Marx Generator

The simulation of lightning

A Marx Generator is a device commonly utilized in museums and labs to simulate and demonstrate the effect of lightning.

Examining the effect of lightning can be useful, not only to advance the understanding of lightning, but also to physically test electronic components such as insulators, high voltage capacitors etc etc.

Typical Marx Generator schematic
(Courtesy of Mike's Electric Stuff)


A gigantic Marx Generator tower in Russia
(Image courtesy of unknown author - 'master')

A Marx Generator basically consists of a very clever way of charging and discharging capacitors such that the output voltage is a multiple of the input voltage.

A typical Marx Generator schematic is shown to the left. It isn't hard to see that the input supply (left-most terminals) charges up the capacitors in parallel. After some time, the capacitors are all charged to the supply voltage.

The terminals in between each 'stage' is a spark gap. The spark gaps are set to a distance so that they will 'fire', or conduct, as soon as the capacitors are charged to supply voltage.

Because of the way they are arranged, when the spark gaps fire the capacitors are essentially discharged in series. Thus, the output voltage is (theoretically) a multiple of the input voltage equal to the number of capacitor stages in the Marx.

I decided to make a small scale Marx Generator. For this, I ordered 20 pieces of 1nf, 20kV rated capacitors from eBay, and they arrived in perfect condition. Given their construction, however, it appeared they probably wouldn't stand 20kV before arcing between the leads, so I knew there was preparation work to do.

Grabbing some oxygen free copper cable, I removed the copper, and threaded the capacitor leads through. A perfect fit, and would definitely insure me against spark overs.

I went for a simple approach. All parts were crudely soldered onto each other on top of a fibre ceramic board, a good insulator. Each of the 1M resistors was made out of two 470K, 1W resistors in series. Close enough.

Spark gaps were constructed, somewhat resourcefully, using the long leads of the resistors. The tips were carefully bent into U shapes to minimize coronal loss.

And here is a test run with 6 stages and the spark gaps set very conservatively. So far, so good.

The high voltage charging supply which I used was based on a flyback transformer using the same driver as the ignition coil driver. This flyback transformer was salvaged from an old computer monitor, but you can find them in most old CRT TVs also.

The output voltage of this supply is approximately 10kV. I increased this later.

After a while, I ran out of resistors. Which is sad, because I only got up to 10 stages.

Anyway 10 stages should theoretically give me an output voltage of 100kV with a 10kV supply. Not too bad. Lets put that to the test...

Here we go! Ruler indicates ~9cm spark length. Which means approximately 99kV. Pretty close.

This really surprised me as it meant that I was incurring very little loss in the circuit. However, in hindsight it occured to me that this these were probably surface tracking sparks given the proximity to the benchtop.

The Marx fires about twice a second. Oh, here's also a short video to give you an idea of what happens...


Output sparks in the air... This was a 2.5 second exposure and there are 5 full length sparks visible, indicating that the firing rate is indeed approximately 2 times per second.

I just could not resist. Mr. Penguin is fortunately made out of plastic, although you can see that one of the sparks has leaked a bit of corona just under his beak... The sparks tend to snake around the penguin... again, surface tracking.

Dull sparks, but I really like this one. A perfect demonstration that plastic/acrylic sheeting or other common household insulators are completely useless at such high voltages.

The spark snakes around a bit then goes straight through the plastic film as it if wasn't there. I was really expecting a few surface tracking sparks but this did not occur.

Here, we have a better demonstration of surface tracking using a thicker piece of perspex.

Got the new resistors, and added ten extra stages. The Marx Generator now consists of 20 stages. The hissing noise of the corona is very audible, and the spark length much less than the theoretical.

The extra stages add on an extra 4cm to the previous spark length. A maximum of 14cm spark length now indicates an approximate output voltage of 150kV.

The effect of surface tracking is again an issue in hindsight here. Realistically the voltage is probably less than 150kV.


Polystyrene balls are great indicators of static electricity, and are commonly seen used in 'rub-the-plastic-ruler-with-wool-and-pick-up-stuff' experiments. Load the video to see what happens when the marx generator charges up a whole container of these!

Here we have a bit of fun with the Marx.

I've always wondered why on earth people think that the rubber tires of a car is what saves them from being stricken by lightning.

Because they're wrong! Given that the lightning bolt has travelled more than a kilometer through the sky, is one inch of rubber going to do much? Lightning cannot strike you in your car because the metal chassis of the car acts as a Faraday cage - the charge flows on the outside of the metal rather than the inside, which is where you are!

Old-school car getting struck

Eiffel Tower having a bad day

Old-school iPod having an even worse day

To everyone who hates ipods! This one was kindly donated to me by Charles. It was already broken through, had a cracked screen, and I had already salvaged the hard drive off it earlier. For the relief of those iPod lovers, you may remember the laws of the Faraday cage!

This setup shows how the Marx output wires are connected during the lightning simulations.

The yellow wire is at high voltage, whilst the white one just visible down the bottom is the ground wire, which in this case is attached to the iPod metal casing.




Back to Penguin's Lab


© Penguin's Lab 2012