Youtube channel

Check out my youtube channel!

Wednesday 31 August 2011

Marx 2 - A Bigger Bang

Last year whilst in America for work, I visited a flea market and came across a box of impressive capacitors. Inside the box were ten 8-inch long glass-encapsulated capacitors, each of them .01uF at 14kV. Considering the size of them, they were obviously rated for heavy discharge use. A quick look at the manufacturer website confirmed that they're indeed intended for pulse applications.

At $50 for the lot of 10, it was a bargain.

I knew what I wanted to do with them, but I didn't have the right parts to hand, so they sat unused for a year or so. When I managed to score 5000 VR37 resistors, I knew I was ready to use them. Naturally, resistors and HV capacitors can only mean one thing - Marx Generator!

With my previous Marx putting out about one Joule of energy per spark, I calculated this one would go an order of magnitude higher - 10 Joules!

Bipolar Marx
For this Marx, I decided to run it in bipolar configuration. It's very similar to a regular Marx circuit, except the "low" voltage is fed into it from the middle of the bank. For a bank of 10 caps, unipolar normally gives 10* input at the top, and zero volts at the bottom. With bipolar, you get 5*Input at one side, and -5*Input on the other side.
This still gives a voltage of 10*Input between the outputs, but only 5*Input difference to ground, reducing corona losses, and thus increasing efficiency.

The Marx is constructed as two 5-cap banks. Each is mounted on a Perspex sheet with good old cable ties, and the sheets are screwed into a frame made from bits of 15mm PVC conduit.

Metal things near a Marx are a bad idea, and must be avoided unless it's supposed to carry current. Mine contains no metal at all in the frame - it's held together entirely by glue and nylon bolts. I tapped the bolt holes into the conduit to avoid having to manipulate nuts inside it.

Spark gaps
For initial testing, I made spark gaps from bits of hookup wire looped over and hot glued into place. With testing resulting in ringing ears (ie success!), I went for the next level. Wire spark gaps are very flimsy, fragile, lossy, and tend to drift over time.
I came across some tiny brass dollhouse doorknobs online which were perfect! 5mm diameter and solid brass, and less than $2 a pop, so got a load of them.
I mounted them into a length of conduit, sawed in half lengthwise. Careful measurements got me with exactly the right mounting hole distances, and they all fired at pretty much spot on 14kV. Some HV cable leads to the caps.
Each half-pipe of gaps has a bit of perspex glued over it, both to contain the UV to propagate the arcing, as well as to prevent arcing from one bank to another.

The VR37 resistors are rated at 3.5kV each. So I took a gamble, and used four of those in each leg of the Marx. Two parallel, two series. This gives 390k at 7kV rating, which should be ok in this Marx, as they dont get overvolted much anyway (only very briefly). I've put most inside heatshrink, with a little hot glue as crude potting.

This Marx is a BEAST. After tuning, I can so far get about 1 spark per 4s, and top out at 18cm per spark, with a super bright flash and massive bang each time. The first time I ran it, I only fired off 3 sparks, but I felt physiological discomfort for half the next day! Most of the rate limit is due to my wimpy flyback driver, which I plan on boosting.

Firing it into CDs eats away at them, and even when confining the spark into a conduit, the flash overdrives a camera CCD.

Plastic of up to 2 millimetres thick is no match for this. It just pops right through it.

The best part about it is that it's really quite solid. With the gaps firmly glued into place and contained inside the frame, and with everything else held pretty solidly, it's quite portable and easy to move into position.

And it makes for great blog masthead photos :)


  1. I'm sure you know how fast that could kill you... awesome!

  2. Do you have a schematic for the BIPOLAR marx that you could post ?