Where Sparks Fly

A lifelong fascination, the Tesla Coil just had to be built one time or another. And now's as good a time as any! A long time ago I started with a Tesla Coil but not grasping the concept of the two tuned resonators I did not get any results and flung the construction in a corner. When I came across websites dedicated to the device I decided to have another go. Initially I wanted to go really fast 'n' easy, converting a computer power supply into a SSTC (Solid State Tesla Coil), but other projects interfered. About a year later I talked to someone and about it decided to revive the project. Only this time, fearing a pile of dead semiconductors, I went for the entertainment electronics workhorse, the venerable PL519.

The choice was easy: I have a whole stack of these tubes and if I melt one down I can easily plonk in another one. Of course this is hard to do, so far the first tube survived heavy torture...

The Design

I have an aversion of power oscillators, so I decided to use a VCO to generate the drive for the output stage:

This is simply a 4046 datasheet implementation for a VCO with a frequency range of 90 to 255 kHz. The tuning voltage is derived from a precision 10-turn potentiometer (2k). Then a driver/buffer stage follows. I had made an inverted output to drive the MOSFETs of the computer power supply, but it's not needed now.

The output stage consists of a voltage amplifier and the PL519. It's failsafe, as the drive fails the control grid is driven into cutoff. The first 2N5401 (an universal 160 V transistor) amplifies the square wave and the emitter followers buffer the output to compensate for the capacitance of the grid. The anode tank is tuned to the Tesla Coil. And finally the screen grid is connected to the anode voltage.

I tried to use the PL519 in the area it's designed to work: low anode voltage and large current. Kilovolts should be applied to transmitter tubes and not to line output tubes. And that brings us to the power supply:

It consists of an old table radio transformer, which under full load (some 50 Watts) drops to 200 V. The filament voltage is provided by the power transformer of one of those crap Chinese stereos, and is a bit low at only 38 volts. Another winding provides the grid bias voltage. The anode voltage is stored in a TV combination electrolytic capacitor totaling somewhere around 300 µF. The bridge rectifier is liberated from a (surprise!) computer power supply.


Actually, the secondary (Tesla) coil was built first. It consists of a PET soda bottle with a diameter of 86 mm. Then about 20 cm was wound with 0.1 mm enamel copper wire. Then a polystyrene sphere was cut out, covered with aluminum foil and fitted on top of the bottle. The wire was wound with a jig with a CD motor, otherwise I'd gone crazy...

The primary is just a bunch of wire tuned to the secondary's frequency of about 175 kHz. The tuning capacitor is from a Philips color TV (8200 pF @ 2 kV). Should be able to handle the 0.9 kV that the PL519 develops.

The rest is just an old setup I modified for the Tesla Coil. Including the Chinese PoS stereo transformer. The meter was fitted with a custom scale. Whenever the needle goes off-scale I know the PL519 draws too much power and will be unhappy...

The frequency of the VCO is adjustable with the 10-turn control:


The performance of this setup is just lousy. I get only about 2 cm sparks from it:

Fortunately, there are many avenues for improvement. First off, the secondary. From the 4hv.org forum I learned that winding the Tesla Coil with 0.1 mm wire is not a good idea. Minimum thickness should be 0.2 mm. Then it could also be a bit higher/longer, so I got myself some nice PVC tubing (80 mm). To suppress the parasitics at 60 MHz I could use a control grid resistor (0.1 kV/div):

There is already a small choke in series with the anode lead, but apparently that doesn't help much. Also the duty cycle of the drive signal could be reduced so the drive can be increased to fully utilize the characteristics of the PL519 tube ("mo' current, mo' powah!"). Now I also understand why plate current "dips" when the anode tank is in tune: the current caused by the drive is exactly in phase with the anode voltage (even negative!) so the average current is minimal. Actually there are two dips because the anode tank and the secondary do not have equal frequencies. I haven't figured out which causes the greater dip.

All in all a great project to lose myself in!



2 April 2009