Two-inch sparks using a television flyback

Following the guidance of one of my favorite Instructables tutorials, this weekend I used a broken CFL bulb and the transformer from an old television to generate a reliable stream of two-inch sparks:

The concept here is pretty similar to that described in my ignition coil relay project: The circuit board at the base of a CFL bulb serves to transform mains power to a high-frequency, high voltage signal that serves as the “spark” that illuminates the argon/mercury vapor inside the fluorescent envelope. If this signal is instead routed to the terminals of the primary circuit of a large step-up transformer, a high-frequency and even high-voltage can be attained (at the expense of a lower current). In this case, the transformer is salvaged from an old CRT-type television screen or monitor, in which a high-voltage beam of electrons is selectively fired at a phosphor coating on the screen in order to create the image.


Remnants of my Tesla coil

My first year of high school I tried to build a functioning, high frequency Tesla coil entirely from scrap parts. This project is almost a cliche nowadays; thousands of dedicated hardware hackers have successfully created ominous and occasionally dangerous coils, and so-called “singing” Tesla coils are the new trend among hobbyists. But the project was one of my first earnest attempts to learn about something on my own and apply that knowledge to a non-scholastic project, and so I wanted to link to a few resources here that I found invaluable when I was first starting out:


The Powerlabs Tesla coil page. This is the most “professional” Tesla coil I have found that was built by a hobbyist. The craftsmanship is impeccable, from the precision of the secondary coil winding to the care with which the capacitor bank was assembled. The care is reflected in the results; I am very confident that this is one of the most efficient Tesla coils I’ve come across, as it appears to regularly generate 18-inch streamers despite its compact size

The trashy Tesla coil. I like this project because the author defiantly avoids using any store-bought components or parts, using piping and wiring entirely scavenged from his local rubbish yard. This site is also home to one of my favorite anecdotes from a hobbyist:

For some funky reason every time I switched on the power, the sprinkler system in the yard turned on. I’m not kidding here. The yard gets watered every time I fire it up.

Primary and Secondary Coil

The red, long coil in image at the top of this post is the secondary coil from my own Tesla coil, which took me about a week of winding 28 gauge enamel-coated wire over oven-baked PVC pipe. That the toroid is a doorknob is a good tip-off that the payload isn’t resonantly coupled. The pancake-spiral in the foreground is a remnant of my original primary coil design, which I based on tutorial found on this page.


I first realized how attainable a homemade Tesla coil would be when I saw just how simple it can be to make high-voltage capacitors at home in the form of Leyden jars, which can be made from a film canister or bottle and some aluminum foil. Using a CD jewel case and some foil, I’ve even made capacitors that can be charged from a CRT screen but which will produce 3-inch sparks upon discharge—although predicting the discharge rate and stability of Leyden jars against dielectric breakdown is almost an art when one is using plastics and glass with aluminum foil. The best page for an intro to Leyden jars and their uses can be found here.

Primary transformer

Most Tesla coils use a step-up transformer even before the current reaches the primary circuit. This allows shielding of the electrical mains from sparks and shorts in the primary circuit, and it also allows one to get by using capacitors made from beer bottles, air gap discharges, etc. because a higher voltage primary circuit requires less finicky specifications (it would also be very difficult to use a spark gap to modulate the frequency if one was only using mains voltage). I originally ran my coil off of car batteries by using an electromagnetic buzzer and a pair of ignition coils in my primary circuit; however, if I were rebuilding it today I would instead use a neon sign transformer, which I believe offers much more reliable and safe performance despite running on mains power. Here’s a buying guide for NSTs for Tesla coils.

Spark Gap

When I was in high school, I always found the spark gap to be the most mysterious component in the Tesla coil primary circuit. After all, the primary circuit is already an AC circuit, and it seems like forcing the current to regularly jump an air gap would induce significant power losses that would reduce the efficiency of the transformer. The latter point is correct, but it turns out that the spark gap is still worthwhile because the timescale of the AC cycles coming out of the HV transformer being used to drive the primary circuit is way too fast to effectively switch most Tesla coil designs, given the dimensions and couplings of the primary and secondary coils. The spark gap allows the capacitors to fully charge and discharge at a rate set by their time constants and the properties of the spark gap itself (since things like pointed electrodes can create corona discharge, reducing the effective dielectric constant of the air in the gap). As a result, the spark gap acts like a high-power switch at a low enough frequency to allow effective transfer of energy between the primary and secondary coils. A good description of the idea behind using a spark gap (instead of a high-power relay and integrated circuit or other solid-state switch) can be found here and here.

Building a high power voltage multiplier


A simple high-voltage circuitry project is the Cockroft-Walton voltage multiplier. I first created this as a demonstration for a class in high school, but I’ve altered it over the years in order to improve its performance. The nice thing about this project is that it can be created entirely using cheap, store-bought components—diodes and capacitors—and so it is thus relatively easy to ensure that it will work and perform at the voltage estimated. I got my components from West Florida Components, and the total cost of everything was under $10. A good tutorial that gives possible specifications for the components can be found on Instructables.

The total voltage drop across many capacitors in series is equal to the sum of the voltage drop across each component—this is a consequence of Kirchoff’s circuit laws, and can be mentally visualized as a charge on one end of a capacitor displacing and equal but opposite charge on the opposite plate, which in turn displaces an opposite charge on any other capacitor plates to which it connects, and so on. The Cockroft-Walton multiplier can be visualized as a fancy way of putting a bunch of capacitors in series and charging them so that they each have a voltage drop of 120V, resulting in a total discharge voltage of 120 V times the number of capacitors. This output is roughly DC, and it has a much lower current than the device draws from the mains, hence preserving energy conservation since power = (voltage)*(current). A simple diagram of the half-wave CW multiplier looks like this:

A circuit schematic for a half-wave Cockroft-Walton voltage multiplier.

A circuit schematic for a half-wave Cockroft-Walton voltage multiplier.

The manner by which the CW multiplier can charge each capacitor separately to 120 V is essentially by charging them in parallel and discharging them in series. The concept borrows from the design of a basic half-wave rectifier, which uses a diode and smoothing capacitor to convert the positive portions of AC sine waves to a smooth-ish DC current. The idea is that the first stage in the circuit (capacitor 1 and diode 1) converts the AC to an approximately constant DC signal, which then gets fed forward through diode 2 to the right plate of capacitor 2. During the first positive cycle, that capacitor charges to +120V. During the “off” cycle (the negative portion of the AC sine wave gets blocked by the first diode), the second capacitor discharges through diode 3 into capacitor 3 because, during the off cycle, there’s now -120V on the bottom plate of that capacitor, leading to a potential difference that allows charging. During the next “on” cycle, the current ignores capacitor 3 because it is fully charged (and so it essentially acts like a break in the circuit there), and so now capacitor 4 gets charged instead. During the next off cycle, capacitor 4 discharges through diode 4 to charge capacitor 5, and the cycle repeats itself until, after (number of capacitors)x(charging time) all the capacitors are charged.

There are several equivalent ways of visualizing what is going on in the CW circuit, but the key things to remember are that the capacitors store the charge (differential) and the diodes force the AC to feed forward and charge each capacitor in sequence. The charging time can be adjusted by adjusting the time constants for each capacitor in the circuit relative to the AC cycle frequency (60 hz in the US).

The device would build up charge twice as quickly if one instead uses a full-wave design (which is analogous to a full-wave bridge rectifier), because it would then take advantage of the negative swing of the AC sine wave, which gets lopped off by the first diode in the half-wave version.

I the video above, I have added a switch and fuse for safety reasons (visible in the upper-left hand portion of the screen; I used a plastic lid as a base for the two components). In the first cut, the ~1 mm spark regularly produced by the device is visible. This spark can be used to drive continuously an 8 inch fluorescent tube (shown in the second section), but, curiously, the frequency of the pulses through the fluorescence in the tube depends on the proximity of other conducting objects—in the fourth clip, it is apparent that touching pliers to the glass reduces the frequency of the pulses, rather than increasing them as I would have expected. My best guess for the cause of this effect is charge build-up on the glass interior beneath the metal, leading to low-frequency discharge for the same reason that high-voltage capacitors decrease the sparking frequency in the primary circuit of a spark-gap Tesla coil. The last clip shows the device discharging through a 1 inch xenon flash tube salvaged from a disposable camera. The firing frequency is low due to the relatively large distance that the spark has to cover, despite the low dielectric constant of xenon gas. In other tests, I’ve noticed that large spark gaps that require charge build-up over periods longer than the ~3-5 s for the flash tube will generally result in short circuits occurring upstream in the capacitors in the CW, which probably cause damage to the solder joints and possible the capacitor ceramic due to dielectric breakdown.

Cockroft-Walton generators have a special significance in the history of physics because they were used to generate steering currents in one of the earliest particle accelerators, enabling their creators to win the first Nobel Prize in Physics ever given to a collider project. For this reason, one of the first large-scale CW multipliers (manufactured by Philips Co.) is prominently displayed in the National Science Museum in London:


A Cockroft-Walton generator built in 1937 by Philips of Eindhoven. National Science Museum, London, England.

A Cockroft-Walton generator built in 1937 by Philips of Eindhoven. National Science Museum, London, England. Image from Wikimedia Commons.



Two dipoles radiating out of phase

I thought I’d write about one of my favorite problems from freshman year. It doesn’t require any math to understand, but it points out many of the risks and subtleties that can arise when physics problems make too many “ideal” assumptions:

Suppose that you have two simple antennae, each consisting of a single, straight length of copper wire through which a single frequency of alternating current is passing. The two antennae are positioned some fixed distance apart, and they are oriented in parallel. If a remote physicist operating the two antennae introduces an appropriate delay between their driving signals, causing the AC waveforms in the two antennae to be 90 degrees out-of-phase (but still at the same frequency), then the electric field in the region between the two antennae will vanish due to destructive interference. Yet the two antennae are still emitting radiation; they are still each drawing current, and presumably the power they consume to create this current must be transferred into the resulting fields they emit. So when the two antenna destructively interfere, where does the energy go?

The conventional response to this question (and the one my freshman lab TA insisted upon) is that the field cancels out in some regions—such as between two two antennae–but it increases by a compensatory amount in other regions where the waves constructively interfere, resulting in the net energy stored in the fields (throughout all of space) remaining constant. While this is certainly a satisfactory answer for most textbook treatments of dipole radiation, it remains troubling because one can easily envision a case in which there are no other regions in which the waves can constructively interfere—for example, if mirrors were used carefully. If, instead of antennae, one pictures two out-of-phase lasers pointed towards each other, then it becomes much less clear where the compensating region between the two lasers would be. However, there’s another way of looking at the problem that sheds light on this inconsistency:

Conventional electrodynamics tells us why the two copper antennae will generate radiation: the moving charges in each antenna beget changing magnetic fields, which in turn create electric fields via Faraday’s law, which then create new magnetic fields as they collapse. This cycle of electric and magnetic fields taking turns forming and collapsing gives rise to self-propagating electromagnetic waves—a collapsing electric field changes quickly, thus inducing a magnetic field which eventually collapses to produce a new electric field, and so on. The power transmitted by the wave is thus determined by the amplitude of the initial magnetic field generated by the antenna, which in turn is proportional to the current through the wire. This current is, in turn, determined by the resistance of the wire comprising the antenna—if the wire were a impossibly perfect conductor, then even the most minor voltage difference between the two ends of the antenna would generate an impossibly infinite current via Ohm’s law. Thus the power put into a single antenna is determined by the resistance of its wire, and this power exits the antenna as electromagnetic radiation—so far, energy has neither been created nor destroyed.

The subtlety of the problem arises because an additional effect that occurs when there are two antennae near each other. The electrons moving back and forth inside one antenna aren’t just limited in motion due to the resistance of the copper wire, but also by the electric field due to the other antenna. If the other antenna is in-phase (no delay), then the electrons will keep experiencing a Lorentz force in the opposite direction to the way that the antenna’s power source wants them to move, and so the power source will need to provide more power in order to generate waves of the same amplitude—the current, and therefore power, drawn by the antenna increases. In the case when the sources are out of phase, or the waves are destructively interfering, the opposite effect occurs: the field from the other antenna actually helps the electrons along, allowing a given electron to oscillate at a certain amplitude without requiring as much energy from the power source. In other words, placing the antennae out of phase reduces the effective resistance, or impedance, of the two antennae, and thus reduces their power consumption by an amount equivalent to the drop in the energy of the electromagnetic field due to their destructive interference.

In the laser formulation of this problem, this explanation would amount to the light from one laser damping excitations in the lasing medium of the other laser, resulting in less power drawn from the source.

What I like about this scenario is the manner in which a very common assumption used in physics problems—that power supplies are monoliths, steadily providing a fixed voltage and current to each component of a system–turns out to be the source of the ambiguity. An electrical engineer who places an ammenter in series with one of the antennas would immediately notice the drop in input power when the antennae are placed out-of-phase. But in the way that the problem is often presented, the power consumption of the antennae seems like a fixed quantity, giving rise to the supposed paradox.

Phase Transitions and Ferrofluid

When I was younger, I came across a tutorial that described a simple way to make thermite entirely from homemade ingredients. The crux of the instructions was that iron oxide, the key ingredient in thermite from which molten iron is created, can be isolated from common sand simply by repeatedly dragging a magnet through a container full of it. At the time, my family happened to live near a beach, and so I resolved to gather as much iron oxide as possible and test out the recipe.

In order to collect the iron oxide, I would drag a bag full of magnets behind me every time I went to the beach. After about two weeks of regular collection missions, I obtained enough iron oxide in the form of magnetite (a black, crystalline solid) that I was able to successfully synthesize thermite, using a recipe I’ve described in previous post.

I eventually moved on to using purified, store-bought reagents for safer reactions, but I still had a large amount of magnetite leftover. Eventually another use of it occurred to me when I read this tutorial, which outlines the unusual properties of a ferrofluid, or magnetic liquid. Ferrofluids consist of ordinary solvents, like gasoline or acetone, that have been mixed with a high concentration of nanoscopic iron particles. The tiny bits of iron essentially act as bar magnets, and so they align in unison with an applied magnetic field just as the magnetic needles of a collection of compasses would. But because the iron particles are so small, Brownian motion (the “mixing” that constantly occurs in liquids due to the chaotic thermal motion of their constituent particles) keeps them suspended within the fluid. As a result, the liquid can shift from behaving like the solvent in the absence of the applied field to behaving as a solid when a magnet is brought near the liquid.

I managed to make a very rough ferrofluid by finely grinding up my leftover magnetite and then using the recipe found on this website. The store-bought ferrofluid used in the video at the top of this post was made using precise industrial methods, and so it naturally behaves in a much more elegant manner because the iron particles inside it are much more uniform. But my ferrofluid still exhibits two key behaviors of ferrofluids: it solidifies in response to an applied field, and it tends to form small, clumped structures rather than a single lump:

The erratic behavior of the ferrofluid can be seen as a simple type of phase transition, in which a system subjected to a smoothly varying stimulus (the proximity of the magnet to the fluid) undergoes a discontinuous change in behavior (the sudden appearance of peaks and lumps in the fluid). Phase transitions are crucial in biological systems in which many autonomous parts (like blood cells or the individual members of a school of fish) must behave as a collective entity for mutual benefits. In the ferrofluid example, the mutual benefit is energetic efficiency—the fluid tends to arrange itself in such a way as to minimize its internal energy. The individual particles of iron are initially independent and diffuse freely through the solvent, but when the external field is applied it becomes energetically favorable for the particles of iron to align and congeal into collective aggregates. In the first video, when the magnet is far from the fluid, large lumps tend to form that have well-defined peaks and arrangements. But when the magnet is brought much closer, these peaks tend to disassemble into many small, hairlike filaments because such structures contain less internal energy. The reason for this behavior is that the energy of the fluid is mostly stored in its surface tension–the collective attractions between different magnetized iron particles on the surface of the liquid hold energy in the same manner as distended springs. When the magnet is brought much closer to the liquid, it becomes necessary for the system to offset the excess energy by further increasing its surface area, resulting in a greater number of small structures.

Jacob’s Ladder

The principle of operation of a Jacob’s ladder is straightforward: A plasma arc is hot and prefers to take paths of least resistance. The hot air made by the spark is more conductive than ordinary air, and so as it rises away from the plasma, it creates a moving conductive channel. This phenomenon is responsible for the “arc” shape of high voltage sparks between terminals, as the spark starts out straight and then bends upwards due to air heating.

If two long, parallel rails are used as electrodes, then the arc will actually travel up them. Generally the rails form a sharp “V” shape, as the rising hot air allows longer arcs to be drawn:

In my Jacob’s ladder, I used a high-wattage Neon Sign Transformer(NST) as a voltage source. Much to the dismay of everyone else using my power gird, the NST takes 120V, 3A wall power and transforms it into a 12,000 V, .03A output signal. Note that conservation of energy holds true– Power = Voltage x Current results in roughly 360W for either side of the transformer. But the high voltage end is much more effective at breaking down air (at the cost of amperage, which affects brightness and loudness of a spark), and so it readily ionizes the air between the rails to start the arc. It is extremely important to have straight, rigid rails– I used two cheap, non-anodized aluminum bars I found at Home Depot (anodization makes the surface of the bars resistive, which is undesirable here).

If our atmosphere consisted of other gases, it actually would be possible to make a Jacob’s Ladder using wall power. All materials are electrical insulators at very low voltages, but once a certain threshold voltage is reached, they break down and begin to conduct. We see this occurring in the air every time lightning strikes– the voltage in the cloud finally becomes high enough to turn the air into a conductor. Copper, due to the relative freedom of its electrons to move about, will gladly allow even modest voltages to create currents, and so we call it a conductor because at most working voltages it acts as such.

Generally speaking, we refer to gases in their conducting states (such as the current-carrying air between the rails of the Jacob’s Ladder) as plasma. The tendency of an insulator to conduct at high enough voltages is quantified by its dielectric constant (k), which determines how readily it breaks down compared to a pure vacuum (which has k = 1). For example, paper has a dielectric constant of 4, which means that making current travel through paper requires roughly 4 times the voltage it would take to make current travel through the same thickness of vacuum. The process by which a dielectric(a word synonymous with “insulator”) turns into a conductor is dielectric breakdown. Lightning and the spark of the Jacob’s Ladder are both cases of dielectric breakdown in air.

Once the arc goes out, the rails need once again to build up enough voltage to bridge the gap at the bottom of the ladder and start a new arc—but once this arc starts, it is easy to maintain. In high-energy circuits, dielectric breakdown usually occurs in solid materials, like glass or plastic, where the heat of the spark is usually sufficient to burn a permanent path through the dielectric. This means that, even once a spark goes out, current is stil able to pass through the path of the original spark, causing damage to the circuit. If this effect occurred in air, then thunderstorms would tend to generate single lightning bolts, which would last for hours while the storm transferred all its excess charge to the ground. Instead, when a bolt of lightning forms, the air molecules it ionizes are quickly replaced by other molecules that drift into their places, preventing the arc from carving a permanent path to the ground and thus limiting the life of the lightning bolt.

Ignition Coil Spark Generator

This is probably one of the easiest high voltage projects I have ever stumbled across. For the cost of an ignition coil and a trip to RadioShack, I was able to intermittently interfere with antenna reception in my household for years.

The cool thing about car ignition coils is that they are essentially miniature solid-state Tesla coils—just like their older cousin, they essentially convert a low voltage, high current power source (such as a 12V, 10 amp car battery) to a high voltage, low-amperage current (such as that irritating 1000V, .003 amp shock you get from a doorknob). Purists may note that, technically speaking, Tesla’s original patent called for an open-air circuit, with energy transfer through a “resonant coupling” between two coils with no electrical connection (like a radio antenna and receiver). But the legacy of the Tesla coil lies in the general idea of transformers, which are vital to almost every modern electronic device.

Transformers (ie, increasing or decreasing voltage while doing the opposite to current) rely on one of Nikola Tesla’s most important inventions: alternating current. Alternating current is typically described as electrons wiggling back and forth in a wire, as opposed to direct current in which they continuously flow through a wire. However, this description obscures the true beauty of alternating current, which is better visualized as waves in a lattice of metal atoms.

In high school chemistry, the transition metals that comprise many conductors are typically described as a “sea of delocalized electrons.” This epithet explains many meaningful properties of metals: they are a fixed atomic lattice containing a bunch of electrons that constantly float around, uncommitted to a single parent atom . This means that something like an incident electric field can very easily cause the electrons to move; and in DC current, a constant electric field makes all the electrons move in one direction. While this carries kinetic energy very well, a common misconception is how it does so: electrons move incredibly slowly, and so the current is not due to a torrent of electrons suddenly cascading through the circuit. Rather, the observed current comes from a large number of electrons getting a tiny boost in velocity– their power is in their sheer numbers, rather than by their individual speeds. In fact, it actually takes about 4 hours for a given electron to make the round trip from your light switch to your ceiling fan– most of the power that travels in the circuit is due to large numbers of slow electrons arriving at the light bulb at the same time.

In alternating current, it’s not the particles themselves carrying the energy—it’s a disturbance in them. In fact, when you flip a light switch, an energetic electromagnetic field travels through the wires in your wall at the nearly the speed of light (because, after all, light is an electromagnetic field). It might slow down some due to other interactions with the metal in the wire (light also slows down when it travels through non-conducting materials, like distilled water), but for all practical purposes the change in current travels super fast. Tesla’s AC was brilliant because it exploited the fact that the particles don’t even have to move at all to carry a lot of energy. Going back to the sea metaphor, imagine how long it takes a single water molecule to travel from Russia to the US, even if currents are steadily eastward. Nonetheless, things like tsunamis can still do a lot of damage; not because of individual particles carrying energy over the entire ocean, but rather because every particle between Russia and the US jumps up and down a little bit to carry the wave. Likewise, when you hit an AC switch, you basically trigger a series of well-timed kicks of energy to the metal lattice (the wire), each of which sends a nice wave traveling through the sea of electrons. Because the light bulb converts the energy of the sloshing electrons into heat and light, you need to keep making waves (using power) in order to keep it going.

The cool thing about kicking the electrons is that the resulting current changes in time– and so it can cause other things to change in time as well. Within an ignition coil, a magnetic field is set up in a wide set of wire loops every time a kick passes through that part of the circuit. When the current stops (the wave passes), the magnetic field collapses. Tesla thought to put a second coil inside the first, so that this collapsing magnetic field has somewhere to go. Just as running current through a coil created a magnetic field, running a dying magnetic field through a different coil will create a current. But say that I’ve made my second coil out of 100 turns of wire, while my original coil only had 10. The voltage induced on my second coil will be 10x my original voltage, and my current will be 1/10 it’s original value. In ignition coils, this ratio is more like 1000:1, allowing me to take a modest input voltage and make a nice, long spark.

It’s important to note that the two coils have no electrical connection between them– the magnetic field literally reaches one coil from the other and transfers the energy (a concept that is the basis of radio communication). Thus ignition coils have four terminals: a primary coil where you can connect a battery, and a set of secondary terminals that are the high voltage outputs.

There’s just one caveat: the magnetic field has to begin to die in order to be resurrected on the second coil. This means that the current that created the field has to come on and off as well– a requirement perfectly suited for AC current. The trouble is that a car battery is DC, and so all it does is pour electrons, rather than make waves.

Your car solves this problem by turning the DC on and off really, really fast using a device known as a distributor, which causes the current to stop and start several times each time a wheel rotates. Each cycle makes a new kick of energy pass through the circuit and thus creates a new ignition spark in the engine each time (and thus creating a steady, continuous set of rapid combustions that allows the car to run smoothly). This used to be done mechanically, but recently integrated circuits have largely replaced physical distributors.

For my project, I faced a similar issue of creating adequate waves in my ignition coil. If I attached the coil to a 12V lantern battery and then rapidly connected and disconnected my leads, I would get sparks, but only as often as I was able to manually connect and disconnect the switch. I considered using modern integrated circuits to do the switching for me, as occurred with distributors in cars, but I realized at the time that I barely knew how to send an email, let alone program an integrated circuit.

A trick I found online is to use what’s known as an electromagnetic buzzer, or a circuit that keeps trying to disconnect itself. Common devices known as N.C. relays have four terminals: two for control, two for output. The control terminals are electrically connected, as are the output (although the latter have a special magnetic switch connecting them). If I connect DC power through the two control terminals, it activates an electromagnet that then pulls open the output switch. If I disconnect power, the field goes away, and so the output terminals close up again. Thus in an ambient state, the N.C. relay is Normally Closed—anything connected to the output terminals will have a continuous circuit through the terminals.

A good way to consider this is that the N.C. relay is a logical NOT gate: if I have power through the inputs, then power is not allowed to run through the outputs. If the inputs lack power, then it is possible for power (in a separate circuit) to run through the output terminals.

N.C. Relay

An N.C. relay blocks current through another circuit when power is run through one of its switches, but allows the other circuit to flow when no power is being run through the input terminals. This illustrates how one circuit can control another, unconnected circuit.

The electromagnetic buzzer uses an NC relay to set up a paradox: what if a circuit were to run into one input terminal, out the other, back in one output terminal, and then out the other and into a circuit? Running power through this configuration would definitely provoke the electromagnet, which would then open the switch- and thus break the circuit. Since the magnet then turns off, the switch closes again—causing the magnet to turn back on, the the cycle to repeat. This means that, if I wire a common relay up correctly, I can end up with an electromagnetic paradox, a circuit that constantly strives to turn itself off.

The two states of an EM buzzer. The circuit constantly oscillates between these positions.

This is the basis of the electromagnetic buzzer, which gets its name from “buzzing” between on and off. If I were to plot the shape of the current in the circuit over time, I would get square waves, where the current runs for a short period at a constant DC value (making a straight line on the plot at this value), and then stops for another regular short period (making another straight line at zero).

Returning to the ignition coil, this means that all I need to put AC through my coil is my battery, the N.C. relay (correctly wired as a buzzer), and my coil. You can see in the video that this process creates a constant spark at a fairly low frequency (you can see it jumping, for example). In true Tesla coils, the sparks jump so fast that they give the illusion of being animated or continuous (at which point they are called “streamers” because they appear to stream out of the terminal).

The full circuit for the continuous ignition coil driver. This should produce 2-3 inch sparks at a steady rate.

For an even higher output voltage (3-5 inch sparks), another coil can be connected in reverse parallel with the original coil primary terminals (ie, in parallel with wires crossed). Sparks will jump between the output terminals of the two coils because they will have opposite polarities due to the reversal.