Tesla coils


Tesla coils are electrical devices designed to generate high voltages at high frequencies that often manifest in the form of large lightning-like discharges, which is often their primary attraction. Below, I describe some Tesla coils I have built. These devices are dangerous due to the high voltage and current involved. If you decide to build or operate your own, you do so at your own risk.






mark I



This is the first Tesla coil I built. It is a fairly typical spark-gap coil. Below are the major parts and descriptions, along with images and videos of it in operation. Click each picture for a larger version.

This is the original power supply for the coil. It consists of four microwave oven tranformers (MOTs) wired in two banks of two. The banks form the two halves of a center-tapped supply, and each bank is wired with the primaries and secondaries in series and in phase to take 240V from a dryer outlet and output roughly 5000 volts. Capacitors sit between the secondaries of each transformer in each bank to provide power factor correction and to limit output current. The output from the whole pack is therefore about 10kV.
This is the current supply. Something, perhaps stray metal shavings, shorted out the secondaries of two of the MOTs on the original supply. They were replaced, connections were made more robust and better isolated, and PVC output bushings were added; you can see them to the right of the image.
This is the original capacitor bank, i.e. 'tank cap'. Its purpose is to store energy from the power supply and be discharged a few hundred times each second into the primary coil. It consists of 19 TDK UHF-9A 2nF 40kV ceramic pulse caps. Ceramic caps are not usually a good choice for Tesla coils, but these were designed specifically for quick discharges in rapid succession.
This is the current capacitor bank. The old one was dismantled and the plates were reused to make this, a bank of 17 4nF 20kV capacitors of the same type as the original. The extra capacitance provides a substantial boost to the energy per discharge, generating longer sparks. A 50Mohm high-voltage resistor was placed across the bank to dissipate charge after turning the coil off. This should not actually be necessary because the capacitor will discharge through the power supply.
The original spark gap was simply two copper pipes with end caps soldered on, held in a wooden rig. During operation, a leafblower was positioned to blow air across the gap to cool the gap and quench the spark. It worked fine, but the leafblower was not affixed to anything and would frequently slide out of position.
The next gap was a synchronous rotary spark gap (SRSG). A 3450RPM induction motor was modified by grinding two flats on the rotor, thus making it a 3600RPM motor that spins in phase with the AC line voltage. The polycarbonate rotor can hold various numbers of electrodes, which bridge enough of the gap between the stationary electrodes that a spark can form. This gap is quickly opened again by the rotation of the disk.

During testing, it was found that this gap had a serious problem of unknown cause: it sucked ass. Spark length was about half that achieved from the previous gap. Maybe more tuning will be done later, by varying the positions of the electrodes, but for now it sits on a shelf.
Here is the latest spark gap. It is a "sucker gap", using a vacuum to suck air through the gap to cool and quench the spark, but it is inside-out, drawing air inward through holes in the gap face. It is enclosed in a PVC tee fitting, which blocks most of the light from the arc, making the streamers from the top more visible. A wet-dry vac is attached to the rubber adapter during operation.
Another view of the sucker gap, showing the electrodes. Note that the heat and UV from the arc have discolored the PVC inside. Arcs really do generate a lot of UV. Welders wear goggles for a good reason.
Streamers from first light. I hung a CD off a nail as a breakout point; it is getting rather fried here.
This time without the CD. Note the leafblower keeping the gap cool. This was not the greatest setup.
Here are some streamers from the fully upgraded version. This is not a particularly good representation of the sparks as they appear in person; for a better idea of what they look like, see this short video.
More streamers. They look blue-white and more substantial in person, and at any one time there is usually only one main streamer flying around. Note also that there is no light visible from the gap, which is quite an improvement from the first revision.

back to top




mark II SSTC



My second coil was a solid-state coil, which drives the primary continuously with MOSFETs instead of pulsing it with a capacitor and spark gap. This allows operation directly from 120V mains, an improvement from having to drag around 50 pounds of high-voltage transformers. However, it is much more complex and costs substantially more. The sparks it produces are shorter, but they have a fibrous quality to them. Due to the low peak current, sparks can be safely drawn to a metal rod held in the hand. Direct contact with the spark will cause a deep RF burn that takes a long time to heal. For this reason it is also important to hold on to the rod firmly, to ensure good electrical contact. Currently the coil is inoperative; I suspect that the last time the power was cranked up, a couple of MOSFETs got smoked.

This is the whole unit. I think it looks nicer than most of my projects. The secondary is about 1600 turns of 28AWG magnet wire on a 4.5" PVC form, coated with enough coats of polyurethane to make it feel smooth. The primary is 10AWG stranded wire, wound on a piece of thin-wall plastic tube that sits in a groove milled in the base. You will note that that same green wire shows up in pretty much all power electronics I make.
This is what is underneath the deck. You can see the section that produces smoothed DC for audio modulation (to the left; only partially successful), the main relay, the power and cap discharge switches, and the electronics.
Here is the power section. The switch on the right discharges the capacitor through the big wirewound resistor next to it. Behind the switch plate is the bridge rectifier that feeds the cap and a single rectifier for half-wave, and behind that the relay that controls power to the power circuitry. The two jumpers on the far right control whether the MOSFETs are fed with half-wave or smoothed DC.
This is the circuit board. I did not design the board; it is a 'PlasmaSonic II' purchased from someone whom I no longer want anything to have to do with. I populated it and made some modifications to get around various limitations, such as the colossal suck demonstrated by the original MOSFET gate drivers.
The gate drivers were TC4421 and TC4422 chips from Fairchild Semiconductor. They can supposedly pump 9A peak to switch MOSFETs with high gate capacitance, but they get extremely hot and burn out driving the IXFN44N50 FETs on the board. You will note above that there are small heatsinks atop the gate drivers. The solution I came up with was to use the same chips, but in a TO-220 case much more capable of dissipating heat than the usual DIPs. Since the heatsink would not fit on top, I moved the sockets to the bottom, fitted each driver to a DIP socket, and plugged them in. Later on I heard about Texas Instruments' UCC3732{1,2} chips that have mostly the same specifications but somehow manage to produce sharper edges with lower power dissipation.
Here is the gate waveform. The horizontal scale is 2us/div and the vertical scale is 10V/div. You will note that the switching is still rather slow, which probably contributed to the death of MOSFETs.
Here would be a picture of the streamers, but the pictures I thought I had do not appear to exist.


back to top




mark III mini twin



This coil is a small twin setup, with two secondaries out of phase so that they arc to each other. The secondaries are only seven inches high. So far the coil has only been run once, producing six-inch sparks between nails taped to the toploads. It currently sits disassembled and without its capacitor bank and primary.

back to top