Farnsworth Fusor

Fred's Fusor

High Voltage Power Supply

Contents of this page


In the Fusor project high voltage (HV) is applied as the power source for 'driving' the reactor and as the power source for the neutron detector. For the neutron detector we make use of one of my two HV power supply NIM modules (max. 3 kV and max. 5 kV) and no further development work is required for these modules, except making the connecting cables. Therefore on this page we will deal only with the HV power supply for the Fusor and we will do that in great detail to explore all potential possibilities.

triangle spin A warning is explicitely given here because the high voltage power from the devices discussed on this page may instantly kill by electrocution. Unskilled, unexperienced and/or HV knowledge lacking experimenters must not deal with these devices. shock

For the Fusor in order to obtain Deuterium fusion a voltage of ≥10 kV is needed.  However, for obtaining a sufficient neutron output, suitable for detecting and measuring, a HV power supply is required that delivers a fully variable, negative high voltage with a minimal capacity of 20 - 25 kV at 15 - 20 mA. These are the basic HV requirements for our type of Fusor. The positive lead of the Fusor power supply is always grounded. See reference 1.

Inert electrostatic confinement fusion is a voltage driven process and the objective of the project is to obtain proof of fusion. Proof of fusion can easier be obtained when the cross-section is made bigger by raising the kinetic energy delivered, i.e. the voltage in order to obtain more acceleration must be raised. At a high voltage of > 30 kV neutron production is better detectable but X-ray production is also raised. X-ray production starts at about 15 kV and at about 35 kV X-rays will start to penetrate the steel walls of the reactor, requiring lead shielding of the reactor.

The HV power supply for the Fusor requires a negative high voltage, fully variable, and this may be obtained from e.g. a variable autotransformer (or variac) powered high voltage transformer with the AC output rectified to DC. In case of a too low HV DC output this could be raised to the desired HV voltage level by means of a voltage multiplier cascade (Cockcroft-Walton generator).

For the operation of the Fusor is does not matter whether a constant DC voltage is applied or a high frequency DC voltage, as long as the basic HV requirements are fulfilled.


HV Transformers

X-ray transformer

X-ray transformers (XRT) are very suitable for constructing a HV power supply but difficult to find. An XRT can have ratings from 50 kV up to 300 kV. Usually an old dental X-ray machine XRT will be suitable for constructing a HV power supply and these XRT's are located in the head of the machine together with the X-ray tube (the head is the part that is brought next to your face when the dentist takes an X-ray picture). The head is filled with an insulating oil (before the 1970s containing PCBs) and care needs to be exercised when taking the head apart (reference 2).
An XRT may have 6 connectors, two connected to the primary coil for mains (the two thickest ones) and four for the secundary coils. One HV coil is at one side connected to the core and must be connected to ground and the other side is the HV output. The other coil is intended for measuring the current and could be connected to a mA-meter but when this is not required one pole must also be connected to the core. See image 1.

xray trafo
Image 1: X-ray Transformator (© FRS 2014)

The XRT cannot be operated as such in the open air (do not try this). It needs to be immersed in oil and any mineral oil will do. Immersion of the XRT in new oil and in a new container requires removing air bubbles from the oil and those trapped in the core windings. The easiest way to do this is by putting the container under a light vacuum for a while. This should not be problematic for a Fusion builder who has vacuum equipment at hand.

XRTs are not current limited and when directly connected to mains they will blow the fuse as they may draw over 60 Amps. It is imperative that the XRT is ballasted but in the primary of the transformer a resistor is not a good choice due to the heat that will be produced. An inductor is a better choice and for this purpose a shorted microwave oven transformer (MOT) can be used.  The primary of the MOT is than to be connected in series with the AC mains and the secundary coil is to be shorted (Image 2). An alternative resistor ballast load can also be connected to the secundary circuit instead of to the primary circuit as shown in the image.

xray trafo with ballast
Image 2: X-ray transformer with inductive ballast (© FRS 2014)

When developing a circuit containing an XRT it should be noted that the open (unloaded) voltage of an XRT is considerably higher than the rated voltage. An XRT rated 70 kV at 10 mA can at the output exceed 200 kV unloaded and the rectifier components should be able to handle that. It means also that the electricity from such a transformer jumps earlier and over larger distances than may be expected.
Being electrocuted by an electrical discharge at these high voltages and currents usually means instant death!


Development of X-ray Transformer Power Supply

X-ray transformers are not easy to obtain because the type of transformer that we need for the project is the rather old model, capable of continuous putput and powered by ordinary single phase 50/60 Hz power.
At an early stage of the project an X-ray transformer was acquired in an auction and it was at that moment decided to abandon our earlier chosen power supply design based on an FBT (see further below).

X-ray Transformer Power Supply

The X-ray transformer (XRT) power supply consist of the actual X-ray transformer, a variac or autotransformer, a rectifier circuit, a ballast resistor and the same additional components to be discussed in the next paragraphs.

X-ray Transformer

The XRT that was acquired is a Siemens, model unknown, rated at an output of 85 kV at 15 mA with an input of 220-240V (image 3).

Image 3: Siemens X-ray transformer (image: Orthochirus)

In image 3, at the right of the image, an additional (dark brown) coil can be seen (detail: image 4) which is the  step-down transformer for providing the power for the  X-ray tube filament (heater). This coil usually has a step-down factor of 10 to 20, yielding an output of 12 to 20 V.

Image 4: Siemens X-ray transformer, heater coil (image: Orthochirus)

The XRT, with a weight of 9 kg, must be operated under transformer oil. It is common practice for XRTs to have a low duty cycle, e.g. 1 : 50 which means one second on and 50 seconds off. It is therefore advisable to store the XRT in a large container with plenty of transformer oil in order to provide as much cooling capacity as possible. Even better would be to have the transformer oil rapidly cooled by means of an external cooling compressor and a high flow pump. This might compensate the problem of a low duty cycle and make the X-ray transformer suitable for use with a Fusor.
It should be noted that rectification with no capacitive filtering improves the duty cycle.

When the X-ray transformer arrived it was still dripping transformer oil from the core and first of all identification of the wires was started (image 5).

Image 5: X-ray transformer with input and output leads (© FRS 2014)

Apparently, the bundle of wires in the image at the right side of the transformer are the input wires, though after reversed engineering the transformer it appeared that probably the wires marked yi1 and bi1 are low AC voltage output wires for the X-ray tube filament.
These wires were grouped and the annotation in the image describes the groups of wires: yi1 + bi1 stands for yellow input 1 combined with blue input 1, etc. Wires in the output have been marked as follows: capitals Bo and Yo stand for the output of the big coils, respectively blue and yellow and the smaller letters yo1 and yo2 stand for yellow 1 and 2 output of the smaller coil at the left of the image.

Roughly a resistance measuring test was performed:
On the input side the YG lead (yellow/green) measured a resistance of 1Ω with the core, which is therefore connected to ground.
Also on the input side the resistance between yi1 and bi1 (yellow and blue in a yellow sleeve) measured 55.7Ω. When connected to mains (230V) this would draw just over 4 Amps or 950W.
The second yellow and blue set of wires on the input side (yi2 and bi2) measured 1.3Ω. It seems unlikely that these wires should be connected to mains.
Also on the input side is another yellow wire (yi0) which appears to have an unmeasurable resistance (nfinite) when measured to the other leads in the input bundle, but it measures a resistance of 80k Ω with the blue wire on top of the right big coil (Bo).

The coil on the outside of the big core (probably the filament transformer) is connected with one yellow wire (yo1) to the big yellow wire (Yo1) that comes from the big coil. The resistance between these combined wires (Yo1 and yo1) and the core measures also 80 kΩ. The resistance between the two wires of the smaller coil (yo1 and yo2) is 0.5 Ω.

Preliminary Conclusion

Thanks to the help of members of the Fusor net forum the following conclusions could be drawn:
The layout of the XRT wires is possibly as follows (image 6):

Image 6: Possible layout of Siemens X-ray transformer (© FRS 2014)

Two of the ten wires then still need to be identified: yi2 and bi2. We could imagine that these two wires are meant as low voltage primary input for the secundary heater coil if not the big heater coil is connected with yo1 to the secundary HV coil Yo1 output. The fact that the big heater coil is connected to the HV side of the secundary HV coil could however probably mean that this coil (that we have named the big heater coil) has a function as the primary coil in a step down transforming action to deliver output (average 10 V and 3 - 5 A) from the inner coil, which is probably connected to the two yet unknown yi2 and bi2 wires. In favor of this assumption appears to be the fact that this additional transformer is a so-called shell core transformer, which clearly can be seen in image 4. The core of the additional transformer runs outside (around) the big coil. Designing a transformer as a shell core type has the advantage that the efficiency is very high due to minimal flux losses in the core. Further meassurements will be required to finally sort out these assumptions and to give a clear picture of the real layout of this X-ray transformer prior to using it as our HV power supply.

Due to the rather modern design of the XRT it needs to be established that the XRT is operating on mains line voltage or that it requires a high frequency power supply. The large iron core does assume that the transformer is not intended for high frequency operation because in that case a ferrite core would have been more appropriate (and more efficient).

xray reviewed
Image 7: Final layout of Siemens X-ray transformer (© FRS 2016)

Finally, image 7 shows the presumed coonections of the XRT as used in the  X-ray machine setup.

Rectifier circuit

The first component after the XRT is the rectifier circuit. To keep things simple, a half wave rectifier circuit can be chosen. It makes no difference for the Fusor if the DC high voltage is more or less pulsating or on an even potential.

In most cases the XRT has a centre tapped secondary coil with an ampčre meter at the centre tap, connected to ground. A full wave rectifier circuit for this type of XRT is shown in image 8:

Image 8: Centre tapped XRT with full wave rectifier circuit (©FRS 2014)

The circuit rectifier in image 8 originates from reference 3 and it has a current limiting inductor in the primary circuit of the XRT as well as a resistor in the secundary XRT circuit with the same function. The current limiting inductor is a microwave oven transformer (MOT) with the primary of the MOT connected in series with the AC mains and with the secundary coil shorted and connected to the core (which is grounded). An MOT usually has an inductance of about 40 mH. The maximum current that will flow through the primary of the XRT equals the line supply voltage (230V AC) divided by the impedance of the ballasting MOT. The impedance of the MOT can be calculated from the formula
Z = 2πfL
where f is the frequency of the AC (50 Hz) and L is the inductance of the MOT (40 mH). Therefore the impedance is 12.57 Ω and the maximum current through the primary of the XRT shall be 18.3 A. When we know the winding ratio n between the primary and the secundary of the XRT we can calculate the maxmum current through the secundary, when short circuited, as IL2+L3  = IL1 /n or with an n of 370 (230 V to 85 kV) and an IL1 of max. 18.3 A we find an IL2+L3 of 80 mA as the maximum secundary current.

The centre tap of the XRT is connected with the core and with ground. The (open) potential over the two HV outputs is in our XRT 85 peak kV. The DC output is just below 40 kV DC and this is sufficient for powering a Fusor.
For this type of rectifier with a resistive load the Vpeak DC = 0.45 x Sec. V AC. For our XRT with 85 kVpeak this would be 38.25 kV DC. The DC current in the output is higher than the AC current of the secundary coil with a factor 1.27 or  IDC = 1.27 IsecAC. For our XRT with max. 15 mA output current the DC output current would consequently be 19 mA. See reference 4.

It is preferred not to add a smoothing capacitor in the HV circuit as it may impose an additional deadly danger. When cutting the power to the HV power supply the capacitor could still remain capable to discharge a considerable amount of energy when working on the switched-off HV circuitry.

An additional remark has to be made concerning the voltage divider for the voltage meter. In the current theoretical setup, the power dissipated on R2 is about 6 W and a 10W resistor is therefore minimally required. It is advised to lower the power by applying the rule of 100MΩ per 10 kV, which for our 40 kV should require a string of minimally 4 resistors of 100 MΩ each or a total resistance of 400 MΩ, dissipating each now 5 W maximally. Consequently R4 should be changed to 4 x 10 kΩ for providing a 104 reduced reading on the 0 -100 V range voltage meter. In this configuration resistor R3 dissipates only 0.5 mW maximally and therefore can be rated at 1W.

An additional advantage of applying a string of resistors is that cheaper HV resistors can be bought, thus reducing the investment costs in the project. This will make quite a difference from a cost point of view, but also the resistors will be easier to obtain, considering one single 400 MΩ resistor rated 50 kV and 20W compared to four identical resistors in series of 100 MΩ rated 20 kV and 5 W.
The reason is that each 100 MΩ resistor in the series string has a voltage differential over it of about 10 kV and dissipates less than 1 W.
The disadvantage of the resistor string is that sufficient space is required in the transformer housing to mount the string in transformer oil.


Neon Sign Transformer

A  Neon Sign Transformer (NST) is a midpoint ground high voltage transformer. It has two equal high voltage outputs, which are each out of phase with the other. The potential difference between the two outputs is the sum of the individual outputs. See image 10.
NST's are current limited stray flux transformers due to the fact that they have iron between the primary and the secundary causing high leakage inductance.
A neon sign requires a high voltage to start and that will cause the neon gas to form a better conducting plasma, which will draw more current. Due to the stray flux design the voltage will drop when the current rises. Image 9 shows the graphical relationships between voltage, current and resistance in a neon sign transformer.
The NST is a perfect transformator design for neon signs and tubes as it does not require an additional ballast inductor for powering the neon tube and it can do that for years continuously but its use for a Fusor remains questionable because in the Fusor at plasma state we encounter the same drop in resistance (and therefore rise in current with consequently drop in high voltage) as in a neon tube. However, we want the high voltage to remain unchanged as we desperately need the high potential difference for accelerating Deuterium ions.
Nevertheless NST's can be used for Fusor HV power supplies when the specific characteristics are taken into account.

The stray flux (or magnetic shunt) transformers were used until the 1990s and gradually replaced by high frequency inverter-converter transformers, which are not preferred for Fusor applications.

Image 9: Graphical relationships between voltage, current and resistance of a NST (source: reference 5)


Oil Burner Ignition Transformer

A possibly suitable transformer for a HV power supply can be found as an oil burner ignition transformer (OBIT), such as the widely available COFI TRK1-30 CVD with an input of 220-240V at 0.18A at 50/60Hz and secundary output of 2x12 kV at 30mA RMS, an Iout Burn of 17 mA and an ED of 100%. This is a transformer with the center of the secundary windings connected at ground (see image 4). Another OBIT with an ED of 100% is the COFI TRK1-20 CVD with a secundary output of 2x14 kV at 20mA RMS, Iout Burn 12 mA. Both OBITs are also available as a single 15 kV trafo with a grounded pole and similar 20 or 30 mA ratings.

OBITs  are usually also magnetic shunt or stray flux transformers such as NSTs and suffer therefore from the same problems as encountered with NSTs, such as the drop in voltage when the current rises.

Centretap trafo
Image 10: Transformer with centre-tap at ground (© FRS 2014)

It will be possible to use OBITs in parallel for raising the current output but they can also be cascaded to raise the high voltage output. See reference 6.


Microwave Oven Transformer

A microwave oven contains a very powerful microwave oven transformer (MOT) usually with a high voltage output of 2 kV at 0.5A. Contrary to NSTs and OBITs the MOT is not current limited though it has a weak magnetic shunt but it cannot be short circuited and therefore must be operated with a resistive or conductive ballast load in the primary or secundary circuit. MOTs have not been designed for continuous use and therefore will overheat easily. This makes them less desirable for use as Fusor power supply. See also reference 7.


Flyback transformers

A flyback transformer (FBT) or line output transformer is used in the operation of television sets and computer monitors with a cathode ray tube. They are very useful for constructing  switched-mode HV power supplies.

FBTs are transformers with an air gap in the ferrite core, which increases the reluctance of the core and that increases its abilities to store magnetic energy. This makes that the FBT acts as a pure inductor during half its cycle and then acts as a pure transformer during the other half of the cycle, i.e. they transfer energy to the secundary and they store energy for a
considerable amount of time.

An FBT cannot directly be connected to mains 230V because for operation it needs a high frequency
of 15 to 50 kHz. For powering an FBT a driver will be required.

Basically the FBT works as follows:
  • When current flows through the primary, a diode in the secundary prevents that no current flows through the secundary and the energy is stored;
  • Switching off the current in the primary induces an emf in the secundary and because the polarities are different the diode conducts, which induces a voltage which is rectified and smoothed, yielding a DC output. Rectifier and smoothing capacitor are components built into the transformer.
FBTs exist in different variations:
  • FBTs from old colour tv sets, consisting of an FBT with a separate cascade (voltage multiplier). The FBT usually has an output of 8 - 10 kVpp (peak to peak) and in combination with the cascade 24 - 30 kVpp.
  • FBTs with a single rectifier from old B&W tv sets. The FBT usually has an output of 15 - 20 kVpp.
  • FBTs with an integrated cascade with an output of 35 kVpp.
  • Dual FTBs, which are suitable for current output doubling. Usually these are of the split diode type.
  • Diode split transformers from more modern tv sets. The secundary coil is split by embedded diodes in several parts with lower voltage, rectifying the output. The output of this type cannot be multiplied with additional cascades, but when the output is pulsed DC they can drive a Cockcroft Walton multiplier (CWM).
  • FBTs can have a DC or an AC output. Some very old (around 1955) FBTs have an AC output and they will require an external rectifying circuit for use as a DC HV power supply.
A possible disadvantage of FBTs is that they usually can only deal with a few mA current due to the low current thin primary windings whereas we need a 20 mA output for the Fusor. This limitation can be overcome by rewinding the primary coil with thicker wire which will raise the current throughput and by using a dual flyback transformer with parallel shunted secundary coils, which will double the current output. The average power of an FBT in television applications is rather low and therefore the primary coil is wound with thin wire. The output of the dual FBT can be raised up to 480W by rewinding the primary coil with thicker wire. A power output of 480W is about the amount of power that we need for a Fusor at basic power requirements: 20 - 25 kV at 15 - 20 mA, or 300 to 500W.

The dual FBT is operated with an input of DC and has a pulsed DC ouput. With an input of 12V DC the output will be10kV and with an input of 24V DC the output will be 20 kV. In both cases the maximum current is 20A and by applying 24V DC at 20A the (theoretical) current at 20 kV output will be 24 mA (losses not taken into account).

Apparently a new challence occurs now at the horizon: the construction of a 24V/20A voltage- and current regulated power supply and selecting an appropriate driver.

See also reference 8, reference 9 and reference 10.
 Reference 10 provides help in finding the pins layout of FBTs.

Flyback Drivers

Mazilli Flyback Driver

As mentioned before the FBT cannot be connected directly to 230V mains supply but it will need a driver.
For a high output dual FBT a driver will be required that can handle and deliver a high frequency and high current at 24V. A popular design is the Mazilli zero voltage switching (ZVS) flyback driver which has a circuit as shown in image 11. Mazilli's are by far the most powerful FBT drivers and are known to be able to pump maximally about a kilowatt of power.

A Mazilli ZVS will require rewinding the primary coil of the FBT with two separate (thick wire) coils, usually 5 turns, because each mosfet in the ZVS circuit needs to power a primary coil. The two windings must be in the same direction because when wound in the opposed directions the output will be zero Volts.

Mazilli ZVS driver
Image 11: Original Mazilli ZVS flyback driver.

The circuit of image 11 is known as a resonant zero switching circuit for driving a flyback transformer with two mosfets in a push-pull configuration. The design of the circuit is somewhat bizar: when evaluated in a circuit simulator it probably will not work. When built on a breadboard it will absolutely work. The reason can be found in the small (manufacturing) deviations that will be present in two similar mosfet components. When a voltage is introduced at V+ a current starts to flow through the primary of the FBT and on to the mosfets' drains. Simultaneously the same voltage appears on both of the mosfets' gates and will start to turn the mosfets on. Because the mosfets are not fully identical in characteristics one of the mosfets will turn on a little faster than the other one and more current can flow through that fet. The extra current that flows through that side of the primary will rob the gate current form the other fet and turn it off. A condensor forms a second order LC circuit with the primary (inductor) and the voltage will proceed to rise and fall sinusoidally. The capacitor prevents the current to continue increasing until the transformer core gets saturated, followed by the mosfets being blowing up.

When the upper mosfet in image 11 is the first one to turn on than the drain of that mosfet will be at near ground while the voltage at the drain of the lower mosfet rises to peak and falls back down when the second order LC circuit  goes through one half cycle. As the voltage at the lower mosfet's drain passes through zero the gate current to the upper mosfet is removed and the mosfet turns off. The voltage at the drain of the upper mosfet is now allowed to rise and the lower mosfet turns on. This mosfet now clamps the voltage at the lower mosfet's drain to ground making sure that the upper mosfet stays off. The same process repeats for the lower mosfet and completes the other half cycle making the oscillator continuing cycling. The inductor in series in the V+ line acts as a choke and prevents the oscillator from drawing high currents and blowing up. In fact, the LC impedance limits the actual current and the choke only mitigates current spikes.

The capacitor and the inductance of the FBTs primary coil determine the frequency at which the mazilli resonant zero volrage switching driver will operate. The frequency can be calculated with:
Hz formula
in which:
f0 = frequency in Hertz
L = inductance of the primary in Henries
C = capacitance of the capacitor in Farads

The two 470Ω resistors in the circuit  limit the current that charges the gates to prevent damage to the mosfet; the two 10 kΩ resistors pull the gates down to ground to prevent latchup; the two Zener diodes prevent the gate voltage from exceeding 12V, 15V or 18V (whatever voltage the Zeners may have) and the two fast diodes (UF4007, FR107, etc.)  pull the gates down to ground when the voltage at the opposite leg of the second order LC circuit is at ground.
The mosfets will need to withstand the resonant rise voltage in the second order LC circuit which is πVcc. or as a rule of thumb four times the voltage to be fed onto the oscillator. In that view the IRFP250 mosfet in the circuit of image 10 could better be replaced by the IRFP260N mosfet as is shown in the circuit of image 13. The capacitor to be used needs to be a MKP, mica or Mylar capacitor.  The primary windings must be wound in the same direction. The maximum (peak) voltage allowed on the circuit is 60V to prevent it from blowing up. A 0.5
Ω wirewound resistor in series with the filter inductor will protect the circuit when the load inductance sharply drops.

A commercially available (China) cheap (< 12 EUR) Mazilli flyback driver is shown in image 12 (lower item) and the circuit is shown in image 13:

Image 12: Flyback driver (below), Flyback transformer (top); Source: Goodie310

Image 13: Circuit of the commercial Flyback driver; source: Goodie310

The manufacturer's (somewhat cryptical) specifications are as follows:
  • Input DC 12V - 30V (24V recommended)
  • High voltage DC output approx. 1000 x input voltage
  • At 12V power approx.  50 - 100W
  • At 24V power output approx.  200W
  • Input current > 10A required.
Driver does not start working: Power input too low. Use minimally 12V/5A power supply.
Heat generation at magnetic core:
When using transformer  as a power supply a filter capacitor of 10000 uf+ is required for preventing large voltage fluctuations in the coil inductor causing excessive heat.

Another Mazilla FBT driver but now with a split diode FBT is shown in image 14:

mazilli with split diode FBT
Image 14: Mazilla FBT driver with split diode FBT

The schematic in image 8 makes use of two Fairchild series SPMS power mosfets of the type FDH44N50 rated at 44A, 500V and 120 mOhm and MUR240, 2A, 400V ultrafast diodes and it is claimed to have an output voltage range of 10,000 - 30,000V (depending on FBT, measured over 300 MΩ/3 MΩ voltage divider) at an input of 10 - 40V, 5 - 10A (50 - 400W). Ideally the primary waveform should be a sinewave; a square wave indicates a too low value for L1. See reference 11.

Royer Flyback Driver

For high power applications with a FBT a Royer oscillator is also a well known, simple concept. An example of a Royer FBT driver can be found in image 15:

Royer FBT driver
Image 15: Royer FBT driver

The capacitor needs to be a polypropylene type, minimal value is 10 nF, and it is used is to suppress spikes which might kill the transistors. The input inductor should have a value around 330 µH with a high current rating. The limiting factor is the input voltage: at 24 V input the voltage across the transistors is 60V; at 70V the transistors will collaps. It is therefore doubted that a Royer driver will meet the Fusor requirements of sufficient voltage and current output.

Other Flyback drivers

A few other FBT drivers are known and these operate either on an astable 555 timer design in combination with a single mosfet (image 16) or even on a single transistor (image 17).

Image 16: FBT Driver with 555 timer (source: reference 12)

Image 17: FBT driver with one transistor (source: reference 13)

We have not further evaluated these driver designs because of their apparent lack of handling high power output.

Summarizing it can be concluded that FBTs offer an interesting possibility for a HV power supply when the appropriate high current enabling driver is used.  However, the basic power requirements for a Fusor power supply are difficult to meet in particular the requirement of "fully variable". A well-tuned Mazilli driven FBT is probably in the input variable between 10 - 40V at 5 - 30A yielding an output of approx. 10 - 40 kV at ≈300 - 5 mA (descending amperage with increasing voltage).


Development of a Flyback Transformer Power Supply

In the beginning of the project and taking the dificulties of obtaining an X-ray transformer into account, it was decided first to explore the possibilities of making a HV power supply based on a powerful Flyback transformer. The impossibility to fully regulate the FBT output over a longe HV range and the fact that the most powerful FBT in the market was of the split diode type, yielding a positive HV output instead of a negative HV output requiring an additional circuit, forced the decision to explore other modalities. When an X-ray transformer became available this FBT power supply project was abandonned, though it yielded over 50 kV at 20 mA.

Flyback Transformer HV Power Supply

The next paragraphs describe the construction of a High Voltage Power Supply consisting of a regulated 24V DC power supply connected to a split diode dual FTB with an output of over 20 kV delivering a power of > 480 W, i.e. an output current > 20 mA.

DC Power Supply

In order to achieve a HV power of > 20 kV at a current > 20 mA (480W max.), the 24V DC power supply therefore needs to deliver a current > 20A and this was created by connecting two HP server power supplies in series, each rated at 12.15V DC at 47 Amps (Images 18, 19 and 20):

power supply apower supply bpowersupply c
Images 18, 19 and 20: Server Power Supplies; left: front, center: back, right: label (© FRS 2014)

The HP server power supplies (HPPS) can easily be found on the internet (eBay) for prices ranging between € 12 and € 20 each. Instructions how to convert the HPPS into a 24V DC, 47A power supply can be found in reference 14 and reference 15.

Once the DC power supply has been converted, a method needs to be found to make the HPPS DC output variable in voltage and current. For this purpose the intention is to use a variable DC-DC converter.

Variable DC-DC Converter

The variable DC-DC converter is a digitally controlled DC-DC  constant voltage and current regulated power supply module rated maximally 60V at 20A (1200W) (image 21).

Dc DC PS 1200W
Image 21: Variable DC-DC Converter (Source: Supplier China)

The technology of the variable DC-DC converter is based on a buck boost converter with a circuit design incorporating two split modules, a power module and a separate control module (image 21  and image 22 lower circuit board including indication of functional items). The DC output has a frequency up to 150 kHz, which may permit direct connection to a flyback transformer.

DCDC PS expl
Image 22: Variable DC-DC converter Control Module (Source: Supplier China)

The variable DC-DC converter has an input voltage range of DC 13 - 62V, an adjustable output voltage DC range of 0 - 60V and an adjustable current range of 0 - 20A. A rotary encoder is used to set the voltage or current with a resolution of 0.01V and 0.01A. An onboard temperature sensor can be set by the user to protect the circuit for too high temperatures occurring under high load; in such cases a ventilator can also be added to cool the circuitry. The display accuracy is 10 mV with an error of ± 0.5% and 1 mA with an error of ± 1.0%.

Split Diode Dual Output FBT
The split diode  dual output FBT has one minus and two HV positive outputs. Both outputs will be connected to deliver double the current. The total power output is not allowed to exceed 480W and a regulation and control of the voltage and current at the input is therefore required with a voltage between 12V and 24V and a maximum current of 20A.

power flyback
Image 23: High Power Split Diode Flyback Transformer (Source: High Voltage Shop - Austria)

The FBT (image 23) delivers a pulsed DC output of 10 kV at an input of 12V DC and 20 kV at an input of 24 V DC. The FBT needs to be connected to a one stage Cockcroft-Walton multiplier for reversing the HV output polarity to negative potential compared to ground. Theoretically the output rises than because of the CWM to 56 kV, but depending on the capacity of the capacitor this will be less, as mentioned in the paragraph about the Cockcroft Walton Multiplier.

Example: As mentioned for a one stage CWM we need two capacitors and two diodes. Suppose we use capacitors of 560 pF, rated at 50 kV, than we need to put two times two capacitors in series in order to increase the rating to 100 kV (for safety rasons).
At the lowest resonance frequency of 15 kHz of the FBT unit we will get with the one stage CWM with a load current of 10 mA (in the Fusor) at 20 kV input voltage on the CWM, an output voltage ranging between 54.8 and 56 kV. The input current in the CWM is than 27.4 mA, or an output power of 548 W, just exceeding the maximum allowable power of the FBT.
Similarly with the highest resonance frequency of 50 kHz we will get an output voltage ranging between 55.6 and 56 kV. The input current in the CWM is than 27.8 mA, or an output power of 557 W, also exceeding the maximum allowable power of the FBT (480 W).

The important issue is here that we do not want to operate our Fusor on 56 kV! Any voltage over 35 kV causes röntgen radiation to escape through the steel walls of the vacuum chamber. Therefore we should limit the voltage on the Fusor to maximally 30 kV.
With an input of 10 kV, 15 kHz, at the CWM in the example we get an output voltage ranging between 26.8 and 28 kV, which variation is acceptable. The input current in the CWM is than 26.8 mA, or an output power of 268 W, well within the maximum allowable power of the FBT (480 W). Unfortunately 10 kV is the lowest possible output voltage for the FBT with 12V DC input and that leaves us no possibility to vary the HV potential over the Fusor, which is a requirement.


Distribution Transformer (Pole Transformer)

These are used to transform HV down to mains voltage, e.g. 10kV to 230V, but of course also work in the opposite direction.They are capable of high currents but also difficult to control and therefore not suitable for use in a Fusor power supply.

Measuring Transducers or Potential Transformers or Instrument Transformer ('Messwandler')

Potential Transformers (hereafter PT) are used to measure high voltage by transforming HV down in a fixed and known ratio to a safe lower voltage, e.g. 30,000V to 300V (ratio 1:100) or even more common 20,000V to 100V (ratio 1:200). These work of course also in the opposite direction and can be used as a HV supply.
A PT has the ability to be subjected for a short time period to potentials higher than the nominal value. This is quite logical because the high voltage network is frequently undergoing transient voltage fluctuations. When using a PT as a HV source is is imperative to connect a current reducing inductive choke ballast in series with the 100V coil in order to prevent drawing an uncontrolled high current from the mains network, causing blowing of mains fuses or blowing up our transformer.
It is not recommended to increase the voltage higher than 120 VAC on the 100VAC coil and to take care that the core does not go into saturation (which can be heard as a 50 Hz humming noise).
PTs are constructed as monopole or dipole transformers. The dipole transformers have both poles "floating", i.e. both HV poles are isolated from ground. The monopole transformer has one side of the HV coil connected to ground as well as connected to ground of the primary coil and the other side of the coil carries the high voltage.
The thermal limiting output of the PT is an indication for the performance of the transformer at constant use at a certain quality class. This performance can be exceeded with a factor 3 up to 4 during a very short period of time.

Our PT is of the TuR brand, a model GE 24, and it was manufactured in Dresden in the former German Democratic Republic (image 24 and 25).
The TuR transformer facility in Dresden was erected in 1948 under the full name "VEB Transformatoren- und Röntgenwerk (TuR) „Hermann Matern“ Dresden", on the remains of the former Koch & Sterzel AG works. In 1989, the end of the German Democratic Republic, the company counted over 5000 employees and finally became part of Siemens AG, who in the next years sold off parts of the former TuR business. The potential transformer division went in 1991 to Ritz Messwandler GmbH in Hamburg, who still produces these and similar transformers. 

PTa  PTb
Image 24 and 25: Potential Transformer (source: shk-engel)

The label on the transformer bears some information about the characteristics. The indication of the voltages, divided by the square root of 3, tells us that the transformer is intended for use in a three-phase network. In such a three phase medium voltage network the transformer will than be connected between one phase and ground. In a three phase network the name for the network refers to the phase to phase voltage, which is in our case 20 kV. The voltage between phase and ground in the primary of the transformer is than 20 kV divided by square root 3 or 11.5 kV and in the secondary of the transformer it is consequently 100 V divided by the square root of 3 or 57.7V.
Does this mean that we can only use this transformer for 11.5 kV to 57.7 V applications (or reversed as we would like to do)? The answer is no as the transformer is designed to accidentally bear the voltage between phases (i.e. the full 20 kV).
The specification sheet of the manufacturer shows that the primary measured voltage is 20 kV with a maximum voltage for operation of 24 kV and also that the maximum applied test voltage during 1 minute is 65 kV, with a real primary voltage of 11.5 kV and a real secundary voltage of 57.7 V. Moreover, the thermal limit power for the primary coil at a given accuracy class is 600 VA and for the secundary coil it is 100 VA.
The crucial issue here is that we only know for sure how the transformer will perform (outside the accuracy class) when we apply a variable voltage to the secundary (low voltage) coil, whilst keeping the primary coil open (no load, because the core flux is at its maximum with no load connected) and when we measure the amperage drawn when we increase the voltage slowly to 100 V, i.e. when we notice the point where the core of the transformer goes into saturation. This usually happens at an amperage somewhere between 2 and 4 A with a PT when no load is connected to the high voltage coil.  This can clearly be seen in graphs showing measurements on PTs (reference 16). From this reference on by moving forwards on the following pages of this website one can see the effect of different types of choke loads in the primary of this particular PT. It should explicitely be noted here that the saturation behaviour of the TP will be different when the transformer will be used under normal circumstances, i.e. with a load connected to the high voltage coil.

The label on the transformer further states for the 100/
3 coil an Ig of 10A, which is apparently the thermal limit current for this coil, and for the 100/3 coil an Id of 4A, which apparently is the continuous current allowable to this coil. The meaning of these current indications come from the German language, where the d in Id means "dauer" or duration, continuance and where the g in Ig means "grenz" or terminal, limit.

So far it is clear that the primary (high voltage) coil of the transformer is not to worry about whether or not it can deliver the minimum of 15 mA that we need at 20 kV for our fusor, as this equals some 300 VA. For the secondary (
100/3 low voltage) coil, that we intend to use as our primary coil, it is also clear that it can process 300 VA as this equals a current of 3 A, because the label states a maximum allowable current of 10A.            

PTc    PTd
Image 26 and 27: Connectors Layout and Coil Schematic (source: Ritz)

When we take a look at the connectors present on our transformer  we see the following:
The a and the n connectors are the normal connectors connected to the low voltage coil, whereas the da and dn connectors in image 26 are connected to an additional winding used for a so-called open delta circuit.

The question now is if the coil connected to a and n is electrically identical to the coil connected to da and dn. To find out if this is true we need to perform measurements on both coils for resistance and inductance.
Should both coils appear to be identical in electrical performance than we possibly have the possibility to connect both coils in series, allowing to feed them with a voltage of maximally 230 V at the lower amperage of 1.3 A to produce an output of 300 VA needed for obtaining 20 kV at 15 mA at the high voltage coil. The consequences of connecting both identical coils in series in relation to transformer performance are not known yet and need to be further investigated. It is however noted that with different elecrical characteristics for each of the coils the maximum allowable current for both coils in series will be the current equal to the lowest allowable current valid for one of the coils.

For the results of measuring the performance of our Potential Transformer with different input voltages we refer to the Experiments and Results page of this website.

The measurements of the PT windings result in the following equivqalent resistance circuit (image  28):

Image 28: Equivalent resistance circuit (© FRS 2016)


Shunting Multiple Transformers

It is an advantage to install a number of transformers in parallel in order to obtain a higher power rating. In general a transformer gives the maximum efficiency at full load. When a number of transformers are run in parallel it will be possible to switch on or off transformers in the parallel chain when more or less load is required.

The conditions for parallel operation of transformers are the following:

  • Same voltage ratio of transformer.
  • Same percentage impedance.
  • Same polarity.
  • Same phase sequence.
  • Secondary cabling from the transformers to the point of paralleling have approximately equal lengths and characteristics

The same phase sequence means that the AC voltage sine wave goes high on both coils at the same time. When phased in opposition a double high current short circuit will be formed.


HV Rectifying and Voltage Multiplying Cascades

The output of an AC power supply needs to be rectified before it can be used for the Fusor (which operates on negative voltage DC). This can be done with a single HV diode and a capacitor. See image 29:

Image 29: Diode rectifier (© FRS 2014)

The diode rectifier supplies half wave rectified DC but this is not a problem for powering a Fusor. It should be noted that when the power supply has an RMS AC output, the actual DC output will be ½√2 or about 0.7Vrms
In a number of cases the HV output from a power supply will be considered as too low for Fusor applications, e.g when (as an example) a high current transformer is used with an AC HV output of 10kV. For operating a fusor we would prefer to have an output of 20 - 30 kV. In that case a voltage multiplying cascade is used. It should be noted that when a voltage multiplying cascade is used the rectifier as mentioned above becomes superfluous as the multiplying circuits will supply DC. The following circuits are quite well-known (reference17):
  • Villard circuit
  • Greinacher circuit
  • Cockroft-Walton multiplier 

Villard circuit
The Villard circuit consists of a capacitor and a diode and it is mainly used in microwave ovens to generate the negative high voltage. The output of this circuit is 2Vpk
and the disadvantage is the high ripple (image 30).

Image 30: Villard circuit (©FRS 2014)

Greinacher circuit

The Greinacher circuit is an improvement of the Villard circuit and produces less ripple (image31).

Image 31: Greinacher circuit (©FRS 2014)

Cockroft-Walton Multiplier

The Cockroft-Walton multiplier (CWM) was discovered (reference 18) independently from the similar cascade developed by Greinacher twelve years earlier (image 32: a two stage, half wave CW multiplier). The CWM in the image will have a positive DC output but a negative DC output can be obtained when the diodes are reversed.

Cpckcroft Walton
Image 32: Cockcroft-Walton Multiplier (©FRS2014)

The CWM has a voltage gain for n stages of Vout = [2n√2*Vin
]-Vdrop where Vin is the RMS input voltage, Vout is the output voltage and Vdrop is the voltage drop. The main disadvantage of the CWM is that the output voltage drops when more current is drawn. This can be compensated by using bigger capacitors or a higher frequency input. The voltage drop can be calculated (reference 19) roughly with the formula:

in which
Vdrop = voltage drop
I = current drawn in Amp.
f = frequency in Hz
n = number of stages
C = capacity of capacitors in Farad

The formula is only a very rough estimation because in practice the real voltage drop measured will be a lot more higher. Moreover, except the voltage drop, when a current is drawn also a higher ripple (proportional to the input frequency) will occur according to the formula:

in which
Vripple=ripple voltage
I = current drawn in Amp.
f = frequency in Hz
n = number of stages
C = capacity of capacitors in Farad

Despite the voltage drop and the ripple a CWM is quite useful in combination with a high frequency power supply such as a Flyback transformer. A (calculated, reference 20) example of an FBT driven one stage CWM (i.e. two diodes and two capacitors) operating at 50 kHz with an input of 12 kVrms, a load current of 15 mA and capacitances of 0.0022 µF (2.2 nF, 50 kV) will deliver 33.6 kV with a ripple of 136 V, requiring a current input (Iin) of approx. 42 mA and a power input (Win) of approx. 502 W.  

Marx Generator

Another method to generate HV from a (relatively) low voltage (DC) source is by means of a Marx generator.

When a DC voltage supply is available but it delivers a too low HV output, a high voltage DC puls can be generated by means of the Marx generator. The Marx generator consists of a number of capacitors in parallel, which after charging are suddenly connected in series. See image 33, reference 21 and reference 22:

Marx generator diagram

Image 33: Marx Generator (source: reference 22)

A high voltage DC power supply charges in parallel n capacitors C through resistors Rc with a voltage V. The spark gaps (used as switches) have the voltage V across them, but the gaps have a breakdown voltage >V, which make them act as open circuits when the capacitors are charging. 
Although the left capacitor has the greatest charge rate, the generator is typically allowed to charge for a long period of time, and all capacitors eventually reach the same charge voltage.
The last gap isolates the output of the generator from the load. Without that gap the load would prevent the capacitors from charging.

To create the output pulse, the first spark gap is triggered to breakdown
; the breakdown shorts the gap placing the first and the second capacitor in series applying 2V over the second gap, which breaks down and connects the third capacitor is series and applies 3V over the third gap, etc.  Ideally, the output voltage will be nV, the number of capacitors times the charging voltage, but in practice the value is less.

The circuit can be improved by using avalanche diodes for the gaps (when the stage voltage is lower than 500V); the charging resistors Rc can be replaced by inductors for faster charging and improved efficiency or be made from plastic or glass tubing filled with diluted copper sulfate; an NPN avalanche transistor fitted with a coil between emitter and base can be used as the switching device, etc.


Additional Components
As additional components we consider an Autotransformer or Variac, a rectifier diode or string of rectifier diodes, a volt and an amp meter, required for measuring the output of the HV power supply, in combination with a resistor divider for reducing the HV to a less dangerous  voltage.


One of the advantages of a HV power supply operating with a PT is the fact that a fully variable output HV power supply is obtained when the PT is powered with a variac or autotransformer. The variac should have sufficient power to cope with the load of the PT, preferably minimally a 10 A Variac should be chosen. We obtained a very cheap Variac with some transport damage (image 34) such as a broken front panel, a broken fuse holder and a voltage meter beyond repair and including some dents in the metal housing.

Image 34: Variable Transformer 2 kVA (© FRS 2016)

The broken parts were shipped with the variac but because the electrical design safety charcteristics did not comply with local and EU regulations we decided to modify the design (image 35).

variac revised
Image 35: Revised design Variac (©FRS 2016)

The revised Variac got a new aluminium front housing (connected to ground), a new fuse holder, a digital voltage meter, a couple of input and output connectors complying with
EU requirements, the Variac metal housing was permanently connected to ground and the dents were hammered out of the housing. The aluminium housing and the digital voltage meter came from China, the connectors and press-on connectors from United Kingdom, the fuse holder from Greece and the total costs of the modification were less than 15 Euros.
The voltage meter was connected across the output connector and this means that when the fuse is accidentally blown the meter will show no output voltage.

Note: Two differents types of Variacs exist: the cheaper ones, which have no secondary winding, and the more expensive ones, which have a secondary winding.
The more simple Variacs without the secondary winding have no galvanic separation from the net and can be considered as the primary of a transformer. This means that the output of the Variac is connected to the full voltage of the net and may cause surprises when touched! See the simple Variac schematic in image 36.

Image 36: Simple Variac schematic (© FRS 2016)


Rectifier Diode

A unipolar Potential Transformer (PT), with one HV output connected at ground, has the disadvantage (for being used as a Fusor power supply) that it permits only half-wave rectifying. The consequences of this setup is that the (average) DC output is 0.45 of the HV transformer output voltage and the (average) DC current is 0.64 of the HV transformer output current (image 37).

DC rectifying
Image 37: Rectifed DC Voltage and Current (© FRS 2016)

Nevertheless, despite the lower average DC voltage such a half-wave rectified transformer still pulses peak DC output at the full voltage, which is equally sufficient to produce fusion though at a lower duty cycle.

When a higher average DC is required we could decide to add a voltage multiplier rectifying circuit such as a Cockcroft-Walton multiplier (CWM), but this will also involve putting in some capacitors apart from other diadvantages. The problem with capacitors that they tend to keep their (dangerous) charge and therefore an additional bleeding resistor, connected to ground, is also required for bleeding the capacitors to zero charge after powering down. Moreover, it is quite difficult to find capacitors of a high capacity and capable of delivering a current of minimally 20 mA and with a power rating of  50 kV or higher. This can be covered by connecting capacitors with a lower power rating in series but this will decrease the capacity, which can be compensated by connecting capacitors not only in series but also parallel. The disadvantage of capacitors in series is, however, that even when taking capacitors with the same value from the same batch, individual differences will be present, which causes e.g. that a "bad" capacitor overcharges the next capacitor in the series and that is not good.For that purpose balancing resistors are connected parallel to each capacitor, which counteracts the effects of variance in capacity and leakage current.
All this is quite annoying and complicating and it supports our opinion to leave out capacitors as much as possible. See also reference 25.

Nevertheless, if we can overcome our objections of capacitors in a HVPS,  we could, instead of using a CWM with all its disadvantages, decide to use a capacitor in the half-wave rectified circuit to smooth the peaks. The capacitor is connected parallel to the load and the higher its capacity the more our output will be smoothed.
The way how we will perform this in our HVPS, with the parts available to us, is discussed in the construction chapter.

The PT will be half-wave rectified by a HV rectifier diode string. The diodes are of the HVM 12 type (image 38).

Image 38: HVM 12 high voltage rectifier diodes (source: supplier)

The HVM 12 rectifier diode is suitable for rectifying with a repetitive peak reverse voltage of 12 kV with an average forward current of 350 mA. The maximum surge current is 30 A, the average forward voltage drop is 12 V, the peak reverse current is 25 µA at 25°C.

The HV transformer outputs in our case a 50 Hz sinus with a peak to peak voltage of nominally 20 kV but this may accidentally rise to over 30 kV when we turn up the Variac too high. The full peak to peak voltage is over the diode and this requires that the diode is capable to deal with the highest possible voltage. Therefore, it is our intention to connect three diodes in series, in order to get to a peak reverse voltage of 36 kV.
A balancing resistor could be connected in parallel with each diode but with modern diodes of the same type and value in a string this is not strictly required and rather old fashioned. The balancing resistors must be able to withstand the voltage of 30 kV, which makes them long and difficult to accomodate in the circuit. Avalanche diodes do never require a balancing resistor.
The string of diodes will be placed in a PVC tube filled with paraffin wax to prevent corona or arcing.



For our rectifying circuit we make use of two (Russian) pulse capacitors with a capacity each of 0.1 µF and a power rating of 80 MW (40 kV at 2 kA). Each capacitor has dimensions of 140 mm (l) x 130 mm (w) x 260 mm (h, including height of large connector; 140 mm height without connectors) and a weight of 4.4 kg (image 39). The power rating is probably a bit over the top for our purposes, but it is all we could get when we were looking for caps that could withstand voltages up to 40 kV.

Image 39: Pulse Capacitor 0.1 µF, 40 kV, 2000 A (source: supplier)

Both ceramic connector isolators have rounded metal caps mounted on M8 thread for the largest connector and on M5 thread for the smaller connector.

The largest connector is the high voltage connector and in our setup to be connected to the negative HV output of the diodes, whereas the smaller connector is to be connected to the ground potential of our Potential Transformer (also connected to the core of the transformer). The housing of the capacitor needs to be connected to ground (earth), for which a soldering contact is present on the housing. We should recognise here the difference between ground and earth, where ground means the line with a potential of 0V (zero volts) and earth the line connected to earth potential (usually in Europe the green/yellow wire of the mains power line).



Ballast Resistor

In the Potential Transformer based HV power supply, as well as in X-ray transformer power supplies, a current limiting ballast is required to prevent damage to the transformer when a sudden heavy load occurs and to prevent blowing the mains fuse. Such sudden heavy loads may occur due to sudden current surges due to Townsend or arc discharges in the Fusor vacuum chamber causing dramatic high voltage impedance changing from almost infinite down to or near almost zero Ohm.

Ballasting is possible in the primary or in the secundary circuit of the transformer, which is equally efficient but sometimes the preference is given to a ballasting resistor in the secundary circuit (reference 23). The disadvantages of such a ballasting resistor in the secundary circuit are the fact that the resistor is in the HV circuit, requiring expensive HV resistors, and the fact that the power of the secundary is dissipated in both the ballasting resistor and the load, i.e. the Fusor. This means that the maximum load dissipated into the Fusor cannot exceed 50% of the total power and this is only the case when both the ballasting resistor and the Fusor impedance have equal values (impedance matching).

Though there is no intention to use the HV power supply at it's highest output of 30 kV DC due to the hazard of producing hard X-rays from Bremsstrahlung, it seems common sense to take the highest output as a basis for evaluation of the ballast resistor as well as for evalaution of the components for the rectifier circuit. Should for instance by accident the input voltage of the PT be turned up to the maximum possible voltage of over 120V, than the result should not be that components from the circuitry are fried! In such a case, however, the resulting production of hard X-rays will immediately be recorded by a gamma detector setting off a visual and audible alarm.

With a ballast resistor in series with the Fusor and the Fusor in "plasma-off" mode with full power over the circuit, the Fusor is the component with the highest impedance and little or no current will be flowing. The voltage over the ballast resistor is than near to zero Volt. In the "plasma mode" the deuterium plasma has -according to literature- an impedance between 0.42 Ω and 2.80 Ω, but this is of no importance as long as no discharge takes place and the plasma remains inside the grid. The impedance of such a plasma may differ with circumstances (e.g. ion density, linear mass, etc) and place of measurement in the plasma. More important is the total impedance of an operating Fusor for which a typical value can be found of 750 kΩ at a 20 kV potential. 

In a sudden flash the Fusor could light up a plasma and -when the plasma enters a discharge mode- the resistance in the Fusor drops to near zero. In the same sudden flash the ballast resistor gets the full voltage load.

Suppose, the ballast resistor has a resistance value of 60 kΩ than at 30 kV the resistor (in theory) will get just over 0.6 A in a transient (disappearing in short time) moment. Suppose also that we have an additional MOT in the primary, reducing the maximum current in the primary to 10A at 230 V AC, this  will permit a maximum current of 80 mA in the secundary of the PT. With a maximum current of 80 mA the resistor will be subjected to almost 3.2 kW of instantaneous power during a transient microsecond before dissipating this energy into the circuit. If the resistor will tolerate this power remains the question but it is advisable to select a resistor with a high power rating, at least at 10% of the expected power peak, i.e. in our case a 300 W HV resistor, which usually will be a wirewound resistor with a high voltage rating. Such a resistor has the windings fully encapsulated in an insulating material (image 40).

Image 40: High Voltage Wirewound Resistor (source: supplier)

Even more important probably is that the resistor is able to withstand the high voltage burst by its physical dimensions, i.e. is it long enough to prevent an arc jumping from one lead to another? To withstand an arc jumping from one lead to another at 40 kV, the length of the resistor should be at least 15 cm (rule of thumb: 5 cm for each 15 kV in dry air). A resistor with that length might be difficult to find. Another problem is corona that will be present at the high voltage. Therefore it is imperative that the resistor has to be encapsulated or immersed in transformer oil. This applies also to the divider resistors that connects to the voltage meter and the capacitor and diode string from the rectifier.

All these limitations seem to indicate that we better could omit the resistor in the secundary of the PT, but this is certainly not advised. The MOT in the primary of the PT should be seen as an additional current limiter that will give us some of the required protection for the PT and our mains network.

The current drawn from the transformer at short circuit conditions is far more than the Potential Transformer is supposed maximally to deliver (for our transformer under normal conditions 20 mA max. at some 20 kV) and this is reflected in an instantaneous raise in the current of the primary side of the transformer. In order to protect the transformer from severe overloading (already limited by the MOT in series with the primary of the PT) it seems therefore logical to have an additional, fast acting fuse or circuit breaker in the primary circuit. Usually such a fuse is already present in the secundary circuit of the variable transformer which is used to regulate the input voltage to the primary of the PT. The fact that high currents may be drawn from the variable transformer also explains why we should have a high power rated variable or autotransformer in the circuit.

However, not too much can be expected from a fuse in the primary coil circuit. The secundary coil has the ballast resistor to prevent too high a current and the primary coil has the MOT as an inductive ballast load. The function of the fuse, rated 250 V and 10 A, is to provide an extra safety for the electrical installation at home. A single power line usually has a fuse in the fuse box rated at 16 A and the general power line entry into the house generally has a 45 A fuse. Blowing the 16 A fuse is inconvenient, but blowing the 45 A fuse requires the electricity company to replace it and they will most certainly write out a fat bill for that!

Earlier we discussed our preference for an inductive ballast load in the primary of the HV-transformer as an extra precaution for high currents.
We also indicated that an MOT probably was a good choice, and we provided a short discussion about the required inductivity of this choke coil. We shall discuss here the same subject with the same formulas now calculating the required inductance for the MOT in some more detail.

Image 41 shows our (typical) MOT with a power rating of 800 Watt. These MOTs usually have an inductance of about 40 mH and with a power rating of 800 W at 230V a maximum current of 3.5 A is permitted (for a short while only as microwave ovens never work full time when heating). The question is will that be sufficcient?

Image 41: Galanz GAL 800E-4 Microwave Oven Transformer (source: supplier)

The total resistance Z of a choke coil consists partially of the actual coil resistance R and partially of the inductive resistance XL in a linear relationship of the frequency f and the inductivity L as
XL = 2 *π*f*L
Z = √ R2+XL2
The resistance of the coil R can be measured directly over the coil with an Ohm-meter but usually R hardly plays a role compared to the inductive resistance.

Suppose we want to reduce the current flowing through our primary of the HV transformer to 10 A maximally at 230 VAC. This means that the choke coil must have a total resistance Z of 230/10 = 23 Ω.
Suppose we measure a resistance R over the coil of 0.5 Ω than the inductive resistance must be √Z2-R2 or √((23Ω2)-(0.5Ω2)) = 22.99 Ω. Indeed the effect of R has hardly any influence on the result!
The next step is to calculate the inductivity by arranging our formula into
L = XL/(2*π*f) 
or L = 22.99/(2*π*50) = 73.2 mH.

To be sure that our  MOT in the function of a ballast choke coil will do its work properly by reducing the current to 10 A maximum in the primary HV coil, we will need to check the inductance of the MOT to be around 73 mH. When this is not the case we may need to add another transformer in series until the correct inductance has been realised.

Please note that the choke transformer coil in series with the HV coil is capable to deal with a current of 10A or more! Our MOT of image 37 only can deal with about 3.5 A! In case of severe amperage non-complaince our MOT will end up in smoke!

Should we want to limit the amperage in the primary HV coil to 4 A maximally, which is sufficient for obtaining at the secondary HV coil 25 kV at 20 mA, than -according to our formulas above- we will need a total resistance Z of  230V/4A = 57.5Ω which equals an inductivity of 183 mH or minimally four transformers of 40 mH inductance in series. If we indeed will need four Galanz GAL 800E-4 transformers in series has to be established by measuring the exact inductance of these transformers. For sake of the price of this type of MOT (new € 6.50 each) we should not hesitate, though shipment costs are considerable due to their weight.

More about ballast coils can be found in
reference 24.

Voltage Divider

A voltage divider is part of the high voltage power supply (HVPS) and safely permits to measure the output voltage. The divider ratio usually chosen is 1:1000 or 1:10000. The divider consists of a high value HV resistor in series with a lower value resistor over which the voltage will be measured.

Image 42: HV resistor 1 GΩ, 35 kV (source: supplier)

The high value resistor in image 42, resistance 1 GΩ at 35 kV has a power rating of 10 W. The dimensions are 147 mm length with a diameter of 11 mm. It has M4 screw-in connectors and can withstand a peak voltage of 70 kV during 5 minutes.

In commercial high voltage power supplies the voltage divider resistance has sometimes a combined function as bleeder resistance. These resistors have a rather high power rating and may be constructed as a wirewound resistor with one or more shunted connecting points. An example is hsown in image 43, showing a typical high voltage divider resistor with a resistance of 251 MΩ and shunts at 50 kΩ and 25 kΩ.

Image 43: HV Divider Resistor (source: supplier)

Bleeder resistor

When one or more capacitors will be added to the rectifying circuit it will be imperative to have a bleeder resistor in the circuit to ensure discharging the HV capacitors when the power is turned off. Usually we can design the circuit in such a way that the voltage divider resistors can be used also as the bleeder resistor.
It is useful to know what the discharge time will be before we are supposed to approach the HV power supply. Even than, prior to servicing the HVPS, we need to short cut the HV output to ground taking care to do this with the short circuit wire connected to a long plastic stick (just in case....).
The discharge time for the capacitor(s) can be calculated with capacitor discharge calculators available on the internet.

Example: suppose we have two capacitors of each 0.1 µF in our HVPS, connected in parallel, and operating at maximally 25 kV with a voltage divider/bleeder resistor with a resistance of 260 MΩ. These capacitors than have a combined value of 0.2 µF and the initial power in these resistors is 2.43 W, which will be discharged trough the bleeder resistor in 51.4 seconds with a total energy discharged of 62.5 J.  Multiply the time with 3 (arbitrary safety factor) and thus wait 3 minutes before approaching the HVPS.


Voltage and Amperage Metering

Primary Windings

We prefer to have an analog voltmeter and an analog ampere meter in the primary circuit and placed in the housing of the Potential Transformer for observing the presence of power on the PT (image 44 and 45)

Images 44  and 45: Analog Volt- and Ammeter

The ampere meter is connected to a 50A/75 mV shunt in the powerline for a 50 A analog meter (image 46), which has a resistance between the connectors of 5.5 mΩ.

Image 46: Shunt for Ammeter (source: supplier)

Additionally we prefer to observe the actual power on the PT also on our operator's panel. For that purpose we will use a panelmount digital multimeter (DMM) as shown in image 47.

Image 47: Digital panelmount Multimeter (source: supplier)

This meter can be switched to read voltage, amperage, power in W, kWh and time lapsed but will mainly be used to indicate the amperage drawn by the PT, as the voltage can also be read from the autotransformer's meter.

Secondary Windings

In the secondary windings a digital voltmeter and a digital amperemeter will be connected for reading the actual potential in the Fusor and the current drawn.

dig panel voltmeter
Image 48: Digital voltmeter, 5 digit (source: supplier)

The voltmeter is a five digit digital panel voltmeter with a range of 0 - 30 V DC. It needs a separate power supply of e.g. 12 V DC. The fact that it has five digits makes it possible to use a 1:1000 ratio voltage divider in the HV line. The read-out of the voltmeter needs to be multiplied with a factor 1000 to find the actual voltage  on the Fusor.

dig panel ammeter
Image 49: Digital amperemeter, 4 digit (source: supplier)

The amperemeter is a four digit panel amperemeter with a range of 0 - 50 A. It
needs a separate power supply of e.g. 12 V DC and a separate shunt for 50A/75 mV.



A high voltage cable can be bought as a special HV-cable from suppliers (reference 26) or by using coaxial cable of the RG-8 (RG-213) family. When used as high voltage cable it is suitable for several tens of kV. Grounding the outer shield makes the field distribution inside the cable very even, reducing the field concentrations that start corona. RG-8 is rated at 5 kV RMS, however, the polyethylene insulation is 3 mm (0.120 inches) thick which corresponds to 120 kV breakdown. It is expected that the 5kV rating (7 kV peak) allows for a substantial voltage standing wave ratio (VSWR) in transmission line use without breakdown. It is known that many systems use RG-8 at 25 kV, and some have been used at 50 kV using RG-8 as a conductor. Also, the field strength at the inner conductor is higher than that at the outer conductor. Source: reference 27.

For NIM electronics including the HV power supply the cable of choice is RG58/U, 
a cable with an impedance of 50 Ohm (reference 28 en reference 29).


Construction of a HV Power Supply

Introduction to the Construction Chapter

This chapter contains the description of the construction of a High Voltage Power Supply which uses a Potential Transformer (PT) as the HV transformer.
Prior to making a choice for a PT as the high voltage source of our power supply we planned two other versions as the source for our high voltage power supply (HVPS): a Flyback Transformer based HVPS and an X-ray transformer powered HVPS. The reason is that initially it appeared to be extremely difficult to find a suitable X-ray transformer and it seemed realistic to focus at a HV power supply based on a flyback transformer. As usual, development of the flyback HV power supply started on this page of the website in theory, while acquiring the necessary components started at the same time. For logical reasons first of all the
HP server power supplies were bought because these were also needed for the 3D printer project. Just prior to acquiring the variable DC-DC converter and the high power split diode FBT, the X-ray transformer as described below turned up on eBay and we started to develop a HVPS for it. This process took a long time due to testing the X-ray transformer and establishing its characteristics, followed by developing a cooling circuit for the oil tank which houses the X-ray transformer. Then, about a year later we got hold of a suitable PT, which is apparently the Rolls-Royce under the high voltage transformers.
Well, it could be the RR under the transformers if it only had been bipolar, which would have made it easier to construct a double wave rectified HVPS suitable for integrating a voltage doubler circuit for raising the poutput voltage. But we will have to deal with what we have got and therefore we started constructing our first HVPS, capable of complying with the minimal requirements for a true Fusor HVPS and consisting of a single monopolar PT.

It should, however, be noted that we can overcome the disadvantages of a half-wave rectified circuit, based upon a monopolar PT, when we can get hold of an identical second PT to be used in the following dual LPT full-wave rectified doubled voltage Cockcroft-Walton multiplier circuit (image 50):

Image 50: Dual LPT full-wave rectifier with CW voltage doubler (© FRS 2017)

For more details on this particular circuit, which is suitable for two identical transformers with one outpout terminal connected to the core, see reference 30.

As described earlier on this page the half-wave rectified transformer in its simplest form
has a pulsed DC output (image 51).

Image 51: Half wave rectified transformer (© FRS 2016)

With a 50 Hz, 100 V AC input, a transformer turns ratio of 1:200, the DC output at high impedance will be (theoretical) a peak voltage of -20.3 kV, an RMS voltage of -10.0 kV and an average voltage of -6.2 kV. On a virtual scope the wave signal is as shown in image 52.

Image 52: Scope signal HVPS half-wave rectified at high Fusor impedance (© FRS 2016)

We can see on the scope signal that the peaks have an interval, where the voltage drops to zero.
In this simulation we have used a load resistance of 1 G
Ω, which has been arbitrary chosen to reflect the high  impedance of a Fusor at start. The lowest resistance will be obtained when the Fusor has a discharge and this impedance is approx. 50 k Ω, the value of the current limiting ballast resistor. Under that condition in the simulation the same DC output voltages will be present, but a higher current will be running of about 400 mA. On the primary winding at 100 V AC input this will result in peak currents of about 80 A (unless limited).   

Adding a capacitor parallel to the load will smooth the pulsating DC voltage. The larger the capacity, the more filtering action will be applied by the capacitor. Because we have two HV capacitors of each 0.1 µF at 40 kV and 2 kA we will put these in parallel in order to increase the capacity (image 53).

HVPShalfwave cap
Image 53: Half-wave rectified transformer with capacitors (© FRS 2016)

The resulting DC output at high impedance is now
a peak voltage of -20.2 kV, an RMS voltage of -19.2 kV and an average voltage of -19.2 kV or, the DC output now has become almost constant. Almost, because the DC output has now a ripple of peak to peak 1,95 V (image 54). Note that this result is load dependent!

Image 54: Scope signal HVPS half-wave rectified with capacitors at 750kΩ Fusor impedance (© FRS 2016)

The difference with the earlier scope signal is that here we only see the amplitude of the ripple of 1.95 kV around a Fusor potential of average -19.2 kV with an arbitrary Fusor impedance of 750 k
At a Fusor impedance of 50 k
Ω (Fusor discharge) the expected HV DC voltage on average will be -15.6 kV at 300 mA and a peak value of -20.7 kV at 415 mA. This represents in the primary winding at 100 V AC input an average current of 60 A or a peak current of 370 A, unless limited.

This is a very theoretical assumption, realised with simulation software, and operating the Fusor will learn what we can expect. It should however be clear that the Fusor in operation will be pending between the highest and the lowest impedances and the trick will be to maintain the Fusor operational at the required parameters.

For the first setup of  High Voltage Power Supply we will use the following schematic circuit (image 55):

Image 55: Schematic circuit High Voltage Power Supply (© FRS 2016)

PT = Potential Transformer: n= 1:200; prim. 100V, sec. 20000V, 10 A max.;
D1 = D2 = D3 = diode HVM 12, peak reverse voltage 12 kV, average forward current 350 mA;
C1 = C2 = Capacitor 0.1 µF, 40 kV, 2000 A;
R1 = 1 G
Ω, 35 kV, 10 W;
R2 = 1 M
Ω, 35 kV, 10 W;
R3 = R4 = 100k, 200 W wirewound;
R5 = 100
Ω, 5 Watt or a commercial shunt for 50 A/75 mV;
Digital voltmeter 0-30 V, 5-digit;
Digital Amperemeter 0-30 A;

Fusor impedance: arbitrary, variable under operation.

Additional components (primary circuit):
Variac, 230 V AC, 10 A;
Ballast inductor 70 mH 10A;
Voltmeter 0-300 V;
Amperemeter 0-50 A;
Circuit breaker 16 A.

The voltage divider with a total resistance of 1001 M
Ω has an additional function as a bleeder resistance for discharching the capacitors after switching off the power supply. The energy to be discharged is 62.5 J with a time constant of 200.2 sec. This produces a discharge time to a safer voltage of 50 V of almost 21 minutes. Applying a safety factor of 3 we should stay clear from the HV power supply for a time period of one hour.

triangle It is not permitted to touch or approach the power supply within a time period of one hour after switching it off.

Always short cut the HV outlet to ground before disconnecting the power supply from the Fusor or opening its case for service purposes.


deuterium monitors

engineering works
control systems
Ion Source


Ref.1: Fusor net: High Voltage FAQ: http://www.fusor.net/board/viewtopic.php?f=29&t=4843

Ref. 2:  X-ray transformers: http://adammunich.com/x-ray-transformers/

Ref. 3:
HV power supply schematic: http://robofusor.aditl.com/?page_id=94

Design Guide for Rectifier Use: http://www.hammondmfg.com/pdf/5c007.pdf

Ref. 5:
Actual data from unknown source; image from http://www.stefan-kluge.de

Ref. 6:
Oil burner ignition transformers: http://www.kronjaeger.com/hv/hv/src/obit/index.html

Ref. 7: 
Microwave oven transformer: http://www.kronjaeger.com/hv/hv/src/mot/index.html

Ref. 8:
Flyback transformer: http://www.kronjaeger.com/hv/hv/src/fly/index.html

Ref. 9: Flyback transformer: http://en.wikipedia.org/wiki/Flyback_transformer

Ref. 10: Flyback transformer pins layout: http://lifters.online.fr/lifters/labhvps/tht.htm

Ref. 11:
FBT driver for split diode FBT: http://www.webalice.it/saverio.filippi/royer_eng.html

Ref .12:
FBT driver with 555 timer: http://www.eleccircuit.com/efficient-flyback-driver-circuit-by-ic-555-irf510/

FBT driver with one transistor: http://www.loneoceans.com/labs/flyback/

Ref. 14:
HPPS conversion: http://www.ultimaterc.com/forums/showthread.php?t=174225

Ref. 15:
HPPS conversion: http://sites.google.com/site/tjinguytech/my-projects/HP47A

Ref. 16:
Core saturation of Potential Transformers: http://www.3d-meier.de/tut15/Seite40a.html

Ref. 17:
Voltage doublers: http://en.wikipedia.org/wiki/Voltage_doubler

Ref. 18:
Cockcroft-Walton (history): http://www.aip.org/history/lawrence/epa.htm

Ref. 19:
Cockcroft-Walton (theory): http://blazelabs.com/e-exp15.asp

Ref. 20: CWM Calculator: http://www.extremeelectronics.co.uk/calcs/index.php?page=cwvoltage.php

Ref. 21: Marx Generastor: http://en.wikipedia.org/wiki/Marx_generator

Ref. 22: Marx Generator: http://www.kronjaeger.com/hv/hv/src/marx/index.html

Ref. 23: Ballast Coils: http://myweb.tiscali.co.uk/teslatutorial/ballast.htm

Ref. 24: Fusor net, HV FAQ: http://www.fusor.net/board/viewtopic.php?f=29&t=9383

Ref. 25: Resistor Network for Series Capacitors: http://www.coilgun.info/theorycapacitors/capacitors2.htm

Ref. 26: High voltage components: http://highvoltageshop.com

Ref. 27: Reference to cable info: https://groups.google.com/forum/#!msg/heatsynclabs/rlzMKPioyDA/GHGLrV30oyEJ

Ref. 28: NIM HV cable: http://en.wikipedia.org/wiki/Coaxial_cable

Ref. 29: NIM HV cable: http://en.wikipedia.org/wiki/RG-58

Ref. 30: High Voltage and X-ray Experiments: Experiments with various multiplier designs (I): http://www.celnav.de/hv/hv9.htm

Last Updated on: Sat Feb 18 17:21:39 2017

Flag Counter

Document made with KompoZer