem with connecting the FET transistors because their pins are so close together that they don’t fit conveniently into a screw connector block. We can get around that problem by cutting one connector off the block, bending the central pin of the FET upwards into a vertical position and using the single cut off connector to make the connection to the central pin of the FET:

The layout of the timer is not at all critical and a layout like this might be used:

The capacitor “C” will be about 10 nF and the variable resistor can be 47K or 50K linear or a higher value could be used.

So, if you were going to build this generator, how might you go about it? Well, you might start by building the timer board shown here, either as shown or to your own layout. I strongly recommend using a socket for the 555 timer chip as transistors, Integrated Circuits and diodes can easily be damaged by heat if they are not soldered quickly. As the generator is for your own use, you can avoid the horrible lead-free solder which is so difficult to work with and I suggest that 0.8 mm diameter multicore solder is the right size for this work. So, to construct the timer board you will need:

1. A soldering iron of about 40 watts, and 0.8 mm cored solder.

2. Stripboard (“Veroboard”) with 14 strips each with 23 holes.

3. A drill bit or a knife to break the copper strips which run between the pins of the 555 chip.

4. One 8-pin Dual-In-Line socket for the 555 chip.

5. Some solid-core plastic covered wire to form the jumpers on the board.

6. The components: One 555 chip, one 8-pin socket, one 1000 microfarad 25V capacitor, two 10 nanofarad ceramic capacitors, one 1K resistor, one 50K or 47K or higher linear variable resistor, one diode which could be 1N4007, or 1N4148, or almost any other diode.

7. A magnifying glass of some description. A cheap plastic one can be quite sufficient. This helps greatly when examining the underside of the board to make sure that solder joints are well made and that there is no solder bridging between adjacent copper strips.

8. A cheap digital multimeter for measuring voltages and resistance.

Not essential but very, very convenient is one of those angled arm clamping devices which are usually supplied with a magnifying glass. If you discard the magnifying glass, the angled arms can hold the board and component in place, leaving both hands free to do the soldering. A cloth wet with cold water is very good for cooling down soldered joints rapidly to prevent heat damage.

Start by breaking the copper strip in columns 10 and 11 on rows 6 to 9. This is needed in order to prevent the strips short-circuiting the pins of the 555 chip. Mount and solder the 555 socket in place (if you bend the legs outwards along their strips it holds the socket in place and makes for a good solder joint. Then, cut solid core insulated copper wire to the correct lengths and solder the five wire jumpers on the board:

Then work from left to right, mounting the remaining components. The capacitor “C” has got a lot of spare space around it so that it can be altered at a later date if you decide that you should.

Finally, connect the variable resistor (which many people mistakenly call a “pot”) and the positive and negative connecting wires using multi-strand copper wire as that is much more flexible, and lastly, the connecting wire from pin 3 out to the distribution block which connects to the FET gates. Check that the circuit has been connected correctly and that there are no soldering errors on the underside of the board – this is much easier with a magnifying glass as the gaps are very small.

Set the variable resistor shaft to about its mid position, connect the board to a 12-volt source of power and measure the voltage coming from pin 3 of the 555 chip. The voltage should be about half of the supply voltage and should not change much when you adjust the variable resistor.

We are now ready to start assembling the generator, getting a suitable board and attaching to it the inverter and the battery:

These two units can be attached to the base board by drilling holes through the board and using string or wire to bind them securely in place.

The timer board can be attached to the base board using a screw or a bolt. The board is very light and robust and a single screw is quite sufficient to hold it neatly in place. The variable resistor and the three connecting strips can be glued to the board. Some constructors hate the idea but my preferred method is to use Impact Evostick as the glue as it is very effective and after a day or so becomes very strong indeed.

The diodes used are 1N5408 types and although each one can handle 3 amps of current, they are grouped in sets of three as that lowers the very slight resistance to current flow through them as well as raising the possible current to nine amps.

My inclination is to use a separate FET with each coil, but the South African developer states that he can detect no difference between driving two coils with one FET and driving those same two coils with two separate FETs.

Patrick Kelly