A Bigger Marx Generator

One of my eBay finds arrived, this time from the US, 100 units of 15n 6kV dry mylar Sprague capacitors. An absolute bargin at $25 US for the 100. I didn't really expect dry mylar capacitors to do very well in Marx service, but I dismantled one to check its construction. Inside I found 3.5 metres of twin mylar films spirally wound, the capacitor is actually three 45n 2kV capacitors in series. Doing the math (there is a 4mm overlap in each cap) I can't get a result that is anywhere near 45n for a film thickness that could hold back 2kV, even with mylar's dielectric constant of 3.2 and strength of 15kV/mm, I took into account the multiple layers near each other, but there must be another effect that helps increase the effective surface area... The construction offers very low inductance, unlike a normal spiral wound capacitor. Destructive tests showed the capacitors are happy to operate at over 8kV with no damage. They are somewhat self-healing if they do punch through as well.

A single capacitor offers an profoundly loud bang and very fat spark. The quality of these devices is exceptional, they are intended for vacuum tube audiophile use, but clearly work very well in pulse discharge service. Cutting to the chase, built a 20 stage Marx generator out of them:

killer dry mylar marx generator

At 148kV, with a stored energy of 8.2J the spark is rather impressive. It can jump 150mm with ease and is absolutely scary in how bright and loud it is. The bank takes about 15 seconds to recharge sufficient to fire when coaxed with the old soda straw trick. It will self fire in about 45 seconds, most of the time. That is the one thing that is rather scary about this unit, it is unpredictable. I really need to develop a triggering system for safety's sake.

marx generator in dark

I used 330k charging resistors this time. I went to DSE to purchase some different resistors to try, and these 1W devices are the only value they had in stock in large quantities. The glaze on these devices is much better than my previous ones, easily handling the voltage with no corona or flash over. If anything this 330k is a bit low, I think I'll stick to 1M and above in the future.

During construction, six stages of this device killed my bench PSU (yes, *expletives removed* again!) despite all the precautions I was taking. This time the thermal overload circuit was destroyed, and lacking a TL071 or compatible device in stock it will stay that way until at least next weekend. In the future all experimentation will be done with the laptop switchmode PSU I've been using, it seems fairly unkillable.

Some more thoughts

This device triggered from the top down, which seemed to work just as well as from the bottom up. In fact it worked just as well when triggered in the middle. I can't see why two tower Marx generators could be charged in opposite directions and triggered in the middle (base of each).

Knowing the voltage of the top capacitor would be a real advantage. Even better would be to trigger the device based on the top capacitor's charge state. To achieve the former I can probably use a neon bulb and suitable resistor network, and the later with a fibre optic link from the voltage monitoring circuit to the trigger.

Three electrode trigger gaps sound the most practical, with the centre electrode biased via large value resistors to half the gap voltage. The trigger pulse can the fed to the centre gap with another gap, and can be sourced from a conventional xenon trigger transformer and a capacitor charged from the supply circuit. A SCR could be used as the trigger switching device. Or something as primitive as a mechanical switch. I have a pair of vacuum relays which could directly switch a few kV into the trigger electrode. A SF6 relay would solve a lot of problems! (I'll troll eBay for one.)

Sealing the entire device into an acrylic tube and using an air compressor to hold back ignition of the gaps could be an option as well. Release of the pressure would be the trigger.


Parent article: Marx Generator.