2003-01-28

Knowing Your Coronatron

I had a little time this afternoon, so I did some work towards a charge indication lamp for the top of the stack. I followed the same procedure as in my plasma diode detector (a big stack of 9v batteries and a 100k pot) to get some real-world numbers for neon ignition and hold voltages (75 and 56 volts for the device under test).

Next I did some measurements on the current required for a reasonable brightness, 10uA gives a visible glow, but at least 100uA is needed for a good solid glow that could be visible from across the room. This lead me to attempt to characterize the coronatron supply output. My back of the envelope calculations from capacitor charge time suggested the supply was putting out about 200uC/s (ie 200uA). The actual tests showed somewhat better performance, and a constant current supply like load curve:

Beizer smoothed, that data looks like this:

Which looks like a typical square-root law cascade (Cockcroft-Walton) multiplier output load curve, except it is around the wrong way, the more you load it the more current it delivers up to some limit, but it is anything by linear. Weird device, it is probably a current-limited supply, as its intended load in a coronatron that makes some sense. The coronatron wires are probably designed to leak as much as possible, the ion current being regulated by the supply limiting. I am too scared to measure its short circuit current with my multimeter, but I will try it in the near future with a D'arsonval micro-ammeter shunted by a neon for protection.

The supply pulls about 450mA from a 19.5V supply, the efficiency increases as the load drops (higher resistance), at peak it delivers about 1.5W into the HV load, or around 17% efficiency. Anyway, with only 200-300uA to play with, I can't really afford to burn up 100uA lighting a neon. I'll have to get cracking on building a neat and controllable kV@mA supply from those LOPTs I picked up last year.

Here is the prototype divider, designed for around 100uA and the nice fact that the strike voltage is about 1/100 of the 'fully charged' voltage, it worked out to around 80M:800K which I approximated with 8 10M resistors in series and 2 10M and a 1M in parallel shunted by the neon:

Check out the new capacitors that arrived today, 47nF 6.3kV units. I've also included a shot of the 3kV 10A vacuum relays I have. I am not sure if I can stretch them to 6-10kV without them breaking down? I guess at fairly modest voltages any x-rays produced would be too soft to escape the glass envelope? Surely no worse than the 148kV output pulse crashing into the strike electrodes.

Thought: a Marx generator as a flash x-ray machine PSU? I must build an ionisation chamber detector, I recently got some "Electrometer JFET" Op-Amp samples from National (bias currents on the order of tens of fA, or a few tens of microseconds between electrons!), they might work, but I'd need to make/buy a huge value (ie Gig-Ohms) feedback resistor. Perhaps I can use the leakage across a 'gimmick' made from twisted insulated wires, or a small value commercial ceramic capacitor?

Leave a comment on this article.

Parent article: Marx Generator.

© 2000 - 2009 Alan Yates - All Rights Reserved.

XHTML | CSS | CSS toggle