Fullscreen HTML5Med Res (??MB)HTML4

How I made a 30mm desktop Spark-Gap Tesla Coil by Flavio Spedalieri My Solid-State “Flame Discharge” Tesla Coil project from the February 2022 issue (siliconchip.au/Article/15196) worked well but lacked the iconic metal toroid ‘top load’ that most people think of when they hear “Tesla Coil”. So I built an even larger device, that while still quite small, is more traditional! This device generates hazardous voltages! While we are not providing instructions on building or operating a Tesla Coil in this article, we advise caution if you build or operate a similar device. All parts of the Tesla Coil operate at lethal voltages and can deliver enough current to stop a heart or cause serious burns. You can also suffer RF burns if you come close to or contact the discharge terminal, even when no discharge is apparent. Always ensure that you are nowhere near the breakout point when the unit is powered up. Keep all parts of your body (or anyone else’s) clear of it until power has been switched off and the discharge stops. Remember that high voltages can still be present even when no discharge is visible. Electromagnetic interference warning This Tesla Coil is an RF generator. The input power is up to 180W and the spark gaps are broadband RF radiators. During operation, it can cause RF interference over a wide range of frequencies, especially the MF band, including the AM broadcast frequencies, MF amateur band and some mobile phone frequencies. Operation within a Faraday cage is advisable. 52 Silicon Chip Australia's electronics magazine siliconchip.com.au B uilding a full-size Tesla Coil is a significant undertaking. Therefore, in August 2020, when my motivation started to peak, I decided to make a small-scale but traditional Tesla Coil. It is traditional both because it is capped with a metal torus and it uses a spark gap-based oscillator. That oscillator drives the transformer that generates the extremely high voltages (tens of kilovolts) required for breakout. While my February 2022 article explained how to build your own very small Tesla Coil, this article is more of a story describing how I built a somewhat larger Coil. I won’t go back over the theory of Tesla Coils that I presented in the February 2022 issue, but I will quickly recap it to explain how this one differs from that earlier version. A Tesla Coil is a type of resonant transformer invented by Nikola Tesla (patented on the 25th April 1891). It transforms relatively low-voltage AC (a few hundred volts to a few kilovolts) to very high voltages (tens of kilovolts to megavolts) via two LC (inductor-­ capacitor) tuned resonant circuits that are loosely inductively coupled. The primary LC circuit comprises the ‘tank’ capacitor, primary coil (inductor), and a ‘switch’ to complete the circuit. The primary circuit can be switched by several methods; In a ‘classic’ Tesla Coil, a basic spark gap is used (see Fig.1). Other topologies use vacuum tubes while, in modern dual-resonant solid-­ state Tesla Coils (DRSSTCs), solid-state transistors (IGBTs or Mosfets) are employed. That latter configuration is what I used in my February 2022 project. The secondary LC circuit comprises Photo 1: the secondary coil has close to 38 turns per centimetre. the secondary coil (the large central tower that is iconic to a Tesla Coil), and the ‘top load’, which provides the capacitance and a place for the high-voltage breakout to occur. In this case, the capacitor begins to charge when power is applied to the primary circuit. Eventually, the voltage across the capacitor increases to the point that the air in the spark gap breaks down. The energy in the capacitor then discharges across that gap and through the primary coil. The energy then oscillates back and forth between the capacitor and the primary coil at a high frequency that is determined by the capacitor value and the primary coil’s inductance. A Tesla Coil’s ability to generate very high voltages and long arcs (streamers) is due to a process known as resonant voltage rise occurring in the secondary LC circuit. Tesla coils can be scaled up to produce many millions of volts. New design concept My initial idea was to build a small Tesla Coil using an arc lighter or neon sign transformer (6kV/30mA) power supply as the high-voltage source. I had some suitable components at hand, including 3nF 20kV AC rated capacitors, a ‘doorknob’ ceramic capacitor, plenty of 0.25mm diameter enamelled copper wire and a 107mm diameter, 27mm high aluminium toroid. Fig.1: a circuit showing one of the most basic arrangements of a Tesla coil. Fig.2: the output data from the JavaTC software that I used to help me design the Tesla Coil. siliconchip.com.au Australia's electronics magazine In the very early stages of the project, I considered a 50mm diameter secondary coil. However, it would have required so much wire that I would have had to special-order a large spool. My local Jaycar Electronics store had a small spool of 0.25mm diameter enamelled copper wire. Given the available length and the wide range of available diameters of PVC tubes, I spent several hours calculating a secondary form that would allow for around a 4:1 aspect ratio. The final design was a 33.5mm diameter, 132mm high coil of the Jaycar wire on a 30mm PVC former. Winding the secondary I cut a 160mm length of 30mm PVC tube to give approximately 10mm of clearance at each end. I then sanded the PVC and sealed the surface with two coats of UltiMeg 2000 electrical varnish (siliconchip.au/link/abha). I used a hand drill to hold the form and slowly guided the wire onto it, taking ~2.5 hours. During the next few days, I coated and sealed the secondary with several coats of clear varnish. I measured 38 turns per centimetre (Photo 1), very close to the 37.88 calculated (shown in Fig.2), giving close to 500 total turns. I sealed the ends of the secondary form using FR-4 (unclad PCB). The construction techniques I used for the SECONDARY COIL OUTPUT DATA Secondary resonant frequency Angle of secondary Length of winding Turns per unit Space between turns (e/e) Length of wire H/D aspect ratio DC resistance Weight of wire Effec. series inductance (Les) Equiv. energy inductance (Lee) Low frequency inductance (Ldc) Effec. shunt capacitance (Ces) Equiv. energy capacitance (Cee) Low frequency capacitance (Cdc) Topload effective capacitance Skin depth AC resistance Secondary Q 1618.28 90 13.2 37.88 0.00936 52.62 3.94 17.6699 0.024 1.848 1.931 1.879 5.233 5.009 8.014 3.997 0.0562 88.3825 213 kHz deg° cm cm mm m :1 W kg mH mH mH pF pF pF pF mm W February 2023  53 Tesla Coil Specifications Primary ] Capacitor: 3nF, 20kV AC ] Tap: 4.25 turns ] Tap frequency: 1527kHz ] Tap inductance: 2.23μH ] Total inductance: 13.5μH Secondary ] Turns: ~500 ] Resonant frequency without toroid: 2489kHz ] Resonant frequency with toroid: 1708kHz ] DC resistance: 19.2W ] Inductance: 1720μH Toroid ] Major diameter: 107mm ] Minor diameter: 27mm ] Calculated capacitance: 4.62pF ] Calculated breakout: 80.62kV Neon Transformer ] Power supply output: 6kV <at> 30mA (180W) ] Primary resistance: 11W ] Secondary resistance: 12.7kW (6.3kW to centre tap) ] Secondary impedance: 200kW secondary are the same as for much larger (high-performance) coils; no part of the winding can penetrate the former. I terminated the ground end of the winding via a copper tab, while the top of the coil is supported by a Nylon screw that I epoxied to the top plate before sealing the secondary. I finished the ends of the windings with black electrical tape and gave the secondary several more coats of clear varnish. I decided upon a coupling method that would allow the secondary to plug into the overall system, allowing a modular approach and safe storage of the secondary when not in use. The coupling also serves as the electrical ground connecting point at the base of the secondary, as shown in Photo 2. With the secondary complete, I characterised it using an oscilloscope and signal generator. Its DC resistance is 19.2W, while its resonant frequency is 1708kHz with the toroid and 2489kHz without it. At this point, I decided that it was going well enough that I would build a high-quality instrument where the aesthetic aspect was important. I also switched from the idea of using the arc lighter to a neon sign transformer (NST) that I had acquired, rated at 6kV and 30mA (180W). Additional calculations suggested that the resonant capacitor should be 15.9nF, with a larger-than-­ resonant (LTR) capacitor being 240nF. Still, I pressed on with the initially selected 3nF capacitor as I could always increase the capacitance later. Using a neon sign transformer meant that some additional components would be required: a protection filter (‘Terry Filter’), power factor correction (PFC) capacitor and an EMI line filter. Primary construction Photo 2: a coupling was added at the base of the secondary so that it could easily be removed when not in use. It also serves as a ground connection. 54 Silicon Chip The next phase of the project was the design and construction of the primary coil, supports and platforms to mount all the main coil components. Some of the crucial criteria when working with high voltages at high frequencies are sufficient clearances (to minimise arcing and insulation breakdown), selection of appropriate materials suitable for electrical work and fastening techniques. I selected “SwitchPanel Type X” as the main support material for the primary coil. It is a fibre-reinforced impregnated phenolic resin designed Australia's electronics magazine Photo 3: a conical primary (30°) was decided on, with an adjustable platform made from hardwood. for electrical insulation (siliconchip. au/link/abhb). I ordered three sections for the coil supports from Vale Plastics, all 180mm × 180mm, with one panel having a 50mm central hole to allow it to clear the coupling. The primary design took considerable analysis, considering the electrical parameters, size and shape (flat spiral, vertical or conical). Due to the way the primary coil’s electromagnet flux is presented to the secondary coil, I decided on a 30° conical primary (see Photo 3). The field of a conical primary coil is more uniform over the secondary coil’s aspect. At the widest point (outermost turn), the primary width is approximately the same as the secondary height (~140mm). I used a 2.14mm diameter copper capillary tube with an inter-turn (edge-to-edge) spacing of 5mm (7.14mm to the centres), giving a total of 10.5 turns for tuning flexibility. The mounting platform of the primary coil should be adjustable to allow for fine-tuning of the coupling to the secondary. I started building it by making the four support wedges, cut from hardwood. I attached them to the SwitchPanel Standard Soldering Ball Soldering Fig.4: the components for the filter were soldered using a technique called “ball soldering”. This technique helps to minimise corona losses at high voltages. siliconchip.com.au using Loctite two-part epoxy, which has good gap-filling characteristics. No metal screws or nails can be used, so all fixed components are glued or fastened using Nylon fixings. With the coil made and on the supports, I glued the final timber caps in place with more epoxy for better mechanical support and to improve the aesthetics. I also glued the central coupler into place. A copper strip, central brass screw, nut and acorn completed the grounding termination for the secondary. At this point, the primary and secondary were almost complete. Terry Filter and safety gaps Intending to use a neon sign transformer (NST) as the power supply, I made the secondary windings from very fine wire. Typically, enamel wire insulation is not very good at handling the fast, high-voltage transients generated in a Tesla Coil each time the spark gap fires, which can shorten the life of the NST. One method of protecting the transformer’s secondary windings is a lowpass RC filter network known as a Terry Filter (www.hvtesla.com/terry. html) – see Fig.5. I started building one by mounting the main capacitors and MOVs on FR-4 laminate board. I came up with the component layout, marked holes for drilling using a piece of ‘perfboard’ (prototyping PCB) and drilled the 1mm holes by hand. I soldered the components using a special technique called ‘ball soldering’, where the joints are made as smooth and spherical as possible to minimise corona losses at high voltages (see Fig.4). The safety gaps are made from three brass drawer knobs. I sanded each ball with fine wet & dry sandpaper to remove the clear lacquered coating, then drilled and tapped them with M4 threads. I repurposed three aluminium blocks as the supports. I drilled and tapped each block with an M4 thread to mount them onto the FR-4 substrate. The position of the two left/right balls can be adjusted for correct operation of the safety gap (ie, so the air gap will break down at an appropriate voltage). You can see how this arrangement is mounted on the Terry Filter assembly in Photo 4. I made the high-voltage cables that connect to the coil itself using 7.5mm2 siliconchip.com.au Fig.5: the protection filter circuit (Terry Filter) for the Tesla Coil. Photo 4: the safety gap uses three brass knobs mounted to the Terry Filter circuitry (components placed but not yet wired up). The knobs were mounted onto aluminium blocks and set up so that their positions can be adjusted. Australia's electronics magazine February 2023  55 t0 Photo 5: after mounting the Terry Filter onto a plate of SwitchPanel Type X, a test was performed to verify the operation of the safety gaps. OFC stranded power cable with two layers of PTFE tape applied. Heatshrink tubing was added, followed by an additional layer of PTFE tape and a final layer of heatshrink tubing. Readily-available white PTFE (plumbing tape) has a high dielectric strength of around 60-70kV/mm (see siliconchip.au/link/abhc). Common high-density PTFE plumbing tape has a density of about 0.3g/cm3 and a nominal thickness of 0.1mm (siliconchip. au/link/abhd), so it has a dielectric strength of about 6-7kV. I mounted the Terry Filter and safety gaps onto a 150 × 250 × 12mm plate of SwitchPanel Type X. I made the electrical connections to the filter with brass hardware and acorn nuts, while the connection to the Tesla Coil is via the two aluminium blocks on the left and right sides. Photo 5 shows my tests to verify the operation of and adjust the safety spark gaps. Main spark gap & tank circuit With the Terry filter finalised, I moved on to the main spark gap and the layout of the tank circuit components. The performance of a Tesla Coil is determined by the performance of the spark gap, which acts as a momentary switch that completes the circuit between the capacitor and the primary coil. The capacitor’s energy is discharged into the primary coil when the gap conducts. A spark gap is a simple device, but the dynamics of its operation are complex. The distance between the electrodes 56 Silicon Chip t1 t2 Fig.6: the times labelled t1 & t3 are the first and second primary notches – the times when the current in either the primary has fallen to zero and the spark gap can be quenched. t3 t0: gap fires. t0 > t1: primary energy transfers to the secondary. t1: all energy is now stored in the secondary (1st primary notch). t1 > t2: remaining energy in the secondary transfers to the primary. t2: all energy is now stored in the primary (1st secondary notch). t2 > t3: remaining energy in the primary transfers to the secondary. t3: all energy is now stored in the secondary (2nd primary notch). This process repeats until the gap stops conducting (quenches). Once quench occurs, an exponential ringdown will occcur. sets the breakdown voltage of the spark gap. With a static gap, the width would be set at the power supply line voltage (6kV in this case) and would be at the correct setting with the gap firing at the full applied voltage of the transformer. This project utilises a static gap arrangement; however, much larger coils employ rotary spark gaps, giving better control and performance. For a coil of this size, that would be slight overkill. The main spark gap consists of the electrode holders and the actual electrodes. For this project, I have employed 6.35mm diameter parallel-­ faced tungsten rods. Tungsten is a favoured material for spark gaps due to its high melting point and, therefore, resistance to burning and pitting. When the gap fires, the arc ionises and heats the air within it, making it highly conductive. Once the gap conducts, it will continue to conduct even when the capacitor’s voltage has dropped below the initial breakdown voltage. This can allow the energy from the secondary to return to the primary; this energy will be lost as heat, sound and light, reducing the coil’s performance. Extinguishing a conducting spark gap is known as ‘quenching’ and is essential to maximise the energy retained in the secondary. Quenching is the action of the spark gap going open-circuit and ceasing conduction, and can only occur when the current through a conducting gap falls to a certain point. Then, the arc may no longer be sustained, and the Australia's electronics magazine air within the gap cools enough to prevent arc-over as the voltage begins to rise again on the next cycle. The total energy transfer time is the number of half-cycles it takes at the resonant frequency to transfer all the energy from the primary circuit to the secondary (not including losses). Ideally, we would like to trap all the energy within the secondary, as any energy that returns to the primary will contribute to inefficiency and, thus, less energy for the output arcs. The only way to trap the maximum energy within the secondary is to stop the gap conducting as soon as the current in the primary circuit reaches zero. Known as the ‘first (primary) notch’, this period is very short, and the amount of energy still within the 1st secondary notch Secondary Envelope 2nd secondary notch Exponential ringdown 3rd primary notch 2nd primary notch 1st primary notch Fig.7: this waveform shows the third notch quenching, meaning the third primary notch is where the gap stopped conducting, followed by the secondary ringdown. siliconchip.com.au Photo 6: a centrifugal blower fan was used on the main spark gap instead of an axial fan, as it provides superior airflow. Photo 7: the spark gap assembly is composed of two electrode holders mounted on a FR-4 substrate. primary is sufficiently high to keep the gap conducting (see Figs.6 & 7). If the gap continues to conduct, the next available opportunity to open the spark gap is at the next point that the current returns to zero (the second primary notch) and so on. Early quenching of the spark gap may be achieved through various methods, including magnetic quenching (siliconchip.au/link/abhe), air blast (siliconchip.au/link/abhf), vacuum (siliconchip.au/link/abhg) or with a rotary spark gap. Using forced air, a vacuum or a rotary gap allows the gap to cool by removing hot, ionised air from it, reducing the chance of the gap re-­ arcing. I decided to use a centrifugal blower fan (drawing 12V <at> 860mA), as such fans generate high-pressure air flows compared to an axial fan (see Photo 6). I was going to use PWM fan speed control but, in testing, it offered little effective control; therefore, I abandoned that idea. Instead, the fan just runs at full speed during operation. I used a copper bus bar to form the support for the capacitor, with a short copper tube to connect it to one side of the gap. One last detail for the coil is the strike rail, which protects the primary coil and primary circuit components from arc strikes, made from a 2.3mm capillary copper tube (Photo 8). The ground rail must present a low impedance path to RF ground, so I made a clip from a copper saddle that is snug fit onto the strike rail. I then added a grounding post to terminate the secondary ground and the strike rail (Photo 9). The strike rail mustn’t form a closed loop, as would otherwise present as a shorted turn. The strike rail supports are made from 9×9×46mm timber sections with a 2.5mm hole drilled through each support. I sanded and stained these before gluing them into place with Loctite epoxy. I then slid the copper tube into place and used heatshrink tubing to cover the open section. Primary tap point Constructing the primary tap connection was a challenge. Early in the project, I drilled four clearance holes with the idea of bringing the tap wire up through the bases. However, this made it difficult to disassemble and reassemble. So instead, I brought up the tap wire from the side, using a clip made from a modified M205 fuse clip, reduced to create a snug fit. I used a length of copper braid to strengthen the clip and provide a better connection. At this point, I had completed much of the Tesla Coil, but was still waiting Main gap & strike rail The main gap is the critical part of the spark gap oscillator. I cut a phenolic resin plinth as the mounting base for both the spark gap assembly and the connections to the capacitor. The gap itself was formed by mounting electrode holders onto two phenolic support blocks, which I then affixed to a strip of FR-4 substrate. I then attached the whole assembly to the phenolic base (see Photo 7). siliconchip.com.au Photo 8: the outermost copper tubing is the strike rail, which was added to protect the primary from arc strikes. Photo 9: after building this, a clip was added to the strike rail to ground it (shown in the photo at right). Australia's electronics magazine February 2023  57 for the high-voltage bleed resistors (10MW, 10kV) for the main capacitor. I searched Digi-Key’s website and found they stock 100MW 10kV 2.5W Maxi-Mox resistors from Ohmite (MOX-1-121006FE). As well as being available, they had the advantage that a 10MW bleeder resistor would have dissipated 7.2W. Increasing to 100MW reduced that below 1W while still discharging the capacitor to a safe level (50V) in 1.5 seconds. Radio frequency (RF) Earthing One of the more overlooked and important areas with any RF system is the provisioning of a suitable low-­ impedance Earth system. Tesla Coils generate heavy RF currents which must be appropriately distributed to Earth. A sound Earthing system is key to a well-performing Coil as the Earth forms the return path for the secondary side of the LC circuit. So I sunk a 19mm diameter, 2.4m-long Earth rod to 1.8m depth, plus a second ‘domestic’ size rod to 1.2m, bonded them together and connected them to the coil via 25mm2 welding cable. Measurements With the Coil essentially complete, I made some measurements to determine the tuning parameters and confirm the resonant frequencies against my calculations. I measured the resonant waveform period as approximately 124μs, corresponding to the total energy transfer time; the first notch came after approximately 8.2μs. Power supply I mounted the neon sign transformer to a 12mm-thick 200 × 300mm base made from SwitchPanel. I added two timber stand-offs to mount the Terry Filter module (Photo 10). I then added the control box, which includes a TE Connectivity 3A EMC-series EFI/RFI line filter to prevent interference from feeding back into the mains. The control box also contains a small mains switchmode power supply (SMPS) to provide 12V <at> 1.2A for the quenching fan. The control box also includes a mains switch, a switch for the fan and a switch to supply power to the transformer, lamps to confirm active power to the circuits and a 2A thermal magnetic circuit breaker (Photo 11). This control box is used with a variac to provide fully adjustable control of the Coil. First tests Following nearly three months of development, I fired it up for a momentary test. I noted a flashover from the end of the primary winding to the strike rail, occurring several times at the same location, causing a tracking burn. I realised this was due to the end of the primary not being smoothed off and sealed. Another overlooked area was that I hadn’t sealed the supports with varnish such as Ultimeg. I repaired the area, smoothed the copper end and applied epoxy resin to seal it. After cleaning up the tracking burns, I also sealed the support. I added more epoxy to all key areas at the primary junction and supports and applied several coats of Ultimeg electrical varnish to the timber supports. I left the primary coil assembly to cure for several days. In hindsight, considering the pulsed nature of the high-voltage present on the primary coil, timber is not the best material to use. A more suitable material would be a phenolic resin; however, it is expensive in small quantities and with suitable dimensions. SwitchPanel Type-X could be used to create the smaller parts, but it would need to be cut from a larger sheet, which would take a great deal of time. With those repairs and improvements completed, I returned to testing, closing down the spark gap to around 3-4mm for the test. The low-power test was successful, with a nice breakout occurring (Photo 12). I then opened the gap to about 5mm and made another run (Photo 13). It was successful, but I noted some random flashover between the final turn of the primary to the strike rail. I obviously needed to improve the insulation between the strike rail and the final turn of the primary coil. I did that by adding more layers of Ultimeg varnish, as well as adding short lengths of clear vinyl tube around crucial points on the strike rail and final primary turn. I let the varnish cure over a week before getting back to testing. I then ran a full power test, applying the full mains voltage to the NST. In doing so, I tweaked the parallel alignment of the spark gap electrodes. The full power test was very successful, with many streamers forming but no arcing at the support points. Photo 10: the neon transformer was mounted to a 12mm thick base made from SwitchPanel. Timber stand-offs were then used to attach the finished Terry Filter above it. Photo 11: the control box for the Tesla Coil. It contains a mains switch, a switch for the fan, another for the transformer, lamps to confirm activity, and more. 58 Silicon Chip Australia's electronics magazine siliconchip.com.au Top-load upgrade I decided to order a larger toroid from a company based in the USA. It took close to two months to arrive. Their customer service left a lot to be desired, so I don’t want to mention the company’s name. The larger toroid has a major diameter of 152mm, a minor diameter of 38.64mm, a calculated capacitance of 6.61pF and a calculated breakout voltage of 114.74kV. The additional capacitance of the larger toroid required re-tapping the primary coil to bring the primary into tune with the lower resonant frequency of the secondary. A further resonant test using an oscilloscope and signal generator on the secondary coil confirmed the new resonant frequency as 1360kHz. I moved the primary tap by one turn to account for the extra load on the secondary, bringing the system back in resonance. The lead photo shows the result of a full-power test with the larger toroid in place. It generated streamers long enough to reach the strike rail; they are equivalent in length to the secondary coil. I conducted another experiment by simply placing the two toroids on the coil, resulting in longer streamers (Photo 14). Re-tuning the coil was not necessary. Photos 12 & 13: on the left is the initial low-power test with a spark gap of 3-4mm, while on the right was another test run with the spark gap at 5mm. Conclusion It was a lot of work, but I am delighted with how this small Tesla Coil turned out. It was interesting to go through the tuning process that Nikola Tesla and other pioneers would have had to figure out. I also learned that it pays to give special attention to insulating everything when working with such high voltages. Tesla was a genius to have come up with such an elegant way of generating extremely high voltages using the very limited technology available at the time. While building a Tesla coil is not for everyone, they are impressive devices and a must-have in any mad scientist’s laboratory! In memory of My Mum (Zina Spedalieri) was amazed when she saw the original article come to print. Sadly, we lost Mum on 2nd of June 2022, It would have been something for her to see the second article come to print. SC siliconchip.com.au Photo 14: placing the newly bought larger toroid on top of the old toroid resulted in larger breakouts. As I was happy with the result, I eventually had the toroids welded together, then cleaned and sanded them to maximise their appearance and performance. Australia's electronics magazine February 2023  59