How to calculate peaking capacitor for Marx? / High Voltage / Forums

Forums

4hv.org :: Forums :: High Voltage

« Previous topic | Next topic »

How to calculate peaking capacitor for Marx?

previous 1 2 3 next

Move Thread

LAN_403

Proud Mary

Sat Nov 24 2012, 06:05PM

Registered Member #543 Joined: Tue Feb 20 2007, 04:26PM
Location: UK
Posts: 4992

ahahn wrote ...

Okay, so as I read Pai, it seems that the risetime of a Marx with peaking capacitor is wholly independent of the Marx itself -- it all depends on the peaking cap. In other words, if I have a suitably fast and low inductive peaker, I ought to be able to pump the laser with the sloppiest Marx in the world and still have it work.

I guess the main characteristic that still matters is the Marx being able to charge the peaking cap fast enough.

So what good would additional pulse-sharpening gaps or saturable reactors do?

I have a growing sense of unreality about all this, so will leave it to others who may have a better understanding of what you are trying to achieve to advise you further.

ahahn

Sat Nov 24 2012, 06:19PM

Registered Member #6075 Joined: Wed Aug 08 2012, 11:48PM
Location:
Posts: 29

Sorry, I was taking things to the logical extreme -- the peaking cap obviously can't compensate for all deficiencies in the Marx. I'm just having trouble understanding where it starts being 'not enough' as it were. This is doubly important since hitting a 1nS rise for atmospheric excitation (it's a target, I realize I may need to break out the vacuum pump) means absolute minimum inductance and therefore short current paths... there isn't much room for additional add-ons, and all the Marx-powered lasers I've seen either had the peaking caps right next to the tube or (more commonly) physically wrapped around or part of it. And nobody's used pulse-sharpening gaps or saturable reactors.

[added]
I think I get it. Since there's no way I can mitigate major losses due to corona etc at full erect voltage -- and since I have to worry about dielectric breakdown if it's charged longer than a few tens of nS! -- the handful of pF in the peaking cap won't stay charged very long. So the voltage coming in to the peaking cap still has to have a substantially quick rise time (say within an order of magnitude or two) else the whole thing falls flat... and there's plenty of room for extra gaps and such in front of the peaking cap to take care of that, rather than behind it like I was thinking when I read your post.

Still not sure if a saturable inductor (you did mean inductor? a saturable reactor is something wholly different, now that I think about it) that can handle the voltages, currents, and risetimes is within reach, though.

ahahn

Sat Nov 24 2012, 06:49PM

Registered Member #6075 Joined: Wed Aug 08 2012, 11:48PM
Location:
Posts: 29

I'm having trouble finding a good reference on saturable inductor pulse compression systems, anyone have any? (This is not the same as a saturable reactor:

ahahn

Sat Nov 24 2012, 07:21PM

Registered Member #6075 Joined: Wed Aug 08 2012, 11:48PM
Location:
Posts: 29

Hmm, I think Pei et al drew their Marx wrong way round... when the first gap fires, the top (+ charged side) of the first capacitor gets shorted to ground... or perhaps it's just intended as an inverting system, such that it generates a negative-going pulse from a positive charging voltage.

Proud Mary

Sat Nov 24 2012, 09:04PM

Registered Member #543 Joined: Tue Feb 20 2007, 04:26PM
Location: UK
Posts: 4992

If you Google Marx wave erection with any or all of the authors below you will find a number of papers addressing the ultra-fast Marx, its problems, and how to overcome them: Mayes, Carey, Nunnally, Altgilbers.

ahahn

Sun Nov 25 2012, 08:29AM

Registered Member #6075 Joined: Wed Aug 08 2012, 11:48PM
Location:
Posts: 29

Okay, so I was thinking about this last night, and I'm wondering if there might be more to an inductively charged Marx than just wave erection. Assuming the voltage stays high enough long enough for the inductors to partially charge -- since after all, the inductors starting to draw off current ought to be one of the main limiting factors to pulse length -- then as soon as the Marx is no longer erect then the inductors have current flowing with no place to go. That's the classic case for creating back-EMF. Only in a Marx we're talking an instantaeous current of many many amps and voltages of just as high.

This would suggest that, given big enough inductors, you'd actually see the Marx erect twice: once to discharge the capacitors, and then once again as the inductors discharge across the gaps. (Perhaps I'm wrong and the back-EMF rises fast enough to take advantage of the ionized gaps... this would explain Steve Ward's results to the effect that 4H (!) inductors charging his Marx made the fattest sparks.)

[added]
There's also the question of whether the inductors influence each other during the Marx's operation, especially if they're all parallel or in a row. This alone could go a long way to ensuring the stages operate simultaneously.

[added]
Though Steve Ward was using 4H inductors, in my (rather less exhaustive) tests I found 500uH inductors to work fine too (charging voltage <10kV :( ).

[added]
Drawing out the schematic of an erect Marx, it seems to me that when the gaps are firing (i.e. short) each stage becomes a parallel-resonant LC circuit, which is kind of odd. Not quite sure what the implications of this are. One might be much higher current through the spark gaps, which (according to the above formula for gap resistance) would mean a much lower gap resistance and potentially higher risetime.
On the other hand, parallel-resonant circuits might reduce the per stage charging voltage (or would they? that's for the case of a tank filter, maybe not applicable here).

[added]
This doesn't explain why Steve got such good results with big inductors. Since the resonant frequency of an LC circuit is 1 / ( 2 * pi * sqrt [L * C] ), the resonant frequency of a Marx stage with 1nF and 4H is 2500Hz -- whereas with 500uH the frequency goes up to 225kHz. So it's probably a stored-energy thing rather than a resonant one -- if anything, the LC circuit's low resonant point ought to be reducing the risetime by acting as a low-pass filter!

Similarly, the circulating current for a parallel-resonant circuit of Q > 10 being Icirculating = Q * Iline (where Iline is the current through the circuit, and Q is the parallel load resistance [of the parallel-resonant circuit, in this case the stray resistance and gap resistance of a single Marx stage] divided by the reactance in ohms of either the inductor or the capacitor) suggests that a high-resistance inductor would actually reduce Q... but perhaps the 4H Ruhmkorff coil inductors were designed for high-Q.

ahahn

Sun Nov 25 2012, 08:31AM

Registered Member #6075 Joined: Wed Aug 08 2012, 11:48PM
Location:
Posts: 29

Proud Mary wrote ...

If you Google Marx wave erection with any or all of the authors below you will find a number of papers addressing the ultra-fast Marx, its problems, and how to overcome them: Mayes, Carey, Nunnally, Altgilbers.

Thanks! I actually read of some of them in your previous posts on the subject, and I've been slowly working my way through their papers. They seem fairly abstract, though.

ahahn

Sun Nov 25 2012, 08:42AM

Registered Member #6075 Joined: Wed Aug 08 2012, 11:48PM
Location:
Posts: 29

Right, so it turns out Pai gives the equation for matching a peaking capacitor, I'd just missed it on first go-round -- Rload has to be greater than 2 * sqrt (Lstray / Cpeaking). Lstray is in this case the stray inductance of the peaking capacitor current loop. If this condition isn't met, the peaking capacitor current loop operates in an oscillatory mode with drastically increased risetime.

However, in one of the other papers I read, the authors described a peaking-cap equipped Marx where the risetime went up when the peaking cap value got too high... so presumably there's more to it than just that simple condition. (That was one of the fancy research-industrial-complex stainless-steel-clad Marxes with a risetime to make you drool and a peaking cap in the 10s of pF.)

[added] Oh, and according to Pai you can approximate the risetime of the peaking cap-stray inductance-load current loop with Trise (is approximately) 2.3 * (Lstray / Rload).

Proud Mary

Sun Nov 25 2012, 10:57AM

Registered Member #543 Joined: Tue Feb 20 2007, 04:26PM
Location: UK
Posts: 4992

There are some useful-looking sections on Marx-related pulse sharpening and peaking capacitors in this here paper:

ahahn

Wed Nov 28 2012, 11:19AM

Registered Member #6075 Joined: Wed Aug 08 2012, 11:48PM
Location:
Posts: 29

Thanks! Added it to the queue.

Research update: so in trying to figure out how to get the pulse to the laser head, I realized that the (infamous) Tesla bifilar pancake coil is just the mutant love child of a stacked transmission line and a pulse transformer... which means I could possibly both amplify and sharpen the output of the Marx by putting a one-turn primary around such a pancake and driving it with the Marx -- essentially creating a very high voltage flyback converter with a spark gap as a rectifier (because the output only gets to very high voltages when the pulse is rising in one direction, when you run a coil in flyback mode).

[corrected -- stacked, not Blumlein transmission line]

Of course, this raises the question whether a higher output voltage will mean more energy in the fast-rise part of the pulse's leading edge, and whether a 30kV delta-V in 1nS is just as good as 0-30kV in 1nS for laser purposes.

previous 1 2 3 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob