3D Printed Wimshurst Machine


Steampunk extraordinaire [Jake von Slatt] has released his latest creation. This time he’s built a Wimshurst machine from mostly 3D printed parts. The Wimshurst machine is an electrostatic generator and was originally invented in the late 1800’s by James Wimshurst. It uses two counter-rotating disks to generate an electrostatic charge which is then stored in two Leyden jars. These jars are also connected to a spark gap. When the voltage raises high enough, the jars can discharge all at once by flashing a spark across the gap.

[Jake’s] machine has a sort of Gothic theme to it. He designed the parts using Autodesk’s 123D Design. They were initially printed in PLA. Skate bearings were used in the center of the disks to ensure a smooth rotation. The axle was made from the fiberglass shaft of a driveway reflector. The vertical supports were attached the base with machine screws.

The Leyden jars were made from sections of clear plastic tube. The caps for the jars were 3D printed and are designed to accept a short length of threaded 1/8″ pipe. Copper wire was used for the interior contacts and are held in place with electrical tape. The metal sectors on each disk were made from pieces of cut aluminum tape.

You may be wondering how this machine works if it’s almost entirely made out of plastic. [Jake] actually painted most of the parts with a carbon paint. This makes them electrically conductive and he can then use the parts to complete electrical circuits. Unfortunately he found this to be rather ineffective. The machine does work, but it only produces sparks up to 1/2″ in length. For comparison, his other machine is capable of 6″ sparks using similar sized Leyden jars.

[Jake] actually tried rebuilding this project using ABS, thinking that the PLA may have been collecting moisture from his breath, but the result is still only 1/2″ sparks. He suspects that the bumpy surface of the plastic parts may be causing the charge to slowly leak away, preventing a nice build up. He’s released all of his designs on Thingiverse in case any other hackers want to give it a whirl.

23 thoughts on “3D Printed Wimshurst Machine

  1. It looks gorgeous!

    But, can we come up with a different way of saying? Let’s stop saying “3D printed xxx” we we actually mean “I made this thing that has some plastic parts among many other things and those plastic parts are 3D printed”

    1. Does that really seem like a better way of saying it? I have a feeling most of the community understands “3d-printed xxx” to refer to an object made largely from 3d-printed parts.

      But I’f you’d rather type 20 words than 3 to convey ultimately the same message, have at’er.

      1. You, I and a few other people understand. But most of general public assume that a 3d printed something is entirely made on that printer.
        I heard this so many times: if now we can print a phone case, in a few years we should be able to print a whole phone.

        1. I’m with you Bogdan, while I too understand that often “3D Printed” means some of the components were 3D printed and everything was then hand assembled using other, more mundanely manufactured parts, but it is misleading. If the whole thing was 3D printed with the exception of wiring and metal parts, it would be a more accurate title.

    2. How about “Partially 3D Printed” ?

      And we can even get rid of the 3d, because that should be apparent from context: “Partially Printed XYZ”.

      Well how do we define printed? Is it FDM, powder and binder, cured resin? “Partially FDM’d XYZ”

      And isn’t the scale of the machine important too? i.e. a device printed on an $200k industrial 3d printer is less exciting to the DIY community than a device printed on a cheap hobby machine: “Partially $300-FDM-machine-fabricated XYZ”

      But then I suppose it’s a bit unfair to penalise those who spend a little extra up front to save themselves a bunch of tinkering time to get the machine up and running. After all, that’s a reasonable thing to do. So for fairness, all maintenance time should be disclosed too: “Partially $300-86Hr-FDM-machine-fabricated XYZ”

      Are we getting warmer?

  2. With Wimshursts there are several key bits to get long sparks. I think the porosity and therefore large surface area in the disks and insulators is a big issue (surface will hold conductive moisture or worse enable an ionised path along the plastic surface under the voltages present). Perhaps a bigger issue is the geometry though. I presume, built as one part, the discs are smaller than on your old Wimshurst? this size is what gives spark length. Leyden jars just determine spark brightness and are optional if you don’t mind looking carefully for a spark. Also your arrangement of leydon jars in front of the discs leaves them very close to the discs causing the air losses to be higher and possibly affecting the influence effect as the plates pass each other.
    I doubt surface roughness is your problem; air in the valley of a rough surface will be at the electrostatic potential of the ‘mountains’ around it, electrostatics works almost with the convex hull of your part unless you bring parts very close.
    To disprove surface loss try running your old machine with a conductive 3d printed part attached and in free air. If it still sparks it isn’t surface roughness.

    1. I have a book that discusses the engineering considerations in making one of these – I’d post the title but I can’t find it ATM. “The Amateur Scientist” (C.L. Strong) also discusses these, but it’s not the book I’m referring to.

      I was surprised at some of the [book] author’s conclusions. For example, he discovered that electrode holders made of wood worked better than non-conducting plastic, due to wood’s dielectric properties. Also there’s some black magic involved in setting the cross-arm angle for maximum voltage.

      From the build blog and my memory of the book, I would suspect the conductive coating. By spraying the plastic with conductive coating, you are making conductive any imperfection in the surface of the 3-d printed piece. Any “point” sticking up from the surface will leak electrons to the air and reduce your voltage.

      As an experiment, try replacing one set of sprayed 3-d printed parts with a non-sprayed non-conducting part with embedded wire for the conduction path, and see if that increases the voltage. Wood will probably work well.

      Specifically, make the comb holders non-conductive but with a buried wire for the conductive path, the leyden lids non-conductive but with a copper pipe going through the middle, and the crossbars out of conductive tubing. Sand down any sharp corners or other points in the conductive path; for example, make the cut ends of your nails “rounded”. Also, grind the tips of the nails to a sharp point, and make sure they’re mounted so that the tips are all the same distance from the wheel.

    2. Well, here’s something I know a little about. I built a 48″ (121cm) diameter Wimshurst that stood about 12′ (350cm) tall.

      Simplified, think of Wimshurst machines being a tiny capacitor that is being slowly charged to an infinite voltage. The leads of the capacitor are the two pickup combs on the left and right sides (not the diagonals, those are the neutralizing bars).

      Eventually, the voltage will climb high enough that the spark will jump the gap. To give it a boost and so that it’s visible when it does jump, you create the Leyden jars which are just big capacitors to increase the natural capacitance of the machine by, oh, 2000%.

      What you want as a Wimshurst designer is to have the spark jump a big gap in the place you want it to jump (between the output electrode poles, which are connected to the combs). Due to other essential parts on the machine that are dead shorts, typically the max spark size is 1/3 the disc diameter *if you design everything else well*. Everything to do with making a good machine is only about making sure that nothing else is a preferred place to spit electricity.

      The big kick in the pants is that at these voltages anything pointy will spew electrons out into the air. At these voltages, “pointy” means “anything other than a gracefully-curving, polished rod or sphere.”

      Think of it like filling a water tower (capacitor) with some holes in it all the way up, and filling it with pipes (conductors) made out of soaker hoses that are leaking everywhere themselves. The higher the voltage, the harder it is to get even higher voltage because the more holes the water is pouring out of.

      Every single conductive thing on the machine (at these voltages, wood is a conductor) needs to be perfectly smooth and graceful or leakage will prevent charge from ever accumulating in the Leyden jars.

      The metal (carbon) “sectors” should be polished metal. Even if you used, say, aluminum ducting tape (common) and got all the crinkles out, the edge of the tape counts as something pointy and electricity will fly out the sides, leaking away. So, carbon sprayed onto bumpy plastic surfaces is probably the worst possible design choice, the surface area is basically one big sheet of “pointy”.

      The next most important thing a Wimshurst needs (though less so than similar machine without the metal sectors) is a perfectly clean and dry disc surface. Machine hasn’t been dusted in the last day or two? Day is slightly humid? Don’t bother, it won’t work. This is for 2 reasons: 1 – Chances are anything “dirty” is also conductive, and 2 – In terms of electrostatics, dirtyiness prevents the magical infinite charge difference that powers the machine.

      Again, for both these reasons, sprayed carbon is the worst possible construction method I can think of. As soon as that wheel spins around once, it’s dragged carbon particles in a circle across the whole disc, creating a dead short.

      You can actually identify where your leakage is going by putting the machine in a perfectly dark room, waiting about 10 minutes for your eyes to adjust, and then cranking. Everything that’s leaking will glow blue or purple or, in worse cases, have actual streamers just jetting electricity out into the air.

      Big fan of Jake von Slatt. Just identifying (probably) why the machine is so disappointing in output. Electrostatics isn’t something that cooperates with wide tolerances.

      1. I’m not entirely convinced that’s the root of the problems. With lots of leaks like that you should be able to overcome it to an extent by turning faster. This will also increase the corona on the culprit. I presume ‘carbon paint’ will have extraordinarily small particle size to the point that they won’t be a significant factor. IIRC people have used spray on galvanizing paint (zinc coating) that has fairly large particle sizes for other electrostatics with good results.

        A few layers of shellac over all the conductors should help suppress any corona regardless of the source, should it not?

        1. I’m, oh, 80% confident that it’s the root of the problem.

          Turning faster gets you better to a point, but, not all the way. For example, on the giant one I built, I was getting 5″ sparks by turning it slowly by hand. I walked away, came back 2 weeks later, and I could not get it to spark at all. I took off the crank, replaced it with a drill, and had to crank it to about 10x the speed to get the same sparks I used to. In daylight, or even room light, the corona isn’t visible, only streamers are.

          Also, the room didn’t smell of ozone like it used to, which it would, if all the losses were corona. So, dirtiness plays a huge factor in even generating the static field to begin with.

          Carbon paint might be okay, maybe, just as a sector material. Carbon paint over a smooth surface might work, but it’s pretty much dead in the water onto a 3d printed surface. You’d never know, because the wipe-off of carbon dragging all the way around the wheel will kill it for sure.

          Shellac helps, a bit, but not much. You have to remember that this is a machine that generates somewhere between tens and hundreds of thousands of volts and causes electricity to fly through air. Air is a pretty good insulator, so even covering the electrodes in shellac is only a mild (significant, but not overly so) improvement over air itself. It can and will still puncture easily. I guess you could compare the breakdown voltage of air to shellac and see how much of an improvement you’d get. My guess is at most 2-3x for a given thickness, and thickness is only as thick as paint before you’re air again anyway.

          1. On this Wimshurst the sectors are aluminium tape so smearing of carbon won’t directly be an issue. But having said that you have raised a couple of points I hadn’t thought of:
            Any loose carbon dust from the painted surfaces is likely to spell disaster. While there will be a natural process of tracking on all the plastic insulators where the field ionises the plastic until it is strongly hydrophilic and becomes conductive from all but the lowest humidities, this tracking effect will be made much worse if it can attract conductive particles. The problem is as soon as there is some surface conductivity the field becomes sharper. The air and plastic ionise at the sharpest tip and the track creeps forward away from its electrode and towards the opposite one. Once this process begins, any loose poweder will be attracted to the tip of this track, both accelerating the process and increasing the conductivity. On my 50kV Kelvins thunderstorm (a perverse machine that uses dribbles of water to generate sparks) you can see the water spray get attracted and rapidly wet a path from one side to the other. Carbon would be similar. (Also on the KT: as your face is earthed it is the other voltage from the drops point of view so it spray paints you if you get too close!)
            The other point is the sectors. Alu tape sectors are normally ok (not the best but a lot less effort) because the dielectric they are stuck down on polarises as they charge making the apparent charged volume thicker. Say a sector is positive, the dielectric will polarise with the opposite side of the disk positive and a rounded field around the edges leaving the sector looking a lot blunter. Normally the issue comes when the edge separates slightly (often due to the brushes rubbing it). In this case however the non-100% print density means the bluntening isn’t happening as much regardless of how well it is stuck down. The field is left sharper causing more ionisation.
            As long as any lumps are much smaller than the diameters of the conductors the surface finish of conductive parts will be a much less significant effect except where they get close to oppositely charged parts because the field will only see the convex hull of the shape. The same reason a faraday cage can have big holes in if you aren’t close to the electrodes, or a cage style toploader works. You want the air near any tip to see conductor for as close to 180degrees of that side of it to avoid effective pointyness. A tiny lump on a nice large radius will achieve that just fine

        2. I’m intrigued by this entire conversation.

          We have an engineering problem and several suggested resolutions.

          My normal recourse would be to experiment and see what fixes the problem. I don’t have access to a Wimshurst machine, but Jake certainly does.

          I’d be very interested if Jake could try some things one-at-a-time and summarize on what effects the output and by how much. It seems like he has to change things anyway, and a test/comparison plan seems like the way to go.

          Specific measurements tabulated in a before/after method would be ideal. It doesn’t need to be particularly onerous or even accurate – “the longest spark I could get while turning at 2 RPM by hand” would be adequate.

          (And this is what actual researchers from the steampunk age would do. If Jake want’s to really be steampunk…)

          1. Oh I do plan to circle back and figure this out, but I’m a little bored with it right now and playing with other things. BTW: my other Wimshurst was really robust and it has worked reliably when completely dusty and at Maker Faire in 2009 when you could smell the salt in the air from the bay and all of the brass tarnished dark over the course of the day due to the salt air and ozone. I’m betting the issue is the disks and that replacing them with sheet acrylic is going to be the special sauce.

  3. This brings back memories from my High-school years. I tried to emulate an experiment I read in a Scientific American’s “The Amateur Scientist” article about a spark-gap pumped nitrogen laser. Imagine the fun of a 6000-volt spark-ignition transformer being fed into a voltage-doubler arrangement into a pair of aluminium-foil duct-taped to acetate sheets capacitors.

    This was my one and only time that mom could have come home to her son dead with smoke rolling out my ears. Damned thing leaked corona from almost every corner.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.