Projects

Wimshurst Lightning Machine

Build your own

Daniel Koch

Issue 60, July 2022

We made a lightning detector last month, now let's make lightning! - by Daniel Koch

Way back in Issue 2, we built a Van de Graaff generator. It was pretty rough, as we had set ourselves the criteria of using off-the-shelf parts available from hardware, camping, and major electronics stores. That limited the performance a bit, particularly in the material for a good alignment of the belt, which the entire thing depends on.

This time, things are different. The landscape is different, too. The criteria this time is performance at all (reasonable) costs. What constitutes reasonable is subjective, we realise that: All the more with the cost of living pressures mounting for many. However, many people like to see awesome projects even though they may not be able to build them, to live vicariously, as long as it's not too often.

On top of that, post-lockdowns, most of us are more comfortable with ordering things and waiting for them to arrive, which changes the options from our 'off the shelf parts' version. The online trend has also increased the products available in some cases. Further to that, 3D-printing is now very common, but not quite universal yet. Even so, made-to-order services can plug the gaps in most cases.

With those points in mind, we are going to build two high-voltage electrostatic generators. We are going to revisit the Van de Graaff generator and make it better, but that will be a near-future project. This month, we're going to build a Wimshurst machine. This machine operates on a different mechanical principle (though deep down, the electrostatics are similar) and for some makers, it will be easier to build. It eliminates the need for a belt and rollers, which can be challenging to get right. Instead, it spins two wheels, so as long as you get the centring right, it is less fussy.

CAUTION: All electrostatic projects involve high voltages. Unlike coil-based generators, these are DC voltages and can reach appreciable currents. While these devices are generally considered safe, electricity is unpredictable and so is the human body. There is no such thing as a guarantee.

Exercise caution when touching these machines, and always use a grounding rod to discharge all active parts before handling anything. This can be as simple as a wire and 1MΩ resistor attached to a grounded metal object in the room, or the floor. We attached the other end of ours to a plastic ruler to keep hands clear while discharging.

OPERATING PRINCIPLES

The Wimshurst machine works by electrostatic induction. It is incredibly hard to come up with diagrams that adequately explain this, because things are happening in multiple planes at once. Instead, we have linked to a video in the 'Reading and Resources' section which explains this really well. The heart of the machine is a pair of discs, which are set up to spin in opposite directions. Around the outer edges of these discs are a bunch of equally-sized metal sectors, usually some sort of foil.

Because no material is perfect, and it would be almost impossible for any human to cut the sectors exactly the same, there will always be a slightly different static charge somewhere on one of the sectors at least. Atoms are fixed in the structure of the metal, but electrons can be made to move if there is a large enough force acting on them. The static charge is such a force.

Like charges repel, and opposite charges attract. Electrons are negatively charged, while protons within the atom are positively charged. All elements except Hydrogen also have neutrons, which have a neutral charge, but these play no part in the electrostatics of the machine. So, an area of material with an excess of electrons will have a negative charge, while an area with a lack of electrons will have a positive charge.

These charges have 'influence' over one another, also referred to as 'induction' but in a different way to the electromagnetic induction many will be more familiar with. Whenever two sectors are aligned with each other, the charges influence each other. So, if the larger charge on either sector is negative, then the negative charge pushes the negative charge in the other sector, meaning the outer faces of both sectors are negatively charged. The positive charges are on the inner faces, locked into the atomic structure and unable to move.

Most of the time, when this occurs, the sectors go back to the way they were after they pass each other. However, there are four points around the rotation that this is not the case. On each side of the machine, is an angled bar containing brushes which contact the sectors at opposite sides. These are called the 'neutralisers' and are at 90° to each other so that, when viewed through the machine, they form a cross.

When a charged sector touches one brush, the charge is instantly shared with the sector on the opposite edge of the disc, 180° around the circle (not the opposite face, there are sectors on the outer faces only). This means the influence of the charge can move the charges further, charging the other sector as well. The charge vacates the sector where the influence is, and moves all the way to the other side of the circle.

As these sectors rotate, they keep their charge because they have been electrically depleted or filled, rather than just the charge moving around within a sector as before they touched a neutraliser. Because the neutralisers are at 90° to each other, and the discs are rotating in opposite directions, the effect continues, with negative charges flowing through both neutraliser bars to leave one end of the machine positively charged and the other negatively charged.

Now, the charge collectors come into play. On one side, all the sectors are negatively charged. There is a U-shaped charge collector mounted to wrap around the outside of both discs, and it contains 'combs' facing the sectors. These are pointed flat pieces of metal, which facilitate the movement of electrons across an air gap, as long as it is not too big, and the electrostatic voltage is high enough. On the negatively charged side, these combs take the excess

negatively charged electrons away along the conductor attached to the charge collector, neutralising them. At the positively charged end of the machine, electrons come from the charge collector, across the gap, and to the positively charged sectors, neutralising them and allowing the process to begin again.

The charge collectors are connected to a capacitor. In its simplest form, this capacitor is a spark gap with metal spheres where the charge can build. When the voltage is high enough to ionise the air, the electrons jump the gap en masse, creating a spark.

It is common practice to include high-voltage capacitors in Wimshurst machines. These often take the form of Leyden jars, but some makers produce their own high-voltage capacitors via other means, as well. The important point is that the dielectric separating the capacitor electrodes can withstand the voltages involved, and that there can be no arcing or leakage between the plates at the edges.

This rules out most commercial capacitors, and certainly any the average maker has available to them. Commercial versions are hard to find rated above 1500V, and the Wimshurst will operate between 20,000V and 40,000V. We did find a few above a 10,000V rating, but they were expensive. That means $50 to $100 per capacitor.

DESIGN OUTLINE

Our Wimshurst machine will be built with two electric motors driving the flywheels. This eliminates the belt drives, which have served for many years but have some trade-offs. They are simple, but getting the right size and profile of belt can be difficult, even with the online landscape.

You need the right length, diametre, and elasticity. The biggest challenge, however, is that one of the belts has to cross over to make the flywheels spin in opposite directions from a common drive shaft. That crossed belt rubs on itself under enough pressure to cause accelerated wear, depending on the belt material.

With hand-cranked machines this is just an accepted part of the process and most machines don't see enough use for it to be a problem.

Instead, we have two motors, mounted on their own support frames and each with its own hub to connect to the wheels. The hub is 3D printed and has a D-flat section to maintain grip on the shaft without being excessively tight or needing secondary securing like a grub screw or glue. The motors are clamped with saddles which we 3D-printed to give some adjustability.

Besides that, the design is largely the same as the standard model described when explaining the operation. The brushes are made from solder braid washed with solvent to remove the flux, and the combs are made from the pointed metal edging of the cutting blade of the box of aluminium foil!

All of the materials are available at major retail electronics, homewares, or hardware stores, except for the acrylic discs. We sourced these from AT Blanks Australia, who sell a variety of acrylic and timber blanks to the public.

A Wimshurst machine can be built with or without capacitors. Capacitors do help give a more energetic spark, but making capacitors which can withstand the voltages involved without breakdown is hard.

While we will experiment with such things, our first build proceeded with no capacitors besides the one formed by the spark gap. We will explore making capacitors after we complete the main build. Before we do that, however, we have a commercial kit to build that is a bit different.

Commercial Version:

Most commercial Wimshurst machine kits are expensive things meant for physics labs and high school science classrooms. They're generally available from educational suppliers. However, a New Zealand company, SparKit Electrostatics, has produced a more consumer-level kit. It is made from PCB, driven by two small DC motors, and runs from two AA batteries.

There are two large PCB sides, which have conduction paths for both the neutraliser brushes and the collector combs. The brushes are conductive polymer threads crimped into a ferrule, and the combs are angled header trimmed down. They are pre-made, so no manufacture has to take place.

The parts have a high-quality finish to them, with thorough tin plating and a thick black mask which really dresses it up. All hardware is supplied. While the Plasma Channel video of the assembly of this kit, linked below, shows an included baby screwdriver, our kit didn't have this. That video is a few years old now, so things may have changed, or it may be an accidental omission. No matter, it's a trivial thing and most makers will have the right screwdriver. It is for the tiny screws for the motors, but even though they're tiny, the Phillips head is big enough that a #1 standard driver will work.

Also, the instructions mention an included cable tie, which we did not see. There was an isopropyl alcohol pad for cleaning the discs prior to assembly, but it had been punctured by parts in the kit and was bone dry. These things are not big issues. If a maker does not have a liberal supply of isopropyl alcohol and cable ties, are they really a maker?

The instructions for the kit are digital, and a QR code with a link to the PDF comes with the kit. We found the instructions reasonable but they could definitely benefit from some bigger pictures and a few macro close-ups of certain details. Just exercise care and caution during assembly, and do nothing without thinking about it. Also, reading ahead helps, too.

On one side, the switch and battery pack are soldered on, before the combs are soldered to the reverse of both sides. This step needs to be done with care, as the combs need to sit flat. We had to clamp down our PCBs well, and use a third hand tool to hold the combs still. Even then, they can slip around when soldering them. Having them at a slight angle would be ok, but having them not level will be a problem. Make sure all legs touch the PCB flat and do not end up lifted, even if solder bridges the gap.

The motors mount next, and we screwed these in before carefully trimming the leads short and soldering them to their respective pads. The wires are somewhat fragile so shorten and strip carefully, if you choose to do so at all. If they snap or you accidentally cut while stripping the insulation, there is no going back.

Next are the brushes. These slide through holes in the PCB and are soldered from behind. Ours were a neat but not tight fit in the holes and it was a challenge to hold them in place at the correct length, and solder them with nothing moving.

With trial and error, and a few reattempts, we got there. It is important that these protrude far enough to touch the sectors firmly but not so much that the fibres are bent too steeply or worse, that the metal of the ferrule touches the sectors.

Now, it is time to fit the discs. There are plastic hubs included, with a centring lug and adhesive tape fitted. The centring lug takes the form of a wide, short cylinder in the middle, and fits into what looks like a shaft hole in the discs. It's a neat fit, but still, take care with centring when you mount them.

Then, the discs need to be cleaned with isopropyl alcohol before being fitted to the motor shafts. Spray the shafts with isopropyl alcohol to lubricate them. The instructions recommend a little bit of soap but we prefer isopropyl alcohol because it evaporates afterwards even from the shaft (capillary action gets it out) and leaves a firm friction fit after. Push the discs onto the shafts enough to stay put.

Finally, the two halves can be assembled. The bottom bolts are shorter, and use just a domed nut and the white nylon spacers. They are electrically active in the circuit and conduct between the sides to power the motor, so make sure these connections are secure.

The top bolts are 5mm longer and have flat washers and spring washers. These clamp the electrodes, and therefore there must be a spring washer, then a flat washer, the electrode wire loop, another flat washer, then the PCB, spacer, other PCB, final flat washer, and the domed nut. Screw these down firmly, but they will still work loose over time as the electrodes are adjusted and experimented
with.

Finally, fit the feet at the bottom. These are not electrically active but have solder pads as a means of retention. Solder them in place so that the whole assembly stands up on its own and is stable.

How it is time to adjust. The discs and their hubs are unlikely to be in the correct position immediately. In fact, we had to squeeze them a bit to assemble the two halves, which is how we wanted it. They need to be close together, but not so much that they rub. We started by squeezing from the outside of the motors, flexing the sides inwards slightly. This resulted in a tiny gap, and we rotated the discs by hand to make sure there was no contact. However, they were not centred. One side was way too close to the combs for comfort, and the other was too far away.

We squeezed again to firmly embed both sides, making the gap much wider now. Then, using a wooden ruler, each side was gently levered a fraction of a millimetre at a time until the position was correct. Both discs were now quite centred and close enough to the combs to do their job, but not so close as to scrape. The discs have to be as close together as possible, as this determines the influence between sectors on each other. If the gap is too big, the charge does not move effectively.

Now, after a final hand-turned test looking for any scrapes or contact, it was time to turn the machine on. We set the spark gap to about 10mm and pushed the button. Even before the wheels had spun up to speed, sparks were jumping the gap and the smell of ozone was permeating. Ozone is toxic in concentration, and the concentration does not have to be unbearable to be a problem, so we opened some windows and doors, and kept going.

At around 18mm, the gap is no longer producing a spark, but sound can be heard and there is definitely fine arcs happening. They are just not visible sparks. We're happy with around 15mm for now, but adjustments to the machine and even a revisit of the brushes and combs, with their respective soldering, might gain improvements. In addition, the cleanliness of the discs is paramount, and the humidity in the room can drastically affect both the sparks, and the electrostatics of the machine as a whole.

The kit can also be explored in other ways. We bought a Corona Spinner, which is a PCB cut onto a 'galaxy' shape and sat in a pointed support. The high voltage ionises the air and the resulting corona discharges make the device spin.

Also, we had fun attaching thin slivers of foil to the spark gap rods, moved further apart. The foil flies in the air and streams not unlike those long-limbed fan-powered advertising figures while flailing their arms around. It was quite an interesting effect. Fun can be had with other electrostatics as well, such as attaching a small metal bowl and watching bean bag beans fly out.

Our Build:

Parts Required:Jaycar
2 x Geared Head 12V DC motorsYG2732
1 x Roll solder wick-
2 x Acrylic Discs 300mm dia by 2mm thickAT Blanks
Copper-Clad 1.6mm Welding WireHardware Store
Aluminium Insulation TapeHardware Store
6mm Aluminium RodHardware Store
42 x 19mm DAR PineHardware Store
Plywood BaseHardware Store
20mm PVC Conduit or 15mm PVC Pressure PipeHardware Store

** Many of the hardware items are down to build preference, and quantity/length needed may also vary.

Build also requires consumables like hook-up wire and glues, as well as 3D-printed parts

That kit arrived, due to international postal delays, right before we were due to go to print. Our build was a bit different. We have based it on some timber components for ease of access and workability, but timber is not ideal. It can absorb moisture and provide a leakage path. Some form of plastic would have been better, but harder to work with. After construction, we'll probably paint this to seal it.

The first step is the discs. It is not as easy as you might think to find the centre of a circle. It is tempting to draw a line across the centre, using a ruler to find the longest measurement and therefore the diameter. However, it is hard to keep one end of the ruler exactly in place. Even a slight deviation means that the lines will not be exactly across the diameter, but instead be a chord. That tiny gap in the centre matters, because the discs must be exactly centred.

Instead, we will use Euclid Geometry for this process. One of the more common methods uses a compass and ruler only, but we cannot use this method. It is photographed below. This method involves using a ruler to draw a chord (a line segment between two points on a circle) and then a compass to set to the length of the chord.

Two circles (or arcs of circles in approximately the right place and not the rest of the circle) are drawn, one at each end of the line, and a line drawn with a ruler where the circles intersect. This line points directly at the centre of the circle. If this is done twice, the intersection of those lines is the centre of the circle.

However, this method cannot be used on a disc because we cannot draw off the side of it. We're introducing enough error just having the point of the compass trying to work on the edge of the disc, without the error of hoping the disc stays exactly still on a surface we place it on to draw outside of it. Instead, we are using a compass set to less than the length of the chord but more than half the length.

Then, two arcs are drawn, above the chord. A square is then used to make a line through this intersection which is exactly perpendicular to the chord. One thing to note is that the chords cannot be parallel. If they are, there will be no defined intersection, as the bisecting lines will be over the top of each other.

Now, all that remains is to extend the lines so that they intersect, and this is the centre of the circle. This is done with both discs. In the photos, the graphite in the compass made marks too faint for the phone camera to see. The centre was marked with the compass point so it could be found later, and the protective plastic removed.

IMAGE CREDIT: Rudy Helmons researchgate.net

Now, with the centre known, a chord can be drawn that is the exact diameter of the disc. A protractor can now be used to mark lines radiating out from the centre at regular intervals. We chose 15°. These are used later to guide the placement of the sectors of foil when we cut them.

A hole is drilled in the centre, as precisely as possible with as small a drill bit as we had available, to make sure it stays exactly centre. Then, a 4mm hole was drilled for later. Drilling through acrylic must be done with care. Ideally, a drill bit with a negative rake should be used, but most drill bits at hardware stores are positive raked. So, drill slowly and carefully, monitoring for cracking.

Two hubs were 3D printed, being little more than a cone with a D-flat shaft for the motor to mount, and a large flat face to glue on the acrylic discs. We went with glue and not screws to minimise holes in the acrylic and to avoid potential points of damage from screw heads between the discs. We used cyanoacrylate glue, which can be problematic with some plastics and if too much or too little is used, but we fluked it. Apologies for the hundreds of reflections in the acrylic: We couldn't find an angle that didn't show the LED strip yet still left a hand free to hold the phone.

One of the more tedious tasks for this build was cutting the sectors from self-adhesive foil strip. This is the sort of aluminium tape used to seal insulation blankets and metal ducting, but is the smooth metal variety. Some are fibre-reinforced, while others are a cloth tape with a foil coating. We laid it on label paper backing to work with it. It is tempting to use foil sheet from a craft supplier, but this is often plastic with a foil effect, or plastic with a layer of foil of only a few microns. As it is we think this tape was really a mylar film for strength with a thick foil on one side but that should be ok. We need forty-eight of these sectors. They need to have rounded corners because sharp points enable leakage of the charge. Again, the reflective surface makes good photography near-impossible.

After tracing the template forty-eight times, cutting begins. The segments were all cut by hand with scissors. If you own a Cricut, or other form of vinyl/plotter cutter or laser cutter, you're in luck. After cutting them all out, they are aligned into the radiant lines drawn earlier, so that there is the same gap between the sides of each.

Be as consistent as possible with both the cutting and placement, but frustration or boredom will set it. You can tell by looking at them which ones of ours were closer to the first cut, and which were closer to the last. The same goes for placement. After the last one was placed, we gave the whole face a clean with isopropyl alcohol to clean off all the marker and all the fingerprints and contaminants.

The next step is the construction of the frame that will support the whole assembly. Because of the use of saddles to hold the motors, there need be no adjustability in the frame itself. As such, we went with a simple upright supported by two angled gussets for the motor mount on each side, both mounted on a scrap of plywood for a base.

The uprights are 200mm high, allowing clearance from the bottom of the discs to the base to avoid leakage. Along one side of each we added a section of PVC pipe, as the insulation on the motor aires is not enough to prevent charge leakage via them when the sectors pass by, or where the neutraliser or collector paths pass by. We used hot-melt glue throughout because of the expectation of modification or adjustment.

The discs are added to the motor shafts and the motors are clamped to the top of the uprights/ Power wires are soldered on, and run down the conduit. Be sure to get the polarity of the motors correct, or the discs will not spin in opposite directions. The DC motors are not polarised as such, but usually have + marked. They will work either + to + and - to -, or + to - and - to +, as long as both are the same! The saddles allow for adjustment of the motors in and out, to get the gap between the discs as small as possible. The saddles also allow a few degrees of rotation in case the uprights are bowed or twisted.

To make the neutralisers, 6mm aluminium rod was used. This cannot bend easily but we had it in the workshop, so copper-clad welding wire was used to bend to a shape to hold the brushes. These are just lengths of desoldering braid tacked on with a very hot soldering iron, so the solder doesn't wick all the way up. The brushes on their wire holders were cable-tied tightly to the aluminium, which itself is secured with a small p-clamp.

The charge collectors are made in a similar manner, but with combs cut from aluminium flashing. We bought this 0.3mm aluminium foil to make sectors from, but cutting it was too slow and difficult to get done with rounded edges. However, the combs feature points, and as such, tin snips do the job just fine.

This was crimped to the welding wire, which was soldered to hook-up wire. This will leak if allowed, so it was run inside PVC pipe and terminated in an eyelet for connection to the spark gap. We found we needed to add two more timber uprights here at the base to hold the PVC.

Finally, the spark gap. The charge collector combs end in eyelets, inside the PVC. The spark gap is made from two aluminium rods, with one end hammered to a slightly flattened shape so a 3.5mm hole could more easily be drilled into it. The eyelet is secured behind the head of an M3 bolt inside the PVC, with a lock nut on the outside. The bolt is much longer, however, allowing the aluminium to be secured with a wing nut. A couple of flat washers are added for good measure.

The other ends of the spark gap rods are finished with small metal drawer knobs from the hardware store. Round surfaces give more area for the charge to accumulate and less tendency for it to leak. Only when the voltage is high enough will the air ionise and a spark occur. These had to be glued into the end and conductive copper tape used to make the electrical connection.

TESTING

Connecting the motors to a plug pack and powering on, we had some adjusting to do on the discs to set the gap. It immediately became clear that there was some misalignment in the mountings, but it did not appear to be in the way the motors were mounted. It appeared to be about shaft centreing.

Either the hubs are not sitting well on the motor shafts, or the shafts are not straight out of the gearbox. This definitely limits the performance of the machine for now. Having said that, we were greeted with the sound of high-voltage crackling and the smell of ozone, so something is working.

With adjustments, we were able to get sparks across the spark gap but they were not impressive. Using a grounded wire, we sniffed around but there does not appear to be significant leakage from any of the places where wires cross near neutralisers and so forth. The leakage, if it is occuring, is likely from sharp edges in the spark gap assembly, and possibly similar other paths. On top of that, the less than ideal disc spacing is likely problematic.

NEXT STEPS

This whole experiment was an interesting exploration and one we will keep going with. In part 2, we will redesign the disc mounting to be a common axle with better hubs, and belt drives to the motors. We'll stick with separate motors anyway, and use toothed timing belts. We will also work on the comb and neutraliser bars, hoping to find a decent source of smaller copper or brass tube that will bend, and material to use as brushes.

Also in part 2, we will cover home-made capacitors, both some attempts with materials to make a rolled plate capacitor, and some Leyden jars. These will significantly improve the brightness (but not length) of the sparks and are almost a project on their own. They will have applications in some other future fun.