You are here : Main projects | The Hyperion Tesla Coil | 2. Construction

The Hyperion Tesla Coil
August 2012

2.  Construction

This page will discuss the design and the construction of Hyperion, and how some components have been improved relatively to my first coil in particular. The general theory of operation, as well as the formulae regarding the components design are covered on Zeus' pages and won't be copied here.

2.1   Transformer

[Back to top]

Actually, the main reason I did this coil was because I had a spare 6kV 25mA NST and didn't quite knew what else I should do with it. With a power of 150W, the basic arc length equation gives us an upper bound of 52cm. Its impedance Z = V/I is equal to 240 kΩ.


Figure 2.1 — The NST partly enclosed in its wooden box, during the early days of the construction, along with the entire chassis.

The resonant-sized capacity of the primary capacitor is given by Cres = (2πZf)-1, where Z is the transformer impedance we just computed and f the mains frequency, which is 50Hz in Europe. We therefore get:

Cres = 13.26 nF

The ideal larger-than-resonance capacity for a static gap is given by φ times the resonant size, where φ ≈ 1.618 is the golden ratio :

CLTR = 21.46 nF

2.2   Primary coil

[Back to top]

On low-power coils like this one, the risk of arcing between the top load and the primary is minimal and one can safely use a conical spiral or a helix-shaped primary. These designs offer a better secondary-primary magnetic coupling, which allows the energy to be transferred in a smaller number of oscillations, which, in turn, decreases energy losses. I used the conical design for Hyperion. The distance between the breaking point and the upper and of the spiral is 50cm, and hopefully, no arc connecting the two has been formed so far (although sometimes, the arc clearly points towards it). Remember this can destroy the Tesla coil, as an extremely high voltage is sent to a primary circuit not designed to handle it.

Primary coil

Figure 2.2 — Hyperion's primary coil has a conical spiral design. Notice the tapping point on the far side.

I used the same construction technique as for Zeus : first deforming the copper tubing into a spiral well enough so no strong fixing force is required later, then fix it on supports with pre-established dimensions. Also like Zeus, I used four supports, placed at each quarter of turn, and the fixing was done using cable ties. Note that the deforming step produced some disgraceful kinks on the inner turns, which is why I placed those wood panels near the centre.

Primary coil Primary coil

Figure 2.3 — Left: The supports of the primary just after cutting (notice how the tubing emplacements are slightly offset from one support to another). Right: close-up when placed.

2.3   Primary capacitors

[Back to top]

Not much to say about them, as their design is exactly the same as Zeus' ones. I used 150nF Cornell-Dubilier 942C caps, as they have proven to be excellent capacitors. Eight of these have been connected in series (each one with its 10 MΩ, 1/2 W bleeder resistor), resulting in a total capacity of 18.75 nF. This design has all the requirements for Tesla coil operation : the total capacitor has a sufficient voltage tolerance and dV/dt for the NST it's connected to, and also possesses a security gap (although I dropped the gap series resistor). The security gap is connected in parallel with the whole capacitor stack, as shown on fig. 2.4 (right) ; its role is to protect the capacitors from overly dangerous voltages by shorting them (the capacitors), i.e. firing.

Primary capacitors Primary capacitors

Figure 2.4 — Left: Close-up on the main capacitor ; notice how the bleeder resistor and the cap unit are on either side of the support. Right: The entire capacitor, with its security gap.

2.4   Spark gap

[Back to top]

This new spark gap is probably the most significant improvement on Zeus' one, along with the top load. Both the gap design and the quenching system have been ameliorated.

Spark gap Spark gap

Figure 2.5 — Hyperion's 2-element adjustable spark gap.

2.4.1   The gap itself

Zeus' gap consisted of five electrodes, because dividing a big arc into smaller ones helps quenching. But as Hyperion has a lower power than Zeus, I decided to resort to only a 2-elements gap. The two electrodes were made out 5cm-long rods of copper tubing which were pinched at their extremities and then folded backwards using a vice. Holes were drilled on these folded section for the screws that would attach them to the PVC structure. The use of copper was more by constrain than by choice. Copper indeed corrodes easily and the electrodes must be cleaned regularly.

Spike effect prevention

Little reminder: the spike effect states that the electrical field on the surface of a conductor is more intense, for a given voltage, on zones of small curvature radius (i.e. spiky). This can be proven using Maxwell's equations (Gauss' equation in particular). This explains why lightning preferentially strikes protruding objects. On rod-shaped electrodes, the electric field near the edges of the rods will be much more intense than it is near their body, which causes the sparks to be much more resilient and harder to break when they reach the edge of the electrodes. This, in turn, decreases the spark gap efficiency, as explained here.

These cushion-shaped electrodes provide a very smoothly-varying gap width for the arc to form and disappear. Indeed, the spark is likely to begin its short life on the middle of the gap, where the distance is minimal ; then, as the hot ionised air channel will naturally travel upwards into the gap, it will smoothly die out when the inter-electrode distance has become too large. This design prevents the spike effect encountered on simple, rod-shaped electrodes (like Zeus' one) and improves quenching.

Spark gap

Figure 2.6 — The progressive formation and disappearance of the arc can be seen on the corrosion pattern, which gradually expands and retracts.


The width of the spark gap can be adjusted for optimal Tesla coil performance. While the right half is fixed, the left one is mounted on springs and mobile. By turning the bolt on the central thread, one can advance or retract the left part, and therefore vary the gap width. Notice there are little epoxy "bulbs" on the end of the three support threads, to avoid arcing between them and the electrode. As for the springs, they simply come from pens.

2.4.2   The quenching system

Hyperion's fan and has been salvaged from an hair drier. It is powered by two rechargeable 9V batteries connected in series (i.e. 18V), and it takes no more than a few minutes to empty them (hence the rechargeable). The duct, which helps focus the flow on the electrodes, comes from the same hair drier. While Zeus' quenching system, consisting of a simple computer case fan, was already quite powerful (as overpowered with 18V too), Hyperion's one blows a lot much more air and concentrates it on the electrodes, and therefore does a much better job at cooling the electrodes and quenching the arcs.

Spark gap Spark gap

Figure 2.7 — The salvaged fan (left) and the entire spark gap zone (right).
The fan itself is enclosed in that wooden square structure. A grid has been added on the back for safety and aesthetics. The duct is the protruding golden part.

2.5   Secondary coil

[Back to top]

The secondary coil is made of ~1000 turns of SWG 27 (0.4 mm) magnet wire, wound around a PVC pipe with a diameter of 9cm. The total height is 40cm. Once the coil was wound, I applied insulating varnish (in spray form) on the entire coil for increased protection. It has an h/D ratio of 4.4, which is just fine. Its effective inductance was calculated (by JAVATC) to be 17.19 mH and its Q factor, 246.

2.6   Top load

[Back to top]

The top load has a toroidal geometry, with a minor diameter of 10cm and a major diameter of 30cm. The main difference with Zeus' toroid is that this one has a much smoother surface. As explained here, the state of the toroid surface is of prime importance for arc production : having a rough surface will typically produce numerous, shorter sparks, while a smooth one will produce fewer but longer sparks. Both can be desirable, but I wanted Hyperion to perform better in raw arc length.

Top load Top load

Figure 2.8 — Photograph and schematic of the toroid.

The main body of the toroid was made out of aluminium duct. A section of adequate length was cut and then folded into a circle ; the junction was made using aluminium tape. Afterwards, many (many) layers of filler were applier to the toroid to fill up the folds of the duct, in order to obtain a surface as smooth and regular as possible. The inner sections of the toroid (not the inside) were left untreated as no sparks arise there.

Once the filler was dry, the last step was to apply bands of aluminium tape on the toroid. These were electrically connected to is as they touched the inner, untreated zone. These band would form the final, arc-producing surface. Although it is almost impossible to avoid creases in the process of applying the bands, the resulting surface is much smoother than before. A good construction guide for this can be found on Philip Tuck's website.

I also added a breaker, which consists of a screw thread that goes through the section of the toroid, to force arcs to form almost exclusively at this unique point. Finally, there are two wooden supports disks for toroid fixation, with the lower one covered with aluminium tape for electrical conductivity. A large screw thread joins the whole assembly to a PVC base that "clips" on the secondary coil structure (see black structure on fig. 2.9). The current from the secondary coil arrives into the toroid via this thread with the secondary wire just rolled around it for easy removing (see fig. 2.9 too).

Top load

Figure 2.9 — Connections of the secondary coil to the top load.

2.7   RF ground

[Back to top]

The RF ground consists of a metallic grid fixed below the main frame, on the lower side of the chassis plate, which makes it much more compact than Zeus' external plate. I had doubts about whether or not this configuration would be sufficient, but Hyperion's great performances seem to indicate it is indeed the case. It is composed of scrapped parts of aluminium grid sheet connected together with small aluminium tape bridges. The grid parts are fixed on the chassis plate with simple screws.

Four cylindrical supports prevent the grid from actually touching the ground. These were made of wood with a extra disk of rubber for additional insulation and stability, and were fixed to the main chassis plate with liberal amounts epoxy glue, which went through the holes of the grid.

RF ground

Figure 2.10 — View from below : the RF ground as well as the supports. The secondary coil - RF ground connection can be seen on the right.

2.8   Protection circuits

[Back to top]

This section will discuss "optional" components, i.e. components that are unnecessary for Tesla coil operation but nevertheless strongly recommended for safety and long coil life. Apart from the main capacitors security gap discussed earlier, Hyperion has a filter circuit protecting the NST as well as an EMI suppression on the power chord.

2.8.1   NST protection filter

This circuit is a low-pass filter with a 3-element security gap. It's role is naturally to protect the NST from the high-frequency noise coming from the Tesla coil circuits and from voltages spikes (more information on how it works can be found on Zeus' page). The design is very similar to Zeus' filter, with the only major difference being the lack of varistors (no problems encountered so far). The spark gap design has also changed a bit, with the centre electrode being maintained with neodymium magnet.

Terry filter Terry filter

Figure 2.11 — The NST protection filter. Notice the corrosion on the spark gap electrodes (far end).

2.8.2   EMI suppressor

This components offers further protection in blocking the harmful high-frequency interference from damaging other electronic devices connected to the mains. More on Zeus' page.

Line filter

Figure 2.12 — The EMI suppressor, mounted upside-down.


All dimensions and characteristics of the Hyperion Tesla coil are regrouped in the JAVATC output file, which can be downloaded from the link on the right.