I hope you like it

Its method of operation is rather simple. The GM tube is a partially evacuated tube containing low pressure inert gas; typically neon, and a small amount of a quench gas; isobutane or a halogen (halogen tubes last forever while isobutane ones have a finite lifetime due to radiation breaking down the organic gas). There is a long cathode wire in the center and a the metal shell is the anode. A high positive voltage is applied to the cathode, this voltage being just under the amount required to ionize the gas in the tube. Usually it’s anywhere from 500 to 900V depending on the tube. When radiation enters the tube it ionizes the gas it transverses, and when a path is made between the cathode and the anode the tube becomes conductive for a moment before the quench gas “puts out the fire.”

There are two ways one can use a GM tube in a circuit:

The first method is to simply use the GM tube in series with an amplifier. When there is an event in the tube it shortly becomes conductive and passes a “blip” onto the amplifier, and while this reduces component count the tube is not properly grounded and it may be possible to receive a small shock from it.

The second method properly grounds the GM tube and the signal is passed onto the amplifier via a small DC blocking capacitor. When there is an event in the GM tube the HV is pulled down to ground and this signal makes it through the capacitor. This is the preferred method of using a GM tube, and in fact I don’t think anyone even uses the other method.

To run a GM tube you’ll need to generate a high voltage at a very low current. Now one could do this using discrete components, but my method is to simply use a CCFL inverter and a Greinacher voltage doubler. By powering this arrangement with an adjustable voltage regulator one can make a variable voltage power supply perfect for a GM tube. Such a circuit also draws very little current and that allows it to be run off a 9V battery for a reasonable amount of time.

As for pulse detection, I find that an LM386 audio amplifier does a good job. This amplifier can directly drive a speaker, though make sure you put a 1000µF DC blocking cap in series with such a speaker or else the amplifier will waste lots of power. The pulses that come out of the amplifier can be counted with a microcontroller.

Types of GM tubes

Geiger-Müller tubes do not equally respond to all types of radiation, and different tubes are designed to better respond to certain types. In general, there are four different types of GM tube:

The typical thin metal-walled Russian geiger tube such as the one shown here will not respond to alpha radiation at all, and it will be able to maybe detect high-energy beta radiation. It is useful only for gamma and x-radiation, and that’s not saying much considering GM tubes are piss-poor at picking up that kind of radiation.

Aside from simply detecting the presence of gamma and x-radiation, these $5 Russian tubes are not much good for anything.

Typical GM tubes respond to only 3% of incident gamma and x-radiation. Certain tubes however, are designed to better detect gamma and x-rays by utilizing a rather thick metal wall. While this may seem counterintuitive, a thick wall actually improves detection efficiency because of the way a GM tube detects this type of radiation. Unlike a beta particle that may start an avalanche in a GM tube, electromagnetic radiation is nothing more than photons; photons that must knock an electron out of the wall  in order to be detected. Thicker wall = more atoms = more electrons to be knocked out.

Metal wall tubes are useless for detecting alpha radiation, such as that emitted by the 241Am found in smoke alarms. In order to detect such a large particle the tube needs to be transparent to them, and the only thing that is both transparent to most alpha particles and is strong enough to hold back a vacuum is mica. Such mica-windowed GM tubes cost more than their all-metal counterparts, but for alpha detection and beta measurement it’s the only tube suited for the task. Be careful though, a mica window is very fragile and may pop if touched!

The last type of Geiger-Müller tube is also a mica-window one, but unlike the other GM tubes which are cylindrical this looks like a pancake. Aptly named pancake tubes, these detectors also have a mica window, though it’s much larger than that of an end window tube. Thus, they much more sensitive to low levels of radiation than any of the other types. Their downfall is the fact that pancake tubes are ridiculously fragile; simply looking at it in an ugly way may be enough to break the mica window.

Mica window GM tubes are harder to find than other types of GM tubes but occasionally they will pop up on auction sites. If you are really bent on getting your hands on one though, LND makes some great quality GM tubes.

More notes…

The time it takes for the quench gas to extinguish the discharge inside the tube is known as the deadtime. Typically this dead time is 100us, though it varies for each tube. During this dead time, the tube is not able to detect any incoming radiation so it puts a limit on the maximum radiation flux a GM tube can detect. When there is too much radiation bombarding the tube it saturates and the quench gas is unable to stop the discharge.  Since a saturated GM tube will read “0″ on a counter and produce no clicks, you may be exposed to dangerous levels of radiation and not know it. Fortunately it takes a lot of radiation to saturate a GM tube; the kind of radiation one would find in an x-ray beam or next to a nuclear reactor core. Therefore unless you are either 100% clueless as to what is going on around you or have had a nuclear explosion nearby, saturation isn’t much a problem.

When it comes to x-rays GM tubes have a very non-linear response, so although they are great for detecting the presence of x-rays they are terrible for making measurements. Just something to keep in mind.

Below I have a video of a mica-windowed GM tube detecting the radiation from different sources. First some DU bits, second a coleman lantern mantle, third a piece of fiestaware and fourth some 241Am from a smoke alarm. Notice that the alpha source must be brought very close to the mica window for the alpha particles to reach the tube.

I’m always on the lookout to buy radioactive things, so if you have anything you’d like to get rid of feel free to contact me.

Since I could not just go out and buy a portable x-ray machine at walmart I was forced to build one! I found it odd that such a device didn’t seem to exist, since such a machine could prove to be very useful in remote or under-equipped areas like army camps or poor countries.

The full paper detailing this endeavor may be downloaded here, but the text below provides a general synopsis of the project.

My Machine

The design has gone through several revisions on its two year journey. The first alliteration was designed to be a self-contained unit that emitted no stray radiation. I even went as far as building a huge lead box for it! The box was constructed by melting wheel weights into a hobo-pie maker to make square plates, then welding them together with a blowtorch. It took 2 weeks, 6 hours a day to make! But at the end I realized how unreasonable it would be as the box was so heavy that it was immovable…

So the lead was melted down and a new design drawn up in autodesk. This version was to be a plywood box with an inner lead box, and it was to emit a beam of x-rays. Simple enough. I began a long tedious process of making yet another lead box, this time by first cutting out the shapes from clay, then casting the shapes in plaster. A form of lost-clay casting which would leave me with the 6 sides I could weld together. Sounded simple, but turned out to be a huge trouble due to water in the plaster that could not be baked out. After 5 attempts at casting the parts, I finally ended up with indecent, but usable shapes. They were then welded together to make an awesome hot box.

BUT THEN… my tube arrived in the mail and it was a bit smaller than I was expecting. So that box wasn’t going to work (I now use it to store my radioactive items). Yet another design was needed, and after a few months of on and off work the whole project turned out like this:

Apologies for the poor image, my good camera is on loan to someone.

The machine is split into two parts: a control box and a ‘tube head’. The tube head is the part that houses the high voltage components and x-ray tube, while the control box houses most of the circuity and the power source (LiPo batteries). Both were designed to be beautiful: burned brass corners, hammered paint finish, unnecessary lighting of the switches, display and meters… Making the device look good took a whole month itself! They are connected via  removable cable which utilizes two octal vacuum tube bases as the connectors.

The control box features a multiplexed nixie display, three dials to set the exposure time, another dial to set the kVp as well as two meters to monitor the coolidge tube current and voltage. Along with those controls we have a button to display the oscillator voltage, as well as a speaker and the on-off switch. And of course, the expose button.

The two meters were a mini-project in themselves. When I bought them from a person on ARF, one measured 0-3mA while the other metered 0 to 10V. After removing the voltage divider resistors on the 10V meter, it turned out that full scale was 50uA. Holy crap that is sensitive. In order to create a 0 to 90kV meter I needed 1.8 billion ohms of resistance in series with the meter! Making the meters glow was achieved with clever placement of EL wire, but I seem to have lost the photos of that process :-/

Compared to the outside, the inside of the control box looks to be a disgraceful, awesome rats nest of wires. The brains of the device is an Atmega 368 programmed with the arduino bootloader.  I chose a micro-controller over state logic so it’d be easier to add features later. In the end there isn’t much room for improvement however considering I used all but 1 pin! 7 pins are used to drive the multiplexed nixie display, 3 more are used as analog inputs, while another two are used to switch a pair of relays. The rest preform various functions such as monitoring voltages.

Method of Operation

The microcontroller constantly polls the three time setting knobs when the machine is idle. By mapping each knob to a 0-9 scale, 3 numbers can be established and combined to form 3 digits: 10′s, 1′s and tenths. These are then printed to the nixie display, one tube at a time over a span of 2 milliseconds. As it takes 2ms to write a number to the display, the printNixie() routine was used as the time-keeping routine of the program. 500 printNixie() calls equates to 1 second of passed time, so by using that routine instead of delay() within the program allows the system to keep time without blanking the nixie display. I suppose this could be considered a hacked version of multitasking… maybe.

After setting the exposure time, the desired anode voltage is set with the big knob. Turning this knob adjusts what voltage the regulator  will be fed to the HV oscillator (14 to 35VDC). Pressing the black button will display this voltage on the nixie tubes, so it’s not just a shot in the dark every time you turn that knob. In the future I would like to add a feedback loop, having the microcontroller monitor the actual anode voltage and adjusting the oscillator accordingly.

Once things are set, the expose button is pressed and a 10 second countdown ensues. At the beginning of this countdown a small relay switches on and supplies 14V to the tube head. That 14V is then stepped down to 3.5V via a buck converter and fed to the tube’s heater.

At the end of the countdown a second relay switches on and powers up the HV circuit. The regulator voltage is now given to a ZVS oscillator which churns out 65kHz AC. This power is then sent into a homemade transformer where it’s stepped up to 20kVAC and fed into a cockroft walton multiplier. After multiplication we are left with anywhere from 40 to 75kV (adjustable) which then gets applied to the tube’s anode and makes x-rays. Oh and yes, the CMOS microcontroller shares the same chassis ground as the 75,000V rail. Take that ESD nannies!

During x-ray production the nixie tubes glow red and a buzz from the speaker is emitted.

X-Rays!

The machine produces copious amounts of radiation, certainly enough for radiography. As an experiment, a geiger-muller tube was placed at different distances from the machine and the results recorded in this video.

As you can see by the GM tube’s screams, not only does the machine in fact work, it’s pretty scary too. At 1 foot there are >8,300 counts per second: so intense that the geiger tube doesn’t even have enough time to recover between events!

Radiography

Of course there is no point in spending years building an x-ray machine if no pictures are taken with it! Below we have two of my first radiographs: a pen and a laptop hard drive. Since lenses do not work in x-ray land, making a radiograph is a shadow process: all images must be ‘shadows’ of the subject being imaged.

X-rays are invisible and are only viewable via the use of a scintillation screen: a plastic/chemical sheet which fluoresces in the presence of ionizing radiation. Scintillation screens, or intensifying screens come in a multitude of brightnesses and colors, but the standard “fast” green one is the one most commonly available and used.

Radiation Sickness

Acute radiation sickness occurs when your body has absorbed a large amount of ionizing radiation, usually on the order of several sieverts. What makes radiation lethal is the effect it has on DNA. When a high energy particle, be it a photon or some other particle collides with DNA it breaks bonds and rearranges the bases. Normally your cells can repair this damage, but if a cell fails at that task it often commits suicide before it divides. For long living cells such as muscle this isn’t too much of a problem, since the other cells have time to replace the dead ones. For short-lived cells though, this apoptosis becomes a major issue as cells are dying too fast to be replaced.

Such short lived cells include the mucus-making cells that line the intestinal wall. When exposed to enough radiation, these mucus cells start to die off en masse, and so are not replaced. No mucus cells means there will be no mucus, and no mucus means there is no protection from stomach acid. The intestine stops absorbing food particles, acid burns the tissue, and eventually you die of sepsis. If somehow you survive this ordeal, you will now need a bone marrow transplant since the short-lived bone marrow cells have died off. Radiation sickness symptoms include nausea, stomach pain and a lack of energy, and a detailed chart of symptoms can be found here.

And that’s why we shield ourselves from ionizing radiation! Keep in mind that it takes a very large amount of radiation to cause radiation sickness, not something a fiestaware plate or even a radium painted clock could ever produce. However, a Coolidge tube is certainly capable of generating very intense radiation.

Radiological Protection

In order to reduce the amount of radiation you are exposed to, shielding is put in between you and the radiation source. This shielding reduces the amount of radiation to an acceptable level. What exactly is an acceptable level though? In the end, that’s up to you to decide, but generally the idea is to go as low as reasonably practicable. In order to help determine what is an acceptable level, I have here a chart of activities that expose a person to radiation.

There are multiple different types of radiation and each type must be treated differently when it comes to radiological protection. Some types require more shielding than others and since this is a guide I will now do some explaining. First with particle radiation, then with electromagnetic radiation. But before we do that let’s discuss energy.

Radiation can have different energy levels, energies which are measured in electron-volts (eV). One electron-volt is defined as the amount of energy gained by one electron as it moves through an electric field of one volt. For example, green light photons usually have an energy of about 2.3eV, while blue light has an energy of 3eV. More energetic radiation is able to cause more damage when it hits something, and this is why microwaves such as those emitted from cell phones (0.00001eV) cause no chemical damage while gamma rays which may have an energy of 5 million eV can cause major damage.

Generally higher energy radiation is harder to shield than lower energy radiation, but when it comes to particle radiation the type tends to play more of a roll when determining penetration. Usually particle particle radiation tends to be the least penetrating.

Particle Radiation

Alpha decay is the most common method of radioactive decay. What happens with alpha decay is the unstable element ejects a duly ionized helium nucleus known as an alpha particle. In fact, all the helium on earth comes from the decay of uranium and other elements underground. Although alpha particles are very high energy, often having energies in the MeV range, they are very large. Because of that they are stopped very easily. In fact an alpha particle cannot even make it past a piece of paper, or even skin for that matter. Alpha particles usually have a hard time making it through more than 3cm of air, so therefore no special shielding is necessary for alpha radiation. Just don’t eat it and you will be fine.
The next type of radioactive decay is beta decay, a process in which a neutron is converted into a proton and in exchange an electron and a neutrino is ejected. The neutrinos are of no concern since they are small, light and neutral, and thus pass through any matter they encounter and fly off into space like a ghost. The speedy electron known as a beta particle has a negative charge though, so it can interact with matter and thus pose a hazard. Fortunately beta particles are not very penetrative; all that is needed to shield them is a piece of aluminum foil.

The last type of particle radiation is known as neutron radiation; something that is created when atoms are either fused together or fissioned apart. Unlike all other forms of radiation, neutrons can actually turn things radioactive! This is because when a neutron smacks an atom it may stick to it, turning that atom into another stable isotope or possibly a radionuclide. Unless you are either playing with Farnsworth Fusors or uranium reactors neutron radiation is not much of a concern, but nonetheless it is best shielded with light materials of all things, materials such as water and aluminum. Large amounts of water make an excellent neutron moderator, but because of this the human body does too. Therefore neutron radiation is especially dangerous to living things so do everything in your power to avoid it.

Electromagnetic Radiation

Now that we have particle radiation out of the way it’s time for electromagnetic radiation:  highly energetic photons. There are two types of electromagnetic radiation you should concern yourself about; gamma and x-rays.

First let’s start with gamma rays. In certain radionuclides the atom’s nucleus is left in an excited state after beta or alpha decay. This energy is then released via a very high energy photon. By high energy I mean several MeV, and due to that gamma rays are very penetrative. It takes quite a lot of material to stop them, so lead is often the material of choice for gamma shielding. If for some reason you have a very active gamma source use plenty of lead to shield it. Something like 5cm or more of that grey metal should be sufficient.

X-Rays

The other type of electromagnetic radiation I have to discuss is x-rays. X-Rays are produced when electrons dump a large amount of energy into a single photon, thus creating a very high energy light particle. X-Rays are a lot like regular light: they travel in straight lines, can be reflected somewhat, and scatter in the air much like a green laser beam. When experimenting with x-rays, always make sure your lab is of light construction. While cinderblock walls are great for stopping x-rays from escaping your lab, they are also great for reflecting them back at you! It’s better to have them escape rather than to have them bounce around.

When possible, be sure to either point your x-ray beams down to the earth or up in the air: anywhere where it is unlikely to be intercepted by an animal or human. NEVER power up an x-ray tube in a shared residence or an apartment without full knowledge that the radiation will be contained, and NEVER intentionally expose yourself to x-radiation.

It is important to shield yourself from x-rays to prevent overexposure. The amount of shielding required is entirely dependent on the energy and quantity of x-rays being stopped. Lead is the ideal shield for x-rays because it is cheap, easily workable and has a high nuclear charge; something that lets it absorb electromagnetic radiation very well. For convenience I have prepared this chart of energy vs. attenuation vs. amount of lead needed using the standards set by the International Atomic Energy Agency.

 Considering the x-ray producing item being shielded is probably a Coolidge tube, much of the shielding should be done via the use of a tube jacket. It’s simply a lead jacket that is fitted around the tube with a hole is punched in the center to let the beam out. Using lead sheet and a soldering iron, a tube jacket can be made in about a half hour. Using one will certainly save you a lot of headache later on.

Despite your best efforts at shielding the x-radiation, Compton scattering and a little bit of reflection will scatter some around and back to you. This is for the most part unavoidable, but always have some sort of radiation detecting device nearby so you know you are in a safe place to stand. An acceptable maximum level of scatter would be about 1000 counts / minute, or 10 times the natural background level. Where I live the natural background level is measured to be 100cpm, so by exposing myself to 1 second worth of scattered radiation I am absorbing the equivalent of 10 seconds of taking a nap, or perhaps the dose received by eating one eigth of a banana.

As with any kind of radiation, be it light, radio, gamma or x-rays, distance is the most useful tool for protection; –the farther you stand from the source the less radiation you will receive. Inverse square law applies here, so by simply doubling your distance from the source the does rate will be 4 times less. Nothing beats getting the hell away from a source of radiation!

That’s about all I have to say about radiation and radiation safety. Be smart, and remember there is no cure for radiation sickness.

PS, a there is a calculator for this stuff.

The Coolidge tube’s method of operation is once again, exactly like a thermionic diode. The heater is given a little bit of current to warm it to incandescence and the hot tungsten cathode then boils off electrons. These electrons being negatively charged, are attracted to the positively charged anode and proceed to move towards it.

When the electrons slam into the anode they lose energy. 99% of the impacting electrons end up grazing atoms and they stop slowly, at the same time dumping their energy in small potions as heat (infrared photons).

 Some electrons however, maybe 1%, generate x-rays in a process called bremsstrahlung, aka braking radiation. This small portion of the impacting electrons comes very close to the anode’s atoms and slingshots around the nucleus, dumping most of its energy at once. These electrons were moving very fast, and this huge abundance of energy is dumped all on one photon.

Now we have a very energetic light photon; an x-ray. This x-ray then exits the glass envelope.

What determines the energy of the x-rays produced is the amount of voltage present on the anode. It’s quite simple actually, more voltage means more attraction, and more attraction means the electrons impact faster. Faster electrons means they slingshot around atoms with more force, and thus “harder” (higher energy) x-rays are produced.

Bremsstrahlung is a continuous spectrum of radiation, akin to a “white light” source. Since most x-ray producing electrons graze some atoms before slingshotting they tend to lose some energy before they make any x-radiation, and thus a whole range of energies are produced. The maximum energy that the x-rays can have is limited to the voltage applied on the anode (kilovolts peak, or kVp), but most of the x-rays produced by bremsstrahlung are low energy x-rays.

Since the lower energy x-rays are not able to escape the tube’s glass envelope the bremsstrahlung spectrum of a glass Coolidge tube actually looks like this; most of the soft x-rays are blocked. To combat this issue some x-ray tubes have a beryllium window that lets the soft rays through. Beryllium windows are expensive though, so most tubes do not have them. These soft x-rays are useless for typical radiography purposes anyway so there is no need to keep them.

The intense heat that the electron beam produces can melt most everything, so tube anodes are typically made of tungsten embedded in copper. This copper protrudes from the tube so heat can dissipated. Coolidge tubes often come with an “anode heat storage capacity” chart on their datasheets, but for the most part the heat storage capacity is 7kJ for a dental x-ray machine tube. If you exceed this rating you can expect the tungsten to melt, but normal operation also causes some pitting and cracking on the anode, the likes of which can be seen in this image.

 Aside from having a high melting point, the tungsten anode also gives the tube another advantage. There is a second mode of producing x-rays, something called characteristic production. In this mode, electrons hit other electrons in the atoms’ lower shells and knock them out. These holes are then promptly filled by electrons from higher shells who emit x-rays on the way down.

Tungsten K-shell electrons have a binding energy of 69.5keV so to kick these out your impacting electrons must have energies greater than 69.5keV. Typically you’ll need to give the anode 72kV or more to accomplish this, hence the standard 75kV x-ray tube.

After you succeed in kicking out a K-shell electron, the hole will immediately be filled by an electron from the L-shell, binding energy 10.2 keV. The difference between these energy states; 69.5 keV and 10.2 keV gives us characteristic tungsten x-ray energy of 59.3 keV. Once 72kV is applied to a tungsten anode Coolidge tube x-ray production at that energy spikes, as seen by this chart here. This is the full spectrum of a 75kV tungsten target x-ray tube.

Anode voltage controls the hardness of the x-rays produced but current controls the intensity of the x-rays. Tube current can be controlled by adjusting the temperature of the heater, something itself controlled by the voltage applied on the heater. More volts = more heat = more current = more x-rays. Typically tubes from dental x-ray machines are designed to be run at 7mA, though this varies. If you can find the tube’s datasheet there is usually a voltage vs anode current chart on there somewhere. Don’t forget that more current also means more heat and too much heat will melt the anode.

In certain Coolidge tubes there is actually a third electrode in front of the heater and much like a triode, different voltages applied to this grid will allow different amounts of current through. However for hobbyist purposes this grid can be left unconnected and the tube can be controlled via heater voltage, which is much easier to adjust in my opinion.

Dental Coolidge tubes are designed to be operated under oil and while they do work in air you still should use them under oil. Oil is a better heat conductor than air and it will allow the tube to handle more power than it could in air. Tubes should also be operated under oil to prevent the high voltage from just “flashing over” the tube rather than going through it. When the tube flashes over the high power sparks may crack it, and that’s no fun. In order to prevent flashover always make sure the tube’s cathode is fully heated before applying high voltage.

Most Coolidge tubes are designed to be surrounded by a jacket of lead containing material. The tube jacket blocks off-axis radiation and makes an overall better quality x-ray beam. Tube jackets aren’t hard to make, just some lead sheet and a solder.  Certain tubes don’t require a lead jacket as they have anodes designed to specifically block off-axis radiation (right), though I still recommend jacketing them.

And that’s about all I’ve got to say about this tube. I’ll now hand the discussion to doctor Coolidge .

You don’t need lots of money to make high voltage capacitors, in fact some pretty decent ones can be made with some cheap and readily available materials. This is because capacitors are very simple devices; consisting only of a dielectric and two plates. Most often a capacitor’s plates are just aluminum foil, and reynold’s wrap is easy enough to obtain, but what about the dielectric?

Enter the overhead projector sheet. Transparencies as they are commonly known as are nothing but acetate film, and while this is not the ideal dielectric for a capacitor it still does quite a good job. Typically a four mil OHP sheet can withstand 14kV before breaking down. As for obtaining them, the cheapest I have found these sheets is $10 for a box of 100, enough for about 16 capacitors.

How you make the capacitors is a rather trivial task, all that needs to be done is some cutting, flattening and rolling. Below I have an image that explains the process. Multiple sheets of OHP sheet are used to increase the capacitor’s voltage rating, and two sets of sheets are used so the capacitor can be rolled up.

Some tips:

• Use a rolling pin to flatten the foil.
• Roll the cap up very tightly.
• Use thin wire to minimize trapped air.
• Secure using zip-ties
• Leave 4cm between the edges of the OHP sheet and the foil to prevent flashover.

If all goes well you should end up with a capacitor that ought to be good for about 50kV. Typically these home-made caps will have a capacitance of about 3nF as can be seen by this LC meter’s screen. That’s definitely not too shabby considering a capacitor such as that would cost you $40 surplus.

Keep in mind that this capacitor is not as robust as a commercial one due to the inclusion of air between the plates and the dielectric. Yet, if you soak the assembly in molten beeswax and simultaneously pull a vacuum the air will be replaced with wax and make a much more robust capacitor. Some sort of modified pressure cooker could do the task, but that’s another story for another day.

Just to prove that these capacitors do in fact work, here is a video of a couple of them placed in parallel with a flyback transformer. They store up the energy and release it near instantly as a bunch of very loud blue-hot sparks.

Remember that they are capacitors and they will store a charge when the power is removed, and being high voltage caps they will give you quite a jolt if you let them bite you. An observant eye will notice that I discharge the capacitors at the end of this video. For those curious, the hum is the from a 48V 6A switch mode power supply.

Uses for these capacitors include filtering the high voltage from a DC flyback, building a large CW multiplier or just annoying the hell out of anyone within earshot. If nothing else they are a fun weekend project.

Because I had nothing to do and had an empty piece of ferrite I decided to make the largest AC flyback transformer that I could. Now if you don’t care how I made it you can skip to the video at the end of the page, but if you do you may read the text below and learn how to make your own.

 First you’ll need to find a suitable ferrite core. The one I had to use was made at least 50 years ago and was part of an old flyback who’s winding pretty much fell apart. Because I’m a klutz the core has a crack in it, though when repaired with superglue it’s good as new.  What’s special about this core is it has a huge winding window, and that allows for quite a large secondary. Cores such as this can be found online if you look hard enough, one this size might cost you $14. 

I had a bunch of 28AWG magnet wire and nothing to do with it, so I decided to use this wire to make the transformer’s secondary. It’ a little thicker than I’d like, but the thinner stuff I had was 40AWG which was way too thin. I chose to make a disk shaped secondary for a number of reasons;

• Because there are few windings per layer, inter-winding capacitance is reduced and the transformer can run at higher frequencies.
• Because there are few windings per layer, there will not be a large voltage difference between layers; something that could cause arcs.
• Because a disk shaped secondary is not very fat I can use tape to make it.
• Disk shaped secondaries look cool.

Ideally I should have used kapton tape to make the transformer since it has very good dielectric strength but I had to make do with what I could get at the hardware store. Therefore I bought some red electrical tape and some teflon pipe thread tape. I bought the nice 3M tape that comes in the little plastic containers because it is very stretchy and very sticky, perfect for making the transformer. After some testing I found that the electrical tape failed at 800V while the teflon tape failed at 1kV. Because of this I’d need a couple layers of tape to prevent failure.

 The first step was to find a nice cardboard tube and wrap some electrical tape on it, then wind some wire around that. In order to prevent failure the wire must not come close to the edges of the tape. I managed to fit about 23 windings nicely in the middle.

 The next step was to provide some insulation for the layer by wrapping 5 layers of teflon tape around it nice and tight. Since teflon is very deformable it conforms to the windings and squeezes out air that could cause corona. This is a plus since corona would eat away at the insulation.

 Then one layer of electrical tape was wound to so the next layer of windings could be put on. After that I repeated these steps until my brain melted and I ended up with 40 layers; a task that took 25 hours, a super mighty and a two liter of loganberry to accomplish.

 Because magnet wire is fragile I attached some silicone insulated wire to the windings and super glued it in place. I chose silicone wire because it is very flexible and it can stand up to HV quite well. Not only that but super glue sticks to it very well. Unfortunately it costs a little more than regular plastic insulated wire.

 Since tape isn’t perfect it would in time slip off center and ruin the secondary. To prevent this I bought a can of liquid electrical tape at the hardware store and coated the secondary with it. This will both keep the windings in place and make the secondary look better. 

 Every transformer needs two windings to work, so the primary winding was made by taking two strands of 18AWG wire and wrapping a 4+4 turn center tapped primary around a paper and tape form. Notches were cut in the coilform to prevent rotation.

After adding some tape to the core pieces to make an air gap I slid the core in the paper tubes and the friction held it in place. When powered up the core will clamp together because of the magnetic field, so no external clamp is needed. And here we have the largest disk AC flyback yet made, a transformer which I have aptly named “the fryback”.

When powered by a ZVS and run at 36V the fryback makes some 5 inch long, 2.5cm fat arcs. The voltage is about 20kV while the current must be 100mA or more. That’s all well and good, but when fed 48V the fryback really shows what it’s capable of. It outputs 8kV (11kVrms) and the current is… a lot. The arcs I can draw from it are 1 foot long!

The fryback can’t handle more than 48V because the core gets very hot at that voltage and it is probably in saturation. The secondary could theoretically handle much more power though because it stays nice and cool, and because the interwinding voltage is only 1250V while I used 4kV worth of insulation.

And what you’ve all been waiting for; the video.

Unlike things such as neon sign and microwave oven transformers, an XRT is not simply “plug and play”. This is because you can’t just buy an x-ray transformer all by itself. Instead you must buy part of an x-ray machine and pull it out of there yourself. X-ray transformers are located inside the “heads” of the x-ray machines; the part that the dentist puts next to your face when you get your picture taken. An x-ray head typically sells for about $100 used and they contain both the x-ray transformer and a nice x-ray tube. You could then sell the tube for about $60, or keep it for later experiments.

Tube heads are typically held together with either lots of screws, solder or epoxy so it takes quite a bit of work to get one open –unless you use a dremel like me. Take these heads apart outside because they’re filled with insulating oil. More recent tubeheads are filled with mineral oil, but some of the pre-1970s ones are filled with PCB oil. Contrary to popular belief, polychlorinated biphenyls are not deadly. Although they do absorb into your body, they are very inert oils that won’t react with anything in your system. Do not be terrified of PCB oil as there is little justification for this over-hyped fear. Regardless of what type of oil is inside, save it for later use because it’s a wonderful insulating oil, much better than the stuff you can get at CVS.

Once you pull out the transformer you’ll have to identify its windings, but before you do that you need to know what to look for.

An XRT has 6 connections. The thickest of these connections will be the primary coil and these two wires get connected to the mains. Since an XRT is a center tapped transformer there will be two HV outputs and these are usually just balls of solder on the windings.

The ground of one HV secondary will be connected to the core, while the ground of the other secondary will be connected to a “mA” terminal on the x-ray head. To measure secondary current you may put a mA meter in between this wire and the core. If no measurement is needed you may solder the wire to the core.

X-ray transformers are not designed to be run in open air, and if you attempt to do just that expect them to die immediately from unwanted arcing. To prevent the destruction of your precious transformer you’ll need to put it under oil. Any oil will do, including mineral, vegetable, peanut or canola oil. If you decide to use an organic oil be warned that you’ll have to replace it every five months or it will go rancid and stink up your transformer. Because the oil I harvested from the x-ray head wasn’t enough to fill the container I made I decided to use canola oil for this transformer. Making an oil tight container is rather tricky if you have no clue what you are doing, so see my guide on doing that here if you want to make one.

The purpose of the oil is to provide better insulation for the transformer to prevent unwanted arcs. If the transformer has been out of oil for more than 20 minutes chances are air bubbles have formed between the windings; air bubbles that will cause insulation issues. Ideally one would pull a low vacuum on the transformer to get all the air out of the windings, but in the case that you don’t have a vacuum pump or refrigerator compressor there is an alternative, yet messier option. That option is putting your hands in the oil and massaging all the air out of the windings until you can squeeze out no more bubbles. It’s not as good as pulling a vacuum but 9 times out of 10 it will work.

X-ray transformers are not current limited; if you try to plug an XRT into the mains it’d be a near short circuit and draw currents in excess of 60A. Needless to say this is bad. Because of the high powers involved with an XRT a resistive ballast won’t be a good idea, mainly because the resistor would get insanely hot. An inductor is a much better option for ballasting XRTs, and the perfect inductor for the task is a shorted MOT. Simply put the MOT’s primary in series with the XRT’s primary and then short out the MOTs secondary. Viola, you now have a reliable ballast that doesn’t get hot and costs next to nothing.

XRTs pose some rather ‘unusual’ hazards for lack of a better term. Although an XRT might be rated at 70kV, this is actually it’s voltage under a 5-7mA load. When unloaded that XRT is capable of producing up to 200kV, and the electricity will gladly jump 8 inches through the air to bite you. This means a long, insulating chicken stick is absolutely required when you draw arcs from an XRT. A good rule of thumb is for every 100kV, make the stick 8 inches long.

Now that we have safety out of the way, here is a video of an XRT arc. It go BZZzzzz.

A CW consists of cascaded stages, with each stage consisting of of 2 ultrafast diodes and 2 high voltage capacitors.

In reality it takes several more AC cycles for the capacitors to charge up. How much the multiplier boosts the input voltage though depends on the number of stages, and there is a simple formula for calculating that voltage boost:

Eout = Ein * √2 * n

Eout is the output voltage
Ein is the RMS input voltage
n is the number of stages in the multiplier

Say you have a 6 stage multiplier and you feed it 7kV. By using the formula above you can calculate that the theoretical maximum output voltage will be 59.397kV.

Why they suck

Like all good things in this world CWs aren’t perfect. The problem with the CW is as more current is drawn the voltage starts to considerably sag. This voltage drop though can be counteracted by using either bigger capacitors or a higher frequency input, and it can be roughly calculated using this formula:

Edrop = I / ( f  * C ) * (2/3 * n³ + n² / 2 - n / 6)

Edrop is the voltage drop
I is the current drawn in amperes
f is the frequency in hertz
n is the number of stages
C is the size of the capacitors used in farads

This formula should definitely be taken with a grain of salt because while it makes sense in theory, the voltage drop in real life will be much higher. For example, I made a 4 stage CW that theoretically should have dropped only 1.4kV per mA; it actually dropped 20kV. That’s electronics for you.

As if voltage drop wasn’t enough, as current is drawn from the CW the output voltage starts to ripple. Once again there is a formula for calculating this;

Eripple = I / ( f * C ) * n * (n + 1) /2

Electrically CW’s blow, big time. Despite all their pitfalls they do work though, and until we find a better way of boosting high voltage they are all we’ve got. Both ripple and sag become bigger problems as the number of stages is increased, so it’s always ideal to use as few stages as possible and as high a frequency as practical in a CW. This means you’ll need a high voltage high frequency source such as an AC flyback transformer to power the multiplier.

Making One

Making a CW is a rather easy task since it’s such a simple circuit, so simple in fact that you don’t even need a PCB to do it. In the CW to the left I decided to use four stages, but because 30,000V capacitors are rather expensive I put two 15kV caps in series (hence why there are 16 capacitors instead of 8). A CW should be under oil to prevent excess corona losses, so I made this one thin enough to be fit inside a PVC pipe. How you make design and make your CW is your decision of course.

Because of the extremely high voltages involved with CWs wire resistance is for the most part irrelevant. Because of this the capacitors are capable of discharging in multi-kiloamp pulses; far more than the little diode at the end of the stack can handle. This means you’ll need either a resistive load such as an x-ray tube, or if you just want to make sparks a resistor in series with the output. Ohm’s law can help you figure out what resistor you need, but expect it to be in the range of several million ohms. Remember that since there are very high voltages in place here there are also very high powers, so make sure your resistor is capable of handling the heat. A 1/4W resistor just ain’t gonna cut it!

The super high voltages that a CW can produce are quite fun to play with. At these extremely high voltages a huge electro-motive field can be made, as well as plenty of ionic wind. Simply sitting next to an operating CW you can feel the field charge up your arm hair (if you have any), and there is no shortage of staticy clicks and pops. Playing with a CW is quite an interesting experience.

The sparks they can make aren’t half bad either.

How they operate is rather simple. A metal clapper swings to one bell and gains a positive charge, where it’s then repelled from the positive bell and attracted to the negative one. Upon impact the charge is lost and the clapper is once again attracted to the positive bell, and so the cycle repeats. Very little current is used in this process, so little in fact that the oxford electric bell, a battery powered Franklin bell, has been running continuously since it was set up in 1840.

The bells must be insulated from each other and the positively charged one must be insulated from ground. The clapper must be insulated as well, hanging it from a cotton or polyester thread does a good job of that.

The potential difference between the two bells should be rather large for the bells to work optimally; anything over 12kV should provide satisfactory results. Higher voltages cause the bells to ring more furiously, with a mild ring at 20kV and all hell breaking loose at 70kV.

Since I was suffering from a lack of proper bells I chose to use aluminum cans for this experiment, and used a pull tab as a clapper. One can was kept at 0V while the other was charged to +35kV via a cockroft walton multiplier. Satisfactory and mildly annoying results were obtained.