High Voltage Transformers – Isolation, Day 9

I go through my collection of iron/ferrite core high voltage transformers. MOTs, OBITs, NSTs, PTs and X-ray transformers. Microwave Oven-, Oil burner ignition-, neon sign- …

TL494 flyback driver

It took some years and someone recently bumping up my old thread about this project for me to write up a article, find the pictures, …

TL494 flyback driver

Published on: Jun 14, 2013. Updated on: Nov 28, 2017.


I wanted to design a versatile driver circuit that could drive a half- or full-bridge of MOSFETs or IGBTs through a gate drive transformer (GDT). This should make a driver that is able to run flyback transformers found in CRT TV sets and computer monitors.

The TL494 IC is designed for maintaining all the functions needed in a switching mode power supply using pulse width modulation (PWM). The output transistors can be run in either single ended mode or push-pull. The pulse width is normally controlled through a feedback signal in the power supply, but for this project we want to control it manually, this is done differently in almost all schematics found.



WARNING!: Working with electricity is dangerous, all information found on my site is for educational purpose and I accept no responsibility for others actions using the information found on this site.

Read this document about safety! http://www.pupman.com/safety.htm



Flyback transformers from a CRT TV are typically driven at 15 kHz and flyback transformers from computer monitors are typically driven between 30 to 150 kHz.

The TL494 IC uses a 5% dead time to insure proper switching and at frequencies over 150 kHz this minimum dead time is higher.

The design goals for this project will be a driver with a variable duty cycle from 0% to 45% and a variable frequency from 50 kHz to 150 kHz.

This should make for a efficient driver and one that works out of the audible spectrum. In order to design with components at hand, the frequency span is not going so low as 15 kHz.



Voltage supply IRFP250N: 0 VAC to 120 VAC
Frequency span 38 kHz to 150 kHz.
Duty cycle span 0% to 43%




25th May 2009

The breadboard prototype is ready to be tested, the tape is to hold the timing capacitor in place since the legs on it was too short.

In the first oscilloscope shot we see the output waveform without pull up resistors, it is about 38 kHz at 43% duty cycle.

In the second oscilloscope shot we see the output waveform without pull up resistors, it is about 38 kHz at 5-7% duty cycle.

In the third oscilloscope shot we see the output waveform without pull up resistors, it is about 150 kHz at 43% duty cycle.


27th May 2009

PCBs was made for both the driver and half-bridge section. The full bridge rectifier used here in the pictures is only rated for a mere 4 A. This is not enough for running a flyback with low input voltage and high duty cycle. A 25 A bridge with heat sink should be used to ensure some overhead.



29th May 2009

In the oscilloscope shot we see the waveform of the primary side of the GDT driving a MOSFET half-bridge. To test the circuit I first used a old half-bridge I had from an earlier project.

The sturdiness of this new driver shines through when I killed a flyback transformer due to over-voltage on the secondary side. Corona glow can be seen in the center towards the ferrite core.



This universal inverter makes it possible to adjust the output voltage and current exactly to ones needs. It makes a great and much more sturdy flyback driver than many simple drivers with just a single transistor, which is of course no surprise as it implements its own control IC, MOSFET driver ICs and a half-bridge of MOSFETs.

For a final constant voltage or current power supply it will not work, as there is no feedback adjusting the pulse width to a certain load.



Simple plasma globe


When I first started experimenting with high voltage, flyback drivers were among the first supplies I built. I did a lot of weird experiments, but never got around to make a plasma globe since I wanted to do it with a big clear bulb.

Now I bought a big bulb, I have some better flyback drivers… 1 + 1 … it is time!

Warning: Touching the globe can give some nasty shocks as the current is much higher in a home made globe, I only tried at very low input voltage and it was easily felt as sparks to the finger.


WARNING!: Working with electricity is dangerous, all information found on my site is for educational purpose and I accept no responsibility for others actions using the information found on this site.

Read this document about safety! http://www.pupman.com/safety.htm

The experiment

1st September 2009

I bought a 125 mm diameter Paulmann globe, the biggest the store I popped into had around.

There is a few conditions that have to be met to get discharges inside the bulb.

  • An AC high voltage supply, old flybacks without rectifiers are perfect.
  • One lead is connected to both terminals of the bulb
  • Second lead is grounded to earth.

Then its all about getting a high enough voltage for the discharges desired.

I noticed that the gas surrounding the actual streamers was glowing green, at first I thought it was overexposure to the eyes optic nerve, but when I got a photograph of it, it got me wondering. Normal light bulbs are mostly filled with Argon and some nitrogen / krypton, but none of these gasses, pure, will emit green. So for now it must be the mix of gasses in the bulb that results in a green glow around the streamers.

Beautiful green glow can be a sign of x-rays, but the supply voltage here is below the limit for x-ray generation. If the supply voltages gets above 50-60 kV the risk of x-rays being generated is present.



This is an easy and fun experiment to do, if you already have a high voltage AC power supply at hand.



Floating waterbridge

WARNING!: Working with electricity is dangerous, all information found on my site is for educational purpose and I accept no responsibility for others actions using the information found on this site.

Read this document about safety! http://www.pupman.com/safety.htm


This experiment was first carried out by Lord Armstrong in 1893. Read the historical tales of this experiment.

De-ionized water acts in a special way when exposed to high voltage. This experiment uses two normal household drinking glasses filled with de-ionized water and a wire in each glass from a 20kV flyback transformer, I used one of my older drivers for the supply.

The experiment

7th may 2009

The drinking glasses are filled to the brink with de-ionized water and put close together, the water will climb up the sides of the glasses and form a bridge of water between the two. Very carefully the glasses can be pulled apart to gain a longer bridge, but eventually it will not be able to sustain due to gravity.

I was able to make a 12 mm long water bridge as seen in these images.


Additional reading about this experiment

Elmar Fuchs and colleagues from the Graz University of Technology in Austria

Peter Terren from tesladownunder.com


This is an easy and fun experiment to do if you already have a high voltage low current power supply at hand, reproducing experiments by scientists is a great way to acknowledge their discoveries and learn some new theory and history.


Kaizer SGTC I


This Tesla coil is my first and was build without any expenses worth mentioning, its the prototype from which I learned a lot about Spark Gap Tesla Coils, high voltage and where to find components in household items and trash.

In the development the first version was more of a proof-of-concept model build only from old microwave ovens, televisions and cable.

Mathematics and theory was not the leading part of this project in the start, but as optimizing went on I learned about tuning the Tesla coil for a better output, resulting in longer sparks.



WARNING!: Working with electricity is dangerous, all information found on my site is for educational purpose and I accept no responsibility for others actions using the information found on this site.

Read this document about safety! http://www.pupman.com/safety.htm



There is no protection against RF spikes going back into the High voltage supply and destroying it. 

The secondary coil is limited by the amount of copper wire I had available from the cooling fan motor from a microwave oven. The diameter / height ratio of the secondary is far from optimal.

I used my flyback transformer as the power supply, it have a very limited current at about 1mA at 20kV, roughly estimated.

In the start I used home made salt water capacitors which are far from optimal.



High voltage supply20 kV from a flyback transformer
Primary capacitor
Primary coilinner 45 mm, outer 90 mm diameter, 1.78 mm diameter isolated copper wire, 6.6 turns.
Secondary coil50 mm diameter, 113 mm long, 800 windings, 0.127 mm enamelled copper wire.
Resonant frequencyTuned at around 655 kHz.
Topload60 mm diameter sphere, tennis ball wrapped in aluminium foil.
Input powerAround 20 – 30 Watt
Spark lengthup to 105 mm long sparks.





10th May 2008

The secondary coil was wound on a plastic tube, 50mm in diameter and 160mm in height. Its bottom was conical so practically there could only be wound wire on 110mm of the height.

The copper wire had a diameter of 0.127mm, varnished its diameter is 0.14mm, this very thin wire made it hard to wound nicely without overlaps.

It took me 2 days to wound the secondary coil, it was hard on the eye, arms and hand to do in one stretch. The plastic tube was mounted on a gear motor controlled by a frequency inverter so I could control the speed by a potentiometer.

11th May 2008

To keep the windings in place and insulate the coil it was varnished with common ship varnish, it was given a layer in the morning and one more in the evening.

12th May 2008

As the power supply I used my 20kV flyback transformer driver at 12Vdc input, the capacitor is a home made salt water capacitor with a capacitance of 2.7nF. The primary coil was wound of ordinary 2.5mm2 wire around a cut up soda bottle.

The topload is a sphere made of bobble wrap, tape and aluminium foil.

Spark gap is just 2 copper wires.

In the first test I could get 4-5mm sparks to a grounded object, the biggest problem was the too tight coupling between primary and secondary. Too tight coupling resulted in alot of racing sparks on the secondary coil, these are dangerous as they can destroy the thin wire on the secondary coil.

Here is a picture of some racing sparks I provoked to get a picture of it.

The primary coil was wound in a bundle and by adjusting the height of it, the coupling could be changed. It was now possible for 40mm sparks to jump to a fluorescent light hold in my hand.

15th May 2008

varnish, varnish, varnish, varnish, glue, glue and more capacitance…

The flyback transformer driver runs at 12Vdc input.

I made another salt water capacitor and installed it in parallel with the other, the total capacitance was now 5.9nF.

As it can be seen in the pictures there is sparks or just “violet light” coming from other parts than the topload of the Tesla coil, its corona loss and decreases the spark length.

To isolate the secondary coil further it was given 4 more layers of varnish and the top of it was glued all over with hot glue. The metal cap was the new topload and 54mm sparks could be achieved.

28th May 2008

The flyback transformer driver runs at 17Vdc input.

85mm sparks can now jump to my fluorescent light.

The bottom from a beer can is the new topload, it results in some spectacular pictures, the higher voltage on the flyback transformer driver is the best improvement towards longer sparks, but also racing sparks on the secondary starts reappearing.

As it can be seen in the picture taken in the dark, there is still corona losses, and with this particular coil it will be impossible to avoid it at these driver input voltages. It is not easy to insulate 100kV. The following picture is taken with long exposure to show the corona around the coil itself, unfortunately its not as clear in the picture as seeing it live. The violet field around the coil is faint, but can be seen in the picture.

This picture is taken directly above the Tesla coils topload.

21st June 2008

I bought 10 old high voltage capacitors at 150 dkr (25$) for the lot, a real bargain in Denmark compared to the joy it has brought me. I only use one of them instead of the 2 salt water capacitors. Its a Fribourg Condensateurs from 1961, 8nF rated for 20kV pulse driving at maximum 2MHz.

A new topload was made from a tennis ball wrapped in aluminium foil, its not as smooth a surface as it should be, but it works.

The spark gap now consists of 2x WT20 tungsten welding electrodes, these are able of withstanding high temperatures without vaporizing as the copper wires were likely to do over time. There is still room for improvement on the spark gap as its still just a single jump static spark gap.

By calculating the Tesla coil in JAVATC I tuned the circuits to theoretically be in resonance, but its only approximating as the construction is far from precise.

With the new improvements 105mm sparks will jump to a grounded wire.

I took a series of pictures with different exposure times. Sparks are about 90mm long.

In the next picture racing sparks can be seen at the top of the secondary windings, without a breakout point or something to jump to, the energy build up is too large for the coil and its under a huge stress.



This small project started as a proof-of-concept model, to see if the theory I had learned would work in practice. It has come a long way since I started on it almost 2 months ago.

I learned a lot more about the theory of the Spark Gap Tesla Coil, the maths behind tuning the circuits and the importance of planning the design before building. This is no surprise. So there have been spend a lot of time trying to optimize a Tesla coil build around a badly designed secondary coil, so the result will never be near optimal.

Despite all this, I am very satisfied with the results of 105mm sparks. There are several things to optimize in a final version of the Kaizer SGTC I, it would be a better spark gap, shorter and more suitable wires, shape of the primary coil and its coupling to the secondary and a topload with a smoother surface.


2n3055 flyback driver


This driver is among the simplest, with just six components it will be able of delivering high voltage with a strong enough current for various experiments.

There is also a push-pull version of this driver, more on that can be found following this link: http://wiki.4hv.org/index.php/Flyback_transformer


WARNING!: Working with electricity is dangerous, all information found on my site is for educational purpose and I accept no responsibility for others actions using the information found on this site.

Read this document about safety! http://www.pupman.com/safety.htm


The transistor used in this circuit should not have a voltage rating higher than 250V, due to the higher base current needed of higher voltage rated transistors, this might in some cases cause problems where the self-resonating circuit can no longer drive the transistor.

At voltages higher than 12 VDC input, the components will dissipate a lot of heat and the 2n3055 transistor is most likely to burn as it is not rated for more than 15 A and it has to be derated even further for case temperatures above 25 oC.

MOSFETs are not suitable in this driver.

The frequency of operation in this circuit is determined by the capacitor across the transistors emitter and collector, experiment with the value of this capacitor to find the best performance, this capacitor have to be a good film or foil type (MKP/MKT)



The driver is supplied with 12 to 17 VDC from a computer ATX PSU.





3rd may 2008

The driver is built on a piece of vero board with multiply resistors to obtain the needed wattage rating, this is far from optimal as the load sharing between them is horrible, to ensure better sharing.

The 2n3055 transistor is mounted on a 30 x 10 cm aluminium profile, this is just about enough to keep it alive doing long runs.

4th may 2008

I experimented with voltages between 12 to 17 VDC, there is no incredible performance with higher voltage compared to the amount of heating the higher voltage generates.

Different capacitors were tried, ranging from 10 nF to 300 nF, 220 nF was found to work the best for my flyback transformer.

8 primary windings and 4 feedback windings gave the best results.



7th may 2008

18 to 20 mm sparks were achieved

Keeping the electrodes just far enough apart for no spark to jump, a beautiful corona breakout is visible.

With the output coupled through a home made salt water capacitor it was possible to have loud and very bright sparks.



This is a very simple and cheap circuit to achieve around 20 kV high voltage, but where it excels in simplicity it lacks a lot in stability and efficiency.

A rough estimate is that 25-50% of the input energy is wasted as heat in the transistor.

It is cheap and simple, but inefficient and unstable.



Mazilli ZVS flyback driver


The Mazilli ZVS flyback driver is well-known throughout the high voltage community for its simplicity and ability to deliver 20-50 kV at high currents for a flyback transformer.

I build this circuit almost a year ago on a vero board, but it kept blowing the thin traces due to high currents flowing. I eventually put the project in a box and forgot all about it.

Inspired by the point to point soldered designs Myke from the 4hv.org forums often uses, I tried to make something in that manner, not as pretty as his work though.


WARNING!: Working with electricity is dangerous, all information found on my site is for educational purpose and I accept no responsibility for others actions using the information found on this site.

Read this document about safety! http://www.pupman.com/safety.htm


The MOSFETs used need a voltage rating about 4 times higher than the supply voltage and a on-resistance below 150 mΩ

5 + 5 primary windings are suitable for voltages between 10 to 40 VDC, at higher voltages additional windings will be needed. Experiment with the number of windings to improve performance. Too few windings will result in excessive heating and too many will result in reduced power output.

A MMC is made from 6 capacitors to avoid excessive heating in a single capacitor.

This driver will push as much power as it can, so be sure to use flyback transformers that can handle the abuse if you want it to live.



Voltage supply 35 VDC from a rewound microwave oven transformer.
MMC 0.66 uF from series string of 3x 2 275VAC MKP X2 capacitors in parallel .
Power consumption 400 Watt.
Longest arc 100 – 110 mm long white arcs



15th may 2009

I have now rebuild the driver using 2.5mm² / 14AWG wire for a good current ability, larger heat sinks and a MMC to avoid as much heating as possible.


16th may 2009

I found 4 different flyback transformers from my collection, among these are a 1980’s Bang & Ollufsen television flyback. A small flyback from a photocopier. A  flyback from a 1990’s portable television, it is without screen and focus resistor networks. A flyback with rectifier tube from a black & white 1950’s television.

The pictures with long arcs about the size of  100 – 110 mm was made with the 1980’s Bang & Ollufsen flyback transformer.


It was well worth it to rebuild this driver. It can now handle long run times with little heating despite pushing out around 400 Watt  of power!


555 Audio modulated flyback


This is a audio modulated arc generator designed for simplicity rather than reliability, its made with very few and common components. There is however some serious trade offs described below in considerations.

WARNING: sensitive audio players might get damaged by this circuit. I bricked my iPod shuffle, seems that the controller chip for the mini jack got wasted as it could no longer detect charger, PC connection or play music as it could not detect headphones.


WARNING!: Working with electricity is dangerous, all information found on my site is for educational purpose and I accept no responsibility for others actions using the information found on this site.

Read this document about safety! http://www.pupman.com/safety.htm


The arc have to be very short in order to limit the distortions of an unstable arc. The sound quality is low due to the way the audio modulation is implemented. If the distance between the elctrodes is too large, the high open loop potential of the high voltage transformer can generate some high transient voltages that through inductive kickback can destroy the MOSFET.

The 555 IC supplied by 12 VDC can not source much more than 140 mA before the voltage drop on the output gets very high. At 140 mA it is already 1,82 VDC. At 200 mA the voltage drop is at 2,5 VDC. The low output will affect MOSFET switching speed and result in higher losses. This graph shows the voltage drop vs. output current of the 555 IC.

In order to optimize the switching of the MOSFET, a small intermediate driver stage can be introduced with two transistors, a NPN and PNP. As illustrated in the red and green graph is the difference between running a MOSFET with proper switching and the other always in linear mode, where losses are very high. This is a future improvement and is not a part of this little project, but it is recommended to add this if you want reliability.

If the circuit can not produce a arc try to reverse the polarity of the primary coil on the flyback transformer.

There are basically 2 kinds of modern flybacks, television flybacks are driven near 15kHz and monitor flybacks are driven between 30-150Khz. Depending on which type we use, we have to adjust the frequency of the 555 timer to match the resonance of the flyback for maximum performance.


Choosing a MOSFET

There are some basic rules of thumb that I will just list here to start with, I will come with an explanation later on.

The voltage rating of the MOSFET (VDSS) needs to be 6 to 10 times higher than the supply voltage, reverse voltage spikes and EMF can be high enough to destroy the MOSFET if its too small. But we still need to use MOSFETs with a reasonable low on resistance (RDS(on)). Try to find a MOSFET with a RDS(on) value not much higher than 0.1 ohm, if you have problems try one with a lower RDS(on) value.

The gate resistor R3 is there to

  • Limit parasitic oscillations that could kill the MOSFET.
  • Limit the current that is needed from the driver stage, in this case our 555 timer.
  • Protect against surge voltages on the MOSFET gate, effectively this would require a much higher resistance, a high gate resistance would lower the operation speed significantly.
  • The values of a gate resistor could be anything between 10 ohm to 200 ohm, it all depends on the MOSFET, experimentation is needed. The alternative is complicated calculations involving data that is usually not available in standard data sheets.

How does the audio modulation work?

Pin 5 on the 555 timer is a direct access to the 2/3 voltage divider point of the upper voltage comparator in the 555 timer. This allows us to pulse width modulate the output on pin 3 of the 555 timer. By applying a voltage to this pin, it is possible to vary the timing of the chip independently of the RC network. When used in the astable mode, as we do with this circuit, the control voltage can be varied from 1,7 VDC to the full Vcc. Varying the voltage in the astable mode will produce a frequency modulated (FM) output.
If the control-voltage pin is not used, it should be bypassed to ground, with a 10n capacitor to prevent noise entering the chip


Both R1 and R2 can be 10K potentiometers.


13th November 2008

I wanted to do a audio modulated flyback arc with few components and a small form factor. I installed the MOSFET on a old CPU heat sink with fan, the 555 timer circuit is also installed underneath this heat sink, its then all put on the side of the flyback transformer with wire strips.

The primary coil is 8-9 windings of 0,75 mm² isolated wire. More windings will stress the MOSFET less but also output voltage will be lower.

The frequency output from the 555 timer is 26,7 kHz at 59,3% duty cycle. This is in the low end for a monitor flyback so further improvements will be adding a potentiometer to adjust frequency to match the resonant frequency of the flyback.

2nd February 2009

Its time to improve the driver with a variable frequency control so the driver can be used with most conventional monitor flyback transformers without changing any parts, but merely turn the potentiometer.

I installed a 9K potentiometer as R1 and a 10K potentiometer as R2, I adjusted the potentiometers till I had a nice silent thin arc at about 15 mm length. 10K potentiometers can be used for both R1 and R2, I just used what I had at hand.

Using a 555 calculator with the measured values of the potentiometers. R1 at 1K3 and R2 at 1K. Duty cycle is 69.7% and frequency is 43700 Hz. Very reasonable for a monitor flyback. Compared to the old frequency I now have a longer and more silent arc.



A quick and very rewarding little project, its fun to play music without conventional speakers. This was also known in the 1970’s as a plasma tweeter and could be found in special hi-fi speakers.

The arc is very very hot and I had to extend the copper wires where it is drawn between to avoid the heat being transferred far enough to start melting the flyback transformers casing.

The 555 IC is not able to supply enough output current to drive a IRFP250N MOSFET at a high duty cycle, so the MOSFET will at times still be in linear mode and this causes excessive heating, which is why the heat sink is necessary. So more notes under considerations about this.