3D Printed Van de Graaff Generator.

"Artistic" impression of my 3D printed Van de Graaff generator.

My small 3D printed Van de Graaff generator.

How does a Van de Graaff generator work?:

I will not explain how a Van de Graaff generator works in this blog entry.

There are thousands of pages on the web about that. Use your favourite search engine for it.

I will, here and there, pop in some technical stuff if it can help explain things else I will rely on a few images.

One point I will highlight here though.

Don't confuse a Van de Graaff generator with a Tesla Coil. It is two quite different beasts. A Tesla Coil works with high frequency, quite often using powerful high voltage neon transformers. A Tesla Coil will, if it works good, deliver continuous sparks that probably will destroy any electronic equipment in the area.

A Van de Graaff generator is a much more "gentle" beast. It builds up the charge, assisted by the rubber belt and charge supply. That means the sparks only fly when enough charge has been build up. The faster the belt runs the faster the charge can be build up.

As this is 3D printed plastic parts you should not go over board when choosing your motor and drive wheels.

Warning:

This Van de Graaff generator uses a very high voltage non insulated Ion generator for the charger source. ~5kV

Some parts of the Ion generator are referenced to 240V~ so no touching those parts.

The top sphere is well insulated from the 240V~ so no electrocution risk - but it will still give you quite a nice zap if you are in to that - If you are reading this post you probably are... I have achieved sparks up to 40 mm (1.5 inch) Looks great in a darkened room - not as in the image above, wouldn't that be great.

Build at own risk and only if you know what you are doing.

3D Printing:

I will not explain a lot about 3D printing in general.

Firstly - I'm self still quite a novice when it comes to 3D printing.

Secondly - Every 3D printer is different, so what might work on my printer, might not work for you. Thirdly - There are loads of other pages and videos on the web with tips, tricks and tutorials.

Instead you can download the free OpenScad programme and all the source files from my Thingiverse page. That means that you can change the design after your own wishes. With OpenScad you can create your own .stl files.

So now we got that out of the way...

I have always wanted to own a Van de Graaff generator. We had a fairy large one at the school I went to and fortunately I was sometimes allowed to "play with it" after school hours.

To build my own version have been on my mind more times than I can count - but it have newer come to more than a bunch of drawings and some small experiments.

The biggest problem building a Van de Graaff generator is the metal top sphere. Without any proper tools it is close to impossible - until I got myself a 3D printer. I thought so at least.

I have actually had my 3D printer for quite a few years but I have never been able to get any good results with neither it or the software that came with it. Only the "demo" objects that came with the printer came out perfect. What ever I tried to create and print, came out as - sorry for the expression - bird shit.

I tried every possible setting, both recommended and experimental. Nothing worked.

So the printer stayed in a corner of my flat with a cover over it for a year or two. Then some time ago I wanted to give the printer an extra go. I knew that it could print properly, the demo objects, so it must be a matter of the self generated files. I decided to do some proper testing and experimenting. I ended up writing a C# programme that translated all G-Code parameter to "proper" values. Very experimental but I was finally able to get some quite acceptable results. Unfortunately, after finally being able to start on a few of the projects that had been waiting in the background, I realised that the printer had a mechanical misalignment. One that I could not fix - and as the warranty on the printer in the meantime had expired, I had to find another solution.

Fortunately the misalignment was only on the X axis - so I added an "anti-misalignment" procedure to my C# programme.

Now finally I could get started on my long waiting childhood dream project. The Van de Graaff generator.

Well - so I thought. There was still quite a few issues to overcome. A few smaller;

The size of my 3D printer (150 * 150 * 150 mm).

The design of the rollers - one metallic and one non conductive.

The rubber belt that carries the high voltage charge to the sphere.

How to drive the generator. Belt driven or with gears - Motor or hand driven.

What motor to use if any.

but the biggest challenge was still that metal sphere.

Other stuff that needed some brain activity;

Power supply for motor.

Power supply for charge injection.

Safety!

and lots of stuff I didn't think about until I ran into whatever issue it was.

And one important thing - I didn't want to spend excessive amounts of money on this project.

A few things I had to get hold of, but that was mainly stuff I would normally have in stock anyway, or at least would be handy to have in stock.

Most non 3D printed parts though, are stuff I had laying around - so recycling at its best.

Size matters

The issue with the size of my 3D printer, I solved by thinking back to my birth country - Denmark- and my childhood favourite toy - LEGO.

I grew up just half an hours drive from Billund but have actually only been in LEGO Land 3 or 4 times. I didn't solved the size issue by printing LEGO blocks but by printing smaller parts that could either be screwed or clicked (the LEGO inspiration) together. I ended up using a mix of small bolts and some clickable parts.

How high can we go.

I ended up splitting the frame of the Van de Graaff generator in tree parts. With that I could get a reasonable height without too many joints and still create a sturdy frame. I ended up with a total height of just around 30 cm (12 inch) so each part ~10 cm (~4 inch)

The frame consists of 4 different parts.

The frame parts

The base

I initially wanted to 3D print the base for the generator but here I ran into real problems I have not been able to fix yet.

Big flat surfaces are very difficult on the best printers, so on my cheap printer it is close to impossible. Therefore the base is, at the time of writing, made from some MDF I had laying around.

I might go for the 3D printed version later, when I get my head around my printers "mental issues".

The bottom of the base. (Not used in my build)

The bottom of the base. Mainly hollow to save PLA and raft material.

The printer currently gets "stuck" half way through the first non raft layer.

Currently not used in my build.

And we are rolling

The design of the rollers was not that big a challenge, I knew from my mechanical experience as a child (I grew up taking everything apart and putting it all back together. Not always successfully, so I think I drove my brothers mad now and then as it was their stuff I took apart.) that the rollers should be thicker in the middle else the rubber belt would not stay in place. For simplicity my first design was just two cones put back to back. That actually worked great. I did decide to base the rollers on a "stretched" sphere instead though for a few reasons:

Firstly, I printed the first set of rollers before I had noticed and fixed my printers misalignment issue, so there was some rattle when testing the generator. That's easy to hear in the video clip at the end of this post.

Secondly, I found the rollers based on a sphere more ecstatically pleasing.

Thirdly, During all the designing and test printing, I had skipped the original slicer software that came with the printer and switched to Cura, that delivers a much better result. Using Cura I don't need to auto correct all my G-Codes. I have now re-written my C# programme to only fix the misalignment issue. As I was fixing the programme anyway I changed it to a process that automatically fixes any *.gcode files generated from Cura. That way I don't need to open the file, convert it and then save it. Saves quite a few mouse exercises. The biggest and probably most important difference between the original software and Cura - it was close to impossible to detach my models from the raft. It left a very rough and un-attractive finish. With Cura my model pops off as if it was a Post-It stamp.

Left roller based on a mirrored cone. Right based on a stretched sphere.

The rollers are mounted to the frame in small ball bearings. One of the things I really like about 3D printing is the precision. The ball bearings I'm using have an outer diameter of 15 mm, thickness of 5 mm and the hole diameter is 5 mm. I did a test print with a 15 mm hole and the ball bearing slides in with a firm press fit. Such a great feeling when stuff fits like that.

The shafts for the rollers and ball bearings was a little challenge in it self. The printed parts are not very strong, specially small parts. To strengthen the shafts, I left them with a 2.5 mm hole for a small screw.

Don't forget the rubber

The rubber belt is quite an important part of the Van de Graaff generator. It is the rubber belt that carries the charge from the bottom roller comb to the top roller comb and sphere. The rubber belt actually doesn't have to be made from rubber. As long as it is non conductive and flexible, it will do its work. Different materials does give very different results though, so a bit of experimenting was needed. I ended up using a stripe of waxed table cloth. This had some advantages; It is non-conductive, it is very flexible and durable, it is easy to cut and I could just sow it together (on my very old black sowing machine) after properly finding the proper length. Furthermore, because the back side is fabric I get a very good grip on the rollers and a very smooth top surface for the charge and discharge comb.

Driving the 3D printed Van de Graff generator.

The generator we had in school was hand driven using a handle and some gears. That it was hand driven made it quite exiting as you had to stand close the generator when using it. The Van de Graaff generator we had was, if I remember right, about 40 - 50 cm tall (16 - 20 inch) and the sphere around 20 cm (~8 inch) these measurements must be taken with a pinch of salt as I was quite a bit smaller when I went to school.

What I do remember was that it took quite some strength to spin the handle on generator. It was, as mentioned, a geared drive, and the more charge the sphere became the harder it was to turn the handle.

My generator is quite a bit smaller than the "professional" ditto we had at school.

30 cm tall (12 inch) and the sphere is 13 cm (~5 inch) in diameter. I'm to lazy now a days to turn a handle - so I wanted my generator motor driven.

I have been experimenting with 3D printed gears. The results have been quite impressive - as long as you don't need high precision, high speed or high torque. In the Van de Graaff generator you need all tree, therefore I decided to go for the belt driven approach. That also made it much easier to adjust the gearing, more or less on the fly (I can just print a new wheel in ~20 minutes) thereby also adjust the speed and torque.

I had a rather large 12 Volt DC motor "in stock", can't remember from where - I think an old HP Ink-Jet printer. It seemed to be strong enough according to my initial experiments also it has a quite thick shaft (3 mm) with some good grip markers. By that, the motor became the part the rest of the Van de Graaff generator would be designed around - more or less.

Model of the motor

The Top Metal Sphere

Finally the biggest challenge - with regards to 3D printing. I still had quite a few unforeseen challenges to overcome - more about that later.

Ideally the top metal sphere should be a perfect smooth hollow sphere. That is so easy to imagine in your mind, so easy to draw in nearly any drawing programme, based on a very simple formulae.

In the real world though - my small flat in the middle of Amsterdam - creating a perfect metallic sphere is a totally different story.

First problem: Every one who has ever been building a Van de Graff generator, have been faced with this problem. From the first very old antique versions to the most modern ones. They all need a hole in the perfect their sphere. Else you will never get the charge in there. And that's the hole point of the Van de Graaff generator.

Next problem: Any sharp points will attract the charge and cause the charge to "spray" away. Smoothness is a must for a good result.

The best compromise for a perfect sphere with a hole, is merging the sphere with an other, in my opinion, perfect object - a doughnut.

A "X-Ray" view of the top sphere

This design is the one I remember from my school days, and the design you see in more or less any (semi) professional Van de Graaff generator. In short this design makes sure any "spraying" of charge stays inside the sphere, no sharp points on the outside - in an ideal world. The smoother and larger the sphere - the larger charge it can hold - the longer and stronger sparks.

It is all coming together.

By now I at least had an idea how to get it all together. I'm still only talking about 3D printed parts. All plastic. Not a lot of very High-Voltage in that.

Fortunately only very few parts need to be conductive, actually the less metal the better. Only the top roller and top sphere needs to be conductive. Apart from that only the corona combs and connecting wires need to be conductive.

How the get those two plastic parts conductive? Searching the internet gives some options. Electroplating was on the top of my list for a while. There are some quite convincing examples on the web. But generally always small objects, and a lot of mess with (homemade) chemicals.

As a child I loved playing with chemicals. Made lot of interesting (and sometimes quite smelly) stuff you wouldn't be allowed to do, legally, today. Now a days I find chemistry interesting on a theoretical level - practically it's not for me.

I went in a different direction. I often build my DIY electronic stuff in to wooden (MDF of multiplex) cases. To create an electric shield I have often used graphite spray. The spray creates a matt conductive surface, with a resistance of couple of kOhms per cm/inch. Good enough for shielding electronic stuff. Also good enough for a Van de Graaff generator? In your dreams.

Next idea. I had a big role of self adhesive aluminium tape. What if I covered the inside and outside of the sphere with segments of that. It would create a nearly perfect surface, of cause creating a sphere from flat segments is impossible - but if the segments are small enough we are getting close enough. I have seen working Van de Graff made from Cola cans. So a few bumps could be allowed.

Well though the aluminium tape is conductive, the self adhesive glue isn't. Back to the drawing board...

Late night, turning in bed, a bright moment. Combine aluminium tape with graphite spray. Wouldn't get a shiny metallic sphere, but it might work.

Next day I read through the data sheet of the graphite spray i was using. One thing that stood out was, if the object coated with graphite spray is polished, with a soft cloth, you not only get a shiny surface, as a free gift, the resistance goes down by quite a bit.

Before applying the aluminium taps I polished the 3D printed sphere to the smoothest surface I could manage. I first used a 120 grain sandpaper, followed by a 800 grain polishing, finishing off with 1200 grain paper.

Making sure the sphere was dust and fat free, I started to apply the segments of aluminium tape. Inside first. The surface of the inside is not that critical, so it was a good for testing. After finishing the inner surface I tried measuring the resistance between two segments and, as expected, no conduction. After that I applied a coating of the graphite spray. Let it dry night over, then gave it a gentle polish with some kitchen paper. Now the resistance between any two points was always below 200 Ohms.

I repeated everything for the outside. Here I was a little more careful with the placing of the segments. After the graphite coating the resistance between inside and outside of the sphere stays at around 200 Ohms.

I finally had overcame the biggest challenge - the top metal sphere. Success? Nahh - there was still a few obstacles before I could drag the first sparks.

Left Aluminium tape only, Right polished graphite spray.

Left Aluminium tape only, Right coated in graphite spray and polished.

At least I now had all the major parts of the generator on the drawing board (and quite a few bits and bobs in the recycling bin)

Before finishing up the 3D printed parts, as mentioned earlier in this post, the upper roller also need a metallic surface, not a must, but a big performance booster. Here I just covered the roller in the self adhesive copper tape I also used for the corona combs.

The next few parts I needed was the charge transfer combs. The bottom comb is needed for transferring the charge from the high voltage source to the belt by corona discharge. The top comb transfers the charge from the belt to the top sphere, again by corona discharge.

For both combs I used self adhesive copper tape. I could have used the cheaper aluminium version, but I had copper tape in stock and big advantage - you can solder wires to it.

To create the comb teeth, I just cut, surprise, teeth in the copper tape. A little fiddly, fortunately the comb does not need perfect contact with the rubber belt as the charge is transferred by corona discharge.

Close up of the charger comb.
Here temporary covered as it is referenced to the 240 line voltage.

The Van de Graaff generator before I coated the top sphere with graphite spray.

The finished build - nearly.

The box under the Van de Graaff generator is a negative Ion generator based on a 11 stage voltage multiplier. It delivers around -5kV to the lower corona comb.

The Ion generator is referenced to the 240V~ net and though the output have a couple of 10 MOhm resistors in series it should not be toughed. It still really hurts.

A closer look at the top sphere and the connection for the top corona comb.

The sphere is covered inside and outside with self adhesive aluminium tape. Above that a coating of graphite spray.

The wire is fastened by fanning out a multi core wire and gluing it to the inside of the sphere, fixing it in place with a squirt of hot-glue.

Model as rendered in OpenScad.

Model as rendered in OpenScad - Exploded view.

Close-up of the base assembly. The 12V= motor is powered by two small switch mode power supplies I had laying around. The charge is transferred to the lower corona comb by the yellow wire,

And we do need a few sparks.

Hello from sparky. A single spark from the video below.

A short Y-Tube video drawing a few sparks.

This is before I fixed the misaligned rollers so the Van de Graaff generator is quite noisy here.

Watch on Y-Tube

Programmes used:

OpenScad: I heard about OpenScad on BigClives Y-Tube channel. I wanted to give it a try. I fell in love immediately. Everything you build is based on a few simple basic shapes; spheres, cubes, cylinders. You write your model in a language that reminds me a little bit about a "real" programming language. You can use for loops, if statements, there are loads of mathematical and geometrical functions.

All the OpenScad source files for the Van de Graaff generator can be found on my Thingiverse page.

Here you can see all the 3D printed parts.